5
Demographic Rates

Knowledge of demographic rates and trends are essential for accurate assessments of sea-turtle populations, as outlined in the discussion of the conceptual model in Chapter 3. The reasons for changes in sea-turtle abundance cannot be diagnosed—nor can management plans to mitigate declines in populations be developed—without demographic knowledge. This chapter introduces the various demographic parameters and methods for generating estimates of them. Chapter 6 describes applications of the different demographic parameters further.

All demographic parameters exhibit variation within and among species and populations and over space and time; some—such as clutch frequency (i.e., the number of clutches deposited by an individual turtle in a nesting season), interbreeding intervals, and somatic growth rates—vary within individuals over time. To develop an accurate assessment, such data need to be collected on all populations, on large spatial scales, and over many years. Caution is needed when extrapolating estimates between species and populations and even within populations for different years and habitats. However, estimation that accounts for variation is expensive. Methods to estimate demographic parameters at reasonable cost are needed so that they can be monitored frequently to detect changes. Moreover, estimation of variance about the mean, not just point estimates, is critical.

The ecological context of demography—that is, the key environmental mechanisms that regulate demographic rates, such as resource availability, temperature, current systems, and oceanic productivity—is necessary for



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 73
5 Demographic Rates Knowledge of demographic rates and trends are essential for accu ­ rate assessments of sea­turtle populations, as outlined in the discussion of the conceptual model in Chapter 3. The reasons for changes in sea­ turtle abundance cannot be diagnosed—nor can management plans to mitigate declines in populations be developed—without demographic knowledge. This chapter introduces the various demographic parameters and methods for generating estimates of them. Chapter 6 describes appli ­ cations of the different demographic parameters further. All demographic parameters exhibit variation within and among spe­ cies and populations and over space and time; some—such as clutch frequency (i.e., the number of clutches deposited by an individual turtle in a nesting season), interbreeding intervals, and somatic growth rates— vary within individuals over time. To develop an accurate assessment, such data need to be collected on all populations, on large spatial scales, and over many years. Caution is needed when extrapolating estimates between species and populations and even within populations for differ­ ent years and habitats. However, estimation that accounts for variation is expensive. Methods to estimate demographic parameters at reason ­ able cost are needed so that they can be monitored frequently to detect changes. Moreover, estimation of variance about the mean, not just point estimates, is critical. The ecological context of demography—that is, the key environmental mechanisms that regulate demographic rates, such as resource availability, temperature, current systems, and oceanic productivity—is necessary for 

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS understanding sea­turtle population status and trends fully. That knowl­ edge is critical for predicting the changes in sea­turtle populations that will occur with climate change and with oceanic regime shifts that have profound effects on many important sea­turtle habitats. Demographic parameters are not of equivalent value for diagnosing status and trends in populations. Some vital rates are influenced more than others by environmental factors—probably acting largely through nutrition. For example, nutrition affects age at sexual maturity, clutch frequency, and the number of years between breeding seasons, but it does not affect clutch size (Bjorndal, 1985). In populations with ample high­ quality food, somatic growth rates, body condition, and clutch frequency will be high, and interbreeding intervals will be small. Populations that have poor food resources or that are approaching carrying capacity, at which competition for food is high, will exhibit the opposite. BREEDING RATES AND ADuLT-RECRuITMENT PROBABILITIES In most species of sea turtles, females generally do not reproduce in consecutive years but at variable intervals of two years or more. The probability that a female will reproduce in any given year (breeding rate) is affected by nutrition (Bjorndal, 1985), environmental factors, and migra­ tion distance between foraging grounds and nesting beaches (Limpus and Nicholls, 2000; Solow et al., 2002; Troëng and Chaloupka, 2007). Knowl ­ edge of breeding rates is critical for understanding the highly variable numbers of clutches deposited in successive years on nesting beaches (Hays, 2000; Broderick et al., 2001; Solow et al., 2002) and for interpreting population trends. Estimates of breeding rates of females have been derived from mark– recapture studies on nesting beaches using an “open robust design”—a specific mark–recapture method—with hawksbills (Eretmochelys imbricata; Kendall and Bjorkland, 2001) and leatherbacks (Dermochelys coriacea; Dutton et al., 2005). Mean remigration interval (the number of years between successive breeding seasons) has been estimated more com­ monly in sea­turtle studies and approximates the inverse of breeding rate. Although not as useful as breeding rate for demographic models, the remigration interval does offer important insights into the productiv ­ ity of the population and population density relative to carrying capacity (Saba et al., 2007; Troëng and Chaloupka, 2007). Remigration interval is usually measured as the number of years that elapse between sightings of individual tagged females at a nesting beach. Thus, values are biased to shorter intervals because of tag loss and human­induced mortality in that the probability of both factors increases with the length of the remigration interval. Values are biased to longer intervals when incomplete sampling

