Abundance, placement and sexual identity of the epizoic barnacle Chelonibia testudinaria relative to the size and species of host turtles in Mabul Island, Malaysia

Abstract Sea turtles are partners in commensal relationships with a variety of epibionts, including barnacles. The acorn barnacle Chelonibia testudinaria is one species commonly found associated with sea turtles and other marine fauna throughout temperate to tropical waters including the Indo-Pacific. We conducted a study to assess the occurrence of this barnacle, relative to host life stage and species in a mixed foraging population of green (Chelonia mydas) and hawksbill (Eretmochelys imbricata) turtles (juvenile, sub-adult, adults (female and male)) in Mabul Island, Sabah, Malaysia (Celebes Sea). Expecting similar relative abundance by life stage, we found instead a significant effect between the combined dependent variables (abundance and size of barnacles) and the life stages of sea turtles in Mabul after controlling for the covariates of size and species of sea turtles (F(8, 428) = 5.77, P < 0.001, Pillai = 0.19). Among green turtles with barnacles, though adult males had larger barnacles compared with the female turtles, the mean barnacle abundance on adult females (43.4 individuals ± 5.19 SD) was higher than all other life stages. Most of the barnacles (85.6%; N = 1931) were found on the plastron of the sea turtles. The highest number of barnacle reacquisition was found among the juvenile turtles. In assessing the complemental males of the barnacles, we found they were consistently attached to the shells of the larger of the hermaphrodites from each region of the host's body despite average shell-size differences with each region.


Introduction
Barnacles (subclass Cirripedia) are sessile crustaceans that are commonly found attached to natural surfaces such as rocks, shells and corals (Chan & Høeg, 2015). However, those in the superfamily Coronuloidea specialize as obligate commensals of mobile marine animals including mammals, reptiles, chelicerates and large crustaceans (Badrudeen, 2000;Cheang et al., 2013;Hayashi, 2013;Hayashi et al., 2013;Zardus et al., 2014;Carrillo et al., 2015;Buckeridge et al., 2018Buckeridge et al., , 2019Dreyer et al., 2020;Kim et al., 2020). They live as epibionts on the exterior body of their hosts of which sea turtles carry the widest variety (Frick & Pfaller, 2013;Hayashi, 2013); therefore, in analyses of associations between sea turtles and epibionts, much attention has been given to barnacles in particular (Zardus & Balazs, 2007;Pfaller et al., 2008;Frick et al., 2010;Fuller et al., 2010). More than 20 nominal species of barnacles from Balanomorpha (acorn barnacles) and Lepadomorpha (gooseneck barnacles) have been reported from sea turtles (Hayashi, 2009). Chelonibia testudinaria is the most cosmopolitan and largest acorn barnacle on sea turtles, reaching a diameter of 120 mm (Zardus & Hadfield, 2004), and associate with all seven species of sea turtles in the world's oceans including the flatback turtle (Natator depressus) (Monroe & Limpus, 1979); as well as the dermiscovered leatherback turtle (Dermochelys coriacea) (Rees & Walker, 1993). Of the four species of barnacles in the family Chelonibiidae, three have been found to be genetically identical and are synonymized under C. testudinaria (Zardus et al., 2014). For individuals attached to sea turtles, a diagnostic trait of C. testudinaria (Figure 1) is its stellate shell patterning formed by open wedges at the sutures between the shell plates or compartments, sculpted along their margins with indentations (Darwin, 1854;Hayashi, 2013).
Parasitic epibionts of sea turtles, leeches for instance, which derive nutrition from the tissues of their hosts can significantly impact the health of sea turtles (Greenblatt et al., 2004). In contrast, barnacles and other commensals use sea turtles primarily as a substratum or foraging platform (Frick et al., 2002) with minor or equivocal impacts on the turtles' health (Stamper et al., 2005;Flint et al., 2010). Depending on the species of barnacle, they fasten onto the carapace and plastron, on the head, or the flippers and skin of their host turtles (Frick & Ross, 2001;Devin & Sadeghi, 2010;Ooi & Palaniappan, 2011;Nájera-Hillman et al., 2012). There are two types of attachment modes across different life stages of barnacles. First, the cypris larva arrives on the substratum and secretes cyprid cement from multicellular cement glands that lead to ducts on the antennules. The release of this cement triggers the metamorphosis of the cypris larva into juvenile and adult barnacles. Second, the adult secretes cement from cement glands to maintain firm attachment. Both stages secrete cement from cement glands that lead through the base (Lacombe, 1970;Walker, 1978).
