Monitoring a population of translocated Grand Cayman blue iguanas: assessing the accuracy and precision of distance sampling and repeated counts

Transcription

1 bs_bs_banner Monitoring a population of translocate Gran Cayman blue iguanas: assessing the accuracy an precision of istance sampling an repeate counts F. J. Burton 1 & F. F. Rivera-Milán 2 1 Blue Iguana Recovery Programme, Gran Cayman, Cayman Islans 2 Division of Migratory Bir Management, US Fish & Willife Service, Laurel, MD, USA Animal Conservation. Print ISSN Keywors Cyclura lewisi; population ensity estimate; istance sampling; population monitoring; repeate counts; restoration; rock iguanas; translocation. Corresponence Freeric J. Burton, Blue Iguana Recovery Programme, PO Box 10308, Gran Cayman, KY1-1003, Cayman Islans. Tel: fjburton@blueiguana.ky Eitor: John Ewen Receive 5 November 2013; accepte 6 April 2014 oi: /acv Abstract In willife translocations, it is important to ientify reliable monitoring methos that can provie accurate population estimates for post-release management. We compare ensity an abunance estimates of istance sampling an repeate counts with complete counts (census ata) to etermine accuracy (percentage relative bias) an precision (coefficient of variation) for monitoring populations of blue iguanas restore to protecte areas on Gran Cayman. We conucte incomplete counts (survey ata) of blue iguanas at 12 transects of unequal length (range = to m) visite 36 times per year, yieling 405 etections in March 2010 an 443 etections in March Distance sampling an repeate counts provie accurate (range of percentage relative bias = 9.52% to 0.00%) an precise (range of coefficient of variation = 0.10 to 0.15) estimates of iguana ensity an abunance, an yiele an estimate of the proportion of iniviuals that were available for etection in the stuy area uring the surveys ( Pˆ a = 088,. se = 0.03). The use of istance sampling an repeate count methos, in carefully esigne surveys with representative coverage an aequate replication, can lea to more reliable monitoring of extant wil an translocate rock iguana populations on other Caribbean islans, where iniviual marking may not be practical, an the proportion of iniviuals that is available for etection is unknown. Introuction Monitoring is essential to guie an evaluate management efforts for the conservation of enangere reptiles (Do & Seigel, 1991). However, monitoring can be resource intensive an excee the uration of typical programs an grant cycles (Kleiman et al., 1991). Therefore, it is important to ientify reliable monitoring methos that can provie accurate an precise estimates of the parameters require to assess an refine management actions. Management of rock iguanas (genus Cyclura) throughout the Caribbean currently involves a range of translocation strategies. For example, heastarting of captive-bre or captive-reare wil iguanas for release at maturity has been shown to reuce juvenile mortality an increase population size (Alberts et al., 2004; Knapp & Huson, 2004; Wilson et al., 2004; Pérez-Buitrago et al., 2008; Burton, 2011); an translocation of breeing age iguanas to unpopulate islans has been use to recolonize or expan the species istribution range (Knapp & Huson, 2004; Gooyear & Lazell, 2006). The nee for reliable monitoring of translocate iguana populations le to a review of methos for estimating ensity an abunance (Hayes & Carter, 2000). These authors argue that istance sampling tens to unerestimate rock iguana ensity an abunance because the metho requires that all iniviuals present in the stuy area are available for etection. They suggeste using markrecapture methos instea of istance sampling, but note that mark-recapture can be costly an mostly applicable to small areas. However, mark-recapture combine with istance sampling has been use successfully on other terrestrial lizars (see e.g. Grant & Doherty, 2010). Other authors have stuie Cyclura populations using raiotracking to estimate the number of iniviuals present at any given time in efine areas (see e.g. Gooman, Echternacht & Burton, 2005, Pérez-Buitrago, Sabat & McMillan, 2010). Population restoration of the