Population monitoring is essential to assess, manage and protect threatened species. Although the South Pacific loggerhead turtle subpopulation is classified as critically endangered by the IUCN, monitoring data are scarce.

This study reports the results of the first long-term monitoring of the nesting population of loggerhead turtles held by Bwärä Tortues Marines on La Roche Percée beach, New Caledonia. From 2006 to 2020, Capture Mark Recapture was used to identify nesting individuals. Time and nesting success were recorded on site.

A total of 452 different females were observed and tagged over 14 years. The number of different nesting individuals observed each year showed a significant increase along the timeframe of the study. A remigration interval of 3.34 years was observed and the overall nesting success was 59.02%. This study also reports the inter-nesting intervals, monthly and hourly variabilities in the visits at the nesting site.

The conservation actions led by Bwärä Tortues Marines seem to be correlated with a higher nesting success. This study provides encouraging results and highlights the need to pursue the monitoring and conservation actions implemented by Bwärä Tortues Marines. Further management recommendations are also provided.

Overview

Over 14 years (2006-2020), daily patrols were implemented throughout each nesting season (from November to March). The timing and activities of the loggerhead turtles were monitored.

Study site

The study was conducted on La Roche Percée, a sandy beach in the Bay of La Roche Percée, New Caledonia (-21.612831, 165.463286). This site is included in a marine protected area registered under the natural reserve status (Decree n° 33-1993/APS 1993 and 293-99/PS, 1999). The beach is oriented South-west and is 2.5 km in length starting from a stand-up rock at the North-western end, to the mouth of the Néra river at the South-eastern end. The Néra river brings dark sediments directly to the beach of La Roche Percée resulting in darker sand compared to nearby beaches. La Roche Percée bay faces a large break in the barrier reef surrounding the area, allowing waves but also marine megafauna to enter the lagoon. At high tide, the beach is 20m (meters) wide in the middle and 100m wide at the northern and southern ends. During the hot and humid season (November to March) the beach can be submerged by water due to extreme weather events.

The only access to La Roche Percée is a road over a dam on its northern shore – blocking the northern bank of the Nera river. Vegetation near the road partially protects the beach from anthropogenic light sources (cars and houses) and there are no street lights in this area. Sand replenishment work took place in 2011 (during the time of this study but not during the nesting season) at the vegetation margin to widen the beach and counter beach erosion. The sand was taken from the Néra river but was not washed and sieved causing its compaction on some parts of the beach, resulting in the inability of turtles to dig in these areas. This compacted area was high on the beach at the margin of the vegetation and was partially covered with sand and vines (Ipomoea pescaprae).

The bay is also famous for being one of the most important surf spots in New Caledonia and is one of the main leisure sites, attracting many locals and tourists all year long. Although pets are prohibited inside the reserve following a government decree put in place in 2009, dogs are regularly seen on the beach. 

In the bay, a second beach named La Baie des Tortues (i.e. Turtle Bay, -21.60655, 165.45478, 280m in length = 1/5 of La Roche Percée) is located 100m north of La Roche Percée beach. The beaches are separated by a high rocky spit. This beach was not monitored during the night as the steep and slippery path leading to it would represent a safety hazard for the team working without light.

Monitoring and data collection

The walking patrol covered the 2.5km of the study site, walking back and forth between the two ends of the beach. Each turtle activity on the beach of La Roche Percée was recorded from November 2006 to March 2020 (14 nesting seasons) by members of Bwärä Tortues Marines. Two patrol sessions were conducted on the beach every day during the nesting seasons: (1) in the evening, usually conducted from 8 pm to 1 am (i.e. night shift), which enabled monitoring of the nesting females encountered on the beach, (2) in the morning (i.e. morning shift) starting at dawn, for 2 to 5 hours, and allowing an exhaustive count of all nesting activities throughout the night. The duration of a patrol could vary if a turtle or a track was detected by a member of Bwärä Tortues Marine before or after the usual hours. For example, if a turtle was spotted at 00:50 am the patrol team stayed on site until the female completed her nesting cycle. Sometimes, turtles crawled on the beach one after the other and the patrol team had to work until dawn. Only extreme weather conditions such as rain downpours or cyclonic alerts led to the cancellation of a patrol (on average eight patrols per year).  

