Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research. (1980)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

CHAPTER 2 BIOLOGY OF HORSES AND BURROS INFORMATION NEEDS The passages quoted from PL 95-5l4 and the BLM/NAS contract in Chapter l itemize in some detail the categories of desired information on horse and burro biology: population dynamics, food and habitat requirements, use of forage and water resources, and population controls. PL 95-5l4 states further, in Section l4 (b)(l), that: The Secretary shall maintain a current inventory of wild free-roaming horses and burros on given areas of the public lands. The purpose of such inventory shall be to: make determinations as to whether and where an overpopulation exists and whether action should be taken to remove excess animals; determine appropriate management levels of wild free-roaming horses and burros on these areas of the public lands; and determine whether appropriate management levels should be achieved by the removal or destruction of excess animals, or other options (such as sterilization, or natural controls on population levels).... Where the Secretary determines ... that an overpopulation exists on a given area of the public lands, and that action is necessary to remove excess animals, he shall immediately remove excess animals from the range so as to achieve appropriate management levels .... This heavy emphasis on the definition of "excess" animals, and the fact that the Committee was invested with the responsibility for helping the Secretaries of Interior and Agriculture make that determination, necessitate a clarification of the term "excess," and of its implications for informational needs. The term has at least three connotations: "excess" can have the sense that (l) the number of animals is detrimental to their own condition and welfare, (2) the number of equids adversely affects the condition and welfare of the other components of their ecosystems, and (3) the number of equids interferes with other management objectives for the public rangelands. The latter two connotations will be considered in Chapters 3 and 4; the first is addressed in this section. An excess of animals threatening to their own welfare will occur when a population has risen to densities at which its members' 19

20 behavior, demography, and state of health fall below some specified levels or standards. Establishment of such standards, and hence the determination of "excess," requires knowledge of the animals' social- and maintenance-behavior repertoire, their demographic characteristics, and facts about their nutrition and other health conditions. It must be determined how these vary with population size, adequacy of forage, weather conditions, competitors, and other environmental variables. Physiology, as well as food preferences and consumption, must also be understood. The sections that follow explore what is known about these subjects. STATE OF KNOWLEDGE Information Sources Committee members conducted preliminary literature searches by examining published research reports, unpublished theses, impact statements, and the popular literature. Bibliographies in these documents were then traced and cross-referenced. The information available on the biology of wild equids was found to be incomplete and superficial in some subject areas, irrelevant to the Committee's purpose in some aspects, and quite complete in others. Wildlife Review (from l936 to the present), the Denver Library (in cooperation with the U.S. Fish and Wildlife Service), and BIOSIS (Biological Abstracts) were also consulted. BIOSIS alone produced over l5,000 citations and associated codes (l969 to l979) that were flagged on the single keyword "Equidae." These citations were obtained and screened. (Broad subject categories were created and all citations were placed in appropriate sections. These files are available at New Mexico State University.) Other computer-based reference sources that were searched included Medline (Index Medicus) and Dialog (through the Lockheed system). Science Citation Index and the bibliographies of individual articles were also consulted. Since BLM and USFS files contain considerable amounts of field data, representative samples were sought by attempting to identify management units that: (a) had a history of intense horse and/or burro activity, and (b) contained information concerning economic and social considerations. Other factors applied to the selection of units included the existence of regional collection corrals, intense public interest, general completeness of available records, duration of available records, history of research programs, adequacy of land-use plans, etc. Management units thus identified were visited, appropriate personnel were interviewed, and available records were copied and centralized for cataloging and evaluation. These units included: Phoenix, Arizona (BLM); Palomino, Nevada (BLM); Susanville, California (BLM); Vale, Oregon (BLM); Lakeview, Oregon (BLM); Burns, Oregon (BLM); Salt Lake City, Utah (BLM); Pryor Mountains, Montana (BLM); Rock Springs, Wyoming (BLM); Modoc, California (USFS); and Jicarilla, New Mexico (USFS). Interviews were conducted during late August and

2l early September l979. Because Committee member Walter Conley conducted a 2-year research project from l976 to l978 at Jicarilla and the records were already complete, no interviews were required there. In addition, the entire records file of the BLM (previously located at the Denver Center) was obtained, reviewed, copied where pertinent, and returned. Data from this review are filed at New Mexico State University. Personal contacts were made with three recognized authorities in the field of equine nutrition: Drs. P. V. Fonnesbeck and L. M. Slade of Utah State University and Dr. H. F. Hintz of Cornell University. They were asked for ideas and sources of information. Mr. Montague Demment, University of Wisconsin Zoology Department, prepared the detailed review included as Appendix A. History and Paleontology of Equids in North America The horse (Equus caballus) and the African ass (E. asinus) were introduced to North America by the Spanish in l495 during Columbus' second expedition to the New World (Denhardt l95l, Brookshire l974). Horses and asses were an integral part of Spanish exploration and colonization and feral populations took hold throughout the Southwest in the l6th and l7th centuries. Native Americans incorporated equids into their cultures and horse and ass populations—both wild and domesticated—spread rapidly throughout the western states. Feral populations increased when animals escaped and were released from ranching and mining activities in the l9th and 20th centuries. Thus horses and asses have a long history in North America and may have been feral in the western United States since the l600s (McKnight l957, l958). The mainstream of equid evolution occurred in North America; only a few types wandered into the Old World in the Tertiary (Romer l966). Native equids were present from the lower Eocene, 55 million years ago (Colbert l969), and remained abundant until ll,000 years ago. Late Pleistocene mammal sites in Arizona (Lindsay and Tessman l974) reveal that Equus was second in abundance only to Mammuthus and was twice as abundant as Bison. The disappearance of equids ll,000 years ago coincided with the extinction of three genera of large mammals, and the immigration of a new predator, Paleolithic man. Martin (l973) has formulated a blitzkrieg model that postulates a spread of Paleolithic hunters so rapid that it resulted in the mass extinction of large mammals within l,000 years. Dating of fossil remains of type Equus (large), as well as other Equus species including E. asinus, has established their ages at between ll,000 ± l00 and l3,3l0 ± 2l0. These remains came from sites in Nevada, Arizona, and California (Haynes l967, Mawby l967, Hemmings l970, Havry l975, Cole and others l979). Skinner (l972) maintains that certain species groups of Equus have existed for 3 to 5 million years. He presents evidence that there is a high degree of similarity between extinct Pleistocene and living

22 equids. Skinner lists the type species of the subgenus Dolichohippus as being Equus grevyi Oustalet, l882, distributed as follows: "extinct in Pleistocene, North America; living, Ethiopia and Northern Kenya." He describes the type species of subgenus Equus (Heminous) as being Equus hemionus Pallas, l775, with the following distribution: "Pleistocene, North America, and living, Asia." Pleistocene deposits show specimens ranging from Texas to Alaska, and Kansas to Arizona. Specimens referred to as Equus (Asinus) cumminsii Cope were found in Pleistocene deposits at Blanco, Texas and Deer Park, Kansas. Equid fossil remains are rarely identified to the species level. However there were ass, horse, and zebra types present in Pleistocene North America, and the skeletal morphologies of the fossil and the reintroduced equids are anatomically indistinguishable (Cole and others l979). Thus although feral horses and asses are considered alien or 'exotic1 today, they represent lineages that have a long paleohistory in North America. This is particularly important to the interpretation of their role in modern ecosystems. The concern on the part of some people that feral horses and asses are detrimental to their habitat is partially based on the assumption that since they are exotic they are particularly disruptive to vegetational communities with which they have not coevolved. However, modern-day equids in North America are not typical exotics. A long period of coevolution between their evolutionary predecessors and the vegetation was broken for ll,000 years, which is a brief interval in geologic time. Whether or not the vegetation today retains the same antiherbivore adaptations it had developed by the time equids became extinct at the end of the Pleistocene is a moot question. Paleobotanical evidence shows distributional changes in the vegetational zones, which were depressed from 600 to l,000 m (Martin and Mehringer l965, Van Devender and Spaulding l979). But to our knowledge, no one has produced any evidence that native plant species have lost adaptations to grazing and/or browsing pressures (e.g., oily foliage, spiny or thorny branches, siliceous stems, or toxic alkaloids) that are the result of selective pressure exerted during millions of years of coevolution with equids. However, several authors (Young and others l976, l979) have postulated that the marked changes in vegetation structure in the Great Basin following the l9th-century introduction of sheep and cattle were due to the lack of heavy, post-Pleistocene pressure from large herbivores. Without such pressures the vegetation had lost defenses, according to this theory, and was vulnerable to the introduction. Some plant species are known to be highly plastic, and to respond to selective pressures in short periods of time (Dyer l968), including grazing (Stapledon l928, Lodge l962). If species optimize traits over an array of selective pressures, as modern evolutionary theory holds, then it is reasonable to postulate that some species may have evolved away from antiherbivore defenses in the absence of grazing pressures. Of course, there is no way to determine differences and/or similarities in behavioral ecology between the equids present in North

23 America ll,000 years ago and the recently introduced species. However, it is an oversimplification simply to dismiss feral asses and horses as "exotics." The possibility of their filling an "open niche" remains (Martin l970). Behavioral Ecology of Equids Genus Equus contains six living species. They are Equus caballus (horse), E. burchelli (plains zebra), E. zebra (Hartman's or mountain zebra), E. asinus (African ass), E. hemionus (Asian ass), and E. grevyi (Grevy's zebra). The genus is characterized by two distinct types of social organization. The harem or stable family group with a dominant male has been described for the plains zebra (Klingel l967, l972), Hartman's zebra (Lingel l968; Joubert 1972a, b), and the feral horse (Feist and McCullough l975, l976; Green and Green l977). The territorial form of social organization, in which stable bonds occur only between mother and offspring, has been described for Grevy's zebra (Klingel l969), the Asian ass (Klingel l977), the African ass (Klingel l972, l977), and the feral ass (E. asinus) in North America (Moehlman l974, l979; Woodward l976, l979). These two types of social organization appear to be the extremes of a continuum that ranges from a system in which territoriality plays an important role and social bonding is limited, to the more socially organized and cohesive family groups. The interrelationship of social organization and environmental parameters has been the focus of review and interpretation by Crook (l970), Jarman (l974), Kaufman (l974), Klingel (l974), and Fisler (l979). Many factors appear to be important in the establishment and/or reinforcement of a social system. In particular, resource availability, feeding ecology, daily activity patterns, and demography are critical in determining social behavior, social bonding, and the type of social organization maintained. Most reviews of social organization and ecology have been on the interspecies level. However, an examination of available data on feral equids does reveal intraspecific behavioral plasticity. Feral Asses Since the early l970s, field studies have been done on several feral ass populations in the southwestern United States. Table 2.l lists the study populations, researchers, and some pertinent facts from each study. The basic pattern of social organization and demography was the same in all southwestern study sites. The only stable unit was mother and offspring. Some typical temporary groups were females and offspring, bachelor males, and mixed groups. Adult males were often solitary. Statistical treatment of group size and composition is only available for two studies (Moehlman l974, Woodward l979a). There was evidence of territoriality by a small percentage of the males on all

24 X *rl b 0 rH i-l rH r.H 1 i i li i i i oi .y •• ft rH i :' i i i rH rH i i i i i i N 0i 0 0 rH O (M i i i i i i JC X i i i i i O 00 tN 01 l i i i • Er u s t-l i-l Sn B 0 0i M X X M i 1 i i i M i i •H rH « i i i i 14 td o) ' i i i i ' rl JC IH oi a. EH X • CO i-l m o in in i i O O 0 9 i i i i td &- fO 00 01 •H in 0) O rH H «y — t n S cu cu I )ii ai (il (II (ll ai I 1 CN t V.s M "A "jj n "g w A 1 A , i , ! o 0 0 00 01 l i 00 i i 0 to i i S rH s s on r- rH a to ro - r- U) vO JS a r- to o1 t- r- 01 00 H rH 01 01 rH H g 1 1 4! 01 & r- S*1 rH 0) 01 g1 | rl tJ 00 i rH H N I 01 en 2 £ on r- ~. Ol Ol 0 •-t II M Q Ol rH !"• to m 0i i-l i^j 00 rH — 01 rH 1 1 Moehlman Norment Woodward Seegmill Walker & O'Farrel Farrell Carother 1 I Morgart Moehlman McCort ( • ^ r^ i0 m CO m in m 01 i i i i rl A 4 4 m i ° in T •o" 2 2 2 01 01 01 on 01 01 en en § rH rH rH 41 rH rH i rl 0) c • 111 JH 3 o) at ^ * •U • £ rH IH A rH 3 California Death Valley Nat'l. Mo n amen Chemehuevi Mts Arizona Bill Williams Lake Mead Nat' Recreation A 3 Grand Canyon Havasu Resourc New Mexico Bandelier Nat' Georgia Ossabaw Island en M i UI 0) 1 M rH

25 study sites, but only the Wildrose population in Death Valley (l970 to l974) and the Bandelier population exhibited strongly expressed territoriality. There appears to be some confusion in Woodward's (l979a) discussion of territoriality. Territorial behavior in both Grevy's zebra and the African ass (Moehlman l974, Klingel l979) does not involve boundary marking and defense against all conspecifics, but rather the domination of a particular area by a single male who has sole access to estrous females within that territory. Short-term observations (l month) were made on a population of feral asses in a very different habitat: Ossabaw Island, Georgia (Moehlman l979). This island is characterized by lush vegetation and a warm, humid climate. There were approximately 40 feral asses on the island, with a density greater than I/km1. Although the Ossabaw observations were limited, they do illustrate the plasticity possible in the social behavior and organization of feral asses. Contrary to the situation in Death Valley, stable groups existed. At the south end of the island there was a stable group that exhibited all the characteristics of a harem. A group of l3 asses at the north end of the island were associated in 63.6 percent of the observations. In this group the dominant male did all the courting and copulating. These observations have since been substantiated by McCort (l979). Feral asses have potentially large home ranges over which they move in a seasonal pattern. Recorded individual home ranges, which probably reflect a minimum, range in mean size per population from 2.86 to 68.l km*. The largest home range observed was l03.6 km2 in Death Valley National Monument. Except for the stable group of mother-offspring (X = 2) all groups are temporary and their composition can be as follows: (a) all male, (b) two or more females and offspring, (c) mixed adult males and females and subadults, and (d) yearlings. Males are often solitary, and in the Death Valley study (Moehlman l974, l979) 23.9 percent of the observations (N = l,l58) on population grouping patterns involved solitary males. The general trend was toward small groups, and 57.8 percent of the groups contained 2 to 4 individuals. Large aggregations (8 to 2l) occurred rarely (3.3 percent) and were associated with scarce resources such as water and/or estrous females. The Chemehuevi population followed this general pattern (Woodward l976, l979a). Data for group sizes are not available from the other studies. Information on spacing between individuals is available from Death Valley (Moehlman l974). Data taken at 5-minute intervals (N = 2,9l5) revealed that adults (male and/or female) spent most of their time separated by distances greater than l0 m. Individuals that spent most of their time within 4 m of one another were usually genetically related. However, adult females allowed the yearling offspring of other females to come closer to them than they would permit their own yearlings. Thus, an observer lacking knowledge of individual relatedness could easily misinterpret how many offspring a female had and how often she was reproducing. This presents obvious problems in present census techniques. Distances between males and females decreased only when the female was in estrus.

26 During the hot summer months asses tend to drink once every 24 hours and water is a critical factor in their distribution. In most study areas, asses were concentrated within 3 km of water sources during the summer months. From September through the winter months, they were less dependent on water and were found 8.0 to 9.6 km away from it. Animals also moved to lower elevations during the winter months. Asses are physiologically well adapted to life in an arid habitat. They can sustain a water loss of up to 30 percent body weight and can drink enough water in 2 to 5 minutes to restore fluid loss (Maloiy l970, Maloiy and Boarer l97l). A recent study (Tomkiewicz l979) on heterothermy and water turnover using temperature-sensitive implants revealed that body temperature could vary from 34.0 to 4l.6°C and was dependent on air temperature. In the summer males had a lower mean body temperature (36.5°C) than females (38.2°C). Mother asses in Death Valley (Moehlman l974) showed a high threat and rejection rate when foals attempted to nurse, presumably because of fluid stress. Nursing rejections began on the first day and increased in intensity as the foal grew older. Rarely did a foal attempt to nurse without being threatened by its mother. Until the foal was a month old its nursing success, calculated by observing each attempt, was 82 percent. By the time it was 3 to 4 months old the success rate had diminished to 35 percent. This rejection is similar to weaning behavior in other equids; it differs mainly in that it commenced when the foal was so young. Lactating females are under more severe fluid stress than other adults in the population. These animals watered more frequently during the hot months (2 to 3 times per day, in comparison with once a day for other adults). Foals ate vegetation, but they did not drink water until they were about 3 months old; the mother provided her foal with most of its fluid intake during that period. Nursing provides certain safeguards for the foal, since it does not have to compete for water at the springs. However, it may increase the fluid stress on the female in what is already an arid and difficult environment. Thus, the female may need to regulate the interval of nursing so as to moderate fluid stress. This physiologically based behavior might in turn stimulate independent and aggressive behavior on the part of the foals, and the entire mechanism may contribute to the lack of bonding between adults in this population. Two-day-old foals were already threatening conspecifics, particularly when they saw their mothers being approached by others. In all populations studied in the Southwest, asses tended to congregate near scarce water sources during the hot months (June through August). Such a high density during one portion of the year results in a more severe impact on the vegetation in the immediate vicinity of the water source (Fisher and others l973, Hanley and Brady l977, Woodward and Ohmart l976, Norment and Douglas l977). This phenomenon will be discussed in greater detail under "Range-Plant-Community Impacts" in Chapter 3.

27 Hanley and Brady (l977) studied precipitation, soil texture, soil moisture, and browse utilization patterns on sites at increasing distances from the water source. They found that: "Browse utilization ranged from heavy to light with increasing distance from the Colorado River. Overgrazing occurred near the Colorado River but decreased to light or moderate use at distances greater than 2.5 km from water." Burro impact was greatest in the secondary wash communities (Cercidium-Larrea). Apparently no species were acting as increasers or invaders under heavy burro utilization pressure in this locale. There is no information in this study on burro densities. Hanley and Brady also point out that assessment of impact is complicated by the need for data concerning the amount of time and the burro densities required to modify the community structure. The same general trend of heavy impact near water sources in the Panamint Range was evident in a report by Fisher and others (l973) . However, at an exclosure approximately 2 km from the water source, measurement of vegetative species density and volume showed higher biomass outside than inside the exclosure. Annual densities and species diversity were higher outside the exclosure, but a bunch grass, Stipa sp., was more abundant inside the exclosure. Secondary production was higher in the area used by feral asses. Two studies in the Grand Canyon (Carothers and coworkers l976, Cole l979) analyzed canopy cover of predominant species. However, their results disagree and have different implications. Cole suggested that the degree of burro impact on vegetation is complicated and changes over short distances on the same slope and substrate. Neither study indicated distances from water or ass densities. O'Farrell (l978), studying feral asses in the Lake Mead National Recreation Area, also described the general trend of ass populations to clump near water during the hot summer months. Maximum density values were X = 0.93 per km2, and the greatest effect on vegetation was observed within a quarter mile of the water source. All of these studies provide interesting information on modification of vegetation by asses. However, all of the data were generated by indirect methodology. There is a distinct need for direct observations of ass feeding ecology and species utilization as they are related to availability, terrain, season, and climatic conditions. The movement of ass populations is restricted in the hot summer months because water is limited. Their seasonal movements and rotational utilization of vegetation within their home ranges has not been examined directly. Carrying-capacity figures alone are insufficient for assessing the relationship of ass populations to the vegetational community. Information on daily activity patterns is available from two studies: one conducted at Death Valley, California and the other at Ossabaw Island, Georgia (Moehlman l974, l979). Death Valley is characterized by a hot, dry climate with low and sporadic rainfall. The vegetation in the study areas was primarily browse. Death Valley offered a relatively unstable habitat in which water sources were of prime importance during the summer months. The feral asses combined

28 in small groups, had large home ranges, were distributed for most of the year at a low density, and characterically exhibited territoriality and a low degree of sociability. In hot weather the population tended to clump around water sources. At these times territorial males resided next to the water source, thus enhancing their chances of copulating with estrous females. On hot, humid Ossabaw Island, vegetation is lush and water supplies plentiful. The population was denser, had a smaller home range, and exhibited greater sociability and group cohesion. One group of five interrelated as a harem. A basic element of the daily activity pattern—percentage of time spent feeding—was quite different in the two populations. The Death Valley asses were primarily browsers and spent 5l.0 percent of their time feeding; Ossabaw Island asses were primarily grazers and spent 38.l percent of their time feeding. Jarman has emphasized the importance of feeding ecology in the determination of ungulate social organization. Browsing species tend to be less sociable, and grazers occupy the more social end of the continuum (Jarman l974). Determining factors in this trend may be the proportion of time required for feeding and the quality of forage. The Ossabaw Island population exhibited almost no aggressive behavior. In particular, females rarely threatened their foals, whose nursing success rate was much higher than that of their Death Valley counterparts (88 percent at 3 to 4 months). This lack of rejection was probably directly related to the fact that lactating females experience little water stress in the Georgian environment. Three categories of behavioral interactions that might affect social bonding and grouping in the two areas are greeting, mutual grooming, and social play. In Death Valley, greeting behavior occurred mainly between adult males (60.2 percent) and often (i.e., 63.3 percent of the time) involved aggressive behavior. Female-to-female greetings were rare (l.7 percent). Foals were involved in 34.0 percent of the greetings, of which they initiated l5.6 percent. Greeting behavior was rare on Ossabaw Island, which may simply be another indication that individuals were well known to each other and relationships were clearly established. Social grooming is usually considered to be important for group cohesion (Sparks l967). In Death Valley the general pattern of low sociability was also reflected in the rarity of mutual grooming (0.3 observation per hour). Ossabaw Island asses performed grooming much more frequently (l.5 observations per hour). Furthermore, three categories of grooming partners were seen in Georgia that were not observed in Death Valley, namely: male-female, female-nonoffspring foal, male-foal. This behavior is consistent with the general pattern of a cohesive group, relatively close spacing of individuals, and a low level of agressive behavior. Social play is the third behavioral category that may influence social bonding. This type of interaction was not observed between foals in the Death Valley population. These foals only exhibited solitary play, which consisted of such activities as using the mouth to pick up and drop objects, running in spurts, stopping quickly in

29 front of adults, pivoting and running away, and biting and mounting their mothers. In a more social equid population, i.e., New Forest ponies (Tyler l972), 3-week-old foals were observed with other foals and/or yearlings, and by the time they were 6 weeks old, social play constituted 55 percent of their play. On Ossabaw Island two male foals spent many hours engaged in social play, which usually consisted of play-fighting. These foals were the same age and associated regularly. There appears to be an upper limit on the size of an equid group in which individuals can express and maintain stable relationships and bonds. The maximum size of permanent groups in Hartmann's zebra (Joubert l972 a,b) is about l3; 65.2 percent of the groups have between 4 and 7 animals. Klingel (l967) reported an average group size of 4.5 to 7.5 for this zebra. The average size of a feral horse harem (Feist l97l) is 5, with a maximum of 2l. Thus, when feral asses did form permanent groups they followed the pattern for Equidae. Behavioral and ecological information on the two feral populations of asses indicates that although Equus asinus normally displays low sociability, this species does have the behavioral plasticity, given a favorable environment, to form highly social and stable harem groups. Feral Horses Feral horse field studies have been conducted in Nevada, Arizona, New Mexico, Montana, and Wyoming. Comparative studies are also available from Canada and islands off eastern North America. Table 2.2 lists field studies for which information was available. The basic pattern of social organization is a family group composed of a dominant male, subordinate adult males, females, and their offspring (Feist l97l, Pelligrini l97l, Feist and McCullough l975, Hall and Kirkpatrick l975, Welsh l975, Glutton-Brock and others l976, Keiper l976 a, Berger l977, Green and Green l977, Rubenstein l978, Salter l978, Nelson l979). Some family bands are single-male harems, but several authors report multi-male groups (Feist l97l, Keiper l976 a, Green and Green l977, Miller l979, Nelson l979). Klingel (l967) described a dominant male in a plains zebra family group who actively searched for a 4-year-old male member that had wandered off. Joubert (l972a) experimented with the effect of removing the dominant male from a band. The females appeared to be very closely bonded and actively searched for the missing male. They rejoined him when he was released a month later. The basic pattern of strong bonding between females persists in semi-feral situations where adult males are removed (Imanishi l950, Tyler l972). However, bands are not completely stable and changes do occur. Feist (l97l) found that immature females accounted for most of the changes between bands, and that these occurred during the breeding season. Klingel's (l969b) study of plains zebras recorded that young females undergoing estrus for the first time had a different and more distinctive standing posture than older females. He felt that this served as a strong advertisement of their reproductive status and

30 5 •3 I I 2 ,r i ! I i i S ct S « f & 1 r. iD c g. I i « C M T) O W ;;j •o a s* 1 M i 1 H C S 41 .H 3 § i i 1} fi 1 M tN a E X *"!§ in iC in A <r vC in r*, H iA m r* ,A tX vD o CNJ fN , CJ N It r i i i i r- i r- i 1£ O 0 g ' 1 o N fN i I."' 1 IA ^H n S n 1 s i R i S ! in 1 a r^ • CT. H 1 ,j *" k 1 S a s S p s ° j, 1* S r. H n . i 13 3 S .H -H n C M •a M • tr . a •S S | =r % •H n & •a « 8 ^ § 3 S I « 5 — t 0) I J s & S 2 § r

3l increased the possibility of their abduction from the family band. Nelson (l979) observed that 6l percent of the females moved away from their primary band. Mature females accounted for 78 percent of these movements. When strange individuals attempt to join a band, not only the dominant male, but also subordinant males and females will drive them away. All-male (bachelor) groups commonly occur, but tend to be unstable in composition. Adult males are rarely solitary. Dominance hierarchies have been recorded in bachelor groups (Feist l97l, Feist and McCullough l975) and appear to be important in determining which males are most likely to acquire females. Miller and Denniston (l979) have reported a further level of social organization in horses. By analyzing l22 encounters among l6 bands, they found a nearly linear interband dominance hierarchy in terms of access to a scarce resource—water. Berger (l977) reported a similar hierarchy among four bands in his study in Grand Canyon. Miller and Denniston propose the designation of the term "herd" for a structured social unit of bands that recognize each other and form a dominance hierarchy. Although horses have a typical social organization characterized by a lack of territorial behavior, this pattern can change under unusual environmental conditions. Rubenstein's (l978) study of the Shakleford Banks horses clearly illustrates this point. Two-thirds of the harems maintain clearly defined and stable territories. The island is narrow and visibility is good, so the energetic costs of defending a territory are correspondingly low. In addition, water and vegetation zones are evenly distributed along the island. Territorial harems are larger than those harems that utilize overlapping home ranges. Thus the structure of the island and the pattern of available resources has clearly influenced social organization in this population. The only other report of territorial behavior is for a feral population in the Wassuk Range, Nevada (Pelligrini l97l). However, territorial behavior in that report was not adequately documented. In hotter, drier habitats, both water and forage availability are major determinants of horse movements (Berger l977, Green and Green l977). During the summer months the animals tend to concentrate their activities within 3 to 7 miles of available water. In Grand Canyon, Berger (l977) found that home-range size decreased in successive warm months and that the time spent in drinking and resting increased, while feeding time decreased. Time spent feeding varied from approximately 50 percent to 30 percent. Green and Green (l977) observed seasonal migrations by part of the population from Stone Cabin Valley south to the Tonopah Nuclear Test Range. The plains zebra in the Serengeti will migrate up to l50 km to get to dry-season foraging ranges (Klingel l972) . Feral horses might also be capable of such long-range movements, but are probably restricted by topography and fencing. Anecdotal accounts of Asiatic wild horses (Mohr l97l) suggest that their home ranges are centered around water sources during spring foaling and the summer dry season, but that they greatly increase their range during the winter when snow is available.

