Using a thermistor flowmeter with attached video camera for monitoring sponge excurrent speed and oscular behaviour

Thermistor flowmeter construction and calibration

A four-channel thermistor flowmeter with integrated video camera was designed, constructed and validated in order to simultaneously measure sponge excurrent speed and observe sponge behaviour. The thermistor flowmeter was designed and built at the Australian Institute of Marine Science (AIMS) in Townsville with all testing done at the AIMS National Sea Simulator (SeaSim). The flowmeter incorporated four independent thermistor probes and a thermometer to determine ambient temperature. Each glass thermistor probe (120 series, 1,000 Ω at 25 °C, Honeywell, USA) was set to hold at 10 °C above ambient temperature. In high flow conditions, the probes drew more power to maintain this temperature difference hence power could be directly correlated to flow rate. This design is similar to thermistor flow meters used by Reiswig (1971) and Labarbera & Vogel (1976); however, in the current model, measurements were recorded digitally and four probes could measure flow simultaneously.

For calibration, each probe was attached to a carriage running along a rail on top of the tank and driven by a stepper motor through a belt and pulley system. The probes were run across the track at 8 different speeds from 5 to 200 mm s⁻¹. Three calibration runs were performed for each speed, and the power used to maintain thermistor temperature was averaged across the three trials. To confirm precision at different temperatures, the same calibration runs were performed between 24–29 °C (at 1 °C intervals). Curves were fitted to a polynomial regression model for power compensation developed by Moore (2003) using R64 (R Core Team, 2013). R-squared values for polynomial regression models for all four probes calibrated at four temperatures were greater than 0.99. A successive approximation method based on the calibration curve was used to back-calculate flow rate from power output. Custom designed software facilitated modification of the calibration equations according to temperature. The reliable range of the instrument was defined as 5–200 mm s⁻¹ with an accuracy of ±5 mm s⁻¹.

As a consequence of having the probes heated to 10 °C above the ambient temperature, no clear trend could be determined between ambient temperature and power, with curves from different temperatures intersecting at two points. Under identical flow conditions, differences in measured flow of 5 mm s⁻¹ could be recorded if temperatures changed by >1 °C. Experiments were therefore conducted at a controlled temperature for which a specific calibration curve was generated. As the logging software records changes in temperature to ±0.01 °C, experiments can be monitored effectively, quality controlled and were aborted if temperatures changed by ≥0.5 °C.

Sample collection and experimental setup

Small sponge explants (50 mm in diameter and containing several oscula) air drilled from large colonies of C. orientalis inhabiting (bioeroding) the skeleton of dead Porites sp. coral. Explants were collected at a depth of 3 m from reefs around Pelorus Island on the inshore Great Barrier Reef (GBR) (S18°32.903′ E146°29.172′, GBR Marine Park Authority Permit:G13/35758.1) and were acclimated to aquarium conditions for two weeks before experimentation. Aquaria consisted of 100 L tanks (Fig. 1) held at 27 °C. Seawater (5 µm filtered) flowed into tanks at 600 mL min⁻¹ (∼9 turnovers per day) ensuring abundant food and nutrients for the animals. Water in each tank was recirculated using an Iwaki MD magnetic drive centrifugal pump (Iwaki Pumps, Sydney, Australia) that collected water from the surface and forced it up through the centre point of the inverted pyramid at the tank’s base (Fig. 1A, IP) in conjunction with a VorTech MP10 pump (EcoTech Marine, Allentown, PA, USA) mounted inside the tank (Fig. 1, V). Hydra 52 LED lights (AquaIllumination, Ames, IA, USA) ramped up to 200 µmol photons m⁻² s⁻¹ from 5:30 to 11:30, held at the maximum for one hour and then decreased to 0 from 12:30 to 17:30. This natural daily pattern of light provided sufficient intensity for the phototrophic C. orientalis.

Observations of oscula behaviour, ES, OSA and Q

Waterproof USB endoscopes (7 mm USB waterproof endoscope; Shenzhen Dowdon Tech Co, Shenzen, China) aided in accurate positioning of the probes over the oscula (item C in Fig. 1). 3D flow patterns were measured for a single osculum. The probe was positioned in three dimensions around the osculum, with flow rates measured for 1 min per location. ESs across the osculum plane were integrated gravimetrically based on the area measured to determine the mean ES over the entire OSA (Reiswig, 1974). The ratio of the mean ES to the centreline ES (ES_centre), i.e., the ES recorded at the centre of the OSA, was 0.90, and this value was used as the profile correction factor (see Eq. (2)), which accounted for any decrease in speed at the edge of the osculum (Reiswig, 1974; McMurray, Pawlik & Finelli, 2014). Based on 3D measures of ES, the excurrent stream was considered to be laminar with a plug-like flow profile.