How the sea turtle life cycle influences barnacle occurrence is unresolved. Most sea turtles undergo long-distance migrations to different habitats throughout their lives (Åkesson et al., 2003). At hatching, newly emerged sea turtles migrate toward oceanic nursery areas then, as older juveniles, subadults and adults, return to neritic foraging grounds (Musick & Limpus, 1997;Lohmann et al., 2008). A recent study by Burgett et al. (2018) has confirmed the ontogenetic diet shifts of green turtles (Chelonia mydas) in a mid-ocean developmental habitat, thus supporting the flexibility of habitat shifts between their foraging and neritic habitats (Hayashi & Nishizawa, 2015). Mature adults periodically migrate between distantly spaced breeding grounds, nesting beaches and foraging areas with a high degree of fidelity for these areas (Bjorndal, 1995;Palaniappan & Abd Hamid, 2017;Shimada et al., 2020). A foraging turtle is defined as a sea turtle that resides in an area where its food source is available (Ceriani et al., 2012;Cheng et al., 2019) and exhibits a strong long-term fidelity to localized foraging sites whereas a resident turtle is defined as a non-migratory individual (Márquez, 1990), that after recruiting from its oceanic nursery resides predominantly at the foraging ground and thus has the probability of being recaptured throughout its resident years (Chaloupka & Limpus, 2001). Foraging turtle aggregations commonly comprise several nesting stocks (Nishizawa et al., 2016). Both foraging and resident female turtles will not nest in the foraging areas. Instead, they return to their natal beach to nest due to their natal homing nature. Nesting turtles are reproductively active females that migrate from distant, long-term residence areas (hereafter, foraging grounds) to their natal nesting beaches during the breeding season (Ceriani et al., 2017;Sönmez, 2019). Documenting the epibiont diversity associated with migrating sea turtles can help provide information about where epibiosis occurs and hence the migratory behaviour and habitat preferences of sea turtles (Hayashi, 2009). The findings of Robinson et al. (2017) suggest that sea turtle epibiont communities are more reflective of where sea turtles feed than where they nest. In addition, stable isotope analysis of barnacle shells has been used to provide insight on foraging distributions, migration distances and habitat use of nesting turtles over time (Pearson et al., 2019), thus aiding in sea turtle population conservation and management.
An initial study of sea turtle epibionts for the region of Sabah, Malaysia by Ooi & Palaniappan (2011) found only one species of barnacle on sea turtles -C. testudinaria, occurring on both green Chelonia mydas (Linnaeus, 1758) and hawksbill Eretmochelys imbricata (Linnaeus, 1766) turtles in the waters of Mabul and Sipadan Islands. Mabul Island (Figure 2) is home to a variety of seagrass species especially Thalassia hemprichii and diverse coral reef communities (Jolis & Kassem, 2011), which attracts a foraging sea turtle population dominated by green turtles, and a small number of hawksbill turtles. So far as is known, both sea turtle species are residents in Mabul Island and some have been caught repeatedly over consecutive years of sampling. Hawksbill and green turtles also share foraging grounds at neighbouring Sipadan Island (14.5 km away) (Chong, 2012;Joseph et al., 2017). Only green turtles hatched in Mabul Island will travel back to their natal beach to nest during the breeding season then return to their foraging grounds, as reported by Joseph et al. (2017).