During the night shift, one or two teams composed of an eco-guard and trained volunteers walked without light alongside the high-water line, searching for turtle tracks. The number of teams deployed depended on the number of volunteers available. When a track was noticed, the team stopped and remained immobile while trying to locate the individual. The team followed the track in a way which meant they were unnoticed by the turtle (e.g., crawling). Observers were able to determine the precise nesting phase by seeing or hearing the turtle. The nesting phases include: ascending the beach, making a body pit (multiple body pit attempts could be undertaken before the next stage), digging the egg chamber, laying eggs, filling the egg chamber, covering the body pit, and returning to the surf (for precise descriptions of the nesting phases see Hailman & Elowson, 1992). To avoid disturbance, the turtles were approached after the beginning of the egg-laying phase, from the back and without light. During the data collection, a red light was eventually used for a short time only and never toward the head of the turtle. If the individual was found returning to the sea, the data were gathered while the turtle stopped to breathe. Data collection included: individual identification and carapace length measure, date, time, location, and nesting phase.

The morning shift consisted mainly of recording all the activities that happened during the late part of the night. It always started at dawn to maximize the chances of meeting the last turtles of the night, to protect those individuals from potential disturbances (i.e. beach visitors or dogs), and to be sure to record all turtle activities.  A later start could have led to the loss of data due to rising tides, strong winds, or beach users who could have erased the tracks. Activities were defined using the sand cues as above. The date, time, and location of the activity were recorded. Each track (crawling tracks, body pit, and mound) was then wiped off so as not to count it a second time during the following night and to hide the nest.

Individuals were identified using Capture-Mark-Recapture (CMR). Every studied individual was tagged with a titanium tag (Titanium Turtle Tag, Stockbrands, Australia) on the trailing edge of the front flipper, in the skin between the first and the second scales adjacent to the axilla following recommendations published in Limpus (1992). During the first five seasons, most of the individuals were tagged on both left and right front flippers, following which it was reduced to one tag, placed on the left flipper, to minimize individuals' stress. Following the CMR protocol, if turtles did not have a tag, they were tagged and considered as Captures (C). Individuals already tagged were considered as Recaptures (R). Individuals could be recaptured from earlier in the same nesting season or from previous nesting seasons. Individual tags were read at the end of the egg-laying phase or during the next phases to minimize disturbance. A few turtles were tagged before this study and observed between 2006 and 2020 (n=9, Limpus, Boyle & Sunderland, 2006). They were included as recaptures in the results. In 2011/2012, the titanium tags were substituted with Passive Integrated Transponder (PIT) tags (Animal Electronic I.D. Systems, Australia). After one season the PIT tag project was considered too intrusive and no longer used. Only one female was tagged with a titanium tag during this season. Each individual studied by the team was measured using the minimum Curved Carapace Length method (Wyneken, 2001) and physical anomalies were recorded. The location of the nest or the nesting attempt was recorded during night and morning sifts. A triangulation method was used to record location from the season 2006/2007 to the season 2016/2017. First, electric poles along the road allowed the team to obtain an approximate location of the nest. Then, precise location of the nest was obtained using triangulation made with salient field cues (e.g., tree, rocks). Since 2017/2018 onwards GPS were used.

If the turtle had already left the beach (i.e. back to the sea), its activity was recorded using visual cues on the sand. The activity was considered a “nesting success” when the pit was filled with disturbed sand, indicating that the individual went through all the nesting phases and laid its eggs (Hailman & Elowson, 1992). The activity was considered as an aborted nesting attempt (also called “false crawl”) when the cues showed either (1) an attempt: the individual went through the first phases of the nesting process but stopped before laying its egg, during either the body pit or the digging of the chamber steps, or (2) a turnaround: the individual ascended the beach, stopped, turned around and returned to the sea (continuous track visible on the sand). Date, time, and location were also recorded.

