Effects of a sudden increase in natural mortality of adults on a population of the common snapping turtle (Chelydra serpentina) Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. Department of Zoology, University of Guelph, Guelph, Ontario, Canada N l G 2Wl Received May 17, 1990 D. A. 1991. Effects of a sudden increase in natural mortality of adults on a BROOKS, R. J . , BROWN,G. P., and GALBRAITH, population of the common snapping turtle (Chelydra serpentina) . Can. J. Zool. 69: 1314- 1320. A northern population of snapping turtles (Chelydra serpentina) centred around Lake Sasajewun in the Wildlife Research Area in Algonquin Park, Ontario, has been studied and individually marked since 1972. From 1972 to 1985, annual mortality and survivorship of adult females had been estimated at 1 and 96.6%, respectively, and only six dead turtles were found. Lake Sasajewun's population of C. serpentina was estimated in 1978- 1979 and 1984- 1985 at 38 and 47 adults, respectively. From 1976to 1987, total number of nests found in the study area remained fairly constant and there were no significant changes in mean clutch size, mean clutch mass, or mean egg mass. On the main nest site, recruitment from 1976 to 1987 was 1.15 (1.8%) new females per year. From 1987 to 1989, we found 34 dead adult snapping turtles in the Wildlife Research Area. Observations of freshly dead animals indicated that most were killed by otters (Lutra canadensis) during the turtles' winter hibernation. A few uninjured turtles also died of septicemia in early spring shortly after emerging from hibernation. The estimated number of adults in Lake Sasajewun was 3 1 in 1988-1989, and the minimum number of adult residents known to be alive in the lake dropped from 47 in 1986 to 16 in 1989. In 1986 and 1987, annual adult female survivorship was estimated at 80 and 55%, respectively, and estimated numbers of nesting females declined from 82 in 1986 to 7 1 and 55 in 1987 and 1988, respectively. The actual number of nests found declined by 38 and 20% over the same periods. Although no significant differences occurred in mean egg mass or mean clutch size between 1987 and 1989 and earlier years, the mean clutch mass in 1988 was larger than in 1977 or 1978. This difference appeared to be due to a gradual increase in the mean age and body size of breeding females rather than to density-dependent changes. Recruitment into the adult breeding female population in 1987- 1989 remained less than two individuals per year. Hatchling survival and number of juveniles were low throughout the study. Our observations support the view that populations of species with high, stochastic juvenile mortality and long adult life spans may be decimated quickly by increased mortality'of adult animals, particularly if numbers of juveniles and immigrants are low. Recovery of such populations should be very slow because of a lack of effective density-dependent response in reproduction and recruitment. BROOKS, R. J., BROWN,G. P., et GALBRAITH, D. A. 1991. Effects of a sudden increase in natural mortality of adults on a population of the common snapping turtle (Chelydra serpentina) . Can. J . Zool. 69 : 1314- 1320. Une population nordique de chClydres serpentines (Chelydra serpentina) rksidant autour du lac Sasajewun, dans la zone de recherche Wildlife Research Area du parc Algonquin, Ontario, fait l'objet d'une Ctude depuis 1972 et des individus ont CtC marquCs. Entre 1972 et 1985, la mortalit6 annuelle des femelles adultes a CtC CvaluCe B 1% et leur survie B 96,6% et seulement six tortues mortes ont CtC trouvkes. La population adulte du lac Sasajewun a CtC estimCe B 38 en 1978-1979 et B 47 en 1984-1985. De 1976 B 1987, le nombre total de nids trouvCs dans la zone Ctudite est demeurt B peu pr2s constant et il ne s'est pas produit de changements significatifs du nombre d'oeufs par portte, de la masse moyenne de la portCe ou de la masse moyenne des oeufs. Au site principal de nidation, le recrutement de 1976 B 1987 a CtC CvaluC B 1,15 (1,8%) nouvelle femelle par annte. De 1987 B 1989, nous avons trouvC 34 tortues adultes mortes dans la zone Wildlife Research Area. L'examen d'animaux morts depuis peu a rCvClC que la plupart des tortues avaient 6tC tuCes par des loutres (Lutra canadensis) au cours de la pCriode d'hibernation des tortues. Quelques tortues intactes sont aussi mortes de septicdmie au dCbut du printemps, peu apr5s leur sortie d'hibernation. Le nombre d'adultes au lac Sasajewun a CtC CvaluC B 31 en 1988-1989, et le nombre minimum d'adultes rCsidants connus dans le lac est pass6 de 47 en 1986 B 16 en 1989. La survie annuelle des femelles adultes a CtC estimCe B 80%' en 1986, et B 55% en 1987, et le nombre estimC de femelles qui ont nichC est pass6 de 82 en 1986 B 71 en 1987, puis B 55 en 1988. Le nombre de nids a dirninut de 38%' puis de 20% au cours de ces pCriodes. La masse moyenne des oeufs et le nombre moyen d'oeufs par portCe n'ont pas changC significativement entre la pCriode 1987-1989 et les annCes prkcedentes, mais la masse moyenne des portCes en 1988 s'est avCrCe supCrieure B ce qu'elle Ctait en 1977 ou 1978. Cette diffkrence semble reliCe B une augmentation graduelle de l'ige moyen et de la taille des femelles reproductrices plut6t qu'8 des changements reliCs B la densite. Le recrutement de la population de