National Academies Press: OpenBook

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

Chapter: WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY

« Previous: SOCIOECONOMIC AND POLITICAL ISSUES
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 190
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 191
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 192
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 193
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 194
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 195
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 196
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 197
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 198
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 199
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 200
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 201
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 202
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 203
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 204
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 205
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 206
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 207
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 208
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 209
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 210
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 211
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 212
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 213
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 214
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 215
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 216
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 217
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 218
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 219
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 220
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 221
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 222
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 223
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 224
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 225
Suggested Citation:"WILD EQUID RESEARCH AND MANAGEMENT METHODOLOGY." National Research Council. 1980. Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research.. Washington, DC: The National Academies Press. doi: 10.17226/18642.
×
Page 226

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 5 RESEARCH AND MANAGEMENT METHODOLOGY FOR WILD EQUIPS INFORMATION NEEDS Chapters 2, 3, and 4 of this report have explored what is known about the biology of horses and burros, their interrelationships with other ecosystem components, and the social and economic values involved in their management. These sections have also outlined research needed to expand knowledge in these three areas. But little has been said so far about actual management methods, or about some of the technical problems that will be encountered in the prescribed research. This chapter addresses some of these methodological problems. A reliable census method is obviously indispensable to a sound management program, and the existing censuses have been criticized by a number of observers. Hence, considerable space is devoted to this tool. Similar scrutiny is given to the commonly used range-survey methodology, inasmuch as a knowledge of range condition is essential to understanding both equid welfare and that of the other ecosystem components. PL 95-5l4 also raises the possibility of "sterilization" in addition to natural control and removal for maintaining equids at desirable population levels. Hence, some consideration is given to the matter of hormonal contraception in the mare. Since horses and burros must be restrained in the course of both management and research efforts, the techniques of chemical immobilization and capture are reviewed. A variety of research techniques is also considered. Alternate techniques of diet analysis as well as some of the unique problems of studying equid nutrition are discussed in this chapter and in Appendix A. Sampling considerations in behavioral studies, and the assay of blood chemistry to provide indices of nutritional state, both for research and management purposes, are explored at some length. STATE OF KNOWLEDGE Information Sources In addition to reviewing the published and unpublished sources described in previous chapters, Committee member Eberhardt visited l90

l9l four BLM districts (Burns, Vale, Lakeview, and Challis) for the purpose of gaining preliminary familiarity with census procedures. He spent 2 hours in the air in one of these districts and considerable time interviewing people on the ground in all four. As a result of these efforts, he has acquired 300 to 400 photographs, and a considerable mass of BLM census data from numerous districts. This document file (currently at New Mexico State University) remains to be cataloged and reduced. Committee chairman Wagner conducted preliminary analyses on the wild-horse census data from 72 BLM districts and national forests covering periods of l to ll years (see Table 2.ll). He detected several biases and calculated preliminary estimates of their magnitude. It was from these data that the preliminary calculations on herd increase rates discussed under "Equid Demography" (Chapter 2) were made. The state of knowledge concerning census is reviewed in the following section. But the data need further analysis, which should be carried out as part of the census research program. Census Potential Methods Animal population measurements fall into one of three basic categories: (l) indices, (2) complete counts, and (3) estimates based on sampling approaches of various kinds. With the exception of a few small-scale research studies, most current and recent efforts to census wild horses and burros have involved some form of aerial counting, and complete counts have usually been attempted. Doubtless this choice has evolved from considerations of cost-effectiveness, topography, and manpower limitations. There are almost no published data on the accuracy of aerial counts of wild horses and burros, and very little data on the variability to be expected in repeated surveys of the same areas. It is essential to obtain such information before making a final decision on a particular method, and in order to do so, aerial surveys for wild horses and burros must be researched. In this review we have assumed that the principal method to be considered is aerial counting, and other potential methods will be touched upon only briefly. l. Indices. Many wildlife problems have been approached by measuring relative abundance, i.e., using an index of abundance. Indices generally do not provide estimates or counts of the actual numbers of animals in an area, but rather give relative measures of numbers. All indices depend on one universal assumption: that the index values are some constant function of, or bear some constant relationship to, actual population size. Where this assumption holds, they are valuable for estimating rates of population change over time, and often can be used in estimating mortality and reproductive rates. They are also frequently useful in comparing relative densities between areas.

l92 It is problematical whether such index methods as roadside counts, track counts, and fecal counts have utility for wild horses and burros. Roadside counts might simply be tallies kept in the course of travel on other business or made on selected routes at prescribed times. In suitable areas, counts at watering sites could conceivably be effective. Track counts might be combined with visual counts, but would probably be most effective if made on fresh snow or after a heavy rain. Fecal counts have not been used much for equids. They have been extensively employed for cervids, particularly for white-tailed deer and elk. Correction factors are available to convert the counts to estimates of absolute abundance. Similar factors might be obtainable for eguids. Development of an effective index method would require that a designed study be combined with censuses conducted by proven methods. Since any management program involving forage allocation to a specified number of equids would require an estimate of absolute numbers of horses and burros, an index would not in itself be suitable. It would have to be "calibrated" somehow so as to provide an estimate of absolute numbers. Actually, it may be appropriate to consider many of the present aerial-survey estimates as indices, as will be discussed below. It is for this reason that the subject of indices has been mentioned briefly here, and because of the tentative inferences on horse-population trends drawn earlier in this report from the aerial censuses. 2. Complete Counts. The present BLM surveys are almost exclusively attempts at complete aerial counts of all animals on a given area. In fairly open areas with low vegetation, a skillful observer working with a cooperative pilot may well be able to tally nearly all of the animals present. Scientific proof of such an assertion depends on some sort of independent check, presumably best obtained by a marking program. In areas of dissected topography and heavy cover, it is quite likely that some proportion of the animals present are missed. Caughley (l974) has assembled data from various surveys of different species indicating that appreciable fractions of the animals present are not seen from the air. A great deal of experience with African "game" has led workers there to use rather narrow strips for aerial counts, and to depend on sampling rather than attempting complete counts (cf. Sinclair l972, Goddard l967). Since much of that experience comes from environments quite similar in topography and vegetative cover to many horse and burro ranges, it is appropriate to give careful consideration to the techniques thus developed. There is also evidence that such factors as airspeed, altitude, width of strip searched, type of aircraft, and observer experience have important effects on aerial counts. These factors are described by Erickson and Siniff (l963), Pennycuick and Western (l972), LeResche and Rausch (l974), Norton-Griffiths (l975, l976), Caughley and others (l976), and Frei and others (l979). The major problem with attempting complete counts is to devise tests of their accuracy and correction factors for deviations from

l93 absolute accuracy. One such test is the "bounded-counts" method. It requires a number of successive counts, made under such conditions that there is some positive probability of counting all animals present in an area in at least one of the counts. The two highest counts are then used to estimate the actual population. The method is based on theory developed by Robson and Whitlock (l964), and was suggested by Regier and Robson (l967). Overton and Davis (l969) provide some further discussion. The bounded-counts method depends on the ability to conduct repeated independent counts of the same population without tallying any animal twice on a given count. The critical assumption in the method is that conditions will permit a definite (although possibly very small) probability that all animals in a population are counted. No technique for checking it seems to have been suggested as yet. As an illustration of the kind of situation in which the method may fail, we have used data on l8 successive aerial counts of black rhinoceros presented by Goddard (l967) , and computed population estimates from the bounded-counts method on 6 successive sets of 3 surveys. The results were 24, 26, 40, 47, 28, and 34 rhinos estimated to be in the population. However, Goddard had independent data to show that there were in fact 69 animals present. Some of the animals were under heavy cover or otherwise concealed so that they could not be seen from the air. Another alternative exists for areas where known numbers of animals are removed. If counts are made before and after the removals, and these counts are consistent—i.e., the degree of undercounting is the same on each count—then some simple algebra employing the counts and the known number removed yields an estimate of the degree of undercounting, and thus an estimate of the true total. One prospect for checking coverage has been described as "ground truthing," or having ground observers maintain intensive coverage of some sample areas in such a fashion that they can determine whether given bands are missed from the air. This scheme has not been attempted with wild horses or burros, but it has been applied elsewhere. An account of the essential aspects of one application may be found in Eberhardt and others (l979). Some trials would be necessary to determine if the method could be usefully applied to wild horses or burros. A likely approach would be that of positioning ground observers in places where they could keep one or more bands of horses under continuous observation, and thus determine whether those bands would be spotted by the aerial observers. An obvious difficulty is that of determining just which bands were sighted by the aerial observers. That may be dealt with by careful mapping of locations from both air and ground. In censusing wild horses, band size and coloration of particular individuals could be used to aid in checking the identity of given bands. Possibly radio telemetry could be used in place of ground observers to establish the positions of particular bands. These data can then be checked against the aerial observer's record. An application of this idea has been given by Floyd and others (l979), who conducted quadrat searches for white-tailed deer. They used