The OSA and centreline ES (ES_centre) were measured for all oscula (n = 10) on two different sponge explants (E1 and E2) for approximately 40 min per oscula. Reported measurements of ES should be considered as ES_centre unless otherwise specified. The OSA was based on the diameter of the osculum as measured by ImageJ (Schneider, Rasband & Eliceiri, 2012), with area calculated assuming a circle. Reduced major axis (RMA) regressions were performed using the lmodel2 package in R64 (R Core Team, 2013) to correlate OSA to ES_centre (sensu McMurray, Pawlik & Finelli, 2014). Additionally one-way Analyses of Variance (ANOVAs) were run to determine if the ratio of ES: OSA was different between oscula on the same explant.

At each time point, the instantaneous pumping rate (Q), i.e., “point rate” (Reiswig, 1974), was calculated as follows: (1)${\displaystyle Q=\mathrm{OSA}\times {\mathrm{ES}}_{\mathrm{centre}}\times \text{profile correction factor (Adapted from Reiswig, 1974)}.}$ RMA regressions were also performed to relate OSA to Q and ES to Q. This equation was used to approximate Q based on ES in future experiments: (2)${\displaystyle Q=5.42\times 10^{-3}{\mathrm{(ES)}}^{1.60}}$ Mean Q values for each osculum were summed to calculated specific pumping rate based on explant volume. Explant volume was approximated by measuring the surface area on the top of the core and the depth that the sponge eroded into the substrate and calculating the volume of the cylinder.

Effects of elevated SSC

If probes were exposed to elevated SSCs, a layer of sediment deposited on them over time. This layer significantly decreased the probe readings. To counteract this effect, probes were cleaned twice a day using small burst of water from a submerged transfer pipette. Two experiments were run exposing C. orientalis to elevated SSCs: (i) a single, short term exposure and (ii) a long term exposure with three pulses of sediment. Sediments were prepared from biogenic calcium carbonate sediment collected from Davies Reef (a clear-water, mid-shelf reef in the central GBR). Sediment was dried and ground using a rod mill grinder until the mean grain size was ∼29 µm with 80% of the sediment ranging in size between 3 and 64 µm, as measured using laser diffraction techniques (Mastersizer 2000; Malvern Instruments Ltd, Malvern, UK). Suspended sediments were added to the experiments gradually (over approximately five min) to reach a level of 100 mg L⁻¹. This level was chosen to reflect SSC within 500 m of dredging activities, where SSCs exceeded 100 mg L⁻¹ for periods of several hours (Jones et al., 2015).

In the short term SSC exposure, four explants (E3–6) were tested individually on separate days. Each of the four trials was independent, and the tanks were cleaned between trials. For consistency, each trial was conducted at the same time of day. To obtain ‘normal’ pumping rates (i.e., unaffected by sediment), excurrent speeds from each sponge were measured for 30 min. Each sponge explant was subsequently exposed to a single pulse of 100 mg L⁻¹ sediment and monitored for 3.5 h. Sediments settled quickly, with SSC decreasing to ∼50 mg L⁻¹ in the tanks after 1.5 h and to ∼40 mg L⁻¹ after 3 h. Temperature was held constant at 28 °C for explants 3 and 4, and 27 °C for explants 5 and 6. Q was calculated using Eq. (2).

In the long term SSC exposure experiment, ES and OSA were recorded for an additional four explants (E7–10) for 2.5 d before and after three separate sediment pulses of 100 mg L⁻¹, each applied over a 4 h period. Again, Q was approximated using Eq. (2). The diurnal pattern of C. orientalis pumping was also assessed. During the first 24 h of observation, diurnal pumping patterns were assessed based on 20 min moving averages of calculated pumping rates.

Since pumping rates oscillated regularly throughout the day, percentage change was recorded based on the local maxima (Reiswig, 1974; Reiswig, 1975). SSCs were monitored using a nephelometer (TPS, Australia), and nephelometric turbidity units (NTUs) converted to SSCs (as mg L⁻¹) by applying a sediment-specific conversion factor. Sedimentation rates were measured using SedPods (Field, Chezar & Storlazzi, 2013), which were capped after 24 h in the tank with deposited sediment filtered through pre-weighed 0.4 µm 47 mm diameter polycarbonate filters, incubated at 60 °C for 24 h, and weighed to 0.0001 g to determine sediment mass. Sedimentation rates during this experiment were ∼6 mg cm⁻² day⁻¹. Tank temperature was maintained at 27 ± 0.2 °C during the experiment.

Observations of oscular behaviour, OSA, ES and Q

The oscula of C. orientalis contracted and partially closed (Figs. 2A and 2B) on a regular basis. When an osculum closed completely the adjacent area also contracted (Fig. 2C), leaving a small indentation in the tissue. Interestingly oscula on the same explant were observed to undergo partial contractions asynchronously. If a single osculum was disturbed mechanically, it closed and ceased pumping within 2 min. If the whole explant was disturbed or moved within the tank, all oscula on the explant closed within two minutes.