32 Studies in Alberta (Salter l978; Salter and Hudson l978, l979) and British Columbia (Storrar and others l977) found that habitat utilization was determined primarily by forage availability and that other factors were of secondary importance. Horses spent 75 percent of their daylight hours feeding in winter, and the authors suggested that there was a decrease in diurnal feeding during the summer. Diet quality peaked in June and was low in January. Meadow and shrubland habitats were mainly utilized in spring, with a shift to forested habitats in summer and fall-early winter. Compared with deer, moose, and elk, horses were more ubiquitously distributed. Skelton (l978), using fecal-sample analysis of horses in the Camargue, found differences in diets according to season, age class, and herd membership. This last factor appeared to be important to the survival of Sable Island horses (Welsh l975). Bands that lost their stallion in the winter of l972 and were taken over by another and presumably less experienced male had the highest mortality. This outcome was attributed to bands moving to areas with poor shelter or food resources during a storm. Studies of feral horses on Assateague Island (Keiper l976a, l977b; Zervanos and Keiper l979) investigated social organization, activity patterns, and feeding behavior. These are the only studies that directly examined vegetation consumption by individuals in terms of time spent grazing per plant species. Watching different bands during summer and winter months, the authors determined that utilization of standing crop biomass by bands ranged from 5.l to 20.l percent in summer, and from 2.6 to 3l.0 percent in winter. These figures do relate to differences in band size. The observers also recorded seasonal differences in home range size (summer, 6.48 km2; winter, 4.32 km2). Horses remained active throughout the night, spending 49.7 percent of the nocturnal hours' feeding. Seasonal distribution of Sable Island horses (Welsh l975) was affected by plant-community type, shelter, and surface water. Welsh calculated that 300 horses (peak population) would eat l3 percent of the annual forage production. In spite of all these studies, there is still a dearth of critical information specifically examining horse energetics and feeding ecology. It is readily apparent that horses are selective feeders but that they can still get by on low-quality forage. However, there are no data on the diversity and abundance of vegetational species or on what type of forage is being selected seasonally by horses per sex and age class. Such data are critical to understanding reproductive potential (Clegg and Ganong l969, Hall l972). A great deal of data is available on equid social organization, demography, seasonal distribution and feeding ecology, daily activity pattern, and reproduction. For the most part, however, the information is of a sporadic and short-term nature. As yet no long-term (5 to l0 years) study has been performed, nor has any study correlated data on population size and demography, behavior, and feeding selectivity with seasonal availability and quality of forage and water. Activity-budget data are critical to determining the energetic requirements of feral horses and burros. The resulting

33 interpretation of nutritional status is particularly critical to understanding the lifetime productivity of females. Bquid Demography The reproductive performance and longevity of an individual animal is partly a function of its genetic make-up, and partly its environment operating through its nutrition, physiology, and behavior. The collective performance of numerous individuals considered together as a population is expressed as a natality or fecundity rate, and as a survival rate. Such rates can be expected to vary annually to some degree within a population as yearly environmental conditions vary, and between populations as genetic make-up and environments differ. These rates, in turn, interact to produce more synthetic population traits like age and sex composition, and rates of population change. The latter parameter alone discloses much about a species' ecological and evolutionary characteristics. It provides an indication of the demographic cushion with which it can absorb such environmental pressures as predation and competition, or human exploitation. Such related parameters as the net reproductive rate (average number of female young produced by the average female in her lifetime), mean generation length, and average reproductive value of an individual female at a given age all provide insights into the species' potential evolutionary plasticity. From the practical standpoint of equid management, estimates of herd increase rates disclose the rates at which animals must be removed in order to hold them at some decided-upon level. Or alternatively, such rates make it possible to predict rates of herd growth in the absence of any artificial controls. They may also provide checks on the validity of census methods. Since this has been a controversial subject in North American equid management, we devote considerable attention to the subject in the sections below. The Wild Horse (l) Fecundity. Considerable work has, of course, been done on reproductive patterns in domestic horses. But very little information is available on natality rates in free-roaming wild horses. The rates that characterize well-managed, confined domestics may constitute the maxima of which the species is capable under optimum nutritional and breeding conditions, and under protection from the elements. The available data from wild herds indicate lower rates, doubtless the cost of field breeding, less-than-optimum nutrition, and the rigors of range life. (a) Age at first female breeding. Although ovulation and conception begin at l year of age in domestic fillies, there appears

34 to be a high fetal loss in mothers of this age. In one study (Ginther l979) of maces confined to pens or small pastures and fed supplementary diets, 69 percent of the yearling females conceived but only 44 percent delivered at age 2. In all the published material we examined, we found only one instance of conception at l year of age and foaling at 2 in wild horses. This was reported by Tyler (l972) who, in a study of New Forest ponies in Britain, observed one such case in l07 2-year-old mares over a 3-year period. Elsewhere, no 2-year foaling was observed in the following studies: (i) Boyd (l980) in a 2-year study of a 400-horse herd in Wyoming. (ii) Keiper (l979) in a 5-year study of Shetland ponies on Assateague Island (Maryland-Virginia), the population averaging about 300 animals. (iii) Welsh (l975) in a 3-year study of 227 to 306 wild horses on Sable Island, Nova Scotia. Youngest mare seen copulating was 26 months of age. (iv) Feist and McCullough (l975) in a l-year study of 270 horses on the Pryor Mountain Wild Horse Range, Montana. (v) Nelson (l979) in a l-year study of ll6 horses in the Jicarilla District of the Carson National Forest, New Mexico. (b) Age-specific and herd fecundity, In one study of confined and well-fed domestics (Ginther l979), the percentages of mares foaling rose from around 65 percent in the younger animals to values of 8l to 89 percent in ages approaching l2, then declined to 50 at age 20. In total, about 80 percent of a mixed-age herd became pregnant, while about 70 percent bore young. Since the gestation period is approximately a year, and mares experience post-partum ovulation and conception, these percentages approximate the percentage of mares in a mixed-age herd that could be expected to produce young each year. Removal of the younger animals, and keeping the herd stocked with the more fecund, 8 to l2-year-old animals could increase the expected annual foaling rate. An important paper by Speelman and others (l944) summarizes reproductive rates of domestic mares raised under western range conditions near Miles City, Montana. These authors reported the foaling rates of 209 mares, each observed through several breeding seasons, so that a total of 953 individual breeding cycles was observed over a l5-year period. Range forage was the primary food source, and foals ran with mares until weaning, when they were given grain and hay until l year of age. From l year to maturity, and thereafter, pastures provided most of the food. No supplemental food was given except during periods of exceptionally severe weather or during work periods.

35 The 953 matings produced 567 pregnancies (59.6 percent) and 568 foals (one set of twins was born). Age-specific fecundity of these animals is shown in Table 2.3. Here, as in the study reported by Ginther (l979), fecundity rises as the mares age from 3 to 7 years, reaches a maximum in about the 8- to-ll-year class, then declines somewhat in the older ages. But the foaling rates are generally lower in these range-reared animals. These authors cited comparable statistics from other studies with percentages varying from 42.3 to 72.0 for herd means. In their own studies, annual mean percentages for the entire herd varied from 43.5 to 73.7. Statistics of comparable volume and detail are not available for wild horse herds, but 3-year-old, 4-year-old, and herd foaling rates have variously been reported in the studies cited above. The manner of reporting has not been uniform among the studies, making comparison somewhat difficult in places. But it seems worthwhile detailing these results, recalculating where possible some of the authors' data in order to allow maximum comparison: (i) Tyler (l972) detailed the age composition of the mares on her study area in each of the 3 years of her study (see her Table l). The number of mares 3 years and older totaled l98, 2ll, and 2l9 in l966, l967, and l968, respectively. She also reported the number of foals born in each of these years at 99, l09, and 84. These values produce annual foaling rates for 3-year-olds and older of 50, 52, and 38 percent. She states that "Most mares foaled for the first time when 3 or 4 years old, but some not until they were 5 years old." But since the foaling performance of these different age classes was not given, no other rates can be calculated. It should be pointed out that the colt foals were removed each year from this population, somewhat similar to the situation on Chincoteague National Wildlife Refuge, as discussed below. (ii) Boyd (l980) reported that ll percent (N = 9) of the 3-year-old mares under observation in l978 foaled, 33 percent (N = l2) in l979. In these years the percentage of 3-year-old and older mares foaling was 78 and 53; of the 4-year-olds and older, 86 and 55. (iii) Keiper (l979) observed two populations on Assateague Island: (l) Those on the Assateague Island National Seashore (AINS, northern portion of island) which were unmanaged and allowed to pursue their own demographic structure and performance; (2) those of the Chincoteague National Wildlife Refuge (CNWR, southern portion of island) which were privately owned, mixed around each year presumably for husbandry purposes, and from which the foals were removed at the end of each summer for sale.

36 Table 2.3. Age-specific Percentages of Mares Bearing Foals Each Year in Range-Reared Domestic Horses (data from Speelman and others l944) . No. Mares % Bearing No. Mares % Bearing No. Mares % Bearing Agel Agel Agel 3 4 25.0 9 78 65.4 l5 24 50.0 4 99 56.6 l0 67 67.2 l6 18 6l.1 5 l25 57.6 ll 56 69.6 l7 l2 4l.7 6 l05 58.l 12 47 57.4 18 7 0 7 96 60.4 13 40 50.0 l9 2 0 8 94 70.2 14 29 58.6 20 2 50.0 ? 48 54.l •"•This is the age at foaling. The foals were sired l year earlier. Keiper reported foaling rates and numbers of mares for the two populations in his Table 4 as follows: Year l975 l976 l977 l978 l979 Mean AINS 58, 64. 70, 70. 8(l7) 3(l4) 5(l7) 0(20) 43.5(23) 6l.4 CNWR 70.9(38) 8l.0(37) 75.0(37) 70.8(24) 74.4 He attributed the differences between the two populations to the foal removal on CNWR, and resulting relaxation on the mares' physiological resources of not having foals suckling through the subsequent pregnancy. One might also speculate that the annual foal removal could have produced an older mare population with higher fecundity. The CNWR age composition was not given, and hence it is impossible to calculate age-specific fecundity rates for the population. Such rates can be approximated for AINS, although we have encountered some difficulty in discerning the consistency between Keiper's Tables l and 4. He states in the footnote to Table l that all foals on AINS were from "mature" (4 and older) mares except for three born to 3-year-olds. The 5-year summary in this table lists 85 mature mares and 52 foals, for a foaling rate of 6l percent, the same mean as in Table 4 (see above). But Table 4 shows a 5-year total of 9l mares and 55 foals. Since the 5-year total of "immature" mares was 74 (his Table l), and somewhere between a third and a half of these were probably 3-year-olds, the 9l in Table 4 cannot have been the combined 3-year-olds and older. Since the 9l is close to the 85 "matures" in Table l, we assume the fecundity rates in Table 4 and above are those for 4-year-olds and older, and are so recorded in our Table 2.4. If we assume conservatively that a third (25) of the 5-year total of 74 "immatures"

37 in his Table l were 3-year-olds and add these to the 85 "matures," then the 52 foals constituted approximately a 47 percent 5-year- average foaling rate for the 3-year-olds and older (52/ll0 X ll0). This would also imply a 3-year average 3-year-old foaling rate of somewhere around l2 percent (3/25 x l00). (iv) Welsh (l975) provided the most detailed analysis of the fecundity of a wild horse herd, summarizing foaling rates over a 3-year period for 3-year-old mares, 4's, and the 5's and older which he called "adults." The results are as follows, the sample sizes given parenthetically after each rate: % Foaling By Year Age 3 4 l970 l97l l972 Mean 0(6) 0(l) 63(76) l9(l6) 50(6) 84(76) 0(6) 25(8) 52(45) ll(28) 33(l5) 69(l97) 5 + In order to compare these with the results of other studies, which have combined results in slightly different ways, it is of interest to calculate the rates for 4-year-olds and older, and for 3-year-olds and older. We have done so as follows for the above columns: 3+ 58(83) 7l(98) 42(59) 60(240) 4+ 62(77) 82(82) 47(53) 66(2l2) (v) Feist and McCullough (l975) considered all mares 4 and older to be "adults," the 2- and 3-year-olds to be "immatures." In the year they observed the Pryor Mountain herd, 78 adult mares bore 33 foals, a foaling rate of 42 percent. In addition, two foals were produced by 3-year-olds. Although the 3-year-olds were not enumerated, the immature age class numbered 27 and we can hypothesize conservatively that about l0 of these were 3-year-olds. On this assumption, the 3-year-old foaling rate was somewhere near 20 percent, and the rate for all mares 3 years and older would be approximately 35/88 x - 40 percent. (vi) Nelson (l979) observed 2l foals born to 38 "mature" (4 years and older) mares, a rate of 55 percent. Apparently no foals were born to younger mares. "Immature" animals numbered l4, and if as many as 5 of these were 3-year-olds, then the rate for all mares 3 years and older would be on the order of 2l/43 x l00 « 49 percent. (vii) Hall (n.d.) presented sex and age composition on the Pryor Mountain herd in l97l, and reported that l8 foals were seen in the area. Mares made up 40 percent of the herd of 80. Of the 80, 9 were l0 years or older, 42 were 4 to 9. Since 3-year-olds made up 9 percent of the herd, they numbered 80 x .09 = 7. If we apply the 40 percent female percentage, then the number of mares in each of these age classes approximated 4, l7, and 3, and totaled 24. The foaling

38 rate for 3-year-olds and older therefore was on the order of l8/24 x l00 = 75 percent. There is no way to subdivide the rates any further. All of these statistics are summarized in Table 2.4, and several generalizations seem justified. First, fecundity rates in wild horses appear to increase with age, at least in the first half to two-thirds of life, as we have seen is the case in confined and range-reared domestics. In the studies summarized in Table 2.4, the percentage of 3-year-olds foaling has varied between years and areas from 0 to 33 percent, and has averaged l3 percent. Most of the studies have not reported the rates for 4-year-olds, and the three values cited in Table 2.4 may or may not be typical. Judging by the abundant representation of rates for 4-year-olds and older, and for 5-year-olds and older, the rates in this latter class commonly rise above 60 percent, in individual years exceeding 80. Because of these age-specific differences in fecundity, the rate one uses to express the performance of a given herd depends on the age classes included. The rate for the 5-year-olds and older animals can be expected to be higher than that for the 4-year-olds and older (the "adults" or "mature" animals of several authors), and in turn the rate for the 3-year-olds and older will be lower than that for the "matures." This is evident in Table 2.4. Furthermore, the rate for a herd will depend on its age composition: a herd with a large number of 3- and 4-year-olds is likely to have a lower rate than one with fewer of these ages and a greater number of older animals, other things being equal. Comparison between herds and years can only be precise when these variables are standardized. Within these constraints, herd fecundity rates for the areas and years covered by the studies in Table 2.4 vary between 38 and 78 percent for the 3-year-olds and older animals, and average 54. The rates for the 4-year-old and older segments of the population vary between 42 and 86, and average 6l. Why fecundity in wild horses should be lower than that of confined domestics, and should vary markedly between years in some herds (cf. Welsh l975, Boyd l980), has been the subject of considerable speculation in the literature. Several authors surmise that the main source of variation is not fertility, but the ability of the mare to carry a viable fetus to parturition. This ability, in the speculation of some, varies with weather conditions, food supplies, and perhaps the age of the mare. One gets the impression from reading the literature that reproduction is a heavy drain on the mare's physiological resources, and that added stress can result in abortion. Thus Boyd (l980), citing published abortion rates of l0 percent in domestic horses, suspected the difference in foaling rates between her 2 years of study to have been due to variations in abortions and stillbirths resulting from the mild winter preceding the first study year, and a severe winter prior to the second. Keiper's (l979) suggestion that the higher foaling rate on CNWR was due to the removal of foals and release of mares from lactation, and consequent lower abortion rate was mentioned earlier. To Tyler (l972), abortions seemed "common" among New Forest ponies during late autumn, winter,

39 CO T •~~. '-* *•% in co CD in rfl O i^ r- sr r-l >-- lt lt V^— '^J lllllll ltl CJ Lf> ID » 0) 10 oo m 1O b 9 00 O + a> r- CM co t- J. r~ o en oo '*' . 2 ^ O C""1* C*" OO irt rH t~i *~H C**' CM CO 4." CO • Xr. M J. CM com CMCsir- CDJ.1-tO^. to r-l tu fd - M ^ OO i/5 (O OO jf l/> CO C** r** ^ LO to V 0) • C p.y E •*H 0) £ r-l tU CM td o rH CD CO CO . " O ^4 .—•' ^.^ .— ' a) u. J i i ll OOtO lllllll lll in •H Irt CM CM • 4^ s t0 10 f^ 0) ^ c c 2 OO^ COOOO1 "rHCO r^tDOO O .H •H 4-i ao w a O OOCM OOa1m rH^.rHi— li— ( a. oi c 3 in ^ x-^ CM CN •— ' •— .^.. i i i i i •— ' ^-^ "— ,— ' "^ E f) "i o CO OOLO OOCO OOrHCM t**O T OCMOO jj 3 00 for^ C^LO LOP^J* ^"^ i/>tooo in in .o 0) .H S-* en c -i U • rfl rfl 15 0 rH rH C «*> •H r> gitM •o 0 rH OlrHtOrHlO CMU1) rfl CO •H c O rH CO O OT rH CM O b '0 m ro m * 0 CM r H CO ' rH rH rH rH 0) 0) 0) H 0 b 0) 0 *j t*> n > a UH .Q P T^ • r. r*. 0 E 3 2O •3 O) to C 0 rH rH •rl tM 'O e £4 r- r- U) T3 O C o rfl OT 01 t0 C rfl b •H 0) rH rH OO C7) 1H OrHCv* cOtDC^OOCDfd O COC^OO vQ 0) II rfl O ^ * OT O) O CJ) CD C71 "O CJ) C/l O) Ol CJ) £ 'iH OT CJ1 Ol TJ rfl OOrH H OHrHrH CrHrHrHrHrH XCrHrHrH 0) x; a) 3 C C C C CO rg bai.H P b § "" fl ffl "H rH tQ JC 03 4-1 •!-• 6 rfl >, 0) P C P CCO > f< >i3.H 0 0 >. O ffl i O) b SZI3 Z Z inzffl o c e e •o 0) v 0) * .3 +•1 x: Vj a ro H 3 3 rH •0.2 in 9 p- O • 9*0 o> rH C j M r H rfl 0) u o v-/ 0 > •H •3 JS 0) 00 •H 4J 60 C•| U X 3 HI 0) 10 &1 O c 3 0) *J tA 0) rH (d N ii ourc 3 « tn o1 ** C•) *-* *-* tj> r* r> CM CJ • o r^ oi 01 P"* 0) t0 .H «i • n 3 C •o rH O 0) C/) £ *O OO Ol rH rH Ol 0) rfl-H rH -r US C rH ~ ^ • 01 rH ^ ^ rH X > tJ O. rfl g M ^ « r. C 00 0) > rfl pj •P JS 0) Ob •H 0> .rl t0 H in rH TJ u> O. w 0) 0) 0) b •H rH >, rH .H rH rH 3 Jd 0) O • ill 0i O ,1> 0) 0) >i H T3 Z H U. X CO 3 M Z H O rH CM

40 and early spring and could have been the major factor contributing to the low foaling rate. She actually saw nine aborted foals. Welsh (l975) surmised that the differences in foaling rates between his 3 years of study (Table 2.4) were due to differences in abortion, these in turn resulting from poor nutrition and differences in severity of the three winters. The first was severe at the beginning but then eased, the second was mild throughout, and the third was long and severe throughout. Nelson (l979) observed differences in foaling between females that did and did not have access to revegetation areas. Females with access to these foaled a month earlier. They comprised 64 percent of the mature female population but contributed 73 percent of the foals. Indications of a tenuous balance between the mare and her resources and the foal and its viability are suggested from several observations by these authors. In Welsh's study, the postnatal mortality of foals born to 3- and 4-year-old mares was higher than that for foals born to older females. In the first and third years of his study, abortion rates were higher and foal mortality lower, suggesting that the weak foals were lost before birth. In the second year, the abortion rate was lower, fecundity higher, but postnatal foal loss was the highest of the 3 years. In the middle year, half of the mares that foaled died, but only l8 percent of those that did not foal died. While not endorsing the idea for her study, Boyd (l980) cites several authors (Klingel l969c, Tyler l972, Moehlman l974, Nelson l979) who have suggested that wild equid mares may commonly foal in alternate years. The phenomenon might reflect their inability to recover reserves sufficiently to bear a foal each year. She noted that older mares that foaled in May l978 did not foal again until late l979. One veterinarian has told us that if range-reared domestic mares are bred at 2 years of age and foal at 3, they are not likely to foal again at 4. It is not known whether these tendencies explain the fact that none of the herds summarized in Table 2.4 continued to foal during each year of observation at rates markedly above or below the means for the age classes. Techniques are available for assessing pregnancy status of freshly captured wild mares and hence to allow estimates of probable natality rates. Appropriate data could be collected, in part, in connection with the adopt-a-horse program and with the projected burro removal operations. A single trained person with the capture teams could accomplish requisite sample collection and examination of the females. This could also be made a part of a capture program for marking purposes. It would be an essential part of field-testing contraceptives, since implanted pregnant animals would continue a normal pregnancy, delivery, and lactation but not return to heat. The effects of the contraceptive would not be evident until the following year in such animals. (c) Foaling season. In captivity, some female domestic horses ovulate year-round with a cycle length of 2l to 22 days, but most tend to be seasonal depending on latitude. In the wild, this seasonality

41 becomes strongly pronounced (Table 2.5), with May the major foaling month and over three-fourths of the foals commonly born in the 3-month period April to June. Even in the exceptional case of Sable Island, where young were born in all months of the year but January (Table 2.5), 77 percent were born during these 3 months. In addition to the studies summarized in Table 2.5, Hall (n.d.) reported that the "majority" of mares had foaled in the Pryor Mountain herd (Montana) by mid-July. And Feist and McCullough (l975) found all of the foals born between l5 April and 30 June during their year of study on Pryor Mountain. (2) Age Composition. In large areas where emigration and immigration are not important variables, the age composition of an animal population—numbers and percentages of animals in each age class—is produced by the interaction of its reproductive and mortality rates. Hence, in a general way it is indicative of the demographic characteristics of a population. More precisely, the age composition can be used to measure mortality or survival rates, provided that certain conditions hold within the population. These aspects will be explored shortly in the section on mortality. At this point in our discussion, the subject of age composition is relevant to the question of herd fecundity. For as we have seen, fecundity appears to vary with the mare's age, and hence a herd's reproductive output and growth rate will depend on its age composition, or at least that of its mares. For these reasons, we devote a substantial amount of space to an analysis of available information on age composition. This section covers the topic with respect to horses. Burros will be treated in a later section. The BLM routinely determines the ages of horses brought in from herd roundups. Over the past few years, thousands of animals have been aged by the familiar tooth-aging method, and records are kept in BLM files on all of these animals. With BLM cooperation, we have obtained a large portion of this information and have summarized it in Table 2.6. These include animals gathered during 22 different periods and/or areas in four states. They total 4,825 females and 3,939 males, a sample equivalent to l5 percent of the 60,000 wild horses reported to occur in the western states. In a normal age distribution, the numbers of animals in each age class should exceed the numbers in the next older class. The predominance of the 2-year-old class in these data (Table 2.7) indicates either that the foals and yearlings are not sampled in proportion to their abundance, or that the aging techniques are unreliable. Since there should be no problem with aging the foals, we are inclined toward the former explanation. Wolfe (l980) used similar BLM roundup data to calculate survival rates from age structure. He, too, found the 0- and l-year-old classes less numerous than the 2-year-olds. A deficiency of foals and yearlings could occur in a localized population following 2 years of reproductive failure and/or low foal survival due to exceptional weather. But the fact that the data in

42 TABLE 2.5 Monthly Distribution of Foals Born in Five Studies of Wild Horse Herds Percent of Foals Born by Location and Source Wyoming (Boyd 1980)1 Md.-Va. (Keiper 1979)2 NewMexico (Nelson 1979) NovaScotia (Welsh 1975)3 Britain (Tyler 1972) Month January 0 0 0 0 0 February 0 0 0 1 . 0 March 0 0 0 1 2 April May June 9 47 33 17 47 33 38 20 35 22 T 96 19 14 1 July 7 11 10 10 .2 August 4 4 5 6 1 September 1 3 0 1 0 October 0 0 0 2 0 November 0 0 0 1 0 December 0 0 0 1 0 No. in Sample 107 142 21 143 294. "1979 only. 1978 not used because observation season delayed > '5-year average. 3-year average. Average for three studied breeding seasons. Winter births have been observed in New Forest ponies.

