Study Area. —

The study area, Barreira do Inferno, is located on the southern coast of Rio Grande do Norte, in the municipality of Parnamirim (lat 5°54′56″S, long 35°15′46″W). The study area consists of 3 beaches: Alagamar, Morro Branco, and Prainha (total length around 5 km). The landscape is composed of cliffs and dunes. Sandy beaches in the area are well suited for marine turtle nesting, as they are free from human occupation, completely in the dark, and without the presence of artificial lighting. The study site is located between 2 very urbanized beaches, making the region a nesting refuge (Fig. 1) (Monteiro et al. 2019). The Barreira do Inferno is a military area, which helps to preserve the natural environment in the region.

View Figure

• Download as Powerpoint
• Download Figure

Data Collection. —

We followed methodologies proposed by the Brazilian Sea Turtle Conservation Program (Tartarugas Marinhas or TAMAR; Marcovaldi and Marcovaldi 1999; Santos et al. 2013), with night patrols starting at 7:00 PM and continuing until 3:00 AM. The field experiment started in December 2018 and extended until June 2019. Each patrol consisted of monitoring on foot the entire length of the beach, with 40-min rest stops at each end. This procedure increases the probability of sighting turtles while they nest, as a successful nesting effort is approximately 40 min in duration (Santos et al. 2013; Marcovaldi et al. 1999). The daytime patrols consisted of monitoring starting at dawn (around 5:00 AM) to record nesting events as evidenced by nesting crawl tracks from the previous night.

After the nesting process, we located the egg clutch with the aid of a stick that we inserted into the sand to probe for eggs, followed by the confirmation of encountering eggs by hand excavation. A nylon string tied to a piece of wood is buried above the egg chamber and tied to a numbered stake inserted adjacent to the nest so that the nest location was identifiable throughout incubation. After nests were marked, experimental treatments were applied to nests. However, nests in the control group were not tagged with a string and marker stake. All nests had their coordinates marked on a GPS device (accuracy of 3 m).

Predation Interactions and Predator Identification. —

Visitations to nests by foxes were classified into 3 categories: 1) visits without further action (named here as visits); 2) nest excavation without predation (named here as excavation); and 3) nest excavation with predation (named here as predation). “Visits without further action” were identified as the presence of predator tracks without disturbance of the nest (Fig. 2A). “Nest excavation without predation” (Excavation) were identified as digging within a circle of 1-m diameter centered on the nest, but without signs of egg or hatchling predation (Fig. 2B). Nest excavations with predation (Predation) were identified when digging at the nest occurred, together with observed broken eggshells scattered around the nest (Fig. 2C).

View Figure

• Download as Powerpoint
• Download Figure

Predators were identified via direct or indirect observation, involving a camera trap or by identifying tracks close to the nest, respectively. Most of the time the identification was indirect; fox paws leave impressions of 4 digits and claws, with an elongated shape and rounded ends. The 2 middle digits are noticeably forward of the 2 outside digits. The cushion is small in relation to the size of the digits, with a paw size varying from 4 cm to 5.5 cm of width (Fig. 3A; Moro-Rios et al. 2008). Only a single camera trap was available for this study, and therefore, its use was restricted. The camera was set up to record at a 90° downward-sloping angle, with a motion-activated trigger time of approximately 0.5 sec, and recording 50-sec videos when triggered (Fig. 3B). This setup was used for direct observation of predators.

View Figure

• Download as Powerpoint
• Download Figure

Nest Treatments. —

Predation mitigation manipulations began in the first week of January 2019 and were completed in the second week of June 2019. Nests were monitored until hatchings emerged, when the team counted the number of live hatchlings emerging from each nest. Eggs were not counted immediately after deposition: they were counted after hatchlings emerged from the nest. We tested 5 strategies to protect nests from predation (Table 1). The evaluated strategies were the following.

Table 1. Brazilian fox Cerdocyon thous behavior according to the nest treatment. N is the number of nests in the experiment.

