*3.1. Coral Cover and Community Composition at Four Sites*

A total of 9521 corals from 51 genera were counted within the 1440 m2 of the 15 m × 2 m belt transects in 2015 and 2017. LIT surveys indicated that benthic composition was significantly different between a three-way interaction, where variables affect benthic composition in conjunction with each other, between year, site, and depth [three-way repeated measures PERMANOVA, pseudo-*F*(3,47) = 2.6888, *p*(perm) = 0.0109]. All sites, except Urban-W, were significantly different between survey years at 10 m while only Urban-W and Rural-N changed at 5 m [*p*(perm) < 0.05]. Dispersion, or variability, was also significant for year [*F*(1,46) = 5.9165, *p*(perm) = 0.0205], and site [*F*(1,44) = 6.7038, *p*(perm) = 0.0012]. Urban-W was significantly more variable in benthic composition than all sites [*p*(perm) < 0.05]. Live hard coral cover (henceforth coral cover) was significantly different with a three-way interaction between year, site, and depth [three-way repeated measures ANOVA χ2(3) = 18.6751, *p* = 0.0003]. In both years, Urban-W at 5 m had the lowest coral cover (mean ± SE; 4.8 ± 1.8% in 2015 and 4.5 ± 1.5% in 2017) and Rural-N 5 m (58.2 ± 1.7%) and Rural-N 10 m (56.9 ± 3.3%) had the highest coral cover, respectively, in 2015 and 2017 (Figure 3). Overall, coral cover was ~25% higher at the rural sites (37.3 ± 5.3%) than at the urban sites (12.9 ± 3.8%).

**Figure 3.** Benthic composition cover from 15 m line intercept transects by site (Rural-N, Rural-E, Urban-W, Urban-E) and depth (5 and 10 m) for November 2015 and June 2017 surveys in Timor-Leste. Major categories include: Hard Coral; CCA—crustose coralline algae; Invertebrates—mobile invertebrates; Macroalgae; Soft Coral; Substrate/Sand; and Turf algae.

Coral community composition, as measured by the abundance of individual coral genera from belt transects, also differed significantly by a site and depth interaction [three-way repeated measures PERMANOVA pseudo-*F*(3,47) = 3.2546, *p*(perm) = 0.0001]. Hard coral diversity, calculated using the Shannon diversity index of hard coral genera counted on belt transects, showed significant site and depth differences [three-way repeated-measures ANOVA χ2(3) = 16.1668, *p* = 0.0010]. Urban-W had the lowest coral diversity of all sites at 10 m but had comparable diversity at 5 m depth (Shannon index 1.7 ± 0.2; 18 ± 2 genera) (Figure S2). The maximum genus richness of 33 ± 2 was present at Rural-N with consistently high (>40%) coral cover in both survey years. This site also had significantly more tabulate [three-way repeated measures ANOVA χ2(3) = 31.5895, *p* < 0.0001] at both depths than all other sites with 21.1 ± 0.7 colonies per transect (*p* < 0.05). Remaining sites averaged less than five acroporid colonies per transect. There was a significant three-way interaction between site, depth, and year for branching acroporids [three-way repeated measures ANOVA χ2(1) = 7.8254, *p* = 0.0498]. Branching Acroporids decreased at all sites and depths between survey years except for Rural-N and Urban-W at 10 m (Figure S3). Although 51 genera were found across the belt transects over all four sites, only a few genera dominated the reef, namely *Porites* (2015 = 17.4%, 2017 = 13.0%), *Fungia* (2015 = 13.7%, 2017 = 19.0%), and *Montipora* (2015 = 12.9%, 2017 = 13.4%).

Community composition measured from belt transects was also significantly different with a year and site interaction [pseudo-*F*(3,47) = 2.1713, *p*(perm) = 0.0002]. Urban-E was the only site with comparable coral community between years, and all sites were significantly different within years [*p*(perm) < 0.05]. In Figure 4, sites were generally distributed along axis two of the PCoA with Rural-N most positively associated with tabulate acroporids, *Galaxea*, *Goniopora*, *Montipora*, and *Stylophora*, while differences between depth were aligned along axis one with more *Pocillopora*, *Platygyra*, and massive *Porites* corals on the 5 m transects. Dispersion, or variability in community composition was also significantly different by site [*F*(3,44) = 15.7770, *p*(perm) = 0.0001] and depth [*F*(1,46) = 5.6388, *p*(perm) = 0.0433]. Specifically, the dispersion (spread of points) was significantly greater at 10 m and highest at Urban-W and lowest at Rural-N (Figure 4).

**Figure 4.** Principal Coordinates Analysis biplot of square root transformed Bray–Curtis similarity matrix of the hard coral genera counted on belt transects. Empty and solid markers indicating 5 and 10 m depths, respectively. Color indicates survey year: blue-2015 and green-2017. Vectors represent top nine coral genera/morphology correlated to axes as calculated by Person's correlation. Abbreviations are coral genera/morphology as follows: ACRtab—*Acropora* tabulate; FUN—Fungiids; GAL—*Galaxea*; MONTI—*Montipora*; GONIO—*Goniopora*; PLA—*Platygyra*; POC—*Pocillopora*; PORmass—*Porites* massive; and STY—*Stylophora*.
