Sediment tolerance mechanisms identified in sponges using advanced imaging techniques

Sponge and sediment collection

I. basta (n = 5, representing a single evolutionary significant unit, see Andreakis, Luter & Webster, 2012, verified via examination of skeletal structure and morphology) was collected at 10 m depth near Fantome Island (S 18°41.028′, E 146°30.706′) in the inshore, central region of the Great Barrier Reef (GBR) in Australia, leaving at least one quarter of the remaining sponge to recover (Duckworth, 2003). The number of sponges collected was limited by to five due to logistical constraints in the field. All collections from the Great Barrier Reef were performed under Great Barrier Reef Marine Park Regulations 1983 (Commonwealth) and Marine Parks regulations 2006 (Queensland) Permit G12/35236.1 and Permit G13/35758.1. Collected I. basta were transported in flowing seawater to the National Sea Simulator (SeaSim) at the Australian Institute of Marine Science (AIMS, Townsville), and then cut into small explants, i.e., clones, (∼1 × 3 cm) and allowed to heal for >2 weeks in large holding tanks (5,000 L) with flow-through seawater (600 mL min⁻¹, 5 µm filtered) at conditions similar to the collection location (28 °C and 36‰ salinity).

Sediments were collected from two geographically diverse locations where I. basta is common: (i) Davies Reef, a mid-shelf reef on the central GBR (S 18°49.354′, E 147°38.253′) and (ii) a coastal habitat approximately 10 km offshore, near the township of Onslow in the Pilbara region, Western Australia (S 21°38.642′, E 114°55.924′). Sediment composition differed between collection sites, with Davies Reef sediment predominantly calcium carbonate (∼80% aragonite) and Pilbara sediment primarily siliciclastic (∼50% quartz) in nature (Ricardo et al., 2015). As such, the sediments are referred to as ‘carbonate’ or ‘siliciclastic’, respectively. Carbonate sediment were representative of sediments in offshore environments as well as sediment from specific dredging operations, including the Barrow Island in north-western Australia (see Bessell-Browne et al., 2017b). Siliciclastic sediment was representative of inshore environments as its composition was highly influenced by terrigenous discharge from a neighbouring river. The collected sediments were ground with a rod mill grinder to a mean grain size of 29 µm (range: 3–64 µm) and 18 µm (range: 2–42 µm) for carbonate and siliciclastic, respectively. Sediments were ground because fine and medium silts (<60 µm) stay in suspension longer and travel farther than other particles (Bainbridge et al., 2012), so these particles are predicted to have the largest impact in coastal areas across space and time during natural sediment resuspension events and during dredging (Jones et al., 2016). Furthermore, dredging and natural resuspension occur in both inshore and offshore environments, so it is important to understand the effects of different sediment types.

Acute exposure to sediments

To determine if passive accumulation of internal sediments could occur without active pumping by the sponge, initial experiments were undertaken to compare sediment accumulation in living, anesthetised, and dead sponge explants. Sponges were anesthetised in a solution of 7.5% MgCl₂ in filtered seawater (FSW) (Messenger, Nixon & Ryan, 1985) for 10 min prior to sediment exposure, resulting in closure of sponge oscula and cessation of pumping (see Strehlow et al., 2016a). To obtain dead individuals, sponges were euthanized by immersion in 100% ethanol for 10 min (Bell, 2004). Living, anesthetised, and dead explants (n = 3 per treatment) were exposed to elevated SSC (siliciclastic sediment, approximately 100 mg L⁻¹) for 48 h. Sediments were kept in suspension using specialised 115 L tanks, as previously described (see Bessell-Browne et al., 2017a; Bessell-Browne et al., 2017b; Pineda et al., 2017b). 3D X-ray microscopy scans were assessed qualitatively for the presence of sediment (see “X-ray microscopy settings and data quantification”). These comparisons showed that sediment accumulated within living explants only. For dead and anesthetised explants, sediment was only present on their surfaces, validating the experimental and analytical X-ray microscopy approach and confirming that water had to be actively pumped by the sponge in order for sediment to enter the aquiferous system.