Gran Cayman blue iguana Cyclura lewisi into its former range in the Queen Elizabeth II Botanic Park, the Salina Reserve an the Colliers Wilerness Reserve (Fig. 1a) has resulte in breeing populations of known iniviuals an their offspring (Burton, 2011). Monitoring of these populations is neee to inform management an to measure progress towar locally agree conservation objectives (at least 1000 iniviuals in the wil, sustaine by natural reprouction). In this paper, we seek to ientify count methos not requiring the ientification of iniviual iguanas in the stuy area, recognizing that it may not be practical to 40 Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon

2 F. J. Burton an F. F. Rivera-Milán Monitoring translocate population of blue iguanas Figure 1 (a) Gran Cayman, showing location of the Queen Elizabeth II Botanic Park (1), Salina Reserve (2) an Colliers Wilerness Reserve (3). (b) Salina Reserve stuy site, showing survey routes an 12 transect ivisions. (c) Occupie iguana retreats mappe in the three release zones in () Occupie iguana retreats mappe in the three release zones in maintain a marke population for long-term monitoring. We use complete counts (census ata) of marke iguanas as a baseline for assessing the accuracy (percentage relative bias, PRB) an precision (coefficient of variation, CV) of ensity an abunance estimates erive from istance sampling an repeate count methos. We focus on the most basic aspects of these count methos in an effort to reach a wie auience of applie scientists collecting an monitoring ata of rock iguana populations in the Caribbean. Materials an methos Stuy area This stuy was conucte in the Salina Reserve on Gran Cayman, which is an area of xerophytic shrublan on limestone an olostone karst with local basins of re earth. The stuy area comprises three contiguous zones where 307 heastarte iguanas were release between 2004 an 2009, with an aitional 98 iguanas release between 2010 an Iguana surveys were conucte at 12 transects of unequal length (range = to m) in March of 2010 an Transects followe preexisting access routes use for iguana restoration work (see Fig. 1b). Transect centerlines were c. 20 m apart, with minimal vegetation clearance. Surface moification was limite to removal of loose rocks, fragile karst blaes an plant ebris to minimize noise from observers walking on transects. Iguana marking All iguanas in this stuy were tagge at the time of release with unique colore glass bea combinations on a wire piercing the nuchal crest (Roa et al., 1988; Fig. 2a). Before release, each iniviual was also injecte with a passive integrate transponer (PIT tag; BioMark HPT series an similar moels, Biomark, Inc., Boise, ID, USA). Photographs of lateral an orsal aspects of the hea were archive for all iguanas to assist photo ientification in cases of bea or PIT tag loss, base on characteristic variations in hea scale morphology (Fig. 2b). Any wil-born offspring of these release iguanas which were capture before this stuy were tagge an photographe in the same way an release at their capture locations. Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon 41

3 Monitoring translocate population of blue iguanas F. J. Burton an F. F. Rivera-Milán istance sampling; namely, etecting all iguanas at transect centerlines, that is g(0) = 1; etermining initial locations of etecte iguanas, before responsive movement to the approaching observer an measuring perpenicular istances accurately (Bucklan et al., 2001). Because iguanas were etecte as singles, we i not have to estimate cluster size. Transect placement was systematic, proviing representative coverage of the stuy area. Transects were surveye 36 times per year, 18 times each between 09:30 12:00 h an 13:00 15:30 h. Base on raiotracking ata collecte in 2006 (F. J. Burton, unpublishe ata), survey ates an times were selecte to span the maximum activity perio of iguanas, increasing their availability for etection in the stuy area. The observer etecting an iguana fixe their gaze on the initial etection location, an immeiately occupie that point or guie their team partner to o so. The team then use a tape measure to recor