Literaturdatenbank
Literaturdatenbank |
|
 |
|
Langtimm, C. A., Dodd, C., & Franz, R. (1996). Estimates of abundance of box turtles (terrapene carolina bauri) on a florida island. Herpetologica, 52(4), 496–504.
Added by: Admin (23 Aug 2008 12:14:15 UTC) |
| Resource type: Journal Article BibTeX citation key: Langtimm1996 View all bibliographic details  |
Categories: General Keywords: Emydidae, Habitat = habitat, Nordamerika = North America, Schildkröten = turtles + tortoises, Terrapene, Terrapene carolina Creators: Dodd, Franz, Langtimm Collection: Herpetologica |
Views: 1/444
Views index: 8% Popularity index: 2% |
| Abstract |
|
{Terrapene carolina ABSTRACT: From 1991-1993, we conducted an intensive mark-recapture study of an islandpopulation of Florida box turtles, Terrapene carolina bauri. We used models based on the Jolly-Seber open population model to estimate abundance, after first examining the capture histories forviolations of the underlying assumptions of mark-recapture analysis with goodness-of-fit tests con-tained in the program RELEASE. Some violations were detected, resulting in low estimates. Weestimated that 544 (415-672 approximate 95% confidence interval) adult box turtles inhabited thesouthern 36.4 ha of Egmont Key in the summer of 1993. This estimate is similar to those ofconspecific populations studied in more northern areas and provides a reliable initial estimate withwhich to monitor population trends in the future. Adult capture probabilities ranged between 0.09and 0.30. Weekly survival probabilities for adults ranged between 0.94 and 1.00. There weresignificant differences in capture histories between adults and juveniles, indicating differences insurvival and/or capture probabilities. Increased movement and activity of individuals on the studyarea during two survey periods, one after a severe storm and the other during the fruiting seasonof a favorite food source, was verified with one of the goodness-of-fit tests. EASTERN box turtles (Terrapene caro-lina) are often perceived as common.However, long-term studies of the popu-lation dynamics of this species have beenconducted only in the northern third of itsrange (Schwartz et al., 1984; Stickel, 1978;Williams and Parker, 1987). Empirical dataare lacking for the Florida box turtle (Ter-rapene carolina bauri). Beginning in 1991,we began a mark-recapture project mon-itoring a population of Florida box turtlesthat inhabits Egmont Key, an island situ-ated at the entrance to Tampa Bay harbor,Hillsborough County, Florida (Dodd et al.,1994). Although the 180-ha island has along history of human occupation (Franzet al., 1992), it presently is a NationalWildlife Refuge and State Park with lim-ited access to the public.In addition to possible comparisons with4 PRESENT ADDRESS: Department of Forestry andWildlife Management, University of Massachusetts,Amherst, MA 01003-4210, USA.more northern populations, monitoring thisparticular population is of interest in termsof its population dynamics. The island onwhich it is found is subject to various nat-ural and human-related influences that canaffect population dynamics, includingflooding from storms and hurricanes (seeFranz et al., 1992, for a review) and habitatmanagement by park personnel (Dodd etal., 1994). A realistic assessment of changesin abundance through time can provideinsight concerning the impact that theseinfluences may have on this long-lived spe-cies.Previous long-term studies of box turtleshave indicated declines in populations(Schwartz et al., 1984; Stickel, 1978; Wil-liams and Parker, 1987). These studies,however, lacked variance estimates fortheir estimates of population size. Rigorousstatistical methods to detect time trendsand to test hypotheses concerning tem-poral differences in population size requireestimates of variance (Link and Nichols,1994; Skalski and Robson, 1992). Open and closed mark-recapture models now pro-vide the statistical methods to estimatepopulation size and variance, but biologistsusing these methods with many aquaticand terrestrial turtle populations oftenoverlooked the assumptions involved inopen and closed population models (re-viewed by Lindeman, 1990). Violations ofassumptions can seriously bias estimates(Carothers, 1973; Gilbert, 1973).New mark-recapture statistical modelsare now available in user-friendly pro-grams that run on personal computers(Nichols, 1992), and goodness-of-fit (GOF)tests have been developed that test for vi-olations of the equal probability of captureassumption that must be met in order toapply the statistical models to the data ap-propriately. Differences in capture prob-abilities identified by these same GOF testsalso can provide information on behaviorpatterns. We used these new methods andcomputer programs to examine our cap-ture data for violations of assumptions andto estimate abundance during 1992 and1993. Our analysis provided an initial es-timate for future comparisons as we con-tinue to monitor the population. We alsocompared our results with estimates of boxturtle numbers in habitats and latitudesdifferent from our study site.METHODSCapture data used in the analysis werecollected on 11 three-day surveys of Eg-mont