*3.1. Benthic Substrate, Coral Distribution, and the Status of Reefs*

The extent of the benthic categories varied significantly across the study sites (Figure 2). The mean percentage cover of benthic categories was compared between the sites on the Palk Bay islands and the sites on the northern coastal reefs in the Palk Strait. Nearly 50% of the substrate at the reef sites on the northern coast was composed of hard corals, while the remaining part was mostly composed of dead corals (RC and NIA).

In contrast, the substrate cover at the island sites was composed principally of dead coral rocks covered with turf algae, with the rest mostly consisting of hard corals, rubbles, and fleshy macroalgae. There was a significant difference in mean percentage cover of hard corals (HC) between the northern coastal reefs (*n* = 16, median = 45.31 ± 10.0 range) and the island reefs (*n* = 24, median = 25.31 ± 25.62 range) (Mann–Whitney *U*-test, *p* = 0.0001). In addition, the mean percentage cover of soft corals (SC) was significantly higher on the northern coast compared to the islands (Mann–Whitney *U*-test, *p* < 0.05). In contrast, there were no significant differences in mean percentage cover of fleshy macroalgae (NIA), of coral rocks covered with turf algae (RC), or of coral rubble (RB) between the northern coast and island reef sites (Kruskal–Wallis test, *p* > 0.05).

**Figure 2.** Radar plot giving an overview of benthic cover distribution on the coral reef sites on the Palk Bay islands (**left**) in comparison with those on the northern coast of the Jaffna Peninsula (**right**). (HC- hard coral, SC- soft coral, NIA- dead coral with macroalgae or nutrient indicating algae, SP- sponge, RB- coral rubble, RC- hard substrate of dead coral more than one year old covered by turf algae or encrusting coralline algae, SD- sand, SI- silt, and OT- others (including sea anemones, tunicates, gorgonians, or non-living substrate)).

Comparing individual sites, live coral cover was greatest at the Point Pedro (HC: 52% ± 7; SC: 4% ± 3), Inbarsitty (HC: 44% ± 3; SC: 3% ± 2), and Thondaimanaru (HC: 46% ± 6; SC: 2% ± 1) (±indicates SD) (Figure 3). The Valithoondal, Karainagar, Pungudutivu, and Kayts sites also had live coral cover of between 25% and 50%, corresponding to fair health status as defined here. Live coral cover was lowest (15–30%) at the five island sites Knr, Pu2, Kt1, Kt2, and Del (excluding Pu1). Punkudutviu and Kayts islands were characterised by the highest mean percentage cover of dead coral (61% ± 7 and 58.5% ± 10.5, respectively) followed by Delft reef front (46% ± 6) (Figure 3). Thus, a high live coral cover (49% ± 7.25) was observed on the northern coast and a low live coral cover (5%) on the island sites, except at Del (Figures 2 and 3). In all cases, the coral mortality index was greater than 0.33 (Figure 3), implying that all the reefs were stressed or impacted.

Figure 4 revealed two major groups of sites that cluster at a 60% similarity level. Sites Knr, Val, Ptm, Tho, Inb, Pu1, and Pu2 fall into the upper group, and Kt1 and Kt2 into the lower group, within which the Delft reef front site (Del) appears to be an outlier. There were three sub-groups in the upper group of the dendrogram, consisting of Inb, Tho, and Ptm (showing >85% homogeneity), Pu1 and Pu2 (>80% homogeneity), and Val and Knr (>70% homogeneity). Thus, all of the north coast sites plus two of the island sites (Pu1 and Pu2) fall into the upper group and the remaining island sites into the lower group, so supporting the distinction made above based on coral cover alone.

**Figure 3.** Percentages of live coral cover and dead coral cover, together with the values of the Coral Mortality Index (CMI) on coral reefs at each of the study sites.

**Figure 4.** Bray–Curtis Cluster analysis based on differences in the benthic cover on the reefs of the Jaffna Peninsula, together with a map showing the locations of the sites. (Abbreviations as shown on the map.)

Turf algae-dominated dead coral rocks (RC) were observed ubiquitously at the island reefs. The greatest extent of dead corals covered with turf algae (48.5% ± 13) occurred at Kayts island and of coral rubble at Pungudutivu island (31.5% ± 7). The Delft reef front site (outlier group) in the dendrogram was distinguished from the other sites by the presence of a very low percentage of live hard coral cover (16% ± 5 SE) and a high percentage of abiotic substrates, such as sand (31% ± 5), although the highest proportion of abiotic forms were observed at Pungudutivu (mean of RB 31.5% ± 7; SD 2.5% ± 2).

Valithoondal and Karainagar reef sites had similar amounts of soft coral cover, and dead coral rocks covered with macroalgae such as *Turbinaria* sp., *Caulerpa* sp., and *Sargassum* sp. (Figure 5, Supplementary Table S1). Valithoondal reef region had the highest live coral cover (45% ± 7) of massive growth forms, including various species (Table 1). At Point Pedro, Valithoondal, and Karainagar, numbers of giant clams, sea cucumbers, tunicates, and sea anemones were present (documented as others, OT) (Supplementary Table S1).

Soft corals were uncommon but were occasionally found on transects on the northern coast. No soft corals were recorded at Pungudutivu or Kayts islands, and only a few at Karianagar island.

**Figure 5.** Principal component analysis (PCA) showing the pattern of differences between study sites and how they correlate with different benthic components (HC- hard coral, SC- soft coral, NIA- dead coral with macroalgae or nutrient indicating algae, SP- sponge, RB- coral rubble, RC- hard substrate of dead coral more than one year old covered by turf algae or encrusting coralline algae, SD- sand, SI- silt, and OT- others (including sea anemones, tunicates, gorgonians, or non-living substrate)).


This distinction between reef areas was further explained by the principal component analysis (PCA) which illustrated the overall pattern of variation amongst sites and the dependence of this pattern on different benthic components (Figure 5). The analysis explains 58.8% of the total variance, of which principal component one (PC1) accounted for 32.3% of the variance and had a positive correlation (PC1 > 0) with the benthic substrates soft coral (SC), dead corals covered with macro algae (NIA), sponges (SP), sand (SD), and others (OT), while principal component two (PC2) accounted for 26.5% of the variance and showed a positive correlation (PC2 > 0) with the substrates rocks covered with turf algae (RC) and silt (SI). Accordingly, sites Pu1, Pu2, Tho, and Inb are assigned to the PC1 < 0/PC2 < 0 domain; sites Ptm, Val and Knr to the PC1 > 0/PC2 < 0 domain; sites Kt1 and Kt2 to the PC1 < 0/PC2 > 0 domain; and site Del to the PC1 > 0/PC2 > 0 (Figure 5).

The shallow areas (1 to 3 m depth) of reef front on the northern coast were dominated by macroalgae from October to April due to the Northwest Monsoon. The data were tested for any relationship between live coral cover (hard coral (HC) and soft coral (SC)) and algal cover (RC and NIA) across all sites, showing a significant negative correlation (r = −0.73, *p* < 0.05) between live coral cover and algal cover (Supplementary Figure S1).