the perpenicular istance from the transect centerline to the initial location. Binoculars were use to rea bea tags. We photographe etecte iguanas to confirm any uncertain bea tag ientities. We also measure substrate temperature at the point of initial location an collecte time-synchronize win spee an air temperature ata at the center of the surveye area using a atalogger an sensors (Hobo Micro Station H21, Onset Computer Corporation, Bourne, MA, USA). Figure 2 (a) Bea tag place through nuchal crest of a young Cyclura lewisi. Combinations of two bea sizes an seven colors allow each iguana to be uniquely ientifie. (b) Dorsal view of hea of C. lewisi. Note that lateral asymmetries an iniviual scale anomalies are conserve as iniviuals grow an can be use to assist ientification in event of tag loss. Line-transect surveys Repeate line-transect surveys allowe ata analysis using a variety of methos for inferences about iguana abunance in the stuy area (for a review, see Nichols, Thomas & Conn, 2009). In this paper, we focus on conventional an multiple-covariate istance sampling an repeate count methos (Bucklan et al., 2001, 2004; Royle, 2004a,b; Royle & Dorazio, 2008). We establishe 12 transects with lengths ranging from 323 to 432 m in three geographic zones (Fig. 1b). Three two-observer teams were eploye over the first 3 weeks of March 2010 an 2013 in each of the zones on a rotating scheule, walking all transects in each zone twice per ay (morning an afternoon runs) as quietly as possible at a spee of 0.16 m s 1. We traine observers in the fiel before commencing the surveys. We i not conuct surveys on overcast or rainy ays because iguanas tene to enter or remain close to their retreats uner these conitions. Teams were rotate throughout the survey to ensure that all transects receive equal observer effort from each of the three teams an equal observer effort at all times within the survey perios. The purpose of having teams of two observers was to increase the chance of meeting three basic assumptions of Complete counts Unique bea tag ientities were tabulate from all three survey team observations on a aily basis. Any non-beae iniviuals recognizable by hea scale photographs, istinctive features (such as missing crest spines, tail loss) or consistent association with a particular retreat were given temporary ientity coes until immeiately after the survey when they were trappe an ientifie by transponer reaing. In 2 weeks following the completion of the linetransect surveys, all geographic clusters of observations of unientifie iguanas were intensively monitore to ientify particularly cryptic an shy iniviuals. Concurrently with the surveys, two aitional observers were assigne to observe all etecte iguanas at the beginnings an ens of the ays to fin their primary overnight retreat(s). Locations of all iguanas were mappe using a Global Positioning System as the survey progresse, enabling use of clusters of sightings to assist in locating retreats. Retreats use by the same iguana on at least two consecutive nights were mappe. Aitional retreats of particularly shy an cryptic iniviuals were mappe in the 2-week perio after the surveys were complete, base on the same clusters of unientifie iguana observations investigate for completion of the bea tag census. We consiere this intensive search effort a census (complete count) of iguanas using the stuy area uring the first 3 weeks of March in 2010 an To evaluate the PRB (PRB = 100 [( N) / N]) of istance sampling an repeate count abunance estimates ( ˆN ), we assume that the bea tag census represente the true abunance (N). 42 Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon

