Diversity Patterns and Community Structure of the Ground-Associated Macrofauna along the Beach-Inland Transition Zone of Small Tropical Islands

: Biodiversity follows distinct and observable patterns. Where two systems meet, biodiversity is often increased, due to overlapping occurrence ranges and the presence of specialized species that can tolerate the dynamic conditions of the transition zone. One of the most pronounced transition zones occurs at shores, where oceans and terrestrial habitat collide, forming the shore–inland transition zone. The relevance of this transition zone in shaping a system’s community structure is particularly pronounced on small islands due to their high shore-to-inland-area ratio. However, the community structure of insular faunas along this transition zone is unknown. Here, we investigated the diversity patterns along the beach–inland transition zone of small islands and tested the hypothe-sis that species diversity increases toward the transition zone where beach and interior habitat meet. By measuring environmental parameters, resource availability, and ground-associated macrofauna diversity along transects running across the beach–inland transition zone, we show that a gradual change in species composition from beach to the inland exists, but neither taxa richness, diversity, nor overall abundance changed signiﬁcantly. These ﬁndings offer important insights into insular community structure at the transition zone from sea to land that are relevant to better understand the dynamic and unique characteristics of insular ecosystems.


Introduction
Understanding the drivers for species distribution and biodiversity patterns at different spatial scales is among the fundamental goals of ecological research [1]. On a global scale, biodiversity changes with elevation and latitude [2]. The main factors shaping these global diversity gradients are temperature, habitat heterogeneity, and area, coupled with environmental stability, predictability, and productivity [3]. With some exceptions (e.g., in marine gastropods in the Southern Hemisphere [4]), these environmental factors allow for the prediction of overall species richness on large spatial scales in the terrestrial and aquatic environment [5]. On smaller spatial scales, however, a given environmental parameter does not determine the occurrence of all species in the same way. Instead, different taxa adapt and respond differently to the same environmental parameter, resulting in idiosyncratic diversity patterns [6].
Shifting from global to local scales, variations in more fine-scale environmental abiotic and biotic interactions become the predominant factors influencing species distribution and, thus, biodiversity patterns [2,5]. The fine-scale gradients determining species composition (for a map of the study site refer to [27]). Sampling was performed on 13 September 2019 between 11:45 to 15:00 on Veyvah (conditions: air temperature: 32 • C, wind speed: 2.4 m/s (avg. 1 min), wind direction: SSW, weather conditions: clouded, partly sunny), on 14 September 2019 between 11:15 to 16:45 on Dhidhdhoo (conditions: air temperature: 31 • C, wind speed: 1.1 m/s, wind direction: SSW, weather conditions: sunny with clouds) and on 15 September 2019 between 10:30 to 12:45 on Veyvah (conditions: air temperature: 33 • C, wind speed: 0.7 m/s, wind direction: SSW, weather conditions: sunny). The field surveying was covered by the legal permit of the Atoll Marine Center field station (10-A/2000/31).

Sampling Preparation
Sampling on both islands was carried out on the accessible beachlines (i.e., exposed sandy beaches not covered by (mangrove) vegetation). These beachlines were mapped using GPS (Garmin GPSMAP 62st, Garmin International Inc., Olathe, KS, USA) and six transects per islands were uniformly distributed along the accessible beachline in a perpendicular orientation to the drift line. The first and last transect were always positioned at 10 m away from the outer edges of the mapped accessible beachline, with the four remaining transects evenly distributed along the remaining space in between. The beginning of the primary vegetation (predominantly Cyperus dubius, Launaea sarmentosa, Ipomoea spp., and Sphagneticoa trilobata) was chosen as the starting point of each transect. Starting from this point, 5.5 m landward and seaward were defined as one transect line. Every transect comprised a total of ten 1 × 1 m plots, four positioned seaward toward the drift line, and four positioned landward in the opposite direction, each equally spaced 0.5 m apart and terminating with an additional inland and beach reference plot, respectively. The beach and inland reference plots did not adhere to the regular 0.5 m spacing and were positioned directly on the drift line and further inland under the dense interior insular vegetation, respectively (see Supplementary Figure S1 for schematic drawing of the sampling design).