43 Tables 2.6 and 2.7 were taken over a 3-year period and a wide geographic range, the fact that 2-year-old predominance recurs in about two-thirds of the individual population samples, and Wolfe's similar result for other areas and years strongly indicates that it is a sampling artifact. Beyond this, there is a strong predominance of animals in the younger age classes (Table 2.7) and a gradual decline in numbers in each successively older class. The sequence is not perfectly smooth, as there are slight divergences from the expected geometric rate of decline. But these could be due in part to sampling error and/or imperfections in the aging techniques. Overall, a gradual shrinkage occurs to the point where the number of animals in each age class l3 years of age and older constitute less than l percent of the total. While the trends are more variable in the individual samples of Table 2.6, doubtless in part because the samples are smaller and more subject to sampling error, the general pattern shown in Table 2.7 tends to prevail in each of these smaller data sets. One can therefore infer that the pattern in Table 2.7 occurs widely over western horse herds. Age composition from the available wild horse studies in the literature is based both on smaller samples and a failure to divide the data by each age class, as in Tables 2.6 and 2.7. But despite these deficiencies, they are of comparative interest and are summarized in Table 2.8. Several generalizations may be drawn from this table, the first about the percentage of foals in the herds. Where the number of foals born is calculated as a percentage of a herd, it constitutes one measure of the reproductive output of that herd in that year. The values in Table 2.8 range from l3 to 23 percent, and average l7.l. Secondly, this percentage can be used to calculate the percentage by which a given herd in a given year increases through reproduction from its numbers just before the foaling period to that immediately afterwards. The calculation is made by: % Foals x l00 = % Increase l00 - % Foals The calculated increase rate is a real property of a herd only if there is no mortality of adults during the foaling season, and if all foals survive. If there is mortality, then the percentage is only the potential by which the herd could have increased, and it overestimates the actual increase. Furthermore, the percentages are potential annual rates of increase from herd size just before foaling to herd size at that point in the following year, again on the condition that no mortality occurs during the year. They therefore set upper limits on potential increase rates in the herds and years for which they are calculated. Bearing these conditions in mind, we can use the above equation to calculate potential increase rates for the herds in Table 2.8, recognizing that they overestimate the actual increase rates if there

44 % 0 -H — OOOOOO0sNOO o o o o ooo OOOOOOo^tmoo 0 O O o o O * N (I. 0 0 oooooooooo o o O O OOO ooooooooooo 0 00 o o ooo O] S! o in oooooooooo o o oo OOO O^OOOOOOOOO O O 0 o o ooo -H f* '-• 0 — oooooooooo o o O O 0oo ooooooooooo O 00 o o ooo * o — oooooooooo o o o o OOO 0.00000-oo0 O oo o o Ooo m O 00 otrtooooovooo o e O O O V O O^.O-^^O O o O o o O V O V °s oooooooooo 0 l-l O O ooo 00*Ooo0oo0oo o o o o o 0O^ S 1~i ,— 0 0i OO oooooooo o - O O Ooo ..0-0oo00,0 o o o ~* o 0 n o W f* tO "«"°°°""^'° rj • 0 M 0 0* O ri x 1-; a; -> --• e — o en o O o o ol n ooo •H —H V O " O fN o at E I S H ° V 2 PM n in rH o 2 5 « * ° - -1 "o R a 3 O» o M , 4.1 v y C h 4J 0 M vfv ^. r.1 „ 1 w S1 % 1 5 r* •H 0 fM fM fN O — , <rt O O 01 O fM O fN fN 00 H M C « O ^ V f*i O 0 tN r* n I • • in O rj NO* »««OOo ^ " o V ^ fN a n 2 ° * o in mo or* CO O N oovOOODt0*o01r^a)'N O 01 H « 0 ,.., g n O OD mo ^ fM O fM S o rM ^ in m o m m o o v i n (N r- ^ *.! o o 1 I r- in C0 00 'J M m t£ o H n fN r* CD f O — i P* 1-1 CM •-j ^ -r r. O fM ft rx ^4 rM "* -i o m p4 *2 n o x m tO o 5 M V r- ^ tM v n N r* rur.r-fN oo^mo n $ fff in H 01 *4 9t 9t 0* PIE 01 r- r- f^ODCDr-CDCDCDCT'tT'tTi t^ r^. r* P. r* tf1 ff. P- S 0t r- ff.ff.t^t^fftff'm1r't^^ m ^ il i r* r«. '- V. !T. at ff. a« tft 0* 0* CD CO & » >l ss . * • , 3 s s • • W J* o SV C 0 £ 3 > i I C o « 4J *.',;•' • *J W -t « at » x .H O OX -. ': IH *^ i 3 r 2 o Z oi 3 o » « • < C n m m " x .H U ** ** »• a «« v d c o u *J C 0 .H .H O W *j *j *j e x TJ 'ii &j t/)uit/>'ve ** «E J 1 So • E e o ai ly -. H o slit is Jj — ( o *j a o o ** 4J W E J< « • . • « *J 3 O ^ 0 .H iM C 3 o 32 5 § § ai u u I a 1 II • • • « ** g w w w a. tn m Q ^ 4 S i/)t/}t/)o.CO m r o . , i 0. 0. 0 O A A 1 X o a. a. m 3 « 9 B Al f) < cr, 2 c « • g c • • •H O h o J' o H O S t Ql .- l tft > at i, "1 il C/ ) > .H N - m C I iiiil^ ^ l Si if • 1*t J 3 oSzJariaa'fl'fl'ifl o O K VI A000ffiiJ>> fal Oi « H a. w 5 3t U OOOOOO 005*1 x z£ 93 5 5 O O 0 n n Q 0 O X • £ Z & S 5 3 Ic z SB z •

45 to 9 OOOOOOOO vD (N n V rH •3 a O'H OOOOO CO O O u s r- CM co vo r- 00 CM i-l VD O vD n <N iH CM 01 in u o X ,a 9 IM O to 0) fr, CO r- • N 1 m « rH 2 • • • • • • CO •H I 5 0)

46 r- 0) o o cooo oo in CD oocncM j. id a. CM COCOlOCOrHCOCOCOr> to *' e ft " Q id CM 00 f^ " CO rH H 00 m H- 00 lO r~ r> rH rH O in • c o1 3 in CD co ^t in in co j. rH f-l > in 0 0 ,Q .O id -^ T CO co T.|- 1t 00 T T" GO CO rH . CM | | rHrH rH T3 o) 0) CO f-l rH R n 00 I rH O rH | CO CM rH 0 0) td 4J £ E 0) U • CO id CO rH O to AGO Ol 1 to O C~ . CD rH " ' JL J. rH rH ' rH O CO r.H U-. 4-1 CO 0) CM si 4J 14-i i-l P O 03 td M 4-1 OP f< rH CO e. o id 00 ID t4-i dj td O £> O X E .rl 0) 3 O to rH rH CM CD J. rH Cn O . O H *• 1-HrHrHrHrH'rH u JM rH M C 4-i rH id • c c P in HI rH .H • H CM C in f • Ol 4-10) i0 •P Uj OO O 'O rH 4-1 C --*. co 01 CD in oo :3. • b C 0) X) C 0) • rH rH rH r H rH rH 05 o) id P id 3 o CO rH (X id H O b • a. E o 0) Cn 10 m 'H co a. to w ^ id rH C Ol 0i 01 .H • •a O rH rH b m rH b vH X) O >, 0i cooo & c*. oo inr-ro. oi P id P t-1 id rH rH rH r-l rH rH u, rH rH 3 CO 0) .H id rH T3 C rH id ft - > 3 4J o x: rH co c O O. CO fl C C co J. j ^ in ID P i-i x: o id • o rH oo en -CM en c^ co r> oo • • "4-i rH 3 CO 4-* * O td tTj "rl Cn rH Ol .rl •a c b en en en en *ri i i O en en en en • • 0 ) 0 id rH rH rH rH 4-1 0 "in O rH rH rH rH 3 O t4n e 3 w 'H X •rl 0) * *. * O f"^ *o r^ *H * 1i • O r O 2 P m id so o en c 01 x c c c 00 Si c-. JJ D. CM id C C C M rH id rH 0) .H t'ti id O >M , 6 O H 0 •o idid.H H X id 0i0i 4-1 « JD o o c PP6.id>1 +J P4-1 c b 'O •*> Q W 9 — i tQ CCO > b S .H coco o) oi o1 x: oi TJ O O >i O iQ 0) b 0)0) O E .P O 0) 00 2: 2: 3 Z 3: Z 03 33 £4 B Q 4 M *0 ^ oi 3 i-H o) ta <.1 * a to 3 c • O C 0) 0) f (M in CO 0) rH b rH OH O id .p id o id b 8. d en c t-H a) a fi id O X) E 0 ) * o rH *• b co co CM. •H P O C Ol rH CH ','I x; XI 'rl M 01 0) O r! Q 00 ro o x: v IL 3 M , tA b CM ^1 C ff u O rH rH 0) " td ^ P J rH ro id > n- o oi "" ** M rH o o (d o .P • E ' t- 1 3 0 3 01 01 "M CM to • c O m r~ r- CM o O r- en o> r- oo UH IM ro o) o) bO 0) b b -O 0) T3 N CB 'ZTJW OrHrHCn (J> us C Cn ^ •« -j f* rH rH rH o o o) xi TD id "P ^ >. e 'a 4-1 x; 3 b b i 3 id C 00 iS CM P"" jCOlOb W b C 0) 0) 0) C 0i .-i H 3 CO rH T3 CO CX to O) t4-t CO X) X) Ol Ol o 0) g •rlrH>« rH.iHrHrH rHZ C 3 ax:.coio1 c § 9 o) id o oi o1 o) >* o 3 id £ 0) (w u: iu 3XZc-t 3a C rH CM CO J. UJ lO

47 is any mortality. For the extremes (l3 and 23), the potential increase rates are l5 and 30 percent. The average is 2l. A second point to be gleaned from Table 2.8 is that the percentages of late summer or fall herds which the foals constitute are lower, on the average (l5.8 with a range of l3 to l9), than the percentages which the newly born foals comprise of the herds. This is doubtless due to mortality between birth and fall. Several authors (Tyler l972, Welsh l975, Keiper l979) stress that most foal mortality occurs in the first month, indeed sometimes the first few days, of life. Thus, the fall foal percentage is only a rough indicator of herd natality. In some areas when foal mortality is especially high, the percentage may be more a reflection of foal mortality than of foal birth. In addition, any loss of older horses between the foaling season and fall is an additional variable affecting this statistic, tending to inflate it. A third inference to draw from Table 2.8 is that the nonfoaling females of the herds (foals, yearlings, and 2-year-olds) average somewhere near 40 percent of the females, the exact figure depending on whether one uses the at-birth foal percentage or the late-summer/ fall percentage. If the low-foaling 3-year-old females (Table 2.4) are added to this group, the new total makes up roughly half of the females and one-fourth of the herds. On average, therefore, the fully breeding 4-year-olds and older females (the "matures" of several authors) comprise only about one-fourth of most wild-horse herds. (3) Survival. It is useful to consider two categories of survival data: those covering the first year and those covering all subsequent years. (a) First-year survival. Appeal, once again, to data on domestic horses may suggest the maximum survival rates that could be expected of wild herds, with the actual values possibly lower. In the confined domestic herds cited above, where about 70 percent of bred mares bear young, foal mortality reduces the ratio to about 55 foals per l00 mares by the end of the foal's first year. This is a loss of about 2l percent, and a first-year survival rate of 79 percent. Speelman et al. (l944) report foal survival from birth to weaning in range-fed domestic herds at 87 percent. Information on survival rates in wild herds comes from two sources, the first of which is direct observation of known individuals and/or herd segments in the intensively studied populations cited above: (i) While not reporting total foal losses, Tyler (l972) concluded that most loss occurred in the first day of life. (ii) Boyd (l979) reported the first-year survival rate of foals in l978 to be 82.4 percent. Field work on the l979 foal crop was not completed when she published, but survival to 2 months of age was 97.7 percent in that year.

48 (iii) Keiper (l979) also did not report total foal loss, but reported that 6 of 7 known foal deaths occurred in the first 3 weeks of life. (iv) Welsh (l975) provides some of the most complete observations on foal loss, as follows: Foals l970 l97l l972 Means No. born 48 70 25 % dying, 0-4 months l3 33 24 25 % dying, 4-l6 months 10 5l ll 24 % dying, 0-16 months 21 67 32 40 Except for what may be unusual 4 to l6-month losses in l97l, the loss rate in the early part of life is heavier than that after 4 months, although the latter is not insubstantial. (v) Nelson (l979) observed no foal mortality between birth and September l among the 2l foals he observed in the Jicarilla herd in l979. (vi) Feist and McCullough (l975) recorded four foal deaths between birth and November l among the 35 foals he observed in the Pryor Mountain herd, a loss of ll percent. (vii) Hall (n.d.) had observed l8 foals on the Pryors in summer l97l. But at a November roundup of the entire herd, foals numbered l5, a number suggesting a loss of 3 (l7 percent) during the period from July to November. A second means of calculating survival rates is with age-composition measurements. In wild populations, the difference between number of foals at birth and the number of yearlings at the same date can be used to estimate first-year survival rate provided the population is a long unchanging one that has assumed a stationary age distribution from constant natality and mortality rates over a preceding period of several years. Few populations meet these criteria, and the ones from which we have cited age composition data do not. Thus Nelson's (l979) Jicarilla herd contained 2l foals and l6 yearlings, suggesting a 76 percent first-year survival rate (l6/2l x l00 =76). But his herd was increasing, a fact that would make this estimate conservative, and the numbers are small. The Pryor Mountain herd (Feist and McCullough l975) contained 35 foals and 30 yearlings, implying an 86 percent first-year survival rate. The above reservations also apply here. Keiper (l979) lists a 5-year total of 52 foals and 49 yearlings in the AINS population, implying a 94 percent first-year survival rate. Yet he observed 7 foal deaths, which would be l3 percent of the composite 5-year foal population, and he did not assume that he had seen all of the animals that died. (Note: annual survival rate s = l - a, the annual mortality rate. Both s and a are expressed as decimal fractions. Both can be multiplied by l00 to convert into percentages.)

49 Because of the apparent bias against foals and yearlings in the BLM roundup data (Tables 2.6 and 2.7), they cannot be used for calculating first-year survival. However, Wolfe (l980) used similar BLM roundup data to calculate survival rates from age structure. He, too, found the 0- and l-year-old classes less numerous than the 2-year-olds, and for this reason we are reluctant to accept his low first-year survival estimates (see his Table 4 with a range of 28 to 70 percent, and a mean of 46). But in his Table 2, Wolfe listed foal percentages based on herd-structure observations made in the field without rounding up the animals. The observations were made on l6 different populations in 6 states, some counted over a 3-year period and in many cases more than once a year. It would seem reasonable to assume that changes in those percentages between late summer after foaling, and the following late winter and spring, result at least in part from foal mortality. In l5 August counts, the mean foal percentage was l8.3 percent. In l7 January to March counts, the percentage declined to l3.3, and a reduction of 27 percent below the August mean. To what extent this could legitimately be taken as a 73 percent August-to-March survival rate is not known. It is difficult to gain any balanced and precise overview of the magnitude of first-year survival from these disparate results. They are derived in many cases from small samples, span periods of time that vary from 2 months (Boyd in l979) to l6 months (Welsh l975), and in some cases involve assumptions that cannot be met or are of uncertain validity. Given these uncertainties, the 79 to 87 percent first-year survival rates in confined and range-reared domestics, Boyd's 82.4 percent first-year rate, Welsh's 79 to 33 percent l6-month rates, and the 73 percent rate calculated above from Wolfe's data suggest that 20 to 25 percent first-year mortality and 75 to 80 percent survival rates might not be uncommon. It seems possible that this is a variable parameter, sensitive to a variety of variables including weather and range conditions, and responding to climatic and vegetative differences between areas. Tyler (l972), emphasizing the importance of loss in the first day of life, observed congenital weaknesses and rejection by the mare as the main natural causes of early loss. Shooting and accidents were the more important human-related causes. She observed 6 foals older than one day that she believed were killed by stallions. Boyd (l980) reported that 30 percent of known foal deaths occurred within the first week of life, 50 percent within the first month. Causes of l9 known foal deaths in l978 were inexperienced mare (37 percent), congenital weakness (l6 percent), miring in mud and accidents (26 percent), and winter debilitation (2l percent). Welsh (l975) reported lower survival of young in small and/or unstable herds, suggesting the role of social factors. Boyd (l980) observed that some desertion occurs in summer when herds concentrate around water holes. Foals may get separated from their dams during the confusion and milling about, and the mares may leave the young behind when they disperse. She suggests that this tendency could increase as population size increases, thereby serving as one density dependent source of attrition.

50 The relationship described above between the mare's physical condition, her ability to carry a foal to birth, and herd natality may also be correlated with survival of foals brought to parturition. Thus, Welsh (l975) points out that Sable Island horses experienced different birth rates in the 3 years he observed them, attributing the differences to different abortion rates. In these same years, there was a rough inverse correlation between early foal survival and natality: l970 l97l l972 % 3 yr. + mares foaling 58 7l 42 % foals surviving, 0-4 months 87 67 76 In Welsh's surmise, the weak foals were aborted in years with low natality, ultimately resulting in higher postnatal survival rates of the foals ultimately born. A somewhat similar pattern may be seen in Boyd's data: 1978 l979 % 3 yr. + mares foaling 78 53 % foals surviving, 0-2 months 82 98 Hence there appears to be some evidence that natality rates and foal survival rates may be inverse to each other, at least in some cases. Welsh further reported that the survival of foals born to 3- and 4-year-old mares is lower than that of foals from older mares, In this case, age-specific fecundity and subsequent foal survival are positively correlated, once again reflecting the dependency of the parameters on the mares' resources. Establishing the magnitude of foal mortality more precisely, and its range of variation under different environmental pressures, would require lengthy and costly research. The most promising approach would probably be to capture large numbers of mares prior to foaling and determine which are pregnant. These would need to be marked or individually recognized by their color and structural characteristics, and then observed in the year after the foals are born in order to document birth and foal survival rates. It might be advantageous to telemeter the mares so that they could be readily located. A sample of 200 foals would not be large enough to provide very high statistical precision, since the confidence intervals around an estimated 80 percent survival rate at 0.025 probability level would be 74 and 85 (Mainland l948). Yet, if only half of the mares foaled each year, 400 would need to be captured and tested in order to provide a sample of 200 foals.

51 An operation of this magnitude would provide an estimate of first-year foal survival for one population for l year. Gaining some view of the extent of year-to-year variation would require similar efforts for each of several years, and determining the extent of variation between areas would require such effort at numerous sites around the West. (b) Survival of yearlings and older animals. Survival in older animals can be estimated with the two approaches used above for first-year survival estimates: direct observation of known individual or herd segments and analysis of age structure. Information from the first source is fragmentary, but is of some value: (i) During her 3 years of observation, Tyler (l972) detected the death of 40 "older" ponies while she captured and saved 3 that were in dire straits. This is 43 out of a 3-year composite population of 979, a mortality rate of 4 percent (43/979 x l00), and an annual survival rate of 96 percent. (ii) In Keiper's (l979) study, he observed ll ponies l year and older that died over a 4-year observation period. This constituted a known mortality of 3 percent in the composite population of 3l5 for that time period, and a maximum survival estimate of 97 percent. (iii) Welsh (l975) constructed time-specific life tables to calculate mortality rates in the Sable Island herd. We have calculated mean annual mortality rates from these by dividing the number of 0- to 4-month animals in his K lx columns by the total numbers in those columns. The results suggest l6.2 percent mean annual mortality for mares (from his Table 54), ll.2 percent for males (from his Table 55), and l3.4 percent for the entire population. The corresponding survival rates are 83.8, 88.8, and 86.6 percent. (iv) Feist and McCullough (l975) observed the death of one mare between May l and November l, l970, out of the Pryor Mountain herd of 270 animals. (v) Nelson (l979) observed the death of 3 older animals in the Jicarilla herd of ll6, while 3 more disappeared during his spring and summer observations. This could imply 5 percent loss through less than half of the year, and a maximum survival rate of 95 percent during this period. In a stationary (constant) population with stable age distribution, the percentage of l-year-olds in that portion of the population l year and older is equivalent to the mean annual mortality rate for that portion of the population. While neither of these conditions may be precisely true of the Pryor Mountain (Feist and McCullough l975) and Jicarilla herds (Nelson l979), the yearling percentages may give rough approximations of the mean annual mortality rates (Table 2.9). These are small samples, and if they were taken from increasing populations (and the Jicarilla was increasing) they

52 overestimate the mortality rates. Hence the mean, annual adult survival rates for those animals l year of age and older appears to be around 80 to 90 percent or higher. Table 2.9 Percentage of Yearlings Among Yearlings and Older Horses, Pryor Mountain and Jicarilla Herds. No. No. Yrlg. 0.25 Area Sex l Yr.+ Yrlgs. * Conf . Intl . Pryor Mountain^ Females l25 19 15 9-22 Males ll0 ll l0 5-l7 Jicarilla2 Females 61 9 l5 7-27 Males 34 7 21 9-38 lFrom Feist and McCullough (l975). 2From Nelson (l979). Similar calculations cannot be made with the data in Table 2.7 because of the inadequate representation of yearlings described above. But adult survival rates can be crudely calculated for those age classes from 2 through l2 years by transforming the numbers of animals in each class to logarithmic form, and regressing these values on age. The slope of the line estimates the mean, annual instantaneous mortality rate. Converting back to anti-logarithmic form provides the mean, annual finite survival rate for the 2 to l2 age classes. The regression for the mares produces an r of 0.9l9 and a calculated, annual survival rate of 78 percent. The same values for the stallions are r = 0.926 and an 80 percent annual survival rate. These .values will be underestimates if the populations are increasing, and are not statistically different. Wolfe (l980) used the Chapman-Robson method (Robson and Chapman l96l) to calculate similar rates in 2- to 9-year-old animals from 6 herds in 4 states. His results average 75 percent for 3 mare samples, 67 percent for 2 groups of stallions, and 74 percent for 3 herds of mixed sex. These results collectively suggest adult survival rates of 67 to 80 percent per year. If the herds are increasing, these are underestimates. The degree of underestimation to be expected for survival rates calculated from age data when the population is increasing can be calculated as: S* = S e -r

53 where S* = apparent survival, S » true annual survival, and r = annual rate of increase. For example, if S = 0.98 and r = 0.20, then the apparent survival rate is 0.80. If S = 0.90 and r = 0.l5, then the apparent rate is 0.77. In sum, the evidence on mortality of yearlings and older animals is more equivocal than that on foals, and more elusive to generalize. At the least, one can say unequivocally that some loss has been detected in every herd that has been studied. With the exception of Feist and McCullough's single loss in a 7-month period, values of 3 to 5 percent known losses over periods of several months to a year recur in several studies and constitute the lowest in a range of available values. Estimates of ll to 25 percent annual mortality and higher have been derived from several analyses, but depend on various assumptions of uncertain validity. Mortality of these older animals, like the first-year rates, are doubtless subject to some between-year and between-area variation associated with weather, climate, and forage differences. However, we suspect that they are less sensitive to these environmental variables than the first-year rates, and are themselves less variable. Elucidating their magnitudes and ranges of variation more precisely would again require lengthy and costly research, probably most effectively carried out with radio telemetry. Once again, large sample sizes would be needed to provide estimates with narrow confidence intervals, and repeated annual studies in a number of areas would be needed to disclose the patterns of temporal and spatial variation. (4) Sex Composition. The sex ratio of wild horses shows a slight preponderance of males in the first year of life (Table 2.l0). As the animals age, the percentage of males gradually declines, reaching a low point somewhere around 4 to 6 years of age (see BLM data in Table 2.l0). This shift suggests higher mortality and/or dispersal of males during this series of years. After the 4 to 6 ages, the percentage of males appears once again to increase progressively through life (BLM data in Table 2.l0). Welsh (l975) emphasized the reproductive hazards to which the mares are subject, suggesting that the costs of pregnancy and lactation may increase female mortality rates. It may be no coincidence that the shift toward males in the sex ratios begins after the 4 to 6 age classes. It is at age 5 that half or more of the mares begin to bear colts (Table 2.4). Welsh (l975) observed that almost half the Sable Island mares that foaled in l97l died, while l8 percent of those that did not foal died. Final cause of death was attributed to heat loss and starvation. Autopsies on 32 dead adults and l5 foals revealed that all but one had little body fat, and bone marrow fat levels ranged between 2.4 and 4l.6 percent. Many had contracted spleens that lacked lymphoid follicles and had large amounts of nemosiderin. However, Welsh emphasized that food quality was similar between years and was rarely if ever limited in quantity. He attributes the susceptibility to starvation to the climatic differences between the mild winter of

54 0) bjl ^ H. m ^ LA rH ' rH WJ I 10 cn J. ^ 1 •rl CO 6 CO x-s CO '"N rH x^ •H O •H J1 +i r*1* +i in • •— J rH cn oo r* LO cn in 1 • in' rH • \^ 0) C"» CO rH tT) s^ O 0 J. >^ a" CO fl c ~ I i 0) "O •H cst •H CO i £ co 1 m 1 00 * 01 c a- cn rt 01 o CO v-* r« o §, co ^> in s-\ CO /-S LO •H a. rH 00 rH a. •o CM CO in 00 +l CM rH o I • r- oo ^x t~1 * "5 # CM CO CM • • * o a. \-> CO H C * 3" C •H § f ^ to x-^ LO ^ • Q CO •H LO rH O CM co in 4J CO bfl rH a. +i co +i rH r^ •H 0) H • oo p» vx 00 ^x cn 10 rH h 0 •—^ • • rH rfl ^H LO co CO \^ 8 ac CO ^ •P DO C a. x^ tO ^^ rH /•^ 3 0) 0) CO •H P« rH LO CM rH O CO o rH a- 0 +1 CO +I CM rH u £^ iJ t CM co vx rH ^^ rH Jjj 0) O CM rH • 3 KM m *^ a- E— .^ m lA O cn u as r- 1 cn TJ rH cn C >^ 0) ^\ ^N t<: t0 CO CO c i^ '-. CM rH rH «N 4J O m * C tfl CO CO 10 .2 'H rH 2 01 0) o S .M 6 «0 0) ^ 1 * rl rH •rl CM o h 5 C rH < 'H tc HJ ^ H £3 3 0 a 0 o 6 e * o o g CO 5 0 >i o a O & fc 2 .H ELt UH U-. H n ^ Ou w 1-3 *** l-l CM CO

55 l97l and the severe winter of l972, and the number and intensity of spring storms. Whatever the mortality causes contributing to the sex-ratio changes in Table 2.l0, the sex ratio for the entire sample in Tables 2.6, 2.7, and 2.l0 is 55.0 percent female. A preponderance of females is common in a wide range of mammalian species and indicates higher male mortality rates. The nearly balanced sex ratio in the older age classes in Table 2.l0 suggests that unweighted mean, age-specific mortality rates for the two sexes and all age classes would be similar. This was already suggested by the similar slopes computed from regression analyses described above. But because such a large fraction (76 percent in Table 2.l0) of the population is concentrated in the foal to 6-year-old age classes when the male-specific mortality appears higher and the sex ratio increasingly favors females, the weighted mean, annual mortality rate for the population would appear to be higher in the males than in the females. But Welsh's (l975) life-table analyses cited above in section (3)(b)(iii) contradict this conclusion. (5) Population Trend. As mentioned above, the question of the rates at which horse herds increase has been one of the more disputed ones in the controversy surrounding the wild horse issue. Annual increase rates of 20 percent or more, apparently originating from the management agencies, have been cited fairly widely, but their origin and basis have been difficult to track down. Some are based on census results. Ryden (l978:294) describes one BLM estimate based on censuses of 20 percent increase between l974 and l975. Cook (l975) and Heady and Bartolome (l977) cite annual increase rates of 20 to 30 percent. Wolfe (l980) calculated finite, annual increase rates (x) from BLM census data on l2 populations in 6 states which ranged from l.08 (8 percent) to 1.30 (30 percent), and averaged l.22 (22 percent) per year, but he was critical of the results. In order to investigate further what preliminary inferences could be drawn from the agency censuses, we obtained data from l0 BLM districts and one national forest. These data include censuses of 67 horse herds, 3 burro herds, and 2 areas which contained both species (Table 2.ll). The districts involved are Susanville and Riverside, California; Burns, Vale, and Lakeview, Oregon; Salmon, Idaho; Billings, Montana; Rock Springs, Wyoming; Salt Lake City, Utah; and Ely, Nevada. The national forest is the Humboldt in Nevada. These were selected on the recommendation of BLM officials as being representative and among the ones with the more complete data series. The number of years through which each herd was censused varied between l and ll, and between l and l8 counts were taken of each herd. Most of the herds had been counted over 4 to 7 years. Analyses disclosed several biases and uncontrolled variables that will be discussed in some detail in a later section on census. However, the two main problems need to be mentioned here: (a) Inconsistent counting seasons. The first is a problem already detected by Ryden (l978:294). In a seasonally breeding animal

56 '>! i |, * « el I i i « 1 i.. i j Ji |S i iN; O r ; J i 5 s !;h iii 5 i ii 11 1 i •H fn 1 1 ! i j 0) i ; i k W ; .M I till CO 2 S S i i! iif! ii n.W i " ! iii i Uii! i 4 l l Iii l 41 ifitiijiaijimii 8 i f •0 0) - i c •rl J3 -- - --- — * * • # •• « ! | 9 :, ti •*.i II 0) as i i s 0 . .._C-.o- l! 3S= 3 i ! i Ii JJM ft . fj... ; !. 1 i . «t e i: s . ;.| iim 0) -"" "" — * • 1 1 : ill j 23SI s 3 (N ?• Ee r- 2 E C ' I J2» = «=*si sa i * j X - 1 S C Si:S!I2S3S O 1 i 2 1 f A S * j 2 1 I i 1 — e o e - . (241 praj.) I I ." :K~2Sg:|K3 ..„„ . 177 . m ;es i i sif! L,S.SB,, fl) 3 Ill iiiiiillli iiiiii C g jS*5 iii* iii i i i i J|jgjsjsyjjs5s5j § i I 1 s i i s ? 1 3 2 i j i i i t i 4 « i f 1 1 . o « 1 CM A 3 I i ! 1 1 * I 1 1 i § i i j ! j Jl i| 1 i Ii I Hi i Ii || i i ]'! 1 i 1 : 1 ii I i ii It |1 « 11

57 S 8 n iijlj tl j! il ijllii I1!:'. 11 ii II! ill I 1 I i V 1" Will a j| • 1] IiI i 1 i. iiii «!• «i. i i il Iill""!,! ii' 1 • |i fli i i Ln !l ii i i : !" ! II !••••fc• I 1 t i !e *l * 1 1 1 i i*ii?i i •S.iiriSiiiS jijuijji II fillllli jiiisiii taffif j • M ill I| M Hi * * M # * j l.lll I 1 ! illili i ! i •B 1 jj.j . 1 M •* 1 ! 1 1 !•!•' I : . f : . rS j..l 1 i Is i I iiiii ' ii iii Mm, HI ! I IjJ3 li'i g gagcg s s 35? Is i I i iitl a* 3 1 a at* * »s«ssi«5iia« * SIS" IIHEMK ""£ ?««»« ] •-|- . '*- = ; I i . « « 58B *s-sc*- i ii *** S*SS----S - 2 I S3 S8C£«« s *Rr»ss*S3a* "-BSSISJg ~."*ait ...,,, -g.,...e. f . «--«-*- I 1 1 iiliigli jrjj:=;jsj; ££:Sc»£c iiiiiii •iliii iiilllli! £«»£££ jcjj5£j 1 a 9 I a a I a ,l u! 1 l l i i i i i il Us i ii« | 1 1 i 1 i i ! 1 |i i i ;, V P i J| 1 ^1 { Ifii 1 ! h j. | MS j: i 1 H& * i! 1 h

58 § CN ! il ii ii i l i 11 - i ? n ih siljiii I'l lihiii S5SSSSE h "Irrrr ill iiliiJi I'l'

