Importance of Sand Particle Size and Temperature for Nesting Success of Green Turtles in Penang Island, Malaysia
Abstract
The microhabitats of green turtle (Chelonia mydas) nests were investigated to identify key factors influencing nesting success by monitoring both successful nests (n = 43) and aborted nests (n = 106) created by the same individuals (n = 9) from September 2013 to September 2014 on Penang Island, Malaysia. The effect of sand particle size on nesting success was significant, suggesting that green turtles tend to abort nesting at sites with sands of particle sizes < 1 mm. In addition, nests were successful at superficial sand temperatures less than 32.95°C.
Sea turtles are marine-adapted reptiles that spend most of their lives at sea, and they come onto terrestrial habitats mainly for nesting. The coastal nesting beaches that sea turtles use for nesting tend to be impacted by various factors that include anthropogenic coastal development and nest poaching (Lutcavage et al. 1997). The success of nesting activities and survivability of nests are directly related to reproductive fitness (Pike 2008b), because most sea turtle species are listed as vulnerable, endangered, or critically endangered on the IUCN Red List (International Union for Conservation of Nature 2016). Therefore, the conservation and management of suitable nesting habitats for sea turtles are required.
Nest site preference has been investigated to understand the suitable nesting habitats for sea turtles. Previous studies indicate that nest site preference corresponds with various environmental factors, such as distance from the high-tide line (López-Castro et al. 2004; Zavaleta-Lizárraga and Morales-Mávil 2013), beach slope (Horrocks and Scott 1991; Wood and Bjorndal 2000; Cuevas et al. 2010), beach elevation (Zare et al. 2012), existence of vegetation (Wang and Cheng 1999; Turkozan et al. 2011), moisture (Bustard and Greenham 1968; Yalçin-Özdilek et al. 2007), and sand particle size (Mortimer 1990). The complexity in the factors influencing nest site preference is due to variation among regions (Liles et al. 2015), species (Cuevas et al. 2010; Turkozan et al. 2011), and individuals (Kamel and Mrosovsky 2005, 2006).
The physical, chemical, and anthropogenic factors that are key to nesting success have been explored from the perspective of the establishment of a strategy for conservation management. The simplest method is to evaluate the habitats of successfully deposited nests (e.g., López-Castro et al. 2004; Turkozan et al. 2011; Zare et al. 2012; Zavaleta-Lizárraga and Morales-Mávil 2013). On the other hand, because sea turtles emerging on a beach sometimes return to sea without successful nesting (called “false crawls”) or excavate several times before successful nesting (Weishampel et al. 2003; Nishizawa et al. 2013; Olgun et al. 2016), the ratio of nesting success or digging success can be a measure of nest site preference (e.g., Wang and Cheng 1999). Principal component analysis, a type of multivariate analysis, can be applied to multivariate data from beach environments to extract key features determining nesting success (Chen et al. 2007; Pike 2008a), but it is sometimes difficult to interpret the results.
This study focused on the microhabitats of both successfully deposited nests and aborted nests where abandoned digging attempts were observed. This study aimed to find key factors that influence the nesting success of green turtles (Chelonia mydas) on Penang Island, Malaysia. The locations of nests on Penang Island in relation to vegetation and the distance from the high-tide line were evaluated by Sarahaizad et al. (2012), but the inclusion of the analysis of sand temperature and sand particle size in this study more accurately provides insights into key factors influencing nesting success and nest site preferences.
Methods
Kerachut (lat 5°27′4″N, long 100°10′58″E) and Teluk Kampi (lat 5°26′20″N, long 100°10′46″E) are beaches located on the northwest coast of Penang Island, Peninsular Malaysia (Fig. 1). Both beaches are managed as the Penang National Park of the Federal Government of Malaysia. These beaches have been utilized mainly by green turtles for nesting (Sarahaizad et al. 2012). The lengths of the Kerachut and Teluk Kampi beaches are 558 and 810 m, respectively.