View Fullscreen

Numbered Stake. —

Only the nylon string and the nest identification stake (PVC pipe with nest number and identification of the TAMAR project) were used in this treatment. This treatment was applied to test if this standard procedure influences predator activity because it is part of the standard methodology used by the TAMAR Project (Fig. 3B). For more details of the methodology see Santos et al. (2013).

Pepper Powder. —

After placing the stake, using a measuring tape, a circle with a diameter of 1 m was drawn and 150 g of cayenne pepper powder was sprinkled inside. After application, a layer of sand was sprinkled on top of the pepper powder to prevent it from being dispersed by the wind. This process was repeated every 15 days until hatchling emergence or when partial predation occurred at the nest. This “aversion taste” methodology is adapted from Lamarre-DeJesus and Griffin (2013), who studied coyote nest depredation. However, we note that they used habanero pepper instead of cayenne pepper as was used here.

Mesh. —

After placing the guide and stake, a ∼15-cm-deep square-shaped hole was dug over the nest into which a 1-m² galvanized mesh screen with 10 cm × 5 cm mesh size. The mesh is a predator exclusion device, but it allowed emerging hatchlings to pass through unobstructed. Eight wooden stakes (30 cm in length) were hammered into the sand to fix each nest screen in place. Afterward, the mesh was covered with a ∼5-cm layer of sand. This methodology has been used previously (Lamarre-DeJesus and Griffin 2013; O’Connor et al. 2017).

Flag. —

After applying the guide and stake, a white, raw cotton flag measuring 80 × 50 cm was attached to a 120-cm rod and planted in the sand next to each treatment nest. The flag was positioned so that wind (when present) would keep the flag waving over the nest. The flag method was proposed because it can work as a potential mechanical, visual, and auditory deterrent for predators. This methodology was adapted from Longo et al. (2009). The wind regimen in this coastal area of Rio Grande do Norte typically flows from the east to southeast. These winds are part of the trade winds that blow steadily from the east across the Atlantic Ocean. The dry season, from September to April, comprises the oviposition period and generally has more consistent wind patterns, with the easterly trade winds being dominant (Medeiros et al. 2001).

Control. —

Control nest group received no treatment to keep human interference to a minimum. A distance 3 m to the left and right of the nest was measured, and a stake placed at each point. These stakes—relatively distant from the nest—were used to mark the nest position during the incubation period; additionally the GPS nest location was recorded.

The sea turtle nesting season spans from November to June, with a peak in nesting in March. The 64 nests selected for this study were divided among the 5 treatments: 13 numbered stake nests, 12 pepper-powder–treated nests, 12 mesh nests, 15 flag nests, and 12 control nests. During high tides in March, a flag treatment nest was washed away. During the high tides of June, a mesh treatment nest had to be moved so that it would not be washed away. Therefore, it was not possible to monitor all nests until hatchlings emerged. Even so, these 2 nests were included in the analysis as visits and excavations, since no signs of predation were recorded up to the final day of observation.

Statistical Analysis. —

All statistical analyses were performed in the R program (R Core Team 2021). To test for differences among nest–C. thous fox interactions and nest treatments we used Kruskal-Wallis tests. We highlight that an ANOVA test cannot be used here because data do not fulfill the normality condition nor do the residuals follow the homoscedascity condition. Indeed, many nests have few interactions, while others have very high numbers of interactions. Moreover, multiple comparisons were performed using a pairwise Wilcox test with Bonferoni correction.

Linear and quadratic curves were fitted to the data to investigate trends in observations. A linear trend indicates a constant increase (or decrease) with incubation duration, whereas a quadratic trend indicates the presence of a maximum (or minimum) in temporal fox interaction. The choice of the best model, linear or quadratic, was performed using the Akaike Information Criterion (AIC), a likelihood tool used in the competition model context.