59 •P o u a I r i ; S i \ \ i i. i i ~ i * s i i ] \ i 1 .1, 1 i i t j5,jjij i i i i \ imm siiiin 2 II! I liiii illll inn inn iiiii iiiii i l!#l 'i I tUIil m I j Pi. 1 Iiiii 2 i a I ! f I,,, HJrrrr t t t 6 L... , . ! rr 11- i i j i I ECJ i I i i i - a i * -• 1 = 1 1 I .---",.'" «-sss .,.•::.• JSSs'sS "*s 3 ,:;;;,; ssssSss iiiii iiliii iiliiS :.iiu. iiliiiii ssssss: iiii C 1 i 3 : ^ s r« 2 3 i l 1 i : 1( | j^ r 5 ii 3 3 3" 3 I 1 i if i s 4 1 1 - i - i 1 1 i i i j 1 f i 1 i :: 1 \ 1 i . \ 1 h h I • \ i h I . i t : !' H

60 . ; i \ f , i 1 \ ~ ' [ j I ; 1 j ' ?* * »l c J 3 i" 1 I ,. ' «„ "l | 1 i i i i'iiii liiii 2JJiJ !••• l II ill lii iii iii lJ.1l iilf iiiiiiltltlj ! j -7 ! | c 8 1 1 ^ 1 I i.... I \ i. . I'' !• • s J i ' 1 i i i j" * § j i 1 j" 111! 111! £ i i ] t i i i 1 jl s in SS8 III 'iJiS.: si: SSS 1 ss SSS SSS i S11!2SI SSiCESE! -li'lii 8=1 • 5S • Or. JSS zat K3SS sasssa S$K± 3 I 2 a~* 8*a**s a*ns «. -SJ .„ I .:t,,-:•._ 1 1 DX 1 Kit * • s 285 sss J:£SEgS; C at j"acx 8* *•*" It — 0 — 3X2 533 its i ,; .t t ~ ^ S - - r s xs±xAisi s f 9 SSSsSsS Ililllll j:£:£:£ »» » » t §s ssS III s5s f 5 j ? 1 1 s » » t » Isjt;£tii.t;t S 9 .5 9 9 9 | 1 I! ii • 2 ii" s 9 I \ *l 3 5 f f i i i Jl 1 1 1 \l ! i i ii ill Hi ii: Hi i i i 1 i i 1 x i i I Hi i I J ] 1 3 i | I | il ii ii m f i i 1 i • s

61 V 8 CN 1 f I 1 ; i. ! ii i . 1 ii * , 1 i i I il, i I i ul! I I i !' i i 1 \ ii ail ijihijj iiJii ititiii f » fllf fiif * + ; - • : - :: + • n iiii. 1 • i If:* it "t s s i •"*" I 8 <s«t « lillcl 1 5 i ;: I 2 3 s; : '- ~- ; Is! -ii!3! i* i IKK - | * |... . !.. ?! 1 • nil III 1 m lil il iiilil ii 3 "2 3 ui -- j i i J 1 , IV. ::sl B ».==*=£ i .5 i ! | ! * j i i -•s.i-sis e-iss » -M«M* i i Kfill is1 ali»S£ sass 33*5 ,„ *2"s HI .saasMSS=tlitst,8l *«*«« 5 1 ssi - * i I S55f£ i 555 ; ? 1 1? 5 555 j 5jjj »»» Kgeee ill »»*;j;:jjjjjij j iiji £££5*55 \ a 1 111 a i a i 11 a , i | i rm » 1 i 1 i • | , i s i |l { ! 1 * •? S ! i 1 i » i i • 1 1 ^ i - | 1 i i i i i h '. i: I « '* . c Ii i i it 1 £ J!J |J i & ! i f? i .* I! 1 1!!' J i H i 1 S i "jit? i I 1 c

62 \- i l\i f - i 4 • • - . s " ' 1 . - ^ ~' i - 4 * S=S£ nil ••. 5. I ' \ i ,? '' . i 11 i 3 a 1 nil i i i i * I i i i i 1 1 1 i i 1 I 1 1 s :f c jiissi iiilSSS •i •i ii ill it Jl {1 ii it ii ii e p i ^ t j nKl | ; . * , ; » - ~ - - - - '- - ; * ti S tl !s '3 A S £ i*ir] •"I**'" - 3 S 1 ii!i!l O 2 ! -•; o * it'sll j t rH • itilis i I* i! 11 ol 1 " i ffl i * s • ,•.38 J 3* *3 i s 3t I 3* ti a 35 s 3* ^ ! J! i in 3 - *' ~3~ }1 111 I ii ii ill ii I li i! Hi ii E s 1 5 3 r. 1 ! • | ? ? i 1 f i i i ' i l % i : J i ^ t ...' * * • o f .. - - * :IiS55R ss?,3il5 85s 5*3 s: = 58S 323 silsi UH E|J!i ar "' 3 S 2^ SS ; ss£ssss ; ; ; J j 5 J SSB ill ~E£ ii; SEE SJSJS ill KJssj SE , " •" • * § i ! 5* 5 '' F S 5 a i l s 1 5T * i .' -' i v 1" i' .? I it '.. s i : ifv ?. 1 i i 1 1 1 iV: I j ' i' '• i 1 ? J i j i 1 j }TS 1 i | i * i i ' 1 1 f t j i'. 1 3 ' I s . J •; II ' j i ii i 1 i j 1 Hi f li in ill • $ . ji * ,. f i1 i "ji (Hi' ii li h 1 i 2 3 ' •' 2 1 :- 'i H

63 \ 1 i . li i \ i • n \ i i ! i i I 1 » i • ii J J l! Is i if s i ffli s i - i 8 L : 1 • 1 i li M i i 1 1 i p si i * M §? K -~ sea 1 n •KM 1 8 s ii iiiii i i 8 m l i i t 1 j i a 5 l £ Jii j h ], 1,— fcf.. **(. • i !¥•»•<•*'*«." : i i i . • i rfhllB fM* MMrln

64 population, numbers increase during the breeding season and decline during the remainder of the year due to mortality. Comparisons between years must be made at the same stage of the life cycle in order to avoid bias from this seasonal cycle in numbers. The counting season has not been standardized in much of the agency census work, counts being made early in some years prior to foaling, in other years being made after foaling. (b) Variation in aircraft used. Some herds were censused in the early years with fixed-wing aircraft, and in later years with helicopters. In some of the censuses in Table 2.ll, the method was not specified. Analysis in the later section on census shows an abrupt increase in the years in which helicopter counts replaced fixed-wing counts, indicating that this method is considerably more efficient at detecting horses. Viewed uncritically, about half of the census series in Table 2.ll tended to increase during the census years, about a fourth were roughly stationary, and about a fourth tended downwards. In the last two cases, a number of the declines were caused by sudden, l-year reductions, and one wonders whether a number of these were due to actual removals that were not recorded in the files we saw. The general impression is that the census values tend to increase, and Ryden (l978:295) concedes that the prevailing evidence points to increases in horse herds since l97l despite her suspicions about the precise validity of the agency censuses. In the 72 herds shown in Table 2.ll, there are 25 sets of consecutive census years that (a) span 3 or more years, (b) were taken with either fixed-wing aircraft or helicopters throughout, and (c) in which the census season was held reasonably constant (i.e., prefoaling or postfoaling). These 25 sets are as follows, labeled by the BLM district and herd name: SUSANVILLE Tuledad: 73-75 Shatter Mtn.: 73-75 New Years Lake: 73-75 Massacre Lake: 73-75 BURNS Smyth Creek: 74-76 Riddle Mtn.:: 74-76 S. Steens: 74-77 Catlow: 72-76 LAKEVIEW Beaty Butte: 74-77 Bluejoint: 74-76 West Hogback: 73-75 SALMON Challis: 7l-74 VALE Three Fingers: 72-74 Three Fingers: 76-78 Cottonwood Creek 74-78 Jackies Butte, 72-74 Stockade: 74-78 Hog Creek: 74-78 Cold Springs: 74-76 Pot Holes: 74-78 Basque: 74-78 Lake Ridge: 74-77 BILLINGS Pryor Mtn.: 69-7l ROCK SPRINGS Desert: 76-78 Continental Peak: 77-79 We converted the individual census values for each of the 25 sets (l5 3-year, 4 4-year, and 6 5-year sets) to logarithmic form and

65 regressed them over time to calculate 25 slopes (and therefore 25 instantaneous rates of change). Both a slope coefficient (b) and its variance were calculated for each set. A weighted average slope (b) was calculated by 25 / 25 E b.w. / / . Wj where 1=1 w i ~ var (b) The variance of the weighted mean b was then calculated by 25 / 25 ]Cwi (bi~ var (b) > w. (b. - b)V (n-l) This enables calculation of 95 percent confidence intervals about the mean rate of change. Finally, the mean rate of change, its confidence intervals, and the individual rates were taken to calculate finite rates of change and confidence intervals. Of the 25 calculated rates of change, 22 were positive while 3 were negative. The weighted, mean annual finite rate of change (x) for the 25 sets and 0.05 confidence interval is l.l621.00l, suggesting a l6 percent mean, annual rate of increase. The 25 sets were subdivided into the fixed-wing and helicopter censuses, and the unweighted means for these two groups were +22 percent (9 sets) and +l5 percent (l6 sets), respectively. The group was also subdivided into a group censused in the earlier years (l97l-75), the intermediate years (l974-76), and recent years (l975-79). These showed increase rates of +25 percent (4 sets), +l8 percent (6 sets), and +l5 percent (l5 sets). These two subtests are confounded because of the prevalence of fixed-wing censuses in the earlier years, and helicopters in the later. Hence it is difficult to say whether this apparent chronological trend is a function of changing techniques or actually declining increase rates. While helicopter counts appear to be more efficient at finding animals, and a change from fixed-wing to helicopter censuses obviously increases the number of animals seen, no such change is involved in the 25 tests. They were restricted to sets of fixed-wing-only or helicopter-only counts. Hence the declining increase rates would appear to be attributable to other causes. At least three factors are conceivable: (a) Increased counting skill, It seems possible that censusing proficiency increased in the early years when agency personnel were first conducting the counts. Hence, the higher apparent increase

66 rates of the early years could result from increasing census efficiency, herd growth, or both. Frei and others (l979) have previously presented evidence to show that observers experienced in horse census see a larger fraction of the animals than inexperienced observers. (b) Increased visibility of large herds. The herd-size bias discussed later in the section on census could be increasing the proportion of horses seen. If the populations have increased, the herd sizes may have grown and their visibility increased. (c) Increased herd density. There may be an actual density-dependent reduction in herd growth. As they increase and begin to press the limits of their food resources, their reproductive and survival rates may be ebbing somewhat. All of this is fraught with uncertainties. There are many unknowns about the accuracy, precision, and validity of the census results. But one might think that more weight could be given to the trends implied by the counts, which thereby would serve the function of indices, than the absolute accuracy of census values themselves. This would acknowledge that there may be biases in the counts, but would assume that those biases are relatively constant. However, these results and the earlier views on rates of increase contrast sharply with the findings of Nelson (l979), Conley (l979), and Wolfe (l980), who used a more deductive, less empirical approach. They calculated rates of horse population increase using Lotka method and Leslie matrix techniques, respectively. These authors used fecundity rates like those cited above, giving most credence to limited, first production of young at 3 years of age, and about 50 percent of all older mares bearing young each year. They then calculated the increase rates that would occur from the interaction of these fecundity rates with a range of survival rates. Both authors found the increase rates resulting from "probable" fecundity and survival rates to fall in a range well below l0 percent per year. Only with unrealistically high fecundity rates in Wolfe's calculations, and survival rates approaching l00 percent in Conley's case, could annual increase rates well above l0 percent be attained. They concluded that rates of herd growth as high as 20 percent per year, like those suggested by both agency personnel and a number of authors, were beyond the biological capability of large, established herds which had assumed remote approximations of geometric age structures. In a further effort to determine what increase rates may be for large segments of the western horse populations, we performed life-table calculations from the age-specific fecundity and survival schedules discussed above. The goal was to calculate r, the instantaneous rate of change, and from it \, the finite rate of change. Calculations were performed as follows:

67 r = (loge R0)/T where 20 RQ, the net reproductive rate = ^ ^"Sc where x - age x=0 lx = probability of survival to age x mx = number of female young produced by a female of age x 20 T, the mean generation length, ^ lxmxx/ £ lxmx X - antiloger Three sets of fecundity schedules were used in these exploratory analyses, in descending order of fecundity rates: (a) a "maximum" schedule involving rates that seemed, on the basis of reports reviewed above, as high as could be expected from any horse herd, namely 25 percent of 3-year-olds foaling, 60 percent of 4-year-olds, and 80 percent of 5-year and older animals; (b) the fecundity schedule of range-reared domestics summarized in Table 2.3; and the average of the seven wild-horse studies reported in Table 2.4. In the latter case, the l3 percent of 3-year-olds and the 66 percent of 5-year and older mares were used. But the mean (25 percent) in Table 2.4 for 4-year-olds was based only on three measurements of small numbers of animals, and thought perhaps to be too low by chance. Hence a 50-percent rate was used for this age class. These schedules would seem to be in the high range for reported wild-horse fecundity, and higher than that used by Conley (l979) in his analyses. A range of survival rates was used that appeared to fall in the higher range of values reviewed above. Nine combinations of rates were used with each of the three fecundity schedules: (a) annual "adult" survival rate of l.0 and first-year rates for foals of l.0, 0.95, 0.90, 0.80; (b) annual "adult" survival rate of 0.95 and first-year rates of 0.95, 0.90, 0.80; and (c) annual "adult" survival rate of 0.90 and first-year rates of 0.90 and 0.80. In total, finite increase rates (x) were calculated for 27 combinations of fecundity and survival rates (see Table 2.l2). The results are similar to those of Conley (l979) and Wolfe (l980) who used similar rates, and like theirs must be considered somewhat abstract. The second and third columns of rates are based on observed fecundity schedules but, because survival rates are imperfectly known, are based on somewhat hypothetical or "for instance" survival rates. The values also apply to hypothetical populations that have age distributions resulting from the fecundity and survival rates used in each test. And finally, population students will recognize that our equation for r produces only an approximation of this parameter. With the somewhat problematic precision of the fecundity and survival rates we have used, we did not engage in the more time-consuming iteration of the r values in the discrete-time form of the Leonhard Euler equation. To what extent do these rates simulate increase rates of real-world herds? They do so to the extent that the fecundity and survival rates in real herds are similar to those in Table 2.l2, and age structures in the herds approximate the exponential series

68 TABLE 2.12 Annual Finite Rates of Population Increase (X) Possible Under Differing Rates of Fecundity and Survival Annual Survival Rates Annual A Values by Fecundity Rates 1 Yr.+ Foal Max. 2 Fecund. Range Dom. This4 Study 1.0 1.0 1.17 1.16 1.16 .95 1.17 1.15 1.15 .90 1.16 1.1M 1.11 .80 1.15 1.13 1.13 .$$ .95 1.13 1.11 1.11 .90 1.09 1.07 1.07 .SO l.CO .95 .97 .SO .SC 1.C9 1.06 1.G6 .SO .99 .95 .95 *ri*»*» ax* ti»e «iitiples by To ch*A?* th«ot to ar.tual p*J the r***ia£tr by ICC. «*iicis the population roenta^e increase * su) would increase ? tract 1.0 and i each year. Eultiply rat*s cf r*=,?«~reinesi cciaestacs sbcwre is Table 2.3 >-T«*r aai 5* a'rera^es si>c*rs ra Table 2.4. A 5C percent rate fc*

69 produced by those rates. As we have said, the fecundity schedules are approximately those of the herd reported by Speelman and others (l944), the averages of seven wild-horse studies, and a hypothetically high set. We have reviewed the available data on herd survival rates above, and the range of values shown in Table 2.l2 would appear to represent the higher, more conservative rates. The age distributions in Tables 2.6 and 2.7, and those reported in Table 2.8, while not exactly smooth, geometric series and subject to the foal and yearling biases, still constitute shrinking frequency distributions that crudely resemble exponential or geometric series. If one were to select the fecundity and survival rates that seem to be indicated by the field data on wild horses discussed above and general experience with wild animal populations, one would probably select the right-hand fecundity schedule in Table 2.l2. All studied herds have been found to experience some mortality, and a 0.95 survival rate for the older animals and 0.9 or 0.8 for foals would seem justified, if not conservative, from the available studies. Yet these combinations produce increase rates well below l0 percent (x - l.l0) per year, and considerably below the rates implied by the censuses. Rates similar to those indicated by the censuses are produced only in Table 2.l2 combinations of virtually no • mortality. A 20 percent per year increase rate requires that no mortality occur, and that half of the 3-year-olds and l00 percent of mares 4 years and older produce young each year. Since even domestic herds sustain some mortality and less than l00 percent fecundity, it is difficult to reconcile the differences between the Table 2.l2 results and the census rates. In general, wild horses have rather conservative demography by the standards of most mammalian species. This conservatism is imposed by the following characteristics: only one foal is normally borne; not all "mature" females produce one foal each year; and the age of sexual maturity is delayed, with only a small fraction of 3-year-olds bearing young and the 4-year-olds being the first effective breeding age. Since some young are produced each year, any herd that has been in existence for 4 years or more will have foal, yearling, and 2-year-old components, which are included in the total from which herd increase rates are calculated, but which contribute nothing to the immediate increase. As Tables 2.6 and 2.7 have shown, these become sizeable components of wild herds. In particular, the late maturity date is a strong determinant of the increase rate, as Cole (l954) has shown. And Conley's (l979) calculations that the reproductive output of females older than ll years of age is an insignificant contribution to the increase rate also match with Cole's findings. We are not able at this stage to explain the difference between the herd increase rates predicted in Table 2.l2, and in Conley's (l979) and Wolfe's (l980) projections on the one hand, and those implied by the censuses on the other. The problem will only be resolved by more intensive research both on horse demography and the census methods. But it needs resolution because it bears on several important management issues.

70 One such issue is the size of the l97l horse population. Some observers have been advocating that horse herds be reduced to the levels that prevailed in the year when the Wild and Free-roaming Horse and Burro Act was passed. That population size could be approximated by projecting backward from a known l979 population, using the appropriate rate of population increase. The BLM has been reporting that wild horses in the western United States increased from a total of about l7,000 animals in l97l to a total of 58,000 in l978 or l979. Projecting backwards from a l979 population of 58,000 horses with an annual increase rate of l6 percent produces a l97l population of l7,ll6. If the more conservative increase rate of l5 percent derived from recent, helicopter counts is used, the l97l population would have been l8,960. If the l979 census is conservative, as it is likely to be to an unknown degree, the l97l projections are similarly conservative. But if an annual increase rate of 7 percent is more realistic, then the l97l population was something approaching twice that projected by a l6 percent increase rate. Ryden (l978:295) has already suggested that the l97l population estimates may be conservative. A second way in which increase rates are relevant to management is the rates at which herds must be reduced in order to maintain them at decided-upon levels. If a policy decision determines that a given herd must be maintained at some specified level by removing the annual increment, the number of animals to be removed annually and the associated cost will obviously be very different depending on whether the herd increases by 7 percent or l6 percent each year. We conclude that research on both demography and census are needed to resolve this question. In the case of demographic studies, data on reproductive rates can be obtained quite readily as indicated below. Determining survival rates will be more expensive and will require more time. Feral Asses As with horses, there is abundant information on the demographic characteristics of domestic donkeys. But the evidence on wild herds is more fragmentary than that on wild horses and considerably less exists than is needed to develop firm generalizations on their demography. Most of the information on wild populations is based on studies in six areas. While thorough and intensive, these have necessarily dealt with small numbers (in most of the cases, less than 200 animals), and for short periods of l to 3 years. This data shortage is all the more tantalizing in light of intriguing hints of fundamental differences between burro and horse demography which, if real, may reflect the different ecogeographic origins of these two species.

7l (l) Fecundity: (a) Age at first female breeding. Burro mares commonly begin sexual activity at about l year of age. Moehlman (l974) observed sexual activities in jennies at ll months of age in Death Valley, as did Woodward (l976) at l0 to l2 months in the Chemehuevi Mountains of California, and Morgart (l978) at l.5 years in Bandelier National Monument, New Mexico. Since gestation is approximately a year, as in the horse, the potential age of first foaling is between 2 and 2.5 years, l year younger than that in wild horses; and this potential has been realized to some degree in most cases so far studied. Although Moehlman (l974) found typical age of first foaling to be 3.5 to 4 years during her study in Death Valley—a situation once again similar to the horse—other authors have observed jennies foaling at 2 years of age. Woodward (l976) considered 2 years (N = 4) to be the first age of foaling in a young population where 75 percent of a sample of 20 "adult-sized" animals were 3 years old or younger. Douglas and Norment (l977) found one jenny foaling at less than 3 years of age in Death Valley. In Seegmiller's study (in litt.), a female that had been captured and aged 5 months earlier was seen with a newborn foal when she was about 25 months old. Ruffner and Carothers (l977) found "reproductively active" (pregnant or lactating) jennies between l and 2 years of age. And among 666 burros rounded up on the BLM Phoenix District, which we will detail later (Table 2.l6), 8 of 8l jennies judged to be 2 years old were followed by foals. Since a third of the foals in the herd had become separated from their dams, and foals were underrepresented in the sample, it is possible that more of the 8l 2-year-olds had borne young. (b) Age-specific and herd fecundity. Shorter periods of study, smaller samples, and failure of most studies to subdivide the mares into as many age classes as have the horse investigators make the discernment of age-specific fecundity schedules in burros more difficult. But further analysis of the different authors' data permits a fragmentary and tentative picture: (i) Moehlman (l974:56-58) found no foaling by 2-year-olds, as mentioned above. Only in the third year of her study (l972), and that a partial year, did she observe foaling by 3-year-olds (two jennies). In l970, 63 percent of "adult" (i.e., 4-year-olds and older) jennies bore young. In l97l, this percentage dropped to 47, and when 2- and 3-year-olds were included, the percentage was 4l. While fragmentary, these results suggest the same pattern we observed in horses: an . increase in fecundity as mares advance through the early age classes. Thus, Moehlman found no foaling in 2-year-olds, limited reproduction in 3-year-olds, and somewhere around half of 4-year-olds and older mares bearing foals. (ii) Woodward (l976:42) observed l4 jennies of "coltbearing age" (2 years old and older), of which 79 percent were accompanied by foals in the Chemehuevi Mountains (California) population she studied. Elsewhere, she generalized that jennies produced foals on the average

72 of every l6.2 months, and this produced an annual natality rate of 75 percent foaling (Woodward l979b). (iii) Douglass and Norment (l977:30) concluded that 56.8 percent of females foaled in the l2-month period l November l975 to 3l October l976 in the Wildrose-Emigrant population of Death Valley. Since he observed one mare foaling before 3 years of age, we assume the rate applies to all 2-year-olds and older females. In a radio-collared sample of l0 females in l976, 9 (90 percent) foaled. (iv) Morgart (l978:33) reported 77 percent of the "adult" females on his study area bearing foals. Since 23 young were dropped (p. 29) this implies a sample of 30 jennies. Since three 2-year-olds foaled, we assume the term "adult" applies to 2-year-old and older females. Of a sample of ll "potentially reproductive" marked females, 8 (73 percent) foaled in l975. (v) Seegmiller (l977:24) reported an "adult" (2 years+) population of 58 burros on his study area in the Bill Williams Mountains of western Arizona. Since 57 percent of these were females, the "adult" jenny population was 33. While he studied the population for l4 months and 20 foals were born in the period, l6 were born in the l2-month period June l974 through May l975. Hence the l-year natality rate was l6/33 x l00 = 48 percent for 2-year-olds and older jennies. Seegmiller did not age all of, or determine the number of, 2-year-olds in the population. But he did ascertain the yearling population at about l5. If we make some allowance for mortality and herd growth, there could have been abut l0 or l2 2-year-olds. Hence, the three 2-year-olds producing foals probably constituted a material fraction of that age class. (vi) McCort (l980:24) reported a l975 population of 27 2-year-old and older female burros on Ossabaw Island (Georgia). Between May l975 and May l976, these animals bore l6 foals. Hence the natality rate was l6/27 X l00 = 59 percent for the year. (vii) The 8l 2-year-old jennies accompained by 8 foals in the BLM roundup data cited above suggest a minimum of l0 percent foaling in that age class of that Arizona population. Since some mortality between birth and the roundup may have occurred, and since some of the unaccompanied foals may have been borne by the 2-year-olds, the percentage is conservative in all probability. (viii) Ruffner and Carothers (l977) reported 89 percent and 78 percent of the jennies collected from two Grand Canyon herds to be "reproductively active," a term that implies pregnant and/or lactating. This manner of reporting the data does not provide a clear estimate of annual fecundity rates because it lumps the present year's live foals with the fetuses in utero. Using their data, we have been

73 able to calculate the range of possible pregnancy rates in these herds as follows, first for Bedrock Canyon. (l) Nine females were collected at Bedrock Canyon, although the ages are not given. (2) Eight of these (89 percent) were pregnant or lactating, indicating that they were at least l year old. (3) Six embryos were recovered from them. (4) If all nine females were yearlings and older, the pregnancy rate was 6/9 X l00 = 67 percent. (5) If only the eight were yearlings and older, the pregnancy rate was 6/8 X l00 = 75 percent. For the Lower Canyon Herd: (l) Twenty-two females were collected, age not given. (2) Of these, l6 (73 percent) were pregnant or lactating, and hence were the minimum number of yearlings and older. (3) Fourteen embryos were recovered from them. (4) If all 22 females were yearlings and older, the pregnancy rate was l4/22 X l00 = 64 percent. (5) If only the l6 were yearlings and older, the pregnancy rate was l4/l6 X l00 = 88 percent. Since some prenatal loss in the form of resorption and/or abortion is possible, these percentages cannot necessarily be taken as natality rates. Furthermore, whatever natality rates they do imply cannot be ascribed to any given age group in the population except that, at the birth of the foals, they would be the 2-year-olds and older animals of the population. The above fecundity estimates, except for those of Ruffner and Carothers, are summarized in Table 2.l3. While they are disjunct and based on small samples, some tentative generalizations can be explored. The first is the clear indication that some fraction of 2-year-olds foal in most populations so far studied. Thus, wild asses begin bearing foals a year younger than do wild horses. The BLM data and speculative analyses on Morgart's results suggest that the fraction of 2-year-old jennies foaling may be of a magnitude similar to that in 3-year-old wild mares (Table 2.4). There is some evidence that the percentage of jennies foaling increases in each age class between the 2- and 4-year olds. It seems reasonable to suspect that this is part of an age-specific increase in fecundity up to some optimum age class or range, as discussed above for horses. Although the data are too few to compare fecundity rates of the older age classes in burros and horses, the net effect of the earlier burro reproduction appears to result in higher rates in two or more of the younger age classes. This can be seen by comparing the means from Table 2.4 and 2.l3 in Table 2.l4.