Sampling Procedure
Sampling started from the beach reference plot toward the inland. Each plot was sampled only once following the same sampling procedure. First, a picture of each plot was taken from the top view to measure vegetation cover (Tevion DC-14, Supra, Kaiserslautern, Germany; Panasonic Lumix DMC-FT5, Panasonic, Kadoma, Japan; Galaxy A5 (2017) -SM-A520F, Samsung, Seoul, South Korea). Next, the plot was scanned for organisms. To detect species that might be hidden beneath vegetation, debris, or human waste, any material obstructing the view of the observers was carefully removed while scanning for hidden or buried organisms. Species were counted and identified up to the lowest possible taxonomic level and subsequently released in their original habitat. Plot temperature was measured in the four corners and in the center by inserting a soil thermometer 1 cm deep into the ground and averaging the five values for each plot (P300W 0-100 • C ± 0.5 • C, Dostmann electronic GmbH, Wertheim-Reicholzheim, Germany). Detritus was collected and classified into two categories (i.e., terrestrial (leaves, branches, deadwood) and marine (seagrass, marine carrion)). The wet mass of the collected detritus per plot was weighed on site using a digital scale (NTP2K 2000 g ± 0.1 g, Nohlex GmbH, Buchholz, Germany). Percentage of grass/herb and shrub coverage per plot was measured using ImageJ 1.49b (Rasband, W.S., ImageJ, U.S. National Institutes of Health, Bethesda, MD, USA, http://imagej.nih.gov/ij/, 1997-2015, accessed on 1 October 2019).

Statistical Analysis
Statistical analysis was carried out using R 3.5.3 extended with the "vegan" and "hillR" packages for multivariate data analysis and calculation of Hill numbers, respectively [28][29][30]. The measured environmental parameters (i.e., soil surface temperature, marine detritus, terrestrial detritus, grass/herb coverage, shrub coverage) were compared between the two investigated islands using permutational multivariate analysis of variance (PERMANOVA) with Bray-Curtis dissimilarity indices and 4999 permutations. For each island, the environmental parameters of the beach-inland transition zone were analyzed using principal component analysis (PCA). The first three principal components (PCs) explained 70.07% of the total inertia and were used for subsequent statistical testing. One-way analysis of variances (ANOVAs) with Tukey HSD post-hoc testing and Bonferroni p-value correction were performed to statistically compare the first three principal components (PC1 to PC3). Taxa richness (Hill number q = 0), exponential Shannon-Wiener (Hill number q = 1), the inverse of Simpson's concentration index (Hill number q = 2), Berger-Parker index (Hill number q = 3), and total number of individuals were calculated for each plot of the transect line (N = 6 per island). The different values for the parameter q of the Hill number quantifies how much rare species are discounted in the calculation of the diversity index. A hill number with q = 0 is simply the number of species, while q = 1 weighs species in proportion to their frequency. Hill numbers with q = 2 and q = 3 add more weight on the more abundant species, while discounting rare ones [31]. Higher values for the q = 3 Hill number (Berger-Parker index) thus indicate that the most common species dominate (i.e., a lower evenness). The diversity indices were statistically compared between the transects of each island using the non-parametric Kruskal-Wallis test with Bonferroni p-value correction for multiple testing, as data did not meet the assumptions for parametric testing. For each taxon, the average occurrence range (i.e., distribution maximum ± standard variation) along the beach-inland transition zone was obtained from the mean taxa's abundances for each plot of the transect (N = 6 for each island). Differences in average occurrence ranges were statistically compared between the two islands using Monte Carlo-simulations (100,000 runs) and Bonferroni p-value corrections.