l94 telemetered deer, and arranged flight plans (without the observer's knowledge) so that the aircraft passed over the area in which the deer were located. The marked deer were identified by the presence of "collars" and the fraction of marked deer actually seen was used to correct the total counts on the survey quadrats. As the authors noted, these checks should be done on quadrats that are part of the actual survey area, and should be conducted within the framework of the actual survey. Also, if complete counts are attempted, there should be no need to conduct a special auxiliary survey of this type. All that is needed is to be sure that the observers do tally any marked animals. Another means for checking the accuracy of absolute counts is by marking specified numbers of animals in areas to be censused, and then determining the proportion of these seen during the census. It is then assumed that the same proportion of unmarked animals is seen. It is an implicit assumption that the probability of seeing a marked animal is the same as that of seeing an unmarked animal, and it must then be true that the mark is not so conspicuous as to make the animal more readily seen. It must also be true that the act of marking does not make the marked animals shy and more prone to flee at sound of an approaching aircraft. Marking has been done either with paint-filled frangible bullets fired from a carbon-dioxide-powered gun, or with specially made darts propelled from an explosive-powered gun. The latter device is said to have much better accuracy and range. The use of natural marks undoubtedly has potential; some investigators have used them, particularly people doing behavioral studies in situations where the results might be influenced by the handling or harassment involved in applying an artificial mark or tag. The main technical question has to do with the problems of misidentification or failure to identify a "known" individual in subsequent samples. Because of the potential for such confusion, it might be advisable to use natural marks only in situations where a few thoroughly experienced observers are working together in a circumscribed area and where photographs can be used to record (and recheck) markings. The technique thus does not seem useful for large-scale surveys. One variant of the use of natural markings may be worth further consideration for use on wild horses. This is an adaptation of the technique developed by Hewitt (l967), who censused male red-winged blackbirds by the Petersen (Lincoln index) method. Birds were "marked" by locations and landmarks described into a tape recorder (distances were measured by a car odometer) during the first trip through a census area. On a second trip, the observer listened to the tape and checked distances to determine which of the birds seen were also observed on the first trip. For wild horses, a combination of distinctive coloration, band size, and location could be used to identify previously sighted bands. Such a census might be practicable only on rather open and accessible ranges. Otherwise, the tendency for bands of animals to occupy particular areas might well bias the results by making "marked" bands more likely to be reobserved on the second trip than unmarked bands. Counts would need to be made on

l95 successive days to reduce the effects of movements from area to area, and thus several observers might be required to cover a given management unit. 3. Estimates Based on Sampling. While most of the current work on equid census attempts complete counts, there may be situations in which accurate, complete counts are not practical or feasible, and in which some sampling approach might be appropriate. Far more wildlife census work depends on sampling procedures than on complete counts. Several of the currently used principles of sampling estimation may be useful for equid census under given conditions. a. Mark and resight methods. This method, a modification of the more traditional capture-recapture method for aerial use, might well have potential. Animals in the first sample from an area are marked after being captured—or from the air as described above—and then released into the population at large. Visual sightings are then used to determine the fraction marked and thus to estimate the population size. The attempts to date to use this method on equids have apparently been restricted to the application of the Petersen method to a single resighting survey (unpublished BLM studies in Arizona). A more effective approach would be to use repeated resighting surveys to improve the precision of the result (narrow the confidence limits). This has been done for deer by Rice and Harder (l977). When repeated resighting surveys are used, it is important to determine the relative efforts that should be devoted to marking as compared to resighting. Relative cost or effort data should be obtained and used, along with variance estimates from the appropriate population estimation model, to determine an optimum allocation of effort. Robson and Regier (l964) have done this for the Petersen method with a single marking and single recapture. If their analysis can be extended to multiple recaptures, it would correspond with the mark and resight situation. Once again, an underlying assumption for this method is that the probability of resighting is the same for marked animals as it is for unmarked animals. However, it may be that the trauma experienced by animals shot with tranquilizer darts from a helicopter will make them take flight when they again hear a helicopter. Thus the probability of their being resighted changes. Very likely this is a situation where natural marks might be used in a behavioral study designed to compare the response of animals that have been marked with those that have not been handled. A companion study might also be done in which telemetered animals are observed during the approach of aircraft at various distances and altitudes. The probability of capture also depends on sighting, so that large bands and bands frequenting open areas are more likely to be marked and to be resighted. b. Plot sampling. Where areas are too large for, or the terrain and cover not suited to, total and thorough search for

l96 complete counts, populations can sometimes be estimated from complete counts on sample areas, and the total number of animals extrapolated from the densities on these areas. Sample areas can be units of convenient size for searching, over which the aircraft is flown on some sort of search pattern, or they can be sample strips of prescribed width on each side—or on one side—of the aircraft. The latter are usually known as strip transects. The more compactly shaped areas, or quadrats, are commonly searched by repeated circling. This technique has the advantage of giving the observers a thorough look at an area from several perspectives. It may also serve to startle resting or concealed animals, making them easier to see. Quadrat searches are thus of special value in rough country, or in areas with substantial vegetative cover. A possible disadvantage is that horses are highly mobile, so that repeated searches of a particular sample unit may drive the animals present out of the unit. It is necessary to take this possibility into account in designing a sampling survey. An alternative is to use photography of the quadrat, as has been done by Norton-Griffiths (l973, l974) and Mathisen and Lopp (l963). Some of the extensive experience in using quadrats is presented in papers by Siniff and Skoog (l964), Evans and others (l966), Goddard (l969), and Bergerud and Manuel (l969). The same major problem experienced in total counting arises in the use of sampling; namely, that of checking the degree of coverage attained (i.e., the fraction of animals present that are actually enumerated) and, if possible, seeking to "calibrate" or adjust such counts for the fraction missed. The same approaches discussed above under "Complete Counts" apply to the plot-sampling methods. Sampling strips have the advantage that they can be spaced far enough apart so that there is little prospect of given bands moving from one strip to another in successive passes by the aircraft. Disadvantages are the prospect of missing some animals on the strip and the limited area coverage sometimes achieved, which may result in substantial variability of the estimates. As with the other methods, it is necessary to do field studies to determine the importance of these potential difficulties. As previously noted, the strip-census method has been widely applied in Africa, where a great deal of practical experience concerning its use has been accumulated. A review and description of the techniques used there can be found in Norton-Griffiths (l975). There has also been an appreciable amount of experience with strip transects in Australia (cf. Frith l964, Caughley l974, and Caughley and others l976), in arid and semi-arid areas. This experience indicates that only a rather narrow strip can be searched effectively, and that low altitudes and airspeeds are required. Even so, an unknown fraction of the animals present is missed (Caughley and others l976). Here again, independent checks on the degree of bias are needed. A particularly impressive demonstration of the effect of strip width has been presented by

l97 Caughley (l974) and Caughley and Goddard (l975). These investigators report a dramatic change in the numbers of elephants counted from the air with changes in strip width from l00 m to 600 m. c. Line transects. In this method, estimates of the distances from the observer to the animals sighted are used to correct for the fact that visibility drops off with distance from the transect line. An important assumption is that all of the animals immediately on the transect line are detected. Some experience with other species (Eberhardt and others l979) suggests that this assumption is likely to be attainable only with specially equipped aircraft: namely, those with a "nose bubble" or similar arrangement. Since the BLM work must be done in many locales and usually with locally chartered aircraft, this requirement may present difficulties. On the other hand, many of the available helicopters do have good forward visibility. At present, it appears that the first step is to assess present experience with aerial census methods that have actually been applied to wild horses and burros or similar species. If such an analysis suggests a limitation due to small numbers of animals observed, then it may be desirable to investigate line-transect methodology. A detailed review of the methods is available in Eberhardt (l978) and Eberhardt and others (l979). The line-transect method is not restricted to aerial surveys, and much work has been done with it in ground-based surveys. Sources and Corrections of Bias and Variability l. Group Size. One source of bias not yet investigated in wild horse and burro censusing is the likely effect of group size on visibility. This problem actually subdivides into two kinds of bias. The first is simply that group size will have a marked effect on the probability that a given group of animals will be seen. One model for estimating group size is that of Cook and Martin (l974), while a more general approach has been explored by Patil and Rao (l978). Jolly and Watson (l978) have suggested some quite simple corrections for missed groups that can be derived from ratio data collected to adjust for members of groups that are observed, but incompletely. It should also be noted that group sizes are likely to vary seasonally, a factor that should be taken into account in survey design. In general, larger groups are much easier to locate from the air, so there is an advantage to working in the seasons when large groups are most prevalent. However, other factors may also affect the choice of season. One disadvantage to surveying large groups is that counting may call for repeated circling over the animals, and this may cause a large group to break up into its component bands, thus changing the group size in subsequent surveys. The second source of group-size bias is the increasing difficulty of accurately counting the numbers of animals in a group as its size increases (Sinclair l973, Bell and others l973). This problem has

l98 been studied by African workers, who recommend photographing groups larger than l0 animals, and checking the visual count against the photographs (procedures are given by Norton-Griffiths l975). 2. Sampling Methodology. A number of sampling issues must be considered as research is continued on censusing horses and burros. Random sampling is generally recommended, but should be properly conducted; otherwise searching one sampling area might affect the animals on, or adjacent to, the next such unit. This problem will not arise if the sampling is truly random, that is, if the units are searched in the order in which they are drawn. However, such a procedure is demanding in terms of travel between units, so many investigators will draw the entire sample and then traverse the units systematically across the study area. An alternative is to use restricted randomization, i.e., to place one sampling unit at random in each of a number of blocks that are large enough to avoid much movement of animals between blocks. The process can then be repeated on successive days of searching. Perhaps a more practical scheme is to use systematic sampling with a new random start on each day of searching. Stratification should be considered, but may have a similar drawback in that movements from one stratum to another are possible. An additional problem with stratification is that strata may be chosen by cover type, which will likely result in differences in visibility between strata. There is then the need to make separate corrections for each stratum. Unless the study units are very large, or animal densities vary markedly, it may be best not to stratify. If sample areas other than strips are to be searched, it may be most practicable to use units of differing sizes. This approach brings in the problem of combining data obtained from units of disparate size. Jolly (l969) has suggested the use of sampling with probability proportional to size (PPS sampling) or of a ratio method for this situation. PPS sampling requires that a sampling unit be included each time it is drawn. Many workers using this method do not search the unit each time, but simply apply the result of the first search to each subsequent drawing of the sampling unit. However, this practice may not be suitable for use with wild horses, since—as already noted—the disturbance of an aircraft working over the area will most likely rearrange the population repeatedly. Hence, in PPS sampling, the units should be searched each time they are drawn. In most cases, the areas to be searched for wild horses and burros will be well defined, with adequate landmarks for laying out and conducting census flights. If such is not the case, then it will be worthwhile considering an approach that requires only demarcation of a baseline. This method is described by Bell and others (l973) and Norton-Griffiths (l973). 3. Variability. The appraisal of variability is obviously a critical issue in the overall problem of censusing wild horses and burros. If an unbiased method turns out to give extremely variable results, then it may be preferable to use a somewhat more biased

