Determining Nest Site Selection by Hawksbill Sea Turtles in the Persian Gulf Using Unmanned Aerial Vehicles
Abstract
Nest site selection is an important aspect of reproductive behavior in sea turtles, and it is influenced by biotic and abiotic beach characteristics. To date, the association between nesting beach characteristics and nest site selection has been explored manually at small scales; yet, hawksbill turtles (Eretmochelys imbricata) nest on a variety of beaches, with knowledge remaining limited on preferred habitat. Here, unmanned aerial vehicles were used to assess all available beaches, identify common characteristics, and determine the preferred nesting sites for hawksbill turtles in the Persian Gulf. High resolution photographs were taken of 236 nests with unmanned aerial vehicles on 7 main islands supporting 10 nesting beaches. The physical characteristics of these nests were photographed (e.g., elevation, slope, vegetation cover, and nest distance from vegetation and the high tide line). Slope was the most important factor regulating nesting activity. Most turtles nested on shallow-sloped beaches with 5° slopes or less (61.4%), at a distance of ≤ 15 m from the high tide line (71.8%), and generally ranged from 1 to 3 m (83.9%) above sea level. Most nests (87.7%) were laid close to the vegetation line (≤ 5 m) in sandy zones. Although there was high geomorphological diversity in landform among the islands, nesting habitat characteristics of hawksbills were consistent. Unmanned aerial vehicles provide a useful approach for identifying the nesting habitat preferences of sea turtles in a time- and cost-efficient manner, covering more surface while minimally disturbing fragile habitat.
Hawksbill sea turtles (Eretmochelys imbricata; Linnaeus 1766) are 1 of the 7 extant species of sea turtles and have been classified as Critically Endangered in the International Union for Conservation of Nature (IUCN) Red List due to long-term overexploitation (IUCN 2020). Similar to other sea turtles, hawksbills spend most of their lives in water, yet they rely on sandy beaches to lay their eggs. Female turtles migrate to nesting beaches after mating and excavate their nests in the sand (Lutz et al. 2002). Nest site selection is an important aspect of reproduction because the surrounding environment has a significant impact on incubation (Wood and Bjorndal 2000; Miller et al. 2003; Caut et al. 2006; Serafini et al. 2009) and offspring survival (Ackerman 1981; Spencer 2002). If nest sites are poorly selected in species without parental care (i.e., choosing a place with inappropriate conditions in terms of temperature, population density, and vegetation), the outcomes can be detrimental (Madsen and Shine 1999; Halloy and Fiaño 2000).
The physical features of the nesting habitat, such as elevation, distance from the high tide line, slope, vegetation coverage, and distance from vegetation, can affect the nest site selection behavior for sea turtles species (Horrocks and Scott 1991; Wood and Bjorndal 2000; Hernández-Cortés et al. 2018; Patrício et al. 2018; Culver et al. 2020). Sea turtle species exhibit different preferences for nesting beach characteristics, probably due to differences in size, weight, and behavior (Culver et al. 2020). Preferences are likely tied to factors that increase nest/ hatchling survival in some way.
Hawksbill turtles tend to nest near or within littoral vegetation far away from the high tide line (Horrocks and Scott 1991; Lewsey et al. 2004; Kamel and Mrosovsky 2006). Ditmer and Stapleton (2012) reported significant relationships between vegetation and hatch success. Another study showed that nesting in vegetated areas increases emergence success (Kamel and Mrosovsky 2005), perhaps due to the fact that vegetated areas improve beach stability and promote a more predictable temperature regime. Vegetation coverage may lower the nest temperature and balance the metabolic heat generated during embryonic development (Kamel 2013). Some studies have reported that hawksbills use slope as a cue for beach selection (Horrocks and Scott 1991; Wood and Bjorndal 2000); in some regions, they tend to nest on beaches with steep slopes and low wave energy. A steep slope reduces the time and energy spent traveling between the nest and the water, thereby reducing predation risk and energy expenditure for gravid females and hatchlings (Horrocks and Scott 1991).
According to the findings of some researchers, turtles may try to select the best compromise between multiple factors. Beach topography can cause variation in the preference for steeper beaches; for example, if steeper beaches have high rock cover, turtles may prefer to lay their eggs on lower-slope locations with less rock cover (Ficetola 2007). The elevation of nests is an important aspect of nest site selection in hawksbills. Also, it seems nests further inland may have higher success, but that depends on the other variables that influence nest site selection such as vegetation cover, dune height, and other habitat factors. Due to the reduced likelihood of flooding, these regions tend to have a greater hatch success rate (Horrocks and Scott 1991).