252
This study investigated the environmental parameters and the diversity pattern of 253 the ground-associated macrofauna community along the beach-inland transition zone of 254 small and remote tropical atoll islands. We show that the environmental parameters grad-255 ually changed from beach to inland. Along this environmental gradient, different ground-256 associated species occupy different sections with a smooth trend for increasing terrestri-257 ality in the type of fauna present (i.e., more insect and spider taxa, fewer crustaceans when 258 shifting along the transition zone from beach to insular interior) as the environmental con-259 ditions become less shaped by the harsh beach conditions. 260 Generally, both islands matched the conditions of reflective microtidal beaches 261 [32,33]. Even though one island (Dhidhdhoo) showed a rapid change (i.e., significant 262 changes in environmental parameters within 0.5 m) at the area of the first pioneer plants 263 in the transition zone and the second island (Veyvah) displayed a more gentle gradient, 264 the general patterns of the environmental gradients were similar on both atoll islands [34]. 265 Terrestrial detritus, which forms the dietary basis for most taxa on the inland, on islands 266 with sufficient in situ primary production [24], along with soil temperature and vegetation 267 coverage increased from the beach to the inland on both islands, displaying the expected 268 patterns also known from continental beach systems [35,36]. The amount of available ma-269 rine detritus, which forms the dietary basis for most beach-dwelling taxa [37], was gener-270 ally higher on the beach on Veyvah. In contrast, on Dhidhdhoo, it peaked at the edge 271 toward the inland habitat where the first pioneer plants appear (Figure 1c). Here, pioneer-272 ing supralittoral vegetation at the upper margin of the beach might act as a barrier holding 273