In the course of examining the sea turtles in Mabul Island, many were found hosting the barnacle C. testudinaria ( Figure 1). In addition, many of the barnacles were found with multiple small male individuals attached to larger hermaphrodites, situated within the narrowest portions of the hermaphrodites' settlement pits as described by Zardus & Hadfield (2004) or attached randomly on the shell surface and apertural orifice (Cheang et al., 2013). Barnacles in general are hermaphroditic and cross-fertilize with neighbouring individuals via the typical barnacle mode of pseudo-copulation (they deposit sperm into a partner's mantle cavity) (Murata et al., 2001), though some barnacle species are capable of performing sperm casting with subsequent fertilization of the internally kept broods (Barazandeh et al., 2013). Darwin (1854) was the first to describe the rare phenomenon of androdioecious reproduction in scalpellid barnacles, wherein small males fertilize larger hermaphrodites, naming the small companions as complemental males. Zardus & Hadfield (2004) found that the complemental males of C. testudinaria possess mature sperm and serve as sperm donors, however it is unknown if complemental males are protandrous and in time develop into full-grown hermaphrodites (Weeks et al., 2006) or whether they can flexibly develop female sexual characters according to their position and partner availability (Wijayanti & Yusa, 2016).
Most sea turtle barnacle investigations have focused on barnacle abundance and diversity relative to sea turtle species (Hernández-Vázquez & Valadez-González, 1998;Pfaller et al., 2006Pfaller et al., , 2008Fuller et al., 2010). This study, in addition to barnacle occurrence, presents fundamental information on the occurrence of complemental males of C. testudinaria as well. To date, few studies have included information on the complemental males of C. testudinaria from wild sea turtles. Ewers-Saucedo et al. (2015) studying C. testudinaria in the south-eastern USA, determined that complemental males were more commonly partners with hermaphrodites from loggerhead sea turtles (Caretta caretta) than with hermaphrodites from crabs, perhaps due to smaller group size and higher mortality rates associated with the latter hosts. The objective of the present study was to determine whether the abundance and size class distribution of the epibiotic barnacle, C. testudinaria, and the occurrence of its complemental males, in the foraging population of sea turtles at Mabul Island, is influenced by sea turtle species and life stage.

Materials and methods
The study was conducted on the foraging population of sea turtles at Mabul Island (4.25°N 118.63°E) located on the east coast of Sabah, Malaysia ( Figure 2). Sea turtles were caught in-water from May 2015 to November 2017. Fieldwork was conducted over a period of 4-6 days once every 6 months per year (a total of 26 days spread over 6 sampling trips) to keep disturbance to the sea turtles at a minimum. The sea turtles were randomly caught by hand while scuba diving during the day at established dive sites (distance between adjacent dive sites did not exceed 500 m) in Mabul Island at depths not exceeding 20 m. The sea turtles were divided into three life stages based on size i.e. straight carapace length (SCL), namely juvenile, sub-adult and adult, with adults further divided into female and male following previous studies (Bresette et al., 2010;Palaniappan, 2017). Sex distinctions were based on size of the tail (van Dam & Diez, 1998). The sea turtles were brought onto the research vessel to be photographed, measured and tagged. The SCL of each sea turtle was measured with 1.0 m Mitutoyo stainless Vernier callipers (accurate to 0.05 ± 0.15 mm) measuring from the anterior point at the midline (nuchal scute) to the posterior end of the supracaudal (Bolten, 1999). Inconel tags (National Band and Tag Company, Newport, Kentucky, USA) were applied to the sea turtles at the posterior edge of both front flippers.
Scaled photographs were taken of the specific locations of the barnacles found on the body of the sea turtles. For each turtle, including repeat captures, all attached barnacles larger than 5 mm were removed before release. Barnacles found on the head, carapace and plastron of the sea turtles were gently removed using a hammer and scraper. Barnacles found elsewhere on the sea turtle were not included in this study. The collected barnacle specimens (minimum size 5 mm) were kept in resealable plastic bags labelled with the turtle's ID number, date of collection and the location that they were found on the turtle's body. The barnacles were identified in the laboratory at the Borneo Marine Research Institute, Universiti Malaysia Sabah (UMS) to the lowest taxonomic level following Monroe (1981). Barnacle size was measured as maximal rostro-carinal diameter ( Figure 1) with 0.3 m Mitutoyo stainless steel Vernier callipers (accurate to 0.02 ± 0.04 mm). The complemental males that were found attached on every hermaphrodite were examined and counted under a Carl Zeiss Stemi stereomicroscope. Barnacles with broken shells were excluded from the analysis.