Citation: Chelonian Conservation and Biology 17, 1; 10.2744/CCB-1266.1
Nocturnal surveys were conducted from as early as 2100–0500 hrs every night for 13 mo from 1 September 2013 to 30 September 2014. The staff of the Kerachut Turtle Conservation Centre, licensed fisherman, and interns from the Centre for Marine and Coastal Studies patrolled the beach approximately 4 times every night. When the patrolling team encountered an emerging turtle, torchlights were immediately switched off, and the nesting activity of the female turtle was monitored from a distance of approximately 5.0 m. When the female turtle started to return to the sea after oviposition, the turtle was tagged using 2 Inconel tags (style 681; National Band & Tag Company, Newport, KY), which were supplied by the Department of Fisheries Malaysia. The tags were clipped between the proximal second and third scales of both the right and the left front flippers (Broderick and Godley 1999) using an applicator. The curved carapace length (CCL) and curved carapace width (CCW) were measured using a 2.0-m measuring tape (± 0.1 cm).
To evaluate the microhabitats of successful nest sites and aborted nest sites, the superficial sand temperature, distance from the high-tide line, and vegetation type were determined. The temperature was recorded using a soil thermometer (± 0.1°C; Corkscrew Stem Thermometer TM40, Extech Instruments, Nashua, NH) at 5.0-cm depth, 1.0 m from the nest site, and at the same distance above the high-tide line (López-Castro et al. 2004). The distance between the high-tide line and the nest site was measured using a 30.0-m measuring tape (± 0.1 m; López-Castro et al. 2004). The vegetation type at the nest site was determined and categorized into one of three types: woody vegetation, grassy vegetation, and open sand with no vegetation. In addition, sand samples were collected from successful and aborted nest sites visited by the same sea turtle, with samples taken from 1 aborted nest site per successful nest site (except in the case in which there were no aborted nests) to determine the sand particle size. Approximately 200.0 g of sand was collected at a 30.0-cm depth and 1.0 m from the nest site and at the same distance above the high-tide line.
The collected sand samples were oven dried for 24 hrs at 105°C (Foley et al. 2006). The sand was then separated into size classes by sieving the sand samples through the following series of sieves: 2.0, 1.0, 0.425, 0.250, 0.125, and 0.063 mm. The separated sand samples were weighted using a weighing balance (± 0.1 g; APTK461 digital weighting scale, Ban Hing Holding Sdn Bhd, Kuala Lumpur, Malaysia), and the percentage based on weight was calculated. As high variation was observed in the percentages of sand particle sizes of 0.425–1.0 mm and 1.0–2.0 mm, the sand particle sizes at each nesting site were considered as one variable, the percentage of sand particles ≥ 1 mm, which included the sand particles of 1.0–2.0 mm and ≥ 2.0 mm, as presented in the following analyses.
Statistical analyses were performed using R version 3.2.4 (R Core Team 2016). Success in nesting was treated as a binomial response variable. Sand particle size, temperature, distance from the high-tide line, and vegetation type were included as fixed effects. The effects of these variables on nesting success were evaluated. First, individual turtles were included as a random effect, and a generalized linear mixed model (GLMM) with a binomial error distribution and logit link function was constructed using the Laplace approximation in the lme4 package (Bates et al. 2015). Because the random effect was ignorable (see “Results”), it was excluded from the model. A generalized linear model (GLM) with a binomial error distribution and logit link function (i.e., logistic regression) was constructed. Stepwise model selection was performed based on the Akaike information criterion value. The significance of the explanatory variables in the selected model was tested using the likelihood ratio test (LRT).
A decision tree was constructed using the rpart package (Therneau et al. 2015). Successful nesting was the response variable, while sand particle size, temperature, distance from the high-tide line, and vegetation type were explanatory variables. The number of cross-validations and the threshold complex parameter were set at 10 and 0.01, respectively, as default values (Therneau et al. 2015). Observations with missing explanatory variable data were analyzed using surrogate variables (Therneau et al. 2015). The constructed decision tree was visualized using the partykit package (Hothorn and Zeileis 2015).
