This study was conducted at the Centro de Protección de la Tortuga Marina in Boca Seca beach, located in Lázaro Cárdenas, Michoacán, México (18° 04′ N, 102° 58′ W; Figure 1). Egg manipulation was kept to the bare minimum and done by the hatchery staff according to protocols stated in Mexican regulation (NOM-162-SEMARNAT, 2014) and a previous report (Herrera-Vargas et al., 2017). Briefly, beach patrolling during three consecutive nights (September 13–15, 2017) allowed identification of nesting females. Ten nests located approximately in the same beach zone (30–60 m away from the shoreline) were chosen and sheltered immediately after the female turtle covered the eggs and left the site. Five randomly selected natural nests remained undisturbed in situ, only fenced with cyclone mesh until hatchling emergence. The complete clutch of the other five nests was carefully collected as soon as the female left the nest, placed in individual plastic bags and transported to the hatchery. There, the eggs were immediately buried in previously built nests and remained undisturbed until emergence. This ensured that conditions related to clutch size (e.g., oxygen availability, temperature, metabolic heat, etc.) remained unaltered. The total time between laying and re-burial lasted less than 2 h. Efforts were made to avoid egg rotation and excessive handling, as well as to emulate natural nest architecture in hatcheries. Ex situ nests were constructed by the hatchery staff according to international norms for L. olivacea, with a narrow neck (20–25 cm) and a wider flask-shaped bottom, at a depth of 40–50 cm and 1 m separation between them (Kutzari, 2006). Nests from both conditions were not shaded or watered. Forty days after incubation started, the hatchery clutches were fenced with cyclone mesh until turtle emergence. This experimental design ensured that in situ nests were not disturbed and that clutches relocated to hatcheries only went through the routine procedures done by the hatchery staff. Egg handling was performed before organogenesis started (Miller, 1985).

Animal sampling, handling and sacrifice protocols were approved by an Animal Rights Committee, under License Number SEMARNAT: SGPA/DGVS/10395/17; in accordance with Mexican regulation (NOM-033-SAG/ZOO, 2014). One hundred and fifty turtles were collected from five in situ and five ex situ nests (15 hatchlings/nest). Fifty hatchlings were used for histological and morphometric observations: for neurogenesis studies, 24 brains per nest type were evaluated, since one brain for each condition was damaged during dissection (48 brains in total); for ovarian cell proliferation quantification, at least two ovaries per nest were used (27 ovaries in total), while all fifty individuals were sexed. The other one hundred hatchlings were used to evaluate motor performance.

Hatchling collection was described by Herrera-Vargas et al. (2017) and Robledo-Avila et al. (2022). Briefly, fifteen emerging turtles from each nest were collected at 5-min intervals, as soon as they surfaced from each nest. Five randomly selected hatchlings per nest were weighted with a digital precision balance (OHAUS™ Scout Pro Sp 602, Max 600 g, d = 0.01 g). Their straight carapace length was measured using a digital Vernier caliper (Mitutoyo™). These same five turtles per nest were used to investigate neurogenesis and gonadal cell proliferation.

To evaluate the effect of ex situ incubation on brain and ovary cell proliferation, hatchlings received an intraperitoneal injection of the cell birth marker 5′-bromo-2′-deoxyuridine (BrdU, a thymidine analog incorporated in the S phase of the cell cycle. Sigma-Aldrich, 100 mg/Kg in 0.9% NaCl) immediately after morphometric data recording and a second injection 2 h after. Turtles were then placed in sand tubs and sacrificed 4 h after the last injection. This procedure (i.e., the timings) minimized the effect of factors other than the incubation condition on cell proliferation. The brain and gonad-mesonephros complex were dissected in situ, incubated in Bouin’s solution (Sigma-Aldrich, Saint Louis, MO, United States) for 24 h and incubated in buffered paraformaldehyde (Sigma-Aldrich, Saint Louis, MO, United States, 4% in 0.1 M phosphate buffer) for 3 days at room temperature.

