On the importance of spatial scales on beta diversity of coral assemblages: a study from Venezuelan coral reefs

Estimating variability across spatial scales has been a major issue in ecology because the description of patterns in space is extremely valuable to propose specific hypotheses to unveil key processes behind these patterns. This paper aims to estimate the variability of the coral assemblage structure at different spatial scales in order to determine which scales explain the largest variability on β-diversity. For this, a fully-nested design including a series of hierarchical-random factors encompassing three spatial scales: (1) regions, (2) localities and (3) reefs sites across the Venezuelan territory. The variability among spatial scales was tested with a permutation-based analysis of variance (Permanova) based on Bray-Curtis index. Dispersion in species presence/absence across scales (i.e., β-diversity) was tested with a PermDisp analysis based on Jaccard’s index. We found the highest variability in the coral assemblage structure between sites within localities (Pseudo-F = 5.34; p-value = 0.001, CV = 35.10%). We also found that longitude (Canonical corr = 0.867, p = 0.001) is a better predictor of the coral assemblage structure in Venezuela, than latitude (Canonical corr = 0.552, p = 0.021). Largest changes in β-diversity of corals occurred within sites (F = 2.764, df1= 35, df2 = 107, p = 0.045) and within localities (F = 4.438, df1= 6, df2 = 29, p = 0.026). Our results suggest that processes operating at spatial scales of hundreds of meters and hundreds of kilometers might both be critical to shape coral assemblage structure in Venezuela, whereas smaller scales (i.e., hundreds of meters) showed to be highly- important for the species turnover component of β-diversity. This result highlights the importance of creating scale-adapted management actions in Venezuela and likely across the Caribbean region.


INTRODUCTION
The importance of scales in ecology has been largely acknowledged for decades (Schneider, 2001;Mac 34 Nally and Quinn, 1998; Wiens, 1989). MacArthur (1972) and Levin (1992) assays deeply influenced 35 modern ecologists by promoting the view that ecological processes act at a variety of spatial and temporal 36 scales, and they generate patterns that may differ from those at which processes act (Chave, 2013). Today 37 it is known that ecological dynamics tends to be stochastic at small scales, but variability is conditional 38 on the resolution of description (Chave, 2013; Levin, 1992). Furthermore, there has been an increased 39 recognition that the problem of scale at which ecological processes act, should be considered as critical if 40 it is wanted to produce general predictions about patterns in space and time (Chave, 2013). Thus, modern 41 ecological thinking agrees that to understand a system (e.g. a community), it is important to study it at the 2010). Generally, the presence or absence of organisms within a community may depend on rare or 48 large-scale (region-specific) dispersal and colonization events, while local abundance is more a function 49 of frequent, fine-spatial scale processes such as biotic interactions and habitat heterogeneity, e.g. Ricklefs 50 (1987). This implies that communities are structured by both abiotic and biotic factors nested along 51 different spatial scales which often occur along environmental gradients (Johnson and Goedkoop, 2002; 52 Whittaker and Heegaard, 2003). Concomitantly, the species richness of a community is also expected to 53 be highly dependent on spatial scales evaluated (Barton et al., 2013;Field et al., 2009;Melchior et al., 54 2017; Whittaker et al., 2001). 55 Coral reefs are one of the most complex and diverse ecosystems of the planet. Reef species diversity 56 has been estimated at 600,000 to more than 9 million species worldwide (Plaisance et al., 2011;Reaka-57 Kudla, 1997). The habitat and shelter for the majority of these species is largely provided by scleractinian 58 corals (Alvarez-filip et al., 2009). There is compiling evidence indicating that ecological processes 59 controlling the structure of coral assemblages (e.g. substrate availability, recruitment, competition, and 60 herbivory) are strongly dependent on spatial scales (Pandolfi, 2002). In addition, oceanographic processes 61 which partly define the environmental setting of a reef are also extremely variable within habitats, across

67
Total species richness of a region, frequently named gamma diversity (γ), can be partitioned in two 68 components: (1) α-diversity (i.e., the number of species by site), and (2) β -diversity (i.e., the variation 69 in the species identities from site to site, sensu Whittaker (1960Whittaker ( , 1972). For decades, ecologist have 70 debated ways to estimate and interpret α and β -diversity; but in recent years, the study of β -diversity has whereas variation portrays variability in species composition among sample units within a given spatial or 78 temporal extent, or within a given category of a factor (such as a habitat type or experimental treatment).

79
On the other hand, Baselga (2010) partitioned the total β -diversity into two components: (1) nestedness, be surprising to find coral assemblages to be extremely variable across spatial scales in Venezuela. We 96 expected that greater changes in community structure and β -diversity of coral assemblages will occur 97 at scales of thousand of kilometers (i.e., between the eastern and western regions) and within sites (i.e., 98 hundreds of meters). This is because of existing contrasting environmental settings driven by upwelling  2003).The goal of this study was two-fold: (1) quantify spatial variation of coral assemblages from and β -diversity from hundreds of meters (sites) to hundreds of kilometers (region). The factor region 128 encompassed three levels (West, Center, and East); nested within region there were two/three localities, 129 four to seven reef sites within each locality and four 30m-long transects within each site, understood as 130 the operational unit.

157
With the PERMANOVA, we tested the spatial variability of coral assemblage structure (i.e., absolute 158 abundance of each species) from meters to hundreds of kilometers. The variance components were 159 estimated depending on each source of variation in the analysis following the procedures outlined by  found that 80% of the sites had a sample coverage greater than 0.7 (Table 1, Supp. 3), which is considered     Our study shows that coral assemblages in Venezuela are much variable within and between localities than 238 we originally expected. We found two fold higher variability at small to medium scales when compared to Manuscript to be reviewed at broad spatial scales (hundreds of kilometers). However similar to our study, they found a higher degree 256 of variability within sites highlighting the relevance of local ecological drivers (e.g. rugosity and wave 257 exposure) in structuring coral assemblages. Other studies have also taken into account the importance of 258 spatial scales for coral assemblages, but must of then have focused on total live coral cover (Murdoch and 259 Aronson, 1999) and total abundance of colonies (Hughes et al., 1999). 260 It is widely acknowledged that behind patterns are ecological processes that shape communities (Barton in touristic use for not all sites within these MPAs are exposed to the same human pressures.

283
In addition, experimental evidence shows that some coral species differ in resistance to environmental   factors producing rapid coral mortality in Cubagua has been shown to be deeply influenced by upwelling 319 events (Rodriguez and Croquer, 2008). Thus, our results indicate that upwelling alone is not sufficient to 320 explain the extremely variable nature of coral assemblage.

321
Patterns of beta diversity 322 We found that differences in coral species composition occurred at spatial scales of hundreds of meters 323 to tens of kilometers. Although it is known that β -diversity depends on the spatial scale at which it is 324 measured, opposite to our results, in most studies β -diversity is assumed to be homogeneous at small

359
In summary, coral assemblage structure in Venezuela is highly variable at different spatial scales but 360 within locality variability seem to be very important. The processes that could underlie these patterns are Program. We thank the support provided by Fundación para la Defensa de la Naturaleza (FUDENA), 373 Instituto de Estudios Avanzados (IDEA) and the Escuela de Ciencias Aplicadas del Mar (ECAM), as well 374 as José Cappelletto, Zlatka Rebolledo and Alejandra Verde who helped with the field work. We also thank 375 Rita Peachey and all the 39th AMLC meeting committee for their help to attend the conference.