74 CO CO + 03 O 00 [-^ , 1 1 1 l 1 LO 1 S.j i — i ,>-* LO O J. 0) bO 1 £, SB bO 0 CO C >1 1 1 1 l l l 1 1 l a •rl CO •0 O S CO C 0) ^ ^ ^ ^ ^ CO ^ ~t~ J" CM o ro r^ M -* S 0 rH J. OO CO CM C CO* •>H U MH CM l rH CD c~- r- co CD i 0 CO 4-i •H O j" r~ LO t~~ Zf LO LO rH CO ^-\ su U. tT) 0 (T CO ^ C TD rH •H 14-i C O 03 > 00 n o 0 0 1 l 1 l l s.x 0) C *~s CM o o o o rH rH .3! ov° 5 S . LO g-H ^M 4n r^ LO rl •H .H . CD r> CO 0) rM nH rH l rj T3 0) 03 J- U-l TJ O 0 rH J- O co ** c^* 0 C C f-. C~- C*~ [-~ t*-' O CD tT) O O3 ^ CD CD CD »* CD O r.H LO CD CO •H a) >.rH rH rH >, rH .H • t-- f* X - CD CD Q) 0! rH rH CO .C DI O 4-1 >- 0) tU id id 03 rH • rH •P 0i 0 T3 rH 14-, rH s: c 0 C Id 'H 03 O •s • •s CO t-l J 03 > rH > 3 N 0) 0! U O id CU 'H .H C CU C ,C O & Z rt M O a, o .4-J P < k N -H 03 • 0) O -r) •O CO CU O 0) Q CO Q o • cu f-. M CO 2 S 0 < co 3 SB 0 CO *O .H C *O U 0 r) *^ •H U-i P •H 4J <Js o x •H cn 0) CO C CU a 4-1 0) O 4J CO l CO CO CO O> CO — ~ a s r. ^j« vo 6 l^ t^ M -^ CTi 00 H -. 3 O CO 0^ 0^ O r- — o •o d d z ^ CO (U •P n cs H n at ' C *O >- (U rH H '^ fj CN §M CO rd fd ^J ^J bO •H M .H 4J i rH S rH D, §- Q M CO CJ £ Q CO U i4 0) 3 C* *Q O^ CO O 3 s Q 00 Z 3 Q S co S ffl ID

75 Table 2.l4. Comparison of Annual Age-Specific Fecundity Rates in Wild Asses and Wild Horses. Annual Percentage Foaling By Age Class l Species 2YO 2YO+ 3YO 3YO+ 4YO 4YO+ 5YOf Horses Burros 0 10 l3 54 (50)2 (>60)3 6l (>60)4 66 (>60)3 60 iNonparenthetical percentages are from Tables 2.4 and 2.l3. 25 percent obtained for this age class (Table 2.4) was based on 3 years' data from a single study and seems low as an average for wild horses. The 50 percent is a speculative value which seems more likely. 3If 60 percent of 2-year-olds and older jennies bear young on the average, and 2-year-olds bear at a markedly lower rate, then the proportion of 3-year-olds bearing should be greater than 60. The value may increase progressively through each age class if there is a progressive rise in fecundity with each advancing year. 4 Since Moehlman's herd was demonstrably less fecund than the other heras studied, we have used this value as more nearly approaching a norm for the species than the 55 percent shown in Table 2.4. This comparison suggests that fecundity rates, at least in age classes 2 through 4, may be higher in burros than in horses. The age structure of those portions of the herds that are 2-year-olds and older appear similar in horses and burros, as we will see shortly. Unless the foaling rates of 5-year-old and older jennies are lower than that of mares, the comparison in Table 2.l4 implies that total herd fecundity in burros tends to be higher than that of wild horses, on the average. (c) Foaling season. In one study of captive donkey mares, Kohli and Suri (l957) kept records on the outcome of l,l92 services by jacks occurring throughout the year. They found no statistically significant variation in fertility rates, the monthly values varying between 42.5 and 5l.3 percent pregnancy from the services. Feral asses in some populations similarly reproduce year-round, as reported by Moehlman (l974), Woodward (l976) for the Chemehuevi Mountains in southeastern California, Seegmiller (l977) for the Bill Williams Mountains in western Arizona, and McCort (l980) for Ossabaw Island, Georgia (Table 2.l5). However, a degree of seasonality has evolved in some populations. Thus Moehlman found most foals born in May, June, and July. Ruffner and Carothers (l977) reported nearly all foaling in the 5-month period March to July, with one fetus projected to a November birth.

76 TABLE 2.l5 Monthly Distribution of Foals Born in Two Studies of Burro Herds Month Percent Foals Born by Location and Source Georgia (McCort 1980) Death Valley , (Moehlman 1974) Grand Canyon (Ruffner S Carothers 1977) January 21 4 0 February 0 1 0 March 7 1 40 April 7 7 25 May 14 15 0 June 7 19 20 July 7 2R 10 August 0 6 0 September 21 6 0 October 7 7 0 November 7 3 5 December 0 3 0 No. in Samples 14 68 20 . "Combined totals for 1970 and 197l.

77 Morgart (l978) reported that all births occurred between June and November during his period of observation on Bandieler National Monument, New Mexico. But he did not begin those observations until May 20, l978. Ohmart and Bicknell (l975) had collected l5 jennies in Bandelier in February of that same year, and among the pregnant ones they found fetuses ranging in age from 90 days to near term. They concluded that colts were dropped "at any time of year" in the area. (2) Age Composition. We have obtained information from BLM files on burro roundups in the Phoenix, Arizona District. This has provided age-composition data on burro populations similar to that on wild horses reviewed in Table 2.7, and is shown in Table 2.l6. These data show fewer foals and yearlings than 2-year-olds, a pattern also evident in the horse data. We attributed the latter to sampling bias, and are inclined to assume the same for the burros. Beyond the irregularity, the burro age distribution is similar to that of horses in having a predominance of animals in the younger age classes and a progressive decline in each older group. The trend is not perfectly smooth, due no doubt in large measure to sampling error. But the general trend is clear. And once again, a large fraction of the population is concentrated in the young age groups that bear few or no foals: over 40 percent of jennies are in the foal, yearling, and 2-year-old classes. Table 2.l6. Age Composition of Feral Asses Rounded Up in the Phoenix BLM District in l979 Males Females Males Females Age No. % No. % Age No. % No. % 0 50 l9.3 5l l2.5 ll 2 0.8 4 l.0 l 34 l3.l 38 9.3 l2 9 3.5 l0 2.4 2 53 20.5 8l l9.8 13 1 0.4 3 0.7 3 28 l0.8 44 l0.8 14 l 0.4 l 0.2 4 23 8.9 34 8.3 15 4 1.5 4 l.0 5 16 6.2 38 9.3 l6 0 0 l 0.2 6 ll 4.3 33 8.l 17 0 0 0 0 7 l3 5.0 27 6.6 l8 l 0.4 0 0 8 4 l.5 l5 3.7 l9 0 0 0 0 9 2 0.8 7 l.7 20 l 0.4 4 l.0 l0 4 l.5 14 3.4 Some additional information on age composition can be obtained by summarizing published data from the various studies (Table 2.l7). While these are generally subdivided into only three age classes, they give some insight not only into age composition, but also into herd dynamics as we discussed above for horses. We have listed separately the foals-at-birth percentages and the percentages of the late-summer or fall populations accounted for by foals, as we did with horses

78 0) rH 1 o 4 <j tN n cn ".H vo CN Co u*i C0 n ^" vo VD tN O U) i rH rH 0i co o r-i fJ r^rHvo 4.1 g 1 1 1 11 11 I 1 I1 1 1 1 1 II II 1 1 II 1 o ^•r- i fN r* cno fNO o ^ inoo vo rH 0) M vovo i^1 i n r^co r^r^ r. p* r*r ^ r. rH I r- • rH U 1 1 1 11 11 1 1 11 1 « fc g •p •g fN g u 0 CT1 at C rH rHrH rH fN rH rH rHi-HrH *o 2 •H rl IX rH rl G *J V C *C GQ g rH u fN 0 3 •O 0) C^ fN ini n rHcn fn T inx in s 0. E rH tN rH rH rH fN rH t.HrH rH tfl **H tw 0 O 0i r.. & 5 9 fN ^" m CM rH fM 3 . jj rH C rH iH 1. i &i o H t0 • * ~ Bi . SP ^ • •o o a; 5 0) $?£ . s ? § s &p"s gls1^ 3 Jk s &Pi?: "s rl .H rH 0) 4J — £ II rH 4J CJ 1 ^^.H .HrH grH | rH g^^ g1"*^ Sfl^ ^ '«^"H ^ •sli > Ids o «cjoj> rt ! C 5H • J9 'o * 9i J3 Q < 2 , C OtNlH N u .H » ac i .H c 0 vM rH • Q Ul S 2 Q U J < O Q < af! ^ 3 ° S C rl rH 4.1 oi aj tr M U E 3 6 — 8.. » S 3 8 | § « .2 r- en SrH • 8 C 5 M T3 3 tT1 Co in en O C 0) rl • r4 0 ^ — 4J « Unweighted Means ^•Number of foals 1 Number of foals : October percental Bedrock Canyon H< Lower Canyon Here 6September-Novemb< August- November ( 8From Table 2.16 ( ^ vo cn — B O M P git' r* r- rH oo C g — Source — — n ^ o cn x C T3 rH L» ^ S S ^ H 4J « V4 — E n) .H M nj 0> M 4.i rH fl m & v 5 o a) M s MH ,.H u .H a: o ooioo P tf l B JC 1J £ It tt X fl K 3 S V ffl

79 (Table 2.8). The rationale for this separation, and the meaning of differences between the two sets of values, were discussed above. The means for both sets of burro values are higher than the analogous means for horses, as would be expected of a species with a higher reproductive rate, although the differences are not statistically significant. But 4 of l4 individual burro values exceed 20 percent, while only l of l4 horse values does so. The burro means are not statistically greater than the horse means, at least in part because of the greater variability in the burro values. Thus, the late-summer/fall values for the latter range between 9 and 23, those for horses between l3 and l9. The respective coefficients of variation are 26.8 and l5.6 percent. This variation is probably the result of between-year and between-area variation in natality or foal survival, or both. (3) Survival. As with the horse data, survival information on burros is considered in terms of first-year survival and survival of older animals. (a) First-year survival. There is evidence from field observations that survival rates are quite variable in burro foals. Thus, Carothers (l976) observed the high rate of "reproductively active" (pregnant or lactating) females in two areas of Grand Canyon described above, but found very few foals in the same areas during the months of July, August, and September. He suspected that this might be due to high pre- or postpartum mortality. More specifically, the data or Ruffner and Carothers (l977) showed 89 percent of "all females" (N = 9) pregnant or lactating in the Bedrock Canyon herd. With six embryos found on autopsy, this is a 67 percent pregnancy rate for all females irrespective of age. Yet this herd only had two foals (ll percent of l8 animals). In the Lower Canyon herd, 73 percent of 22 females were pregnant or lactating and carried l4 embryos (64 percent pregnancy). This sample contained 4 foals (9 percent of the 45 animals sampled). Moehlman (l974) observed three pregnant jennies during her study of the Wildrose-Emigrant Canyon population in Death Valley. These animals were not accompained by foals at a later date, suggesting that they had aborted or lost their young. She counted the number of foals and yearlings each year in her population between l970 and l972. This allows first-year survival estimates based on the shrinkage of two cohorts. These results are as follows: Year: Foals: Yearlings: Implied survival l970 l97l l972 (yrlg./foal): Three years after Moehlman's study, Douglas and Norment (l977) observed high foaling rates in the same population. The large discrepancy that they observed (Table 2.l7) between the percentages of foals in the population and the percentages of yearlings could reflect

80 the foal mortality that they suggested was occurring. White (l980) calculated a first-year survival rate of 0.66 between l977 and l978 for the Butte Valley population in Death Valley. Dead foals are occasionally found. National Park Service (NPS) officials found a live foal being fed upon by coyotes in Death Valley and gave it to Moehlman (l974). Norment and Douglas found two at Death Valley in a little over a year's observations, and Norment (Norment and Douglas l977) found a still-born fetus along the Colorado River in Grand Canyon. In contrast to the above findings, Ohmart and others (l975) and Seegmiller (l977) found no evidence of foal mortality in Arizona, nor did Morgart (1978) in the Bandelier Monument herd. In sum, these varied sources of evidence suggested that foal mortality in burros may vary between areas from very low rates (Ohmart and others l975, Seegmiller l977), Morgart l978), through the moderate ones calculated from Moehlman's (l974) data, to the high ones implied by Douglas and Norment's (l977) Death Valley yearling percentages and Carother's (l978) observations in Grand Canyon. The great range of yearling percentages in Table 2.l7 further suggests this variability, with a 44 percent coefficient of variation about the mean yearling percentage being more than half again as large as the coefficient for the foal percentage. Year-to-year variation within areas probably adds further to the variability. (b) Yearling and older survival. We attempted preliminary estimates of survival rates in the 2-year-old and older animals in Table 2.l6 by regressing logarithms of the number of animals in each age class on age, as we did above with horses. (We did not extend the regression through the yearlings because of the apparent sampling bias against their numbers.) The calculated finite, annual survival rates are 0.78 for jennies, 0.70 for jacks. If the population was increasing at the time of the roundup, these rates underestimate the true population values, as discussed above. The slopes for the two sexes are not statistically different, and are roughly comparable with those calculated above for horses. Very little adult mortality has been observed by any investigators. Both Moehlman (l974) and Douglas and Norment (1977) saw only occasional dead animals. The latter authors observed four natural deaths out of a population of 86 2-year-olds and older animals in a little more than a year's observations, an observed loss of about 5 percent. One of the deaths was a jenny that died foaling. Seegmiller (l977) saw only one natural death of older animals in the Bill Williams herd, although he saw ll carcasses of animals which apparently had been shot. Most authors agree that predatory loss is almost nonexistent. These varied sources suggest that adult mortality has beeen light in all of the areas in which it has been studied. (4) Sex Composition. The studies reporting sex composition do not present a consistent picture. Woodward (l976), Douglas and Norment (l977), G. A. Ruffner (as cited in Douglas and Norment l977),

81 and Walker and Ohmart (l978) all report appreciable excesses of males ranging from 53 percent (Ruffner) to 68 percent (Douglas and Norment). Woodward also concluded that jacks survive longer than jennies. To the contrary, Seegmiller (l977) reported 46 percent males in the Bill Williams herd while Morgart (l978) reported 43 percent in Bandelier Monument. And the percentage of males in the Phoenix District roundup is 39 (Table 2.l6). This preponderance of females also parallels the sex composition in horses discussed above. Furthermore there is a suggestion (Table 2.l8) of the same age-specific changes in sex ratios that were observed in the horse (Table 2.l0): a balanced ratio at birth, increasing distortion toward females in early and mid-life, and a partial restoration of balance in the older ages. Table 2.l8. Age-Specific Sex Ratios in Burros Rounded Up in the BLM Phoenix District, l979.x Age Foals l-3 4-6 7-l0 ll-20 % Males + 0.025 C.L. Sample Size 50+l0 l0l 4l+6 278 33+7 l57 27+9 86 4l+l5 46 •'•Data from Table 2.l6. Whether these sex-ratio discrepancies reflect real differences between populations or sampling bias is not known. Douglas and Norment (l977) suggest that there may be a tendency to see males more readily when they display partially turgid penises or announce their presence by braying. (5) Population Trend. Because the demographic parameters we have been reviewing are both more variable and less precisely known than those for horses analyzed above, life-table calculation of increase rates becomes more problematical. In review, we have seen that: (a) Fecundity rates vary over a wide range. The low end, as reported by Moehlman (l974), is comparable with that of horses, with the 4-year-old jennies the youngest effective breeders and foaling rates of adult females 4l to 63 percent. At the high end, some fraction of 2-year-olds foal; and annual foaling rates as high as 79 percent (Table 2.l3), and in one small sample as high as 90 percent, have been reported. (b) First-year survival rates appear similarly variable, with some values conceivably below 0.5, and others approaching l.0. (c) Adult survival rates appear high and comparable with those of horses, probably falling somewhere in the range between 0.7 and l.0.

82 One of the first questions we can address is the advantage burros gain over horses by the fact that they breed l year earlier. Cole (l954) has shown that population increase rates are extremely sensitive to the age of first breeding, more so than to moderate variations in total number of young produced. We have performed life-table calculations on annual increase rates of both horses and burros, using a hypothetical rate of first foaling in 3-year-old horses and in 2-year-old burros of 25 percent. All older age classes were assigned 50 percent foaling rates. All age classes were assigned a 0.9 mean annual survivorship. The results produced a potential, annual increase rate of 4 percent in horses and l3 percent in burros. Clearly, the difference in this variable alone provides the burros with a materially higher potential rate of increase, provided that survival rates are similar. As in horses, some high rates of increase have been reported for burro populations: (a) Douglas and Norment (l977) calculated an l8 percent (X = l.l8) annual increase rate for Death Valley burros in the mid l970s following herd reduction. (b) Ohmart and others (l975) estimated 20 to 25 percent for the Havasu Resource Area in California-Arizona over an l8-month period, doubtless utilizing the work of Woodward (l976) and Seegmiller (l977) who arrived at similar conclusions for the same locale. (c) Morgart (l978) inferred a 29 percent (X = l.29) annual increase rate for Bandelier Monument. We can now calculate what demographic patterns would be required (and we do not suggest that they necessarily occur in nature) to attain rates of these magnitudes. We performed calculations based on the following three hypothetical conditions: (i) Half of 2-year-olds and 60 percent of all older ages foaling each year. (ii) Half of 2-year-olds and 80 percent of all older ages foaling each year. (iii) Half of 2-year-olds and l00 percent of all older ages foaling each year. Given these schedules and l.0 survival rates, burro herds could increase at the following three, respective rates: (i) x = l.l6 (l6 percent per year) (ii) X = l.20 (20 percent per year) (iii) X = l.22 (22 percent per year) One additional implicit assumption is an age distribution approximating a geometric series.

83 Obviously, the three published rates of increase reported above reach or exceed the biological limits of the species. In populations with age distributions like that in Table 2.l6, they could only obtain with no mortality and maximum possible reproductive rates. But since several of the values are based on intensive observations of small herds over periods of time, one hesitates to question their validity. Not all burro herds have been reported to increase at these high rates. The Bedrock Canyon Herd in Grand Canyon had experienced periodic reduction, and Ruffner and Carothers (l977) found high reproductive rates, a high proportion of young animals (78 percent between l month and 4 years of age), excellent physical condition, and presumed population increase prior to their removal effort. But the Lower Canyon Herd had not been subject to human interference, had only 49 percent of its individuals in the first 4 age classes, was not in as good physical condition, and was presumed to be roughly stationary or growing only slightly. Similarly, Moehlman's (l974) Death Valley herd had not been exploited for some years, exhibited a low reproductive rate, and appeared to increase at 4.8 and 9.7 percent from l970 to l97l and from l97l to l972. While observing substantial fractions of foals in the Butte Valley population of Death Valley (Table 2.l7), White (l980) measured almost no population change from l976 to l979. He concluded that the population was stabilized by emigration and perhaps the low first-year survival described above. He further concluded that densities were roughly 5 to 6 times Moehlman's (l974), 7 to 9 times those of Douglas and Norment (l977), and some of the highest ever reported. What ecological patterns can now be inferred from the disparate results that we have reviewed in the past several pages? No definitive answers are possible, but we suggest some hypotheses that might serve as leads for future research. It seems instructive to consider asses in comparison with horses. The domestic, and now feral or "wild," horse is derived from the wild Przewalski's horse (Equus caballus prezewalskii). Its original distribution was the high-latitude Asian and European steppes with their strong summer-winter contrasts, and a semi-arid to mesic precipitation pattern. Such a climatic regime placed survival value on a warm-season foaling period and marked reproductive seasonality. With a moisture regime of intermediate dependability, probably made less risky by strong nomadic habits, relatively fixed and unvarying reproductive rates could evolve and endure. We cannot speculate on why those rates are so conservative. The North American wild horse is now settled in the New World ecogeographic analog of its ancestral range: the western Great Plains and the Intermountain sagebrush and salt-desert-shrub steppes. Its reproductive pattern is highly seasonal (Table 2.5) and of limited variation (Tables 2.4 and 2.8). The domestic donkey, and now feral burro, is descended from the African wild ass (Equus asinus). Its native range is the low-latitude (l0°N), arid regions of northeast Africa. A combination of minor, seasonal changes in temperatures and photoperiod permitted the evolution of a year-round reproductive capability as it has in so many low-latitude species.

84 Similarly, the North American feral burro is largely a denizen of the continent's "hot" deserts, particularly the Mojave and Sonoran. At the southerly extremes of its U.S. distribution in southern California and Arizona (latitude ca 34 to 34.5°N), and on Ossaoaw Island, Georgia (ca 3l°N), there is no statistically demonstrable seasonality to its breeding. But at the slightly higher latitudes and/or altitudes of Bandelier Monument (ca 36°N), Grand Canyon (ca 36°N), and Death Valley (ca 36.5°N), a degree of seasonality has developed (Table 2.l5), although it is not as marked as that in horses (Table 2.5). One trait that has definite survival value in arid regions is demographic flexibility. Deserts are the most variable of all terrestrial systems in terms of precipitation and primary production. Many desert animal species the world over, ranging from snails and insects to birds and mammals, possess great flexibiity in a variety of demographic parameters. They reproduce and increase rapidly during periods of above-average rainfall, and take advantage of temporarily favorable conditions. But they curtail reproductive activities during drought periods. Some species have abandoned all reproductive seasonality and developed the ability to breed at any time of year when capricious rain occurs. It seems possible that the variability in the burro data we have reviewed may reflect the demographic flexibility of a desert species. Its population processes may adjust to such vicissitudes as subnormal rainfall, inadequate forage, and population density. Several observers whom we contacted during the course of this study expressed the view that burro foaling rates are higher during years of above-average precipitation and abundant forage than in years when rainfall is below. Several authors have suggested that observed variations in reproductive and/or survival rates are due to variations in density. Douglas and Norment (l977) postulated that the differences between the foaling rates they observed and those of Moehlman (l974) were a function of the differing population densities (Figure 2.l). The herds studied by Douglas and Norment (l977) and Woodward (l976) had been lowered by herd reduction prior to the studies, and their high increase rates and high proportions of young animals were thought to be responses to reduced density. Ruffner and Carothers (l977) proposed a similar hypothesis for the two Grand Canyon populations, the Bedrock Canyon herd having a lower density, a higher reproductive rate, a larger proportion of young animals in the population, and a presumed higher increase rate than the Lower Canyon herd. White's (l980) report of high emigration and low foal survival rates associated with exceptionally high density may fit this same pattern. If these hypotheses are correct, they may have important management implications. Burros appear to have materially higher potential increase rates than horses, and to be capable of populating an area rapidly. Indeed, Woodward (l976) called it "[t]he epitome of a successful colonizing species...." If the management goal for a given area is to maintain burros at low to moderate numbers, or as in the national parks to eliminate them and prevent their repopulation, it may require diligence and continuous effort.

65 • DATA FROM MOEHLMAN (1974) A DATA FROM PRESENT STUDY I DATA 1NCOMPLETE FOR YEARU972) IOCH 80- <60- o u- 111 _J |40' UJ .2 .4 .6 DENS1TY/Km2 1.0 Figure 2.l Relationship of the percentage of jennies foaling to population density in Wildrose-Emigrant population of Death Valley (after Douglas and Nonnent l977).

86 However, burros may also have considerable ability to limit their own density, once their numbers begin to weigh on their own populations and their environments. Conceivably this phenomenon could reduce the need for herd control. The important question is whether that limitation is effected at densities below which vegetation impacts and conflicts with other values are excessive in terms of specified management goals. Our later discussions on the effects of burros on other ecosystem components bear on this question. All of this is speculative and supported by a thin data base. The subject of demography needs considerable research aimed at measuring more fully the reproductive, foal-survival, and adult-survival rates, and their variation due to different environmental pressures. They will need to be measured in a number of areas to disclose the basis for differences cited above, and for several years in each area to record the range of variation and its causation. Foaling rates should be fairly accessible at nominal expense by examining animals at roundups. But as with horses, research on survival rates will be considerably more demanding of time, logistics, and funds. In all probability, telemetry will be an indispensible tool. One other problem that concerns some observers is uncertainty about accurately determining burro ages. Many of the hypotheses about burro demography developed above—first breeding age, age-specific fecundity, and survival rates—presume that the authors cited were able to age the animals accurately. In fact, age criteria have not been specifically worked out for burros. Burro investigators use the tooth-eruption and wear pattern worked out for horses (see Welsh l975 for review). This has been justified by Woodward (l976), Ruffner and Carothers (l977), and McCort (l980) on the grounds that the pattern has been found similar in three different species of Equus: the plains zebra (E. burchelli), the mountain zebra (E. zebra), and E. caballus itself. The assumption is that the same pattern holds for E. asinus. Woodward (l976) was able to make limited tests on the assumption by checking dentition of animals she had earlier captured and marked while they were foals. She found no discrepancies. But additional research would give more confidence in the validity of the assumption. Demography-Related Characteristics of the Mare Reproductive Cycle The domestic mare begins ovulating at about l year of age—that is, the first season after the year of birth. The adult mare is seasonally polyestrous, with a cycle length of 2l to 22 days between ovulations. Spontaneous prolongation of luteal activity is common in the mare and can extend the cycle length to 2 months. Horse mares can cycle year-round; however, they tend to be seasonal depending upon latitude, with perhaps 2 to 6 months of anestrus during the winter. The pony mare is more strictly seasonal (as are the Pryor Mountain wild horses), with all animals cycling by May and none after September.

87 Multiple ovulations are common in the mare and may occur in l0 to 25 percent of cycles. Multiple ovulations and the multiple conceptions that can occur from them may represent a significant part of the apparent early pregnancy failure rate. Multiple ovulations will not make a significant demographic contribution (less than 0.l percent of thoroughbred foals are survivors of twin pregnancies), since twins are rarely carried to term in domestic mares and have significantly lower birth weights and reduced rates of survival. Multiple ovulations and twinning very rarely occur in ponies. It is unknown whether multiple ovulations and multiple implantations occur in the wild horse. Another curious feature of the horse is that unfertilized ova do not leave the fallopian tubes and persist over a period of months while degenerating. Thus, the recent ovulatory history of the mare can be ascertained by flushing and examining tubal washings for ova. The onset of estrus, as well as the occurrence and frequency of ovulation, have been shown to be modulated by nutrition. Furthermore, embryonic survival is reduced by poor nutrition. Thus nutritional status in late winter and early spring may affect age of first breeding and time of onset of estrus and successful establishment of pregnancy, even if ovulation and breeding occur. The seasonal cessation of estrus in the fall may be a function of genetics and nutrition and also of light, since time of reproduction in the mare and pony has been shown to be modulated by artificial changes in photoper iod. Sexual receptivity in the horse mare is said to average 5.7 days, whereas the pony mare responds for about 8 days—5.7 days before ovulation and 2.3 days after. Mares can exhibit estrous behavior without ovulating and animals can ovulate while exhibiting very low levels of estrous behavior (Asa l979). Occasional animals can exhibit more or less continuous estrous behavior and willingness to copulate regardless of ovulatory status. The sequence of endocrine and gonadal events after conception have been concisely represented in Figure l0 of Stabenfeldt and Hughes (l977). Gestation length is about 340 days in horses. These events provide the rationale for a series of tests for pregnancy, the usefulness of which is related to the stage of pregnancy and viability of the fetus. Some of these tests are summarized in Ginther (l979). They include rectal palpation of the reproductive tract (days 24 to 45), pregnant mare serum gonadotropin (PMSG) (days 45 to l20), progesterone (days 45 to l30), and estrogens (days l20 to 340). An elevated PMSG test combined with a low progesterone result indicate a dead or resorbing embryo. The level of progesterone in milk can be used to detect pregnancy in a lactating mare. Urinary estrogens are useful beginning about l00 to ll0 days past conception. Ultrasound detection of fetal heartbeat can be used from 45 to 90 days (approximately l00 percent effective after 90 days) to term. Delivery usually occurs at night and lasts about l0 to l5 minutes in uncomplicated births. Premature delivery can be induced by administration of exogenous corticosteroids (l00 mg of dexamethasone around days 32l-324 with delivery on day 328). These foals survive.