In the laboratory, brains were rinsed with 70% ethanol and transferred to buffered sucrose (Sigma-Aldrich, Saint Louis, MO, United States, 30% in 0.1 M phosphate buffer) at 4°C until they sank. Then they were frozen in the Peltier module of a cryostat (Microm) and sectioned coronally at 30 μm. Free-floating sections were collected in Tris-buffered saline (Sigma-Aldrich, Saint Louis, MO, United States, 50 mM Tris–HCl, 150 mM NaCl, pH 7.6) and processed for immunohistochemistry. Cell proliferation in neurogenic niches was evaluated by immunoreactivity for BrdU and neuronal integration was evaluated by immunoreactivity for the anti-neuronal nuclear protein (NeuN) in separate brain sections. Briefly, the tissue was incubated in ImmunoDNA retriever 20× with citrate (Bio SB), then in 2 N HCl for 30 min at 65°C and finally in 0.1 M sodium borate buffer at room temperature. Blocking of non-specific binding was done by incubating the sections in 0.1% bovine serum albumin (Sigma-Aldrich, Saint Louis, MO, United States) for 30 min. Sections including the dorsal and medial ventricular zones were incubated with mouse anti-BrdU monoclonal antibody (1: 500, Roche) and independent cortical sections were incubated with mouse anti-NeuN monoclonal antibody (1: 1000, Millipore) for 16 h at 4°C. Then they were incubated with a donkey anti-mouse biotin-conjugated secondary antibody (1:500, Vector Laboratories) for 2 h. Sections were incubated in avidin/biotin horseradish peroxidase (Vectastain Elite, PK-6100) for 2 h at room temperature and then incubated in diaminobenzidine as a chromogen, with peroxide and buffer for 10 min (Vector Staining Kit, SK-4100). Finally, the tissue was mounted onto gelatinized slides, dehydrated and cover-slipped using Cytoseal 60 (Richard Alan Scientific).

To evaluate cell proliferation in neurogenic niches, three equivalent non-adjacent brain sections containing the dorsal and medial ventricular zones were selected per turtle according morphological criteria (appearance of lateral ventricles) along the antero-posterior axis. BrdU + immunoreactive cells were quantified in two microphotographs per section per neurogenic zone at 1000× magnification. To evaluate neuronal integration, three equivalent non-adjacent brain sections including the dorsomedial and medial cortices were selected per turtle according morphological criteria (opening of the lateral ventricles). NeuN + immunoreactive cells in the cellular layer were quantified in three microphotographs per section per cortex at 1000× magnification. Microphotographs were captured with a Zeiss microscope using the Axio Vision 4.6 software and analyzed using NIH ImageJ software.

For gonadal sex determination, one gonad-mesonephros complex was dehydrated using increasing ethanol concentrations, embedded in paraffin, sliced transversally (7 μm) in a microtome (Leica) and stained with hematoxylin-eosin (Merck), as previously described (Herrera-Vargas et al., 2017). Gonadal histology showed that all fifty specimens were females, thus all gonadal analysis were performed in ovaries. To evaluate ovarian cell proliferation, at least two gonad-mesonephros complexes per nest were frozen, cryosectioned at 30 μm and immuno-stained for BrdU as described for the brain (the other gonad-mesonephros complexes were frozen for RNA quantification). Five adjacent gonadal sections per hatchling were selected from the central ovary. BrdU + immunoreactive germ cells were quantified in the cortex of ovaries throughout the whole section at 400 x magnification, as described for brain sections. After quantification, selected ovarian sections were stained with hematoxylin-eosin to observe cell density and cytoarchitecture.

To evaluate the effect of ex situ incubation on motor skills, ten hatchlings per nest were randomly selected and separated in tubs with sand for 15 min, to prevent lethargy from interfering with performance (Booth et al., 2013). Turtles were placed upside-down on a tray full of dry sand and the time they took to self-right was recorded with a chronometer (Sper scientific 810015 5 channel timer). Hatchlings that exceeded 10 min to self-right were discarded from the analysis. Thus, 48 turtles from in situ nests and 31 ex situ hatchlings were analyzed.