Key from March 1991 through July1993 (see Table 2 below for dates). Becauselogistical problems prevented us from sys-tematically sampling the entire 180 ha ofthe island, we restricted our analysis todata collected on the southern part of theisland, south of the Cross Island Trail (Fig.1 in Dodd et al., 1994). The area totaledapproximately 36.4 ha. The most extensivehabitats were forest composed of cabbagepalm (Sabal palmetto), Australian pine(Casuarina equisetifolia), and Brazilianpepper (Schinus terebinthifolius), andmowed lawn at the compound of the Tam-pa Bay Pilot's Association.Turtles were captured by hand and clas-sified according to sex and adult or juvenilestatus. Juveniles were <10 cm carapacelength and could not be identified to sex.Each individual was marked with a pat-tern of unique shell notches for futureidentification (Cagle, 1939) and released.We compiled capture histories by record-ing the presence or absence of individualsduring each of the 11 survey periods. Ad-ditional descriptions of the habitat and fieldmethodology are contained in Dodd et al.(1994).The choice of a statistical approach toestimate abundance from mark-recapturedata depends on whether the populationis closed or open to birth/immigration anddeath/emigration. A population may beconsidered closed if the sampling periodis short in duration such that birth anddeath or immigration and emigration arenon-existent. If data are collected overweeks or months, individuals entering orleaving the population confound the anal-ysis and open population models must beused. Closed population models permit re-laxation of the assumption of equal captureprobabilities, and hence they provide morerobust estimates of population size thanopen models when this assumption is notmet (Pollock, 1982). However, logisticalconstraints prevented us from making ex-tended trips to the island to collect capturedata under a closed design. We thereforeused the open population approach.Before applying open models to the data,we assessed how well the capture historiesmet the following assumptions necessaryfor mark-recapture analyses. (1) Everymarked individual present in the popula-tion immediately after the ith sample hasthe same probability of survival (0i) untilthe following sample (i + 1). (2) Everyindividual in the population at the time ofthe ith sample has the same probability ofcapture (p,). (3) Marks are not lost or over-looked. (4) All samples are instantaneousand each release is made immediately af-ter the sample.Assumption 3 was not a problem in thisstudy, because the shell notches provideda permanent and unambiguous mark. Infour years, we never observed any weararound the notches that would cause themto be misread. Assumption 4 was also met by our methods. Because each survey wasrestricted to three days, the sampling in-terval was short enough not to be affectedby death/emigration or birth/immigra-tion. We ascertained the validity of as-sumptions 1 and 2 with several GOF testsin program RELEASE (Burnham et al.,1987).We examined the equal probability ofcapture and survival assumptions bygrouping individuals a priori according tocategories that one might consider as beingsubject to different mortality regimes oractivity patterns and then testing for dif-ferences in capture histories among groupswith TEST1 of program RELEASE. Thethree groups that we considered were adultmales, adult females, and juveniles. Wespecifically examined the equal probabil-ity of capture assumption for each groupand for all groups combined using TEST2and TESTS. TEST2 detects differences incapture probability within individual cap-ture histories due to a response to beingcaptured (i.e., "trap happiness" or "trapshyness": Pradel, 1993). TESTS detects dif-ferences in capture probability among in-dividuals due to previous capture history(the most common case being the presenceof transients, individuals caught only onceand then never recaptured: Burnham etal., 1987; Paradis et al., 1993).We estimated population size, captureprobabilities, and survival probabilities foreach survey period with program JOLLY(Pollock et al., 1990). Three variations ofthe classic Jolly-Seber open populationmodel were compared for suitability to thedata for box turtles: Model A, the generalJolly-Seber model which allows capture (p)and survival (tq) probabilities to vary overthe survey periods (i) (4i, pi, in the notationof Lebreton et al., 1992); Model B, whichassumes that survival probabilities are con-stant over the study period while captureprobabilities vary (q, pi); and Model D,which assumes both the survival and cap-ture probabilities are constant during thestudy period (q, p). The best fitting modelfor estimating the parameters was chosenbased on likelihood ratio tests betweenmodels.RESULTSWe captured and marked 718 box tur-tles during the study. There were 365males, 248 females, 102 juveniles, and threeadults of unknown sex.Tests for Violations of AssumptionsSome violations of the assumptions ofequal probability of capture and survivalwere indicated. TEST1 of RELEASEshowed no significant differences in cap-ture histories between adult males and fe-males (x2= 19.45, 19 df
Added by: Admin |