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

1969 or 2016. Also, 94% of more common species/groupings of associated of macroalgae (excluding Caulerpa species) and both coral taxa were most frequent in 1994 compared to 1969 and 2016, and multivariate analysis showed that community composition was less similar between 1994 and 1969 or 2016 than between 1969 and 2016. Ordination analysis showed that these patterns in seagrass community composition across years may be related to an inverse pattern in the frequency of occurrence of sea urchins (Tripneustes ventricosus group), which was minimal in 1994 compared to 1969 and 2016. The importance of sea urchins as major grazers of seagrass and macroalgae is well established [67, 68] and some species are known to graze coral recruits [69–71]. Sea urchins were intensively harvested in Grenada (less so in Car- riacou) for export in the early 1990s, leading to the collapse and closure of the fishery by 1995 [72]. Disease or mass mortality from other causes, reported at other Tropical Atlantic locations in the 1990s [73, 74], may have contributed to the low abundance of sea urchins that we observed in 1994. Herbivorous fish also are important grazers of macroalgae in the seagrass beds [75] and increased fishing pressure [37] may have amplified the effect of decreased sea urchin abundance in seagrass beds. The frequency of occurrence of Caulerpa species did not follow the trend for other macroal- gal species: the highest frequency of occurrence of the more common 4 of 6 species of Caulerpa occurred in 1969 and 2016, when sea urchins were relatively common. Secondary metabolites

in Caulerpa species (notably caulerpenyne) that could limit urchin grazing [76] may account in part for this discrepancy from the general inverse pattern of macroalgal and sea urchin abundance across sampling years in our study. Caulerpa prolifera was the most common spe- cies, except in 2016 when C. scalpelliformis [77] was more frequent. C. scalpelliformis, an Indo- Pacific species first reported in the Caribbean at Barbados in 1966 [78] and Antigua in 1967 [79], was absent at Carriacou in 1969 and rare in 1994 but had expanded to over a third of the stations by 2016, including 5 stations where it co-occurred with C. prolifera. Erosional features and expansion of the inshore seagrass-free zone The greater extent of the seagrass-free inshore zone (SFI) and the higher frequency of erosional scarps and blowouts on transects across fringing seagrass beds at HWE sites compared to LWE sites are consistent with studies showing that, in the absence of significant anthropogenic disturbance, variation in structural features and cover by seagrass is related to differences in wave energy or current speed required to initiate sediment movement, combined with factors that affect erodibility of sediments [80–82]. At HWE sites, the average extent of the SFI more than doubled between 1969 and 2016 (an increase of 19 m) and for the 2 of 3 transects also measured in 1994 (T6, T7) most of the increase occurred between 1994 and 2016. As our transects began at the MLW mark, and not at a geographically fixed position, landward recession of the MLW contour due to erosion

could account, at least in part, for an increase in SFI. All HWE sites are on shorelines without any man-made structures (e.g., walls or dwellings) within 50 m of shore and where there has been no attempt to reduce shoreline erosion. Cambers [83] found that beach erosion on 5 vol- canic islands in the Eastern Caribbean averaged 0.65 m y-1 between 1985 and 2000. In Carria- cou, Fitzpatrick et al. [84] estimated an erosion rate of 1 m y-1 between 1999 and 2006 at a sandy beach in Grand Bay (between HWE transects T8 and T9) where sand mining contrib- uted to erosion. Assuming a conservative erosion rate of 0.5 m y-1 between 1969 and 2016, landward recession of the MLW contour at our HWE sites would be 23.5 m, suggesting it could account for most or all of the increase in SFI, as opposed to net loss of seagrass at the inshore margin of beds. Average landward recession of the MLW contour associated with sea level rise alone from 1969 to 2016 (9.6 cm, based on a mean of 2.04 cm y-1 for 1950–2010 [85]) PLOS ONE Seagrass bed biodiversity and structure and across 5 decades PLOS ONE | https://doi.org/10.1371/journal.pone.0306897 August 1, 2024 17 / 25