Bias in sea turtle productivity estimates: error and factors involved

Abstract

The conservation and management of endangered species, including sea turtles, require consistent long-term monitoring of productivity (e.g., number of hatchlings emerged per nest, per female, per nesting site, per population). In sea turtle species, some of the relevant data are obtained by estimating the number of hatched eggs from fragments found in the nest after hatching. Yet, no formal assessment of the associated error has been published. Here we provide an estimation of the error associated with the most widespread method used to determine sea turtle productivity (count of shell fragments > 50% of the egg size) using a large dataset (n = 8539) of actual and estimated clutch sizes of four sea turtle species (Caretta caretta, Chelonia mydas, Dermochelys coriacea, Eretmochelys imbricata). The data are analyzed through linear mixed models with several explanatory variables. Results show that the error can be large in certain cases and, when the associated error rate is not adequately considered, may represent a serious problem in studies on reproductive parameters such as clutch size. Some significant explanatory variables suggest that some sources of error are linked to species-specific biological traits (e.g., clutch size, egg size, nest depth), and others imply human error. Other biotic and abiotic factors may also be involved. We recommend that—whenever possible—errors be assessed and adequately reported by studies that estimate clutch size, hatching and emergence success, or hatchling production.

Introduction

The conservation and management of endangered species require consistent long-term monitoring of various parameters. One of the most important to the understanding of population dynamics and its implication for conservation is productivity, i.e., the recruitment of new cohorts to the population (Weaver et al. 1996; Anders and Marshall 2005; Carrete et al. 2006).

Sea turtles are long-lived, late-maturing, and highly migratory species of conservation concern that are studied primarily on nesting beaches, on which they are easily accessible. Sea turtles are tied to land for reproduction, which makes it easier to monitor them than other migratory marine taxa (e.g., Evans and Hammond 2004; Holmberg et al. 2009). Breeding females come ashore to lay egg clutches in nests (hereafter, nests). Nests can be easily counted along beaches, providing an indirect proxy for female abundance (e.g., Ceriani et al. 2019). Moreover, nests can provide information on natural productivity (number of hatchlings emerged per nest, per female, per nesting site, per population) if three reproductive parameters are adequately assessed: clutch size (number of eggs per clutch), hatching success (number of hatched eggs/clutch size), and emergence success (number of hatchlings that emerge from the nest unaided/clutch size). The accuracy of the methods used to estimate these parameters is fundamental to obtaining unbiased productivity estimates necessary to model population dynamics (e.g., Crouse et al. 1987; Marcovaldi and Chaloupka 2007; Mazaris et al. 2009; Dethmers and Baxter 2011; Casale and Heppell 2016; Piacenza et al. 2017). Accordingly, identifying reliable methods for estimating demographic parameters is one of the key questions for sea turtle conservation and management (Rees et al. 2016).

Clutch size can be determined in four ways, all of which may be subject to error or have a negative impact on sea turtle nesting or reproductive success: (i) counting eggs while the female is laying (error possible in counting dropping eggs with limited visibility and potential disturbance of the nesting female and thus the nesting process); (ii) excavating the nest shortly after laying to count eggs (which may alter the incubation environment); (iii) counting eggs during nest relocation (which may cause movement-induced mortality or alter the incubation environment); and (iv) counting eggs at nest inventory (when the number of eggs that are not whole must be estimated from eggshell fragments).

A nest inventory, conducted after hatchlings have left the nest, is a component of many sea turtle nesting monitoring programs and does not have conservation implications because it does not impact the female or the developing eggs. Estimated clutch size at nest inventory has been used as the reference clutch size for a population (e.g., Tucker and Frazer 1991; Niethammer et al. 1997; Godley et al. 2001; Cardona et al. 2014) and as the reference clutch size in productivity (e.g., Tiwari et al. 2006) and population assessments (e.g., Piacenza et al. 2017), when the actual clutch size obtained through the other methods was not available. Nest inventory is also the only way to determine the number of unhatched eggs, count live and dead hatchlings remaining in a nest, and estimate the number of eggs that hatch and the number of hatchlings that emerge from the nest (Fowler 1979; Brost et al. 2015). The latter cannot be determined empirically apart from rare instances in which all hatchlings are intercepted and counted, as when using restraining cages (e.g., Garcia et al. 2003). From these numbers, hatching and emergence success are also obtained and annual hatchling production can be estimated (Tiwari et al. 2006; Tomillo et al. 2009; Brost et al. 2015).

At nest inventory, the number of hatched eggs is estimated by counting eggshell fragments. This can be done by counting only shells that make up more than 50% of the egg size (Miller 1999), by piecing together fragments (e.g., Tucker and Frazer 1991; Bouchard and Bjorndal 2000; Kornaraki et al. 2006), or by a combination of the two (Whitmore and Dutton 1985; Turkozan et al. 2011). The first method (shells > 50% of the egg size) is possible because sea turtle eggs have flexible aragonite shells (Miller 1999; Miller et al. 2003) that do not fragment during hatching, compared to eggs of other Sauropsida whose eggshells are much more calcareous (Mikhailov 1997; Hincke et al. 2012). Thus, the eggshell remains mostly intact. Miller (1999) formally described the method that he acquired from Limpus in the mid-1970s (Miller pers. comm.), and Limpus (pers. comm.) confirms to have been using this approach since then. It is likely, however, that this intuitive method was developed independently by different researchers (e.g., Ehrhart pers. comm.) at that time. In the recent literature, counting only shells that make up more than 50% of the egg size (Miller 1999) has become the most widely used method across sea turtle species and continents (Tiwari et al. 2006; Bell et al. 2007; Pérez-Castañeda et al. 2007; Tomillo et al. 2009; Bevan et al. 2014; Pendoley et al. 2014; Balazs et al. 2015; Brost et al. 2015; Behera et al. 2018; Olendo et al. 2019).