femelles reproductrices en 1987- 1989 est rest6 infCrieur ii deux individus par annCe. La survie B 1'Cclosion et le nombre de juvtniles est rest6 faible durant toute la pkriode de 1'Ctude. Nos observations supportent la thCorie selon laquelle les populations d'esp&ces B forte mortalit6 stochastique de juvCniles et B 1ongCvitCClevCe des adultes seraient exposCes B un dCclin rapide rdsultant d'une mortalit6 accrue des adultes, particulibement si le nombre de juvtniles et d'immigrants est faible. De telles populations mettent sans doute beaucoup de temps ii rCcup6rer puisqu'elles ne poss&dentpas de syst5me de reproduction ou de recrutement liC ii la densitC. [Traduit par la redaction] Introduction ~~~t population and management models are based on species that have relatively short life-spans and high annual recmitment (Miller and Botkin 1974). For example, most birds, mammals, and fish that are managed for killing fall into this category. Long-lived species present different problems for ' ~ u t h o rto whom all correspondence should be addressed. 2Present address: Biology Department, Queen's University, Kingston, Ontario, Canada K7L 3N6. wildlife managers and population modellers, particularly if such species fail to show density-dependent reproductive responses 1988). (Obbard 1983, 1985; Brooks et generally are long-lived (Gibbons 1987), and although evidence is accumulating that they are susceptible to extinction through exploitation studies that (fitchard 1980, 1989)9 there are few would allow a direct test of this hypothesis. Recently, we reported on the high survivorship of female adults in a northem population of snapping turtles (Chel~dra serpentiruz) (Galbraith and Brooks 1987). However, from 1987 BROOKS ET AL. TABLE1. Number of adult snapping turtles in Lake Sasajewun based on Petersen estimates and on number known to be alive in the lake (based on trapping and visual observations) Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. Sample period Total trap nights Number of captured adults Number of recaptures from previous period Known number of resident males <10 kg Known number of resident males > 10 kg Total known resident males Petersen estimates MNKA (male and female) NOTE:The data within parentheses represent the interval within 95% confidence limits. MNKA, minimum number known to be alive. to 1989, the number of adults of both sexes in this population declined dramatically accompanied by a marked increase in the number of documented mortalities. We present evidence of this decline in numbers of adults and nests, and we test the hypothesis that the decline is not compensated by density-dependent adjustments such as increased survival of young, increased recruitment of young or adults, earlier age of maturity, or by greater individual reproductive output. As well, we present evidence concerning the causes of the observed mortality, and speculate on the long-term effect on the population. ' Methods The study was conducted in Algonquin Provincial Park, Ontario, at the Wildlife Research Area (WRA) (45"35'N, 78"301W)on the North Madawaska River watershed. Methods and descriptions of the study area are in earlier papers (Obbard and Brooks 1979, 1981; Galbraith and Brooks 1987; Galbraith et al. 1988) and are summarized here. From 1972 to 1975,72 adult turtles were captured and tagged: 53 were females captured at the main nest site, a gravel dam on Lake Sasajewun (Loncke and Obbard 1977), 7 were males trapped in the lake, and 12 were females that resided outside the study area. Since 1975, we have tried to capture, mark, and measure all turtles nesting on the dam. Females also use about 20 other nesting sites in the WRA surrounding Lake Sasajewun. These sites have been monitored since 1975, but only since 1979 have we attempted to capture all females at these sites. The identity of the nesting female is known for more than 90% of clutches on the WRA after 1979. In addition to captures by hand, extensive live trapping was conducted in Lake Sasajewun from 1977 to 1980 and from 1983 to 1989. All captured turtles were weighed, measured, tagged, and notched. Between 1983 and 1989, we released 1997 marked hatchlings either in the fall of the year they hatched or in the following spring. These hatchlings were from eggs collected late in summer from natural nests or from eggs collected in spring immediately after oviposition, and reared in constant-temperature incubators in the laboratory. All hatchlings were released on the WRA site close to their original nest sites. Methods to estimate changes in population numbers and survivorship We used six methods to estimate changes in numbers and survival of adult snapping turtles on the study area from 1976 to 1989. First, we tallied the number of known mortalities of turtles. Second, the adult turtle population of Lake Sasajewun was estimated from trapping data for 1978- 1979 (Obbard 1983), .1984- 1985 (Galbraith et al. 1988), 1987-1988, and 1988-1989, using the Petersen estimate (Krebs 1989). We used the same method for all sample periods to standardize the results. Third, using trapping records and positive field identification, we tallied the number of male turtles known to be alive in Lake Sasajewun for 1986-1989. Fourth, we determined the number of nesting females on the dam and alternate sites to estimate