l99 procedure. Unfortunately, most decisions about variability require actual data collected under field conditions, and these have not been available for wild horses and burros except in a fragmentary and limited way. Hence, the issue of variability must be included as an important part of the needed field research program. 4. Double Sampling. One possible means of testing the completeness of counts—either total counts on an entire area, or counts on sample plots—is that of "double sampling." This approach involves using an accurate, unbiased method on a series of test areas—when and if one is devised—in combination with some other admittedly biased, but perhaps less expensive and more readily applied, method (Cochran l977) on the same area. Ratio or regression methods are then used to extend the results from this "calibration set" to a larger set of areas on which only the less expensive method is used. The practical difficulty is that present recommendations (Cochran l977) call for samples of approximately 30 pairs for the calibration. Very possibly this requirement can be reduced somewhat, but we suspect that the sample ought to be at least 20 pairs. Such an approach does not seem feasible for the circumstances discussed here. A possible exception concerns estimation of the entire population of horses (or burros), in which case it may be feasible to use the most recent BLM district estimates as the "auxiliary variable" in double sampling, and survey a subsample with the accurate method to obtain an improved estimate of the total. The double-sampling approach has other potential uses. For example, Jolly and Watson (l978) advocated that a subsample of the groups of animals sighted in a survey should be circled and observed until the observer is satisfied that all animals in the group have been sighted. These more accurate estimates are then used in a ratio correction to adjust the overall group size observed on the transect line (i.e., observed without circling). A problem with this approach is that some of the smaller groups will not be seen at all. Jolly and Watson suggest an approximate method for correcting for this problem, based on the ratio data. Their approach would probably work best on strip transects, since it is likely that nearby groups will be disturbed while one band is being circled for accurate counting. Preliminary Analysis of BLM and USFS Census Data The accuracy and precision of existing horse and burro census procedures need to be investigated in connection with the proposed research on this subject (Project l7). It is only with that research that any firm assessment of the procedures can eventually be obtained. The various and extensive data on horse and burro censuses in BLM and USFS files should be analyzed thoroughly as part of the census research effort. As discussed under "Squid Demography" (Chapter 2), we obtained census data from the files of l0 BLM districts and one national forest on 67 horse herds, 3 burro herds, and two areas that contained both

200 (see Table 2.ll). The data were analyzed to gain some preliminary estimates of the accuracy of the censuses, and to see what biases and variables could be detected. Upon first examination, several characteristics became evident: l. The time interval over which each herd was censused varied between l and ll years, and the number of counts varied between l and l8. Most of the herds had been counted over 4 to 7 years. 2. The time of year during which a herd was counted often varied between years. Thus a herd might be censused in March of one year and August of another. The numbers in all seasonally breeding animal populations vary during the year, rising to a maximum at the end of the birth season, and then declining until the next as a result of mortality. Since wild horse foals are dropped seasonally, counts in March and August of even the same year could vary by l5 or 20 percent. It is a basic principle in censusing animal populations that inferences about year-to-year trends and comparisons between areas must be based on censuses taken at the same season each year, or at the same stage in the annual life cycle. While burro populations produce colts throughout the year, the number of births intensifies in late spring and early summer. Hence, this same principle should be applied to burro census. 3. While most of the counts were made from the air, the mode of census varied; sometimes censuses were conducted on the ground, sometimes from fixed-wing aircraft, and sometimes from helicopters. In many cases the method was not specified. Casual inspection of the data showed one seemingly obvious difference between census modes. It appeared that in most cases where fixed-wing aircraft were used in one year and helicopters in the next, the censuses increased abruptly in the second year. An effort was made to explore this variation further. Out of the 72 herds, there were l9 horse herds in which a change from fixed-wing to helicopter counts was made between years and recorded, and in which the census season was held reasonably constant. The unweighted mean difference between the l9 pairs of counts showed the helicopter counts to be 88 percent higher than the fixed-wing counts of the previous year. Even after discounting this percentage by l4 percent for population increase, the helicopter counts appear to be some three-fourths more efficient in finding animals than the fixed-wing counts. Since fixed-wing counts were used more commonly in the earlier years, and helicopters more generally in recent years, it is quite likely that earlier counts underestimated the actual numbers to a considerable degree. The helicopter counts also appear to be less variable than the fixed-wing counts. The coefficient of variation about the mean count for the fixed-wing censuses was l70 percent, and that about the mean count for helicopters was l3l percent. Changes in herd size and length of individual record sets will of course influence this comparison. As mentioned earlier, it seems possible that very nearly all of the horses in open terrain can be seen and counted by a skilled

20l observer working under favorable weather conditions. If the censuses err in any direction, they are probably conservative because some animals are missed. It seems likely that the degree of error is much lower in horses, which are more likely to run when approached by an aircraft, than in burros, which are more likely to remain motionless. In addition, the burros' grey to brown coloring blends in with the desert terrain they inhabit and further reduces visibility. One insight into the magnitude of error in aerial burro census is provided by Ohmart and others (l978). These authors flew a l5,360-ha area in the Black Mountains, Mohave County, Arizona, making 7 hours of low-level helicopter flights similar to those used for census counts. They marked every burro they saw with a CO2-charged pistol that expelled oil-base paint pellets. After 5 to 9 days, they flew the area again, recorded every animal they saw, and determined whether or not it was marked with paint. About one-third of the animals they observed were marked, and the inference drawn was that typical aerial census operations for burros could be expected to count only about one-third of the animals present. The authors also suggested that a mark-resight method would be more appropriate for burro census than attempts at total counts. Review of Range-Survey Methodology Although the Public Rangelands Improvement Act of l978 authorizes the Secretary of the Interior to take necessary action to remove excess animals on rangelands where several species utilize the same forage resource, the question of which herbivores are in excess is difficult to resolve. Principal among the biological factors to be considered in making such a determination are forage and habitat preferences and specific grazing impacts on soil and vegetation resources (see Chapters 2 and 3). Other factors, including season of use and characteristics of the land that influence distribution and grazing patterns, must also be weighed in range analysis. Much of BLM decision making in these regards is based on the range inventory or survey. The first standardized range-survey method was developed in l937 and is sometimes referred to as the Ocular Reconnaissance Method (Interagency Range Survey Committee l937). It is still discussed in BLM manuals and is still being used by that agency in some areas. Other agencies have slowly developed other methods ("weight estimate," range sites and soil surveys, SCS; Range Environmental Analysis, USFS). Around l978 to l979, BLM began using the "Soil-Vegetation Inventory Method." It is based primarily on the SCS methods (U.S. Dept. of Interior l979). Range inventories made by BLM or any other agency are usually used for more than one purpose. BLM is probably the only agency in the United States that is still very concerned about establishing an initial grazing capacity. Other purposes of inventory are (a) to establish a baseline for monitoring range "trend" and (b) to provide

202 basic information about resources and conditions to be used for planning and management. Hence, in talking about "inventory" there is more at issue than just the establishment of stocking rates. However, those inventory factors and processes that are most controversial are the ones applied primarily to obtaining stocking rates. Historically, initial determinations of grazing capacity on public lands were based on range inventories or surveys that attempted to estimate average, usable forage production. The available forage was then allocated to domestic livestock producers through an adjudication process that considered their relative privileges under existing laws and regulations. A percentage of the available forage was usually set aside for wildlife, or the survey methodology was presumed to reserve wildlife forage before grazing capacities for livestock were established. However, few—if any—forage reservations were ever made for wild horses and burros. These processes were completed on most public lands many years before wild horses and burros became by law the responsibility of the federal government. Recently, the BLM has begun to make determinations of grazing capacity (initial stocking rates) as part of its land-use planning procedures and environmental impact studies of livestock grazing problems. These are being made primarily because many people believe that earlier studies did not adequately consider the suitability of the land for livestock grazing, nor did they properly allocate range resources among all uses and users; e.g., wild horses and burros, habitat as well as forage for wildlife, protection of threatened and endangered species, cover for watershed protection, enhancement of recreational experiences, and others, as required by recent legislation. Initial grazing capacities (or "stocking rates"), whether determined through the historical survey and adjudication procedures, or the newer land-use planning processes, were made to be adjusted periodically on the basis of continued monitoring studies in the grazing units. At the least, monitoring studies should employ actual use records, permanent range-condition and trend-monitoring sites, and utilization measurements. They may also include such studies as water-quality monitoring, surveys of nongame species, analyses of grazing impact on cultural resources, and soil-loss evaluations. Forage for wild horses and burros was not generally provided for in either public land adjudications or land-use plans prior to passage of the Wild and Free-Roaming Horse and Burro Act in l97l. However, many domestic horses were legally licensed to use public lands. Since l97l, BLM and USFS range managers have had a legal mandate to manage wild horses and burros to achieve and maintain "a thriving natural ecological balance on the public lands" (U.S. Congress l978) . An attempt has been made by this Committee to review some recent BLN and USFS documents that consider wild horse numbers in terms of current land-use planning and management. The purpose was to evaluate how adequate range-survey methodology has been in addressing questions of proper equid numbers based on range biological factors. No detailed review of the literature on range-survey methodology has been published, although that literature is extensive, vallentine (l978)