4 F. J. Burton an F. F. Rivera-Milán Monitoring translocate population of blue iguanas Distance sampling Conventional istance sampling is base on estimation of a etection function, ĝ( x) in the case of line transects, which ecreases with istance (x) an is neee to estimate iguana etection probability given availability ( ˆP ). However, because iguana etection may be influence by species, environment an observer covariates, we use conventional an multiple-covariate istance sampling for the analysis of line-transect survey (Bucklan et al., 2001, 2004). In the case of multiple-covariate istance sampling, ensity was estimate as ˆ n D =, 2wLPˆ ( z ) where ˆD is the number of iguanas per hectare; n is the number of iguana etections; ˆP is etection probability moele as a function of perpenicular istances an other covariates represente by vector z (i.e. g[x, z]); L is the total length of transects an w is the transect half-with. We poole the istance ata from repeate visits to each transect an recore survey effort as transect length multiplie by the number of visits per year (Bucklan et al., 2001). Detection probability was estimate as an effective strip with as w ˆ 1 P( zi)= ˆ, w g ( x z i) r 0 ESW ˆ = w Pˆ ( z ) after right-truncation of the istance ata at w = 8.98 m (i.e. about 5% of etections within a maximum istance of m). We conucte exploratory ata analysis to etermine the best truncation istance for moeling the etection function (Bucklan et al., 2001). Abunance was estimate by extrapolating estimate ensity to the stuy area (i.e. = A; where A = 7.6 ha). We evaluate the fit of etection moels (uniform, halfnormal an hazar-rate key functions with an without cosine an polynomial ajustment terms) with quantilequantile plots an gooness-of-fit tests (Burnham et al., 2004). Moel selection was base on minimization of Akaike information criterion correcte for small sample sizes (AIC c). We consiere moels with ifferences in AIC c 2 to be equally supporte by the ata. We use non-parametric bootstrapping to account for moel selection uncertainty through moel averaging an for robust estimation of stanar errors (se). The half-normal an hazar-rate key functions with an without cosine ajustment terms were use to explore the effect of factor an numeric covariates on iguana etection; for example, sampling year (2010 or 2013), sampling perio (three 6-ay subsets from the 18 survey ays), time of ay (minutes after sunrise an before sunset), run time (morning or afternoon), survey zone (A, B, C), age (ault or juvenile), i i sex (male or female), activity state (alert or moving in staning posture, or resting in prostrate posture), sun exposure at times of etection (sunny or in clou shaow) an several win spee an temperature measurements. We also use stratification an post-stratification to compare iguana etection an ensity estimates, for example, between survey zones an sampling years using conventional istance sampling (Bucklan et al., 2001). We use the program DIS- TANCE 6.0, Release 2 ( for istance sampling analyses (Thomas et al., 2010). Repeate counts We use the Poisson-binomial mixture moel to estimate iguana etection an abunance (Royle, 2004a,b; Royle & Dorazio, 2008). Repeate transect counts (c it) were efine as binomial ranom variables with inex N i (abunance at transect i = 1, 2,..., R) an etection probability p it (iguanas observe at transect i in occasion t = 1, 2,..., T). We analyze the repeate count ata collecte uring three 6-ay sampling perios (ays 1 6, ays 7 12 an ays 13 18) an an 18-ay sampling perio per year, assuming inepenence between transect visits (12 or 36 per sampling perio) an population closure (i.e. no permanent emigration or eaths an no permanent immigration or births uring the surveys). Through repeate counts, we were able to estimate the probability that an iguana was available for etection an the probability that it was etecte given availability (i.e. P a = P a P ). Assuming that abunance can be represente by a Poisson istribution (P) an counts by a binomial istribution (B), the integrate likelihoo of repeate counts was L ( R T λ, pit cit )= ( cit Ni, p ) it Ni λ Ni max c B i t P( ). i= 1 = ( ) = 1 We accounte for extra-poisson variation in mean abunance λ with an aitive normal ranom effect for log(λ); for example, log(λ i) = α + β 1(Z i)+β 2(Y i) for the survey zone (Z) an sampling year (Y) covariates. A logistic regression moel was use to account for extra-binomial variation in etection; for example, logit(p it) = α + β 1(Z it) +β 2(Y ij), where Z it an Y it are covariate values for transect i an time t. Abunance was interprete as ensity at transect level (i.e. ˆλi πw 2, using the line-transect truncation w = 8.98 m; Royle, 2004b). For the analysis of repeate counts, we use the program PRESENCE, Version 6.1 (US Geological Survey, Patuxent Willife Research Center, MD, USA) (Bailey et al., 2007). Results Census ata The number of unique bea tags etecte in line-transect surveys reache a stable maximum within 18 ays in March 2010 an 2013, representing a census of resient, territorial Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon 43