Results
In the surveys, 43 successful nests produced by 9 individuals (average ± SD of CCL: 102.4 ± 3.9 cm; CCW: 92.6 ± 4.0 cm) were identified. In addition to these successful nests, 106 aborted nests were identified (Table 1). Sand temperature was not measured for 7 of the 43 successful nests (6 nests in woody vegetation and 1 nest in open sand) because of bad weather conditions. However, the sand temperature was measured at all other successful and aborted nests (range = 25.2°–35.3°C; Table 1; Fig. 2). Higher numbers of successful nests were observed in woody vegetation (19 nests) and open sand (18 nests) compared with grassy vegetation (6 nests), but there were more aborted nests in the former 2 vegetation types (37 and 62, respectively; Fig. 2). Sand samples were collected at 42 of 106 aborted nests (Table 1).
Citation: Chelonian Conservation and Biology 17, 1; 10.2744/CCB-1266.1
In total, full data sets were obtained at 36 successful nests and 42 aborted nests (78 in total), and the GLMM and GLM analyses of nesting success were applied to these data sets. The GLMM indicated that the variance derived from individual differences (i.e., the random effect) was 0. Stepwise model selection in the GLM analysis resulted in a logistic regression model containing only sand particle size as an explanatory variable (final step is shown in Table 2). The effect of sand particle size on nesting success was significant (LRT = 81.36, p < 0.01). The selected model indicated a threshold percentage of sand particle sizes ≥ 1 mm of 50.4%, representing the point at which 50% of nests were successful (Fig. 3).
Citation: Chelonian Conservation and Biology 17, 1; 10.2744/CCB-1266.1
A decision tree was constructed based on the 149 observations of successful and aborted nests (Fig. 4). Sand particle size appeared at the first node, indicating that 50.3% of sand particle sizes ≥ 1 mm contributed to nesting success. The threshold value of 50.3% was almost the same as that estimated from the logistic regression model (50.4%). In addition to the dominance of small sand particles, sand temperatures ≥ 32.95°C resulted in aborted nests.
Citation: Chelonian Conservation and Biology 17, 1; 10.2744/CCB-1266.1
Discussion
By investigating both successful and aborted nests, this study identified key features determining nesting success. Previous observations of green turtles on Penang Island indicated that the majority of nests were located 10–40 m from the high-tide line or woody vegetation (Sarahaizad et al. 2012). This study showed a larger number of nests in woody vegetation than in grassy areas, but there were also more aborted nests in woody vegetation. In addition, while no successful nests were observed < 10 m from the high-tide line, successful nests were observed at > 40 m unless the sand temperature was high. Therefore, evaluating the habitats of only successfully deposited nests is not sufficient to identify key features determining nesting success.
One of the key features identified in this study was sand particle size. This key feature causes green turtles to abort nesting at sites where sands of particle sizes < 1 mm dominate. This is contrary to the expectation because Mortimer (1990) reported that the inhibition of green turtle digging and reduction of hatching success on Ascension Island are owing to large sand particle sizes. The negative effects of large sand particle size may be caused by desiccation (Mortimer 1990) or the high compactness of sand (Chen et al. 2007). However, large particle sizes may instead be preferable in terms of gas exchange between the nests and surrounding sand (Ackerman 1980) and resistance to slippage of the egg chamber wall, as indicated by the higher friction angle and angle of repose (Vangla and Latha 2015). The sand particles of relatively larger size on Penang Island may not be coarse enough to negatively impact digging and hatching, as indicated by Mortimer (1990). In this study, the sites where successful or aborted nests occurred were dominated by sand particles of 1–2 mm or 0.425–1.0 mm, respectively, which were widely observed at beaches where green turtles nest (Hirth 1971; Stancyk and Ross 1978; Mortimer 1990; Chen et al. 2007).