203 lists l38 references to herbage-removal studies alone. We obtained and examined l0 wild horse capture plans with accompanying environmental analysis reports (EARs) (see literature cited under BLM for titles). Of these, 8 were BLM plans and 2 were joint BLM/USFS plans; 9 were from Nevada, and l from Oregon. Two capture plans (East Range and Paradise/Denio) were proposed primarily because of the high proportion of nonfederal land within the horse-use area. Thus range condition and forage competition with other herbivores were not direct considerations. The remaining eight wild horse reductions were proposed primarily for the purpose of either correcting depleted or unsatisfactory range condition and overutilization of forage, or adjusting existing horse populations "to numbers established in the land-use plan." Few of the plans and EARs provided much information to support the contention that range conditions were unsatisfactory or explained how those conditions were determined. Rarely was any attempt made to show which herbivores (horses, cattle, or wildlife) were responsible for the range condition presented in the report. In one case (Lucky C), there was obvious disagreement between state and district office personnel as to actual conditions. The BLM district claimed there had been no livestock grazing the range in question for at least 2 years, and yet the BLM state specialist concluded that wild horses were only a minor contributor to admittedly deteriorating range conditions. The most recent EAR (Pine Nut Mountains), however, had detailed supporting material to document the impact of horse populations. This information included actual use records by allotments, as well as data from 38 vegetative trend plots and 4 years of range utilization studies. We conclude that although adequate range studies may not always have been used to evaluate and support proposed numerical adjustments of wild equids, the technology to do so exists and appears to be adequate. Therefore, we believe that additional research on range-survey methodology is not justified at this time in the present context, except where it can or must be included as a part of other studies. Limitations of Techniques for Diet Analysis Most studies assessing dietary composition for both horses and burros have relied on fecal analysis for determination of the relative contribution of each forage species to the diet (e.g., Hansen and Martin l973, Hubbard and Hansen l976, Hansen and others l977, Salter l978, Vavra and Sneva l978, Salter and Hudson l979). Other techniques can be used for quantifying diet composition. Moehlman (l974) observed actively foraging animals daily in her study of Death Valley burros. Browning (l960) analyzed stomach contents. However, such approaches are often expensive, time consuming, or impractical for use on free-ranging animals. Consequently, fecal analysis has been widely used. The BLM currently prescribes the use of this procedure in contract studies that are conducted to provide a basis for forage

204 allocation. However, the accuracy and precision of the technique, at least for research purposes, is presently subject to question. Smith and Shandruck (l979) found that fewer plant species could be identified in the feces of antelope than in their rumen contents. In addition, the amount of browse and forbs found in the feces of mule deer correlated poorly with the known composition of their diets. Extreme disparities can occur between known diet composition and that indicated by fecal analysis: Slater and Jones (l97l) found the epidermal tissues of white clover to be completely absent from the feces of sheep, even when the animals were fed diets containing 37 percent clover. These studies were conducted with ruminants, and the digestive system of equids could conceivably minimize these problems. Casebeer and Koss (l970) determined that fecal and stomach-content analyses of zebra diets agreed best when the diet was composed mostly of grasses. However, grasses in the early stages of growth and forbs were poorly represented or absent in fecal samples. Owaga (l977) also compared stomach and fecal contents of zebra, although her methodology differed somewhat from that of the other studies cited in that she identified intact plant parts in samples rather than cuticular or epidermal fragments. She determined that of some 60 grass and sedge species that constituted the diets of her zebra, the l0 most important ones in stomach samples were also the most important species in fecal samples, although the order of importance sometimes varied. Thus fecal analysis apparently sacrifices some degree of accuracy even when applied to equid species under the relatively unrigorous conditions of a grass-dominated diet. The more general weakness of any technique relying on microscopic identification of cuticular or epidermal fragments was highlighted by Owen (l975). Plant species are highly variable with regard to the proportion of epidermal fragments having one or both identifiable surfaces. Large errors in diet quantification occurred in his study unless correction factors were established for each species. However, Sparks and Malechek (l968) found that the dry weights of four grass and four forb species in known mixtures were directly proportional to the number of epidermal fragments identified and counted in subsamples of the mixtures. The analytic technique outlined by Sparks and Malecheck (l968) has been the basis for most of the laboratory analysis of horse feces carried out in the western United States since their paper appeared. Inferences about dietary composition based on fecal analysis should be drawn cautiously until the technique is further validated with a proven procedure such as microscopic analysis of fistula extrusa (Theurer and others l976). Data from fecal analysis will apparently suffice for such purposes as ranking the dietary importance of various plant species (Vavra and others l978). For situations where a high degree of accuracy is not necessary, such as in certain management applications, fecal-analytic data are apparently useful. However, the user should recognize the limitations of such data and should not make inferences implying high accuracy.

205 A potentially serious limitation to fecal analysis that has not been evaluated is the lack of assurance that fecal samples represent forage consumed in the vicinity where the samples were collected. The transit time of food through the gastrointestinal tract of horses has been estimated at 37 hours for long hay (Wolter and others l974). Thus, in situations where animals range over a variety of sites or vegetation types in their daily feeding circuit, it is improbable that the botanical composition of the diet can be related to the kinds and amounts of forage available in the vicinity where feces are collected except in a very general sense. In the cases of deer and elk, Collins (l979) found that roughly half of all defecations occurred during 3 to 6 percent of the day, when the animals were traveling between preferred feeding sites. More often than not, the vegetational subunits separating the preferred feeding sites were of marginal or no value to the deer or elk as food resources. Similar criticisms can be leveled against diet studies based on rumen content analysis, and particularly against fecal-analytic studies of ruminants, whose food transit time is considerably longer than that of equines. Another important consideration regarding any technique based on microhistological analysis of dietary materials or residues is the adequacy of reference materials. Typically, this aspect of the technique involves collecting all forage species in the areas where fecal material or diets are sampled. Most researchers who have dealt with diet analysis, by one technique or another, will readily admit that selectively feeding herbivores often ingest considerable quantities of plant species that the researcher finds difficult or nearly impossible to locate, even by intensive systematic sampling of the plant community. Consequently, reference collections often lack representation of a few plant species that may, on occasion, contribute significantly to diets. This is especially true when the reference species are collected by someone other than the person doing the dietary or fecal analysis. Extreme caution should be exercised in interpreting studies that do not report at least some small (say, <l0 percent) proportion of a diet as unidentifiable material. Sample size and frequency of sampling are also matters of great concern to those who wish to make strong inferences from dietary data. Van Dyne and Heady (l965), working with sheep and cattle that had esophageal fistulae, calculated that l to 298 samples were required from a single population of animals in order to estimate dietary composition within l0 percent of the mean with 90 percent confidence. The greatest variability in diets occurred in early summer, when an abundance of different forage species was available, but it still remained high in late summer. Similarly, Harniss and others (l965) showed that large numbers of samples of fistula extrusa were required to determine accurately the diets of sheep grazing a sagebrush-grass range in spring. They found that for species contributing 40 percent or more to the composition of diets, the coefficient of variation (s/x X l00) was around 30 percent. For such species, 26 samples would provide estimates within ±l0 percent of the population mean 90 percent of the time. If accuracy of only ±20 percent of the population mean was required, then the number of samples necessary dropped to eight.

206 Unfortunately, roost studies on horse and burro diets have not presented variance statistics for dietary composition. Thus a comprehensive survey of sampling variability is not possible. However, Hubbard and Hansen (l976) did present standard deviations for their mean dietary composition data. From their published figures, approximate sample sizes were calculated according to Snedecor and Cochran (l967) that would estimate within ±20 percent of the population mean of a species 90 percent of the time. The required number of samples for the three most abundant grasses in their horses' diets were l7 for Stipa comata, 33 for Agropyron spp., and 48 for Bromus spp. If less common species were to be estimated accurately, more samples would be required. Aggregate categories such as "grasses" could be estimated with far fewer samples, particularly in view of the apparently small variation in grass composition of horse diets suggested by Table 2.22. Not only is variability high from animal to animal in diet studies, but there is often extreme short-term variation within a season as well (Van Dyne and Heady l965; Collins l977, l979). Thus frequent sampling is necessary—often every 2 weeks—if one is to have a true picture of seasonal preferences. Vavra and others (l978) even suggest that important daily variations in diet occur. Even if adequate data have been obtained, questions remain concerning their interpretation and inferential value. In this regard several problems arise. The most common approach for relating dietary-composition data to plant-community composition has been to calculate preference quotients. These quotients are produced by dividing the quantity of a forage in a diet by the quantity of that forage available to the animal (see Krueger l972 and Petrides l975 for detailed discussions of this concept). Forages that are relatively more abundant in the diet than in the plant community are assumed to be "preferred." However, failure of an animal to exhibit a "preference" for any environmental factor, whether it be food or something else, does not necessarily indicate that the factor is unimportant. Rather, there may simply be more of it in the surroundings than the animal needs. Smith (l954) suggested that the simple percentage occurrence of a species in the diet may accurately represent preferences for plants where availability does not limit consumption. Determining Range-Forage Digestibility in Equids In studies of ruminant nutrition, in-vitro techniques have proven valuable in evaluating forage digestibility. In-vitro approaches, while possibly less accurate than in-vivo techniques, are often the most practical in situations where the forages are difficult to collect in sufficient quantities for feeding trials. However, we know of no in-vitro procedures that have been specifically modeled after equid digestive physiology. L. M. Slade (Department of Animal, Dairy and Veterinary Science, Utah State University, personal communication, l979) stated that