OCR for page 73
 DEMOGRAPHIC RATES on the nesting beach results in females being missed in intervening breed­ ing seasons. Breeding rates of male sea turtles have been poorly studied, and more information is needed. Males may breed at greater frequency than females; substantial proportions of males may breed annually (Hamann et al., 2003). Newer techniques, such as ultrasound, are useful and mini ­ mally invasive for evaluating the reproductive condition of adult sea turtles of both sexes. If the kidney is located first as a landmark in the male, the size and density of the testis and epididymis (parts of the male reproductive system) can be determined and the diameters of epididymal tubules measured for comparative studies (Blanvillain et al., 2008). Male breeding rates will inform our understanding of the proportion of males in a population required for successful reproduction and our understand­ ing of possible depensation effects. (Depensation is described in the sec­ tion “Density Dependence” later in this chapter.) Recruitment of females into the breeding population and the propor­ tion of first­time breeders in a nesting population are critical for assessing population trends. For example, if a nesting population is increasing in abundance, is the increase the result of increased recruitment of first­ time breeders, increased survival of mature females, or both? In nesting populations subject to saturation tagging (tagging of every female) for a duration longer than the remigration interval with no loss of individual identification through tag loss and no immigration due to low fidelity, recruitment can be measured directly as the number of females that arrive with no tags (Richardson et al., 2006; Dutton et al., 2007). Few studies, however, meet those requirements. Another technique, laparoscopy, can be performed on female sea turtles at rookeries to determine the propor­ tion of females that are first­time breeders or performed on foraging grounds to assess the proportion of female recruits that are preparing to breed in that year (Hamann et al., 2003). However, a method that is less invasive and more rapid is needed to distinguish recruits from females that have nested in previous seasons. FECuNDITY Fecundity is the reproductive output of an individual or a popu­ lation. In sea turtles, fecundity is usually measured as the number of eggs deposited during a nesting season, which when combined with breeding rate (see above) yields an estimate of lifetime fecundity (aver­ age breeding rate multiplied by average reproductive lifespan). Within a nesting season, egg output of an individual is the product of the num ­ ber of clutches deposited (clutch frequency) and the number of eggs in each clutch (clutch size). Egg size is usually not considered a measure of

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS fecundity. However, because egg size is both a predictor of offspring qual­ ity and a component of estimates of resource allocation to offspring pro ­ duction, it is included in this discussion. Production of healthy hatchlings is another and perhaps better measure of fecundity than the production of eggs; therefore, the committee also addresses fertility, temperature­ dependent sex determination, and hatching success. Egg Production Clutch frequency is an extremely important demographic parameter for both population models and assessment of trends in population abundance. Many monitoring programs on nesting beaches rely on nest counts to generate estimates of and trends in population abundance with the explicit assumption that clutch frequency is constant, but clutch fre ­ quency requires continual monitoring because it may vary among years (Broderick et al., 2003). For example, clutch frequency varied substan­ tially with quality of nutrition in green turtles (Chelonia mydas; Bjorndal, 1985), and this indicates that changing resource and environmental condi­ tions affect clutch frequency. Attempts to measure clutch frequency have been based largely on saturation­tagging programs on nesting beaches. Because of the length of the reproductive interval and the distance over which females deposit nests in a given season, intercepting females at each emergence is challenging. In Florida, for example, individual logger­ heads (Caretta caretta) have been recorded nesting up to eight times in one season over an 82­day interval (Tucker, 2009), and one female deposited nests over a range of 182 km along the east coast of Florida during one season (Bjorndal et al., 1983). Hence, many published estimates of clutch frequency need to be viewed with caution. Other approaches have been used to estimate clutch frequency and deserve further development. Radio and satellite telemetry have both been used. Radio telemetry is limited by the relatively short transmission distance and labor­intensive nature of monitoring. The relatively large location error of satellite telemetry has limited its application but does not preclude its application (Tucker, 2009). This technology will become more valuable as telemetry systems that generate more accurate locations are developed. Rivalan et al. (2006b) estimated clutch frequency in leather­ backs in French Guiana by using mark–recapture data to model stopover duration. A recent initiative used genetic markers from one egg in each clutch deposited in Georgia to identify the individual female that had deposited the clutch and thus the number of clutches deposited by each female (Brian Shamblin, personal communication). Methods of estimating clutch frequency that are relatively inexpensive and can be applied repeat­ edly on nesting beaches around the world are greatly needed.