After sediment uptake by live sponges was verified, a second, acute sediment exposure experiment was performed. The experiment used the same tanks and siliciclastic sediment, and tested three SSCs: high (135 ± 30 mg L⁻¹, mean ± standard error), medium (42 ± 8 mg L⁻¹), and control (0 mg L⁻¹), with three tank replicates per treatment, resulting in a total of nine tanks, each with five sponge explants. Sponges cloned from different ‘parent’ individuals were randomly distributed across the tanks and treatments to minimise the potential impact of genetic identity. Sediment was added every 6 h over 48 h, after which sediment addition was stopped and explants were observed for a subsequent three-week recovery period. During this recovery period, both high and medium sediment treatments returned to control levels. These SSCs correspond to levels that can be observed during dredging (Jones et al., 2015; Fisher et al., 2015) and following flooding and natural resuspension events (Bainbridge et al., 2012). Sedimentation rates, i.e., the rate at which sediments deposited on the bottom of the tank, was measured for each SSC treatment using SedPods (Field, Chezar & Storlazzi, 2013). Sedimentation rates were 1.29 ± 0.07, 1.03 ± 0.05, and 0.14 ± 0.01 mg cm⁻² d⁻¹ in the high medium and control, respectively. The deposition detected in the control treatment was comparatively low, and was likely caused by algal growth or the particulate waste produced by the sponges, i.e., not sediments. The natural orientation, i.e., erect and perpendicular to the substratum, of I. basta was preserved during all experiments. Therefore sediment deposition on the sponge surface was not equal to the sedimentation rates on the SedPods, and the effect of sedimentation on the sponges was presumed to be minimal. Daily light integrals were approximately 1.2, 3.0, and 6.5 mol photons m⁻² d⁻¹, respectively. However, as I. basta is a heterotroph, decreased light levels have no effect its physiology (Pineda et al., 2016). Therefore, due to its orientation and heterotrophy, the potentially confounding effects of light attenuation and sediment smothering were eliminated in I. basta.

Sponges were sampled 24 h after sediment dosing commenced and after 1 h, 24 h, 78 h, and 3 weeks (w) of recovery under control conditions and fixed for X-ray microscopy and SEM analysis. Three separate sponges were processed per time point, one from each tank. Oscula on each sponge at each sampling time, as well as after 3 h of recovery, were recorded as either open or closed where visible. Sponge explants contained between 3–10 oscula.

Chronic exposure to sediments

In the longer-term (chronic) exposure experiment, large (∼5 × 5 cm) I. basta explants were exposed to elevated SSCs (76 mg L⁻¹), reduced light (daily light integral: 0.15 mol photons m⁻² d⁻¹), and high sedimentation rates (5.2 mg cm⁻² d⁻¹) for four weeks using the carbonate sediment. In the control treatment, sponges were exposed to no suspended sediments, no sedimentation, and a higher daily light integral of 6.53 mol photons m⁻² d⁻¹, typical levels for clear water reef environment (Jones et al., 2015; Fisher et al., 2015). Experiments were performed in 4 × 1,200 L fibreglass tanks with 5 µm filtered seawater in an environmentally controlled room at the SeaSim. The chronic exposure formed part of a larger collection of studies (Pineda, Duckworth & Webster, 2016; Pineda et al., 2017a; Pineda et al., 2017b; Pineda et al., 2017c) using carbonate sediment. Sediment was kept in suspension using two recirculating pumps (see Pineda et al., 2017a). When sediment concentrations dropped, doses of concentrated sediment (∼6 g L⁻¹) were added automatically. The doses were monitored and controlled throughout the experiment by integrated nephelometers, which measured sediment levels, connected to a programmable logic controller (PLC) (see Pineda et al., 2017a). Two tank replicates were used for each treatment, making four tanks in total. At each time point, one sponge was taken from one tank and two sponges were taken from the corresponding tank replicate, making three replicates per time point per treatment and a total of 12 sponges. A piece (∼1 × 3 cm) was fixed for X-ray microscopy, as outlined below, after four weeks of exposure to both control and high sediment conditions from each replicate (n = 6). After four weeks, tanks with high sediment loads were returned to control conditions to allow sponges to recover. After 2 w of recovery, a piece (∼1 × 3 cm) was taken from each replicate (n = 6) as described above and fixed for examination using X-ray microscopy. For a full account of the methods and specifications of this experiment see: Pineda et al. (2017b).