207 in-vitro methods commonly used for studying ruminant diets were likely to be applicable to equids, providing the inoculum is buffered at a pH of 7.0 to 7.2 during the fermentation stage. Roller and others (l978) used a modification (Goering and Van Soest l970) of the Tilley and Terry (l963) technique to investigate cell-wall and dry-matter digestion in ponies. The in-vitro technique was not as responsive to change in the ponies' nutritional status as was an in-vivo (nylon bag) method. The authors acknowledged that the effect of the predigestion by gastric secretions of the foregut and small intestine on fiber entering the cecal and colonic environments is not known. Hemicellulose digestion may be enhanced by exposure to the intestinal environment prior to cecal fermentation (Keys and others l969, l970). It would be worthwhile to investigate how in-vitro techniques suitable for ruminant nutritional research could be adapted to studies of range equids. How well and how accurately these procedures can be transferred is uncertain. Further study comparing the results of appropriate in-vitro procedures with in-vivo digestion is needed. Possibly a procedure that more closely duplicates the sequence of equid digestion should be developed (i.e., acid-pepsin digestion should precede anaerobic fermentation). As an alternative to in-vivo digestion trials where fecal output is collected and measured, indicator methods (Kotb and Luckey l972) might prove suitable. Pulse and others (l973) demonstrated no significant differences among estimates of fecal output determined by three techniques: total collection, 0^O3 indicator, and a polyethylene powder marker. Similarly, Knapka and others (l967) reported that polyethylene and Cr2C>3 methods are reliable indicators of fecal output to be used in determining digestibility in burros. The possibility should also be investigated that the free-ranging horse's ability to digest forage can be adequately estimated from chemical data by a regression procedure or summation equation such as that proposed by Van Soest (l967) for ruminants. Given the horse's apparent propensity to eat grass almost exclusively, this approach could prove highly successful. Problems might be anticipated where shrubs and other unconventional forage items form a large proportion of the diet. Certain published data (e.g., Fonnesbeck l968, l969; Vander Moot and Trout l97l) would provide a valuable departure point for research on such an approach. Blood Assays as Possible Indices of Nutritional State in Horses and Burros Assays of easily taken blood samples may have potential both for evaluating the nutritional condition of individual animals and for using an animal's condition to indicate nutritional adequacy of the range (Seal l977) . The evaluation of individual animals may include nutritional condition in terms of very recent food intake (protein and energy), and of the quality of that intake during recent months. One may also evaluate specific nutritional deficiencies, as well as the

208 physiological effects of various capture and handling techniques for horses. The literature relating nutritional status to breeding capability of mares is extraordinarily sparse (Button and others l977, Ellis and Lawrence l978). A study in South Africa (van Niekerk and van Heerden l972) compared two groups of mares (one kept on pasture and the other kept on pasture with supplemental grain) in terms of the occurrence of estrus, ovulation, and successful conception. The pasture grass contained approximately 3.2 percent protein and the supplements contained l2 to l5 percent protein. Six of eight animals maintained on grass alone exhibited estrus, but only two of these ovulated and both conceived. In contrast, the seven mares maintained on grass plus supplement exhibited estrus and ovulation, and the six served conceived. The supplemented animals gained a total of 53 kg over the 53 days of the study, whereas the unsupplemented animals did not gain weight. The starting weight averaged 370 kg. These data, like those for other species, suggest that variations in nutritional intake can affect ovulation. Horses are also vulnerable to abortion and stillbirth, to lactational failure, and to loss of foals too weak to nurse properly; however, quantitative studies addressing these points and their relation to nutrition do not appear to be available (Ginther l979). Nitrogen intake and absorption in horses can be assessed by measuring plasma urea nitrogen concentration (Fonnesbeck and Symons l969, Reitneur and Treece l976, Owen and others l978). Fonnesbeck and Symons studied diets ranging in protein content from 8.3 to l6 percent. This span neglects the very low protein content of range forage that may be encountered during some seasons of the year. The correlation of plasma urea nitrogen with apparent nitrogen absorbed was r = 0.78. There were indications of variations in plasma protein, sugar, and cholesterol concentrations with the several diets. However, there was no direct correlation, except that those diets with the highest soluble carbohydrate content resulted in the lowest serum cholesterol concentrations. To encompass the range of conditions likely to be encountered in natural habitats, a study including diets of 3 to 4 percent protein and 6 percent protein would be required. Serum protein and hemoglobin levels will decrease at a lower protein and energy intake, as was found with animals maintained on pastures without supplementation (Owen and others l978). However, there are no controlled experiments available concerning these possibilities in horses. The effects of fasting for periods up to 9 days have been described in ponies (Baetz and Pearson l972, Wensing and others l975, Gronwall l975, Baetz l976). These authors described increases in bilirubin, cholesterol, phospholipid, free fatty acids, triglycerides, other serum lipids, pyruvate, lactate, and alpha-one-globulins. Serum urea, phosphorus, and magnesium decreased. It appears generally possible to use several of these assays to distinguish fasting or total food withdrawal from reduced food intake, and thus it is not necessary to confuse these conditions. There are differences, however, between ponies and larger breeds of horses, including Morgans

209 and thoroughbreds (Robie and others l975). Apparently ponies maintain higher triglyceride levels in cold weather and are also more vulnerable to fasting hyperlipemia than the large animals. These results imply the desirability of analyzing data from captured wild animals in terms of season and lipid metabolism. There is no evidence that season affects serum urea nitrogen unless the animals are on pasture without supplementation (Owen and others l978) . Literature on hematology and blood chemistry of burros (Brown and Cross l969, Yousef and others l97l) and domestic horses is available. In addition, there have been several careful studies of water metabolism in the burro under desert heat conditions (Yousef and others l970, l97l). Data for newborn and young foals are available as well (Kitchen and Rossdale l975, Medeiros and others l97l). Studies on zebras (Seal and others l977) provide evidence that equids are similar to other mammals in their blood responses to trauma and disease. Blood proteins that increase during any disease process or tissue injury are called acute-phase reactants. The response pattern of each protein is characteristic, and careful studies in humans have demonstrated a semiquantitative relationship to such traumas as myocardial infarction, surgery, and various infectious diseases. Measurement of such proteins as fibrinogen, haptoglobin (Allen and Archer l97l, Spooner and Miller l97l) and perhaps C-reactive protein provide simple yet powerful tools for assessing the general clinical condition of an animal without undertaking a full spectrum of clinical diagnostic procedures. Prior exposure to various infectious diseases in currently well animals can be detected by seriological assays, which are readily available for certain widespread agents of major domestic importance. Detection of the effects of handling, exercise (Hensley l978; Keenan l979; Lucke and Hall l978; Krzywanek and others l976; Rose and others l979; Snow and MacKenzie l977a, b) , and sustained chasing and capture is a matter of continuing concern and interest for those involved in the capture and transport of both wild and domestic animals (Harthoorn l976). Increased hormone secretion by the adrenal cortex and medulla (Anderson and Aitken l977) is considered to be a physiological response to such disturbances. A study of horses (Kirkpatrick and others l979a) comparing different techniques for handling found no differences in serum cortisol concentrations and concluded that the several techniques were minimally stressful in terms of their impact on serum corticoid levels. Additional sensitive techniques for detecting the effects of different levels of exercise and muscular exertion in horses appear to be measurement of creatine phosphokinase (CPK) activity (Anderson l976), uric acid (Keenan l979), and bilirubin (Hensley l978). The enzyme CPK is a sensitive indicator of exercise and significant muscle trauma. Animals in initially poor condition tend to respond with much greater increases in serum CPK after graded levels of exercise. If captured animals die—particularly after chases and during their first 5 days in captivity—then the possibility of metabolic acidosis and myopathy resulting from the exertion need to be considered. This condition has been well documented in zebra by

2l0 Harthoorn in a long series of studies in Africa. He described an effective treatment that consists of giving a liter of electrolyte solution with l,000 mEq sodium bicarbonate added, which counteracted the decreased blood pH and resulted in survival. There are available pH meters suitable for field use that would allow diagnosis of the problem. Other techniques for evaluating animal condition include hair bulb diameters, autopsy, direct measurement of rump fat thickness; gross and microscopic evaluation of tissues; estimation of fat thickness by ultrasonic measurement (Westervelt and others l976); and measurement of body weight in relation to age and bone growth. Extensive growth data on domestic horses exists from about l40 days of gestation up to maturity at 24 to 36 months. Most domestic horse strains appear to attain about 45 percent of their mature weight at 6 months, 65 percent at l2 months, and about 80 percent at l8 months. At the same time intervals, thoroughbreds achieve 83, 90, and 95 percent, respectively, of their mature height at withers; apparently guarterhorses and Arabians do also (Hintz and others l979, Platt l978). Behavioral Sampling: Methods for Comparison Between Treatments and Experiments Since different experiments within the research projects have been organized to analyze particular aspects of equid biology, but at the same time to allow comparability across experiments, it is necessary to measure behavioral responses within experiments and relate these responses to other treatments within and between experiments. For example, the data in feeding trials will relate nutritional plane to reproductive success. In order to relate the findings on nutritional condition in trial animals to animals in enclosures or on the open range, it will be essential to be able to compare the level of activity between groups. Since activity determines part of the nutritional requirement, the ability to measure and compare activity is critical to the ability to generalize the results of the feeding/nutrition experiment. Likewise, an understanding of the social effect of contraception on the horse will require good experimental measures that are comparable with measures used on the behavior of wild populations. An overall lack of comparability severely limits the ability of researchers to generalize their experimental findings to wild populations. A primary goal of the behavioral sampling scheme is to establish a set of data that are comparable across experiments. However, each experiment has unique questions that require information unnecessary for other experiments. Therefore, the discussion on behavioral sampling is presented in two separate sections. The present one deals with data to be collected that will be used for comparisons across experiments. The next section will concern data used within a single experiment or a subset of experiments.