88 Since capture and handling of late pregnant mares often results in early parturition, delivery could be induced in a controlled manner by means of these drugs. The whole issue of premature delivery might be avoided with drug-capture techniques that use tranquilizers rather than succinylcholine, which does not alter the animal's awareness of environmental events and produces symptoms of "stress" in white-tailed deer (Wesson and others l979). A "foal heat" (estrus and ovulation), occurs over about 7 to 9 days and within l3 days after foaling in domestic mares. Successful breeding is possible at this time and is commonly done in domestic animals. Foal heat is likely to occur in wild horses, but its modulation by nutrition has not been studied. Thus, it is important to test captured mares that have accompanying foals or are lactating for pregnancy (milk progesterone, PMSG, serum progesterone) to establish the frequency of this annual rebreeding in wild populations and its relationship to age. This testing could be initiated in the adopt-a-horse program, especially with animals captured in the fall, using the combined techniques of PMSG and progesterone from blood samples and ultrasound detection of fetal heartbeat. Live foals are produced by about 70 out of l00 well-managed domestic mares bred. About 80 mares in l00 are estimated to become pregnant; thus failure to conceive is about 20 percent. This failure rate may be due to frequent use of hand-breeding or to inbreeding depression, since 90 to l00 percent of pasture-bred quarter horse mares are reported to become pregnant. The death rate of live foals varies widely depending upon season of birth and source of data: about 55 percent of mares bred may have foals that survive to l year of age. This figure is close to that reported for wild horses, but that fact may only be coincidental given the multiple uncertainties of the domestic animal data. We cannot safely conclude that mares breed each year on the average in the wild (i.e., breed at foal heat if previously pregnant) rather than in alternate years. Direct study will be necessary. Although l-year-old domestic fillies do ovulate and can become pregnant, many of them lose their fetuses: 69 percent conceived and 44 percent delivered in one study (Mitchell and Allen l975). It is unlikely that animals of this age conceive in the wild. However, the actual year of first breeding will require direct study. It has been suggested that first breeding is delayed until the age of 3 or 4 years in the wild and that it is modulated by nutrition. Age-specific fertility in domestic mares has been reported to vary between 65 and 79 percent up to age 7, between 8l and 89 percent to age l2, and then to decline gradually to 50 percent at age 20 (Ginther l979). Age-specific mortality rates useful to wild horse demographic studies cannot be inferred from domestic data, except that juvenile mortality can be high and strongly dependent upon management conditions. In conclusion, if age-specific fertility and mortality data of wild mares are necessary for management purposes, then they will have to be collected from studies of wild populations. The techniques are available for assessing pregnancy states of freshly captured wild

89 mares and hence to allow estimates of fertility and probable natality rates. Methods for applying these techniques to captured wild horses are discussed under "Feral Horses: Fecundity" in the section on "Equid Demography." Genetic Polymorphism A knowledge of equid genetics can contribute to management in at least two significant ways. First, an animal population must embody a certain amount of genetic diversity in order to be able to adapt to marked environmental changes and survive. The genetic composition of a population is presumably adjusted to render it most fit for its contemporary environment. However, a large population will contain considerable diversity, the extremes of which may not necessarily be the most fit individuals for that environment. But given a profound environmental change, such as climatic oscillation, some of the previously marginal genotypes may now be the best adapted and serve as a breeding source for a new, better-adjusted population. In order for a population to have a sufficient probability of survival, it must be large enough to contain enough genetic variability to meet likely environmental contingencies. A management program could well determine what this size is, and set as an objective the maintenance of individual populations at or above this level. If forage allocation or other considerations prohibited populations of minimum size in certain areas, this situation might be the basis for deciding to remove entire, small herds from such areas. A second management implication relates to the contentious issue of the lineage of wild horse herds. Many of the individuals and groups who strongly support wild horse management (e.g., Ryden l978) contend that many of these extant herds are derived from previously abundant Spanish mustangs. Opposition groups, such as livestock and wildlife interests, argue that most wild horses today stem from escaped or released draft horses, cavalry mounts, and saddle animals. This dispute could be settled by genetic studies. The biophysical characteristics of blood enzymes and proteins have been found to be so closely associated with genetic make-up that these serum constituents can be used as delicate and precise genetic markers. It is quite likely that the serum proteins of the Spanish barbs and the Arabian breed (the apparent progenitors of the Spanish mustangs), as well as those of the different domestic breeds, could be precisely identified and described in contemporary, captive animals and then sought in wild herds. In all probability, the complex lineage of wild animals could be worked out with considerable precision. At present, information on equid population genetics is scarce. There are apparently no systematic studies of the levels of genetic variation in wild equids. Data, however, do exist with respect to some domestic breeds. The genetics of coat color of the horse has been under intensive study for more than half a century. Because the genetics of color variants is fairly well-known in many cases, it seems that the coat color information available from the BLM census

90 studies could be used to obtain some information on gene frequencies. Nevertheless, for tne reasons advanced in the section on "Needed Research" (below), it seems that such information by itself is not sufficient to obtain estimates of genetic variation in wild equid populations comparable to the estimates that exist for many other kinds of organisms. The study of blood enzymes and other proteins is, at present, the best method for obtaining such estimates. An important study of the genetics of blood systems was published by stormont and Suzuki in l964. These authors studied l6 blood factors in equine blood and established that a minimum of eight gene loci are responsible for the control of blood groups in horses. Extensive polymorphisms exist in at least four of the eight loci: the minimum number of alleles is six at one locus, five at another, and three at each of two other loci (Stormont and Suzuki l964). Allelic frequencies at the various loci were found to be different in different breeds; two breeds in particular (Shetland ponies and thoroughbreds) are markedly different with respect to those frequencies, as shown in Table 2.l9. A practical extension of the genetic analysis of blood groups was presented by the same authors in l965. In cases when a mare had mated with two different stallions, it became possible to exclude one of the two stallions from paternity in about 65 percent of the cases (Stormont and Suzuki l965). The genetics of blood groups has also been used to confirm the inheritance of coat color in equids (Trommershausen-Smith and others l976) . The chestnut rule in coat-color genetics asserts that matings between chestnut horses never produce bay, black, brown, or grey offspring; the grey rule asserts that grey offspring must have at least one grey parent. The authors studied nine alleged exceptions to the chestnut rule, and eight alleged exceptions to the grey rule. Study of the blood groups of the animals involved demonstrated that the "exceptions" were due to incorrect parentage assignation. These two applications of blood-group genetics illustrate the resolving power of genetic studies. An early investigation of genetic variation in horse proteins other than those involved in blood-group determination was also published by Stormont and Suzuki (l963); it involved variation in albumin phenotypes. Gahne (l966) studied variation in four blood proteins, each encoded at a different gene locus, by means of gel electrophoresis. The number of alleles found at each of the four loci was: albumin, 2; prealbumin, 4; esterase, 4; and transferrin, 6. Stormont (l979) has recently reviewed the genetic studies of blood proteins in horses. Table 2.20, adapted from his publications, gives the number of alleles known at each of the eight loci responsible for the polymorphisms observed in each of the corresponding proteins. (For an additional review, see Kaminski l979). Other genetically determined proteins studied in horses are: o-galactosidase (Beutler and Kuhl l972), butylcholinesterase (Chiu and others l972), liver alcohol dehydrogenase (Ryzewski and Pietruszco l977), and hemoglobin (Sandberg and Bengtsson l972). Several of the protein polymorphisms studied in horses have also been investigated in burros: transferrin (Niece and Kracht l967),

9l TABLE 2.l9 Estimates of the Frequencies of Alleles of Eight Blood-group Loci in Shetland Ponies and Thoroughbreds Frequencies in Loci Alleles Shetlands Thoroughbreds .M 0.3107 0.7050 A' 0.2852 0.0290 A a" 0.0358 0.0036 A'H 0.060l 0.0000 • 0.3082 0.2624 1D 0.1392 0.0000 D dJ 0.12l5 0.l503 d• 0.7394 0.8497 EPl 0.34l5 0.2058 EP' 0.0483 0.09l0 E 0.6102 0.7031 ae 0.1519 0.5082 R 0.3869 0.0000 51 s 0.0103 0.l038 a Q a 0.1306 0.0756 RS 0.l893 0.3l25 a a 0.13l0 0.0000 C* £C 0.652l 0.7317 K* k* 0.l796 0.0635 T* t.T 0.4505 0.6594 U* u 0.3l74 0.l485 *The frequency of the alternative allele at each of these loci is sirupIy 1 minus the figure shown. SOURCE: Data from Stonnont and Suzuki (l964).

92 TABLE 2.20 Number of Alleles Associated with Various Blood Proteins in Horses Protein Number of Alleles References Serum proteins Albumin Prealbumin Postalbumin Esterase Transferrin 2 Stormont and Suzuki (l963); Gahne (l966) 8 Gahne (l966); Braend (l970); Trommershausen-Smith and Suzuki (l978a) 2 Braend (l970); Stormont (l972); Trommershausen-Smith and Suzuki (l978a) 4 Gahne (l966); Scott (l970) 6 Gahne (l966); Braend and Stormont (l974) Red-blood-cell proteins Carbonic anhydrase Catalase 6-Phosphogluconate dehydrogenase 3 Sandberg (l968); Deutsch and others (l972a,b) ; Deutsch and Bray (l975) 2 Kelly and others (l97l) 2 Sandberg and Bengtasson (l972) References not included in Stormont's review.

93 hemoglobin (Trujillo and others l967, Isaacs l970), esterase (Kaminski l970), 6-phosphogluconate dehydrogenase (Bergman and Gustavsson l97l, Hof and Osterhoff l973), and albumin (Blake and Douglas l978). Esterase (Kaminski l970, Kaminski and others l978) and 6-phosphogluconate dehydrogenase (Hof and Osterhoff l973) have been studied in other equids as well. The present state of knowledge concerning the genetics of natural populations of wild and free-roaming horses and burros may be summarized as follows: l. No information exists about these populations concerning any of the relevant genetic questions, such as the amount of genetic variation within populations, the amount of genetic differentiation between populations, and the pattern of genetic relatedness ("phylogeny") of the wild populations and the domestic breeds. 2. No studies exist of domestic horses or donkeys that would give valid estimates of the amount of genetic variation within a breed, the amount of genetic differentiation between breeds, or the pattern of genetic relatedness among breeds. To obtain these estimates would require the systematic study of a random sample of genes (i.e., random with respect to the degree of variation). The existing studies concentrate on one or another gene (responsible for a given coat color, blood group, or protein) and, in general, pay particular attention to genes known to be polymorphic. 3. Genetic information exists, particularly for domestic horses, with respect to coat colors, blood groups, and various proteins; polymorphism appears to be pervasive (and, it would seem, as extensive at least as in other vertebrates). The successful application of the techniques of gel electrophoresis to the study of a number of protein polymorphisms indicates that such techniques could be used to obtain the kind of population genetic information needed (see the two previous paragraphs). Food Habits of Horses and Burros Since North American wild burros and horses occupy rather different biomes—burros in the hot or southern deserts, and horses largely in the plains, steppes, and cold or northern deserts—marked differences in food habits are to be expected, and possibly in nutritional characteristics as well. At least part of the burro's ability to adapt to the harsh environmental conditions of the desert Southwest probably relates to the broad spectrum of plant species it will accept as food. Most studies show that burros are highly opportunistic feeders, and can greatly alter their diets in response to forage availability and phenology (Browning l960, Hansen and Martin l973, Koehler l974, Woodward and Ohmart l976, Jordan and others l979a, Seegmiller and Ohmart l980). They apparently prefer green grasses and forbs when these are available (Hansen and Martin l973, Woodward and Ohmart l976). However, virtually all researchers have remarked that, during dry seasons or periods when forage is scarce, burros utilize

94 plants and plant parts not usually considered forage for domestic ruminants. For example, Ohmart (l975) observed burros consuming large palo verde (Cercidium microphyllum) branches, and Koehler (l974) reported them eating yucca (Yucca spp.) plants and bark from cholla cactus (Opuntia imbricata). In their comprehensive review of literature on equine nutrition, Robinson and Slade (l974) observed that domestic donkeys may have a higher capacity for digesting crude fiber than either horses or cattle, and that feral equids of both species can survive on diets deficient in total nitrogen as well as specific amino acids. These physiological capacities might well contribute to the burro's adaptability to such low-quality forages. Table 2.2l displays a representative, though not exhaustive, summary of food-habit studies conducted on burro populations in several environments. Variations in the grass component, compared with that of horses (see Table 2.22 and later sections of this report) further emphasize the plastic nature of burro feeding habits. In contrast to the dearth of information on grazing impacts, considerable research effort has been devoted to studying the diets of horses over a wide range of conditions in the West. Much of this research has been supported either directly or indirectly by the BLM. In some cases, diets of sympatric ungulates have also been studied to ascertain the degree of dietary overlap. Results of several of these studies are summarized in Table 2.22. While some studies have reported that forbs (Smith l949, Archer l973, Hansen l976, Wagner l978, BLM l979c) and browse (Frischknecht l975, Hansen l976) are periodically important in horse diets, most studies (Hubbard and Hansen l976, Hansen and others l977, Olsen and Hansen l977, Salter l978, Vavra and Sneva l978, Salter and Hudson l979) have emphasized that gramineous species constitute the major proportion of the annual diet. This state of affairs is emphasized by the low variability associated with the grass component in Table 2.22, which is based on studies conducted over all seasons and under many different environmental conditions. In a parallel situation, Casebeer and Koss (l970) found that zebra diets more closely matched the grass composition of the vegetation type where they grazed than did those of three species of ruminants: cattle, wildbeest, and hartebeest. Archer (l977) and P. V. Fonnesbeck (Utah State University, personal communication, l979) both found that horses can be highly selective feeders, even showing preferences for short, new growth in a sward while rejecting the older growth of grasses present there. Stoddart and others (l975) noted that horses are the most selective grazers of domestic livestock and, under conditions of forage scarcity where heavy utilization has been forced, no other animal matches the horse's ability to crop the vegetation closely. Arnold and Dudzinski (l978) have made similar observations. Several authors reported that horses and cattle often consume very similar diets in sympatric situations (Olsen and Hansen l977, Salter l978, Vavra and Sneva l978). Vavra and Sneva (l978) found that similarity of diets was greatest in dry years and during seasons of dormancy when forage is least available. Some potential

95 TABLE 2.2l Diets of Free-Roaming Burros Over a Range of Vegetation Types and Seasons, as Determined by Fecal and Stomach-Content Analysis Month/ Season (s) Dietary Composition (%) Study Vegetation Type Grasses Forbs Browse Woodward and Ohmart (1976) Mojave Desert (lower Colorado River- Chemehuevi Mts. Jan. Feb. 0.0 l.2 22.7 46.9 73.8 36.0 Mar. 2.2 77.4 19.5 Apr. 7.7 58.2 34.l May 0.2 5l.9 38.l June 0.0 37.2 58.0 July 2.0 l2.l 82.3 Aug. 2.4 l3.9 78. B Sept. 2.3 8.4 83.8 Oct. 12.6 8.0 74.0 Nov. 2.9 10.9 82.9 Dec. 14.3 11.2 73.l Annual mean 3.9 30.1 6l.l Hansen and Martin (l973) Mohave Desert (in lower Grand Canyon) Mar. Annual mean 79.6 68.6 l5.8 l3.1 5.7 13.7 Jordon and others (l979) Mohave Desert (Grand Canyon Nat'l. Park) July Aug. 47.8 34.4 23.3 l7.4 l5.2 l0.l 3l.8 48.7 64.2 Sept. Annual mean 35.2 l4.2 48.2 Seegmiller and Ohmart (l976) Colorado Desert (lower Colorado River- Bill Williams Mts.) Jan. -Mar. Apr. -June July-Oct. l.8 30.l 33.l 56.5 34.5 11.2 39.6 30.4 48.6 Annual mean 22.0 33.0 40.0 Browning (l960) (stomach contents) Mohave Desert (Death Valley) Spring and fall l0.0 39.0 50.0 Mean and coefficient of over all five studies variation (%) 27.3 (84.2) 22.6 (46.0) 40.8 (49.0)

96 TABLE 2.22 Diets of wild, Free-Roaming Horses 0ver a Range of Vegetation Types and Seasons, as Determined by Fecal Analysis Dietary Composition (%) Study Vegetation Type(s) Season (s) Grasses Forbs Browse Hansen (1976) Desert grassland Spring 58 28 9 Summer 58 13 28 Fall 36 31 28 Winter 36 42 21 Annual 47 28.5 21.5 Hubbard and Hansen (1976) Mountain shrub Pinyon- juniper Annual Annual es 1 0 12 89 7 Eco tone Annual 97 0 2 Hansen and others (1977) Saqebrush-qrass and pi nyon - j unipe r Annual 94 0 5 olsen and Hansen (1977) Sagebrush-grass and saltbrush Annual 91 8 1 Salter and Hudson (1979) Upper foothills of boreal forest Jan-Mar Apr-May 87.5 92.5 0.6 0.8 5.9 3.0 June-Aug 98.5 0.8 2.3 Sept-Oct 95.2 1.9 1.5 No v- Dec 89.3 0.4 5.1 Annual 92.6 0.8 3.6 Vavra and Sneva (1978) Desert-forest fringe of the cold desert Normal Year Spring Fall 99 — — 98 Winter 100 — — Annual 99 — — Dry Year Spring 95 — — Summer 100 — — Fall 100 — — Winter 97 — — Annual 98 — — Vavra and Sneva (1978) Sagebrush-grass (4 locations varying in sage- brush dominance) Annual Annual Annual 92.8 95.9 85.8 7.0 3.6 0.2 0.5 1.2 12.9 Mean and coefficient of variation (%) over all seven studies Annual 95.2 2.5 2.4 89.4 (15.1) 5.8 (146.7) 5.1 (125.7)

97 food-competition relationships are also reported to exist between horses and elk during winter and spring (Salter and Hudson l979), and particularly between horses and such climax species as bison, bighorn sheep, and pronghorn antelope (Wagner l978), which apparently have more restrictive dietary niches. Some caution is necessary, however, in interpreting dietary overlap as a strong indication of competition. Mere dietary overlap does not directly translate into exploitative competition unless shared forage resources are in short supply. In this connection, Salter and Hudson (l979) noted that horses were ubiquitous in their distribution over several major plant-community types. There was little contemporaneous spatial overlap between horses and cattle, even though fecal analysis showed 67 percent overlap in their summer diets. Noticeably absent from most studies on horse diets is any type of quantitative description of food availability. Likewise, few studies have adequately described the spatial relationships of horses and sympatric ungulates. Hence, very little, if anything, can legitimately be said about competition for food between horses and domestic or wild ruminants. Certainly, no populations of horses and sympatric species have been monitored for purposes of identifying competitive exclusion resulting from exploitation of the food supply. Equid Forage Requirements and Nutrition Knowledge of forage requirements and nutrition is important to effective management of horse and burro populations and the ranges they occupy. From the standpoint of animal welfare, the horse or burro biologist must understand the animals' nutritional needs for maintenance and reproduction in relation to the nutritional plane that the range is capable of providing seasonally. From the standpoint of range welfare, the manager must have reasonably good estimates of the quantities of forage consumed daily by a given animal population in order to establish stocking rates and, if necessary, allocate forage to other herbivores occupying the range. The two needs are not distinct and call for close collaboration between the range manager and the horse or burro biologist. The following discussion will first address factors affecting forage requirements and then discuss equid nutrition in the rangeland setting. Theoretically, the configuration of the equid digestive tract makes these animals less limited than ruminants by the quantity of forage dry matter they can turn over per unit of time (Bell l969, Janis l976). The implication to the range manager is that horses may be able to consume more forage per unit of body weight than cows can. Hence the animal unit (AU) requirement for forage might be more for horses than for equivalent-sized cows, especially on ranges where forage is of poor nutritional quality. (An animal unit is considered to be a 454-kg (l,000-lb) cow or her equivalent by Stoddart and others [l975].) Apparently, some grazing-capacity assessments conducted by the BLM for cows and horses have gone forward on the assumption that a mature horse is roughly equivalent to l.25 AU.

98 Heady (l975), in speaking of exchange ratios for various animal species on rangelands, states that "... if the food eaten is reasonably the same for both species being compared, the ratio of metabolic weights gives the exchange." He used the widely accepted interspecies mean of body weight raised to the fractional power of 0.75 to define metabolic body weight. Thus, using this procedure, a 454-kg horse and a 454-kg cow both have metabolic body weights of 98 kg and both are equivalent to l.0 AU. In practice, however, such exchange ratios are profoundly affected by the kind of range, the age and physiological status of the animals, and the kinds of forage consumed. Hence they appear to offer little basis for comparing animals as dissimilar as cattle and horses. Another possible approach in comparing ruminant species is the use of recommended nutrient standards, such as those assembled by the National Research Council for sheep (NAS l968) and cattle (NAS l976). Such a comparison could be based on either digestible or metabolizable energy requirements. However, direct comparisons of such energy standards between horses (NAS l976) and ruminants are not advisable because of differences in digestible-energy utilization by equines and ruminants. Considerable energy from easily digestible foods is absorbed as glucose in the foregut of the equid. Undigested material, consisting largely of plant cell walls, is subsequently fermented in the cecum, yielding volatile fatty acids (VFA) that are absorbed and utilized as energy substrates by the animal. In ruminant digestion, however, fermentation occurs in the foregut, and consequently there is a major dependence on VFAs as energy substrates, with little or no direct utilization of glucose. As forage matures and becomes less digestible, fermentation in the cecum becomes relatively more important to the equid (Hintz and others l979). Under such conditions, decreasing amounts of energy are absorbed as glucose in the foregut and increasing amounts are derived as VFAs in the hindgut. More metabolizable energy is available from foregut utilization of glucose than from hindgut utilization of VFAs, because the fermention process is attended by losses of methane and heat that do not occur in foregut gastric digestion. Hence, the equid obtains less metabolizable energy from each unit of digestible energy as the digestibility of its food declines. This shift is not as much of a problem in ruminants, since the proportion of food that is fermented remains relatively constant over the entire range of forage digestibility. Moe and Tyrrell (l976) found that for ruminants, metabolizable energy expressed as a percentage of digestible energy ranged from 80 percent for forage diets to about 88 percent for high-concentrate diets. The relationship of digestibility, dietary fiber, and energy nutrition of the equid are discussed in detail (with numerous appropriate reference sources) in Appendix A. Estimates of daily forage requirements for horses, based on required total digestible nutrients (TDN), have occasionally been suggested. Data on TDN requirements are readily available for both horses (NAS l973) and cattle (NAS l976) and would, as first approximations, appear to offer a simple method for equating forage

99 requirements of the two species. However, such comparisons must be interpreted with caution. An example follows. For a 400-kg mare and a 400-kg cow, both in the last 90 days of gestation, recommended TDN levels are 3.72 kg daily for the mare (HAS l973) and 4.00 kg daily for the cow (NAS l976). Forage diets are assumed to supply 4,000 kcal of digestible energy per kg of TDN for horses (Fonnesbeck l968). Thus, according to this approach, cows would require some 7.5 percent more TDN than mares of equivalent size and physiological status. The difference of 0.28 kg TDN per day (4.00 - 3.72 = 0.28) can presumably be attributed to the less efficient utilization of digestible energy by the ruminant, which is the result of fermentive losses discussed earlier. The main deficiency in this approach is that there is no direct way of translating TDN to quantities of forage dry matter actually consumed. The latter quantity is of most interest to the range manager who must allocate a fixed forage resource to one or more grazing animal populations. Presumably, if equids digest highly fibrous forage diets less thoroughly than do ruminants, and the rate at which digested material is passed through the equid is not limited by the configuration of the digestive tract as it is in ruminants, a horse might consume more forage in order to extract its required quantity of TDN (or energy) than would an equivalent-sized ruminant. Detriment (Appendix A) approached this question on a theoretical basis and, using published data and some assumptions, constructed a family of curves comparing intake of similar-sized horses and cattle as a function of dietary crude-fiber content. At dietary crude-fiber levels lower than 40 percent, horses would consume less dry matter per day than would cows; however, above about 45 percent dietary crude fiber, horses would consume more dry matter per unit of body weight than cows, and the difference would increase at an increasing rate. In this connection, Stoddart and Greaves (l942) reported that native grass species (including Agropyron, Bromus, Stipa, and Poa spp.) from northern Utah mountains contained about 28 percent crude fiber in spring and 36 percent in fall. Crude-fiber values higher than 40 percent of dry matter are uncommon in native range forages (NAS l97l). Although the accounts are few and some of the data are questionable, the published literature seems to support the contention that equids may, in fact, have relatively higher rates of forage intake than do ruminants of equivalent size. After extensive review of the literature, Cordova and others (l977) concluded that realistic estimates of intake for grazing ruminants fall in the range of 40 to 90 g/kg BW°' , where BW°'75 is the metabolic body weight of an animal. Intake rates at the lower end of the range were associated with mature and cured forages and those at the upper end with immature, easily digestible forages. Thus, forage intake by horses and burros should approach or exceed 90 g/kg BW before being accepted as evidence that their intake requirements are higher than those of ruminants. Koehler (l974), citing an obscure "U.S. Forest Service personal communications" reference, stated that daily intake by burros on Bandelier National Monument was approximately ll Ibs (5 kg), but he

l00 did not give the average weight of the animals. The NFS (l979) assumed that the daily intake by Grand Canyon burros was 5.l kg of forage per l67-kg animal, versus l.6 kg of forage per 64-kg sympatric bighorn sheep. These values convert to ll0 and 7l g/kg BW°*75 for burros and bighorns, respectively, but the difference is complicated by differences in body size of the two species. Apparently, the NFS used the same estimates of intake requirement for burros in both Bandelier National Monument and Grand Canyon National Park. This value was taken from work done by Maloiy (l970) with the Somali wild ass. Darlington and Hershberger (l968) found that ponies voluntarily consumed timothy hay at 82 g DM/kg BW°'75 per day (DM = dry matter), Intake levels of orchard grass and alfalfa were 65 to 90 and 80 to 83 g DM/kg BW°'75, respectively. The fact that the intake rates of confined ponies approached or equaled 90 g/kg BW^-75 indicates that under free-ranging conditions their intake requirement may, in fact, be higher than that of ruminants. Recalculation of intake data presented by Ngethe (l976) suggests that zebras are capable of daily forage-intake rates of at least l57 to l65 g/kg BW°*75. Ngethe's zebras were fed a cafeteria-style diet of cut grasses, but apparently the pens he used allowed the animals to exercise. Both of the foregoing studies measured intake in the highly controlled pen situation, and thus there is no reason to question the accuracy of the data. However, they probably underestimate intake by free-ranging animals, since the latter presumably have somewhat higher maintenance energy costs. Obviously this question requires research, particularly in terms of the mid- and low-quality native forages that characterize rangeland during much of the year. Janis (l976) suggested that, due to their presumed greater intake rates, equids are superior to ruminants in dealing with high-fiber forages, providing that intake is not limited by the actual quantity of forage available. In other words, equids might have a competitive advantage over ruminants in situations where a critical nutrient such as nitrogen is present in low concentrations in the available forage. This is not to say that equids digest protein any more efficiently than ruminants (Vander Noot and Gilbreath l970), but they possibly have "access" to a greater amount of the nutrient by the fact that they can process a larger quantity of food (Robinson and Slade l974, Hintz and others l978). The ruminant's throughput rate is limited by the capacity of the rumen and the fermentation rate there. The question of equid digestive physiology is of considerable theoretical, as well as practical, interest. Several recent reviews of equine nutrition (Slade and Robinson l970, Mehren and Phillips l972, Robinson and Slade l974, Hintz and Schryver l978) have covered various aspects of the topic. Janis (l976) discussed the evolutionary aspects of equid digestion and possible ramifications regarding interactions with other sympatric ungulates. The reader is referred to these papers, as well as to Appendix A, for more detailed discussions of digestive physiology and possible ecological relationships.

