Shifts in biodiversity and physical structure of seagrass beds across 5 decades at Carriacou, Grenadines

We examined the assemblage structure for all three years and all stations using multivariate analysis. Difference between years were analyzed using a one factor permutational multivariate analysis of variance (PERMANOVA) based on a Bray–Curtis similarity matrix generated from arcsine sqrt transformed data [50]. Year was a fixed factor and analysis used type III sums of squares. Non-metric multidimensional scaling (nMDS) ordination, based on the same Bray– Curtis similarity matrix, was used to examine multivariate patterns in the composition of the benthic assemblage among years and stations. Spearman’s rank correlation was used to exam- ine relationships between ordination scores and the frequency of occurrence of main func- tional groups (green algae, red algae, brown algae, corals, and seagrass). To test the hypothesis that variation in the sessile benthic community was related to variation in sea urchin abun- dance, we also used Spearman’s rank correlation for the ordination scores and examined the frequency of occurrence of Tripneustes ventricosus at each station. All multivariate procedures reported here were performed using Primer 7 with the PERMANOVA add-on [51]. Results Seagrass cover on transects Total cover by seagrass along 50-m transects (Fig 2) was higher at LWE sites (mean: 69%) than at HWE sites (48.2%) in 1969. By 2016, seagrass cover had declined more than 2-fold at HWE sites (21.4%), but there was little change at LWE sites (64.7%). An extreme shift occurred at one HWE site (T5) where the elevated, sand and pebble-based seagrass bed (83% cover) that

existed in 1969 had been completely scoured down to bedrock by 2016, with only a thin cover of shifting, coarse sand and T. testudinum persisting (21% cover) in sand-filled crevices. 2-way ANOVA showed a significant effect of wave exposure on total seagrass cover within transects (F1,16 = 11.5, P = 0.004) but no effect of year (F1,16 = 3.9, P = 0.065) or interaction between year and wave exposure (F1,16 = 2.7, P = 0.118). A similar result was obtained for total cover of a transect by the seagrass bed per se (which includes seagrass-free patches within beds). Given the unbalanced design, high variability between transects at HWE sites and the marginal test result for the effect of year, we conducted a paired-sample t-test to compare cover of the fring- ing seagrass bed between years (1969, 2016) within each wave exposure level. This showed a highly significant effect of Year (higher in 1969) at HWE sites (P = 0.003), but no effect at LWE sites (P = 0.771). There were no differences in cover of seagrass-free patches within sea- grass beds (Fig 2) across years or wave exposures (2-way ANOVA, p > 0.47). This indicates the pronounced decrease in the extent of the bed and the total cover of seagrass at HWE sites over time is related to the expansion of the seagrass-free inshore zone (SFI), which was not evi- dent at LWE sites. Mean cover by monospecific T. testudinum (Fig 2) changed very little at LWE sites between 1969 (33.8%) and 2016 (34.2%) but declined by 4-fold at HWE sites (from 19.5% to 4.3%)

although this difference was not statistically significant (paired samples t test, P = 0.132). In 2016 the invasive seagrass H. stipulacea was observed within 3 of the 4 LWE transects (Fig 2); at the remaining transect (T2) H. stipulacea occurred beyond our 50-m limit for analysis (at 93–100 m over the interval 0–180 m). H. stipulacea was absent from the 6 HWE transects. Sea- grass cover in pure stands and mixtures was highly variable between transects (Fig 2). Seagrass abundance and morphology at stations Quadrat observations at 17 stations in 1969, 1994 and 2016 provide indicators of changing conditions within contiguous seagrass beds (Fig 1). Ten of the stations (S2, S7, S8, S9, S10, S13, S16, S19, S23) were located within fringing seagrass beds and 6 stations (S3, S5, S6, S14, S24) were in ‘patches’ or ‘streaks’ or ‘cobble banks’ seaward of fringing beds. The remaining station PLOS ONE Seagrass bed biodiversity and structure and across 5 decades PLOS ONE | https://doi.org/10.1371/journal.pone.0306897 August 1, 2024 7 / 25