Introduction

Marine animals exhibit diverse reproductive strategies to adapt to their environment and maximize the fitness of their offspring1,2,3. These strategies are particularly important in the face of contemporary climate change4,5,6. Evidence suggests that climatic variables are altering the ecological and phenological aspects of marine species7,8,9,10. For instance, elevated sea temperatures can disrupt the reproductive activity in fish11,12, while increased storm frequency not only destroy nesting sites of sea turtles13 but also influence the movement behavior of fish and dolphin14,15. Ocean acidification can also affect the metabolism and metamorphosis of coral larvae, reducing their survival rates and dispersal potentials16. These impacts not only affect the reproductive activity and success of marine animals but also threaten their population persistence. Therefore, understanding reproductive ecology is crucial for the conservation of marine species, especially those facing significant environmental pressures.

Among oviparous marine animals, sea turtles exhibit high fecundity coupled with high offspring mortality17,18,19. The offspring fitness of animals in the order Testudines can be evaluated by hatchling morphology. Some evidence supports the “bigger is better” hypothesis, suggesting that larger hatchling size confers post-hatching advantages20,21. Additionally, fitness can be anticipated by examining residual yolks, which are considered an energy source that fuels locomotor behaviors at the beginning of the post-hatching period22,23. However, reproductive strategies among sea turtles are complex and involve trade-offs between egg size and clutch size, which can vary among different species24,25. For example, both leatherback (Dermochelys coriacea) and loggerhead sea turtles (Caretta caretta) follow a strategy where larger females lay larger clutch sizes but not necessarily bigger eggs26. In contrast, olive ridley turtles (Lepidochelys olivacea) tend to lay smaller eggs but in larger clutches27. Additionally, there are notable conspecific variations among different locations. For instance, the egg mass of loggerhead sea turtles varies significantly across populations in Florida, Brazil, and Greece, with eggs being larger in Florida and smaller in Greece28. These regional differences in reproductive strategies underscore the importance of understanding local variations to effectively manage and conserve sea turtle populations.

The morphological characteristics of neonatal sea turtles, which are critical determinants of their survival, are influenced by a variety of factors, including maternal and environmental factors29,30,31. Maternal traits, such as body size and health, combined with the nutrient availability from their foraging habitats, play a pivotal role in shaping hatchling morphology32. For instance, larger female green sea turtles (Chelonia mydas) produce larger eggs, resulting in larger hatchlings33,34. Similar patterns are observed in loggerhead sea turtles, where maternal size correlates with hatchling size35. Additionally, the egg components are also influenced by maternal effects. Larger eggs tend to have more albumen and yolk, providing the hatchlings with a greater energy supply for early development36,37. Environmental conditions, such as nest temperature and humidity, also play a crucial role in shaping hatchling morphology, significantly affecting their development, sex ratio and survival38,39,40,41. Higher nest temperatures can accelerate embryonic development, often resulting in smaller hatchlings due to the faster depletion of yolk reserves42. Conversely, optimal temperature ranges promote healthy development and larger hatchling sizes43,44. Humidity levels within the nest environment impact the yolk absorption and overall morphology, with adequate moisture contributing to better energy reserves and more robust hatchlings45,46,47,48. It is worth noting that excessively high humidity, such as during heavy rainfall or inundation, can also affect the development of sea turtle eggs, potentially leading to lower hatching rates or extended incubation time, which can negatively impact hatchling survival49,50. Additionally, previous studies have shown a correlation between morphological traits and the behavior performance of hatchlings. For example, research on leatherback sea turtles found that heavier hatchlings tend to have faster crawling speeds51. In contrast, a study on green sea turtles revealed that heavier hatchlings exhibited poorer self-righting performance52. Similarly, research on hawksbill sea turtles (Eretmochelys imbricata) found that smaller body size and a larger umbilical scar size were associated with slower self-righting times40.