OCR for page 73
 DEMOGRAPHIC RATES Clutch size may be the only demographic parameter on which there are adequate data. The most accurate counts of clutch size are made dur­ ing the egg­laying process, but with proper training and experience, accu­ rate egg counts can be determined from pieces of egg shells during nest inventories after hatchlings have emerged (Miller, 1999). Unlike clutch frequency, clutch size apparently is not greatly affected by environmental factors (Bjorndal, 1985; Bjorndal and Carr, 1989) although it does vary substantially within populations or individuals over time (van Buskirk and Crowder, 1994; Broderick et al., 2003). Female body size accounts for some of the variation, as does time within the nesting season (Frazer and Richardson, 1985; van Buskirk and Crowder, 1994; Broderick et al., 2003). A better understanding of the reasons for the variation would be valuable for determining the importance of clutch size as a basis of population assessment. For a parameter that is so easily measured, there are surprisingly few data on sea­turtle egg size. Egg size is measured most commonly as egg diameter, but egg mass and volume have also been measured. Egg size varies widely among sea­turtle species (van Buskirk and Crowder, 1994) and perhaps between populations and individuals of the same species. Substantial variation in hatchling size has been shown recently in flatback (Natator depressus) populations (Whiting et al., 2008). Accounting for the variation in egg size, evaluating the relationship of egg size to hatchling size (see van Buskirk and Crowder, 1994), and determining whether egg size is substantially affected by environmental factors would be valu ­ able in assessing the importance of egg size as a factor in population assessment. Hatchling Production Survival from egg deposition to the emergence of hatchlings from the nest is the best quantified life stage of sea turtles. Given the accessibility of this stage, however, the number of quantitative studies, particularly for natural nests that have been subject to management interventions, is surprisingly low (National Research Council, 1990). The paucity of published data is due primarily to three factors: the difficulty of marking and following nests, the substantially longer monitoring period that is required to quantify hatching success throughout the season, and the lack of publication of many of the studies. Determining hatching success is critical for the assessment of sea­turtle populations. Relying solely on the number of nests deposited to estimate hatchling production can lead to serious overestimates. In Tortuguero, Costa Rica, which has the largest green turtle Atlantic nesting popula ­ tion, Horikoshi (1992) reported that hatchling success was substantially

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS reduced by high groundwater that drowned many nests, although no problem was apparent in surface observation of the beach. Many natural and anthropogenic factors can affect embryo survival and reduce hatching success (Lutcavage et al., 1997). Techniques for eval­ uating hatching success have been summarized by Miller (1999). Loss to predators—both natural predator populations and those introduced or subsidized by humans—can be high (Stancyk, 1982). For example, raccoon populations that had increased above natural numbers as a result of human activities were responsible for predation of up to 97% of loggerhead nests on some Florida beaches (National Research Council, 1990). Although the fertility of eggs deposited by sea turtles is gener­ ally high, probably exceeding 95% (Miller, 1997; Bell et al., 2003), low egg fertility can be a problem so egg fertility needs to be monitored in studies of hatching success. Decreased egg fertility of leatherback eggs in Terengganu, Malaysia (Chan, 1989), probably resulting from a reduc­ tion in the ratio of males to females, has been identified as a factor in the dramatic decline of nesting in that rookery (Chan and Liew, 1996). All species of sea turtle exhibit temperature­dependent sex deter­ mination (Wibbels, 2003). That is, the temperature at which an embryo develops is primarily responsible for determining the sex of the hatchling (but see LeBlanc and Wibbels, 2009). In sea turtles, females are produced at higher temperatures and males at lower temperatures. Therefore, the primary sex ratio—the sex ratio of hatchlings—can vary greatly among clutches, among months within a nesting season, among nesting seasons, and among nesting beaches. Environmental changes, such as construc­ tion of tall buildings in Florida that shade the beach and reduce sand temperatures (Mrosovsky et al., 1995) and removal of trees behind the nesting beach in Terengganu that result in higher sand temperatures (Chan and Liew, 1996), can have substantial effects on the sex ratio of hatchlings. Hatchling sex can be identified reliably only with gonad his ­ tology or morphology (Ceriani and Wyneken, 2008); a nonlethal, accurate technique that could be used on a large number of hatchlings is greatly needed (Wibbels, 2003). Such a technique will be critical for monitoring responses of populations to climate change. As temperatures increase, pri­ mary sex ratios may shift toward females. Because many nesting beaches already produce primary sex ratios strongly biased toward females, there is concern that the proportion of males will be insufficient and that fertil ­ ity of eggs could decline (Hawkes et al., 2007; Poloczanska et al., 2009). Laparoscopy may constitute a nonlethal technique for determining sex in hatchlings (Wyneken et al., 2007) and needs to be investigated.