General Overview of Fox-Nest Interactions. —

During the 2018/2019 nesting season, a total of 85 nests were recorded in the Barreira do Inferno area. Of these, 15 nests were recorded prior to the commencement of the controlled scientific study, and 2 nests were relocated to a protected incubation area due to the risk associated with tidal flooding. Among the 64 nests considered suitable for study inclusion, 54 records were of E. imbricata, 7 of the green turtle (Chelonia mydas), and 3 remained unidentified. Nonidentification occurred in cases where the nesting female was not sighted or hatchlings could not be identified due to the nest being destroyed by predation or waves.

Predators’ visits to nests were attributed to crab-eating foxes (C. thous) (96%) and the domestic dog Canis lupus familiaris (1%), while in some cases it was not possible to identify the visiting species (3%). Importantly, C. thous was responsible for all (100%) recorded nest excavation and predation events. The presence of ghost crabs (Ocypode quadrata) was also documented in the nests. Although crab burrows were frequently observed during the nesting season, no direct predation of eggs or hatchlings by ghost crabs was recorded. The distribution of ghost crab activity across the treatments varied: 22 records were associated with the numbered stake treatment, 183 with pepper powder, 49 with mesh, 21 with flags, and 37 in the control group.

Interactions with nests by C. thous were recorded and organized by date of occurrence. For convenience, the data were grouped into periods of 7-day interval (weeks) to illustrate the temporal pattern of predator interactions throughout the nesting season. The highest frequency interactions by C. thous with marine turtle nests occurred between week 6 and 15 (between February and April). Notably, this period aligns with the peak of nesting in the study area, which typically occurs from 12 February to 2 March (Oliveira et al. 2020).

The proportion of nests predated by Cerdocyon thous was analyzed for each treatment to assess their effectiveness. Predation frequency was highest in the “control” (50%) and the “numbered stake” (46%) groups, followed by the “pepper powder” (42%). The “mesh” (17%) and the “flag” (13%) treatments exhibited the lowest rates of predation (see Table 1 and Fig. 4).

View Figure

• Download as Powerpoint
• Download Figure

Cerdocyon thous Visiting and Predation. —

In this study, we examine the interactions with turtles from the behavioral perspective of C. thous, which includes visits without further action, excavations without predation, and excavations with predation. Evidence of nest predation by foxes was characterized by the presence of eggshells fragments, fox tracks, signs of excavation, and broken eggshells around a nest. During event predation, foxes typically consume 2 to 5 eggs, and the same nest can be visited or predated several times during the incubation period. To analyze the frequency and patterns of fox behavior, we quantified the number of interactions per nest and used these data for statistical analyses. Figure 4 illustrates the dominance of visiting behavior over other behaviors. A Kruskal-Wallis test (df = 2, F = 147.8, p < 0.0001) revealed significant differences among the 3 behavioral categories. Pairwise Wilcox test further indicated that the visiting behavior occurred at significantly higher frequencies than either predation or excavation, while no significant difference was observed between predation and excavation (p < 0.05).

To test the differences among the treatments, we also perform a Kruskal-Wallis test. The result revealed a significant difference among the treatments (df =  4; χ² = 32.7; p < 0.0001; Fig. 4). A pairwise Wilcox test with Bonferronin correction showed that the “flag” treatment significantly differs from all other treatments (p < 0.0001). However, no significant differences were found among the other treatments. These findings suggest that the flag treatment is more effective in reducing visitation and predation by C. thous foxes compared to the other treatments.