Despite numerous studies examining the relative contributions of maternal effects and environmental factors on sea turtle hatchlings, findings often vary inconsistently across different geographical locations28,53. There is a notable lack of comprehensive research that simultaneously assesses the impact of both maternal and environmental factors on hatchling morphology and their behavior performance in the South China Sea. This region is critical for the green sea turtle population; however, in recent years, this population has faced threats such as incidental capture, illegal trade, and habitat loss54,55,56. Therefore, understanding the ecological status of green sea turtles, particularly their hatchlings, is helpful for future conservation efforts in the South China Sea. This study aimed to examine the effects of maternal traits (such as carapace length and width), clutch variables (including clutch size, egg weight, and egg size), as well as nest environmental conditions (such as distance to the highest high tide line, inundation days, and clutch temperatures), and incubation duration on hatchling morphology on Taiping Island. In addition, we examined the size of the umbilical scar, which has been linked to self-righting performance40 and may reflect residual yolk mass, a potential indicator of energy reserves57. Furthermore, we utilized both umbilical scar size and hatchling morphology to predict hatchling self-righting performance. By analyzing the relationships between nesting sea turtle characteristics, clutch variables, and environmental factors, this study identifies the key determinants of hatchling development in this nesting habitat. Understanding the factors influencing hatchling morphology and umbilical scar size provides valuable insights into the physiological and ecological processes that shape hatchling fitness, particularly for local green sea turtle populations.

Methods

Study area and sampling procedure

The study was conducted at beaches along the coasts of Taiping Island (10° 22’ N, 114° 22’ E), which has an area of 0.48 km2 in the Nansha Archipelago (Fig. 1). Beach patrols were performed at night from April to September 2021. Nesting green sea turtles were randomly selected, with a total of 35 nesting sea turtles included in the study. To collect morphological data on the nesting sea turtles, the curved carapace length (CCL) and curved carapace width (CCW) were carefully and quickly measured during oviposition, and their nesting sites were marked using ropes. After the nesting sea turtles returned to the sea, we excavated the nests and recorded various clutch variables, including clutch size, egg weight, and egg size (diameter). All eggs of each clutch were measured to ensure comprehensive data collection. We then installed a temperature probe at the center of each clutch connecting with the temperature data logger (Elitech, GSP-6) placed on the ground, and each nest was reburied within two hours to minimize potential impacts on the eggs58,59. Forty days after egg laying, each clutch was enclosed with fences (105 × 105 cm2, with a height of 35 cm) to collect hatchlings. Each night, routine patrols were conducted along the nesting beaches. Upon detecting signs of hatching, the emerging hatchlings were carefully collected in buckets and transported to the laboratory station, located approximately 5 min from the sampling site. In the laboratory, all measurements and behavioral tests were performed, ensuring that the entire process was completed within two hours of hatchling collection. Upon completion of the experiments, the hatchlings were promptly released back to their original beach. This approach was designed to minimize the potential stress on the hatchlings while ensuring the timely and accurate acquisition of data. Hatchlings that emerged near dawn or during the day were placed in buckets lined with sand and were kept moist with water until the evening before the experiments were conducted, and these hatchlings were released at their original beach within 24 h of emergence.

Fig. 1
Fig. 1
Full size image

A map of Taiping Island in the Nansha Archipelago is shown, with the island highlighted by a red dot and pointed to by a red arrow. Thirty-five clutches of green sea turtles (Chelonia mydas) are represented by different symbols, with white dots showing successfully hatched clutches, black triangles indicating unhatched clutches, and grey cross marks for clutches where only the nesting sea turtles and their eggs were measured, but no temperature data loggers were set up.

Measurements of hatchling features and behavior performance

A total of 426 hatchlings were included, with 30 to 62 hatchlings randomly selected from each clutch. To ensure statistically meaningful and representative data, a minimum of 30 hatchlings were selected per clutch. However, when resources allowed, additional hatchlings were included to further increase the sample size. If a clutch contained fewer than 30 hatchlings, all of them were included in the measurements and behavior tests. To evaluate hatchling morphology, the hatchling weight, straight carapace length (SCL), straight carapace width (SCW) were measured. To provide an estimate of the energy reserve of individual hatchling, the umbilical scar length (USL) and umbilical scar width (USW) were also measured57. Additionally, to assess the defensive abilities of hatchlings, their self-righting responses were observed. This was determined by the number of successful righting attempts and the time taken to reposition from a carapace-down position to an upright posture32. Each hatchling was placed on a flat surface with sand collected from the beach, and the trials were recorded from above using a camera (Panasonic Lumix, DC-ZS80). A successful righting event was defined as one in which the hatchling righted itself in under 10 s. If the hatchling took longer (≥ 10 s), an additional 30 s was allowed for rest. The trial was concluded once the hatchling had completed three successful righting events or six unsuccessful ones. The time taken for each successful righting event was recorded by reviewing the video footage, while each unsuccessful attempt was assigned a 10-s time cost.