5 Monitoring translocate population of blue iguanas F. J. Burton an F. F. Rivera-Milán Figure 3 (a) Quantile-quantile plot showing the fitte cumulative an empirical ensity functions an (b) etection probability of blue iguanas (both sexes combine) base on conventional istance sampling (n = 848 etections). (c) Detection probability of male an female blue iguanas, base on multiple-covariate istance sampling (n = 729 etections) after right-truncation of ata at w = 8.98 m. Table 1 Conventional istance sampling (CDS) an repeate count (RC) estimates of etection probability ( P ˆ an Pˆ a ), ensity ( D, ˆ iguanas per hectare) an abunance ( N ˆ ) of blue iguanas surveye in the Salina Reserve, March 2010 an 2013 Metho Year Parameter Mean SE 2.5% 97.5% CDS 2010 Pˆ Pˆ Overall Pˆ RC 2010 Pˆ a Pˆ a Overall Pˆ a Point estimates (mean an SE) are presente with 2.5% an 97.5% quantiles, base on non-parametric bootstrapping. For parameter estimation, we use the CDS an RC ata collecte uring 18 sampling ays (i.e. 36 visits per transect each year). iguanas, as well as those occasionally present in the stuy area. The corresponing area census of active retreats accounte only for permanent resients with stable territories. We mappe 32 retreats an foun 46 iguanas in March 2010, an 30 retreats an 42 iguanas in March 2013 (Fig. 1c,). Survey ata Quantile-quantile plots an gooness-of-fit tests showe no major problems with the iguana istance ata collecte in March 2010 an 2013 (Fig. 3a,b). The largest absolute ifference between the fitte cumulative istribution function an the empirical istribution function i not iffer significantly (Kolmogorov-Smirnov test: D n = 0.04, P value = 0.11). Moreover, the functions were tie over the entire range of the ata (Cramer-von Mises family tests: W 2 = 0.18, P value > 0.30, C 2 = 0.13, P value > 0.20), proviing evience of goo moel fit an little measurement error (Fig. 3a). Among the conventional istance sampling moels consiere, the half-normal key function with no ajustment term an no covariates provie the best fit (χ 2 statistic = 20.55,.f. = 15, P value = 0.15; Fig. 3b). The inclusion of sex as a factor covariate in the half-normal etection moel receive strong support from the istance ata (AIC c = , ΔAIC c = 0). However, sex ifferences in etection probability cause minor heterogeneity in the etection function (Fig. 3b,c). Detection probability an effective strip with was 0.54 (se = 0.03) an 4.85 m (se = 0.24) for male iguanas an 0.58 (se = 0.02) an 5.19 m (se = 0.22) for female iguanas (etection ifference: z statistic = 0.16, P value = 0.44; Fig. 3c); an their perpenicular istances i not iffer significantly (unpaire t statistic = 0.942,.f. = 755, P value = 0.35). Age class (AIC c = 3,017.03, ΔAIC = 48.63), activity state (AIC c = , ΔAIC = ), an other covariates (ΔAIC > ) ha a negligible effect on iguana etection. Moreover, conventional an multiple-covariate istance sampling provie similar ensity estimates, suggesting that moel selection was of seconary importance for abunance inferences from the blue iguana survey ata collecte at the Salina Reserve in March 2010 an Distance sampling etection given availability ( ˆP ) was 0.57 (se = 0.02) for male an female iguanas combine (Table 1). In comparison, repeate count etection, which is the prouct of the probability that iguanas are available for etection an the probability that they are etecte given availability ˆP a ), was 0.50 (se = 0.03). Therefore, the probability that iguanas were available for etection ( ˆP a ) average 0.88 (se = 0.03) in March 2010 an Conventional istance sampling ensity estimates i not iffer significantly among years (z statistic = 0.28, P = 0.39; Table 1). Overall, estimate ensity was 5.28 iguanas per hectare (se = 0.57) an abunance was 40 iguanas (se = 4). Conventional istance sampling an repeate counts provie similar ensity an abunance estimates (Table 1). However, base on point estimates of abunance, PRB 44 Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon

6 F. J. Burton an F. F. Rivera-Milán Monitoring translocate population of blue iguanas Table 2 Percent relative bias (PRB) of estimate abunance ( N ˆ ) base on conventional istance sampling (CDS) an repeate counts (RC) of blue iguanas in the Salina Reserve, March 2010 an 2013 Metho Year 2.5% 97.5% PRB 2.5% 97.5% CDS % 30.43% 15.22% % 35.71% 21.43% Overall % 27.27% 11.36% RC % 28.26% 21.74% % 33.33% 47.62% Overall % 31.82% 34.09% True abunance from census (N) was 46 iguanas in March 2010, 42 iguanas in March 2013, an 44 iguanas taking the average of both years. an PRB estimates are presente with 2.5% an 97.5% quantiles, base on nonparametric bootstrapping. For the evaluation of PRB, we use the CDS an RC ata collecte uring 18 sampling ays (i.e. 36 visits per transect each year). Figure 4 Coefficient of variation (CV) of ensity estimates in 2010 an 2013, as a function of the number of survey ays. Inclue survey ays were selecte from the full series of 18 ays at ranom, an ensity was estimate by conventional istance sampling using the etection function generate from all ata in both years. tene to be more negative for conventional istance sampling (range = 9.52 to 8.70%) than for repeate counts (range = 2.27 to 0.00%; Table 2). The 2.5 an 97.5% quantiles of both count methos inclue true abunances in all instances (Table 1). The count methos showe aequate precision in both years (range of CV = 0.10 to 0.15). Density estimates from conventional istance sampling reache maximum precision within 6 survey ays (12 transect visits), with no ecrease in CV resulting from increasing the survey uration to 12 or 18 ays (Fig. 4). Rather there was some inication that ensity CV increase with longer survey perios (Fig. 4). Discussion The low number of iguanas in the stuy area compare with the numbers release implies that the majority of the release iguanas emigrate from the stuy area or ie. Because we saw release iguanas outsie the stuy area, an saw no evience of any mortality in the stuy area, we hypothesize ispersal from the release area accounts for the unchange population ensity 2010 to 2013 espite release of iguanas between the surveys. The results have immeiate implications for conservation management, focusing attention on strategic questions such as population containment in protecte areas, versus allowing ispersal to recolonize an ancestral range now largely in private ownership an open to unmanage threats. Given the highly mobile nature of the iguanas uring their active perios an the high egree of transect coverage in the stuy area, it shoul be expecte that all iguanas resient in the stuy area woul be encountere by the observer teams at some time uring 18 ays of surveys sprea over 3 weeks, even though some i require follow-up work to establish their actual ientities. This view is consistent with the cessation of new iniviual iguana etections before the en of each survey. The mappe retreat census shoul be consiere an unerestimate of abunance because it i not inclue known iguanas which were active within the stuy area, but were not observe to occupy the same retreat(s) in consecutive nights. The bea tag census can reasonably be consiere a complete count of all iguanas which were either resient, intermittently resient or transiently present in the stuy area at the time of the surveys. March is on average the riest month of the year in the Cayman Islans, an the iguanas are highly active as they are entering their breeing season. The value for ˆP a implie by comparison of conventional istance sampling an repeate counts in our ata may, therefore, represent a high point in seasonal availability of iguanas for etection. Assumption that P a < 1 for istance sampling is avisable for this, an probably other iguana species, because a proportion of iniviuals may remain persistently unergroun in rock