annual rates of recruitment of new females to the nest sites. Fifth, from recaptures of nesting females from 1976 to 1989, we calculated Jolly-Seber estimates (Krebs 1989)of the number of females nesting on WRA sites each year, and their probability of survival to the following year. We also recorded total number of nests located on the WRA sites. Sixth, we monitored the number of nests on two sites near the main study area (Whitefish Lake, 1983- 1989; Cache Lake, 1987-1989). Both are part of the Madawaska River system, However, no turtle caught or observed at either site has been observed at the other or at the WRA site, so it is safe to assume that the three populations are relatively isolated from each other. These sites were monitored originally to estimate rates of predation on undisturbed nests, but here the data are used to estimate annual changes in the number of nests. Methods to estimate reproductive output From 1977 to 1989, clutch size, clutch mass, and mean egg mass were recorded for most clutches located on the study site. In 1976, only clutch size was recorded. These measures were compared among years using ANOVA and if significant variation was detected, Scheffk's test was used to determine which years differed significantly (procedure of the generalized linear model, SAS Institute Inc. 1985). Specifically, we tested the hypothesis that these measures would increase significantly ( p < 0.05) as numbers of turtles declined on the study site (one-tailed test). Results Known mortalities From 1972 to 1986, we found six dead adults (two males and four females) and one dead juvenile. Four of these animals died, probably during winter, between 1978 and 1983. Two adult females died in spring soon after emergence from hibernation. One of these females and an untagged male with a deformed carapace appeared to have drowned in the water chute at the main nesting area. The cause of death of the other five animals was unknown. Six other adult females were killed by researchers and are not included in any calculations of survivorship or of the numbers of nesting females. From 1987 to 1989, 34 dead adults were found in the study 1316 CAN. J. ZOOL. VOL. 69, 1991 TABLE2. Number of snapping turtle nests found at various locations in Algonquin Park from 1976 to 1989 LakeSasajewun(dam) Alternate sites Total 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 54 10 50 19 43 17 41 23 30 30 36 32 35 36 39 28 34 30 32 34 32 36 25 44 20 29 16 23 64 69 60 64 60 68 71 67 64 66 68 69 49 39 Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. Whitefish Lake Cache Lake area around the WRA. All but three of these died during winter. Of the three exceptions, one was killed by a car and two died shortly after emergence from hibernation. From the location of the bodies and from data obtained by radiotelemetry, it appears that in winter the turtles move to a small number of restricted areas to hibernate (G. Brown, unpublished data). In the springs of 1987 and 1988 particularly, almost all of the dead animals were located in these sites. Those individuals fitted with radio transmitters were found in the same spot in which they had spent most of the winter. Most individuals that died overwinter and were not decomposed, were mutilated in a very consistent manner. In mutilated animals, with one exception, the viscera of both the thoracic and abdominal region were entirely gone, and all these turtles had large holes in the inguinal and (or) axillary regions. Most also hac! grooves, apparently made by teeth of a predator, on the lateral edges of the plastron adjacent to the holes and were found lying on their backs in less than 1 m of water or on shore. Usually, there were no injuries to the head, limbs, or tail, and if found before the onset of decay, these turtles, when viewed from the dorsal side, appeared intact. Five turtles were found on shore. These animals were more decomposed so it was difficult to determine whether the limbs, tail, and head had been eaten or simply decayed or both. One radio-tagged individual still had thoracic organs and could move when first observed in spring 1989, but his intestines had been eaten and he had extensive damage to one hind limb. During the winter of 1988-1989, an otter (Lutra canadensis) hole was found in the ice within 15 m of the hibernating site of this turtle. In March 1989, the turtle moved (or was moved) about 30 m to another otter ice hole where he was found in May. Otter holes were observed also in the ice near hibernation sites in Lake Sasajewun in the winters of 19871988 and 1988-1989. One other large male was found alive in spring 1989, but with the characteristic grooves in the right rear plastron and a deep, 2.0-cm gash in the inguinal pocket. This injury was treated with antiseptic and cornstarch and has since healed. The sex ratio of marked adult turtles that died over winter at these hibernacula from 1987 to 1989 was 13 males : 9 females. The overall sex.ratio of dead adults (that could be sexed) was 15 males : 13 females. There appeared to be no relationship between size and mortality; the dead turtles included both our oldest and youngest marked adults. Population estimates for Lake Sasajewun Petersen estimates of the Sasajewun population for four sample periods are given in Table 1. These indicate that