Environmental variables

Clutch depth was measured during the excavation of each nest, and the distance from the surface directly above the nest to the highest high tide line was manually recorded. Inundation days, referring to the days when the entire area directly above the nest was submerged by seawater, were determined through daily patrols and estimated using daily tidal heights taken from the marine station on Taiping Island. Additionally, the in situ clutch temperatures were automatically recorded every 20 min using temperature data loggers.

Ethics statement

This field study was conducted in accordance with the Act on Wildlife Conservation 2013 in Taiwan, reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of National Taiwan University (NTU-110-EL-00088), IACUC of Academia Sinica (Protocol # 21-07-1691), and Ocean Affairs Council (1,100,008,557), and conducted under their authority. No extra animal discomfort was caused for sample collection for the purpose of this study.

Statistical analyses and modeling

To assess the relationships between the morphological characteristics (CCL and CCW) of 35 nesting sea turtles and their clutches (clutch size, egg weight, and egg size), Pearson’s or Spearman’s correlation coefficients were calculated, depending on the normality of the individual variables. The p-values were then adjusted using the Bonferroni correction due to small sample size.

Of the 35 clutches monitored, 2 were lost due to coastal erosion, and 4 had not reached the end of the incubation period when the study ended. Therefore, a total of 15 hatched clutches were included in the following analysis. For hatched clutches, the entire incubation period (from egg laying to hatchling emergence) for each clutch was divided into three equal intervals, referred to as the first (1st), middle (2nd), and last (3rd) incubation windows, and the mean clutch temperature (mean ± SD) was then calculated for each of these windows. Pearson’s correlation coefficients were used to examine the relationships between clutch size and clutch temperatures during each incubation window, with p-values adjusted using the Bonferroni correction. Pearson’s correlation coefficients followed the Bonferroni correction were also used to examine the relationships between hatchling morphology (including weight, SCL, and SCW) and umbilical scar length and width.

A linear mixed-effects model was employed to investigate the influence of maternal and environmental factors on hatchling morphology and umbilical scar size. Hatchling morphology was represented as a composite variable, formed by summing the hatchling weight, SCL, and SCW for each individual, while umbilical scar size was represented by multiplying the USL and USW for each individual hatchling. The model included both fixed effects, such as maternal characteristics of nesting sea turtles (CCL and CCW), clutch-related variables (clutch size, egg weight, and egg size), and environmental variables (clutch depth, distance to the highest high tide line, inundation days, and clutch temperatures during each incubation window), and incubation days, as well as individual variability among mother green sea turtles as a random effect. The fixed effects were incorporated to assess their direct influence on hatchling characteristics and umbilical scar size, while the random effect was included to model the inter-individual variability. To assess model significance, a Likelihood Ratio Test (LRT) was conducted by comparing the mixed-effects model with a null model (random effect only) using Analysis of variance (ANOVA). Additionally, Akaike Information Criterion (AIC) was calculated for both the mixed-effects model and the null model. AIC provides a measure of model fit while accounting for model complexity, with lower AIC values indicating a better balance between fit and complexity. Variance Inflation Factor (VIF) was calculated to evaluate multi-collinearity, and variables with a VIF greater than 5 were removed from the model to mitigate multi-collinearity. The goodness of fit was assessed using both marginal and conditional R2 values, which represent the proportion of variance explained by the fixed effects and random effects, respectively.