X-ray microscopy settings and data quantification

Sponges were fixed in 2.5% glutaraldehyde and 1% paraformaldehyde in FSW for 1 h at 23 °C and stored at 4 °C. Sponges were stained in diluted Lugol’s Iodine (I₂: 1%, KI: 2%) for 24 h, to increase the contrast of tissue elements in X-ray scans. Three-dimensional X-ray scans were performed using an Xradia Versa XRM520 X-ray microscope, using the following specifications: ∼5 µm pixel size; voltage: 80 kV; power: 7 W; exposure time: 12 s; and binning: 2. Scanning time was approximately 6 h per sample. Projections were reconstructed using XMReconstructor software (Xradia Inc., Pleasanton, CA, USA).

Avizo Fire software (FEI, Hillsboro, OR, USA) was used to extract quantitative data from the 3D reconstructions. The Edit New Label Field tool was used to segment scans. Due to the porous nature of sponges, the sponge tissue could not be segmented out from the background using thresholding. Instead, the lasso selection tool was used to manually select the internal edge of the pinacoderm (outer layer) on 2D X–Y cross sections. Selections were performed every 50 cross-sections (∼250 µm), and then the interpolate function was used to create a 3D selection between the two selections. This process was repeated across 800 cross-sections (∼4 mm) to select the internal volume of the sponge. This procedure also served to exclude sediments on the surface of the sponge since the concentration of these sediments was highly variable due to handling effects and mucus production (Pineda et al., 2017a). Internal organisms, such as polychaetes and barnacles, were manually excluded from analysis using the lasso selection tool. The internal sediment, within the selected volume, was then segmented via the Interactive Thresholding tool, and the number of sediment particles and sediment particle volumes was quantified using the Label Analysis tool. Sediment particle diameter, was estimated assuming that the particles were spherical. Sponge tissue and skeletal elements were also segmented, labelled, and quantified within the selected volume. The remaining volume in the selection was labelled as aquiferous system because it matched threshold levels of the background solution, and was quantified using the same approach. The number of sediment particles in each sample was standardised to the selected volume, i.e., aquiferous system volume plus tissue and skeletal volumes. During tissue regression in I. basta, the choanocyte chamber density decreases and the tissue itself appears denser (Luter, Whalan & Webster, 2012b), indicating that the volume of the aquiferous system channels had decreased. The potential for rapid recovery from this regressed state implies that there is little change in the total number of sponge cells present (Luter, Whalan & Webster, 2012b). Therefore, in order to quantify tissue regression, the percent of the total volume occupied by the aquiferous system was calculated. Due to the pixel size, the aquiferous system volume analysed is more representative of the channels of the aquiferous system, i.e., incurrent channels, excurrent channels, atrium, etc., than of the choanocyte chambers, which are difficult to visualise at this resolution and within these relatively large samples.

Sediment particle position was determined by examining cross sections of the 3D volumes. A random cross section was selected for each sample. If sediment particles appeared tightly surrounded by sponge tissue, they were considered to be in the mesohyl. Particles present in the channels were also counted. The percentage of total particles in the mesohyl and in the channels was determined for each sample and averaged for each collection point to determine sediment position.

SEM analysis

Fixed sponges were dehydrated in increasing concentrations of ethanol using a PELCO Biowave microwave fitted with a PELCO Coldspot, then immersed in liquid nitrogen and fractured using a chisel (Johnston & Hildemann, 1982). This fracturing technique revealed the full aquiferous system including the choanocyte chambers. After fracturing, samples were returned to anhydrous ethanol, critical-point dried, and then mounted on stubs using carbon tape. Mounted samples were coated with 10 nm of gold and 10 nm of carbon to ensure sample conductivity. Images were acquired at 5 kV using a Zeiss 55 field emission scanning electron microscope. After 24 h exposure, all samples from the acute exposure experiment were assessed qualitatively for the presence of sediment. One control sample from each time point was also mounted without being freeze fractured in order to view the outer surface. Ostia, incurrent pore diameter, and surface area was determined from images using the ImageJ software package (Rasband, 2012).