In this analysis, time 0 is defined as the moment the turtle digs the nest, and the final time corresponds to the hatching of the nest, which takes an average duration of 54 days (Santos et al. 2016). Linear regression analysis (Figs. 5 and 6) revealed a decreasing trend in nest visitations over time (df = 62; t = −9.02; R = −0.75; p < 0.00001). Similarly, excavation without predation decreased over time (df = 62; t = −2.5; R = −0.30; p < 0.015). In contrast, predation interactions did not exhibit a linear increase or decrease throughout the incubation period (df =  62; t = −0.09; R = −0.012; p < 0.97).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

A second analysis was conducted based on studies suggesting greater predatory behavior at the beginning and end of the nest incubation period. To test this hypothesis, we applied a quadratic regression model to the data and evaluated the statistical significance of the fitting. The quadratic regression analysis showed a poor fitting for visitation data (df = 62; t = 0.62; p = 0.55). However, the analysis for “excavation without predation” over time showed a significant quadratic trend (df = 62; t = 3.58; p < 0.0006) as did “excavation with predation” (df = 62; t = 3.96; p < 0.0002). In fact, both excavation and predation were up to 10 times higher at the beginning and end of the incubation period compared to the middle.

A comparison between linear and quadratic models was conducted to identify the optimal model for each behavior, with the minimum value indicating the best fit. The AIC values presented in Table 2 suggest that the linear trend provides a better for the visitation behavior, whereas the quadratic model is more suitable for excavation and predation behaviors. The interpretation of these results is that visiting behavior decreases as incubation time progresses. In contrast, the quadratic model outperforms the linear model for excavation and predation, indicating that these behaviors exhibit 2 peaks of maximal activity—1 at the beginning and another at the end of the incubation period. Specifically, peak activity occurs during the laying and hatchling emergence stages.

Table 2. Comparative between linear and quadratic modeling for visiting, predation, and excavation records. The Akaike Information Criterion (AIC) points to the same trend, the visiting linear decreases along nesting time, while excavation and predation show an activity maximum at the beginning and ending of nest life. The best results are in bold.

View Fullscreen

Nest Predation. —

During the nesting season, our observations were consistent with those of Longo et al. (2009), indicating a higher frequency of nest visitations by C. thous specimens during ovipositing. This suggests that predator presence is directly linked to the availability of egg resources. The notable absence of the armadillo predator, Euphractus sexcinctus, is concerning, as it is 1 of the most significant turtle nest predators in the region (Gandu et al. 2013). This decline, likely attributed to hunting activities, as supported by field observations, may have a broader ecological implication. The reduction of the armadillo population, an important source of protein and a component in traditional medicines (Barboza et al. 2011), could disrupt ecological processes, such as soil and water cycling, and affect the insect population (Vale et al. 2023). Additionally, this decline may also contribute to an increase in the C. thous population, since they partially compete for prey such as turtle eggs in coastal environments.

Regarding the presence of ghost crabs at the nest, it is important to consider that these crabs possess a highly developed sense of smell and taste, enabling then to detect food-derived molecules on the sand. As detritivores of organic matter, they are particularly adept at locating food sources (Lucrezi et al. 2014). Ghost crabs have been identified as significant predators of turtle eggs, especially in areas where larger crab predators are either absent or dispersed. Additionally, burrows located near turtle nests during the incubation period may serve as entry points for other predators, increasing the risk of nest infestation or predation (Ali and Ibrahim 2002).

Regarding C. thous predation and population balance in the area, it is important to note that carnivorous mammals, such as C. thous, are also predators of ghost crabs (Mendonça et al. 2010). Ghost crabs are significant predators of marine turtle nests and are considered major nest predators on many nesting beaches (Carmo et al. 2023). The reduction or absence of carnivorous mammals can lead to a notable increase in the abundance of ghost crabs, which may, in turn, result to an overall rise in turtle nests predation (Barton and Roth 2008; Marco et al. 2015).

The literature suggests that the vulnerability of nests to predation varies throughout the incubation period, with the highest risk occurring during oviposition and hatchling. A similar pattern of predation has been observed in the Caribbean region (Leighton et al. 2011). The increased risk immediately following egg laying is often attributed to detectable cues such as the scent of the oviposition fluid, the presence of the female during laying, and the disturbance of the soil (Stancyk 1995; Riley and Litzgus 2014). At the end of the incubation period, it is hypothesized that the odor associated with the hatching process, particularly the breaking of eggshells by emerging hatchlings, serves as an olfactory signal for predators (Stancyk 1995). This is especially relevant to nocturnal predators like foxes (C. thous), which rely heavily on their olfactory senses for navigation and locating food (Faria-Correa et al. 2009). Another possibility is that the sounds produced by the hatchlings as they emerge from the nest may also attract predators (Monteiro et al. 2019).