This work was authorised by the Province Sud of New Caledonia, provincial decree numbers: 2015-2010/ARR/DENV, 2607-2012/ARR/DENV, 2770-2013/ARR/DENV, 2973-2015/ARR/DENV, 3553-2016/ARR/DENV, 3385-2017/ARR/DENV, 4276-2018/ARR/DENV, 2837-2019/ARR/DENV.

Data analysis

All data from 2006 to 2020 were gathered in one global database (available on Dryad, Barbier et al., 2023). This global database was used for further analysis.

The calculations were carried out in Excel 2016 and R Statistical software (v4.2.2; R Core Team 2022). Statistical analyses were performed using R Statistical software (v4.2.2; R Core Team 2022) and the package ggplots2 was used to create the figures (Wickham, Chang & Wickham, 2016).

Calculation of the total number of different individuals

The left front flipper tag was used to identify individuals. If this tag was missing or unreadable, the tag located on the right front flipper was used. The individuals tagged with PIT tags during the season 2011/2012 were not recorded as Captures for this season. They were probably tagged again with a titanium tag and considered as new individuals during the following seasons of this study as it was impossible for the team to recognize those individuals. Counting those individuals as Captures in 2011/2012 and as Captures in the following seasons would have led to an overestimation of the population size. Indeed, it is likely that the individuals encountered in 2011/2012 came back at least once during the eight following years of the study (remigration interval = 3.34 ± 1.77 years, 2 to 4 years on average, reviewed in Miller, Limpus & Godfrey, 2003). Full identification of the individuals was impossible for 2.1% of the observations (36/1682). Among those, 13 individuals possessed a tag although no previous reference to the tag was mentioned in the database, 16 individuals presented scars of tag loss and seven individuals had a tag but the patrol team was not able to fully read it. When compared to the database, those part-read tags could correspond to more than one already-registered tag. Those individuals were named ND1 to ND7 and were taken into account for the inter-nesting recapture calculation only. All 36 individuals were then considered as already known but unidentified. Finally, nine individuals observed during the study were tagged before our monitoring (Limpus, Boyle & Sunderland, 2006). The tags were known and registered, and individuals were counted as Recapture.

Metrics calculation

The inter-nesting interval (interval between two successful laying attempts) was calculated using individual identification and the dates of successful laying attempts within a nesting season. For each individual, the number of days between two successful laying attempts was determined.

The clutch frequency extrapolated (CFE) was calculated using the formula: total number of clutches laid within a nesting season / total number of different individuals observed within this season. The clutch frequency observed (CFO) was the number of clutch laid per identified individual within a nesting season (Frazer & Richardson, 1985; Tucker et al., 2018).

The nesting success was calculated using the formula: number of successful nesting attempts within a season / total number of nesting activities within this season (i.e. crawls on the beach).

The Percentage Increase (PI) in nesting individuals was calculated using the formula: ((number of individuals at seasons - number of individuals at seasons-1) / number of individuals at seasons-1) x 100.

Statistical analysis

Temporal variation in the number of new individuals, the total number of different individuals, the percentage of nesting success, and the number of activities were estimated using a Generalized Additive Model GAM (mgcv package, Wood, 2017; Wood, 2022). REML (Residual Maximum Likelihood) was used for smoothness selection and a penalized cubic regression spline fitting procedure was implemented. The models included either number of individuals, number of activities (Poisson distribution), or percentage of nesting success (beta distribution, fitdistrplus package, Delignette-Muller & Dutang, 2015) as the response variable and year as the smooth fixed effect. Ten different models with different number of knots (four to 13 knots) for the smooth fixed effect “season” were tested. Model selection was performed using AIC criteria and visual fit to the data. Best models showed 13, 9, 12, and 5 knots respectively for the total number of different individuals, the number of new individuals, the number of activities, and the nesting success. Model validity was verified using the statistical and visual output from the gam.check function. Finally, linear models (“season” included as a linear fixed effect) were used to test linear trend in the temporal data (LM package lme4, Bates et al. 2015). Normality and homoscedasticity of residuals were verified for each linear model. A chi-square test was used to test if the presence of the team affected the nesting success. Averages are given along with standard deviation (Mean ± SD).

R Statistical software (v4.2.2; R Core Team 2022) was used to run the statistical analyses.