As with any method, however, estimating the number of hatched eggs is subject to error and depends, in part, on the skill of the person doing the excavation, any undetected underground predation, and the level of fragmentation of the eggshells. Miller (1999) did not provide a validation of the method or estimate associated error, although he suggested that it should be done. While many studies report on hatching and emergence success, most lack details on the criteria used to count hatched eggs (e.g., Eckert and Eckert 1990; Niethammer et al. 1997; Broderick and Godley 1999; Kamel and Mrosovsky 2005; Caut et al. 2006). Studies rarely include some appropriate measure of counting error (e.g., Marco et al. 2015), while more often the information is either incomplete (Fowler 1979; Turkozan et al. 2011) or based on a small sample (e.g., n = 9, Bouchard and Bjorndal 2000; n = 15 per species, Broderick et al. 2003). Formal evaluations of counting error might be a component of individual projects (e.g., Marco et al. 2015; sea turtle nesting projects in Australia, Limpus pers. comm.) but the problem of how well a count of egg fragments might match the number of eggs that hatched has not been carefully examined, formalized, validated, and published.

The aim of this study was to assess the error associated with the most widespread method of estimating hatched eggs and consequently clutch size and hatchling production, using a large dataset of actual and estimated clutch sizes for four sea turtle species.

Materials and methods

Data collection and potential sources of error

This study was conducted in the Unites States (Florida) and Panama. Data on loggerhead turtles (Caretta caretta), green turtles (Chelonia mydas), and leatherback turtles (Dermochelys coriacea) were collected in Florida from 2001 to 2019 from 55 permit holders monitoring 130 beaches around the state. The Florida Fish and Wildlife Conservation Commission (FWC) oversees sea turtle nesting monitoring in Florida. The FWC, under a cooperative agreement with the U.S. Fish and Wildlife Service, issues the permits, provides training in nest survey techniques, and compiles nesting data. Data are collected by a network of individuals (including representatives of conservation organizations; local, state and federal government personnel; academics; and consultants) who hold permits from the FWC allowing them to conduct research and conservation activities on sea turtles. Each permit holder, in turn, oversees several individuals involved in data collection. FWC encourages as little manipulation and intervention as possible and considers relocation a management technique of last resort because moving eggs creates opportunities for adverse impact (e.g., Limpus et al. 1979; Whitmore and Dutton 1985; Mortimer 1999; Pintus et al. 2009); thus, only a small fraction of nests is relocated each year in Florida (~ 1% of nests/year; Florida Fish and Wildlife Conservation Commission 2016). Nests can be relocated only by the permit holder or authorized personnel with appropriate training (Florida Fish and Wildlife Conservation Commission 2016). Relocated nests are monitored and inventoried following a standardized protocol found in the FWC’s Marine Turtle Guidelines (FWC 2016; https://myfwc.com/media/3133/fwc-mtconservationhandbook.pdf). The number of hatched eggs is determined by counting only shells that make up more than 50% of the egg size, per Miller (1999).

Data on hawksbill turtles (Eretmochelys imbricata) were collected in Panama from 2005 to 2012 and during 2016 at the Zapatilla Cays (9.266° N, − 82.060° W), two adjacent islands on the Caribbean coast that are part of Bastimentos Island National Marine Park (Meylan et al. 2013). Nest surveys in Panama are conducted primarily by indigenous community members assisted by a small number of international students. Six to eight people are involved each year, alternating between the two cays. Nearly all nests are left situ, but each year a small number are relocated to a higher position on the beach due to threat from erosion. As in Florida, the number of hatched eggs is determined by counting only shells that make up more than 50% of the egg size. Surveyors receive training in relocation and nest inventory methods from project leaders. Permission to carry out the work is granted by the Areas Protegidas division of the Ministerio de Ambiente.

To investigate the error associated with this method (eggshells > 50% of the egg size; Miller 1999), we analysed records of nests not impacted by erosion and with no evident sign of predation where both the actual number of eggs laid (counts obtained during nest relocation) and the estimate obtained from eggshell fragments at nest inventory were available. We cannot, however, exclude the possibility that a few eggs may have been removed by underground predators like crabs and snakes.

The error (ERR) in estimating the number of hatched eggs was defined as:

$${\text{ERR}}\, = \,{\text{EST}}\, - \,{\text{EXP}}.$$

where EST is the number of hatched eggs estimated from the eggshell fragments (eggshells > 50% of the egg size, Miller 1999) found at inventory, and EXP is the expected number of hatched eggs, calculated as.

$${\text{EXP}}\, = \,{\text{EGGS}}\, - \,({\text{P}}\, + \,{\text{W}}\, + \,{\text{D}}).$$

where EGGS is the number of eggs counted while relocating a nest; P is the number of live or dead pipped hatchlings (still in their shells); W is the number of whole and undamaged unhatched eggs; and D is the number of damaged unhatched eggs. P, W and D were counted at the inventory, as were live and dead hatchlings in a nest (H).

The following types of error are hypothesized: (i) erroneous estimation (EST) of the actual number of hatched eggs using the fragments found at inventory; (ii) erroneous count of EGGS, H, P, W or D; (iii) undetected predation; (iv) mistaken nest ID at inventory causing a mismatch between relocation and inventory data (i.e., they came from different nests); (v) errors recording on datasheets (typographical errors, data placed in wrong field); and (vi) errors in entering data into databases. While the aim of this study was to assess (i), it should be considered that (i) may be confounded with all the other sources of errors.

Data analysis

All analyses were performed in R version 4.0.3 (R Development Core Team 2020), and graphs were generated through R packages ggplot2, sjPlot and ggpubr.