the numbers in the Lake did not decline until 1988. In the Petersen estimates of 1987-1988 and 1988-1989 the low rate of recapture gives wider confidence limits and the high value of 1987-1988 is due to the low rate of recapture in 1988. The FIG.1 . Jolly-Seber estimates of numbers of female Chelydra nesting on the study site each year. Vertical lines indicate 2 SE on either side of the estimate. , minimum number of resident adults known to be alive in the Lake declined from 47 in 1986 to 16 in 1989 (Table 1). Of the 16 found in 1989, only 8 were considered full-time residents of the Lake; the other 8 were captured only once or were known to have moved to other lakes later in the summer. Overall, the minimum number known to be alive appears to have declined by about 65% from 1986 to 1989. The males, particularly those larger than 10 kg, were more sedentary than the females, and therefore easier to define as residents. Only one male heavier than 10 kg died (or was unaccounted for) from 1976 to 1987, but five died in 19881989. Smaller males tended to be less sedentary than large males, but from 1976 to 1986 the number of these males living in Lake Sasajewun was fairly constant (Table I). By 1989, only seven resident males remained and of these only four were present throughout the summer; the other three moved to other lakes or into creeks leading into Sasajewun (Table I). Number of nesting females Since 1976, we have located every nest laid on the Sasajewun dam. The number of nests declined from a peak of 54 in 1976 to a low of 16 in 1989 (Table 2). This decline (1976- 1978) can be explained partially by a shift of nesting from the dam to other (alternate) sites in the WRA. The increased number of nests found on the alternate sites is not only due to this shift, but also to more intensive and extensive surveillance of the alternate sites. From 1972 to 1976, 63 females were captured on the Sasajewun dam. In 1986, 27 of these nested on the dam and 14 nested on the alternate sites (65% of the original 63). In 1989, only 11 of these 41 females nested on the dam and three on the alternate sites (22% of the original 63). Of the 14 turtles marked on the dam in 1972, 11 were observed nesting in 1986; 8 in 1987, 4 in 1988, and only 1 in 1317 BROOKS ET AL. TABLE3. Mean clutch size, mean clutch mass, and mean egg mass of female Chelydra serpentina on the main WRA study site from 1976 to 1989 Clutch size Clutch mass (g) Egg mass (g) 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 34 (43) nd 33 (53) 375b (52) 11.4 (52) 33 (39) 370c (38) 11.2 (38) 34 (37) 387 (37) 11.4 (37) 35 (46) 393 (46) 11.6 (46) 35 (62) 380 (62) 10.9~ (62) 36 (69) 401 (69) 11.1 (69) 36 (62) 403 (62) 11.2 (62) 36 (56) 426 (56) 11.8 (56) 34 (65) 397 (65) 11.7 (65) 35 (67) 407 (67) 11.6 (67) 37 (67) 431 (66) 11.6 (66) nd Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. NOTE:Values in the row followed by the same letter are significantly different (Scheffe's test, p FIG. 2 . Jolly-Seber estimates of annual survivorship of nesting Chelydra. Vertical lines indicate 2 SE on either side of the estimate. 1989. For the other turtles captured on the dam from 1973 to 1976, a similar pattern of decline occurred in 1986- 1989. From 1977 to-1986, only 12 new females nested on the dam. In 1986, nine of these (75%) were observed nesting (seven on the dam and two on the alternate sites). In 1989, only 4 of the 12 new females nested, (3 on the dam and 1 on the alternate sites). Since 1987, when large numbers of dead turtles began to be found, only three females nested on the dam for the first time. Two of these had nested or resided in the study area prior to 1987. Therefore only one new female (P14 in 1987) nested on the dam after 1986. This female was much older than the average age of maturity of 17- 19 years (Galbraith et al. 1989). Similarly on the alternate sites, although seven females nested for the first time in 1987-1989, only five of these were new to the area: two nested in 1987, two in 1988, and one in 1989. Only one of these females has been observed nesting in more than 1 year. All five were much larger and had more annuli than first-time breeders, and were likely new immigrants or transients. Average annual recruitment to the main nest site (the dam) from 1977 to 1989 was 1.15 females (a recruitment of 1.8% per year if we assume 63 as the population size). After 1986, only 0.33 females have been recruited per year. Recruitment on the alternate sites is more difficult to estimate because many of the females were marked when we discovered the location of these sites rather than as they first appeared at the sites. From 1982 to 1988, 18 "new" females nested on these sites, a rate of 2.3 per year. However several of these females were tagged in the study area in earlier years. Number of females found nesting on the alternate sites peaked in 1987, and only three new females have nested since then. The actual rate of "recruitment" is probably about one female per year, but most of these seem to be transient = 1988 38 (46) 459bc (46) 12.1~ (46) 1989 38 (39) 440 (39) 11.6 (39) 0.05). Data in parentheses represent the number of clutches. as they are observed to nest only once, and then are not seen again. Jolly-Seber estimates Jolly-Seber estimates of the total population of nesting females