Discussion
This study investigated the environmental parameters and the diversity pattern of the ground-associated macrofauna community along the beach-inland transition zone of small and remote tropical atoll islands. We show that the environmental parameters gradually changed from beach to inland. Along this environmental gradient, different ground-associated species occupy different sections with a smooth trend for increasing terrestriality in the type of fauna present (i.e., more insect and spider taxa, fewer crustaceans when shifting along the transition zone from beach to insular interior) as the environmental conditions become less shaped by the harsh beach conditions. Generally, both islands matched the conditions of reflective microtidal beaches [32,33]. Even though one island (Dhidhdhoo) showed a rapid change (i.e., significant changes in environmental parameters within 0.5 m) at the area of the first pioneer plants in the transition zone and the second island (Veyvah) displayed a more gentle gradient, the general patterns of the environmental gradients were similar on both atoll islands [34]. Terrestrial detritus, which forms the dietary basis for most taxa on the inland, on islands with sufficient in situ primary production [24], along with soil temperature and vegetation coverage increased from the beach to the inland on both islands, displaying the expected patterns also known from continental beach systems [35,36]. The amount of available marine detritus, which forms the dietary basis for most beach-dwelling taxa [37], was generally higher on the beach on Veyvah. In contrast, on Dhidhdhoo, it peaked at the edge toward the inland habitat where the first pioneer plants appear ( Figure 1C). Here, pioneering supralittoral vegetation at the upper margin of the beach might act as a barrier holding back seagrass and other marine detritus that gets washed ashore during high tide or storm events, resulting in its highest availability not directly at the drift line, but at the margin of the inland habitat [38].
Despite the apparent gradient in environmental conditions from the drift line at the beach across the beach-inland transition zone and into the interior insular vegetation, the ground-associated fauna's overall richness, abundance, and diversity did not change significantly across the investigated transition zone on a species/genus level. However, summarizing the overall species distribution into larger taxonomic groups revealed that the ground-associated macrofauna community is changing completely along the beach-inland transition zone, albeit the diversity indices did not change significantly (Figures 3 and 4). This species turnover is in line with earlier findings that showed comparable turns in community composition in other highly dissimilar adjoining ecosystems [16,17].
Crustaceans (Coenobita spp., Ocypode sp., Pachygrapsus minutus, Talitrus sp.) dominated the beach environment and rarely-if at all-were observed dispersing over the margin of the beach across the transition zone into the vegetated supralittoral habitat. The gradually increasing soil temperature, which also indicates decreased soil moisture and salinity [36,39], toward the upper margin of the beach and the beach-inland transition zone likely forms a natural dispersal barrier for the beach-associated crustacean taxa, which depend on, for example, fine-grained, moist soil for burrowing [27,40]. Although terrestrial hermit crabs, Coenobita spp., are known to disperse further landward into densely vegetated mangrove or forest habitats [41][42][43], these results indicate that they do not reside in large numbers in the sparsely vegetated beach-inland transition zone, at least during the day, probably to avoid desiccation [44]. On the other hand, hexapods and their common predator, the house gecko Hemidactylus frenatus, were recorded rarely-if at all-on the beach side of the transition zone. Their natural dispersal range seaward might be limited by available supralittoral vegetation, which offers a heterogeneous microenvironment and serves as a shelter, and food and nutrient supply [45]. The absence of primary production and terrestrial detritus further seaward might additionally deter the herbivorous and detritivorous taxa from dispersing further into the beach side of the transition zone beyond the line of first pioneering plants [46]. This bipartite distribution pattern of crustaceans (beach) and hexapods (inland), following their natural dispersal limits along the beach-inland transition zone, would also partly explain why energy fluxes between beach and inland communities on small islands with sufficient primary production are limited [24].
Although the dispersal ranges of crustaceans and hexapods only marginally overlapped around the beach-inland transition zone, a third group of taxa was found on both sides of the transition zone: non-web-building spiders (Lycosidae, Philodromidae, Salticidae). Since vegetation structure and prey availability determine the distribution of spiders [47], it is likely that the observed wolf-, jumping-, and running-spider taxa are all adapted specifically to sparse vegetation structure, like as occurred along the beachinland transition zone in the studied system [48]. Jumping-, wolf-, and running-spiders can be fairly thermotolerant [48], allowing them to withstand the dynamic environmental conditions at the beach-inland transition zone. They are not dependent on vegetation as they do not build webs for hunting prey [47] and thus can advance further into the beach environment for foraging than other predators [49,50]. Therefore, other than the initial hypothesis, the beach-inland transition zone of small tropical islands is not an area of elevated species diversity where beach and inland taxa overlap, but instead, is primarily occupied by non-web building spiders. Their occurrence between the beach and inland fauna results in the observed consistent total taxa richness when shifting across the beach-inland transition zone on the investigated small tropical insular system.
A noteworthy feature of the observed biodiversity patterns across the beach-inland transition zone is the remoteness of the investigated system, the Maldivian archipelago in the Indo-Pacific. Many studies in various continental systems identified transition zones as regions with increased biodiversity due to overlapping distribution ranges of species and additional specialists that are adapted to tolerate the rapidly changing environmental gradients [11,12]. This pattern could not be observed in the present system. The overall remoteness of the Maldivian islands coupled with their small island sizes allows only a limited number of taxa to reach and successfully colonize these islands [51][52][53]. Furthermore, the absence of any transition-zone-specialists could be caused by the highly dynamic character of these small atoll islands, which display constant changes in their insular morphology [54,55]. When systems are dynamic in space and time, transition zones and ecosystem boundaries become less diverse with fewer specifically adapted taxa present than in more stable ecosystems [56]. Additionally, due to the Maldivian islands' overall young geological age (although the archipelago itself is about 55 million years old, the present Maldivian islands started forming ca. 3000 years ago after being fully submersed during the last glacial period), processes of speciation and adaptive radiation to tolerate the environmental conditions of the beach-inland transition zone might have not (yet) taken place. These factors probably result in a paucity of endemic and specialist taxa in the beach-inland transition zone of the investigated remote atoll islands compared to continental systems [54,57]. Taking the factors above together could explain why neither a highly diverse beach and inland fauna, nor any changes in taxa richness or abundance along the transition zone were observed. Instead, an overall low but consistent taxa richness across the beach-inland transition zone in the investigated system was recorded, albeit the macrofauna community composition changed completely along the transition zone.

Conclusions
The present study shows how taxa richness and diversity of the ground-associated macrofauna on remote tropical atoll islands in the Indo-Pacific are structured along the environmental gradients of the beach-inland transition zone. While the relevant environmental gradients that determine the biodiversity patterns of the beach and supralittoral communities have already been investigated on global to regional scales [58], this study demonstrated how the biodiversity of the beach and supralittoral macrofauna is structured on a small spatial scale. In light of globally accelerating habitat conversion, especially drastic in the coastal environment [59], these findings also provide new and relevant insights into community dynamics and diversity patterns of the coastal-and supralittoral-associated macrofauna.