OCR for page 73
 DEMOGRAPHIC RATES SuRvIvAL PROBABILITIES One of the greatest gaps in developing the conceptual model is in esti­ mates of survival of immature turtles and nesting females of all species. Survival of turtles through embryonic development to their emergence from the nests is discussed above (see the section “Fecundity”). Adult Females Estimates of survival for adult females have been derived from mark– recapture studies that used open robust design for hawksbills (Kendall and Bjorkland, 2001) and leatherbacks (Dutton et al., 2005). This analysis is the best available approach for estimating survival probabilities based on mark–recapture data on nesting beaches if sufficient data are avail ­ able. Survival estimates have also been generated from recovery analyses (Campbell and Lagueux, 2005; Troëng and Chaloupka, 2007) and a model of remigration intervals (Solow et al., 2002). Applying more than one approach to a population can increase confidence if the independently derived estimates are similar. For green turtles nesting at Tortuguero, Costa Rica, four analyses that used three techniques yielded similar esti ­ mates of probabilities of annual survival of adult females (Solow et al., 2002; Campbell and Lagueux, 2005; Troëng and Chaloupka, 2007). Despite multiple calls for new studies (see Table 1.2; Turtle Expert Working Group, 2000; Heppell et al., 2003), there have been few attempts to update estimates of survival of loggerhead turtles nesting in the United States with mark–recapture analysis (e.g., Hedges, 2007), and current models still rely on results from the 1970s when mark–recapture studies were conducted on Little Cumberland Island, Georgia (Richardson et al., 1978; Frazer, 1983). The survival rates from those studies were not estimated with the open robust­design methods that have been devel­ oped to account for detectability of nesting females (Kendall and Nichols, 2002) but did account for tag loss. Efforts to assess loggerhead status and interpret trends in nests with lifecycle and simulation models have been stymied by the lack of new estimates (Turtle Expert Working Group, 2000; National Marine Fisheries Service Southeast Fisheries Science Center, 2001). That has also prevented proper evaluation of the effectiveness of management actions, such as the implementation of turtle excluder devices (Epperly and Teas, 2002). Survival of nesting female Kemp’s ridley (Lepidochelys kempii) turtles was estimated in a model­fitting exercise in which a simple age­structured model was fitted to nest census counts from Mexico to obtain a point estimate of annual survival before and after 1990 (Turtle Expert Working Group, 2000; Heppell et al., 2005). That was a unique circumstance in that all nesting of this highly endangered species was restricted largely

OCR for page 73
0 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS to one well­monitored nesting beach, and the population had exhibited changes in trends that provided contrast for model fitting. However, the estimate was not empirically based, and additional analysis of existing mark–recapture data on females tagged with passive integrated transpon­ ders (PITs) is needed (Heppell et al., 2007). Immature Turtles The paucity of estimates of annual survival of immature sea turtles on their oceanic and neritic (nearshore) foraging grounds limits the ability to assess sea­turtle populations. Mark–recapture models based on tag­ ging studies (Chaloupka and Limpus, 2002, 2005; Bjorndal et al., 2003c; Campbell and Lagueux, 2005; Braun­McNeill et al., 2007) and catch­curve analyses (Frazer, 1987; Bjorndal et al., 2003b) have been used to generate estimates. A serious limitation of both approaches, particularly in Atlantic populations in which immature turtles tend to move among foraging grounds to a greater extent than in the Pacific, is the confounding of emi ­ gration and mortality in estimates of apparent survival (usually referred to as phi). Differences between apparent survival and true survival can be substantial in populations of immature sea turtles (Bjorndal et al., 2003c). Estimates of survival not confounded with emigration are possible with Burnham models (Burnham, 1993; Catchpole et al., 1998), joint analyses of live­recapture and dead­recovery data (Bjorndal et al., 2003c; Seminoff et al., 2003), if sufficient data are available. Transients, which are usually identified as marked animals seen only once in a study area, can lead to biased estimates of survival probability (Pradel et al., 1997). Accounting explicitly for transient behavior of marked sea turtles has been under­ taken in a few studies of sea­turtle survival probabilities (Chaloupka and Limpus, 2002; Sasso et al., 2006) but needs to be explored further. Another common technique in fisheries, catch­curve analyses, requires knowledge of size­at­age, which can limit applications to sea­turtle populations, and needs to incorporate differential growth rates and recruitment. Data on strandings of sea­turtle carcasses cannot be used to estimate survival probabilities. However, stranded carcasses can be used to assess abrupt changes in mortality due to changes in fisheries or disease out­ breaks or to track the incidence of diseases. Stranding data are most valu­ able in hazard­specific analyses (Crowder et al., 1995; Chaloupka et al., 2008b) because the proportion of the population represented by stranded turtles is unknown. A major anthropogenic hazard for sea turtles worldwide is incidental capture in shallow­set pelagic longline fisheries (Lewison et al., 2004). Many turtles caught in such fisheries are alive when released from the gear (Gilman et al., 2007), but it is widely assumed that a substantial number

OCR for page 73
 DEMOGRAPHIC RATES will die soon after because of injuries caused by hooks or line entanglement (Lewison et al., 2004). However, there are few reliable estimates of post­ release mortality in sea­turtle species despite their being essential for risk assessment and hazard mitigation. Chaloupka et al. (2004a) and Sasso and Epperly (2007) used satellite telemetry to estimate post­hooking mortality in loggerhead sea turtles but pointed out limitations of the method, includ­ ing inadequate sample sizes and premature release of satellite tags, that make it difficult to derive reliable cause­specific mortality estimates. DISPERSAL PROBABILITIES Movement of Adult Females Between Rookeries To date, all measured probabilities of female movements between rookeries are too low to influence management plans. Nesting females are highly philopatric (i.e., they return to their birthplace), but the degree of site specificity varies among species. Loggerhead nesting populations may show population structure (mitochondrial DNA differentiation) on a scale of less than 100 km (Bowen et al., 2005), green turtles on a scale of 500 km (Dethmers et al., 2006), and olive ridley (Lepidochelys olivacea) and leatherback turtles on a scale of more than 500 km (Lopez­Castro and Rocha­Olivares, 2005; Dutton et al., 2007). That information is important because the degree of site specificity and the scale of population structure determine the appropriate sizes of management units (see Chapter 2) and determine the extent to which nesting populations will reinforce each other. Those geographic scales are supported in some species by tag­recapture data from renesting females. However, long­distance relocations of nest­ ing females (beyond the geographic ranges outlined above) have been documented. LeBuff (1974) reported a loggerhead female relocating from southwest Florida to southeast Florida, and at least two tagged females have switched from Tortuguero to other locations in the Caribbean (cita­ tions in Bowen et al., 1992). A low level of switching between nesting sites is beneficial and probably necessary for the long­term persistence of sea­ turtle species. In view of epochal changes in climate, oceanography, and geography, the appropriate nesting sites of, for example, the Pliocene are not the same as the ones today. Shifting among nesting beaches allows sea turtles to respond to a changing world. Dispersal of Immature Sea Turtles Immature sea turtles generally undergo two phases of dispersal (both of which are poorly understood): (1) hatchlings disperse away from the