holes in any given sampling perio, an temporary emigration from the stuy area may occur (e.g. in response to human presence). Representative coverage an aequate replication of linetransect surveys are essential to increase sample size (number of iguana etections), account for the ifferent components of etection probability (P an P a in our case) an obtain precise ensity estimates (Bucklan et al., 2001; Royle & Dorazio, 2008; Nichols et al., 2009). Replication also allows for rotation of teams an the orer of transect visits to average out any observer an time of ay biases. Practice trials before survey ata collection is avisable to allow for observer training an initial habituation of iguanas to passage of the observers, therefore reucing responsive movement an facilitating recoring perpenicular istances to initial locations. Inaequate replication an failure to formally assess an correct for P a < 1 may explain Hayes & Carter s (2000) observation that istance sampling tene to unerestimate true population ensity of rock iguanas. For rock iguanas, generally, P a shoul be expecte to vary between surveys, an so shoul be assesse to justify abunance inferences (Royle & Dorazio, 2008; Nichols et al., 2009). There are logistical an resource limitation issues associate with obtaining an inepenent measure of P a by raio telemetry Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon 45

7 Monitoring translocate population of blue iguanas F. J. Burton an F. F. Rivera-Milán or mark-recapture methos. Therefore, the combination of istance sampling an repeate counts appears to be an effective option for population monitoring because it generates an estimate of P a without having to mark an iniviually ientify iguanas. Because we wishe to etermine the reliability of istance sampling an repeate count methos using census ata, this stuy occupie a month in total uration an some 133 person-ays in the fiel. Our census an survey efforts in such a small area may not be representative of the reality face by other scientists seeking to conuct population assessments with fewer resources over much larger areas. In our stuy area, we showe that accurate an precise ensity estimates can be obtaine with 12 transect visits, provie transect length (x = m in this stuy) is generally sufficient to obtain one or more iguana etections on each visit. If careful attention is given to the survey esign an observer training (see, e.g. Bucklan et al., 2001: chapter 7), then conventional an more avance istance sampling an repeate count methos can be use (see, e.g. Royle & Dorazio, 2008: chapter 7) to monitor the populations of other Cyclura species in the Caribbean, accounting for spatial an temporal ifferences in P an P a. Finally, as a result of this stuy, we have etermine that annual monitoring of blue iguanas in the Salina Reserve can be effectively conucte with a thir of the transect visits mae in March 2010 an Acknowlegments We thank D. Bell, L. Casolino, C. Easby, J. Freeman, M. Goetz, C. Lee, M. Rasmussen, C. Robertson, P. So, C. Stine an S. Whitaker who (with the first author) constitute the iguana survey teams in 2010 an Geenlight Re (a reinsurance company base in the Cayman Islans an Irelan) provie financial support throughout. Roger an Mary Bumgarner provie free accommoation for the survey teams. The Blue Iguana Recovery Program is a program of the National Trust for the Cayman Islans, with local an international partners. Proucts an commercial technologies mentione in this article are not necessarily enorse by our organizations. References Alberts, A.C., Lemm, J.M., Grant, T.D. & Jackintell, L.A. (2004). Testing the utility of heastarting as a conservation strategy for West Inian iguanas. In Iguanas: biology an conservation: Alberts, A.C., Carter, R.L., Hayes, W.K., & Martins, E.P. (Es). Berkeley an Los Angeles, California: University of California Press. Bailey, L.L., Hines, J.E., Nichols, J.D. & McKenzie, D.I. (2007). Sampling esign trae-offs in occupancy stuies with imperfect etection: examples an software. Ecol. Appl. 17, Bucklan, S.T., Anerson, D.R., Burnham, K.P., Laake, J.L., Borchers, D.L. & Thomas, L. (2001). Introuction