First, a suspected source of error of type iv in clutches relocated in hatcheries (a protected and enclosed area on the beach where multiple nests are transferred from other sites) was investigated by comparing ERR between clutches relocated in hatcheries and those relocated just more landward from the deposition site. This comparison was implemented only on C. caretta clutches (the only species relocated to hatcheries) through a Mann–Whitney’s U test.

The effect of species (SPECIES), year (YEAR), team that collected the data (TEAM), number of hatchlings found in the nest (H), and EXP on ERR was investigated through linear mixed models (LMM) or linear models (LM) run by the functions lmer or lm, respectively, of the R package lme4. Test of the significant effect of terms was performed through a stepwise backward elimination of terms of the most general LMMs by the step function of the R package lmerTest. The most general LMM was in the form:

$${\text{ERR}}\,\sim \,{\text{SPECIES}}*{\text{ TEAM}}\, + \,{\text{H}}\, + \,{\text{EXP}}\, + \,({1 }|{\text{ YEAR}}).$$

where YEAR was included as a random factor and SPECIES and TEAM as interacting factors. For SPECIES and TEAM, the level with the most records was set as the first (reference) level. Significance level was considered as p < 0.05. H was hypothesized to contribute to ERR because hatchlings staying in the nest for a prolonged time might have further fragmented the shells while moving, resulting in fewer shell fragments > 50% to count. EXP was hypothesized to contribute to ERR for the same reason and because a greater number of hatched eggs may increase the probability of a counting error (error source of type i or ii). Three LMMs for each species were also run: one with the same explanatory variables (except SPECIES).

$${\text{ERR}}\,\sim \,{\text{TEAM}}\, + \,{\text{H}}\, + \,{\text{EXP}}\, + \,({1 }|{\text{ YEAR}}).$$

one with an additional explanatory variable V (number of days from emergence to nest inventory).

$${\text{ERR}}\,\sim \,{\text{TEAM}}\, + \,{\text{H}}\, + \,{\text{EXP}}\, + \,{\text{V}}\, + \,({1 }|{\text{ YEAR}}).$$

and one with a different additional explanatory variable I (incubation period: number of days from egg laying to emergence), where interaction of I and H or EXP were included:

$${\text{ERR}}\,\sim \,{\text{TEAM}}\, + \,{\text{H }}*{\text{ I}}\, + \,{\text{EXP}}\, + \,{\text{I}}\, + \,({1 }|{\text{ YEAR}}).$$
$${\text{ERR}}\,\sim \,{\text{TEAM}}\, + \,{\text{H}}\, + \,{\text{EXP }}*{\text{ I}}\, + \,({1 }|{\text{ YEAR}}).$$

The latter two LMMs were run on subsets of records for which data about V or I were available. V might affect ERR if a longer permanence of live hatchlings in the nest caused more extensive fragmentation of eggshells, while I may affect ERR through longer effects of environmental factors. Because clutch size (EGGS) is the most common parameter used, the relationship ERR ~ EGGS was also provided through a separate LM for each species.

Results

A total of 8539 records of clutches with the necessary data were obtained. Error was significantly greater for C. caretta clutches placed in hatcheries (n = 1017) than for those not placed in hatcheries (n = 6561) (Mann–Whitney U test, p < 0.001, n = 7578). Consequently, clutches placed in hatcheries were excluded, and further analyses were conducted on a sample of 7139 clutches (see Table 1 for clutch size and incubation duration statistics).

Table 1 Clutch size and incubation duration (days from nesting to first emergence) of clutches of four sea turtle species (n = 7139)

In the general LMM, the random factor YEAR and the explanatory variables SPECIES, TEAM and EXP showed a significant effect on ERR, while H and the interaction SPECIES * TEAM were not significant (see Supplementary Table 1 for results of the LMM where the non-significant terms are removed). Error for D. coriacea and E. imbricata, but not for C. mydas, was significantly greater than error for C. caretta. Therefore, separate LMMs were conducted for each species. The distribution of the error by species is shown in Figs. 1 and 2, with E. imbricata showing the greatest error (median =  − 5; IQR: − 15 to − 1; n = 145), followed by D. coriacea (median =  − 2; IQR: − 7 to 0; n = 146), C. mydas (median = 0; IQR: − 1 to 0; n = 287), and C. caretta (median = 0; IQR: − 1 to 0; n = 6561). LMMs conducted on each species provided different results. C. caretta showed significant effects on the error by all the random factor (YEAR), TEAM and EXP, while H was at the threshold (p = 0.051). No significant variables resulted for C. mydas. D. coriacea showed significant effects by TEAM, while H was at the threshold (p = 0.050). The LMM for E. imbricata did not include TEAM (because only one TEAM collected this dataset) and showed a significant effect by EXP only. The relationship between the error and EXP for the two species where EXP was significant is shown in Fig. 3. Given that clutch size (total number of eggs) is the most common parameter used, the relationship between the error and EGGS for the same two species is also provided in supplemental Fig. 1. No significant effect of V (n = 5549) or I (n = 6394) was observed in any species. A significant effect of I (n = 6386) was observed only in C. caretta and when interaction with EXP was considered.