In the Persian Gulf, hawksbill turtles nest along the coasts of Iran, Qatar, Saudi Arabia, Kuwait, and the United Arab Emirates (Meylan and Donnelly 1999; Pilcher et al. 2014). Some of the most important populations are found in Iran, and the species nests on several islands (Zare et al. 2012; Askari Hesni et al. 2016; Pazira et al. 2016). There are few studies on hawksbill turtle nest site selection in the Persian Gulf. Askari Hesni et al. (2019) monitored nesting density on four hawksbill nesting sites (Kharko, Nakhiloo, Nayband Bay, and Ommolkaram) in the northern Persian Gulf. Zare et al. (2012) studied hawksbill nesting success and nest site characteristics (e.g., elevation, slope, distance from high tide line, temperature, vegetation cover, and sand particle size) on Shidvar Island, finding that the maximum nesting success (∼ 80%) occurred at sites 2 m or less above the high tide line.
Unmanned aerial vehicles (UAVs) have increasingly been used for wildlife monitoring, especially for marine animals in recent years (Iv et al. 2006; Koski et al. 2009; Hodgson et al. 2013). This is due in large part to the fact that they are a more efficient and cost-effective alternative to manned aerial surveys (Koski et al. 2009; Koh and Wich 2012; Hodgson et al. 2013). These systems typically have airframes equipped with cameras and sensors that collect environmental data. Standard red, green, blue (RGB) photos and videos are often collected for behavioral sampling and habitat assessments (Puttock et al. 2015), as well as 3-dimensional (3D) topographical reconstruction using the “structure from motion” (SfM) method (Mancini et al. 2013). For sea turtles, UAVs have the potential to complement and enhance many existing research and conservation projects, particularly on nesting beaches (Bevan et al. 2016; Rees et al. 2018; Schofield et al. 2019). In recent years, researchers have used UAVs to estimate the nesting population of adult female green sea turtles (Chelonia mydas) using the mark–resight approach (Dunstan et al. 2020) and also to calculate offspring and breeding (operational) sex ratios to achieve mating dynamics in loggerhead sea turtles (Caretta caretta) (Schofield et al. 2017). UAVs can also be used to map and profile major turtle breeding and foraging sites with greater efficiency and resolution (Dunstan et al. 2020). For example, Culver et al. (2020) used lidar data to assess the relationship between beach geomorphology and Kemp's ridley (Lepidochelys kempii) nest site selection. In another UAV study, olive ridley (Lepidochelys olivacea) nesting beach surveys were surveyed in Sinaloa, Mexico, to determine the percentage of vegetation cover using orthomosaics, and digital elevation models were created to estimate physical characteristics connected to nesting sites (Flores et al. 2020).
There has been little research on the influence of the physical characteristics of nesting beaches on hawksbills' nest site selection in the Persian Gulf, and no large-scale comparison of these variables has been made among islands in the region. Therefore, the drivers behind nest site selection remain understudied in the Persian Gulf, and more scientific studies are needed to inform conservation and management decisions. Hawksbills nesting in the Persian Gulf occur in areas with 3 types of protection status (marine national parks, wildlife sanctuaries, and seasonal protection areas) and are an important flagship species for local conservation efforts. However, there is high diversity in the shape and structure of the Persian Gulf islands (flat sand, sand and rock, sand with flat vegetation, and sand with hilly vegetation), and little is known about the specific habitat features that are linked with nesting by this critically endangered species.
The aim of this study was to investigate the physical characteristics of the most important hawksbill nesting beaches in the Persian Gulf using UAVs and to determine the importance of several environmental factors in the selection of oviposition sites. UAV photos were used to measure the physical properties of hawksbill nest sites and compare elevation, slope, and vegetation for 7 regions: 6 islands (7 locations) and 1 bay (2 locations).