To assess the effects of hatchling morphology and umbilical scar size on hatchling behavior, the individual mean time spent in the self-righting response was analyzed using a linear mixed-effects model. The model included both fixed effects, such as hatchling weight, SCL, SCW, USL, and USW, as well as individual variability among mother green sea turtles as a random effect. As aforementioned analysis, model significance was assessed using a Likelihood Ratio Test comparing the mixed-effects and null models. Additionally, multi-collinearity was evaluated using VIF, and model fit was assessed with both marginal and conditional R2 values.

All analyses were conducted in R 4.4.3 (R Core Team, 2025) and SigmaPlot 14 (SYSTAT, NA) softwares.

Results

Nesting sea turtles and clutches

Of the 35 nesting sea turtles observed on Taiping Island, the measurements of curved carapace length (CCL) and curved carapace width (CCW) ranged from 73.0 to 106.0 cm and 69.0 to 103.0 cm, with averages of 96.5 ± 6.1 cm and 88.6 ± 6.6 cm, respectively (Fig. 2A). Additionally, the clutch size, egg weight and egg size ranged from 31 to 146 eggs, 29.5 to 48.8 g, and 37.8 to 44.4 mm, with averages of 97.3 ± 23.5 eggs, 40.4 ± 4.1 g and 42.0 ± 1.4 mm, respectively (Fig. 2B–D). Both the CCL and CCW of nesting sea turtles were positively correlated with egg weight (r = 0.54, adjusted p = 0.012 and r = 0.51, adjusted p = 0.023) and egg size (r = 0.57, adjusted p = 0.005 and r = 0.53, adjusted p = 0.016), but were not correlated with clutch size (r = 0.31, adjusted p = 0.810 and r = 0.33, adjusted p = 0.682, Table S1). On the other hand, clutch size showed no significant correlation with either egg weight (r = -0.19, adjusted p = 1.000) or egg size (r = -0.09, adjusted p = 1.000). These results indicated that larger green sea turtles nesting on Taiping Island lay bigger, heavier, but not necessarily a greater number of eggs.

Fig. 2
Fig. 2
Full size image

The boxplots show the distributions of (A) the length of curved carapace length and curved carapace width of nesting sea turtles, and (B) the clutch size, (C) mean egg weight, and (D) mean egg diameter of individual clutch.

Hatchling morphology and its predictors

We assessed hatchling morphology across 15 successfully hatched clutches (Table 1). The mean hatchling weight ranged from 17.7 ± 0.44 g to 24.7 ± 1.69 g, straight carapace length (SCL) from 43.9 ± 0.52 mm to 49.3 mm, and straight carapace width (SCW) from 34.9 ± 0.84 mm to 38.9 ± 1.39 mm. The mixed-effects model showed that both egg size (β = 2.00, p = 0.003) and distance to the highest high tide line (β = 0.74, p = 0.003) significantly contributed to hatchling morphology (Table 2). This suggests that hatchlings from larger eggs and those farther from the highest high tide line tend to exhibit larger body sizes. In the likelihood ratio test, the mixed-effects model had a lower AIC (2313.6) compared to the null model (2326.4), with a chi-square value of 31.79, 9 degrees of freedom (Df), and a p-value of < 0.001. This provides strong evidence for the inclusion of these predictors in the model. Moreover, the VIF values (ranging from 1.67 to 4.76) indicate minimal multi-collinearity, and the model explained 38% of the variance through the fixed effects (marginal R2) and 57% through both fixed and random effects (conditional R2).

Table 1 The measurements of clutches that successfully hatched on Taiping Island in 2021.
Full size table
Table 2 Prediction of hatchling morphology and umbilical scar size: Summary of the results from the linear mixed-effects model.
Full size table