To evaluate the contribution of abiotic variables to the developmental traits, nest temperature, moisture and sand grain size were determined. Nest temperature was recorded from developmental day 11 until emergence, since this period includes the bulk of hippocampal neurogenesis previously described for Emys orbicularis (Goffinet et al., 1986) and the critical time for gonadal development in L. olivacea (Merchant-Larios et al., 1997). Temperature was registered by data loggers (Onset HOBO^® Bluetooth Pendant MX2202 series; accuracy ± 0.2°C) carefully located outside the nest to avoid disturbing the clutch. They were placed in sand 30 cm from the center of the nest and 50 cm deep, 11 days after the incubation period began. They were programmed to record the temperature every hour; results were averaged by nest.

Moisture and grain size were determined from 100 g of sand, sampled 10 cm deep inside the nests, immediately after hatchling emergence. The sand was placed in a sealed plastic bag, weighed, dried at 105–110°C in a standard oven and weighed again. Moisture content was calculated as the ratio of wet to dry sand mass (Head, 1992). Grain size analysis was performed by particle sieving, using international parameters (gravel: >2.0; coarse sand: 2.0 ± 0.2; fine sand: 0.2 ± 0.02; silt: 0.02–0.002 mm; Brady and Weil, 2008), and subsequent weighting with an analytical balance. Gravel, coarse- and fine-sand, as well as silt proportions were calculated dividing by the total dry mass (Gee and Or, 2002).

Preliminary analysis showed that only one nest had a different gravel composition from the rest, thus gravel was discarded from further examination. Similarly, fine sand was collinear with coarse sand, hence only the latter was used. This was done because multivariate analyses are sensitive to collinearity between variables, which causes interpretation problems (Harrison et al., 2018). Accordingly, the selected parameters for analysis were abiotic variables within nests (temperature and moisture, coarse sand and silt), as well as turtle biological variables (cell proliferation in the dorsal and medial ventricular zones, as well as the ovary; neuronal integration in the cellular layer of the dorsomedial and medial cortices; body mass and length, as well as self-righting time).

Multivariate principal components analysis (PCA) was performed to reduce data dimensionality and investigate the distribution of samples in two-dimensions. This allowed the assessment of possible differences between conditions, based on the abiotic variables within nests and turtle biological data.

The outcomes of in situ (n = 5) vs. ex situ clutches (n = 5), abiotic variables and their interaction (in situ vs. ex situ condition interacting with each abiotic variable) were studied with linear mixed models to avoid violation of independence assumptions (turtles within clutches). Biological results were used as the response variables in these models; which included the following effects: in situ vs. ex situ clutches (main); abiotic variables and their interaction (fixed) plus turtles within clutches (random). All abiotic variables (temperature, moisture, coarse sand and silt) were standardized by subtracting the mean from every value and dividing by the standard deviation (Harrison et al., 2018), due to wide differences in their ranges. Outliers were removed from most turtle biological data (all except body mass and length). Ovarian cell proliferation and self-righting were also transformed to meet normality and homocedasticity assumptions; with square-root and natural logarithm, respectively.

Stepwise backward elimination followed by selection with the conditional Akaike information criterion (cAIC) were performed to obtain the best linear mixed model (Supplementary Data 1). The marginal R² for each model was determined as a measure of the proportion of the variance explained by the model. The effect size for in situ vs. ex situ clutches was evaluated by Glass’s Δ (Sink and Mvududu, 2010; Sullivan and Feinn, 2012). Graphs presented in the results for each of the eight biological variables were obtained back-transforming the predictions of the final models. Residuals for each model were plotted to assess the distribution of the model fit (Supplementary Data 1).