The most significant finding of this study is that the predominant behavior of C. thous foxes in the vicinity of nests is not egg predation, but rather nest visitation. The majority of the observed interaction between foxes and nests were characterized by exploratory behavior. Specifically, the foxes approached and inspected the nests, often passing by without incident or engaging in minimal excavation activities. These actions are hypothesized to represent a strategic mapping process, whereby the foxes catalog potential food resources for future exploitation.

This exploratory behavior observed in C. thous foxes bears resemblance to the spatial memory, spatial navigation, and mapping strategies well documented in other canid species. Although such cognitive capability was not specifically investigated in C. thous, related canids such as wolves exhibit sophisticated spatial memory, which is critical for effectively navigating their territory and optimizing foraging efficiency (Gurarie et al. 2022). These cognitive skills not only facilitate the effective location of prey but also aid in avoiding potential threats and revisiting resource-abundant areas (Fujita et al. 2012). Similarly, domesticated dogs have shown the ability to recall the location of hidden food and to utilize human-provided cues to locate food within spatial contexts (Vetter et al. 2023). Such documented cognitive capabilities likely confer an evolutionary advantage and may also be present in the C. thous. As these traits are deeply rooted in the Canidae family, they have been retained in domesticated dogs originating from their wild ancestors (Pravosudov and Roth 2013).

The behavior of C. thous foxes in mapping nest locations may be interpreted as an example of opportunistic prey switching, particularly during the peak of the turtle nesting season. While this peak occurred from October to December in southern Brazil, the exact timing in northeast Brazil remains uncertain (Faria-Correa et al. 2009). During periods of resource abundance, such as the studied period, a satiated fox might prioritize identifying and memorizing food locations over immediate consumption. This behavior likely contributes to the formation of a cognitive map of potential food resources, which could prove essential for survival in seasons of scarcity. An alternative explanation, however, warrants consideration: C. thous individuals may frequent beaches for other available food sources and are attracted by sea turtle nests by olfactory cues or curiosity about human disturbance in the area, without the intention of engaging in mapping behavior.

Treatments: Cost and Efficacy. —

At first glance, it is essential to consider the multifaceted nature of nest protection strategies against predation. The methodologies employed in this study were evaluated not only for their efficacy but also with regard to their practicality and cost effectiveness. A comprehensive analysis was conducted to balance the benefits of these strategies against their potential limitations.

The numbered stake treatment was used to verify whether the methodology, used by the TAMAR Project, facilitates the predator‘s action in finding the nest. Bear in mind that in the numbered and stake treatment it is necessary to dig into the egg chamber to place a nylon string. In this context, the control treatment was characterized by no human action in the nest area. We showed that the usual numbered stake treatment had no influence on the C. thous fox–nest interaction, since the control treatment had visiting, excavation, and predation rates similar to the numbered stake treatment.

To access the economic viability of the treatments, we calculated the costs associated with each method. Among the treatments analyzed, the most cost-efficient approach to minimize C. thous predation was the use of flags, with an average cost of $4.0 per nest. In contrast, the highest cost was associated with the application of pepper powder, averaging $10.10 per nest, due to the requirement of approximately 150 g of powder applied 4 times per nest. The use of mesh had an intermediate cost of $6.80 per nest. Although these differences in cost may not appear substantial, they are noteworthy within the context of Brazil’s limited budget for conservation research. It is important to highlight that, beyond the materials required for each treatment, additional items were necessary during the application process. For instance, the use of pepper powder necessitated personal protective equipment, including disposable gloves, masks, and protective glasses, to safeguard against airborne pepper particles that could irritate the eyes, mouth, and nose of the individuals applying the treatment. Furthermore, other materials (tape measure and measuring tape) were also essential to take measurements for applying the pepper powder and the control treatments.