2ll Comparative Methods in Behavioral Sampling Two initial decisions are required for a behavioral sampling scheme. First, it must be decided how animals from a population are to be chosen for sampling. Second, it must be determined how the behavior of the chosen animals is to be recorded. There are two widely used techniques for choosing the individual(s) for sampling. The scan sample is a method in which all animals in a group or some subset of the group are sampled simultaneously. The method provides a good sample of a population's response. Furthermore, since the sample is created instantaneously, this method removes temporal and spatial variability to a greater extent than methods where behaviors of individuals of a population are lumped but recorded on different days and locations (e.g., focal-animal sampling, discussed below). However, since several individuals are monitored at once, data are usually recorded in a way that makes it impossible to reconstruct which individual performed which behavior—as a raw percentage score. Thus, individual profiles of animals cannot be constructed. This weakness not only eliminates the capability of considering individual performances and requirements, but also makes it impossible to compare variability between individuals, between treatments, and between experiments. The second method for choosing subjects is the focal-animal sample, in which the researcher focuses attention solely on one animal for the sampling period. The advantages of this technique are that a great deal of detailed data can be recorded. From these data individual profiles and variability can be determined, a capability that allows more extensive statistical testing. One disadvantage of the technique is that a generalization to population response is more difficult, since data are recorded on only one individual at a time. "Population responses" inferred with this technique are really collections of individual records taken on different days, and so variability is increased because of changes in time, season, and location between samples. Another disadvantage is that a great number of individual records must be amassed to provide a sample size as large as that possible with scan samples. Furthermore, the work time necessary to characterize population responses by means of focal-animal sampling is great. The series of experiments in the recommended research projects will require both the ability to characterize behavioral differences across treatments (i.e., mean treatment responses) and the ability to measure intra-treatment variability (i.e., differences in responses to a treatment). Because of the size and observability of equid groups and the ability to distinguish individuals within groups, a subject-sampling method is proposed that combines the focal- and scan-sampling techniques. If data are taken by scanning the group being watched and attributing behavioral scores to the individual who performed the behavior, then the strengths of both sampling techniques will be combined. This method will allow data on a number of animals to be gathered simultaneously, thereby increasing sample size and reducing temporal and spatial variability. At the same time,

2l2 individual variability and profiles can be calculated. Because of the rate at which equids perform behaviors, and their size, it is possible to combine these two techniques. The second decision in behavioral sampling is that of how to record the behavior. Again, two techniques are commonly used. The instantaneous or point sample records the behavior of an individual at a given point in time. Usually this sample is taken at regular intervals throughout the sample period. Point samples are often combined with scan samples and the behavior of several or all animals in a group is recorded at an instant in time. Point samples allow the amount of time spent in activities to be estimated. However, they underestimate the duration of rare events and behaviors that are instanteous or short lived. They cannot be used to estimate the duration of activity nor to investigate accurately the sequential dependency of behaviors. Point samples would allow the comparison of behavioral rates and activity rates between treatments and experiments, but would not provide an estimate of individual variability. The second method of recording behavior is to note the state of the animal continuously. In this method—called continuous sampling—the onset, termination, and sequence of behaviors are recorded (usually against a real-timed record). These data are more detailed and therefore more difficult to record than are point samples. They are usually restricted to one sample animal and usually employed with focal-animal sampling. Continuous data have the advantage of preserving a much greater amount of information about the individual's activity and behavior. They can be analyzed for frequency, duration, and sequence as well as for rate means and other common statistics. Continuous data also can be recorded along with, e.g., point-sample information to generate statistics that can be compared to other data sets. Continuous data have the disadvantages of being difficult to collect, time consuming and costly to transform to an analyzable format, and of requiring large numbers of samples to characterize populations. Since they are commonly used with focal-animal samples, they share the disadvantages inherent in those samples as well. The data collected for comparative purposes should be gathered by means of a technique that yields large sample sizes easily, is statistically viable, and can be recorded and transformed for analysis with ease. For these reasons, we favor the point sample over the continuous sample for comparative purposes. Continuous data are suitable for—and are used for—several of the individual experiments. For the purpose of the general comparative behavioral samples, the FSP method, which combines the focal-scan sample with point samples, can provide a data base from which the following might be analyzed: l. The mean and variance of behavioral rates within treatments and thus the capability of testing for differences in responses between treatments. 2. Differences in behavioral rates between experiments.

2l3 3. Definition of rates of activity within experiments and extrapolation to rates measured in other experiments (e.g., the activity-level example discussed earlier). 4. Tests for correlations in response within treatments. For example, weight gains for a given nutritional level may show individual variability. Behavioral data gathered by FSP methods would permit differences in weight gain to be tested for correlation or regression with differences shown by the individuals in their activity levels. 5. Construction of individual profiles of animals. Animals may show serial dependency or compensation between behaviors. For example, individuals who are active in the morning may be inactive in the afternoon or vice versa. Population measures would not detect this effect if horses showed no preference for morning or afternoon activity. Consideration of individual energy budgets requires measurements that can be identified at the individual level. FSP Sampling Scheme l. Behaviors to be Recorded. The following recommendations are for minimum data recording. Recorded data are to reflect (a) activity states that provide information for calculating energy budgets and (b) social behaviors that characterize 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. Social interactions constitute a large body of often species-specific behaviors that will require precise and common definitions across experiments. Researchers involved in these studies should be required to standarize these definitions after initial work has begun. 2. Sample Animals. Where experimental design permits, FSP horse samples should include the dominant male and at least one subordinate male, two females, and a foal. For burros, whose groups are not as spatially coherent, samples cannot be simultaneous on different classes of animals. FSP burro samples should include territorial and nonterritorial males, females in groups, and females with offspring. In experiments where mixed groups will not be used—such as the reproduction/nutrition trials—the sufficiency of animals within a treatment will be determined by the variability of response with the treatment. 3. Sampling Schedule. To eliminate diurnal variability, the following schedule should be followed. FSP samples should be taken for 30 minutes each hour from 0600 to l800 once every 2 weeks. Points are to be recorded every 2 minutes within the 30-minute sample. Ideally, the full range of samples should be done on the same day so that they will provide daily profiles and allow the determination of interdependence of behaviors within the day. Two minutes has been

2l4 chosen as being a sufficiently long period for a trained observer to record behaviors and identify individuals for an FSP that includes up to 5 or 6 focal animals, while providing a reasonable number of sample points (l5 per 30-minute period). Behavioral Sampling for Specific Experiments Contraceptive Experiments on Penned Animals l. FSP a. Focal animals. The groups for this experiment should be composed of stallions and mares. Stallions of different social rank (if multi-male groups are used) should be chosen as focal animals. Mares should be picked to include animals of different ranks in both the control and treated (contraceptive) groups. b. Sampling schedule. The sampling schedule should be intensified during the breeding season so that good activity profiles are available for different phases of the breeding cycle. 2. Specific Behavior Sampling a. Rationale. Behavior of animals is important to the determination and functioning of their social groups. Since contraception changes the hormonal activity of the female, and hormones are often linked closely with behavior, it will be essential to measure the differences in behavior exhibited by normal and contraceptive mares. The effect of individual behavioral changes on the group's social dynamics will be monitored and the ecological and management implications of these effects may be suggested. b. Sampling scheme. Refer to paragraph (4) under the "Methodology" section of Project 6 (Chapter 2) for a description of this sampling scheme. Nutr ition/Reproduction Exper iments l. Purposes of the Data Collection a. To determine activity levels so that nutritional levels can be accurately generalized to free-roaming populations, in which activity levels are likely to be different. b. To correlate differences in reproductive performance and condition to activity measures. c. To compare nutritional treatment effects on activity. 2. Specific Behavior Sampling. When females in the experiment are mated, behavioral samples identical to those used in the contraceptive experiment should be recorded. (Sampling scheme has been detailed in the "Methodology" section of Project 6.) These data will show how the nutritional plane affects the sequence of sexual behaviors between the stallion and mare.

2l5 Field Feeding Trials (Small Enclosure) l. FSP. The purposes of the data collections are to compare activity levels between treatments and to be able to compare activity levels between experiments. 2. Specific Behavioral Sampling a. Rationale. Assessment of the feeding response of cattle and horses to inter- and intraspecific competition will require detailed observation of the feeding animals. The purpose of these samples will be to record diet and dietary proportions to assess treatment affects, to generalize feeding behavior within the habitat type of the small enclosure, and to record feeding behavior in sufficient detail to determine which characteristics of the food are important factors determining diet in horses and cattle. b. Sampling scheme. The sampling scheme for monitoring detailed feeding behavior should be a continuous focal-animal sample. Data are to be recorded in a continuous timed sequence on an individual animal. The schedule for sampling should be identical to the FSP schedule described in the previous section. Data are to be 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 (see subsections on the FSP sampling scheme and the contraceptive experiments for behavioral categories) should be recorded against a real-timed record. The identity and size of all plants and plant parts eaten are to be recorded. One must have this detailed record in order to understand the basis of dietary selection in the horse and to measure changes in feeding response when inter- and intraspecfic 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 also the specific rates at which particular 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. The results of these balance trials will serve as an intermediary condition for translating results of the highly controlled nutrition/reproduction feeding trials to the free-ranging feeding observations. 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. Habitat-Choice Experiments l. FSP. These data will allow the activity and social interaction rates to be compared between treatments within the

2l6 experiment and between experiments to determine if differences are due to interaction with cattle (as opposed to horses) or to the habitat choice of the horse. 2. Specific Behavioral Sampling a. Rationale. The purpose of these data is to measure the use of the space and habitat type within these enclosures by horses and cattle to determine the effect of intra- and interspecific competition on the use of space. b. Sampling scheme. At the beginning and end of each 30-minute FSP sample period, the location of the group being sampled is recorded on a topographical map of the enclosed area. The type of habitat occupied is also recorded. Because these locations are likely to be variable depending on the season and weather, sampling frequencies for this experiment are to be increased from 2 to 5 days per month. c. Other behavioral measures. Data on the feeding and social behavior of animals in this group would be of benefit, but the feasibility of recording data requiring closer observer contact with the animals is doubtful. Since the primary goal of the experiment is to determine use of space, any space-related reaction by the horse to observers will damage the accuracy of the results. The ability of the horse to become habituated to the observers will determine the degree to which other behavioral data can be recorded. Wild Populations with and without Contraception l. FSP. These data will be used for comparative measures with other experiments and comparisons of general activity between normal and contraceptive mares. 2. Specific Behavioral Sampling a. Rationale. Continuous focal-animal samples of activity, feeding, and social behavior are necessary to make specific comparisons. Activity is to be compared to levels in the feeding trials to determine how range conditions comparable to feeding levels would affect animals active under free-ranging conditions. Feeding behavior is to be compared with that measured in the small enclosures to determine if the feeding responses are similar in the enclosed and free-roaming condition. For example, does the increased amount of locomotion required of free-ranging animals change their dietary choices by shortening the amount of time spent feeding at each plant? Finally, social behavior—and in particular sexual behavior—is to be compared with measures in the penned contraceptive experiment to determine how the free-roaming populations respond to contraceptive implantation. Effects of the treatment on social behavior and organization may differ between the confined and free-roaming conditions.