Data analyses were done using R (R Core Team, 2020; version 4.1.1) and RStudio (RStudio Team, 2020; version 1.3.1073). Linear mixed models were performed using: readxl (Wickham and Bryan, 2019), lme4 (Bates et al., 2015), lmerTest (Kuznetsova et al., 2017), and cAIC (Säefken et al., 2018). Model parameters, including residuals, were evaluated with the performance package (Lüdecke et al., 2021). Principal component analysis results were graphed with ggbiplot (Vu, 2011).

To evaluate the effect of ex situ incubation while minimizing the effect of turtle retention, two important events for early neurogenesis were evaluated 6 h after nest emergence: cell proliferation in neurogenic niches and neuron integration in cortices homologous to the mammalian hippocampus. The results showed that the ex situ incubation strategy is associated with a lower number of proliferating cells and differentiated neurons in turtle hatchlings. In vertebrates, hippocampal neurogenesis is a highly regulated process that includes cell proliferation, migration, differentiation and integration (Altman and Das, 1965; Gage, 2002; Bayer and Altman, 2004; Kuhn et al., 2016; McDonald and Vickaryous, 2018). Proper development of the mammalian hippocampus is required to achieve ecologically relevant cognitive tasks such as spatial learning and memory, as well as to regulate emotional responses (Gould et al., 1999; Deng et al., 2010).

In reptiles, newly born cells from the dorsal and lateral ventricular zones migrate radially for several days and mature as neurons as they approach the cortices. Neurogenic niches in the postnatal gecko brain produce neurons exclusively (McDonald and Vickaryous, 2018). Thus, it is highly likely that the BrdU + cells we observed herein were neuronal progenitors. Nonetheless, future cell-fate mapping studies should confirm the lineage of BrdU + cells produced in newly emerged sea turtles. In this study, BrdU + cells were only present lining the ventricular walls in both conditions, confirming the idea that proliferating neurons migrate several days after birth.

Offspring from ex situ clutches showed fewer BrdU + proliferating cells early after nest emergence, suggesting that a lower number of neural precursors will migrate to hippocampal cortices postnatally. Thus, less neurons (or glial cells) differentiate and integrate into functional circuits. Similarly, the lower density of NeuN + cells observed in the medial and dorsomedial cortices of hatchlings from ex situ clutches could reflect deficient prenatal cell proliferation or precursor migration. The difference between in situ and ex situ conditions for NeuN + cells was higher in the medial cortex, suggesting differential cortical sensitivity to early non-optimal conditions, as observed in mammals (Alkadhi, 2019). These findings, together with our prior work in male hatchlings (Herrera-Vargas et al., 2017), suggest that ex situ incubation alters neurogenic events during critical prenatal and early postnatal windows. The functional relevance of our findings for sea turtle cognitive and behavioral performance in the short and long-term are still unknown. However, recent evidence in lizards supports prior studies in mammals and birds, showing that a disturbance in neurogenesis during development could impair spatial memory and migration either during early life or adulthood (Amiel et al., 2017; Dayananda and Webb, 2017). Interestingly, these studies have related non-optimal incubation temperatures with decreased hippocampal neurogenesis in lizards (Amiel et al., 2017; Dayananda and Webb, 2017). Herein, nest temperature, moisture, and substrate composition differentiate in situ vs. ex situ nests; however, neither isolated abiotic variable was directly associated to altered neurogenesis in hatchling turtles. Follow-up studies should experimentally assess the effect of either variable separately or as a whole on hippocampal cell proliferation and neuronal integration early during ontogenesis in sea turtles.

Brain development begins at stage III (incubation day 4) in Caretta caretta turtles (García-Cerdá and López-Jurado, 2009). Thus, early relocation to hatcheries is not likely to account for the observed developmental alterations. However, future studies should be done to rule-out this possibility.