Field survey results
A total of 403 sea turtles were captured from May 2015 to November 2017 in Mabul Island, consisting of 364 green turtles and 39 hawksbill turtles. All life stage groups, as determined by SCL, were represented in the population of green turtles (77.8% juveniles, 6.0% sub-adults, 6.9% adult males, 9.3% adult females) but hawksbill turtles were mostly juveniles (94.9%) with only one sub-adult and one adult female (Table 1). The low number of subadult and adult hawksbill turtles suggests that juvenile hawksbill turtles eventually move to neritic habitats and utilize Mabul Island as a temporary developmental foraging ground (Pilcher, 2010;Joseph et al., 2017).
Mabul Island consists of a mixed population of foraging, resident and nesting sea turtles. All the sea turtles caught in-water during this study were tagged upon capture. We have had no reports of these tag numbers among the green turtles that have nested in Mabul Island. Green turtles show natal homing on a very broad, regional scale well before their first reproductive migration (Allard et al., 1994). When sea turtles leave the open ocean to establish coastal feeding sites, they choose foraging grounds within their general natal region, which are often a considerable distance from the natal beach (Bowen et al., 2004;Bowen & Karl, 2007). A total of 2255 barnacles summed across all hosts were collected, all of a single species, Chelonibia testudinaria. This is consistent with previous studies of green turtles in Mabul and Sipadan Islands, Malaysia (Ooi & Palaniappan, 2011) and in the waters of Japan (Hayashi & Tsuji, 2008). Intriguingly, Dobbs & Landry (2004) did not find this barnacle species in a nesting population of Australian hawksbill turtles but did document the presence of other species of commensal coronulids, including Chelonibia caretta (the dominant species with 81% occurrence), Platylepas sp. and the burrowing barnacle Chelolepas cheloniae. Nevertheless, a recent study by Razaghian et al. (2019) had reported C. testudinaria on the nesting hawksbill turtles in Iran. The overall incidence of sea turtles carrying this barnacle was similar between the two host species with 208 green turtles (57.1%) and 22 hawksbill turtles (56.4%) hosting it. The percentage of sea turtles with C. testudinaria was also similar (∼50%) across the host life stages present (Table 1) with no significant difference (χ 2 (3) = 0.15, P > 0.05). However, the quantity of barnacles per turtle varied with turtle size, species and sex.
Abundance and size variation of C. testudinaria Overall, on sea turtles with barnacles, barnacle abundance ranged from 1-232 individuals per turtle (mean = 9.80 individuals ± 23.20 SD), across both hosts. The highest number of barnacles was found on a large adult female green turtle (SCL = 910 mm) whereas several individuals of both host species hosted only one barnacle each. In comparison between the two sea turtle species, juvenile green turtles had substantially more barnacles (mean = 6.9 individuals ± 9.34 SD) than juvenile hawksbill turtles (mean = 3.5 individuals ± 2.98 SD) ( Figure 3). This could be due to the greater body surface area of green turtles which on average is larger than hawksbill turtles, a finding that is consistent with Hayashi & Tsuji (2008). There was a statistically significant difference between the sea turtle life stages on the combined dependent variables (i.e. barnacle abundance and size of barnacles) after controlling for turtle size (SCL) and turtle species (F(8, 428) = 5.77, P < 0.001, Pillai = 0.19). We found, to a lesser degree, C. testudinaria on the foraging population of hawksbill turtles in Mabul Island. The small sample size of adult hawksbill turtles captured in this study did not allow us to fully address the relationship between the abundance of barnacles and turtle size, therefore, no further analyses were performed.