l0l Nutritional Value of Diets Consumed on Rangelands Cook (l975) expressed the widely held belief that most western ranges lack suitable year-round forage resources to sustain resident populations of horses and burros, and that eventually this lack will limit their populations by fostering poor reproduction, disease, and starvation. Although we are not aware of any documented evidence (i.e., demographic studies) that such limitations presently exist, animals in poor body condition have periodically been observed on some horse ranges. Young and others (l976) suggested that Great Basin plant communities have not evolved in association with equids since their Pleistocene extinction, and that feral horses were only able to expand their ranges into the Basin after the advent of Europeans. At any rate, now that wild horses and burros are established on many ranges, it must be determined whether the seasonal quality and quantity of their food are currently creating limitations. Generally, forage availability and nutritional quality are greatest during the season(s) of active plant growth. As plants mature, crude fiber, lignin, and cellulose increase. Part of the change in nutritional composition is the result of decreases in leaf-to-stem ratios, changes in chemical constituents within plant structures, and some leaching. These lead to lower digestibility in most cases. After maturation, leaching of soluble carbohydrates, proteins, and certain minerals and vitamins greatly accelerates and continues throughout the period of dormancy. Some species are more susceptible to leaching than others, and it is more severe among herbaceous plants than among shrubs. Shrubs maintain protein and vitamin levels better than do herbs, but they do not provide enough metabolizable energy to sustain weights in domestic livestock (Cook l975). Leaf drop and shatter also contribute to qualitative, as well as quantitative, losses during dormancy. The areas in which the major horse and burro herds roam differ in climate, and thus support somewhat different regimes of plant growth. In southwestern regions, rainfall patterns range from a single winter mode in southern California, through a bimodal precipitation pattern in Arizona, to a single summer mode in New Mexico. In areas where the bimodal pattern holds, two annual peaks in forage quality are often observed: one in late winter and early spring (provided winter rains occur), and one in late summer (Cable and Shumway l966). Grasses mature during the latter period. In the Great Basin region and adjacent high plains areas of Wyoming and Colorado, precipitation occurs mainly from September to April, much of it as snow. Most plant growth occurs during spring and early summer, with the vegetation becoming dormant by mid- to late summer and remaining so throughout the ensuing fall and winter. Moderate temperatures and adequate precipitation in fall may sometimes result in some "greenup" of vegetation (primarily gramineous species) at that time. Thus, forage quality is generally highest during late winter and early spring—and again in late summer—in the southwestern region, and during spring and early summer in the Great Basin-Intermountain region. Drought conditions in either region can reduce the availability of forage

l02 during growing seasons and during periods of dormancy. Snows often greatly reduce forage availability during the winter in the Great Basin-Intermountain region. Raleigh (l970), working in the sagebrush-bunchgrass region of Southeastern Oregon, reported that nitrogen in the diets of grazing cattle declined from 3.0l percent (l8.8 percent crude protein) in early May to 0.46 percent (2.9 percent crude protein) in mid-September. Further declines into the fall and winter may have been observed if the study had continued past September. In the same study, digestible nitrogen available from forage had fallen below recommended levels for cows nursing calves (NAS l976) as early as late June. Digestible energy reached levels deficient to cows with calves in mid-July. If one substitutes recommended levels of crude protein and digestible energy for lactating mares (NAS l973) into Raleigh's (l970) curves for forage quality, digestible protein (or digestible nitrogen x 6.25) would appear to become deficient in early June, but a deficiency of digestible energy would appear unlikely, even into September. This exercise assumes that horses would consume diets similar to those of cattle, and—on the basis of horse-diet data published by Vavra and Sneva (l978) for the same general area—this may not be an unreasonable assumption. But it does not consider the possibility that horses may compensate for low-quality forage by eating more of it (Janis l976). Cook and Harris (l968) and Rittenhouse and Vavra (l979) provide a comprehensive summary of nutritional data for domestic animals in salt-desert shrub and sagebrush vegetation areas of the Great Basin. In addition, Murray and others (l978) have recently published a detailed account of nutritional values for some 20 important forage species common to the Great Basin area. Their data were presented in such a way that the effects of plant phenology or stage of maturity can be clearly ascertained. The studies mentioned above, as well as numerous others (Vallentine l978 lists some 7l5 bibliographic entries on range-animal nutrition specific to the western United States), have focused entirely on the nutritional value of range forage for domestic cattle and sheep. Fewer studies (e.g., Smith l952, l954, l957; Dietz and others l962; Short and others l966; Urness and others l975) have considered wildlife species. But virtually all studies, whether concerned with wild or domestic creatures, have been carried out on ruminants. Essentially no data exist on the nutrition of the free-ranging equid in the western United States. Moreover, the studies on domestic animals have generally been done from the perspective of economically effective levels of animal production. It is conceivable that, in the case of wild horses and burros, the question of mere survival may occasionally be the important one. Studies with ruminants have shown that nutritional deficiencies in animals cannot necessarily be determined by comparing the nutritional value of hand-harvested forage plants with animal requirements. By selectively feeding on specific plant parts, the grazing animal usually ingests food that is of considerably higher quality than that

l03 of the plant as a whole. The detailed data presented by McCulloch and Urness (l973) on white-tailed and mule deer in Arizona chaparral show that by selectively feeding, deer maintained a relatively uniform level of protein intake all year long, even though the gross amount of protein available in the vegetation fluctuated markedly from season to season. Similar findings are reported for cattle (Hardison and others l954) and sheep (Arnold l964). Langlands and Sanson (l976), however, using cattle and sheep as experimental subjects, demonstrated that there are limits to what can be gained by selective feeding. As forage availability declined during the course of their study, a point was reached at which the most nutritious forage became too difficult to obtain in sufficient quantities to make seeking it worthwhile. If utilization becomes heavy enough and animals can no longer afford to select just the most nutritious food, the nutritive value and digestibility of their intake will eventually decrease. The utilization level at which animal selectivity no longer meets nutritional requirements becomes a primary concern. This type of information is scarce for domestic ruminants (Pieper and others l959, Cook and others l962) and is totally lacking for wild horses and burros. When considering possible nutritional deficiencies in equids, it is important to recognize that the recycling and storage of some nutrients may buffer some apparent deficiencies in the diet. Nitrogen recycling in horses (Robinson and Slade l974) may lower the maintenance requirement during seasons of plant dormancy when crude-protein availability is low. Church and Pond (l976) reported that ruminants can store reserves of vitamin A in the liver for 90 to l20 days. Recycling and storage are perhaps more important to horses than burros, however, because the latter typically have year-round access to evergreen plants in the types of habitats they occupy. Furthermore, parts of the Great Basin-Intermountain region support palatable shrubs that contain good sources of vitamin A all year (Cook and Harris l968). If, as Janis (l976) and Hintz and others (l978) suggest, equids do compensate for a low availability of critical nutrients by processing larger volumes of forage, then the most important nutritional consideration to wild horses and burros may be the quantity rather than the quality of forage. On horse and burro ranges, forage availability is most often limited by the degree of current annual production, utilization, and sometimes snow cover. This last limitation may not be as important for horses, since they are often reported to be capable of pawing through snow for forage. Salter and Hudson (l979) reported that horses were "adept" at obtaining forage beneath snow as deep as 0.6 m, and that they foraged in snow-covered areas even when south slopes were bare. In the same study, winter diets reached a low of 6 percent crude protein and a high of 52 percent acid detergent fiber, probably acceptable levels for maintenance. Probably the most critical criterion for evaluating horse and burro forage supplies is the extent to which energy is available and the degree to which availability coincides with requirements. As is

l04 true for most mammals, the nutritional demands of horses and burros are highest during the last trimester of gestation (when 90 percent of fetal growth occurs), and during lactation and growth (see HAS l973 for specific values). Energy required for thermoregulation may also be substantial at times during winter months, but data inadequacies prevent evaluation of this topic. Some of these periods of high requirement coincide with the period(s) of peak forage nutritional values. Energy for maintenance of body temperature in winter is a notable exception, but such a demand is generally not as enduring or as great as those for reproduction and lactation. Wild and free-roaming-horses generally foal in the spring when forage is generally nutritious. Thus, nutrient requirements of lactating mares and foals are probably met except when droughts are extreme or ranges severely depleted. Although it is not documented, fetal development may sometimes be inhibited, especially on ranges where forage is scarce in winter. Yearlings may also suffer insufficient intake for growth on depleted ranges and during harsh winters. Burros, by contrast, foal year-round, with a peak in spring and summer, as discussed above under "Equid Demography." Consequently, performance of jennies, survival of foals, and growth of colts are likely to be less dependent on seasonal greenup than on annual productivity of the range. Greenup will certainly favor reproductive success, but this event coincides with the peak requirements of only a portion of some herds. In other herds foaling is virtually as seasonal as it is in horses. Basically, we assume that wild horses and burros are limited by the same constraints in food supply as are grazing ruminants, the main difference being that horses and burros may be more resistant to depleted forage supplies than are ruminants. This presumption, however, must be verified. Habitat Preference and Use, and Interspecific Competition Forage preferences and consumption rates alone do not provide enough information to support decisions for a given land unit on the amount of forage to be allocated to horses and/or burros, livestock, and wildlife, and on the numbers of each type of animal to be carried on the unit. Several different patterns of interaction could be envisaged between these categories of species—patterns that would influence decision making. l. The different groups might select mutually exclusive habitat types, and whether forage preferences were similar or not, would not affect each other's populations. No interspecific competition would occur in this situation. 2. The different groups might have overlapping habitat preferences but through behavioral interaction, segregate into discrete portions of the habitat. If competition here is gauged by what a species is capable of in the absence of the other, this example

l05 could be classed as competition if forage in one (or more) of the habitats became limiting to the species occupying it (them). 3. The different groups might have overlapping habitat preferences and remain sympatric. If they had different forage requirements, they would not compete. If they had overlapping forage requirements, but were not present in sufficient numbers to deplete the forage to limiting amounts, they would not compete. But if the forage were reduced to the point of affecting the welfare of one or more of the groups, then competition would occur. Clearly, when developing criteria for site suitability it is important to understand habitat preferences and uses, and whether competition is a possibility or a reality. Possible competition between wild horses and cattle was reviewed before the following comments were made. Broadly viewed, habitat analysis and evaluation—the techniques of which are reviewed in deVos and Mosby (l97l)—address two complementary sets of questions. One set involves the structural characteristics of the habitat in terms of vegetation, topography, soils, and water. Extensive literature exists on the measurement of habitat in such diverse organisms as small mammals (Rosenzweig and Winakur l969; M'Closkey l972, l975, l976, l978; Rosenzweig l973; M'Closkey and Lajoie l975; M'Closkey and Fieldwick l975; Conley l976; Lemen and Rosenzweig l978), medium-sized mammals (Conley and Southward unpublished), molluscs (Green l97l), a variety of birds and other taxa (Shugart and Patton l972), and elk (Cervus elaphus) and mule deer (Odocoileus hemionus) (Sivinski l979) . A second set of questions involves the behavioral responses of animals to the habitat. These questions are concerned with the habitat aspects that are required for such activities as feeding, breeding, parturition, escape, and protection from weather, and collectively are subsumed under the term "habitat selection." In establishing suitability criteria for horses and burros, it is important to recognize that while there is some optimum for them that might appropriately be termed the "preferred habitat," the animals may be forced to occupy suboptimum conditions due to habitat degradation, competitive displacement by other species, or simply the absence of the optimum. This type of habitat may be called the "subsistence habitat." The parallel terms of animal response are "habitat preference" and "habitat use." The distinction is made here because it cannot be automatically assumed that an animal's presence in a given type necessarily implies that that type is optimum. Since interspecific competition can profoundly affect a species' habitat-use patterns, and must be taken into account in allocating forage among the several species groups dealt with here, it should be considered in some detail. By most definitions, interspecific competition is judged on the basis of two criteria (Milne l96l, Conley l976): (l) two species compete when they both use some resource that is present in short supply; and (2) in using the resource, each species reduces the other's population performance, and ultimately fitness, to levels below what these measures would be in the absence

l06 of the other species. The important implication is that two species can use the same resource, but it their joint use does not reduce it to the point where each limits the other's demographic performance, they are not competing. This point seems to have escaped the writers of much of the literature on competition. The extensive writing on dietary similarities among wild equids, livestock, and wildlife often infers competition without evidence of resource limitation or population effect. These reports (Cole l954, Tueller and Lesperance l970, Hansen and others l973, Hansen and Reid l975, Hubbard and Hansen l976, Hansen and others l977, Hansen and Clark l977, Olsen and Hansen l977, unpublished data obtained from BLM files) provide important data for describing the biology of each species, assuming the efficiency of the techniques used (Hansen and others l973, Todd and Hansen l974, Deardon and others l975; but see Keiss l977, Smith and Shandruk l979, Vavra and others l978). But such information is not sufficient to demonstrate competition among the species involved even when the vegetation is surveyed (e.g., Jordon and others l979), much less when it is not (e.g., Hansen l976). Two species may use more than one common resource, and they may conceivably compete for one without competing for the other(s). It was mentioned earlier that two species may segregate into separate areas of the habitat as the result of behavioral interactions, and thus may eliminate competition for food. There are reports in the literature of this situation occurring between domestic animals and wildlife. Jeffery (l963) and Mackie (l970) reported that when cattle were present elk vacated areas that they otherwise occupied. These flexible patterns of resource use are explained in Hutchinson's (l958) construct of the niche occupied by a species. In the absence of competition, a species occupies some broad portion of the resource spectrum for which it has tolerance; Hutchinson terms this portion the "fundamental" niche. In the presence of competitors, the species may constrict its distribution into some subset of the fundamental niche for which it is best adapted, and this subset is called the "realized" niche. If a species expands its use or occupancy of resources upon the removal of another species, circumstantial evidence of competition has been established. This expansion has been termed "ecological release" (Ricklefs l973, Pianka l974). The concept has been used by various workers to infer competitive pressures (Ayala l970; Koplin and Hoffmann l968; Peterson l973; Neill l974, l975; Rosenzweig l973; Schroder and Rosenzweig l975; Davis l973; Simon l975; Grant l969, l97l; Morris and Grant l972; Grant l975; Crowell l973; Crowell and Pimm l976). Avoidance is more generally the response between competing species than outright aggression (Andrzejewski and Olszewaki l963, Kikkawa l964, Colvin l973, Grant l978), but exceptions exist, particularly under experimental conditions (Conley l976). It has long been recognized that "niche space" is a highly complex phenomenon, and that in any given circumstance only a portion of such a theoretical construct can actually be measured. This process, called "a partial analysis of niche" by Maguire (l967), has been followed by a number of recent workers.

l07 All of this discussion comes down to the point of discerning and measuring competition between wild equids, livestock, and other wildlife so that it can be provided for in management plans. Since an essential criterion of competition is the creation of a population effect, its existence cannot absolutely be established without experimentally manipulating one species and ascertaining whether the other responds. If eguids are competing with other species for food, the effect is presumably on nutrition and ultimately demographic performance. Since equid demography is so conservative, a demographic change following experimental reduction in a suspected competitor would be difficult to measure in the time available for this project. It is hoped that a nutritional change could be detected through blood analysis (described below under "Needed Research"), and that the demographic results could then be assumed. Research on possible nutritional response is outlined in Chapter 5. If equids partition the habitat with other species through behavioral interaction, this state of affairs can be identified readily through experiment. Habitat use can be measured in the presence and absence of suspected competitors. Such experiments are outlined in Chapter 5 of this report. NEEDED RESEARCH Overview Rationale for Projects l Through 7 It is clear from the "State of Knowledge" sections in this and later chapters that there is not enough of the information needed to formulate effective horse/burro management plans. Broad gaps in understanding exist in nearly all aspects of management. In order to fill these gaps, the Committee is advocating the l8 research projects listed in Chapter l of this report. We believe that the data base provided by these projects will help develop the more effective equid management programs toward which the BLM and USFS are moving, and which PL 95-5l4 prescribes. The projects are divided into four groups, which correspond to the four main chapters of this report. The first three groups also correspond roughly to the areas of concern of the three subcommittees and to the three connotations, discussed above, of the term "excess" used in PL 95-5l4. In considering what is implied by the term "excess," we have approached the matter from the following perspective. Any given tract of land has a certain ecological potential to support herbivores in an equilibrium state. That potential is determined by the site's climate, topography, soils, water, and vegetative characteristics, as well as by the nature of the collective herbivores themselves. Policy decisions based on biological, sociopolitical, and economic considerations can allocate some portion of that potential to equids. As brought forth in our discussion at the beginning of Chapter 2,

l08 equids reach excess numbers when they increase to population levels that (a) threaten their own health and welfare, (b) threaten other components of the ecosystem they occupy, and/or (c) interfere with other management goals for that area. In order to allocate a portion of the area's potential to equids, the manner in which they use it in terms of habitat selection, food consumption, and interaction with other animals must be understood. Once the allocation is made, their biological performance—in terms of nutrition, demography, behavior, and genetics—must be monitored to detect when the animals are approaching excess numbers in the first sense, and are threatening to exceed the portion of the area's potential allocated to them. Accordingly, the seven research projects that are discussed in this chapter are designed to assist in making the allocation decisions, and to monitor the welfare and biological performance of the animals. These are: Project l. Habitat Preference and Use by Co-occurring and Separately Occurring Feral Equids and Cattle Project 2. Food Consumption Rates and Nutrition of Wild and Free-Roaming Horses and Burros and Their Associated Species Project 3. Nutritional Plane, Condition Measures, and Reproductive Performance in Domestic Mares Project 4. Blood Assay of Experimental Equids and Livestock in Projects l, 2, 3, 5, and 8 Project 5. Demography of Wild Horses and Burros Project 6. Social Structure, Feeding Ecology, and Population Dynamics of Wild and Free-Roaming Horses and Burros Project 7. Genetic Polymorphism. The projects set forth in Chapter 3 address the welfare of other ecosystem components and the second connotation of what constitutes "excess" numbers, while those in Chapter 4 address the third connotation. Time Constraints There is one aspect of the research on which the Committee strongly and unanimously concurs, and we believe that we must declare our view on the matter forcefully. PL 95-5l4 and the BLM/NAS contract decree that the research shall be carried out essentially in the 2-year period covering l980 and l98l, and that the Committee is to marshall the evidence and complete a final report in l982. It is the Committee's opinion that much of the information needed to provide a sound base for management programs cannot be generated in a 2-year period. The area in question encompasses semi-arid and arid regions. Such climatic types are the most variable on earth. Rainfall, varying from year to year in a largely random and unpredictable fashion, may differ by a factor of l0 between 2 successive years. Vegetative production is equally variable.

l09 Consequently, the composition and quantity of forage available to horses and burros, as well as to other herbivores, varies markedly between years. Any hope of ascertaining the overall patterns of food preference, nutritional condition (and therefore behavioral and demographic performance), competition with other herbivores, and impacts on vegetation and watersheds depends on long-term research that covers the full range of climatic variability. At the very least this work would take 6 to l0 years. The research program that we propose in the following pages will increase our current knowledge of horses and burros substantially. But we wish to make it clear that a 2-year effort would fall considerably short of supplying the informational needs that have been discussed in the foregoing pages. In a number of cases, a 2-year effort would add so little to what we now know that one could question the wisdom of making the commitment to it. Integration, Scale, and Geographic Distribution Project l should be conducted with study plots no smaller than 5 to 6 square miles per experimental treatment (to be outlined shortly), and possibly much larger. The types of scientific observations to be made will include equid behavior and habitat measurement. Project 2 (and Projects 8 and 9, to be outlined later) should be conducted in study plots ranging in size from about l00 to 300 acres per experimental treatment. The scientific observations to be made include range ecology, feeding behavior and nutrition, and measurement of various watershed parameters. Thus several disciplines will be needed, and the location and design of experiments will need to consider the availability of these capabilities. In addition, Project 4 should be carried out in conjunction with Projects l, 2, and 8, and therefore should be planned and designed in coordination with them. Project l0 (discussed later) is likely to need a study area of about the same size as that needed for Project l, but will require the presence of a riparian zone. Therefore, where possible (i.e., where scientists of both disciplines are available) both Projects l and l0 should be carried out in the same sample plots. The ideal integration of these projects can be seen below in the scheme for a single replication. Grazing Intensity Class of Animals Moderate Heavy Horses Projects l, 4, l0 Projects l, 4, l0 Proj. 2, 4, 8, 9 Proj. 2, 4, 8, 9 Cattle Projects l, 4, 7 Projects l, 4, l0 Proj. 2, 4, 8, 9 Proj 2, 4, 8, 9 Horses and Cattle Projects l, 4, l0 Projects l, 4, l0 Proj. 2, 4, 8, 9 Proj. 2, 4, 8, 9 Neither Proj. 2, 9 Proj. 2, 9

ll0 Each of these experiments should be replicated three or four times, perhaps once in each of three or four states. It is not essential that the large-scale (l, l0) and small-scale (2, 4, 8) studies be combined at a single site as shown here. If expertise for Project l—but not for Projects 2, 8, and 9—exists in a single area, then a replication of Project l could be carried out in one area, and a replication of 2, 8, and 9 elsewhere. But if the combined expertise can be brought together in a single area, then integration is desirable and efficiency is gained. First priority is given here to horses and cattle because the possibility of competition, both behavioral and nutritional, seems to be greatest. However, if funds permit, sheep could be included, although their addition to the above scheme would double the number of experimental treatments in each replication. A similar rate of increase could be anticipated with the addition of each additional wild ruminant species. The scheme could also be repeated using burros instead of horses. Details of the individual projects follow. Project l. Habitat Preference and Use by Co-occurring and Separately Occurring Feral Equids and Cattle Rationale As discussed above, before proper resource allocation and development of site-suitability criteria can be carried out, it is necessary to understand the habitat preferences of equids and other large herbivores on the range, and how their use of preferred habitat might be modified by interspecific competition. Since preferences—as well as the presence and distribution of the different species—may vary seasonally, it is desirable to initiate studies on this phase as early as possible. They should be conducted year-round throughout the period of this program. Objectives l. Determine utilization patterns in relation to habitat structure (vegetation, topography, soil, and water) demonstrated by feral equids and domestic bovids in the presence of, and the absence of, potential competitors (i.e., equids alone, equids and cattle, cattle alone). 2. Develop an appropriate multivariate statistical model that allows testing of the null hypothesis: "there is no difference in utilization patterns between equids and bovids." 3. Use the statistical model in (2), above, to describe probabilities of habitat utilization patterns that can be used to develop site-suitability criteria for both equids and cattle.

Ill 4. Synthesize the results from (l) through (3) above with the nutrition program, and evaluate the existence of, and potential for, competition between the two species. Methodology l. A single experimental block should be a large area containing wild horses or burros. 2. Such a block should be subdivided into six treatment cells about 5 to 6 square miles each, and containing adequate habitat and topographic diversity. a. Cattle stocked at proper, long-term carrying capacity levels. b. Cattle stocked at levels considered to be excessive with regard to long-term health of the vegetation. c. Equids stocked at levels described in a. d. Equids stocked at levels described in b. e. Equids and cattle stocked as in a. f. Equids and cattle stocked as in b. These cells must be large and diverse enough to allow natural segregation of habitat use patterns if differences exist, and to allow for sufficient numbers of individuals to ensure statistical reliability. 3. To eliminate bias, the sampling program should be a suitable random or stratified-random design, and should concentrate on activities related to (a) feeding and watering, (b) escape (from pests or disturbance), and (c) parturition and care of newborn. For each of these categories, multivariable vectors representing characteristics of habitat structure are to be obtained. Variables chosen for measurement should reflect local conditions and incorporate consideration of vegetative structure, aspect (topography), edaphic structure (in the broad sense), and availability of water and/or other special requirements. It is desirable to use telemetered individuals to facilitate location for observation. In keeping with the nutrition experiments, the use of breeding-age females should be emphasized in order to maximize the potential for detecting demographic effects if they exist. 4. Preliminary measurements should be evaluated for variation, and sample sizes computed (e.g., Cochran l977) for stated levels of precision. Statistical procedures should follow the example of workers cited above (e.g., M'Closkey l975, Shugart and Patton l972, Sivinski l979). Multivariate analyses that allow testing of null and alternate hypotheses and projections to group membership should be emphasized. Suitable divisions of data sets should be included to maximize predictability. 5. This entire experiment should be replicated three or four times in as many states so that generalized statements can eventually be made about equid and boyid habitat preferences and use patterns, and possible effects of competition.

ll2 6. Blood assays should be taken (see Project 5), and the reproductive condition of mares should be checked in the appropriate season. Project 2. Food Consumption Rates and Nutrition of Wild and Free-Roaming Horses and Burros and Their Associated Species Rationale Knowledge of animal impacts on plant communities addresses only half of the complex question of grazing animal management. The range manager must also understand the autecological limitations on animal production. Studies of this nature provide the critical link between the plant community and the demographic response of the animal population in question. Considerable research has been conducted on the nutrition and feeding ecology of sheep and cattle (e.g., Cook and Harris l968; Cook and others l953, l962, l967), mule deer (e.g., Dietz and Nagy l976, Smith l952, Smith l959, Hansen and Reid l975), and pronghorn antelope (e.g., Beale and Smith l970, Mitchell and Smoliak l97l, Severson and May l967, Severson and others l968). However, little comparable work has been done on either the horse or burro under range conditions. Virtually no information exists on daily forage intake by horses. This information is critical in that it enters into calculation of grazing capacities under the BLM's forage allocation system, and it supplies the quantitative perspective for any nutritional evaluation of habitat. From an experimental and technical point of view, studies on nutritional value (i.e., chemical composition) of range forage ingested by grazing horses and studies on forage intake can logically be integrated and can proceed simultaneously. However, the dietary chemical composition has secondary priority and could be delayed or deleted if necessary. Studies on food habits can also be integrated into such an inquiry, and the data can be easily obtained in conjunction with work on nutrition and intake. All of these components are presented as objectives on the assumption that they will proceed simultaneously under a single experimental design. Objectives 1. Determine seasonal botanical composition of diets in relation to kinds and amounts of available forage. 2. Determine daily forage intake as related to animal size and physiological status (i.e., maintenance, gestation, lactation) and to kind and amounts of available forage. 3. Determine nutritional value of the diet as related to animals' nutritional demands and as affected by season and kinds and amounts of available forage. Major attention should be given to

ll3 dietary nitrogen, phosphorus, digestible energy content, and fiber components according to the system of Van Soest (l967). Methodology l. Objectives in this study can be met by using domestic horses and tractable livestock (cows or sheep) as experimental subjects. 2. Diets should be determined by a technique other than fecal analysis. Use of animals fitted with esophageal fistulae and cannulae, with subsequent microscopic analysis of fistular extrusa (reviewed by Theurer and others l976) is a valid technique that is probably feasible under the constraints of the present design. Alternatively, a bite-count technique (Wallmo and Neff l970) or a suitable modification thereof would be appropriate if carefully controlled and validated with a proven procedure, such as the fistulated animal technique. These specifications do not preclude the use of the fecal-analysis technique. Indeed, this study would provide an opportunity to validate it under field conditions. However, it should not be relied upon as the sole procedure, since its accuracy and precision are presently subject to question (Smith and Shandruck l979, Vavra and others l978). 3. Diets should be sampled on at least a monthly basis during times when animals inhabit the particular range being studied. Detailed information on forage available (such as that derived from Project 8, the "grazing impacts" study) should be obtained in conjunction with dietary sampling. 4. Nutritional value of diets should also be studied on a monthly basis, and such studies could be linked with diet-analysis studies, as outlined under objective l above. Laboratory analysis of samples obtained either from animals (via fistulae) or by hand plucking should include as a minimum: dry matter, nitrogen (crude protein), phosphorus, gross energy, cell walls, cellular constituents, and lignin. 5. In addition, estimates of forage digestibility should be obtained for ruminant species grazing with horses. In-vitro techniques are recommended, preferably those of Tilley and Terry (l963). However, since in-vitro procedures have apparently not been perfected for equid species, a limited number of in-vivo digestion trials with horses will be required. Such trials should employ classical techniques in which penned animals are fed controlled quantities and total feces are collected. The major forage species constituting the seasonal diets of horses should be fed in mixtures approximating the proportions normally ingested by the animals (as determined under objective l, above). This approach is limited by the logistics of harvesting suitable quantities of native forages for feeding trials. However, alternative methods such as the lignin-ratio method appear undependable. These trials will offer the opportunity for development of in-vitro procedures directly applicable to equid species. This technique development should be pursued, as the

ll4 potential payoff is large in relation to the small marginal investment of research time and funds. 6. Forage intake by the grazing animal should be measured by the procedure of Garrigus and Rusk (l939), described by the equation: Dry matter = (Fecal output - % dry matter) x l00 intake l00 digestibility 7. Fecal output should be measured using total collection procedures with animals wearing fecal collection bags. Alternatively, indicators of indigestibility (e.g., Cr2C>3 or appropriate rare-earth compounds) can be used to estimate fecal output if adequately validated and quantitated. 8. Intake studies on horses should focus on the reproductive female and the field experiments should be phased so that they cover the three major physiological phases: maintenance, gestation (especially the last trimester of pregnancy), and lactation. Project 3. Nutritional Plane, Condition Measures, and Reproductive Performance in Domestic Mares Rationale The nutritional condition of the female is considered to be of primary importance to estrus, ovulation, and early survival of the offspring from conception to weaning. The rate at which offspring are produced is a critical demographic parameter when estimating rates of population increase. If levels of energy and nitrogen intake can be successfully linked with reproductive success, and these levels of response can be predicted by measures of condition of the animal, then the ability to estimate reproductive rate from range condition or nutritional condition of the animals is greatly enhanced. Objectives l. Determine the relationship between the nutritional plane of a mare and the probability that she will produce a foal. 2. Assess condition measures for their ability to predict the relationship between nutrition level and reproductive success. Should some condition measures prove to be successful predictors of reproductive rate, they could be used to estimate these rates in wild populations. Methodology l. Confine groups of l0 domestic mares for 2 years (or l5 to 20 if conducted in l year) in each of the nine blocks to be described

ll5 below. Do not use ponies. Ideally, the study should run for 2 years, since the given nutritional plane may have a cumulative effect. Reproduction in one year often reflects nutrition from a previous year, although acute nutritional deprivation can delay onset of estrus or perhaps inhibit it completely if there is no improvement in food supply. The mares should range in age from 4 to l0 years. The age of onset of estrus is a separate question. 2. The nine feeding blocks are all the possible combinations of three levels of energy and three levels of nitrogen intake. This nine-block design can be achieved in several ways: a. Levels are set in absolute intake values for energy and nitrogen and not varied throughout the experiment. This method is the simplest to conduct. The constant feeding system has the advantage that a particular level of energy and nitrogen intake can be related directly to success in a particular stage of reproduction. b. Three levels of energy are fed, simulating various winter conditions. After "winter," each of these three groups is divided into three groups and fed at high, medium, and low spring conditions. This design examines the nine possible conditions of a poor-, medium-, or high-quality winter being followed by a poor-, medium-, or high-quality spring. The advantage of this design is that it simulates the real sequence of events in wild populations more accurately than would a constant ration. However, the experiment will contain 8l cells if the design includes nitrogen as well as energy, and hence is too large to accomplish with a reasonable effort. c. Nutritional levels are fed as some percentage of the NRC requirements for maintenance, gestation, and lactation. In this case nutritional intake at a particular level would increase during gestation and lactation according to some percentage of the increment suggested by the NRC. The difficult aspect is to establish the levels for feeding the nutrients. Investigators might choose to employ conditions existing on very good and very poor ranges and to include an intermediate level. d. Nutritional levels are set as a percentage of dry weight and horses are fed ad libitum. This technique allows the horse to respond naturally to poor food by increasing its intake. The percentage of dry matter as a measure of nutritional level is more easily translated into range conditions than other measures used in the three previous designs. The horses will not be on a constant plane of nutrition and the establishment of high, medium, and low nutritional levels will be difficult. However, some preliminary feeding trials might establish the capability. 3. The following reproductive parameters should be measured to establish quantitative relationships between nutritional plane and reproductive success, which may then be used for demographic models of specific ranges:

ll6 a. Estrus (onset and behavior) b. Ovulation rate (rectal palpation) c. Number of estrus cycles before pregnancy (behavior and progesterone), extended cycles (prolonged corpora-lutea life). d. Conception e. Implantation rate (rectal palpation, PMSG, progesterone) f. Embryonic loss (PMSG, progesterone, urinary estrogens) g. Foaling rate (count live abortions and stillbirths) h. Length of gestation (dates of breeding) i. Parturition (independent or requiring assistance, duration of birth) j. Weight of foal at birth k. Lactation behavior (ability to nurse, acceptance by mare) l. Nutritional composition of milk (colostrum at 2 and 6 weeks, protein, fat, carbohydrate) m. Milk production (probably difficult to quantify) n. Foal growth and behavior (weekly weight, withers height, etc., and simple estimate of activity level). 4. Condition measures should be made every 2 weeks in order to establish measures that correlate with empirical intake data, general condition measurements, reproductive parameters, and seasonal variation. These standards would allow evaluation of particular wild herds and assist demographic projections of population changes. Recommended tests are: a. Hematology (CBC) b. Chemistry (serum urea nitrogen, glucose, triglycerides, serum protein and albumin, uric acid, bilirubin, free fatty acids (NEFA), CPK, ketones, haptoglobin, transferrin) c. Hair bulb diameter. 5. Behavioral sampling should be performed for the following purposes: determining activity levels so that nutritional levels can be accurately generalized to free-roaming populations, whose activity levels are likely to be different; correlating differences in reproductive performance and condition to activity measures; and comparing nutritional treatment effects on activity. a. Specific behavioral sampling: When females in the experiment are mated, behavioral samples identical to those used in the contraceptive experiment (Project l8) will be recorded. These data will show how the nutritional plane affects the sequence of sexual behaviors between the stallion and the mare. b. Focal animals: The groups for this experiment will be composed of mares from each nutritional group and an experienced stallion.