Statistical analyses

Linear models were fitted to predict internal sediment variables (i.e., sediment particles mm⁻³ and sediment particle volume) and the percent volume of the sponge aquiferous system in terms of time and treatment (both as fixed factors). Control sponges had no internal sediment at any time, so control treatments from internal sediment variables data were excluded from the models. Internal sediment data were then compared between sediment treatments and over time. The percent volume of the sponge aquiferous system from control treatments was included in the models. These models were compared to linear mixed effects models with an additional random factor for tank, using the Akaike information criterion (AIC) in R (R Development Core Team, 2016). Based on the lowest AIC values, there was no indication that tank had an effect on any variable in either experiment, so the more parsimonious linear models, i.e., without tank as a random factor, were used (Table 1). In order to minimise genetic effects, sponge explants from each original replicate were randomly distributed between the tanks. However, due to the low number of original replicates (n = 5), there was still a chance of false positives in the models due the limited genetic independence of the explants. For models of sediment particles mm⁻³ over time and treatment, variances were weighted using a power law function within a generalised least squares (GLS) model for data from both acute and chronic exposure, in order to ensure homogenous variance of standardised residuals. Power-law weighted GLS models were the best fit according to AIC value for sediment particles mm⁻³ in both exposures. All other variables were modelled using linear models, i.e., analyses of variance (ANOVA). Data were arcsine transformed to model the percent volume of aquiferous system relative to tissue and skeletal volume over time and treatment in the chronic exposure, to meet the assumption of a normal distribution; otherwise raw data were used. Significance of model terms was evaluated using F-tests following the standard ANOVA approach. Tukey HSD was used to test for differences between time points and treatments. Welch t-tests were performed for each time point to identify differences in internal sediment mm⁻³ in each treatment level, i.e., medium and high sediment, in the acute exposure. Unless otherwise specified, reported values represent mean plus or minus standard error.

Acute exposure to sediments

After 24 h of exposure, sediment was present in I. basta from both high and medium SSC treatments, but was not evident in control sponges (Figs. 2–4). No sediment particles were evident in the control sponges at any time point (Figs. 2–4). Internal sediment particles per cubic millimetre of the total volume, i.e., of the sponge tissue and the aquiferous system, (sediment particles mm⁻³) increased significantly following exposure and then decreased during the recovery (P < 0.01, Table 2A), and there was a significant interaction of treatment (i.e., sediment level) with time (P < 0.05, Table 2A). After the first hour of exposure, the number of sediment particles in the high treatment sponges (180 ± 61 particles mm⁻³) was double that of the medium treatment (90 ± 32 particles mm⁻³), however this was not significantly different (P > 0.05). After 48 h of exposure and 1 h of recovery, sponges in the high and medium treatments had accumulated 500 ± 218 and 435 ± 118 particles mm⁻³ respectively, and again there was no significant difference between the two sediment treatments (P > 0.05, Table 2A, Fig. 2). Two dimensional (2D) cross-sections, of 3D X-ray microscopy images for all treatments are shown in Fig. 3. These cross-sections show the sponge skeletal and aquiferous system elements as well as the mesohyl and any internal sediments. Representative 3D X-ray microscopy images, thresholded to show only internal sediments are shown in Fig. 4. Full 3D renderings of sponges and sediments are presented in Movies S1–S5 .

After ∼24 h recovery (78 h after initial sediment exposure), sediment levels inside samples from the high treatment (747 ± 146 particles mm⁻³) remained significantly higher than those of the medium treatment (39 ± 10 particles mm⁻³, Welch two sample t-test P < 0.05, Table 2A Fig. 2). After 3 w of recovery, little to no sediment was detected in either of the sediment treatments or the controls (13 ± 8, 21 ± 21, and 0 ± 0 particles mm⁻³ for the high, medium, and control treatments respectively).