Differences in the effects of sand particle size on nesting success could also be attributable to regional differences in beach origin or climate. Differences in beach origin could result in differences in sand particle shape and texture that influence sand desiccation, sand compactness, and water and gas exchange in eggs (Kam and Ackerman 1990; Mortimer 1990). In fact, sand texture and organic content have been reported to be factors determining nesting success in loggerhead turtles (Caretta caretta; Mazaris et al. 2006). Mortimer (1990) found differences in clutch mortality in green turtles between biogenic and volcanic beaches and negative effects of coarse sand at biogenic beaches. In addition, desiccation is more severe on Ascension Island owing to the small mean annual precipitation (194 mm; Mortimer 1990). On Penang Island, granitic soils are dominant, and the mean annual precipitation is more than 2000 mm (Ahmad et al. 2006; Pradhan and Lee 2010). Therefore, it is reasonable to assume that the effects of large particle size on nesting success are not equivalent to those reported by Mortimer (1990). Regional differences in the effects of sand particle size are also supported by findings of the higher hatching success of sea turtles in coarser sand at some beaches (Fadini et al. 2011), while negative (Mortimer 1990) or insignificant (Foley et al. 2006; Yalçin-Özdilek et al. 2007) effects of coarse sand on hatching success have also been reported.
In addition to sand particle size, the decision tree indicated that sand temperature is also an important factor influencing nesting success. Successful nests were created in superficial sand at temperatures less than 32.95°C. The preference for establishing nesting sites where the superficial temperature is approximately 32°C was found in olive ridley turtles (Lepidochelys olivacea; López-Castro et al. 2004). This preference might prevent drastic changes in temperature for the eggs of sea turtles, whose body temperature generally ranges from 25°C to 33°C (Mrosovsky and Pritchard 1971; Spotila and Standora 1985). All successful nests in this study were observed in this thermal range of sand temperature. Nest site selection based on temperature may also contribute to the avoidance of the excessive heating of eggs, especially because of large sand particle size, during embryo development, which has a negative effect on survivorship and the sex ratio (Fisher et al. 2014; Howard et al. 2014).
Interindividual differences are a factor that should be considered when evaluating nest site preferences (Kamel and Mrosovsky 2005, 2006). However, the GLMM indicated no significant interindividual differences in nesting success among the green turtles nesting on Penang Island. This result may partly be due to the small sample size. The relatively small sample size, owing to the relatively small number of nesting females, limits statistical power and the inclusion of additional candidate variables, such as moisture, precipitation, salinity, beach elevation, and anthropogenic factors (e.g., as investigated in Wood and Bjorndal 2000; Foley et al. 2006). In addition, while this study focused on the various microhabitats on a beach, the selection of nesting beaches should also be addressed because sea turtles use multiple environmental cues in the different steps of the nesting process (Mazaris et al. 2006; Kaska et al. 2010). Further continuous investigations, including studies focusing on the effects of microhabitats on hatching success, will contribute to the comprehensive understanding of nest site selection.
In conclusion, this study identified key features influencing nesting success and estimated their threshold values. Although green turtles nest on beaches of various sand particle size compositions (Hirth 1971; Stancyk and Ross 1978; Chen et al. 2007), the effect of sand particle size on nesting has not been comprehensively studied. However, in addition to sand temperature, this study provided insights into the importance of sand particle size in nesting success.
Geographic location of Penang Island and beaches.
Nesting success (successful: hollow circles; aborted: filled circles) in relation to superficial sand temperature, distance from high-tide line, and vegetation type. Seven successful nests (6 in woody vegetation and 1 in open sand), where sand temperature was not measured, are not shown.
Logistic regression line for the relationship between nesting success (1 means successful, while 0 means aborted) and sand particle size presented as the percentage of sand particle size ≥ 1 mm. Horizontal and vertical dashed lines indicate a 50% probability of nesting success and the corresponding percentage of sand with a particle size ≥ 1 mm, respectively.
Decision tree for nesting success. Ratio of successful to aborted nests is illustrated for each leaf.
Contributor Notes
Handling Editor: Jeffrey A. Seminoff