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS nesting beach into oceanic habitats after emergence from the nest and (2) immature turtles disperse from oceanic habitats when they recruit to neritic habitats, usually years before reaching sexual maturity. Once on neritic foraging grounds, immature turtles tend to move among foraging habitats. Knowledge of movements of immature sea turtles has improved through increased flipper tagging of immatures, satellite telemetry, genet­ ics, and stable isotopes and has revealed a more complex series of dis­ persals of some turtles (Eckert and Martins, 1989; Eckert, 2002; Bolten, 2003a; Harrison and Bjorndal, 2006; McClellan and Read, 2007; Reich et al., 2007). Evaluation of dispersal of hatchlings has been limited to direct obser­ vations (Frick, 1976; Witherington, 1991), tissue transplants or “living tags” (Wood and Wood, 1985), shell notching (Limpus, 2009), and evaluation of current patterns (Blumenthal et al., 2009b). Over 43,000 Kemp’s ridley hatchlings were marked with internal wire tags in 1996–2000 (Caillouet, 1998; Snover et al., 2007). All those techniques have well­documented limitations. The greatest challenge for any mass­hatchling tagging pro­ gram (e.g., with wire tags or PIT tags) is to intercept and recognize these marked turtles in their juvenile stages. The feasibility of an improved program of marking large numbers of hatchlings so that they can be rec­ ognized when they appear in oceanic or neritic foraging grounds could be explored. In 2009, neonate loggerheads were tracked successfully with highly miniaturized satellite transmitters that had been designed for birds (Mansfield et al., 2009b). In addition, application of hatchling­dispersal models coupled with multitrophic biophysical models,1 such as the Spatial Ecosystem and Population Dynamics Model (Lehodey et al., 2008), now being applied to pelagic fish can be used to predict movements and habi ­ tat occupancy through the first years of life. The recruitment of sea turtles from oceanic to neritic habitats can occur over a range of sizes and, presumably, ages (Bolten, 2003b). Sufficient numbers of recruits from the mass tagging of Kemp’s ridley hatchlings with internal wire tags were identified to estimate the age of recruitment as 2.2 years (Dodge et al., 2007). Identifying new recruits on neritic for­ aging grounds is challenging; a number of techniques have been used but with uncertain success. Arrival of turtles without tags in areas with saturation tagging (Bjorndal and Bolten, 2008) with epibionts (organisms that live on the surface of other living organisms) from oceanic habitats (Limpus and Limpus, 2003a) or, in green turtles, with clear plasma color (Bolten and Bjorndal, 1992) has been used to identify recruits. Stable iso­ 1 Models that integrate effects of biological and physical parameters over several trophic levels.

OCR for page 73
 DEMOGRAPHIC RATES tope signatures of carbon and nitrogen in scute tissue (the keratin cover­ ing of the upper shell that is inert after deposition) provide a history of diet and habitat that can be used to identify recent recruits (Reich et al., 2007). Reliable, rapid, and non­invasive methods of identifying recruits are needed. SOMATIC GROWTH AND AGE AT SEXuAL MATuRITY Somatic growth has been measured in a number of sea­turtle popula­ tions. Adult females essentially stop growing after attaining sexual matu ­ rity, at which point resources are allocated away from somatic growth to reproduction. In immature turtles of a given species, growth varies spa­ tially and temporally (Diez and van Dam, 2002; Balazs and Chaloupka, 2004b; Chaloupka et al., 2004b; Kubis et al., 2009). Known sources of variation are body size (Chaloupka and Musick, 1997), population density (Bjorndal et al., 2000a), habitat quality (Diez and Van Dam, 2002), nutrient quality of diet (Wood and Wood, 1981), disease status (Chaloupka and Balazs, 2005), and compensatory growth (Bjorndal et al., 2003a; Roark et al., 2009a). A combination of somatic growth rates with indexes of body condition is the best current measure of habitat quality and population status on foraging grounds (Bjorndal et al., 2000a; Diez and van Dam, 2002; Kubis et al., 2009). The most common method of measuring growth rates in turtles has been mark–recapture study. Because population and environmental conditions can be monitored throughout a mark–recapture study, this technique offers the best approach for evaluating the mechanisms that regulate growth. Mark–recapture studies are of necessity long term and labor intensive and are successful only when recapture probabilities are relatively high. Because that condition is not always met, other techniques have been used. Skeletochronology, the use of markers in skeletal material (primarily humeri and eye ossicles), has been used in many studies to estimate somatic growth rates (Zug et al., 1986; Bjorndal et al., 2003a; Snover and Hohn, 2004; Snover et al., in press). Caution in the interpretation of marks is critical, the technique is not practical for live animals, and remodeling of internal bone layers can be problematic. Those and other challenges in the application of skeletochronology have been well reviewed (Snover et al., 2007; Avens et al., 2009). Advantages of the technique are that turtles do not have to be captured, skeletal elements can be gathered from the large number of carcasses that strand on the U.S. coast each year, and longitudinal sampling of individuals can be exploited. Longitudinal sam ­ pling is possible only with multiple recaptures in mark–recapture studies. With skeletochronology, growth­increment analysis of the humeri can be