Factors influencing umbilical scar size

When examining umbilical scar size, the mean umbilical scar length (USL) and umbilical scar width (USW) of hatchlings ranged from 9.8 ± 1.00 mm to 12.4 ± 1.31 mm and 2.5 ± 0.57 mm to 4.6 ± 1.22 mm, respectively (Table 2). No significant correlations were found between these umbilical scar measurements and hatchling morphology (all r < 0.31, adjusted p = 1.000, Fig. 3). However, the clutch temperature significantly predicted umbilical scar size. Specifically, clutch temperature during the middle incubation window inversely affected umbilical scar size (β = − 3.99, p = 0.028), while clutch temperature during the last incubation window had a positive effect (β = 6.25, p = 0.017). These results suggest that varying incubation temperatures during different stages of development can influence the size of the umbilical scar. In the likelihood ratio test, the mixed-effects model had a lower AIC (3214.9) compared to the null model (3224.6), with a chi-square statistic of 27.65 (Df = 9, p = 0.001). Additionally, the VIF values ranged from 1.67 to 4.76, indicating no significant issues with multi-collinearity. The model explained 32% of the variance through the fixed effects (marginal R2) and 48% when considering both fixed and random effects (conditional R2).

Fig. 3
Fig. 3
Full size image

The scatter plots represent the relationships between hatchling morphology and (A) mean umbilical scar length, and (B) mean umbilical scar width of hatchlings of each clutch. The variables of hatchling morphology include mean weight (blue), mean straight carapace length (red), and mean straight carapace width (yellow) of hatchlings of individual clutch.

Self-righting performance and predicting variables

In our analysis of self-righting performance, we found that the time required to right themselves ranged from 1.6 ± 1.32 s to 10 s across clutches (Table 3). The mixed-effects model revealed that both hatchling morphology and umbilical scar size were significant predictors of self-righting performance. Specifically, straight carapace width (SCW) was inversely associated with self-righting time (β = -0.26, p = 0.007), indicating that hatchlings with wider carapaces took less time to self-right. In contrast, umbilical scar width (USW) was positively correlated with self-righting time (β = 0.43, p = 0.003), suggesting that larger umbilical scars were associated with longer self-righting responses. This finding highlights the importance of both body size and scar dimensions in influencing hatchling behavior. In the likelihood ratio test, the mixed-effects model had a lower AIC (1786.6) compared to the null model (1819.1), with a chi-square statistic of 42.58 (Df = 5, p < 0.001). The model also showed minimal multi-collinearity, with VIF values ranging from 1.24 to 2.99. The model explained 10% of the variance through fixed effects (marginal R2) and 43% when considering both fixed and random effects (conditional R2).

Table 3 Prediction of self-righting performance: Summary of the results from the linear mixed-effects model.
Full size table

Discussion

Through investigation of nesting green sea turtles, clutch variables, environmental conditions, and hatchlings, we have provided evidence to clarify the effects of maternal characteristics, clutch, and environmental factors on the hatchling morphology and the umbilical scar size on Taiping Island. Our study further showed that both hatchling morphology and umbilical scar size can be used as indicators for predicting hatchlings’ self-righting behavior (Fig. 4). These findings contribute to the understanding of the embryonic development of green sea turtles in the South China Sea and offer valuable biological data for their conservation in the future.

Fig. 4
Fig. 4
Full size image

A conceptualized graph illustrates the predictions for hatchling morphology and umbilical scar size of green sea turtle hatchlings, as well as how these traits influence the self-righting response. Variables with significant positive impacts are highlighted in red, while those with inverse influences are shown in blue.

In the present study, we found that the size of nesting sea turtles was positively correlated with egg weight and egg size, but not with clutch size. Additionally, clutch size was not significantly correlated with egg weight or egg size (Table S1). These results are consistent with the findings of previous studies on green sea turtles at Tortuguero, Costa Rica33 and loggerhead sea turtles at Masirah Island, Oman60. However, they differ from reports for both leatherback and loggerhead sea turtles26,61, suggesting that reproductive behavior exhibits distinct strategies due to species and geographic variations. This highlights the need for regional surveys to understand the “optimum trade-off” of reproductive investment. Nevertheless, a similar phenomenon was observed in green (our study), leatherback, and loggerhead sea turtles, where egg mass, including egg weight and size, determines hatchling morphology (Table 2). This suggests a conserved mechanism where maternal investment dominates embryo development29,35,37,47. In addition to maternal influences on hatchling morphology, environmental conditions play crucial roles in hatchling development. For instance, events such as inundation were found to negatively affect hatchling growth and development50,62. This may help explain the positive relationship observed between the distance from the nest to the highest high tide line and hatchling morphology in our predictive model. A greater distance from the highest high tide line suggests a reduced likelihood of nest inundation, which may mitigate environmental stressors and promote more favorable conditions for hatchling development. These findings highlight the significance of nest location in influencing hatchling survival and development.