Among green turtles, females had considerably more barnacles (mean = 43.4 individuals ± 67.96 SD) than males (mean = 12.4 individuals ± 14.39 SD), sub-adults (mean = 14.5 individuals ± 14.79 SD) and juveniles (mean = 6.9 individuals ± 9.34 SD). Few studies have compared barnacle occurrence among sea turtles in populations of mixed life stages though several have compared epibionts among mixed sea turtle species (Fuller et al., 2010;Lazo-Wasem et al., 2011;Ooi & Palaniappan, 2011;Robinson et al., 2017). The number of barnacles per turtle varied widely in our study with the highest number (i.e. 232 individuals) collected from an adult female at Mabul Island, considered a high amount for green turtles (Bugoni et al., 2001;Hayashi & Tsuji, 2008;Fuller et al., 2010;Lazo-Wasem et al., 2011;Nájera-Hillman et al., 2012;Robinson et al., 2017). Our results indicate that the number of barnacles attached to green turtles increases with turtle size, and this finding is consistent with Hayashi & Tsuji (2008) and Devin & Sadeghi (2010). The attachment area selected is due to the free space availability (Minchinton & Scheibling, 1993;Ihwan et al., 2018), thus resulting in a higher abundance of barnacles in larger sea turtles.
The size of barnacles differed significantly between collecting trips as determined by one-way ANOVA (F(5, 2249) = 15.18, P < 0.001), except those collected in May of 2015 and 2016. Barnacles collected in May were larger (mean = 19.1 mm ± 8.41 SD) than those collected in November (mean = 16.7 mm ± 8.81 SD), across all sampling trips, indicative of either periodic acquisition or growth. The size frequency distribution of barnacles across the duration of the study was skewed significantly from normal as determined by the Shapiro-Wilk's test (Shapiro & Wilk, 1965) (W = 0.95363, P < 0.001). A skewness test confirmed the data were skewed to the right, driven by the presence of a few large individuals (Figure 4). However, Hartigan's dip test (Hartigan & Hartigan, 1985) for unimodality indicated our data   (2015) suggested that the size frequency distribution of each age class in a natural population should be normally distributed and multimodal if several age classes are present. While our data were positively skewed, there was no evidence of multimodality in our data despite the presence of a few large individuals on host sea turtles. A lack of age classes in our study may be due to year-round reproduction of C. testudinaria in a tropical location compared with the previously mentioned studies from more seasonally influenced climates. Upon visual inspection, most of the barnacles collected in Mabul Island were in the size class of 5.1-10 mm, suggesting that these individuals were rather young. Ten et al. (2019) suggested that the distribution of C. testudinaria in neritic waters could result from synergistic effects of its flexibility in host selection and its short planktonic phase (about 9 days). Among the juvenile green turtles, large juveniles (501-750 mm) acquired larger barnacles (mean = 19.8 ± 9.11 SD) than the small juveniles (351-500 mm) (mean = 17.9 ± 7.09 SD) after controlling for SCL (F(2, 158) = 4.93, P < 0.05, Pillai = 0.06) ( Figure 5A). Such difference in barnacle size between the two juvenile groups suggests these barnacle individuals are new recruits from different locations. Post-hatchling sea turtles can travel up to 12,000 km from their natal regions to juvenile foraging sites (Hays & Scott, 2013), and might be colonized by various epibionts including barnacles along their migration pathways. The presence of bigger barnacles on larger juvenile green turtles suggests that they have acquired the barnacles prior to arrival in Mabul Island as the growth rate of the colonized barnacles is likely driven by enhanced feeding conditions caused by their host turtle's travelling activity (Trager et al., 1990;Doell et al., 2017).