Fig. 1
figure1

Frequency distribution of the error in estimating the number of hatched eggs from their fragments in four sea turtle species: Caretta caretta (Cc; n = 6561), Chelonia mydas (Cm; n = 287), Dermochelys coriacea (Dc; n = 146), and Eretmochelys imbricata (Ei; n = 145)

Fig. 2
figure2

Error in estimating the number of hatched eggs from their fragments in four sea turtle species: Caretta caretta (Cc; n = 6561), Chelonia mydas (Cm; n = 287), Dermochelys coriacea (Dc; n = 146), and Eretmochelys imbricata (Ei; n = 145). Box: 25th, 50th (median), and 75th percentiles; whiskers: 95% range; dots: outliers

Fig. 3
figure3

LM-predicted relationship between the error in estimating the number of hatched eggs from their fragments and the number of expected empty eggshells in the two sea turtle species where the latter variable showed a significant effect: Caretta caretta (Cc; n = 6561), Eretmochelys imbricata (Ei; n = 145). Shadow: 95% CIs

Discussion

This study provides the first formal evaluation of the error associated with the estimation of the number of hatched eggs through the most widespread method used in sea turtle studies (i.e., using eggshells > 50% of the egg size; Miller 1999). These results can help projects worldwide to improve the estimation of the number of hatched eggs and, therefore, the estimation of parameters based on the number of hatched eggs (clutch size; hatching and emergence success) needed to estimate productivity. We found that the error can be large in certain cases and might represent a serious problem for studies of the above-mentioned reproductive parameters if not adequately considered. The results confirm the most intuitive and expected sources of error of type (i) (erroneous estimation of the actual number of hatched eggs), at least in some species, although other sources may be implicated. Some of them are biological traits, while others imply human error. Both types should be carefully considered in future studies.

First, clutch size—or, more specifically, the number of hatched eggs—had a strong effect on error for two species (C. caretta and E. imbricata). As explanations, we propose the increased probability of making an error while counting a large number of eggshells and the greater fragmentation of eggshells produced by a higher number of hatchlings in the nest. The same explanation can be proposed for the almost significant effect on error, for C. caretta and D. coriacea, of the number of hatchlings found in the nest.

Second, an effect of incubation duration on error was observed, but only in C. caretta, perhaps because of the larger sample size available for this species. As explanation, we propose that eggs incubating for longer periods are more likely to show a higher degree of decomposition due to increased exposure to environmental variables (e.g., moisture [rainfall, tidal exposure] and bacteria that may increase the likelihood that project personnel conducting nest inventory will break eggs while pulling them out of the egg chamber.

Third, a difference among species was observed. The error was smallest for C. caretta and C. mydas, intermediate for D. coriacea and largest for E. imbricata. We hypothesize that the differences in counting error among species resulted from a combination of the following interspecific differences: (i) clutch size, (ii) egg size, and (iii) location and microenvironment in which the eggs develop. However, since E. imbricata data were collected by only one team, the effects of species and team are confounded here, and a human error cannot be excluded (see below). Of the species considered, D. coriacea lays the smallest clutches, followed by C. caretta and C. mydas, while E. imbricata lays the largest clutches (Table 1). Clutch size values reported here for C. caretta, C. mydas, and D. coriacea in Florida and E. imbricata in Panama are comparable to those available in the literature (Miller 1997). Egg size is comparable between C. caretta and C. mydas (C. caretta: egg weight = 32.7 g, egg diameter = 40.9 mm; C. mydas: egg weight = 46.1 g, egg diameter = 44.9 mm; Miller 1997), while E. imbricata lays the smallest eggs (egg weight = 26.6 g, egg diameter = 37.8 mm, Miller 1997) and D. coriacea the largest (egg weight = 75.9 g, egg diameter = 53.4 mm, Miller 1997). C. caretta tends to lay on open ocean beaches, and the average depth of the bottom of the egg chamber is 49.2 cm (Tiwari and Bjorndal 2000). C. mydas prefers to lay closer to the dune or even in the dune system, and the average depth of the bottom of the egg chamber is 58.5 cm (Hays et al. 1993), while D. coriacea prefers nesting in open sand, and the average depth of the bottom of the egg chamber is 70 cm (Billes and Fretey 2001). E. imbricata nests preferentially in the vegetation and the average depth of the bottom of the egg chamber is 40 cm (Limpus et al. 1983). We hypothesize that the greater error in D. coriacea than in C. caretta and C. mydas may be due to a combination of the following characteristics of D. coriacea nests and nest inventories: (i) nest depth; (ii) long duration of incubation (although such an effect was not observed in this species, it might become more obvious with a larger sample size, as observed in C. caretta); (iii) D. coriacea hatchling size (large hatchlings with elongated flippers that may result in greater fragmentation of eggshells compared to hard-shell species, i.e., Cheloniidae) (iv) D. coriacea morphology (i.e., since the species is characterized by extreme reduction in the bones of the carapace and plastron and extensive cartilaginous structures (Pritchard 1997), dead in nest hatchlings might decompose faster than those of hard-shell species); (v) the higher level of decomposition found in D. coriacea nests than those of hard-shell species (M. Koperski pers. comm.), likely a result of (i–ii and iv) and this might mask effect (iii) because the hatchlings producing damage may decompose and may not be identified easily at inventory. All the above may increase the likelihood that project personnel conducting nest inventory will break eggs while pulling them out (i.e., it is more difficult to carefully pull out eggs from egg chambers that are difficult to reach and where there is more decomposition). High counting errors in E. imbricata nests may, in part, be due to biological attributes of the species such as large clutch size and small egg size, but we hypothesize that additional factors at the study site may have contributed, including (i) highly organic nesting substrate with extensive roots, (ii) tropical climate and high rainfall totals (3.5 m/yr in study years, Smithsonian Tropical Research Institute, Physical Monitoring Program, https://biogeodb.stri.si.edu/physical_monitoring/research/bocas) which may accelerate decomposition inside the nest, and (iii) undetected subterranean removal of eggs and hatchlings by two species of crabs (Gecarcinus sp. and Ocypode sp). In the study areas, the median error for C. caretta and C. mydas appears to be minimal, and only a little higher for D. coriacea. For C. caretta, the relationship between error and clutch size can provide correction factors to improve clutch size estimates. However, caution is needed for E. imbricata where a much larger error was observed.

