We examined filtration by sponge assemblages in the shallow waters (~2 m depth) of Florida Bay (Florida, USA), where water residence times are often high and filtration by dense communities of sponges were hypothesized to deplete the water column of food, primarily picoplankton and dissolved organic matter (DOM). We transplanted three sponge species into replicate locations that differed by an order of magnitude in natural sponge community biomass. Sponge transplants were clones, enabling us to control for sponge genotype effects across all sites. The growth of sponge clones was recorded seasonally for 18 – 30 months. Growth of transplants placed in areas devoid of sponges was 10 times greater than in areas with dense sponge communities, and three times greater than in areas with average sponge biomass. Sponge mortality was similar regardless of background sponge density. Measures of picoplankton, DOM, and PO₄ concentration confirmed an inverse relationship with sponge community biomass, whereas nitrogen concentrations in seawater were highest where sponge species replete with nitrogen-fixing symbiotic microbial communities were most abundant. This is striking evidence that filtration of waterborne resources by sponges in shallow, coastal environments can deplete those resources sufficiently to cause exploitative competition that limits sponge growth, with cascading implications for tropical hardbottom environments where sponges dominate the animal biomass.

Study Sites and Experimental Set-up

Our study sites were located in the shallow, coastal waters of the Florida Keys (Fig. 1), an archipelago south of Florida (USA) where seagrass and hard-bottom habitats prevail behind a 250 km long barrier coral reef. Hard-bottom covers ~ 30% of the shallow seafloor in the region (Zieman et al., 1989; Herrnkind et al., 1997) and is characterized by low relief, limestone bedrock overlain by a thin veneer of sediment (Schomer and Drew, 1982; Chiappone, 1996). Approximately 60 species of sponges occur in this back-reef region where their densities can exceed 80,000 sponges/ha with some individuals exceeding 1 m in diameter (Torres et al., 2006; Stevely et al., 2010, 2011; Butler et al., 2021). Depth at our study sites was ~ 2m, which is typical for the shallow seas surrounding the Florida Keys, and all sites were separated by a minimum of 0.5 km (Fig. 1). Water turnover and vertical mixing in this region of the Florida Keys is largely wind driven (Wang et al., 1994; Nuttle et al., 2003; Gilbert et al., 2009; Lee et al., 2016), except near channels where tidal mixing predominates. Our study sites were at least 3km  from channels andexperienced similar current velocities (3 – 12 cm/s) as measured at each site using a WaterMark USGS current meter (Model 6205) during spring tides in March 2016.

We initially began our study with two species of HMA sponge common to the region (Ircinia campana - Vase sponge; Spongia barbara - Yellow sponge) that were transplanted into sites with either medium or high natural densities of sponges where their growth was monitored for 30 months; 18 months into the study we added a third sponge species (Spongia graminea - Glove sponge). To control for genotype, large sponges of each species were collected from the seafloor and cut into 12-15 smaller pieces (~ 3500 cm³ each) that we attached to brick baseplates with plastic cable ties and numbered tags. These “clones” were left on the sea floor at the collection site for two months to heal and affix to the brick baseplates prior to relocation (Fig. 2). After the healing period, clones were out-planted equally onto six transplant sites that differed in natural sponge abundance (i.e., medium and high sponge abundance) to test for site-specific differences in the growth of sponge out-plants.

Prior to out-planting sponges, divers surveyed each site to document the density, biomass, and diversity of natural sponges that occurred on the sites. Following established methods for hard-bottom surveys (Butler et al., 1995, 2018; Herrnkind et al., 1997), divers identified, counted, and measured (height, diameter) all sponges > 20cm diameter within four, non-overlapping 25 m x 2 m belt transects positioned haphazardly on each site. “Site” is the replicate for the sponge density treatments, so an average from the data from the four transects/site was calculated to represent the values for each site. Those data were used to compare sponge diversity, density, and volume among sites (n = 3 sites per sponge density treatment). The volume of sponges counted on the seafloor were estimated based on their shapes and volume formulae those shapes (e.g., sphere, cylinder, tube, hollow cone, etc.). Prior to out-planting, we used the same measurements and formulae to estimate the volume of our clone out-plants.

To begin the experiment, individually tagged sponge clones (14 of I. campana and S. barbara) were transplanted in March 2016 in a haphazard arrangement within a 30m² area on each of the six medium and high natural sponge density sites. Sponge clones were remeasured approximately every six months for 18 months: August 2016, March 2017, August 2017. A year into the study we added another species (S. graminea) to the study and thereafter continued to monitor the growth of all three sponge species at the six study sites every 6 months for another 12 months (i.e., a total of 30 months of monitoring for I. campana and S. barbara; and 18 months of monitoring of S. graminea). We consider the growth measured each March as indicative of “winter” growth (Sept – March) and the measurements made in August as indicative of “summer” growth (April – August). During each measurement, sponges and the bricks to which they were attached were cleaned of epizoic organisms. Near the end of the study, we took advantage of a sponge die-off event that decimated over 90% of the sponges over a large (500 km²), thus allowing us to test sponge growth under a third treatment condition (low natural sponge density) into which all three sponge species were transplanted.  We monitored the growth of those transplanted sponges only once, 6 months after transplantation, because a second sponge-killing cyanobacteria bloom swept over the area again killing all of our transplanted sponges and precluding further measurements. Mortality of sponges was also monitored during the study.

Water Column Characteristics

To document the potential relationship between sponge effects on water column characteristics at each experimental site, we took water samples at each site in August 2017 at slack tide during a week in which prevailing winds were less than 5 kts. A total of 1 liter of seawater was collected in high-density polyethylene bottles 1 m from the seafloor by sampling 200 ml of water from five locations within each 30 m² experimental area on each site. Storage and preservation of water samples followed methods presented in McMurray et al. (2016). In the laboratory, water was filtered using pre-combusted, 7 micron GF/F filters and 100 ml aliquots were stored in centrifuge tubes at -20 °C for nutrient analysis. To quantify the concentrations of POM and DOM at sites of varying sponge densities, seawater was filtered through pre-combusted, 7 micron GF/F filters. Filters were wrapped in aluminum foil and frozen, and 20 ml of filtrate was stored in acidified (100 ul of 50% phosphoric acid) glass vials at -20 °C. Total dissolved nitrogen (TDN) was quantified using an Antek 9000N analyzer that was run in tandem with the Shimadzu TOC 5050 used to measure DOC. Particulate organic carbon and nitrogen (POC and PON, respectively) were measured using a CE Elantech NC2100 elemental analyzer. Nutrients (nitrate-nitrite, ammonia, phosphate) were measured with a Bran+Luebbe AutoAnalyzer III, and dissolved organic nitrogen was calculated as the difference between TDN and dissolved inorganic nitrogen (DIN).

In addition, 5 mL water subsamples were preserved in electron microscopy grade glutaraldehyde (0.1% final concentration), frozen in liquid nitrogen after 10 minutes, and stored at -80 °C for analysis of picoplankton concentrations.  Phytopicoplankton (picoeukaryotes, Synechococcus, Prochlorococcus) were enumerated using a BD FACSCelesta Flow Cytometer as previously described (McMurray et al., 2018) and aliquots of each sample were stained with Sybr Green-1 to quantify high nucleic acid (HNA) bacteria, low nucleic acid (LNA) bacteria, and viruses. Live particulate organic matter for each picoplankton cell type was estimated using standard cell conversions used in previous studies of sponge feeding (McMurray et al., 2018).