METHODS
Study Site. — The Persian Gulf is surrounded by deserts and is bordered by 8 rapidly developing nations. It is 1000 km long and 200–300 km wide and lies between 24°N and 30°N and 48°E and 57°E. Two small uninhabited islands (Nakhiloo and Ommolkaram) and one mainland site (Nayband Bay) are located along the coast of Bushehr Province in the northern Persian Gulf. Nakhiloo and Ommolkaram islands are part of the Dayer-Nakhiloo National Park to the south of the Mond River Delta. The sandy beach of Nayband Marine-Coastal National Park, located in Nayband Bay in the north, was the only mainland beach surveyed in this study. All these areas are designated as marine protected areas by the Iranian Department of Environment.
Four other islands (Qeshm, Kish, Lavan, and Shidvar) are located along the coast of Hormozgan Province. Qeshm island, the largest island in the Persian Gulf, has an area of about 1491 km2. Kish Island has an area of 90 km2 and its coast has been strongly affected by development. The coasts of Kish and Qeshm islands are under severe ecotourism pressure. Lavan Island is the third-largest Iranian island after Kish Island, with an area of about 81 km2. Because of its oil deposits, this location has been exploited by oil companies. Shidvar Island, located east of Lavan, is a small, uninhabited island with a surface area of 1 km2; it has international importance for wildlife and coral reefs (Fig. 1).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 21, 2; 10.2744/CCB-1552.1
Data Set. — We used UAVs to study the beaches and determine distance from the vegetation line, vegetation coverage, slope, nest surface elevation, and nest distance from the high tide line. Monitoring efforts using UAVs occurred during the peak (7–16 May) of the 2020 nesting season. Airborne raw photos (20-megapixel RGB photo in JPG format, 3000 × 3000 pixels) were collected with a multirotor UAV Phantom 4 PRO with a focal length of 8.8 mm. Automatic camera settings were used (ISO: 100, mechanical shutter speed: 8-1/2000). The beaches were photographed between 1000 and 1400 hrs, and the average speed of the flights was 3.9 m/sec. In some cases, more than one beach was photographed on the same day. Survey details for each nesting beach are listed in Table 1.
A manual image quality assessment was first used to filter usable images. Images with obvious artifacts and high levels of blurring, or images taken too near to the ground, were not processed (Brovkina et al. 2018). Acceptable UAV images were processed using Agisoft Metashape Professional v1.6.3 to generate orthomosaics (Fig. 2A) and Digital Elevation Model (DEM) (to 5-cm precision) using the structure from motion (SfM) technique. The SfM technique generates 3D point clouds from overlapping 2D images. It works by matching key points in one set of images to the same points in another set of images over the same area (Otero et al. 2018). The relevant parameters used for photogrammetric reconstruction are summarized in Table 2.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 21, 2; 10.2744/CCB-1552.1
Physical Characteristics Extraction. — Field data and aerial images were cross-checked and used to determine the distribution of nests. The nest data were converted to a shapefile of points, which were used to create a database containing vegetation coverage and raster topography data sets for each nest.
The DEMs were used to obtain the topographic variables of slope in degrees with the Terrain Analysis tools in ArcMap 10.5 software. DEM and slope maps were then classified into equal intervals (0.5 m, 5°) (Flores et al. 2020). Depending on when the aerial photos were taken, tides could have been at different levels, so we used a Real-Time Kinematic-Global Positioning System (RTK-GPS) in the field to calibrate the orthomosaic products with high tide line elevation as a constant. Finally, by overlapping the nest points on these maps, we created maps of the elevation (m) and slope (degree) of the nest locations to determine the two mentioned features (Fig. 2B–C).
The orthomosaics were used to visually classify the percentage of vegetation coverage within a 3-m radius around the location of each nest. To begin, a 3-m buffer was established around the nests, 1 m (radius of the nest) + 2 m (edge of the vegetation zone) (Serafini et al. 2009; Flores et al. 2020). For this purpose, we had a field survey to collection the data needed. About 10 normal nests were selected for trait calculations. Then, the boundary of the laying area was determined from the high tide line to the edge of the 3-m radius of the focal nest. The laying area referred to the larger polygon encompassing all nests and their 3-m radii. Vegetation coverage areas were calculated in square meters after creating the polygons around the plants at the defined boundaries (Fig. 2D). We overlaid nest sites with orthomosaic images of each island to determine distance from the high tide line (Fig. 2E) and distance from vegetation.