Regarding sea turtle species not included in this study, if our interpretation of the biological traits that drive error is correct, we predict that the counting error for Lepidochelys olivacea (olive ridley turtle) and Lepidochelys kempii (Kemp’s ridley turtle) will be comparable to that observed for C. caretta and C. mydas because L. olivacea and L. kempii are hard-shell species and lay clutches of similar size and egg diameter (Van Buskirk and Crowder 1994; Miller 1997), although at shallower depths (Pritchard and Marquez 1973; Limpus et al. 1983). On the other hand, Natator depressus (flatback turtle) is a hard-shell species and lays the smallest clutches of all sea turtles and the clutch is shallower (Limpus 1971) than C. Caretta but its egg is large and comparable to those of D. coriacea in diameter (Limpus 1971; Van Buskirk and Crowder 1994; Miller 1997). Therefore, depending on which parameters are more important for the error, we could expect the error in N. depressus to be similar to that for C. caretta, D. coriacea or intermediate. Future studies should investigate and quantify counting errors in these species and in other populations of the four species studied here.

Fourth, the different error seen with different teams strongly suggests human error and the importance of adequately training personnel. Human error can include several error types, from the ability to excavate a nest without overlooking any of the contents or breaking the eggshells into smaller fragments, to correctly counting those representing > 50% of the eggshell, to correctly labeling the nest, to compiling field data sheets and databases. Human error is the most likely explanation for those cases where the estimated number of hatched eggs were more than the expected ones, suggesting either that an error was made in labeling or compilation or that fragments < 50% of the size of a whole egg were counted. Basic human error (e.g., in nest labeling) is a likely explanation for the high error (in both directions, over- and underestimation) observed for clutches relocated to hatcheries. In the hatchery dataset, we discounted the possibility that the underestimation error was due to undetected or unreported predation because the only hatchery included in the study was a fully enclosed hatchery on a beach in Florida. If a predator had entered the hatchery, most (if not all) nests would have been impacted and predation would not have gone unnoticed. We believe the greater counting errors observed in the hatchery dataset were made primarily because nests were placed near each other, which may have increased the likelihood of mislabeling or misidentifying nests. The same problem may have occurred also for E. imbricata because the study site was characterized by high nesting density in the wooded zone (Panama, Supplemental Fig. 2), in contrast to the other three species (Florida, Supplemental Fig. 3).

The number of days from emergence to nest inventory (V) was not a significant factor in error in this study but this may be an artifact, since most of the inventories were conducted 3 days after emergence (FWC 2016) and other researchers around the world may not wait the same amount of time after an emergence to conduct an inventory. Environmental factors (e.g., rainfall, drought) could also be involved.

In conclusion, although estimating clutch size at nest inventory is the least impactful approach available (e.g., it does not disturb nesting females or alter the incubation environment), results suggest that error associated with this method may be sufficient to warrant concern about using the data for certain purposes (e.g., establishing the reference clutch size for a population, conducting population assessments). For some species and study sites, determining an accurate clutch size may be particularly challenging, but biological differences between subpopulations of a species and different conditions at study sites could also affect error. If so, the present results may not necessarily apply to other areas or populations. We recommend that: (i) whenever possible, errors in counting hatched eggs should be estimated, adequately reported (e.g., median and IQR) and incorporated into studies estimating hatching and emergence success, productivity, and the associated estimated clutch size; (ii) data lacking such validation should be regarded with the necessary caution, especially concerning E. imbricata and to some extent D. coriacea; underestimation of clutch size should be assumed in these cases; (iii) this study should be repeated for other species and populations where relocation is undertaken for other reasons; (iv) for the areas included in the present study, results should be considered when attempting estimations of clutch size, hatching and emergence success, and hatchling production; and (v) when clutch size is reported in the literature, the source of the data should be given.

Data availability

The dataset analysed during the current study will be provided upon reasonable request submitted to the corresponding author.

Code availability

The code to assist in data analysis will be provided upon reasonable request submitted to the corresponding author.

References

  1. Anders AD, Marshall MR (2005) Increasing the accuracy of productivity and survival estimates in assessing landbird population status. Conserv Biol 19:66–74. https://doi.org/10.1111/j.1523-1739.2005.00543.x

    Article  Google Scholar 

  2. Balazs GH, Van Houtan KS, Hargrove SA, Brunson SM, Murakawa SKK (2015) A review of the demographic features of Hawaiian green turtles (Chelonia mydas). Chelonian Conserv Biol 14:119–129. https://doi.org/10.2744/ccb-1172.1

    Article  Google Scholar 

  3. Behera S, Tripathy B, Sivakumar K, Choudhury BC, Dutta SK, Pandav B (2018) Nesting space dynamics and its relationship with the arribada of olive ridley turtles at Gahirmatha rookery, east coast of India. J Herpetol 52:381–386

    Article  Google Scholar 

  4. Bell CD, Solomon JL, Blumenthal JM, Austin TJ, Ebanks-Petrie G, Broderick AC, Godley BJ (2007) Monitoring and conservation of critically reduced marine turtle nesting populations: lessons from the Cayman Islands. Anim Conserv 10:39–47. https://doi.org/10.1111/j.1469-1795.2006.00068.x

    Article  Google Scholar 

  5. Bevan E, Wibbels T, Najera BMZ, Martinez MAC, Martinez LAS, Reyes DJL, Hernandez MH, Gamez DG, Pena LJ, Burchfield PM (2014) In situ nest and hatchling survival at Rancho Nuevo, the primary nesting beach of the Kemp’s ridley sea turtle, Lepidochelys kempii. Herpetol Conserv Biol 9:563–577