All sediment exposed sponges (n = 3 per time point) closed all of their oscula within the first 24 h of sediment dosing in the high and medium treatments, but re-opened them at differing times. In the medium treatment, oscula reopened 1 h after sediment exposure ended, i.e., 49 h after the start of the experiment. Whereas, in the high treatment, oscula stayed closed for 3 h after sediment exposure ended. Control sponges did not exhibit closed oscula at any of the sampling times.

The size of sediments (volume: 41 ± 10 µm³, diameter: 4.2 ± 0.8 µm) inside the sponge changed very little over time for either sediment treatment. Although there was a significant effect of time on sediment size (P < 0.05 Table 2B), this effect was likely caused by the decrease in particle volume in the medium treatment after 3 w (Fig. 2B). However this decrease was not significant in the Welch t-test (P > 0.05), and no other post hoc comparisons were significantly different. The percentage of sponge aquiferous volume relative to tissue and skeletal volume similarly did not significantly decrease (P > 0.05, Table 2C) in any treatment, remaining at ∼56% throughout the experiment and recovery period for all treatments (Fig. 2C).

In the high sediment treatment, internal sediment was mostly located (93%) in the mesohyl of sponges after 24 h. After 48 h of exposure and during recovery, most internal sediment (61–93%) was found in the channels of the aquiferous system. In the medium sediment treatment, the majority (>65%) of internal sediments were located in the channels of the aquiferous system at all time points.

SEM revealed that all sponges, including those not exposed to sediments, produced considerable amounts of mucus on their outer surfaces (Figs. 5A and 5B). Rates of mucus sloughing and production were not determined in this study. Mucus in the control treatment appeared to trap and facilitate any fouling on the sponge surface or pinacoderm (Fig. 5A). The mucus in the siliciclastic sediment treated sponges also appeared to trap and facilitate the removal of sediment particles on the sponge’s outer surface (Fig. 5B). For the siliciclastic sediment, no major differences were observed in choanocytes of sediment-treated and control sponges (Figs. 5C and 5D). In sediment-treated and control sponges, choanocyte chambers were present and not clogged with sediments. Some sediment-like particles were observed in the channels of the aquiferous system and choanocytes, but internal sediment levels were not quantifiable. The ostia of some sponges exposed to sediment were clogged with sediment particles (Fig. 5F), whereas the ostia of control sponges were always clear (Fig. 5E). In general, ostia (n = 15) in control sponges were found to be ellipsoid with an average diameter of 20 ± 1 µm and an average area of 268 ± 34 µm³.

Chronic exposure to sediments

Following chronic exposure to carbonate sediments, no sediment was detected in control sponges. There was accumulation of sediment, with high internal sediment concentrations of 185 ± 165 particles mm⁻³, in the treated sponges after 4 w of exposure to elevated suspended sediments, light attenuation, and increased sedimentation. Internal sediment concentrations decreased to a very low level of 23 ± 13 particles mm⁻³ after 2 w of recovery in control conditions (Fig. 6A); however, this reduction was not statistically significant (P > 0.05, Table 3A).

The average amount of sediment particles mm⁻³ in sponges from the chronic treatments after 4 w of sediment treatment (185 ± 165 particles mm⁻³) was approximately 63% lower than the highest observed value in the acute treatments, despite the longer exposure time. The average particle volume of internal sediments in the chronic exposure was 62 ± 23 µm³ (diameter: 5.0 ± 1.3; Fig. 6B), and it did not significantly change over time (P > 0.05, Table 3B).

The percent volume of the aquiferous system (relative to the tissue and skeletal volumes) in the high sediment treatment (56 ± 3%) was significantly lower after 4 w exposures than that of control treatments (75 ± 5%) (Tukey HSD: P < 0.01, Table 3C Fig. 6C). The percent volume of the aquiferous system did not fully recover to control levels by the end of the recovery period, returning to 63 ± 3% (P = 0.075). After 4 w of chronic exposure, the majority of internal sediments appeared within channels (66 ± 33%). After the 2 w recovery, the majority of internal sediments appeared in the mesohyl (75 ± 20%).