ISTS Symposium43 Program/Agenda

Background. Variations in ecological roles of sea turtle species may lead to differentiations in ocular design and visual sensitivity to the colour spectrum. The capability of green turtles and loggerhead turtles possessing visual components necessary for colour vision has been supported through electroretinograms and microspectrophotometry studies. Behavioural colour preference studies in air and in water on hatchling and post hatchling green turtles found evidence of a longer wavelength preference and a blue hue attractiveness when given a choice between blue, red, and yellow colours. This paper assessed and compared the colour preferences to singular colours via behavioural responses of hawksbill turtles and green turtles at 15 months of age and again at 22 months of age.

Methods. Eleven hawksbill turtles and twelve green turtles were presented with one coloured water balloon per day (purple (400-450 nm), dark blue (450-490 nm), cyan (490-520 nm), green (520-560 nm), yellow (560-590 nm), orange (590-635 nm), and red (635-700 nm)) with a seven-month gap before repeating the study. Balloons were secured mid water and turtles were individually placed into their housing compartments opposite the stationary balloon. Latency time for turtles to contact balloons with beak and behaviours exhibited by turtles were recorded.

Results. Interaction with balloon colour was significantly influenced by species. Hawksbill turtles had the highest levels of interactions between balloons of shorter wavelengths with hue preference being between 450-490 nm. Green turtles consistently had the highest level of interaction to longer wavelengths with a yellow (560-590 nm) hue preference seen in both experimental phases. Smaller Hawksbill individuals have a longer cumulative contact time than larger individuals. On the other hand, no trend has been found for the green turtles.

Discussion. This study supports physiological differences between these two turtle species that may reflect variations in visual capabilities as a result of different ecological niches of hawksbill turtles and green turtles.

During their reproductive seasons, multiple paternity occurs across various taxa, including marine turtles, terrestrial turtles, reptiles, and mammals. In Florida, an estimated 31-70% of sea turtle nests show evidence of multiple paternal contributions. For sea turtles, multiple paternity is hypothesized to increase fitness benefits for males, increase fertilization success, and increase genetic diversity. Increasing genetic diversity could also result in females indirectly increasing their fitness if their offspring are of higher quality and are successful. The rate of multiple paternity varies due to factors such as mate selection and mate availability. Sperm competition may increase the genetic diversity within a clutch of eggs and could potentially increase the genetic quality of the offspring if the strongest males’ contribution were used for fertilization. Predation of hatchlings directly after emergence is a considerable risk, with approximately 10% of hatchlings dying on their way to the water. Hatchlings with better locomotive efficiency have a higher chance of reaching the shoreline, which increases their chances of survival. 15 nesting females and a subset of their hatchlings were examined in 2024 in Casey Key, Florida, one of the Gulf of Mexico’s largest loggerhead turtle rookeries. For this study, females that were observed completing one successful nesting event were selected, to allow for the possibility of greater genetic diversity, and potentially differing paternal contributions, between nests. By analyzing loggerhead sea turtle hatchling locomotion and comparing those values to paternity, we can discover if multiple paternity rates impact hatchling locomotive efficiency. Mean clusters comparing self-righting response and crawling test results of individual hatchlings within the same clutch were created. Using DNA analysis to compare individual hatchling paternity, these mean clusters of locomotive efficiency may represent different fathers within the clutch, suggesting that hatchling fitness is impacted by their paternity, which may encourage sperm competition in females. This study aims to quantify the incidence of multiple paternity within loggerhead clutches for females observed nesting a single time on Casey Key, Florida and to evaluate whether the paternity of individual hatchlings affects their locomotive efficiency, influencing their survival rates.

As hatchling sea turtles emerge from subsurface nests on the beach, they exhibit an innate hyperactive state: rapidly crawling to the water, followed by almost continuous swimming for 24-48 hours as they escape coastal, predator-rich waters. This “frenzy period” is dominated by powerstroking – characterized by elaborate ranges of motion of both forelimbs (flippers) bilaterally, as dorsoventral “flapping”. Cheloniid sea turtles incorporate dogpaddling or “rear-flipper kicking” for routine swimming after the frenzy. In contrast, leatherbacks seem to continue powerstroking almost exclusively throughout life. How can we understand why leatherbacks only use the powerstroke? Unlike cheloniid sea turtles, sightings of neonate or juvenile leatherbacks at sea are virtually nonexistent. Their open-ocean life history makes the study of ontogenetic shifts and dispersal challenging and rare. We compared the powerstroke of over 60 leatherback individuals at two different life stages, as hatchlings and then as neonates. The turtles came from nests on the same nesting beaches in Southeast Florida, USA between 2022 and 2024. We hypothesize that morphological and behavioral shifts between life stages are reflected in kinematic differences in the powerstroke at different life stages. To test this hypothesis, hatchlings were collected on the day of nest emergence, measured, marked with high-contrast dots for motion tracking, placed in a sea water-filled tank, and swimming was video recorded at high resolution. Force data were collected synchronously with videos to measure thrust patterns in concert with flipper motions. Both force measures and video were sampled at high rates, allowing us to uniquely pair when in the stroke force is produced (i.e., upstroke versus downstroke) and identify the corresponding motion of the flipper (i.e., location and angle of the flipper) with peaks and troughs of force values within a single stroke cycle. Turtle swimming was compared as hatchlings and after 4-8 weeks of growth. We found (1) ratios of flipper length to straight carapace length decreased with age. (2) Flipper movements and thrust production patterns (force/time per stroke) differ across life stages and among individuals at the same life stage. These results identify ontogenetic shifts and individual variation. By using the same methodology to compare the well-studied movement of hatchlings, to the understudied movement of older neonates, we identified functional and behavioral differences that likely contribute to the mysterious early dispersal of leatherback sea turtles.