    Google Scholar 

  6. Billes A, Fretey J (2001) Nest morphology in the leatherback turtle. Mar Turtle Newsl 92:7–9

    Google Scholar 

  7. Bouchard SS, Bjorndal KA (2000) Sea turtles as biological transporters of nutrients and energy from marine to terrestrial ecosystems. Ecology 81:2305–2313

    Article  Google Scholar 

  8. Broderick AC, Glen F, Godley BJ, Hays GC (2003) Variation in reproductive output of marine turtles. J Exp Mar Biol Ecol 288:95–109. https://doi.org/10.1016/s0022-0981(03)00003-0

    Article  Google Scholar 

  9. Broderick AC, Godley BJ (1999) Effect of tagging marine turtles on nesting behaviour and reproductive success. Anim Behav 58:587–591. https://doi.org/10.1006/anbe.1999.1183

    CAS  Article  PubMed  Google Scholar 

  10. Brost B, Witherington B, Meylan A, Leone E, Ehrhart L, Bagley D (2015) Sea turtle hatchling production from Florida (USA) beaches, 2002–2012, with recommendations for analyzing hatching success. Endang Species Res 27:53–68. https://doi.org/10.3354/esr00653

    Article  Google Scholar 

  11. Cardona L, Clusa M, Eder E, Demetropoulos A, Margaritoulis D, Rees AF, Hamza AA, Khalil M, Levy Y, Turkozan O, Marin I, Aguilar A (2014) Distribution patterns and foraging ground productivity determine clutch size in Mediterranean loggerhead turtles. Mar Ecol Prog Ser 497:229–241. https://doi.org/10.3354/meps10595

    Article  Google Scholar 

  12. Carrete M, Donázar JA, Margalida A (2006) Density-dependent productivity depression in Pyrenean bearded vultures: implications for conservation. Ecol Appl 16:1674–1682. https://doi.org/10.1890/1051-0761(2006)016[1674:Dpdipb]2.0.Co;2

    Article  PubMed  Google Scholar 

  13. Casale P, Heppell SS (2016) How much sea turtle bycatch is too much? A stationary age distribution model for simulating population abundance and potential biological removal in the Mediterranean. Endang Species Res 29:239–254

    Article  Google Scholar 

  14. Caut S, Guirlet E, Jouquet P, Girondot M (2006) Influence of nest location and yolkless eggs on the hatching success of leatherback turtle clutches in French Guiana. Can J Zool-Rev Can Zool 84:908–915. https://doi.org/10.1139/z06-063

    Article  Google Scholar 

  15. Ceriani SA, Casale P, Brost M, Leone EH, Witherington BE (2019) Conservation implications of sea turtle nesting trends: elusive recovery of a globally important loggerhead population. Ecosphere 10:e02936. https://doi.org/10.1002/ecs2.2936

    Article  Google Scholar 

  16. Crouse DT, Crowder LB, Caswell H (1987) A stage-based population model for loggerhead sea turtles and implications for conservation. Ecology 68:1412–1423

    Article  Google Scholar 

  17. Dethmers KEM, Baxter PWJ (2011) Extinction risk analysis of exploited green turtle stocks in the Indo-Pacific. Anim Conserv 14:140–150. https://doi.org/10.1111/j.1469-1795.2010.00404.x

    Article  Google Scholar 

  18. Eckert KL, Eckert SA (1990) Embryo mortality and hatch success in in situ and translocated leatherback sea turtle, Dermochelys coriacea, eggs. Biol Conserv 53:37–46

    Article  Google Scholar 

  19. Evans PGH, Hammond PS (2004) Monitoring cetaceans in European waters. Mammal Rev 34:131–156. https://doi.org/10.1046/j.0305-1838.2003.00027.x

    Article  Google Scholar 

  20. Florida Fish and Wildlife Conservation Commission (2016) Marine Turtle Conservation Handbook. http: //www.myfwc.com/wildlifehabitats/managed/sea-turtles/conservation-guidelines/. Accessed 15 October 2020

  21. Fowler LE (1979) Hatching success and nest predation in the green sea turtle, Chelonia mydas, at Tortuguero, Costa Rica. Ecology 60:946–955

    Article  Google Scholar 

  22. Garcia A, Ceballos G, Adaya R (2003) Intensive beach management as an improved sea turtle conservation strategy in Mexico. Biol Conserv 111:253–261

    Article  Google Scholar 

  23. Godley BJ, Broderick AC, Downie JR, Glen F, Houghton JDR, Kirkwood I, Reece S, Hays GC (2001) Thermal conditions in nests of loggerhead turtles: further evidence suggesting female skewed sex ratios of hatchling production in the Mediterranean. J Exp Mar Biol Ecol 263:45–63

    Article  Google Scholar 

  24. Hays GC, Adams CR, Speakman JR (1993) Reproductive investiment by green turtles nesting on Ascension Island. Can J Zool 71:1098–1103. https://doi.org/10.1139/z93-149

    Article  Google Scholar 

  25. Hincke MT, Nys Y, Gautron J, Mann K, Rodriguez-Navarro AB, McKee MD (2012) The eggshell: structure, composition and mineralization. Front Biosci 17:80

    Article  Google Scholar 

  26. Holmberg J, Norman B, Arzoumanian Z (2009) Estimating population size, structure, and residency time for whale sharks Rhincodon typus through collaborative photo-identification. Endang Species Res 7:39–53

    Article  Google Scholar 

  27. Kamel SJ, Mrosovsky N (2005) Repeatability of nesting preferences in the hawksbill sea turtle, Eretmochelys imbricata, and their fitness consequences. Anim Behav 70:819–828. https://doi.org/10.1016/j.anbehav.2005.01.006