In terms of post-hatchling umbilical scar size, there was no significant correlation between hatchling morphology and umbilical scar size (Fig. 3). This suggests that multiple factors may influence this relationship, including genetic variations, nutrient consumption efficiency, and environmental differences across clutches on Taiping Island33,63. In the mixed-effects model, the umbilical scar size was inversely correlated with the middle incubation window, while positively correlated with the last incubation window (Table 2). Considering that the middle incubation window plays a crucial role in shaping the morphology of hatchlings, including sex determination and the development of many organs and body parts, an increase in clutch temperature during this window accelerates nutrient conversion and yolk absorption64,65,66. However, during the later stages of development, increased metabolic activity tends to elevate nest temperature, often making it higher than that of the middle incubation window67,68. Given that the average clutch temperature during the final incubation window ranged from 30.20 to 35.59 °C on Taiping Island (Fig. S1), which is already near the critical tolerance threshold (33–35 °C)69,70, an increase in clutch temperature accelerates embryonic development but reduces the time available for nutrient transformation71,72. This trade-off between survival and energy conversion results in an accelerated hatching process, leading to larger umbilical scars in nests from the higher last incubation window. Notably, the clutch temperature during the last window showed a positive trend with clutch size (r = 0.61, adjusted p = 0.216, Fig. S1), likely due to gains in metabolic heat. These findings support those reports on green sea turtles and loggerhead sea turtles, where metabolic heating of developing embryos and microbial activity in dead eggs are inferred to be determined by clutch size73,74,75.

Although a previous study suggests that umbilical scar size may serve as an indicator for estimating residual yolk mass, there is no direct evidence linking their relationship57. This is because the umbilical scar gradually disappears within a few days after hatching, and its size may be influenced by variations in emergence timing, which could affect the measured values. Nevertheless, both residual yolk mass76,77 and umbilical scar size40 are influenced by and correlated with clutch temperatures. Additionally, both umbilical scar size and residual yolk mass are related to behavior performance of hatchlings. A previous study suggests that hatchlings with smaller yolk reserves are expected to fatigue more quickly, as they rely solely on their yolk reserves during the frenzy period to fuel swimming behaviors22. Another study also found a positive correlation between umbilical scar size and the time required to complete the self-righting response40, which is consistent with our findings. Therefore, it is possible that, like residual yolk mass, umbilical scar size could serve as an indicator of post-hatching energy reserves and predict behavior abilities and survival capabilities after hatching.

In our study, whether using maternal variables, clutch variables, and environmental variables to predict hatchling morphology and umbilical scar size, or using hatchling morphology and umbilical scar size to predict hatchlings’ self-righting performance, we found that maternal variability as a random effect cannot be ignored. This suggests that the random variability associated with individual mothers plays an important role in shaping both hatchling morphology and their subsequent behaviors26,30. However, after adding the fixed effects, the model’s predictive accuracy improved, indicating that incorporating these variables enhanced the model’s ability to predict hatchling traits and behaviors. Overall, both hatchling morphology and post-hatching energy reserves may play significant roles in determining the survivability of hatching sea turtles78. Although some evidence suggests that bigger is not always better79,80,81, post-hatching energy reserves may have no impact on the fitness of hatchlings due to potential overprovision by their mothers82. Understanding the ecological traits and potential influencing variables at local nesting habitats is helpful for regional conservation activities. This experimental field study not only assesses the reproductive ecology and embryonic development of green sea turtles on Taiping Island, which has not been surveyed in the past two decades, but also reveals that hatchling morphology and umbilical scar size can be explained by distinct variables. The findings of this study offer valuable insights into the factors influencing hatchling development and offspring fitness, providing regional biological data that can serve as a critical reference for understanding the potential impacts of climate change and anthropogenic factors on green sea turtles in the future.