indicate a constant number from 1977 to 1980, and then an increase corresponding to more thorough searching of the alternate sites and addition of new sites beginning in 1981. The estimated breeding population peaked in 1986 at 83, and by 1988 it had declined to 55 (Fig. I). Jolly-Seber estimates of survivorship of nesting females indicated that annual survivorship remained above 95 % until 1985, and then declined to 80% in 1986 and 7 1% in 1987 (Fig. 2). Other populations The number of nests at the Whitefish Lake site declined sharply from 21 in 1983 to as low as 4 in 1988 (Table 2). The decline seemed to occur from 1983 to 1985. At Cache Lake, the number of nests seems to be declining but estimates have been made for only 3 years (Table 2). Changes in reproductive measures From 1976 to 1989, there was significant annual variation in both mean clutch size and mean clutch mass ( F = 2.43, p < 0.01, and F = 3.99, p < 0.001, respectively) (Table 3). However, annual differences accounted for only about 5% of the variation in both cases. The average clutch size and mass increased over this period by about six eggs (19%) and 100 g (27%), respectively. However, Scheffk's test indicated no significant differences in mean clutch size among years (Table 3). Clutch mass in 1988 was significantly heavier than in either 1977 or 1978. Mean egg mass showed no obvious trend over the 14-yearperiod, but the ANOVA indicated a significant difference among years ( F = 4.41, p < 0.0001), although again r2 (= 0.07) was very small. Scheffk's test indicated that the only significant difference in mean egg mass among years was between 1981 and 1988 (Table 3). Numbers of juveniles and marked, released hatchlings In all years, the number of captured juveniles was low, never exceeding six in any year. From 1976 to 1980, Obbard (1983) captured a total of only seven different juveniles. From 1987 to 1989, seven different juveniles were captured, although only three of these were caught in either 1988 or 1989. Only one marked juvenile (ElO) has been observed to mature and enter the breeding population. Recapture of previously marked, released hatchlings has been consistently low after the year of release. In 1989, only 6 hatchlings were recovered from 1722 released between 1983 and 1988. The oldest of these was an individual from 1983 (mass = 258 g). Two other recaptures were hatched in 1985 and 13 18 CAN. J. ZOOL. VOL. 69, 1991 Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. another two in 1987. The year of hatching could not be determined for the remaining individual. Discussion From 1976 to 1986, the number of snapping turtles in Lake Sasajewun and the number and survivorship of nesting females remained constant on the study area, despite low recruitment into the breeding population. However, from 1987 to 1989, known mortality of adults increased 20-fold. Concomitantly with this increase in known mortality, the estimated number of turtles in Lake Sasajewun, the number of females nesting on the WRA, and the known number of males in Lake Sasajewun all declined by about 65%. Although the population declined, mean clutch size remained stable and recruitment of new adults, numbers of juveniles and hatchlings, and growth rate of individuals remained low. Therefore, we reject the hypothesis that there was been a compensatory (density-dependent) response to increased adult mortality by increased reproductive effort, recruitment, or juvenile growth rate. The major source of observed mortality was predation by otters. However, most reports of feeding habits of otters revealed no evidence that they had consumed turtles or indeed any reptiles other than occasional snakes (Lagler and Ostenson 1942; Greer 1955; Hamilton 1961; Sheldon and Toll 1964; Melquist and Hornocker l983; Tumlison and Kaines 1987). We found only one reference describing otters eating turtles frequently. Park (197 1) reported a letter from a correspondent in Wisconsin stating that, in winter, otters regularly pulled hibernating snapping turtles out onto the 'ice to consume them. In one winter, this observer counted 27 adult snappers killed by otters on a 3-mile stretch of the river (Park 1971). - Most of our turtles were consumed under water, perhaps because the lake ice was too thick (0.5 m) for the otters to get the turtles through. Most turtles found on shore had overwintered along small streams where ice was thin or where fast water remained ice-free all winter. Because the otters did not consume the skeletal muscle of most carcasses, we speculate that they did not suddenly begin to eat turtles because of a loss of their normal prey. Their consumption of only the more nutritious and more easily consumed viscera appears to be an example of optimal foraging when food is plentiful (Nudds 1987). Snapping turtles often hibernate at high density in small areas (Ultsch 1989; Meeks and Ultsch 1990; G. Brown, unpublished data), and because their bodies are cold (1-2°C in our Algonquin population, G . Brown, unpublished data), the turtles move slowly and are an easily exploited and locally plentiful food resource for the otters. It seems likely that one or more otters have "discovered" this resource recently, because otters have been common throughout the study area since the study began. A small number of turtles died