    Article  Google Scholar 

  28. Kornaraki E, Matossian DA, Mazaris AD, Matsinos YG, Margaritoulis D (2006) Effectiveness of different conservation measures for loggerhead sea turtle (Caretta caretta) nests at Zakynthos Island, Greece. Biol Conserv 130:324–330. https://doi.org/10.1016/j.biocon.2005.12.027

    Article  Google Scholar 

  29. Limpus CJ (1971) The flatback turtle, Chelonia depressa Garman, in southeast Queensland, Australia. Herpetologica 27:431–446

    Google Scholar 

  30. Limpus CJ, Baker V, Miller JD (1979) Movement induced mortality of loggerhead eggs. Herpetologica 35:335–338

    Google Scholar 

  31. Limpus CJ, Parmenter CJ, Baker V, Fleay A (1983) The Crab Island sea turtle rookery in the north-eastern Gulf of Carpentaria. Aust Wildl Res 10:173–184

    Article  Google Scholar 

  32. Marco A, da Graça J, García-Cerdá R, Abella E, Freitas R (2015) Patterns and intensity of ghost crab predation on the nests of an important endangered loggerhead turtle population. J Exp Mar Biol Ecol 468:74–82. https://doi.org/10.1016/j.jembe.2015.03.010

    Article  Google Scholar 

  33. Marcovaldi MA, Chaloupka M (2007) Conservation status of the loggerhead sea turtle in Brazil: an encouraging outlook. Endang Species Res 3:132–143

    Article  Google Scholar 

  34. Mazaris AD, Kramer-Schadt S, Tzanopoulos J, Johst K, Matsinos G, Pantis JD (2009) Assessing the relative importance of conservation measures applied on sea turtles: comparison of measures focusing on nesting success and hatching recruitment success. Amphib Reptil 30:221–231

    Article  Google Scholar 

  35. Meylan AB, Meylan PA, Espinosa CO (2013) Sea turtles of Bocas del Toro Province and the Comarca Ngöbe- Buglé, Republic of Panamá. Chelonian Conserv Biol 12:17–33. https://doi.org/10.2744/ccb-0948.1

    Article  Google Scholar 

  36. Mikhailov KE (1997) Fossil and recent eggshell in amniotic vertebrates: fine structure, comparative morphology and classification. Spec Pap Palaeontol 56:1–80

    Google Scholar 

  37. Miller JD (1997) Reproduction in sea turtles. In: Lutz PL, Musick JA (eds) The biology of sea turtle. CRC Marine Science Series. CRC Press Inc, Boca Raton, pp 51–81

    Google Scholar 

  38. Miller JD (1999) Determining clutch size and hatching success. In: Eckert KL, Bjorndal KA, Abreu-Grobois FA, Donnelly M (eds) Research and Management Techniques for the Conservation of Sea Turtles. IUCN/SSC Marine Turtle Specialist Group Publication No. 4, pp 124–129

  39. Miller JD, Limpus CJ, Godfrey MH (2003) Nest Site Selection, Oviposition, Eggs, Development, Hatching, and Emergence of Loggerhead Turtles. In: Bolten AB, Witherington BE (eds) Loggerhead sea turtles. Smithsonian Books, Washington, DC, pp 125–143

    Google Scholar 

  40. Mortimer JA (1999) Reducing threats to eggs and hatchlings: hatcheries. In: Eckert KL, Bjorndal KA, Abreu-Grobois FA, Donnelly M (eds) Research and Management Techniques for the Conservation of Sea Turtles IUCN/SSC Marine Turtle Specialist Group Publication No 4, pp 175–178

  41. Niethammer KR, Balazs GH, Hatfield JS, Nakai GL, Megyesi JL (1997) Reproductive biology of the green turtle (Chelonia mydas) at Tern Island, French Frigate Shoals. Hawaii Pac Sci 51:36–47

    Google Scholar 

  42. Olendo MI, Okemwa GM, Munga CN, Mulupi LK, Mwasi LD, Mohamed HB, Sibanda M, Ong’anda HO (2019) The value of long-term, community-based monitoring of marine turtle nesting: a study in the Lamu Archipelago, Kenya. Oryx 53:71–80. https://doi.org/10.1017/S0030605317000771

    Article  Google Scholar 

  43. Pendoley KL, Bell CD, McCracken R, Ball KR, Sherborne J, Oates JE, Becker P, Vitenbergs A, Whittock PA (2014) Reproductive biology of the flatback turtle Natator depressus in Western Australia. Endang Species Res 23:115–123

    Article  Google Scholar 

  44. Pérez-Castañeda R, Salum-Fares A, Defeo O (2007) Reproductive patterns of the hawksbill turtle Eretmochelys imbricata in sandy beaches of the Yucatan Peninsula. J Mar Biol Assoc UK 87:815–824

    Article  Google Scholar 

  45. Piacenza SE, Richards PM, Heppell SS (2017) An agent-based model to evaluate recovery times and monitoring strategies to increase accuracy of sea turtle population assessments. Ecol Model 358:25–39. https://doi.org/10.1016/j.ecolmodel.2017.05.013

    Article  Google Scholar 

  46. Pintus KJ, Godley BJ, McGowan A, Broderick AC (2009) Impact of clutch relocation on green turtle offspring. J Wildl Manage 73:1151–1157. https://doi.org/10.2193/2008-103

    Article  Google Scholar 

  47. Pritchard PCH (1997) Evolution, phylogeny, and current status. In: Lutz PL, Musick JA (eds) The biology of sea turtles. CRC Marine Science Series. CRC Press Inc, Boca Raton, pp 1–28