ll7 c. Sampling schedule: The sampling schedule should take place primarily during the breeding season so that good activity profiles are available for different phases of the breeding cycle. d. Specific behavioral samples: Behavior of the animals is important to the determination and function of their social groups and to successful breeding. Since nutrition affects the reproductive success of the animal, it will be essential to measure the differences in behavior between nutritional groups to determine the sensitive points in the reproductive cycles. The effect of behavioral changes in individuals on the group's social dynamics might be monitored and the ecological and management implications of these effects can be suggested. e. Sampling scheme: Besides the activity and frequency of social behavior sampled by FSP (see Project 6) method, the reproductive behavior needs to be sampled in detail. Reproductive behavior can be divided into functional categories. These divisions are established because breakdown in the reproductive act can be attributed either to the mare or the stallion and the point at which failure occurs can be assigned to a particular phase of the reproductive interaction. The objective of this sample will be to monitor behavior of mares of different nutritional planes and that of their stallions to determine differences in each of these functional categories. Differences can then be related to failure of behaviors at these stages of reproductive sequence. Continuous focal samples should be taken. The timing of the samples and their intensity can be determined by the researcher. The continuous samples will record the occurrence of sexual behaviors upon a time record. From these samples the rates of behavior can be calculated and tested against estimates for wild populations. Spacing data will provide information on the distance maintained between individuals in the groups. Such data can be taken at a point in time where the distances and identity of the individuals are recorded. These samples would be taken at the beginning and end of all half-hour FSP samples. These data will allow a comparison not only of sexual states but also of dominance rank. Spatial displacements, usually a reliable indicator of relative dominance between individuals, will be recorded as part of the continuous focal samples. Project 4. Blood Assay of Experimental Equids and Livestock in Projects 1, 2. 3. 5. and 8. in Projects l, 2, 3, 5, and 8. Rationale It has been suggested above that various blood characteristics might be used to provide sensitive and easily obtained indices of nutritional plane. These could, on the one hand, be used as predictors of the animals' demographic performance, and on the other

ll8 hand, as indicators of the range condition. This project recommends routine blood sampling and assay on experimental animals described in Projects l, 2, 3, 5, and 8 to test the utility of these measures in describing ecological and nutritional characteristics. Objectives l. Systematically collect blood samples during habitat- interaction studies from cattle and horses for hematology and chemistry assays to provide independent evaluation of animal condition and to establish correlates with range-condition evaluations. 2. Systematically collect blood samples during range-nutrition studies and assay to establish correlates with condition and known levels of food intake and utilization. 3. Systematically collect blood samples from mares in the reproductive-performance study to establish correlates with condition, reproductive success, and mechanisms of reproductive suppression. Methodology l. At each sampling, extract 50 ml of blood from the jugular vein of animals that are roped and pole-restrained, immobilized in chutes, or drugged. The procedures are well-known and are commented upon by Kirkpatrick and others (l979a) in his article on restraint procedures for evaluation of stress. If drugs are used, they must be identified and specified, since the various immobilization drugs have an impact upon laboratory studies (excluding genetic studies), In particular, xylazine (Rompun*, Chemgro Corp.) should be avoided if the clinical metabolic studies for evaluation of condition are to be performed, since its administration results in a steady increase in blood urea and blood glucose for a period of 4 to 6 hours. It also would be desirable to collect the samples from the animals as soon after capture and handling as possible, and certainly before they are transported to more permanent holding facilities and fed.. 2. Samples should be preserved until assay by a single laboratory to minimize interlaboratory variation. Procedures should be established to document quality control, including reproducibility and accuracy. A high degree of laboratory precision can be achieved in each of the assays listed below and excellent procedures and equipment are available for this purpose. 3. Approximately 2 ml of the sample should be drawn into a tube containing EDTA (ethylenediaminotetraacetate) as an anticoagulant for hematology, 5 ml into heparin for preparation of red cells and plasma for molecular genetic studies, 5 ml into Alsevier's solution for blood group typing, and the remainder into untreated tubes for preparation of serum. 4. The following blood assays should be evaluated for correlation with measures of animal nutritional condition, evaluation of range condition, measures of reproductive success, and estimation of the

ll9 presence of disease: serum urea, hematocrit and hemoglobin concentration, serum protein concentration, haptoglobin, creatine phosphokinase, triglycerides, uric acid, bilirubin, free fatty acids and ketones, white cell count and differential, transferrin, and various special clinical studies (serology). Project 5. Demography of Wild Horses and Burros Rationale It is clear from the discussion on demography in the first part of this chapter that the particulars of this aspect of wild horse and burro biology are not well known. We have seen that the basic demographic parameters of (a) age-specific fecundity, (b) first-year survival rates, (c) adult survival rates, (d) age structure, and (e) sex structure are known only by the roughest approximation, and are often based on extremely small samples that make for an equally low precision of estimate. Even less well-known are the ranges of variation in these parameters and the role of such environmental variables as weather/climate, forage condition, nutrition, competition with other herbivores, and population density in producing this variation. BLM officials have frequently accorded demography a high priority among the research needs on horses and burros. While we have not asked the reasons for this ranking, we surmise that the research is desired to dispel the disagreement over BLM's stated rates of population increase, and in general to resolve the question of population growth. The relevance of demographic understanding to an equid management program depends on the level of management. At its least intensive, such understanding is not essential. Once a decision is made for a given tract of land on how much forage is to be allocated to equids, and on the population level that this implies, the population can simply be held at this level. Assuming that the census method is valid, the population can be counted annually, or at some other interval. As its numbers exceed the desired level, the excess can be removed periodically irrespective of what the increase rate may be. Another situation in which there is no need to know increase rates is if some populations are allowed to limit themselves, irrespective of their impacts on the ecosystem. At a slightly more sophisticated management level, if the approximate rates of increase are known, then needed removal rates can be anticipated. Reduction operations can be planned well in advance. A considerably more knowledgeable program would incorporate a thorough understanding of equid biology and of the relationships between environmental variables and demographic performance. The fact that performance would probably vary from herd to herd, and from year to year—as well as the causes of the variation—would be understood and anticipated. Such a program would result in a more insightful and

l20 sympathetic understanding of equid management. It would also have some predictive capability. The Committee is divided on its own assessment of priority for demographic research. Those who accord it low priority do so on pragmatic grounds. A limited research commitment, in this group's view, would not add materially to what is now known. To refine by a few percentage points the survivorship estimates, or sharpen somewhat the fecundity parameters, would not alter the basic picture of increase rates developed in the earlier demography section. On the other hand, the mounting of a research program sufficient to delineate precisely the range of variation in these parameters that occurs between areas and years—as well as the functional relationships between that range and associated environmental causes—would entail a cost beyond the resources likely to be available for such research. Those who assign high priority to demographic research contend that the parameters are too poorly known for workers to have any confidence in the present level of understanding of eguid demography and in the various calculations of herd increase. Any increase in precision over what is now known would be a gain and worth the cost. Among the parameters to be measured, there would appear to be some grounds for assigning priorities on the basis of (a) cost, (b) present level of understanding, and (c) the extent to which more information would contribute to the present level of understanding. Evaluations of these three measures are presented below for each parameter. l. Age-specific fecundity a. Least costly. Could be elucidated as outlined under "Methodology," below. b. Inadequately known at present. Because it may be one of the parameters more sensitive to environmental factors, it may therefore be one of the more variable parameters, especially in burros. c. An increased data base would refine the present estimates and could give some understanding of the range of variation and its relationship to environmental conditions. 2. First-year survival a. Extremely costly, especially if designed to provide a good understanding of the range of variation and associated environmental causation. b. Perhaps the least understood parameter at present, and one that is likely to be highly sensitive to environmental variation, especially in burros, and therefore highly variable within and between populations. c. The value of an increased data base would depend on the scope of the research. Just to measure first-year survival for l or 2 years in a few populations would not add substantially to what is now known. We currently have a range of scattered, first-year survival estimates, crude though they may be. The next higher level of understanding will come when we have measured the range within each of an array of populations, along with

l2l measurements of the relevant environmental variables. Such a study would involve a commitment to several years of research in each of a number of populations, and would ultimately allow formulation of the functional relationships involved. 3. Adult survival. a. Costly. b. Poorly known, but one that is probably least sensitive to environmental variation and therefore least variable within and between populations. c. An increased data base would refine the present estimates, but would, in the opinions of some (but not all) members of the Committee, not add a great deal to present levels of understanding. Objectives The objectives of this project are to measure age-specific fecundity rates, first-year survival rates, and adult survival rates to two different levels of precision. The more definitive level will, of course, be the more costly. Level l: Obtain an array of measurements more precise than those currently available from a number of different populations. Level 2: Obtain a series of measurements in each of a number of populations to determine the range of variation in each, and to correlate those with relevant environmental variables. Methodologies for each of the three rates are described for Levels l and 2 in the following section. Methodology l. Age-Specific Fecundity Rates, Level l: Determine pregnancy and corpus-luteum rates by rectal palpation of horses and burros captured in the course of herd-reduction efforts, or by autopsy of burros shot on national parks and monuments. Observations would need to be made in late winter or early spring in the case of horses and seasonally breeding burro populations, and perhaps again in the fall in burro populations breeding year-round. 2. Age-Specific Fecundity Rates, Level 2: Ideally, the range of variation between years—within and between populations—would be determined by rounding up the individuals within studied populations each year, palpating them, and releasing them. Level of precision—i.e., the confidence intervals around the estimates—would be decided upon before sampling, and this factor, along with prior measures of variability of the measured parameters, would dictate the sample sizes needed. Concurrently, weather records would be kept; the condition of the ranges occupied by the herds would be measured annually; blood assays (Project 3) would be taken of the sampled

l22 animals; and the total populations would be censused to provide measures of population density. Numbers of potentially competing ungulates should be determined, along with total forage available and its allocation to all herbivores present. Properly, each herd should be studied in this way for not less than 5 or 6 years, and preferably l0. If this ideal is not possible, then some level intermediate between it and Level l could be achieved by recording range condition, weather pattern in the l0 to l2 months prior to sampling, and population density for the locales in which each roundup providing animals for Level l studies is carried out. In addition, blood assays should be taken from the animals that are palpated and from males and young females in the herds. This collective information could not only give fecundity estimates that are more precise than those currently available, but could also provide background environmental information with which to relate them. 3. First-year survival rates, Level l: Determine first-year survival rates by: a. Rounding up samples of animals from within populations; b. Palpating females in study design (l), above; c. Placing radio transmitters on pregnant animals; d. Releasing the telemetered mares, locating them periodically from the ground or air during the year after birth of their young, and determining which ones have foals surviving through the year; and e. Predetermining sample sizes by deciding on the desired level of precision beforehand. A scattering of l- or 2-year studies in different populations should give some idea of the range of values for this parameter, and provide more precise estimates than are now available. 4. First-year survival rates, Level 2: Determine the range of variation within and between populations in first-year survival rates and the environmental causes of that variation by: a. Measuring first-year survival rates as in study design (3) each year in each of several populations for at least 5 to 6 years, and preferably l0. b. Recording each year the condition of the home range of each population; maintaining weather records; taking blood assays of the studied animals as they are caught, palpated, and telemetered; and censusing each population annually. Determine numbers of potentially competing ungulates, total forage available, and its allocation to all herbivores present. 5. Adult survival rates, Level l: Determine adult survival rates by: a. Rounding up samples of animals from within populations. b. Placing radio transmitters on pregnant animals.

l23 c. Releasing the telemetered animals, locating them periodically from the ground or air during the year, and determining which ones are still alive. d. Predetermining samples sizes by deciding on the desired level of precision beforehand. A scattering of l- or 2-year studies in different populations should give some idea of the range of values for this parameter, and provide more precise estimates than are now available. 6. Adult survival rates, Level 2: Determine the range of variation within and between populations in adult survival rates and the environmental causes of that variation by: a. Measuring adult survival rates as in study design (5) each year in each of several populations for at least 5 to 6 years, and preferably l0. b. Recording each year the condition of the home range of each population; maintaining weather records; taking blood assays of studied animals as they are caught, palpated, and telemetered; and censusing each population annually. Determine numbers of potentially competing ungulates present, total forage available, and its allocation to all herbivores present. c. Predetermining sample sizes by deciding on the desired level of precision beforehand. 7. Adult survival rates, a third approach: A great deal of age-structure data exists from the roundups. As shown in the demography discussion in the first part of this chapter, survivorship can be calculated from such data, although the method contains certain biases and demands large samples. A knowledge of population trend is also vital to its use. The available data should be explored further in populations that are censused annually, and ideally in those being studied by methodologies 5 and 6, so that results of the latter studies can be used as a check. Project 6. Social Structure, Feeding Ecology, and Population Dynamics of Wild and Free-Roaming Horses and Burros Rationale Data on demographic patterns, feeding ecology, and population dynamics are important to determining what constitutes a healthy and viable breeding population of wild equids that will also permit maintenance of vegetation, soil, and water resources. Social structure, distribution and movement, and breeding patterns are particularly important to interpreting data on genetic variability and the potential effect of different cropping schemes on the genetic structure and viability of a population. Since activity levels determine part of nutritional requirements, enclosure studies will be insufficient for determining habitat preference and use, grazing

l24 impacts on range-plant communities, food-consumption rates and nutrition, and impact on riparian water quality of wild and free-roaming horses and burros. In order to relate the findings on nutritional condition in the enclosure animals to those on the open range it will be essential to compare the level of activity between study groups, so as to make realistic generalizations concerning the feeding and nutrition experiments. If one is to understand the social effects of contraception in equids (see Project l8) it is necessary to have comparable measures of the behavior of experimental animals and wild animals. If there is a lack of comparability between studies, then the ability of researchers to generalize their experimental findings to wild populations will be severely limited. A major point of controversy is the rate of increase in horses and burros. It is extremely important to document and integrate in free-roaming populations (a) population size and structure in terms of sex, age class, and social grouping; (b) daily activity pattern, feeding ecology, and energetics; (c) seasonal movements and home-range utilization; (d) nutritional level; and (e) age-specific reproduction and mortality. Objectives l. Determine the social structure, seasonal habitat preferences, and vegetation and water utilization of free-roaming equids on public lands. 2. Determine the daily activity pattern and dietary composition in relation to age, sex, physiological status (i.e., maintenance, gestation, lactation), and climate. 3. Assess nutritional status and general condition, and correlate these with activity, reproduction, and survival. 4. Determine age-specifc reproduction and survival. 5. Investigate the applicability of contraception as a means of managing wild equid populations. Methodology l. Establish three study areas each for wild horses and burros. In order to have the broadest possible application, research should be performed on populations that are not manipulated (i.e., cropped). One study area should be located in what is considered the most "marginal" habitat. Another area should be located in typical habitat, but be used to investigate the effect of contraceptive devices on social behavior, spacing, feeding ecology, and populaton dynamics of free-roaming equids. 2. For determining daily activity patterns and energetics, we recommend that at least the following data be recorded. Recorded information is to reflect activity states that provide data for calculating both energy budgets and social behaviors characterizing

l25 the rates and directions of social interactions (i.e., who does what to whom). The following categories are to be scored: standing, standing eyes closed, lying down, walking, trotting, running, feeding, urinating, defecating, grooming, and social interactions. The term social interactions covers a multitude of what are often species-specific behaviors and which will require precise and common definitions across experiments. Researchers involved in these studies should be required to standardize these definitions after initial work has begun. Point samples on these categories should be taken for 30 minutes each hour from 0600 to l800 (light permitting), once every 2 weeks. Points are recorded every 2 minutes within the 30-minute sample. Two minutes has been chosen as being a sufficiently long period for a trained observer to record behaviors and identify individuals for a scan point sample of up to 5 or 6 focal animals, while providing a reasonable number of sample points. At the beginning and end of each 30-minute sample, interindividual spacing, location, habitat, and climatic data are to be recorded. In horses, focal animals should include at least one dominant male, two females, one subordinate male, and one foal. Ideally, in horses, this group of focal-scan-point samples should be done on the same day, to provide daily profiles and to allow the determination of interdependence of behaviors within the day. For burros, whose groups are not as spatially coherent, samples cannot be simultaneous on different classes of animals. In burro studies, animals should include one territorial male, one nonterritorial male, two females, and one foal. In the contraception study (see below) one of the two females should be "contraceptive" and the other "normal." 3. The sampling scheme for monitoring detailed feeding behavior is to be a continuous focal-animal sample. Data are to be recorded on an animal in a continuous timed sequence. The schedule for sampling is identical to the FSP schedule. Data are recorded on one individual for 30-minute periods of each hour from 0600 to l800. In these samples the onset, termination, and sequence of all behaviors (categories recorded for the FSP and reproductive behavior samples) are recorded on a real-timed record. The identity and size of all plants and plant parts eaten are recorded. This detailed record is necessary to understand the basis of dietary selection in the horse and to measure changes in feeding response when inter- and intraspecific competition occurs. The detailed data on food intake are also necessary to estimate energy and nutrient balances in animals whose fecal output is collected. For both of these calculations it is necessary to know not only what is eaten, but to know specifically the rates at which forage species are eaten, the quantities ingested as a function of time, the sequence of ingestion through the day, and the diurnal patterns of gross intake rates. Spacing data should be recorded at the beginning and end of each 30-minute sample. These data, coupled with displacement data collected as part of the continuous record, will allow dominance interactions to be related to feeding behavior.

l26 Location, habitat, and climatic data should be recorded at the end of each 30-minute sample. Data should be recorded on the same focal classes of animals (N = 5) as were used for the scan-point sample. 4. Besides the activity and frequency of social behavior sampled by the FSP method, the reproductive behavior should be sampled in detail. Reproductive behavior can be divided into functional categories. These divisions are established because breakdown in the reproductive act can be attributed to either the mare or the stallion and the point at which the failure occurred can be assigned to a particular phase of the reproductive interaction, preceptive behaviors are those that the female uses to make her estrous condition known to other animals. These behaviors include raising the tail, urinating at high frequency, urination stance, and approaching the male or altering her spacing relative to him. The male's responses to preceptive behaviors are called attraction behaviors. His approaches, spacing relative to the female, and mounting attempts signal his perception of her state. Her response to the male's attraction behavior is termed reception and is evinced in either acceptance or rejection of mounting attempts. The object of the sample will be to monitor behaviors of "normal" and "contraceptive" females and their stallions to determine differences in each of these functional categories. Differences can then be related to failure of behaviors at particular stages of the reproductive sequence. Samples should be continuous focal samples. The timing of the sample and its intensity can be determined by the researcher. The continuous samples will record the occurrence of sexual behaviors with a time record. From these samples the rates of behavior can be calculated and tested against estimates for wild populations. Spacing observations give data on the distance between individuals of the groups. Such data can be taken at a point in time where the distances and the identity of the individuals is recorded. The samples should be taken at least at the beginning and end of all 30-minute FSP samples. Since spatial relationships often reflect social relationship, these data will allow a comparison not only of sexual states but also of dominance rank. Spatial displacements, usually a reliable indicator of relative dominance between individuals, will be recorded as part of the continuous focal samples. 5. Population size and sex and age-class structure should be monitored monthly. Age-specific reproduction and survival should also be monitored. Project 7. Genetic Polymorphism Rationale Rational management of natural populations requires knowledge of the genetic structure of the population. In particular, it is necessary to ascertain the degree of genetic variation within and between populations. The amount of genetic variation in a population

l27 determines its "evolutionary potential," and hence must be preserved as the fundamental natural genetic reservoir of the species. The process of mutation may give rise to new genetic variations and may restore variants that have died out. However, the process of mutation is exceedingly slow—in vertebrates the typical rate at which a given allele at a given locus appears per generation is around one in a million individuals—and most new mutants are likely to be lost by random drift in the first few generations after their appearance even if they are adaptively favorable. Hence, it becomes important to preserve genetic variation in order to protect the welfare and evolutionary potential of the species. The genetic variation of a group of populations may be present more or less equally in the various populations or each may have a different amount. Which of these situations obtains in a particular case must be ascertained, because each one leads to different strategies for the preservation of genetic variation. If several populations are nearly identical genetically, the preservation of any one population (subpopulation) is less important, because its genetic endowment will also be present elsewhere. On the contrary, if populations are genetically different from each other, the preservation of genetic variation requires that every one of the differentiated populations be preserved. In the case of wild and free-roaming horses and burros, it becomes important to study not only the wild populations but also the domestic breeds. For the reasons just mentioned, it must be ascertained whether or not the variation present in the wild populations is the same as the variation present in domestic animals. The comparison between domestic and wild populations may also make it possible to determine the ancestry of the wild populations. None of the required information mentioned is presently available. Objectives l. Determine the amount and kind of genetic variation per population of wild and free-roaming horses and burros. 2. Determine the degree of genetic similarity among wild and free-roaming populations of horses and of burros. 3. Determine the degree of genetic similarity between wild populations and domestic breeds. Methodology l. Population Sampling. The information gained with a given amount of work is maximized if the populations are sampled in a "nested" fashion. The appropriate model to be followed should be coordinated with other projects, because blood (and tissue, if possible) samples from individual animals are needed. Ideally, one breeding wild group should be intensively studied (a sample of 50 to l00 animals might be sufficient, depending on the size of the group).

l28 Two or three other breeding groups should be sampled within a given region, but fewer animals (l0 to 20) would be needed from each group. In addition, one group from each of the other major regions (l0 to 20 animals per group) in which the wild animals range should also be studied. This pattern of sampling should be followed for horses as well as for burros. It will then be necessary to compare wild animals with domestic ones. About l0 animals for each of the main domestic breeds should suffice, although of course the more animals that are studied, the greater will be the amount of information obtained. The point is, however, that more information will be obtained by sampling a few animals from each of a number of breeds (or populations) than by studying a larger number of animals from fewer groups. 2. Genome Sampling. In order to obtain a valid estimate of the amount of genetic variation in a population, it is necessary to study a number of gene loci that are randomly selected with respect to how variable they are. That is, it is necessary that the probability of including a given locus in the sample be independent of how polymorphic the locus is. Morphological traits (such as coat color) are generally useless for this purpose. The reason is that the genetics of morphological traits is determined empirically by traditional Mendelian methods; that is, by making crosses between individuals that manifest different phenotypes with respect to the trait. The pattern of segregation in the Fz progeny allows one to determine whether or not genetic differences underlie the morphological differences, as well as to ascertain the number of genes involved. Thus, the only genes that can be studied with the traditional methods of Mendelian analysis are polymorphic genes; if a gene is invariant, such methods will not reveal it existence. It follows that use of such methods makes it impossible to obtain a random (with respect to variation) sample of the genome, because only polymorphic genes can be included in the sample (and usually the more variable a gene is, the more likely it is to be included in the sample because its existence is more easily detectable). The methodological handicap derived from the traditional Mendelian method of genetic analysis also applies to blood-group studies. As in the case of morphological variation, the study of blood groups depends on the detection of differences. Obtaining a sample of genes that is random with respect to variation has become possible owing to the advances of molecular genetics. It is now known that genes consist of DNA and that (structural) genes code for proteins. It is possible to select for study a number of proteins, without having previous knowledge as to whether each protein is variable or how variable it is. If a given protein is found to be polymorphic, it can be inferred that the gene coding for it is also polymorphic. If a protein is found to be invariant, the inference is that the corresponding gene is also invariant. (Actually owing to the redundancy of the genetic code, not all gene variations result in protein differences; hence, the overall amount of variation is underestimated.)

l29 The techniques of gel electrophoresis make it possible to study variation in a large number of proteins in a sufficient number of individuals without prohibitive investment of labor and money. Experience indicates that about 20 genes coding for proteins are sufficient to obtain acceptable estimates of genetic variation (Ayala and Kiger l980), although 30 or more genes should be studied, if feasible. 3. Statistical Techniques. The amount of genetic variation is best measured by the heterozygosity per locus or per individual, although the proportion of polymorphic loci per population also provides useful information. The degree of genetic similarity can be summarized with standard measures, such as Nei's genetic distance. The formulae with which to calculate heterozygosity, polymorphism, and genetic distance and their significance can be found in recent textbooks (e.g., Ayala and Kiger l980). 4. Biochemical Techniques. The techniques of gel electrophoresis and selective assay of enzymes are the most appropriate and least expensive with which to accomplish this project's objectives. Description of these techniques can be found in Ayala and others (l972). Information about a few specific enzyme assays as used in equids can be found in the references cited within "Genetic Polymorphism," in the first part of this chapter. Blood samples (a few cc per individual) can be used for the electrophoretic studies, although tissue samples (liver, in particular) should be used whenever available, because these make it possible to assay additional enzymes.