2l7 b. Sample scheme. Since observational conditions and habituation of the animals to human observers will differ widely in free-roaming populations, a strict sample scheme cannot be projected at this time. The researcher should employ the same sampling methods and schedule for continuous focal studies as are used in the experiments to which comparisons will be made (i.e., sexual behavior should be recorded as per the instructions outlined in contraceptive experiments, feeding behavior as described under the small-enclosure feeding trials, etc.). c. Sampling schedule. Continuous focal-animal samples should be 30 minutes in length and conducted each hour between 0600 and l800, light permitting. These samples should follow one animal for the entire day. Each focal animal is to be sampled in this manner once a month. Focal horses should include at least one dominant male, two females (one contraceptive, one normal), one subordinate male, and one foal. Focal burros should include one territorial male, one nonterritorial male, two females (one normal, one contraceptive), and one foal. Spacing, location, and habitat data should be recorded at the beginning and end of each 30-minute sample. Behavioral Sampling Summary l. Contraceptive Experiments on Penned Animals a. FSP: 30-minute samples, l2/day, 2 months b. Continuous focal samples (CFS) on sexual behavior (schedule determined by researcher) c. Spatial data: beginning/end each 30-minute FSP sample 2. Nutrition/Reproduction Experiments a. FSP (schedule same as above) b. CFS during mating 3. Field Feeding Trials (Small Enclosure) a. FSP b. CFS on activity, feeding, social behavior: 30-minute samples, l2/day, 2 months c. Spatial data: beginning/end each FSP sample 4. Habitat-Choice Experiments a. FSP: 5/month b. Location/habitat data: beginning/end each FSP sample c. Other behaviors measured as possible without disturbance of animals 5. Wild Populations a. FSP: 2 months b. CFS on activity, feeding, social, and reproductive behavior c. Spatial data: beginning/end each FSP and CFS sample d. Location/habitat data: beginning/end each FSP and CFS sample

2l8 Contraception in the Horse Contraception is not being advocated in this section as a population-control measure in horse management. That is a matter of policy decision. It is the purpose of the following discussion to point out some of the technology that is available if contraception is chosen as a management tool to be used in a particular situation. The efficacy of the techniques can be evaluated in a relatively small sample of domestic horses but determining their usefulness in free-roaming populations will require field trials. Fertility control or sterilization of the harem stallion has been suggested as a means of limiting the growth rates of wild horse populations. This approach is based on the premises that the females of a harem or band are bred only by this male and that this social structure is in some measure constant. It assumes that the behavior will be constant while the females undergo repeated estrus cycles during the breeding season, and that the females will continue to have a limited breeding season. The approach is currently being tested in field trials using hormone injections effective for a single season. Evidence reported by investigators in Wyoming and Montana suggests that these assumptions may not apply to all wild-horse populations. Several males have been observed breeding females in estrus; males may change band affiliation; a given stallion may maintain dominance only about 4 years; and some mares spend long periods of time outside of a band. A study of burro bands, whose dominant males were vasectomized, found that the females were then bred by young males previously considered nonbreeders (McCort l979). These considerations, plus the fact that the reversible contraceptive techniques for males currently being tested are effective for only 6 months, indicate that this approach will require annual reapplication and will not be effective in horse populations with more fluid social structures. Computer simulation of population effects of female contraception indicate that the treatment may need to be effective for 5 to 7 years to prevent population growth or to achieve a decline in population numbers. Artificial manipulation of the estrous cycle in horses (Ginther l979) and other domestic animals has been oriented toward relatively short-term applications. These have included: (a) synchronization of estrus and breeding for scheduled pregnancies, especially for artificial insemination; (b) treatment of infertility; (c) early induction of estrus and ovulation; and (d) suppression of estrus in feedlot cattle to reduce activity and interactions. Long-term control of fertility has been intensively studied only in human beings, and more recently there has been work with hormones and mechanical devices in dogs and with hormones in a few exotic wild species (Seal and others l976) and in white-tailed deer (Matschke 1977a, b; Harder and Peterle l974). Approaches to fertility control that might have application for female horses include: (a) hormones, hormone analogs, and hormone antagonists; (b) mechanical devices such as intrauterine devices (lUDs) and pessaries; and (c) surgical intervention. Surgery—including a laparoscopy approach for females—would be

2l9 impractical under field conditions because of the infection risk and the time required for each animal. Surgery is also permanent. The IUD has been an appealing possibility because its effectiveness lasts indefinitely, it is potentially reversible, and it would not interfere with normal hormonal and behavioral patterns except for pregnancy. However, it is unsuitable (a) because variations in uterine size require individual fitting; (b) considerable care and skill is necessary in placing the device so as to avoid damage and infection; and (c) retention has been variable in some species. Detailed, extensive information is available on the reproductive cycle of the domestic mare and stallion (Ginther l979, Rowlands and others l975). The feral mare appears to be more strongly seasonal than the domestic mare (Kirkpatrick and others l979b). However, published data on the long-term effectiveness of contraceptive agents in horses are nonexistent. The literature on artificial control of the mare's estrous cycle is oriented towards synchronization of estrus, treatment of infertility, and early induction of estrus and ovulation (Ginther l979). Zimbleman and others (l970), in studying melengestrol acetate, found that intravenous administration of this agent had no effect on the occurrence of estrus. However, the study by Ganjam and others (l975) offered a rationale for the failure of this approach. They demonstrated that there is little or no specific plasma-protein binding of progesterone in the horse, and so the half-life of progesterone (2.5 minutes for compartmental mixing and 20 minutes for disappearance) and presumably other progestagens is very short. This observation would also explain the fact that very large daily doses of progesterone have been required to inhibit estrus (l00 mg/day) and ovulation (200 mg/day) in the mare (Loy and Swan l966). Such information suggests that a different approach needs to be taken. Other compounds tested in domestic horses include chlormadinone acetate (Arbeiter and Jochle l975) and an orally administered synthetic progestin, l7-a-allyl-estratiene-4-9-ll, l7-B-ol-3-one (Webel l975). The latter compound (allyl trenbolone) has potential for use in implants. It has been administered to pregnant horses with no apparent adverse effects. A range of contraceptive agents is available, and effective implants are either available or can be produced that release small amounts of active agents on a continuous basis for periods up to 5 years, depending upon the compound and the nature of the implant (Dziuk and Cook l966, Kinel and Rudel l97l). They have been tested in humans in extended field trials, and several agents have been used in more than 400 animals now covering a range of species (Seal and others l976). Given the logic for contraception in the mare (Nelson l979, Miller l979), these agents would appear to be a choice for testing in females. Implants can be installed in minutes under field conditions, but they require handling of the animal. Preparations that could be delivered from a dart gun are available, but they are only effective for about one year in their present form. The animals must be identified in some permanent fashion and appropriate records maintained so that the effectiveness can be monitored. A full 5 years

220 of study of domestic animals would not be required to determine the likely span of effectiveness, since it is possible to measure release rates of the compounds from the implants as well as their levels in the blood, and thus to predict their effective lifespan. It is only necessary to establish that this mode of administration and the particular compounds chosen would provide effective contraception. Studies of domestic mares to determine the effectiveness of this approach might employ the following design: use animals with proven reproductive histories, install the implants, observe the animals for signs of estrus, collect blood samples for hormone assay, use rectal palpation or serum progesterone to determine if ovulation has occurred, test with a stallion to detect behavioral responses characteristic of estrus, and test for pregnancy if breeding occurs. It is possible that estrous behavior might occur but that ovulation might not. It is also possible that ovulation and breeding could occur in some circumstances, and yet the condition of the fallopian tubes or uterus could prevent implantation. In summary, the use of reversible endocrine contraception in mares or stallions is feasible from an endocrine point of view. Single treatments could provide periods of effectiveness ranging from l to 5 years depending upon the hormones used and the mode of delivery selected. The choice of a particular strategy would depend upon the characteristics of the specific horse population and the management constraints. Chemical Immobilization and Capture of Wild Equids The capture of wild equids and their restraint and handling in zoos and nature preserves has provided an impetus for development and testing of chemical immobilization agents (Harthoorn l976, Haigh l978, Fowler l978). Use of these drug techniques requires considerable skill and is best accomplished by people with adequate knowledge of physiology and anesthesiology if losses are to be kept to a minimum. Many of the problems that have arisen with immobilization drugs have been the result of their administration by inexperienced and untrained personnel (Harthoorn l976, Haigh l978). Veterinary practitioners also commonly have to deal with horses at pasture and fractious animals that cannot be handled by standard techniques. The experiences have resulted in the development of techniques for using tranquilizing and immobilizing agents. Chemical immobilization of individuals is not an efficient procedure for the primary capture of large numbers of animals within a reasonable period of time. It is frequently more effective to employ appropriate trapping, corralling, or herding techniques. If necessary after capture, animals can be immobilized by drugs for further handling, but it is possible to use other handling techniques. Chemical immobilization can be effective and useful when applied selectively by trained personnel and when it is appreciated that it is a time-intensive technique (Harthoorn l976). It is possible to come within shooting range of horses or burros with dart guns by pursuing them in helicopters, waiting at water