without injury. Necropsies indicated severe systemic bacterial infection in two dead turtles that bore no external wounds. Turtles are long-lived, slowly growing animals with late sexual maturity and a long reproductive life-span; they often nest in large aggregations, which leads to high rates of mortality of eggs and hatchings (Dodd 1985; Gibbons 1987). This suite of life-history traits is poorly adapted to high rates of adult mortality, because such species often cannot replace adult losses quickly and hence are especially susceptible to extinction when such losses occur, as for example from "harvesting" by humans (Miller and Botkin 1974; Lambert 1982; Internation- al Union for Conservation of Nature and Natural Resources Species Survival Commission 1989). An important part of understanding the population ecology of turtles is to relate the effect of adult mortality to conservation requirements. However, as Dodd (1985) has noted, controversy arises when survivorship, sex ratio, actual numbers, and threats to survival are unknown (e.g .,Mrosovsky 1983; Pritchard 1983). This lack of knowledge of long-lived species has been cited as a serious limitation to adequate tests of current life-history theory (Wilbur 1975), because most differences in suites of correlated lifehistory characteristics are predicted to occur between short- and long-lived species (Congdon et al. 1987). Snapping turtles have been given very little attention despite their wide distribution and high biomass in wetland communities, because they require long-term studies and because they are of little interest to wildlife biologists or managers (Congdon et al. 1987; Brooks et al. 1988). They are also predators of more favored species (e.g ., waterfowl, game fish) and so are rarely afforded protection (Brooks et al. 1988). Although there are some reports of harvest levels at the state or provincial level (Helwig and Hora 1983; Brooks et al. 1988), total stocks are unknown and so we are forced to rely on local studies for quantitative data on effects of variation in demographic features (Hogg 1975; Stubbs et al. 1985; Brooks et al. 1988). In snapping turtles, at least in northern parts of their range, there appears to be high, stochastic juvenile mortality (Congdon et al. 1987; Brooks et al. 1988). This and other factors, such as variation in age-specific size and fecundity, limit the value of classic static life tables. Removal of adults should produce a decline in numbers and recruitment, but there will be a time lag in the decline if juveniles are numerous and can replace lost adults quickly (Stubbs et al. 1985; Congdon et al. 1987; Pritchard 1989). This lag is especially likely if a decline in adult density leads to an increase in juvenile growth rates, as demonstrated in the Mediterranean tortoise (Testudo hermanni). In this case, the adult population recovered rapidly from a catastrophe (fire) because a large proportion (49-74%) of the population was immature before the fire (Stubbs et al. 1985). Similarly, Davenport (1988) suggested that excessive exploitation of adult sea turtles may occur for 20-50 years before the effects of reduced recruitment will be observed, because their age of maturity may vary from 15 to 50 years. In a desert tortoise (Gopherus agassizii), high adult mortality over a short term (4 years) apparently was compensated over the next 4-5 years by increased survival of juveniles and by immigration (Germano and Joyner 1988). In the WRA population prior to 1987, the probability of survival from eggs to 19 years (estimated age of maturity) was estimated to be only 0.000692 and from hatching to 19 years, 0.7541 (Brooks et al. 1988). Life tables using these estimates and annual survivorship of 96.6% after 19 years still indicated that the population would decline (i .e. , intrinsic rate of increase r, < 0). When r, is low, then late maturity must be compensated by increased fecundity, longer reproductive lifespan, or high prereproductive survival (Roff 1981). Females in our population appear to have little ability to raise r, especially when adult mortality is high. Our females mature later (17-19 years) and at a larger body size than in most populations of C. serpentina (Galbraith et al. 1989). If they matured earlier, they would be smaller and fecundity would decline because fecundity varies directly with body size (Obbard 1983; Galbraith et al. 1989). If they delayed maturity, fecundity would increase, but fewer females would be recruited, BROOKS ET AL. Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. especially with high adult mortality. If females increased fecundity by reducing egg size, it is likely that hatchling survival would decline because present egg size is probably optimal (Brockelman 1975). Egg size is also usually considered to be less variable in response to changes in resource availability than is clutch size, although this is not always true (Schwarzkopf and Brooks 1986; Congdon and Gibbons 1987; Congdon et al. 1987). A final option to increase r, would be to increase growth rate and (or) survival of hatchlings and juveniles. We have no evidence that either of these options is occurring in our population. Marked and released hatchlings appear to have very low survival (<1% after 2 years) and only three unmarked hatchlings have been encountered