Principle Components Analysis (PCA). — Multivariate statistics were used to analyze the percent contribution of the main physical factors. We used PCA with PAST 4.09 from the University of Oslo Natural History Museum (Hammer et al. 2001) to investigate the structure of the habitat variables data set and to determine if and how the variables were correlated. Variables were normalized by calculating the Z scores of all observations, resulting in equal means and variances for all variables. A correlation matrix was then used to extract the principal components (PCs) with eigenvalues > 1, and a Varimax rotation was employed to help the interpretation of the resulting PCs. Finally, the outcomes of these analyses were compared across all sites for 236 nest points.
RESULTS
Physical Characteristics. — Hawksbill nesting activity was located between 0.88 to 5.68 m above mean sea level. Ommolkaram Island had the most restricted elevation range (between 1.11 and 1.85 m). The widest elevation range was measured at Qeshm Island, 1.64 to 5.68 m (Fig. 3A).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 21, 2; 10.2744/CCB-1552.1
The slope range was generally between 0° and 15°, and most hawksbill nests were on slopes ≤ 5°. On Nakhiloo Island, where all nests were detected on slopes between 0° and 10°, the most restricted slope was for nesting (2.50° ± 1.81°). Because nesting activity was reported on all slopes on Ommolkaram Island, it was more diverse than other islands (12.52° ± 6.37°) (Fig. 3B).
Some islands have a rocky or hilly landscape with narrow and limited beaches. The minimum range of distances from the tide line was related to Lavan Island, Shidvar Island, and Nayband Bay (5–20 m). The maximum range was observed on Nakhiloo Island (5–35 m) and Qeshm Island (5–45 m). The results showed that most (71.8%) of the nests were laid at a distance of ≤ 15 m from the high tide line. About 28.2% of the nests were located at distances from high tide line further than 15 m (Fig. 3C).
Nesting activity was observed in both sandy and vegetation zones, but was more frequently encountered in sandy areas; only 12.3% of nesting activity occurred in the vegetation zone. Most of the nests (87.7%) were laid in sandy zones, but within 5 m of the vegetation zone border (Fig. 3D).
Kish Island had the largest nesting surface area for hawksbill turtles, whereas Ommolkaram Island had the smallest. Except for Lavan Island, which had no vegetation, vegetation covered less than 10% (5.94% ± 2.38%) of the total nesting area at all study sites. Nayband Bay (2.1%) and Ommolkaram Island (9.5%) had the lowest and highest vegetation cover, respectively (Fig. 4).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 21, 2; 10.2744/CCB-1552.1
PCA Analysis. — The resulting PCA axes were rotated using a Varimax rotation. The first 3 components were retained, and these explained 98.37% of the variation (PCA1: 60.45%; PCA2: 23.70%; PCA3: 12.22%). PCA1 was highly correlated with slope (beach angle) and negatively correlated with distance from the tide line, PCA2 was highly correlated with all parameters, PCA3 did not show a high correlation with environmental factors, and PCA4 was highly correlated with elevation. The first and second components are shown for the 6 nesting areas (Fig. 5).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 21, 2; 10.2744/CCB-1552.1
DISCUSSION
Our findings indicate that hawksbill nesting activity decreases as beach slope increases. Most of the nests were recorded on a slope of 0.15° to ∼ 15°. However, nesting was also seen on islands with steeper slopes < 35°. Green turtles prefer beaches with slightly steeper slopes, primarily in the berm zone, as opposed to hawksbill turtles that have a tendency to choose a wider variety of beach physical features (Cuevas et al. 2010). While the lower slope was preferred in this study, hawksbill nesting has been observed on islands with varying slopes.
Observations showed that 85.6% of hawksbill turtles in the Persian Gulf excavate nests at an elevation of 3 m or less (83.9% nests at 1–3 m). This compares also with the results found by Zare et al. (2012), who indicated that the majority (80%) of the nests on Shidvar Island were deposited at a height of 2 m or less above the high tide line. Overall, elevation seems to be an important factor in nest site selection for hawksbills in the Persian Gulf. Laying at a higher elevation was seen in Qeshm Island and Nayband Bay (Fig. 3A). These two areas have the highest recorded tidal fluctuations among the Persian Gulf. Also, some nests in Qeshm Island were recorded at the farthest points from the high tide line (Fig. 3C). Thus, nest site selection appears to correspond with the maximum high tide line and beach elevation. Water flowed fastest between the Lavan and Shidvar islands, where turtles nested at higher slopes, possibly reducing the risk of inundation (Fig. 3B). On Nakhiloo Island, nests were further away from the vegetation line (Fig. 3D). This phenomenon might be driven by the high territoriality of 3 tern (Sternidae) species that nest in the vegetation on this island, with turtles perhaps avoiding the dense colonies of birds. Orthomosaic photographs also showed that the wide beach at Nakhiloo Island provided more nesting opportunities away from vegetation cover compared with other beaches.