22l holes, or stalking. The range of the currently available rifles for projecting darts varies from 30 to about l00 yards. Capture of free-ranging wild animals using dart-gun techniques has been facilitated by the recent development and successful use of a small radio transmitter suitable for use with darts. This device has a range of up to a quarter of a mile, depending upon terrain, and allows animals to be located and followed in rough country after they have been shot. Discussion of the drugs used for immobilizing equids can be found in Harthoorn (l974, l976), who did much of the work on developing techniques for zebra capture in Africa. It appears that either of two drugs can be used for capture if circumstances are appropriate. One is succinylcholine, a muscle relaxant, which is neither a tranquilizer nor an anesthetic agent. It can be administered intramuscularly by way of a dart gun, and when used successfully can immobilize an animal within 5 to l0 minutes. The dosage of succinylcholine needed for capture can vary with season of the year, sex, and age, so a suitable dose range for a particular population would need to be established from experience. Borchard and others (l979) have recorded experience with this agent, and in skilled hands the death rate would probably run no more than 2 to l0 percent. Respiratory arrest in successfully immobilized animals may occur in 20 percent or more of animals. Therefore, rapid access to the animal is necessary and the ability to support respiration is essential. The use of succinylcholine has been greatly restricted by court orders based upon humanitarian concerns. Animals do not lose consciousness or sense of pain with this drug. The other drug is etorphine, a morphine-type compound that is effective at low doses. It acts as an immobilizing agent, an analgesic, and an anesthetic (Harthoorn l976, Bogan and others l978, Hillidge and Lies l978). There is an antagonist available for use with this drug—diprenorphine—that reverses the immobilizing drug's effects. An animal can recover spontaneously over a period of time ranging from 30 minutes to 2 hours without administration of the antidote. Etorphine appears to be a suitable drug for capturing wild equids in the field: it also can be used with the wild ass (Woodward l979). Etorphine is customarily used in combination with acepromazine or xylazine. Effective doses of etorphine range for equids from 5 to l0 mg per animal. Equids have a wide tolerance for this compound, and as much as l0 times the effective dose has been administered without fatal results. Xylazine has been used alone and in combination with etorphine for capture of wild equids. It should be used with great care in combination with etorphine, since both are respiratory depressants and the combined effect can lead to respiratory failure, which, if not treated promptly, results in death. Xylazine, in combination with hetamine, has been used for anesthesia of domestic horses. The combination is also being used for immobilization of many wild species and may be useful for wild equids. Xylazine can effect dramatic alterations in metabolic pathways (elevations in glucose and urea) that would make blood samples useless for evaluating the nutritional status of the animal (Short and others l972, Eichner and others l979).

222 If succinylcholine cannot be used for horse immobilization, it will be necessary to test other drugs and combinations, such as etorphine with a tranquilizer, or xylazine and hetamine (Alford and others l974, Hoffman l974, Holmes and Clark l977, Kaha and others l979, MacKenzie and Snow l977, Muir and others l978). NEEDED RESEARCH Project l7. Census Methods for Wild Horses and Burros Rationale Section l4(b)(l) of PL 95-5l4 states that the Secretary of the Department of the Interior shall "maintain a current inventory of wild free-roaming horses and burros." Effective management, both in terms of resource use and communication with the interested public, requires that such inventory data be accurate and reliable. Little is known about the accuracy of the existing estimates. There is thus an urgent need for research to evaluate the existing methods, and to develop satisfactory alternate methods where needed. A research project is outlined below that would investigate three approaches to horse and burro census: "complete" counts (the approach currently used by the management agencies), mark-resight estimates, and strip-transect estimates. The three approaches would be used together in each study area to serve as checks against one another, and to determine which provides the greatest accuracy and precision: (a) under a given set of environmental conditions (terrain, vegetation, etc.) and (b) for horses or burros. For the present, quadrat and line-transect estimates are not suggested for inclusion in the initial study. When an initial set of data has been collected and analyzed, it may be found desirable to test these and other methods. This project might benefit from being directed by two people: one with expertise in statistical methodology—especially with interest and experience in animal-population measurement—and the other a wildlife biologist with experience in aerial census. This suggestion does not preclude direction by one person with adequate training and experience in both areas. The first project will direct attention only to wild horses, and will be considered a pilot effort. Subsequent efforts will investigate burro census and alternate census methodology, if adequate progress in the work with horses and sufficient funding are assured. Objectives l. Test the accuracy and precision of three approaches to wild horse census: a. Complete counts, including bounded counts b. Mark-resight estimates c. Strip-transect estimates

223 2. Develop a set of criteria by which to choose the appropriate approach for a given area with particular habitat characteristics. 3. Investigate the role of such variables as weather, vegetation, terrain, herd size, and horse distribution on probability of observation. 4. Prepare a report that would outline procedural details for carrying out the three approaches. It should especially document any further work needed to establish methodology (including a reference manual) for future BLM censuses. Methodology l. Select, with the help of BLM officials, three or four areas in the western United States in which to evaluate the three approaches. It is desirable that these areas contain geographically discrete populations—physiographically isolated or fenced if possible—to minimize or prevent ingress and egress. Modifications of the methodology suggested here may be required for individual sites. 2. Conduct a typical complete-count census of each area with helicopters in late summer or early fall. 3. As soon as possible after censusing, mark approximately half (if possible) of the horses on each area from helicopters with paint capsules. Costs may impose a limitation on the proportion marked, and a cost function should be applied to obtain an optimal balance between effort expended in marking and resighting. 4. After marking, conduct a series of overflights at 3- to 7-day intervals, counting the number of marked and unmarked animals seen and tallying group sizes. Two overflight patterns are possible. One consists of parallel, straight-line flights across the entire area, spaced at l- or 2-mile intervals and covering the entire area. Each such set could be oriented toward a different compass direction, and should be selected a priori without relation to the area's terrain and vegetational characteristics or to the locations of marked horses. The second alternative is to use random starting intervals rather than the equidistant ones oriented to random compass directions. Terrain or visibility factors may dictate this option. 5. Use additional overflights for strip-transect sampling of horses in the line of flight. First priority should go to the mark-and-resight effort as a check on the complete counts. However, strip transects may be deemed especially suitable for some sites. 6. As soon as possible after the overflights, conduct two or three more complete counts from helicopters to provide an array of values for the bounded-count method of checking the accuracy of the complete counts. This is regarded as having lowest priority. 7. With regard to objective 3, conduct intensive interviews with personnel experienced in large-animal census to gain information on the effects of vegetative, climatic, and equid behavioral variables on observability. This step should be an initial one, and may lead to necessary modification of the proposed methodology. Group size should be investigated as a factor affecting observability.

224 8. Take opportunities to conduct surveys over ground-truthed research projects wherever possible and convenient. 9. Encourage potential contractors to offer improvements to the above procedures. Tests of complete-count methods are, however, regarded as essential. Project l8. Contraception Studies Rationale Those who manage wild equids will have to deal with multiple populations of differing size, composition, and accessibility. If population control is desired, but removal techniques cannot be applied, contraception might be a useful technique. The studies proposed here will provide data on short-term effectiveness in mares. The long-term limits of effectiveness can be projected by determining blood levels of the compounds and by calculating measured rates of release. Tests in stallions in wild bands and in captivity are in progress. They are designed to be effective over the span of a single breeding season (Kirkpatrick and others l979a). Compounds effective in mares for several seasons could be adjusted to last for only a single season. Objective Establish a method for reproductive inhibition in female horses and burros that (a) is at least 95 percent effective (i.e., of l00 animals treated, no more than 5 become pregnant during the specified test period); (b) requires only a single treatment or administration under field conditions; (c) could be adjusted to last up to 7 years; (d) is potentially reversible; and (e) will not adversely affect the health or behavior of the animal. Methodology l. Establish five experimental treatments each of five captive mares, as follows: a. Control: no treatment (placebo/vehicle) b. Compound A: estimated dose for 5 years c. Compound A: five times base dose d. Compound B: estimated dose for 5 years e. Compound B: five times base dose. Use nonpregnant breeding-age females (4 years of age or older) , preferably ones whose reproductive status has been established (i.e., those known either to have cycled or to have been pregnant the past year). 2. Place implants intramuscularly early in the breeding season, if possible. Consideration should be given to implants of solid

225 silastic polymer in the form of rods. Use known amounts of drug in each implant. 3. Measure the following: a. Reproduction Estrus: does it occur or is it completely suppressed? (Observe behavior and appearance of genitalia.) Ovulation, determined by rectal palpation and serum progesterone Breeding: will the female accept the stallion? Pregnancy, established by rectal palpation and hormones Foaling Foal heat. In this sequence, if an agreed-upon percentage of the animals becomes pregnant, the trial is a failure and should be terminated. b. Condition Hematology (CBC) Blood chemistry High priority Lower priority Serum urea nitrogen Free fatty acids (NEFA) Glucose Ketones Triglycerides Haptoglobin Serum protein and albumin Transferrin Uric acid CPK Bilirubin Behavior Detailed guidance for measuring behavior can be found earlier in this chapter ("Behavioral Sampling for Specific Experiments: Contraceptive Experiments on Penned Animals," and in Chapter 2 ("Project 6: "Methodology," paragraph 4). Hormone levels: the following should be measured regularly: Blood levels of contraceptive hormone Serum progesterone Rates of release of contraceptive from device

Next: REFERENCES »
Wild and Free-Roaming Horses and Burros: Current Knowledge and Recommended Research. Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!