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS used to detect individual variance in growth rates (Vaughan, 2009). A greater effort needs to be made to archive humeri from sea­turtle carcasses of known size, sex, location, and date for age and growth studies. Length­frequency analyses, which rely on maximum­likelihood algo­ rithms to detect ageclass modes in size distributions, have been used widely in fisheries and successfully in sea turtles (Bjorndal et al., 1995, 2000b, 2001). A disadvantage of the technique is that, with currently available software, only von Bertalanffy growth models can be used. Greater overlap of body lengths in older ageclasses may limit the use of the technique. Its main advantage is that it requires only data on size distributions. Two other techniques for measuring growth of sea turtles have been investigated. Hays and Marsh (1997) estimated growth rates of the very early stages by analyzing drift times to remote locations and the size of small turtles at those locations. And the use of RNA and DNA ratios, which have been used extensively in studies of fish growth, has been tested in sea turtles with some success (Roark et al., 2009b). Both tech ­ niques deserve further evaluation. Age at sexual maturity is a critical demographic parameter. Estimat ­ ing age at maturity on the basis of somatic growth rates is problematic because, for all Atlantic populations, few data are available on growth rates of large subadult turtles (i.e., above 70 cm in carapace [upper shell] length in green turtles). A high priority might be to determine growth rates of large subadults so that estimates of age at sexual maturity can be based on a stronger foundation. SEX RATIOS Because sea turtles exhibit environmental sex determination, primary sex ratios are determined by environmental factors, as described above (see the section “Fecundity”). Variation in secondary sex ratios (i.e., the odds of a hatchling will be male) on foraging grounds may result from variation in primary sex ratios, sex­specific mortality, or sex­specific dis­ persal. Data on secondary sex ratios of immature and adult sea turtles are needed to develop sex­specific population models and to evaluate “opti­ mal” sex ratios (i.e., ratios at which reproductive output is maximized in a population). If it becomes necessary because of global warming, the latter will be critical for programs to manipulate primary sex ratios at nesting beaches (Mrosovsky and Godfrey, 1995). DENSITY DEPENDENCE Rates in a population are said to be density dependent if they vary with the abundance or density of the population. For example, in the

OCR for page 73
 DEMOGRAPHIC RATES classic logistic model of population growth, the per capita population growth rate increases linearly as the population declines. That kind of density dependence is termed compensatory because it tends to stabilize population size. When the population is small, the per capita growth rate is high, and the population increases toward its carrying capacity. As the population nears its carrying capacity, the growth rate declines as births and deaths become equal, and the population reaches a stable abundance. If other characteristics are equal, a depleted population will begin to recover relatively rapidly when compensation is present if the limiting factor (e.g., harvesting, bycatch, disease) is reduced or eliminated. The most common cause of compensation is competition for food, space, or other resources. In contrast with compensation, some populations exhibit a form of density dependence termed depensation, in which over some low range of abundance the per capita growth rate decreases as abundance declines. Depensation is said to be critical when the per capita growth rate becomes negative at low abundance. When depensation is operating, a depleted population will tend to recover slowly and therefore be vulnerable to extinction shocks. If depensation is critical, the population may become extinct despite the elimination of the limiting factor. Although the classic explanation of depensation (also referred to as the Allee effect) is the rarity of high­quality mating opportunities, other factors may be involved (Liermann and Hilborn, 2001). Turtle population growth has been evaluated with stage­based matrix models that typically assume that vital rates are independent of density. That reflects primarily a lack of information about density effects on these rates. The effect on model predictions of ignoring density dependence remains an open question. Clearly, if vital rates are strongly density depen­ dent, a model with fixed rates will be, at best, applicable over a narrow range of population size. Despite that limitation, model predictions may be correct qualitatively (Heppell et al., 2000). Chaloupka and Balazs (2007) developed a statistical state­space model for Hawaiian green turtles that allows for density dependence (either compensatory or depensatory). Of necessity, most work on identifying density dependence in sea­ turtle populations has focused on processes that occur on nesting beaches. For example, Girondot et al. (2002) identified density­dependent nest destruction in the leatherback turtle on a beach in French Guiana; the rate of destruction increased with nesting numbers (presumably as a result of nesting­habitat limitation). That was accompanied by a density­dependent feminization of the hatchling sex ratio (Caut et al., 2006). Tiwari et al. (2006) also reported density­dependent nest destruction and predation on hatchlings in a Caribbean green turtle population. In a study not involv ­ ing nesting processes, Bjorndal et al. (2000a) found evidence of density