The ex situ incubation strategy was associated with a dramatic reduction of ovarian cell proliferation in turtle hatchlings at nest emergence, evidenced by few BrdU + cells. In L. olivacea, gonadal development starts in the middle third of incubation (day 16 of development), when primordial germ cells accumulate at the base of gonadal ridge (Merchant-Larios et al., 1997). In this species, warm temperatures (32–33°C) promote ovary formation by stimulating the production of estradiol and aromatase from the undifferentiated gonad and adjacent tissues: the mesonephros and inter-renal glands (Wibbels et al., 1991, 1993; Ewert et al., 2004; Freedberg et al., 2006; Díaz-Hernández et al., 2015, 2017). The highest peak of ovarian cell proliferation is not known for L. olivacea hatchlings. However, folliculogenesis and oocyte entry into meiosis occur in the 3rd–5th post-hatching months (Merchant-Larios et al., 1989), suggesting that this peak occurs perinatally. The consequences of reduced germ cell proliferation in female juvenile or adult turtles have not been evaluated. In rodents, germ cell absence does not allow ovarian follicle development (McLaren et al., 1984) or results in follicle degeneration (Ray and Potu, 2010; Bishop et al., 2019). In mammals, reduced cell proliferation may result in delayed ovarian formation or even complete infertility (Monniaux, 2018).

In this study, ovarian weight was not formally evaluated, since each was dissected together with the mesonephros. Interestingly, qualitative histological observations of ovarian sections did not obviously evidence a reduction in cell density. Herein, the main variable explaining the diminished ovarian cell proliferation was the ex situ condition, which probably included the effect of incubation temperature plus moisture and substrate composition. Future studies are needed to identify the effects of ex situ incubation on ovarian cell density, evaluate the contribution of each abiotic variable on the ovarian phenotype and elucidate the long-term effects of poorly developed ovaries.

Hatchlings from ex situ clutches showed a lower body mass and straight carapace length than those from in situ nests. Linear mixed models for body mass and length suggested that the nest condition (in situ vs. ex situ) was an important factor influencing them. A direct contribution of isolated abiotic variables could not be identified. However, it is known that temperature plays a chief role determining reptile body size (Stewart et al., 2019). Although sand temperatures were above the threshold to promote female differentiation in both conditions, they likely differentially affected hatchling body size, as previously suggested (Robledo-Avila et al., 2022). Mean sand temperatures registered for in situ clutches were approximately 1°C below those in ex situ clutches, whereas mean maximum temperatures registered for both conditions showed broader ranges (35.68°C ± 0.83 SD for in situ nests and 37.03°C ± 1.29 SD for ex situ nests). Accordingly, the incubation duration was shorter by one day for ex situ clutches.

The mechanisms that may account for a differential temperature effect on the growth rate include a direct action on cell, tissue, or organ differentiation, as well as long-term neuroendocrine changes possibly via epigenetic alterations (Singh et al., 2020). Additionally, it is recognized that moisture also plays a role on body size, although the mechanisms are less well known. Modifications to gas exchange could explain the effects of moisture on development (Wallace et al., 2004). Other variables, such as nesting female size (Chatting et al., 2018), egg mass (Wallace et al., 2006), metabolic expenditure (Rusli et al., 2016; Gammon et al., 2020), or yolk absorption (Stand, 2002) also may contribute to determining body size in reptiles. A study showed an interaction between the maternal component, sand temperature, moisture, and body length in the loggerhead sea turtles. Moreover, it described differential effects of moisture on body length throughout development (i.e., a more prominent role of nest moisture on body mass during the last third of development; Tezak et al., 2020). Thus, the combined contribution of several abiotic and biotic variables could explain our results, as discussed below.

A larger body size has been consistently related with better motor performance and thus with better survival chances (Fleming et al., 2020; Martins et al., 2020). In this study, principal component 1 showed a negative relationship between body mass and length with self-righting ability, supporting prior observations. Moreover, self-righting has been associated with incubatory conditions like temperature and substrate composition, therefore constituting a good indicator of microenvironmental conditions (Stewart et al., 2019; Reboul et al., 2021). Herein, nest silt and temperature were related to increased turning time of hatchlings, in accordance with prior studies (Stewart et al., 2019; Reboul et al., 2021).