The conservation of sea turtles is essential for maintaining the balance of marine ecosystems, making it crucial to identify them individually to keep accurate records in conservation programs. By leveraging their morphological features, it is possible to identify them individually. In recent years, non-invasive photo-identification methods have been developed, which, despite representing technological advancements, face challenges in uncontrolled environments, posing a significant challenge for computer vision.

Extracting the morphological features of sea turtles through image content analysis and representing them specifically using mathematical and computational algorithms is highly important to enable these photo-identification methods to accurately learn to identify each sea turtle. In this context, graph-based algorithms are particularly useful for capturing detailed information about the morphology of sea turtles. The shape of the scales, shields, head, shell, and flippers is represented through a structure of nodes or specific points using graphs, and the relationships among these shapes are determined by linking their structure, coloration, and texture, facilitating precise identification. These algorithms analyze patterns, shapes, and unique markings, enabling individual distinction.

This work presents a comparative analysis of the performance of two graph-based algorithms: the Region Adjacency Graph (RAG), which prioritizes spatial contiguity, and the K-Nearest Neighbors Graph (KNN), which relies on metric proximity. Both algorithms are designed to generate graph structures that capture relevant morphological features derived from images of sea turtles. The graph structure is integrated into an optimized architecture of Graph Attention Networks (GAT), developed for the automatic identification of sea turtles. GAT leverages attention mechanisms, enabling graph nodes to extract information from their neighborhoods. This enhances both the accuracy of sea turtle identification and the adaptation to the complexities of graph representations.

The results show that the RAG algorithm plays a crucial role in the identification process, outperforming KNN in terms of accuracy and computational efficiency. Its ability to extract only the key information of each sea turtle allows for more robust and efficient representation, even under uncontrolled image capture conditions. This RAG-GAT development framework is efficient, enabling its application across various image capture conditions and eliminating the need for controlled environments for individual identification. To the best of our knowledge, this is the first study to explore these parameters in a GAT architecture using a real and diverse dataset, such as that of sea turtles, promoting the use of innovative models for the conservation of these species. The ability to quickly and accurately identify each turtle through photo-identification methods not only supports conservation efforts, detailed population records, and the development of non-invasive methods, but also provides effective technological tools for their monitoring and preservation in their natural habitats. Moreover, it deepens our understanding of their ecology and behavior, which is essential for ensuring their long-term protection.

Hawksbill sea turtles are listed by the IUCN as critically endangered. Due to their endangered status, conservation measures are important for the continuation of the species. To implement best practices for Hawksbill conservation, the biology of the animal should be well understood. Prior work indicates that the breeding strategy of free-ranging Hawksbill sea turtles in Bocas del Toro with regard to how they support their metabolism remains unknown. In general, captive and wild animals have different physiologies, but comparing the two can provide valuable insight on how wild animals may behave when inaccessible to humans. Captive Hawksbill samples are instrumental in providing a perspective to interpret the wild free ranging animals’ results. The goal of this study was to explore biomarkers of foraging and reproduction in captive, female Hawksbill sea turtles. Captive animals were sampled at Okinawa Churaumi Aquarium in Okinawa, Japan for an entire year in 2018 and exhibited mating and laying over the course of sampling. We measured the concentration of β-hydroxybutyrate (BHB), triglycerides (TRG), and testosterone (T). BHB and TRG are foraging biomarkers, and at elevated concentrations should indicate fasting and foraging, respectively. T is a reproductive hormone which has been observed to gradually decline throughout a female sea turtle nesting season. BHB concentrations were significantly higher during the breeding season (April – July, p<0.05) compared to the non-breeding period (November – March). Analysis revealed that TRG concentrations were highest early in the breeding season (April – May), gradually decreased throughout the season, and showed a significant decline in October (p<0.05). Testosterone concentrations were significantly elevated during the breeding season, particularly in May (p<0.05), which marked the onset of mating. These results indicate that the animals are behaving as capital breeders and fasting during breeding season. Results also suggest TRG concentration is more related to reproduction than foraging in these animals.

Arginine vasotocin (AVT) is a neuropeptide involved in various physiological processes. In reptiles, AVT has been shown to stimulate smooth muscle contraction in the oviduct, facilitating egg deposition. In sea turtles, AVT concentration has been shown to increase significantly during egg laying, peaking around mid-clutch. Before and after egg laying, sea turtles exhibit a distinct series of behaviors; however, not all individuals successfully oviposit. Some females display disruptions in their typical nesting behavior, returning to the water without laying eggs, referred to as a non-nesting emergence (NNE). We tested the hypothesis that external disturbances contribute to NNEs. We predict that increased corticosterone, the sea turtle stress hormone, inhibits AVT production, thus preventing the animals from nesting. To address this hypothesis, we collected blood samples (n=32) from nesting loggerhead turtles at the X’cacel-X’cacelito sanctuary in Quintana Roo, Mexico from June 7 to June 18. We collected samples from the dorsal cervical sinus during successful oviposition and NNEs and categorized samples according to oviposition stages and nest behavior: mid clutch (around 50% full; n=6), end of laying (more than 75% full; n=7), covering of the egg chamber (n=14), and NNEs (n=5). The primary aim of this study is to establish baseline data on the physiological mechanisms occurring during NNEs. AVT and corticosterone concentration analysis will help determine if these hormones are linked to disruptions in nesting behavior, and whether they can be considered physiological markers for NNEs.