to istance sampling: estimating abunance of biological populations. Oxfor: Oxfor University Press. Bucklan, S.T., Anerson, D.R., Burnham, K.P., Laake, J.L., Borchers, D.L. & Thomas, L. (2004). Avance istance sampling: estimating the abunance of biological populations. Oxfor: Oxfor University Press. Burnham, K.P., Bucklan, S.T., Laake, J.L., Borchers, D.L., Marques, T.A., Bishop, J.R.B. & Thomas, L. (2004). Further topics in istance sampling. In Avance istance sampling: estimating abunance of biological populations: Bucklan, S.T. et al. (Es). Oxfor: Oxfor University Press. Burton, F.J. (2011). Population restoration for a critically enangere reptile the Gran Cayman blue iguana (Cyclura lewisi). Reptile Australas. 1, Do, C.K. Jr. & Seigel, R.A. (1991). Relocation, repatriation an translocation of amphibians an reptiles: are they conservation strategies that work? Herpetologica 47, Gooman, R.M., Echternacht, A.C. & Burton, F.J. (2005). Spatial ecology of the enangere iguana, Cyclura lewisi, in a isturbe setting on Gran Cayman. J. Herpetol. 39, Gooyear, N. & Lazell, J. (2006). Status of a relocate population of enangere Iguana pinguis on Guana Islan, British Virgin Islans. Restoration Ecol. 2, Grant, T.J. & Doherty, P.F. (2010). Monitoring of the flattaile horne lizar with methos incorporating etection probability. J. Will. Manage. 71, Hayes, W.K. & Carter, R.L. (2000). Population monitoring. In West Inian iguanas: status survey an conservation action plan: Alberts, A. et al. (E.). Glan, Switzerlan an Cambrige: IUCN. Kleiman, D.G., Beck, B.B., Dietz, J.M. & Dietz, L.A. (1991). Costs of a re-introuction an criteria for success: accounting an accountability in the Golen Lion Tamarin Conservation Program. In Beyon captive breeing: re-introucing enangere mammals to the wil, Vol. 62: Gipps, J.H.W. (E.). Oxfor: Symposia of the Zoological Society of Lonon. Knapp, C.R. & Huson, R.D. (2004). Translocation strategies as a conservation tool for West Inian iguanas. In Iguanas: biology an conservation: Alberts, A.C. et al. (Es). Berkeley an Los Angeles, California: University of California Press. Nichols, J.D., Thomas, L. & Conn, P.B. (2009). Inferences about lanbir abunance from count ata: recent avances an future irections. In Moeling emographic processes in marke populations: Thomson, D.L. et al. (Es). New York: Springer. Pérez-Buitrago, N., Garcia, M.A., Sabat, A.M., Delgao, J., Alvarez, A., McMillan, W.O. & Funk, S.M. (2008). Do heastart programs work? Survival an boy conition in heastarte Mona Islan iguanas Cyclura cornuta stejnegeri. Enanger. Species Res. 6, Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon

8 F. J. Burton an F. F. Rivera-Milán Monitoring translocate population of blue iguanas Pérez-Buitrago, N., Sabat, A.M. & McMillan, W.O. (2010). Spatial ecology of the enangere Mona Islan iguana Cyclura cornuta stejnegeri: oes territorial behaviour regulate ensity? Herpetol. Monogr. 24, Roa, G.H., Bock, B.C., Burghart, G.M. & Ran, A.S. (1988). Techniques for ientifying iniviual lizars at a istance reveal influences of hanling. Copeia 1988, Royle, J.A. (2004a). Generalize estimators of avian abunance from count survey ata. Anim. Bioivers. Conserv. 27, Royle, J.A. (2004b). N-mixture moels for estimating population size from spatially replicate counts. Biometrics 60, Royle, J.A. & Dorazio, R.M. (2008). Hierarchical moeling an inference in ecology: the analysis of ata from populations, metapopulations an communities. Lonon: Acaemic Press. Thomas, L., Bucklan, S.T., Rexta, E.A., Laake, J.L., Stinberg, S., Heley, S.L., Bishop, J.R.B., Marques, T.A. & Burnham, K.P. (2010). Distance software: esign an analysis of istance sampling surveys for estimating population size. J. Appl. Ecol. 47, Wilson, B.S., Alberts, A.C., Graham, K.S., Huson, R.D., Kerr-Bjorklan, R., Lewis, D.S., Lung, N.P., Nelson, R., Thompson, N., Kunna, J.L. & Vogel, P. (2004). Survival an reprouction of repatriate Jamaican iguanas. In Iguanas: biology an conservation: Alberts, A.C. et al. (Es). Berkeley an Los Angeles, California: California University Press. Animal Conservation 17 (Suppl. 1) (2014) The Zoological Society of Lonon 47