Mean nest elevation in the Persian Gulf Islands ranged from 1.52 ± 0.69 m to 3.24 ± 1.40 m. Furthermore, the slope ranged from 2.50° ± 1.81° to 12.52° ± 6.37°. The variance of the distance from the high tide line was much higher than that of slope and elevation (i.e., from 7.87 ± 3.50 m to 17.08 ± 8.12 m). In this limited slope range, the elevation of nests increases with increasing the distance from the tide line. Consequently, environmental factors impacted the study areas differently to topography (landform). Factor analysis showed the different effects of 4 environmental factors. The PCA loading bars for elevation and slope were generally in different directions to the high tide line distance. The magnitude of effect of elevation and distance from vegetation seems smaller than slope and distance from high tide line (Fig. 5). This finding is consistent with Horrocks and Scott (1991), who report that hawksbills nest site selection in Barbados is more sensitive to elevation than to distance from the high tide line.
The supralittoral zones of some islands in the study region were characterized by sand dunes with herbaceous and shrubby saltbush vegetation such as Atriplex leucoclada and Arthrocnemum macrostachyum. Halophytic plants are the upper boundary of the sea turtle nesting locations on nesting beaches. It is likely that the distance from vegetation is a covariate rather than a parameter to be investigated, and that turtles and vegetation both select for these higher elevations—not that turtles select nests based on where vegetation is, or is not, found. Persian Gulf hawksbill turtles clearly avoid nesting in the vegetation zone. They generally dig nests at a distance from vegetation of 5.88 m in a sandy area close to the vegetation zone. In our study, only 14.3% of nests were laid in the vegetation zone (Fig. 3D). In comparison, preference for sand vs. vegetation zone was not detected at Arembepe Beach, northeastern Brazil (Serafini et al. 2009). Hawksbill nesting in and near beach vegetation cover was reported for Gardens of the Queen Marine Park in Cuba (Medina Cruz et al. 2010), and a tendency for nesting in association with mangrove woody vegetation has been observed in El Salvador (Liles et al. 2015).
Field inventory methods tend to differ with the site, as recording all environmental parameters of large numbers of nests is costly and time consuming. Such issues lead to the risk of error or insufficient sampling effort. Furthermore, foot traffic on nesting beaches could risk damage to nests (particularly in the final days of development). Therefore, remote sensing techniques provide a suitable alternative for enhancing sampling effort both spatially and temporally. As nest sites are extremely small (1 m2 maximum), satellite images, which have a medium spatial resolution, are not suitable. Therefore, the use of aerial drones allows for fine-scale images to be collected over large regional scales at a relatively low expense. However, one potential limitation of UAVs is that this technology often cannot be deployed in habitats close to airports, military facilities, or other access-controlled areas. Nevertheless, we believe using UAVs is a preferred choice for studying the nesting habitat of sea turtles, and we suggest that they can be used for a wide variety of sea turtle nesting beach studies, especially in important turtle areas (Pilcher et al. 2014) around the world.
Map of Persian Gulf's islands showing the location of nesting beaches of hawksbill sea turtles. (A) Nakhiloo, (B) Ommolkaram, (C) Nayband Bay, (D) Lavan, (E) Shidvar, (F) Kish, and (G) Qeshm.
Physical characteristics of hawksbill turtle nesting beach in Ommolkaram Island regarding UAV photos: (A) orthomosaic, (B) elevation, (C) slope, (D) vegetation coverage, and (E) distance from the tide line.
Distribution of hawksbill nests: (A) elevation above mean sea level (m), (B) slope, (C) distance above high tide line (m), and (D) distance from vegetation.
Comparing diversity of 7 nesting beaches: vegetation coverage (m2) and nesting area (m2).
Principal component analysis with 95% ellipses for distance from the tide line, elevation, slope, and distance from vegetation along 236 hawksbill nests for 6 locations: (A) Kish, (B) Qeshm, (C) Nakhiloo, (D) Nayband, (E) Ommolkaram, and (F) Shidvar.
Contributor Notes
Handling Editor: Jeffrey A. Seminoff