on the study area since 1981, despite intensive annual searches. Our sample of juveniles is too small and transient to allow us to assess whether their growth rates have increased, but their survival certainly has not, and adult recruitment has remained less than 1% annually. Similarly, adult females do not appear to have increased growth rates (Galbraith et al. 1989; R. J. Brooks, unpublished data). Therefore, most of the "options" for our population to raise r, are unlikely to be successful in theory, and, empirically, they do not appear to be occurring anyway. Mean age of nesting females was estimated at 33-40 years (Brooks et al. 1988; Galbraith and Brooks 1989). Gibbons (1987) suggested that adults of large species of turtles such as C. serpentina and Macroclernys t~rnrninckiiare likely immune to predation. Again an extended life-span and low predation rates on adults suggests a life history dependent on extended iteroparity . Failure to increase fecundity, growth rate, or juvenile survivorship after increased adult mortality also suggest that recovery of our population will be very slow unless immigration increases. In the Murray River, introduced foxes (Vulpes vulpes) have reduced recruitment of the turtle Ernydura rnacquarii by destroying 96% of eggs. This added mortality led to an increase in the mean age of the population (Thompson 1983). Similar changes may have occurred in our population prior to 1987 (Brooks et al. 1988), and the apparent very gradual increase in clutch size and clutch mass from 1976 to 1986 could reflect an aging, slowly growing cohort of females. Possibly, as the number of nests declines, the rate of predation of nests will also decline, but our observations in Algonquin suggest that individual nests are as likely to be depredated as are those in groups, and predators find nests throughout the incubation period (e.g., Snow 1982). Sandhill cranes (Grus canadensis) have late maturity (5 years), a relatively long reproductive life-span (15-20 years), and low estimated annual recruitment (4-8%) (Miller and Botkin 1974); life-history traits that resemble those of turtles. Models show that from an initial population of 194 607 cranes, an annual hunting kill of only 12 770 (6.5%) would lead to extinction in 19 years, and even with an annual loss of only 3% for 10 years, it would take 50 years for the population to recover (Miller and Botkin 1974). Such models indicate that long-lived species are extremely susceptible to unusual adult mortality. Our population of Chelydra, with lower recruitment and later age of maturity, combined with apparent lack of density-dependent response in reproductive output, should be even more susceptible to unusual adult mortality. Our results strongly indicate that populations of turtles with the age structure and life-history traits of the WRA population would not tolerate increased adult mortality rates. Hence one would predict that unregulated harvesting or other unusual 1319 mortality of adults would decimate such populations quickly, and that recovery would be very slow. Exploitation of such populations even at low levels would have long-term consequences, and probably such populations should be given complete protection from human predation. Acknowledgements We thank the personnel of the Wildlife Research Station for their assistance over the past 15 years. Numerous field assistants helped to collect the data and we grateful to them. In particular, we thank M. L. Bobyn, J. Howell, J. A. Layfield, E. G. Nancekivell, M. E. Obbard, D. Schluter, C. M. Shilton, and G. L. Stephenson. This research was funded by the Natural Sciences and Engineering Research Council of Canada, grant no. A 5990, and by grants from the Ontario Ministry of Natural Resources and the Canadian National Sportsmen's Show to R.J.B. BROCKELMAN, W. Y. 1975. Competition, the fitness of offspring, and optimal clutch size. Am. Nat. 109: 677-699. BROOKS, R. J., GALBRAITH, D. A., NANCEKIVELL, E. G., and BISHOP, C. A. 1988. Developing management guidelines for snapping turtles. In Management of amphibians, reptiles, and small mammals in North America. Edited by R. C. Szaro, K. E. Severson, and D. R. Patton. Proc. U.S. Dep. Agric. For. Serv. Symp. Gen. Tech. Rep. RM-166. pp. 174-179. CONGDON, J. D., and GIBBONS, J. W. 1987. Morphological constraint on egg size: a challenge to optimal egg size theory? Proc. Nat. Acad. Sci. U.S.A. 84: 4145-4147. SELS,R. C., and CONGDON, J. D., BREITENBACH, G. L., VANLOBEN TINKLE, D. W. 1987. Reproduction and nesting ecology of snapping turtles (Chelydra serpentina) in southeastern Michigan. Herpetologica, 43: 39-54. DAVENPORT, J. 1988. The turtle industry of Bali. Br. Herpetol. Soc. Bull. 25: 16-24. DODD,C. K., JR. 1985. Conserving sea turtles: constructive criticism is still needed. Herpetologica, 41: 103- 1 1 1. GALBRAITH, D. A., and BROOKS, R. J. 1987. Survivorship of adult females in a northern population of common snapping turtles, Chelydra serpentina. Can. J. Zool. 65: 1581-1586. 1989. Age estimates for snapping turtles. J. Wildl. Manage. 