    Google Scholar 

  48. Pritchard PCH, Marquez MR (1973) Kemp’s Ridley Turtle or Atlantic Ridley. Lepidochelys kempi, IUCN Monograph No, p 2

    Google Scholar 

  49. R Development Core Team (2020) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna

    Google Scholar 

  50. Rees AF, Alfaro-Shigueto J, Barata PCR, Bjorndal KA, Bolten AB, Bourjea J, Broderick AC, Campbell LM, Cardona L, Carreras C, Casale P, Ceriani SA, Dutton PH, Eguchi T, Formia A, Fuentes MMPB, Fuller WJ, Girondot M, Godfrey MH, Hamann M, Hart KM, Hays GC, Hochscheid S, Kaska Y, Jensen MP, Mangel JC, Mortimer JA, Naro-Maciel E, Ng C, Nichols WJ, Phillott AD, Reina RD, Revuelta O, Schofield G, Seminoff JA, Shanker K, Tomás J, van de Merwe JP, Van Houtan KS, Vander Zanden HB, Wallace BP, Wedemeyer-Strombel KR, Work TM, Godley BJ (2016) Are we working towards global research priorities for management and conservation of sea turtles? Endang Species Res 31:337–382. https://doi.org/10.3354/esr00801

    Article  Google Scholar 

  51. Tiwari M, Bjorndal KA (2000) Variation in morphology and reproduction in loggerheads, Caretta caretta, nesting in the United States, Brazil, and Greece. Herpetologica 56:343–356

    Google Scholar 

  52. Tiwari M, Bjorndal KA, Bolten AB, Bolker BM (2006) Evaluation of density-dependent processes and green turtle Chelonia mydas hatchling production at Tortuguero, Costa Rica. Mar Ecol Prog Ser 326:283–293

    Article  Google Scholar 

  53. Tomillo PS, Suss JS, Wallace BP, Magrini KD, Blanco G, Paladino FV, Spotila JR (2009) Influence of emergence success on the annual reproductive output of leatherback turtles. Mar Biol 156:2021–2031. https://doi.org/10.1007/s00227-009-1234-x

    Article  Google Scholar 

  54. Tucker AD, Frazer NB (1991) Reproductive variation in leatherback turtles, Dermochelys coriacea, at Culebra National Wildlife Refuge, Puerto Rico. Herpetologica 47:115–124

    Google Scholar 

  55. Turkozan O, Yamamoto K, Yilmaz C (2011) Nest site preference and hatching success of green (Chelonia mydas) and loggerhead (Caretta caretta) sea turtles at Akyatan Beach, Turkey. Chelonian Conserv Biol 10:270–275

    Article  Google Scholar 

  56. Van Buskirk J, Crowder LB (1994) Life-history variation in marine turtles. Copeia 1:66–81

    Article  Google Scholar 

  57. Weaver JL, Paquet PC, Ruggiero LF (1996) Resilience and Conservation of Large Carnivores in the Rocky Mountains. Conserv Biol 10:964–976. https://doi.org/10.1046/j.1523-1739.1996.10040964.x

    Article  Google Scholar 

  58. Whitmore CP, Dutton PH (1985) Infertility, embryonic mortality and nest-site selection in leatherback and green sea turtles in Suriname. Biol Conserv 34:251–272

    Article  Google Scholar 

Download references

Acknowledgments

We thank the coordinated network of Florida sea turtle permit holders, who collected the Florida data analyzed here, for their dedication and effort on behalf of sea turtle conservation. In Panama, thanks are due to many local community members, especially Inocencio Castillo, Arcelio González Hooker, and Luis Baker, who have worked many years on the project. We thank Cristina Ordoñez (Sea Turtle Conservancy) for logistic support and help with data management, and the Smithsonian Tropical Research Institute for assistance of many kinds. We thank J. Miller, C. Limpus, L. Ehrhart and J. Mortimer for providing historical context regarding the method used to conduct nest evaluation. We also thank M. Koperski and S. Pessolano for their insight into D. coriacea nest excavations. We thank A. Foley, L. Soares, B. Crowder and two anonymous reviewers for their constructive suggestions to improve the manuscript.

Funding

Funding for the Florida sea turtle nest counts program has come from the U.S. Fish and Wildlife Service and from the Marine Resources Conservation Trust Fund (thanks to the Florida Sea Turtle License Plate program, https://helpingseaturtles.org/get-a-plate/). Funding for the work in the Zapatilla Cays has come primarily from a series of grants to A. and P. Meylan from the Marisla Foundation.

Author information

Affiliations

Authors

Contributions

SAC conceived the study. BB collated the dataset for Florida. PC conducted the statistical analysis. ABM and PAM contributed the dataset from Panama. SAC and PC led the writing of the manuscript with contributions from ABM and PAM.

Corresponding author

Correspondence to Simona A. Ceriani.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

All applicable international, national and institutional guidelines for handling sea turtle eggs have been followed and all necessary approvals have been obtained. The research conducted for this project was undertaken with the authority and under the supervision of SA Ceriani and AB Meylan as Sea Turtle Nesting Program Coordinators for the Florida Fish and Wildlife Conservation Commission, the institution responsible for regulating sea turtle monitoring and conservation and issuing permits in Florida. Research activity and data collection in Panama were authorized under a series of permits from Autoridad Nacional del Ambiente and Ministerio de Ambiente, the most recent (2016) being SE/A-54–16.

Informed Consent

Consent for participation is not applicable to this study as there were no human test subjects.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Reviewed by Y. Kaska and M. Ware.

Responsible Editor: L. Avens.

Supplementary Information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ceriani, S.A., Brost, B., Meylan, A.B. et al. Bias in sea turtle productivity estimates: error and factors involved. Mar Biol 168, 41 (2021). https://doi.org/10.1007/s00227-021-03843-w

Download citation