We conducted an ANCOVA test on barnacle size between the foraging adult male and female green turtles, with turtle body location as the covariate. While the size of barnacles found on  the plastron, carapace and head of sea turtles differed significantly (F(3, 906) = 3.58, P < 0.05), the interaction between the body location and the sex of sea turtles had no effect on the barnacle size (F(1, 905) = 0.04, P > 0.05). Nevertheless, the foraging adult males had acquired larger barnacles as compared with the foraging adult females ( Figure 5B). This difference could be caused by variation in each sea turtle's historical environmental exposure (Stamper et al., 2005). Male sea turtles migrate between foraging and breeding grounds more frequently (on an annual basis) compared with female sea turtles (once every 2-3 years or more) (Balazs, 1983;Hays et al., 2014). The growth rate of C. testudinaria is attributed to food availability, which is dependent on the sea turtle's activity level, diet and feeding behaviour (Doell et al., 2017). The travelling activity associated with migrating sea turtles can cause an increase in water flow and nutrient availability, providing potentially enhanced feeding conditions for barnacles (Trager et al., 1990), which may promote rapid growth of barnacles on host sea turtles (Doell et al., 2017). This may explain why adult male green turtles had larger barnacles as compared with females in our study.

Barnacle site selection
Enumeration of barnacles by body location for each sea turtle species of Mabul Island revealed a consistent pattern, that barnacles had a strong predisposition for the plastron (85.6%) over the carapace (13.7%) and low affinity for the head (0.6%) ( Table 2).
Among green turtles, although the foraging females had more barnacles than foraging males, the number of barnacles by body location is consistent for both sexes where higher numbers of barnacles were found on the plastron on both adult male and female turtles compared with the carapace. The occurrence of C. testudinaria on sea turtles is affected by the host sea turtle's behaviour (Frick & McFall, 2007), size (Hayashi & Tsuji, 2008), interactions among epibionts (Pfaller et al., 2006), and tolerance to desiccation and physical trauma (Pfaller et al., 2008). Swimming and resting behaviours displayed by sea turtles on hard substrates may affect the settlement of barnacles on a turtle's body (Razaghian et al., 2019), while an increase in water flow could affect both settlement patterns of the larvae (Larsson & Jonsson, 2006) and further augment the feeding opportunities of the barnacles (Schärer, 2003). Sea turtles commonly visit specific cleaning stations on reefs to seek 'free' cleaning services by fishes and shrimps to get rid of inconvenient epibionts and parasites (Sazima et al., 2004;Grossman et al., 2006;Schofield et al., 2006;Ooi & Palaniappan, 2011). Cleaning by most reef fish focuses on the carapace and skin of the sea turtles (Sazima et al., 2010) while the sea turtles rest on the reef motionless with the plastron facing down. Thus, small barnacles attached on the carapace may get removed more frequently than on the plastron due to higher exposure to the cleaner organisms. The presence of distinct sea turtle cleaning stations in Sipadan Island (Ooi & Palaniappan, 2011) supports this hypothesis. Moreover, healthy sea turtles always practice self-cleaning by wedging themselves into coral
crevices or scraping their carapace against ledges and other hard substrata to remove barnacles (Frick & McFall, 2007), which may also help explain why more barnacles were found on the plastron compared with the carapace. While there was no difference between the enumeration of barnacles by body location among the adult green turtles, this suggests that both sexes share the same cleaning behaviour in Mabul Island.

Reacquisition of barnacles on recaptured turtles
Our study is one of the few to report on reacquisition of barnacles on sea turtles. Of the 230 sea turtles with barnacles, 51 turtles were recaptured individuals (hereafter, resident sea turtles) from which all barnacles larger than 5 mm had been collected for analysis at each previous capture. Among these resident sea turtles, 39 turtles were recaptured once, eight turtles were recaptured twice, three turtles were recaptured thrice and one turtle was recaptured four times. The total quantity of reacquired barnacles was 365 (16.2% of 2255) and, among turtle life stages, was highest for juveniles for both host species (Table 3). The shortest interval for barnacle reacquisition was ∼175 days (one sampling trip) whereas the longest was ∼730 days (four sampling trips). The geographic localities where sea turtles are colonized by barnacles can give an understanding of the movements and habitat use by sea turtles (Ten et al., 2019) and conversely, where and when barnacles accumulate on sea turtles can provide insight in barnacle life history. There were no large turtles (i.e. sub-adult, adult males and adult females) recaptured after Recapture Event One, but repeated captures after the first recapture were evident among juvenile turtles (Table 3). This could be attributed to the sitefidelity of juvenile turtles, similar to the findings of Pilcher (2010) where the author reported that young sea turtles exhibited minimal movement within their foraging areas over several recaptures in Mantanani Island (west coast of Sabah). The reacquisition of barnacles on recaptured juvenile turtles in our study verifies that Mabul Island is at least one source for barnacle larvae in the region.