Several nest-related abiotic variables were measured to determine their potential contribution to the observed developmental effects in L. olivacea hatchlings. The results showed that in situ clutch conditions were related to a better turtle phenotype (i.e., greater body size, higher neurogenesis, increased ovariogenesis, and lesser time to self-righting). In situ nests showed lower temperatures, were located higher on the beach and in coarse sand with lower silt levels, while ex situ clutches showed the opposite. Accordingly, sand composition, temperature, and moisture were important for differences between in situ and ex situ clutches. However, a differential contribution of each abiotic variable, in isolation, to the developmental traits could not be identified. One possible explanation is that emergent properties of microenvironment-associated abiotic variables affect the turtle hatchling development. Recent reports support this idea (Tezak et al., 2020; Tanabe et al., 2021).

Interestingly, although in situ nests showed lower temperatures than ex situ nests, the average temperatures for the incubation period, as well the mean maximum temperatures for both conditions were above the thermal tolerance reported for L. olivacea embryos (Valverde et al., 2010; Maulany et al., 2012). Temperature is a chief parameter that determines brain, gonadal and motor system development (Reece et al., 2002; Amiel and Shine, 2012; Paredes et al., 2016; Amiel et al., 2017; Fleming et al., 2020), as well embryo survival inside the nest (Robledo-Avila et al., 2022). Thus, it likely also plays a major role in explaining the observed phenotypes. However, its direct contribution could not be determined.

Ex situ clutches showed a higher silt proportion, which was negatively associated with hatchling development. Previous studies have shown that successful embryo development occurs in sandy substrates (grain diameter: 0.063–2 mm), whereas substrates with a high silt content (grain size < 0.063) cause mortality and diminished egg weight, reduced hatchling mass and size, as well as lower fitness (Sarmiento-Ramírez et al., 2014; Marco et al., 2017). The proportion of silt observed in both incubation conditions was below the value described as detrimental for embryo development (0.02; Abella-Pérez, 2011). Thus, although its effects on neurogenesis, ovariogenesis, body size, and motor abilities should not be ruled out, silt is unlikely to completely explain the observed phenotypes. The proportion of silt may affect moisture, gas exchange and/or microbial load (Marco et al., 2017).

Alternatively, the scarce contribution of sediment size, temperature, and moisture to completely explain the observed phenotypes could result from procedural details. Sand temperature was recorded before the thermosensitive period began (developmental day 11), until turtle emergence to avoid disturbing egg development. This interval includes the peak of hippocampal neurogenesis and the critical window for sex determination. Surface sand (10 cm deep) was sampled for moisture and particle size measurements right after turtles emerged. Future studies should record the temperature inside the nest during the entire incubation period and collect sand surrounding the eggs. This should verify the microenvironmental contribution to the effects of clutch relocation on development of sea turtle hatchlings. Moreover, future experiments should measure temperature inside the nest to consider the effect of metabolic heating on the observed phenotypes.

It must be noted that the ex situ incubation strategy, per se likely did not directly affect hatchling development. However, the combined effect of the incubation microenvironment was probably directly to blame for the phenotypic effects. This idea is supported by our prior study on the action of ex situ nests on the configuration of the immune system (Robledo-Avila et al., 2022).

Besides the measured abiotic factors, other variables such as micro-biological parameters (Patino-Martínez et al., 2012) may explain the developmental changes. Recent split-clutch designs have highlighted the maternal contribution to hatchling body size and self-righting response (Kobayashi et al., 2020; Tezak et al., 2020). This study did not consider the maternal component to privilege normal conditions (i.e., leaving natural nests undisturbed). However, parental origin, egg movement, and reburial should be studied to identify the factors that determine the altered phenotypes. These studies will result in recommendations for hatchery management to maximize the developmental potential of sea turtles.