53: 502-508. W. L., and GALBRAITH, D. A., BISHOP, C. A., BROOKS, R. J., S~MSER, LAMPMAN, K. P. 1988. Factors affecting the density of populations of common snapping turtles (Chelydra serpentina serpentina). Can. J. Zool. 66: 1233-1240. GALBRAITH, D. A., BROOKS, R. J., and OBBARD, M. E. 1989. The influence of growth rate on age and body size at maturity in female snapping turtles (Chelydra serpentina) Copeia, 1989: 896-904. GERMANO, D. J., and JOYNER, M. A. 1988. Changes in a desert tortoise (Gopherus agassizii) population after a period of high mortality. In Management of amphibians, reptiles, and small mammals in North America. Edited by R. C. Szaro, K. E. Severson, and D. R. Patton. Proc. U.S. Dep. Agric. For. Serv. Symp. Gen. Tech. Rep. RM-166. pp. 190-198. GIBBONS, J. W. 1987. Why do turtles live so long? BioScience, 37: 262-269. GREER,K. R. 1955. Yearly food habits of the river otter in the Thompson Lakes Region, Northwestern Montana, as indicated by scat analysis. Am. Midl. Nat. 54: 299-313. HAMILTON, W. J., JR. 1961. Late fall, winter and early spring foods of 141 otters from New York. N.Y. Fish Game J. 8: 106-109. HELWIG,D. D., and HORA,M. E. 1983. Polychlorinated biphenyl, mercury and cadmium concentrations in Minnesota snapping turtles. Bull. Environ. Contan. Toxic01. 30: 186- 190. HOGG,D. M. 1975. The snapping turtles of Wye Marsh. Ont. Fish Wildl. Rev. 14: 16-20. Can. J. Zool. Downloaded from www.nrcresearchpress.com by University of Sydney on 02/05/13 For personal use only. 1320 CAN. J. ZOOL. INTERNATIONAL UNIONFOR CONSERVATION OF NATURE AND NATURAL RESOURCES - Species Survival Commission. 1989. Tortoises and freshwater turtles. An action plan for their conservation. InternationalUnion for Conservation of Nature and Natural Resources. Kelvyn Press Inc. Broadview, IL. KREBS,C. J. 1989. Ecological methodology. Harper and Row, New York, NY. K. F., and OSTENSON, B. T. 1942. Early spring food of the LAGLER, otter in Michigan. J. Wildl. Mange. 6: 244-254. LAMBERT, M. R. K. 1982. Studies on the growth, structure and abundance of the Mediterranean spur-thighed tortoise, Testudo graeca, in field populations. J. Zool. (Lond .) 196: 165- 189. D. J., and OBBARD, M. E. 1977. Tag success, dimensions, LONCKE, clutch size and nesting site fidelity for the snapping turtle, Chelydra serpentina (Reptilia, Testudines, Chelydridae) in Algonquin Park, Ontario, Canada. J. Herpetol. 11: 243-244. MEEKS,R. L., and ULTSCH,G. R. 1990. Overwintering behavior of snapping turtles. Copeia, 1990: 880-884. MELQUIST ,W. E., and HORNOCKER, M . G. 1983. The ecology of river otters. Wildl. Monogr. 83: 1-60. MILLER, R. S . , and BOTKIN,D. B. 1974. Endangered species: models and predictions. Am. Sci. 62: 172- 181. MROSOVSKY, N. 1983. Conserving sea turtles. British Herpetological Society, London. NUDDS,T. D. 1987. The prudent predator: applying ecology and anthropology to renewable resources management. In Wild furbearer management and conservation in North America. Edited by M. Novak, J. A. Baker, M. E. Obbard, and B. Malloch. Ontario Trappers Association. Ministry of Natural Resources, Toronto. pp. 113-1 18. OBBARD, M. E. 1983. Population ecology of the common snapping turtle, Chelydra serpentina , in north-central Ontario. Ph .D . dissertation, University of Guelph, Guelph. 1985. Can we safely harvest snapping turtles in Ontario? Can. Amphibian Reptiles Conserv. Soc. 23: 1-4. OBBARD, M. E., and BROOKS, R. J. 1979. Factors affecting basking in - a northern population of the common snapping turtle, Chelydra serpentina. Can. J. Zool. 57: 435-440. 1981. A radio-telemetry and mark-recapture study of activity View publication stats VOL. 69, 1991 in the common snapping turtle, Chelydra serpentina. Copeia, 1981: 630-637. PARK,E. 1971. The world of the otter. J. P. Lippincott, New York, NY. PRITCHARD, P. C. H. 1980. The conservation of sea turtles: practices and problems. Am. Zool. 20: 609-617. 1983. Review of: Conserving sea turtles by N. Mrosovsky . Copeia, 1983: 1108- 1111. 1989. The alligator snapping turtle. Biology and conservation. Milwaukee Public Museum, Milwaukee, WI. ROFF,D. A. 1981. Reproductive uncertainty and the evolution of iteroparity: why don't flatfish put all their eggs in one basket? Can. J. ZOO^. 38: 968-977. SAS INSTITUTE INC.1985. SASISTAT guide, version 6 edition. SAS Institute Inc., Cary, NC. SCHWARZKOPF, L., and BROOKS, R. J. 1986. Annual variations in reproductive characteristics of painted turtles (Chrysemys picta). Can. J. Zool. 64: 1148-1 151. SHELDON, W. G., and TOLL,W. G. 1964. Feeding habits of the river otter in a reservoir in central Massachusetts. J. Mammal. 45: 449-455. SNOW,J. E. 1982. Predation on painted turtle nests: nest survival as a function of nest age. Can. J. Zool. 60: 3290-3292. STUBBS, D., SWINGLAND, I. R., and HAILEY, A. 1985. The ecology of the Mediterranean tortoise Testudo hermanni in northern Greece (the effects of a catastrophe on population structure and density). Biol. Conserv. 31: 125-152. THOMSON, M. B. 1983. Populations of the Murray River tortoise, Emydura (Chelodina): the effect of egg predation by the red fox, Vulpes vulpes. Aust. Wildl. Res. 10: 363-371. M. 1987. Seasonal changes in food habits TUMLISON, R, and KAINES, of river otters in southwestern Arkansas beaver swamps. Mammalia, 51: 225-231. ULTSCH,G. R. 1989. Ecology and physiology of hibernation and overwintering among freshwater fishes, turtles, and snakes. Biol. Rev. Camb. Philos. Soc. 64: 435-5 16. WILBUR,H. M. 1975. The evolutionary and mathematical demography of the turtle Chrysemys picta. Ecology, 56: 64-77.