Complemental males of C. testudinaria
Complemental males of the turtle barnacle C. testudinaria were encountered on 37.5% of the barnacles from the sea turtles in Mabul Island, constituting the first report of complemental males for this species in the region ( Table 4). The occurrence of complemental males was greater with barnacles from green turtles (38.3%) than with barnacles from hawksbill turtles (6.8%). But, lacking larger sample sizes across other life stages of hawksbill turtles, how meaningful this difference is remains uncertain. Turtle barnacles with attached complemental males were common in this study. The complemental males of C. testudinaria have a unique settlement pattern among barnacles. They attach to the shells of hermaphrodite adults in the depressions between the shell plates, which are perhaps specialized for their settlement (Zardus & Hadfield, 2004), or at the margin of the orifice of the hermaphrodite (Crisp, 1983). The size of the hermaphrodites with complemental males ranged from 6.3-49.0 mm in rostrocarinal diameter (mean = 24.75 mm ± 6.86 SD) and the maximum number of complemental males arranged on a single adult hermaphrodite (25.3 mm) was 34 (mean = 3.4 individuals ± 5.93 SD). The mean size of adult hermaphrodites of C. testudinaria in Mabul Island was about 50% smaller than in other studies (Zardus et al., 2014) while the maximum number of attached complemental males (N = 34) was a little more than 100% greater. The causative factor for this difference remains unknown. It is common for several complemental males to occur on a single hermaphrodite and for multiple complemental males to be found within a single settlement pit as described in Zardus & Hadfield (2004). The males that attach nearest to the opercular rim, where the depressions are wider and are larger, presumably have the advantage of growing larger and of having a better success rate in copulating with their 'host' hermaphrodite (Ewers-Saucedo et al., 2016). Complemental males in the present study were mostly found attached away from the opercular rim within the 'settlement pits' of the adult hermaphrodite shell where they were protected from abrasion and greater risk of dislodgement (Zardus & Hadfield, 2004). Hermaphrodite size was on average larger for those with complemental males than for those without, but varied by their location on the body of the sea turtles (Table 4). For instance, although hermaphrodites collected from the head were generally smaller than those of the plastron and carapace, it was the larger of the head-attached individuals that hosted complemental males. Though adult hermaphroditic barnacles appear to prefer the plastron to the carapace and head of sea turtles, there was no detectable difference in the number of complemental males associated with hermaphrodites from any body location after controlling for the effect of hermaphrodites' size, (F(2, 841) = 1.99, P > 0.05). Complemental males were found on hermaphrodites from all regions of the turtle body, but less frequently on the heads of sea turtles, where the overall size of adult hermaphrodites was the smallest, suggesting that substratum space availability is the key for settlement. Complemental males in androdioecious species may evolve under lower population densities (e.g. mating group size; Yamaguchi et al., 2008;Dreyer et al., 2018a) where hermaphrodite-hermaphrodite sperm competition is relaxed and chances of fertilizing the broods of a hermaphrodite partner become greater (Ewers-Saucedo et al., 2016). In these conditions androdioecy increases opportunities for fertilization more than can be realized with two separate and autonomous sexes (Kelly & Sanford, 2010). Selection may also favour the small size of complemental males in C. testudinaria particularly by reducing the weight of mating units and minimizing hydrodynamic drag, thus lengthening the lives of the attached barnacles (Zardus & Hadfield, 2004). An ANOVA test verifies that the total number of complemental males increases with the size of the adult hermaphrodite (F(1, 841) = 21,74, P < 0.001). Our results suggest that large hermaphrodites host higher numbers of small complemental males, likely linked to space availability (Minchinton & Scheibling, 1993). Hence, we hypothesize that a larger hermaphrodite would have a larger settlement pit, thus can house more cyprids that will specifically develop into males. The settlement pits of C. testudinaria can be analogized with the receptacles (areas on the rim of mantle cavity; Dreyer et al., 2018b) of the androdioecious barnacle Scapellum scapellum, where the small complemental males are exclusively attached in the receptacle area of a hermaphrodite partner and never elsewhere (Spremberg et al., 2012;Høeg et al., 2016). Dreyer et al. (2018b) found that inward curvature of the ventral transparent lamellae (a thin, cuticular membrane between the mantle rim and mantle cavity) in the receptacle area provides more space for cyprid settlement in larger species, a similar reasoning that can apply to the settlement pits of larger C. testudinaria. However, addressing this hypothesis requires additional measurements of the size of the pits on each hermaphrodite.
