Prior to the 2017 hurricanes, six shallow (8-15 meter depth) reef sites had been selected from the Virgin Islands Territorial Coral Reef Monitoring Program’s (TCRMP) permanent monitoring sites to study variation in sponge communities in St. Thomas, U.S. Virgin Islands. These sites included Black Point (N18° 20.665’, W64° 59.107’), Coculus Rock (N18° 18.734’, W64° 51.613’), and Magen’s Bay (N18° 22.459’, W64° 56.077’), which are in embayments with heavily developed watersheds. Buck Island (N18° 16.717’, W64° 53.925’) and Savana Island (N18° 20.437’, W65° 04.939’) are located near undeveloped offshore cays. Botany Bay (N18° 21.433’, W65° 02.071’) is a nearshore site in a bay with a low level of watershed development.

For this study, we used three randomly selected transects out of the six permanently established 10-meter TCRMP transects at each site. The same three transects at each site were re-surveyed repeatedly in August 2016 (pre-hurricanes), December 2017 (10 weeks post-hurricanes), March 2018 (24 weeks post-hurricanes), November 2018 (61 weeks post-hurricanes), and July 2019 (93 weeks post-hurricanes).

Sponge assemblages were surveyed on each transect using multiple measures. Sponge density was quantified by a diver in situ by counting every sponge individual (i.e., ramet) within 0.5 meters (m) of each transect (resulting in 0.5 x 10 m belt transects) for all sites except Black Point in 2016, where 0.5 m on both sides of the transects (1 x 10 m belt transects) were surveyed. Density (sponges/m2) was calculated as the total number of individual sponges per transect, divided by the transect area (Gochfeld et al. 2020).  

At each site, sponge volume was calculated within permanent 1 square meter (m2) quadrats centered on the transect line within the initial and/or final meter of the transect. Sponge volume was calculated for 3-5 quadrats per site. As there were over 80 species of sponges within our survey areas, representing a wide diversity of morphologies, we used a standardized measurement approach for all sponges, rather than calculating the true volumetric measurement for each sponge based on its actual morphology. Thus, sponges were essentially treated as cuboids. We used a flexible sewing tape to measure the longest dimension of the sponge, then one to several width measurements perpendicular to the initial length measurement, and one to several height measurements, as needed to represent the shape and dimensions of each sponge. Multiple measurements for each dimension were averaged and length x width x height was calculated. For large tubes, of which there were relatively few, we subtracted the dimensions of the interior cavity from the exterior dimensions of the sponge. Generated data were in centimeters cubed (cm3) of sponge per meter squared (Gochfeld et al. 2020).

The percent cover of sponges was determined from videos of the three transects at each site following established methods from the Virgin Islands Territorial Coral Reef Monitoring Program (Smith et al. 2016). A diver swam at uniform speed while videoing the substrata from a height of approximately 0.4 m (the height of a guide wand). Consecutive, non-overlapping images, each approximately 0.64 x 0.48 m in planar area, were captured for each transect, for an average of 21 images per transect. Twenty random points were superimposed on each image (average of 1282 points per site, per sampling period) and the benthic cover underneath each point was identified to the lowest identifiable taxonomic level and used in the calculation of percent cover by transect. Specific sponges were not identified in the benthic cover analysis and were instead grouped into the overarching category of “Sponge”. The number of points per image required to adequately characterize the percent cover of each of the benthic categories (sponges, hard corals, macroalgae, epilithic algal community [EAC; i.e., diminutive turf algae and other low complexity filamentous algal communities; Smith et al. 2016], non-living substrata, calcareous algae, cyanobacterial mats, gorgonians, zoanthids, and other/unknown living substrata) was determined by visual inspection of the running means. For all categories, the mean value stabilized at no more than 17 points per image per transect, indicating that the 20 points analyzed per image were sufficient to accurately reflect the percent cover at these sites (Gochfeld et al. 2020).  

To determine whether different sponge morphologies were differentially affected by hurricanes, each sponge from each transect was assigned to a broad morphological category (sensu Wulff 2006). These categories included excavating sponges (“excavating”); low relief encrusting sponges (“encrusting”); thicker cushions, massive, tube, vase, or other amorphous shapes of medium relief (“massive”); and upright, branching, and rope sponges (“upright”). These groupings differ slightly from those used by Wulff (2006) but represent the morphotypes found within our transects in St. Thomas. The percent of the entire sponge community represented by each morphological category was calculated as the number of individuals in each category divided by the total number of individuals for each transect and multiplied by 100 (Gochfeld et al. 2020).

To describe the sponge assemblages across dates and sites, all sponges recorded in each transect were identified to the lowest possible taxonomic level.  For those that could not be identified in situ, photographs and voucher samples collected into 90 percent ethanol were used for spicule preparations and sectioning and identified with the help of collaborator Dr. M. Cristina Diaz. For each transect, the abundance of each sponge taxon was converted into a proportion (number of individuals/total number of sponges in that transect) for analysis.