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS dependence in the somatic growth rate of immature green turtles in the Caribbean region. They found a negative correlation between population density and both the mean annual growth rate (as measured by carapace length) and an index of body condition. That suggests that Caribbean green turtles are food limited when abundance is high. Bell et al. (2010) evaluated evidence of depensation in green turtles and loggerhead turtles. They focused on the relationship between rookery size (as measured by total clutches per season) and fertilization success, hatch success, and hatchling emergence success; they used data on the Cayman Islands and a meta­analysis of global data. The study found no evidence of depensation in either species in either the Cayman Islands data or the global data. However, because the analysis was based on a mixture of cross­sectional and time­series data, the result needs to be treated with caution. A more complete analysis would treat the data as multiple time series with depensation operating within, but not between, the component series. STRANDINGS DATA A substantial proportion of the effort expended to collect sea­turtle data in the United States is invested in the Sea Turtle Stranding and Salvage Network (STSSN). Because the usefulness of the data generated in that program has been debated (Epperly et al., 1996; Turtle Expert Work ­ ing Group, 2000), the committee addresses STSSN here in some detail. Sea­turtle strandings occur when animals have washed up on a beach or into shallow water. Stranded animals may be dead or dying because of anthropogenic causes, such as interactions with fisheries, or natural mor­ bidity, such as disease or “cold stunning” when they have been exposed to lethal cold­water temperatures. Strandings include all life stages that are present in neritic habitats, including juveniles and adult males. Carcasses provide opportunities for data collection that are difficult or impossible with live animals, such as collection of data for evaluating maturational status and removal of the humeri for age and growth studies. Carcasses are checked for tags; this is an important source of tag recoveries that are used to evaluate growth and dispersal of individual turtles. Strandings can also provide some information on mortality and have been correlated with levels of fishing effort and enforcement (Lewison et al., 2003). With careful consideration of the many sources of variability that affect the probability of stranding and detection of carcasses, strandings may also provide distribution and trend information that is relevant to population assessment (Chaloupka et al., 2008b). The density of strandings has been used as a trigger for manage ­ ment action in some areas and has resulted in spatial closures in fisheries

OCR for page 73
 DEMOGRAPHIC RATES (Santora, 2003). Strandings have been clearly linked to fishing activity (Caillouet et al., 1996; Chaloupka et al., 2008b), and changes in the relative abundance of strandings have been used as an indicator of management effectiveness (Crowder et al., 1995; Lewison et al., 2003). However, many factors are related to the frequency of strandings, including cause of death and condition of the carcass, location of death and water currents, water temperature (which affects the decay rate), and salvage effort. Physical processes that affect stranding rates may also change, and this poten­ tially necessitates regular evaluation of the correlation between ocean conditions and the probability of stranding as climate­driven forcers vary in time and space. Concerns about these caveats have led to disagree ­ ment about the value of strandings for population assessment (Epperly et al., 1996; Turtle Expert Working Group, 2000). Expert working groups and recovery­planning teams have agreed that strandings are highly stochastic events that provide information about local mortality events and a minimum estimate of regional mortality, but it may be difficult to extrapolate trends in strandings to changes in population abundance. Strandings account for an unknown proportion of total mortality and probably varies among regions. Nevertheless, patterns of strandings in time and space can provide information about seasonal distribution and interactions with fisheries when carcass­recovery efforts are standardized and data are pooled over broad spatiotemporal scales (Chaloupka et al., 2008b; Tomás et al., 2008). STSSN in the United States operates in each coastal state and is coor­ dinated through the National Marine Fisheries Service (NMFS) and state agencies. The network is run by a state coordinator and depends heavily on local volunteers. Coordinators are responsible for training programs for the volunteers and for the weekly or bi­weekly data reports that are sent to NMFS. Although the specific goals of each salvage program vary, most are designed to evaluate carcass abundance and trends that are assumed to be indicative of the living population of turtles in the monitored area. The programs provide data that can be used to quantify seasonality, species composition, population structure, life­history stage, sex ratio, and spatial distribution of turtles that wash ashore. All STSSN­recovered animals are identified as to species, checked for external tags, and recorded by date and location. Carapace length and general condition of the carcass are also recorded for most animals. Many recovered dead turtles are necropsied by the state coordinator and staff to identify sex and state of maturation, to record ingestion of plastic (Bjorndal et al., 1994), and to conduct a general evaluation of the potential cause of death (although this can only rarely be determined). Carcasses may also be checked for PIT tags or magnetic wire tags. Some samples are collected from necropsied animals for specific projects, including tissue