Beyond their effectiveness and cost, it is crucial to evaluate the technical challenges associated with each treatment method. Pepper powder, while effective in deterring mammalian predators such as foxes, is the least effective overall and the costliest, and highly susceptible to environmental conditions. Its efficacy diminishes when it is washed away by rain or tides, and application is not feasible during rainfall. Additionally, wind complicates its use by dispersing pepper particles into the air, potentially causing discomfort to personnel applying the treatment. An interesting ecological consideration arises from the use of pepper powder as a deterrent. The active compound, capsaicin, activates TRPV1 receptors in mammals, producing a deterrent effect (Koivisto et al. 2022). However, TRPV1 receptors in reptiles, such as marine turtles, are present but not necessarily sensitive to capsaicin (Koivisto et al. 2022; Seebacher and Murray 2007). This suggests that emerging hatchlings are unlikely to be affected by the presence of pepper powder. Although mammals respond to capsaicin with aversive behaviors due to TRPV1 activation, marine turtles and other reptiles may not exhibit the same sensitivity. This distinction in receptor functionality underscores the ecological safety of pepper powder as a predator deterrent in marine turtle conservation efforts. By effectively deterring mammalian predators without posing a risk to hatchlings, pepper powder offers a targeted approach to reducing nest predation. However, its limitations due to environmental factors and application challenges must also be carefully considered in the broader context of conservation strategies (Seebacher and Murray 2007; Koivisto et al. 2022).

Although more expensive, the mesh method proves effective in protecting nests from C. thous fox predation. Importantly, the mesh size allows the passage of the hatchlings. An additional advantage of this method is its reusability for protecting subsequent nests. However, the installation process is labor-intensive and time-consuming. We recommend the use of 30-cm stakes to secure the mesh. These stakes should be coated with a synthetic material, such as varnish or boat hull paint, to prevent wood decay caused by sand moisture and to facilitate their reuse. During this study, we mistakenly utilized galvanized iron mesh, which may interfere with the earth’s magnetic field, a critical signal used by hatchling sea turtles to imprint on their natal beach and navigate (Irwin et al. 2004). Future projects should use either nonmagnetic metals such as aluminum or plastic mesh, which have been used in other studies (e.g., Lei and Booth 2018). These alternatives not only mitigate the potential disruption of magnetic signal, but also align with best practices in marine turtle conservation.

The flag method, while the least expensive among the effective treatments, is limited to a single use due to its susceptibility to degradation. The fabric of the flag often tears, and the stake tends to rot at the buried end due to sand humidity by the end of the nest incubation period. This degradation can be mitigated by applying a synthetic coating to the stake. Nonetheless, the efficacy of the flag method is heavily reliant on consistent wind to keep the flag in motion, as this movement is essential for deterring predators. In the absence of sufficient wind, the effectiveness of this method may be significantly reduced, highlighting a critical limitation that must be considered when implementing this treatment in varying environmental conditions.

In summary, the use of mesh and flag treatments emerged as the most effective strategies in deterring C. thous from engaging in predation activities. However, the limited number of nests monitored during our study’s nesting season underscore the necessity of extending testing of these methods across additional reproductive seasons to validate their effectiveness. Furthermore, giving the cognitive adaptability of canids, it is essential to assess the potential for these predators to develop counterstrategies against the implemented deterrents over time. Despite these considerations, we propose that the combined use of mesh and flag treatments may enhance protection against fox predation. Specifically, the strategic deployment of flags near the expected time of nest hatching could add another layer of deterrence by creating visual and auditory stimuli that may obscure olfactory cues associated with hatchling emergence. This integrated approach may increase the overall efficacy of predator management strategies in marine turtle conservation.