In conclusion, the turtle barnacle C. testudinaria was found in abundance on the foraging populations of sea turtles in Mabul Island. The highest number of barnacles found on an individual sea turtle was 232 and barnacle abundance was slightly affected by turtle size only for green turtles. Colonization of this barnacle species was the highest among adult female turtles with SCL of 828-991 mm. The majority of hawksbill turtles that were captured in the current study were juveniles, suggesting that Mabul Island serves as a temporary developmental foraging ground for this species. The repeated captures of juvenile turtles over several sampling trips, together with the abundant small-sized barnacles found among the smaller juvenile green turtles, suggest that barnacle larvae were present in Mabul Island. Among the adult green turtles, adult males had larger barnacles compared with the adult females. It is generally known that the adult males travel more frequently than females during the breeding seasons. Such intensive travelling activity may directly enhance barnacle feeding conditions, thus leading to an increase in the growth rate of barnacles on the adult male turtles particularly. Although the overall sizes of barnacles collected in May were larger than those collected in November, we were unable to relate this to seasonality due to the limited duration of sampling. The size frequency distribution of C. testudinaria in the current study was unimodal and not normally distributed. Barnacles were found to recolonize the same sea turtles several times, but the factor that drove the temporal succession of barnacles on these turtles is unclear. The highest abundance of C. testudinaria was found on the plastron (85.6%), followed by the carapace (13.7%) and the least on the head (0.6%) of sea turtles. Reacquisition of barnacles on recaptured turtles and the presence of complemental males of C. testudinaria were recorded for the first time on the foraging population of sea turtles in Mabul Island. More complemental males were found on the larger hermaphrodites, suggesting that substratum space availability is a factor in complemental male recruitment. This is supported in scalpellid barnacles, where the size and shape of the male-housing receptacle area on the scutal shell plates are positively correlated with the number of males attached (Dreyer et al., 2018b). Future insights on the aggregation pattern of complemental males on adult hermaphrodites could be done by using a zonal model to determine settlement behaviour and survival of barnacles in relation to distance of the nearest conspecific from a given individual. Future studies should consider barnacle shell isotopes to trace movements and residence time of sea turtles in each habitat, such as has been done for inferring ancient whale migration routes and patterns (Buckeridge et al., 2018;Taylor et al., 2019).
Acknowledgements. The Mabul Sea Turtle Research and Conservation Project headed by P.P. is sanctioned by the Sabah Ministry of Tourism, Culture and Environment, supported with the presence of the Minister during the launch of the fieldwork events, as well as the Sabah Wildlife Department which provided the turtle flipper tags used in this study. The project is also sanctioned by Universiti Malaysia Sabah. We would also like to thank the organizers and all the volunteers of the Mabul sea turtle outreach programme as well as Universiti Malaysia Sabah for assistance in the data collection. Mean ± SD (mm) 13.0 ± 6.52 24.7 ± 6.80 17.8 ± 7.60 25.9 ± 7.40 17.3 ± 6.11 18.6 ± 5.02 Measurements shown are rostro-carinal diameter of the barnacles (mm) plus or minus the standard deviation (SD) and the number sampled (N).