OCR for page 73
 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS samples for contaminants evaluation and bones (humerus and eye ossicle) for aging (Snover and Hohn, 2004), and body­size data have been used to generate curves of somatic growth (Bjorndal et al., 2001). The proportion of recovered turtles that are evaluated thoroughly varies by state and frequency of strandings. Although sampling of sea­turtle carcasses for specific research projects does occur, the extent to which samples are col­ lected and archived is variable in the southeastern United States, where most sea­turtle strandings occur (Laris Avens, personal communication). In Hawaii, the Pacific Islands Fisheries Science Center regularly archives tissue and humeri samples from recovered turtles. Strandings data are compiled and reviewed by the Southeast Fisheries Science Center, but researchers must request access to the data from each state individually, and state coordinators vary in their criteria for sharing data. Changes in size distributions of strandings may be a valuable indi­ cator of shifts in age structure or distribution of juveniles (Shoop et al., 1999). Turtle Expert Working Group reports (2000, 2009) have included examinations of trends in total strandings by region and the size dis ­ tributions of strandings. Kemp’s ridley and loggerhead strandings size distributions were converted to age distributions with an age­length key and used to estimate total instantaneous mortality with a simple catch­ curve method (Turtle Expert Working Group, 2000). This approach raises a number of issues that were discussed by the report authors, includ ­ ing the unknown relationship between sizes of strandings and those of the population­at­large, the need to pool strandings across several years because of small samples, and variable growth rates that confound the age­length relationship. Recently, a data review by a loggerhead turtle working group included plots of turtle sizes observed through time that showed a good correlation between the size distributions of strandings on the east coast of the United States and of turtles observed at a power­plant intake in Florida and juvenile mark–recapture surveys (Turtle Expert Working Group, 2009; Vaughan, 2009). That suggests that strandings may be a reasonable indicator of what turtles are in the nearshore population, at least on broad spatial and temporal scales. Confirmation of the congru ­ ence is needed, particularly if researchers want to continue to use strand ­ ings to estimate mortality. If the size composition of the nesting population is known and a strandings event can be linked to a particular fishery or environmental event, the observed size distribution can help to determine selectivity of the mortality source (what size classes are susceptible to the fishing gear used or an environmental event, such as red tide). Further research on the “selectivity curve” for strandings would be helpful to determine how the probability of carcass recovery is affected by the size and condition of the animal and its environment. A study using drifter bottles deployed in

OCR for page 73
 DEMOGRAPHIC RATES the South Atlantic Bight provided a rough estimate of a 20% probability of reaching shore for a wind­ and current­driven carcass, with strong seasonal and spatial variability (Hart et al., 2006); similar studies need to be conducted in the Gulf of Mexico and the northeast United States, with an emphasis on establishing the likelihood of detection and statistical discrimination among spatial scales (Wiens, 1989). The level of environmental monitoring needed to identify relation ­ ships between oceanography and strandings may be substantial, given the complexity and variability of nearshore ocean processes. It is possible that environmental variance will nullify strandings as a source of trend and distribution information for nesting populations, but large strandings events are still valuable indicators of local conditions (e.g., harmful algal blooms, intense fishing mortality), and samples from dead animals can provide important information through diet evaluation about local popu­ lation structure, growth rates, maturation rates, and habitat use. Every recovered carcass can be a valuable source of information for assessment if recovery efforts are standardized; proper measurements are taken; and samples are collected, processed, and archived according to established protocols. To improve the value of strandings data for assess­ ment, each state program needs to be reviewed and evaluated for con­ sistency in recovery effort, volunteer training, and protocols. Areas that have low or inconsistent sampling effort could be identified to improve extrapolation methods. Programs for evaluating size distributions and growth rates from turtle hard parts need to be supported and enhanced to maximize the amount of information obtained from each stranded animal. Flipper collection could become standard protocol for STSSN volunteers in the southeastern and Gulf states, but a considerable investment in time and resources will be needed to process and evaluate those samples. RECOMMENDATIONS • Researchers should give high priority to generating estimates for the following parameters: survival of immature turtles and nesting females, age at sexual maturity, breeding rates, and clutch frequency. • Because demographic rates can vary over time and space, researchers should collect data over both dimensions so that population trends can be detected and evaluated adequately. • Researchers should be aware that evaluation of point estimates of demographic parameters is not sufficient for population assessment; characterizing uncertainty and variance is also necessary. • Researchers should strive to understand the mechanisms regu­ lating variation in demographic rates; this is essential for diagnosing changes in population abundance and mitigating population declines.

OCR for page 73
0 ASSESSMENT OF SEA-TURTLE STATUS AND TRENDS • NMFS and the U.S. Fish and Wildlife Service should arrange for a review of data now being collected under the auspices of, or with the sup­ port of, their agencies and evaluate the costs and benefits. For example, the sea­turtle stranding and salvage networks should be evaluated, per­ haps with the assistance of the U.S. Geological Survey’s National Wildlife Health Center. • STSSN should collect—in addition to data on abundance, size, condition, and sex—samples of tissues and hard parts that can be used to identify stock of origin, to assess diet through isotope analysis, and to evaluate age and growth.