When the oscula opened, the OSA, ES and Q increased. Once open, the osculum partially contracted regularly again (decreasing OSA), but it did not fully close (Fig. 3A). These partial contractions coincided with decreased ES. As Q is the product of OSA and ES, it follows the same periodic increase and decrease. The same partial contractions occurred when a different osculum was pumping actively (Fig. 3B). Again, Q followed a regular period. When a third osculum closed (Fig. 3C), the ES decreased proportionally with a lag of approximately one min.

Summaries of centreline ES, OSA, Q and partial contraction period for ten oscula of two separate explants are presented in Table 1. Factoring in all oscular flow rates, explants 1 and 2 had a mean specific volume flow (per litre of sponge tissue) of 0.020 ± 0.008 and 0.007 ± 0.002 L water s⁻¹ L sponge⁻¹ (mean ± SD). The estimated mean cycle times, i.e., the time required for all oscula to pump a sponge’s volume, were 41 and 151 s, respectively. During active pumping, the osculum partially contracted and then reopened following a regular period of 20.84 ± 2.86 min.

During partial contraction, full contraction, and reopening, ES was positively related to OSA (Eq. (3); R² = 0.431, P < 0.001, n = 1,335, Fig. 4A). On closer examination of the ratio of ES to OSA for each oscula, significant differences were demonstrated between values from O2 and all other oscula, O3 and O4, O4 and O1 in E1 (P < 0.05). The difference in O2 was due its active closure, whereas all other oscula were undergoing regular partial contraction (Table 1). There are also significant differences between O1 and O3 as well as O3 and O6 in E2. These differences probably account for the ∼50% of variation not explained by the RMA model. Q was positively related to OSA (Eq. (4); R² = 0.880, P < 0.001, n = 1,335; Fig. 4B) and ES (Eq. (2), R² = 0.946, P < 0.001, n = 1,335, Fig. 4C). Despite the variation caused by different oscula, both ES and OSA were strongly correlated to Q. The regression equations were as follows: (3)${\displaystyle ES=250.48\left(\mathrm{OSA}\right)+1.35}$ (4)${\displaystyle Q=539{\left(\mathrm{OSA}\right)}^{2.24}}$

Effects of elevated SSC

In the short term SSC exposure experiment, small temperature fluctuations did not have an effect on recorded ESs (Fig. 5). Immediately after exposure to the high SSC, three of the four explants decreased Q by 42–90%, and minimum pumping rates were reached after 15 min (Fig. 5). E3 started closing immediately after sediment exposure but partially reopened its osculum and restarted pumping at ∼50% of pre-treatment levels after ∼25 min. E4 exhibited a short burst of increased pumping, and then partially closed its osculum in response to SSC before reducing pumping rates to ∼50% of the pre-treatment level. After 80 mins of reduced pumping, both E3 and E4 closed their oscula and ceased pumping. E5 continued pumping in a “normal,” periodic manner for ∼20 min after exposure, then maximal pumping decreased by ∼50% and osculum closure and pumping cessation occurred after 180 min. E6 closed its osculum and pumping ceased after 30 min. All explants exhibited osculum closure and pumping cessation after treatment. On average, explants kept their oscula open for 99 ± 4 min (mean ± SD) after exposure to sediment. When explants reduced their pumping rates, partial contraction periods of ∼20 min were similar to those explants not exposed to SSC (see Fig. 3).

In the long term sediment dosing experiment, explants 7 to 9 increased pumping rates towards midday, reaching a maximum around the time of highest light intensity (Fig. 6), and then decreased into the night. Although the ∼28 min period of partial contraction of each oscula was consistent throughout the day, excurrent speeds and pumping rates varied during the day. Generally, ES and Q increased during the morning to peak approximately 1 h after midday (maximum light intensity), and decreased during the afternoon and were very low during night-time. During the night, pumping activity was markedly depressed and all oscula were either fully or partially contracted. Some pumping occurred at night, but maximal rates were <50% of those during the day.

On the second day, pumping rates were generally lower and the diel cycle was less obvious, although night-time speeds decreased for all individuals (Fig. 7). The first sediment pulse reduced excurrent speeds for all measured oscula (Fig. 7). Maximal excurrent speed decreased by 41%, 11% and 36% for explants 7–9, respectively (mean: 29 ± 9%). Calculated maximal pumping rates decreased by 57%, 27% and 51%, respectively, with a mean decrease of 41% ± 12% mL s⁻¹ (Fig. 7B). In contrast, the maximal excurrent speed of E10 increased by 7%, and pumping rates increased by 12%. All four explants closed their oscula following the first sediment dose after an average of 56 ± 67 min. E9 exhibited a very rapid closure, arresting pumping and closing after 6 min. After the first sediment pulse, all explants except E8, reopened their oscula and restarted pumping 68 ± 26 mins after closure, albeit at lower flow rates than before.

Maximal excurrent speeds after the second sediment dosing decreased by 59  ± 19% from pre-treatment maxima and pumping rates decreased by 71  ± 19%. However, variability was again detected between explants. E7 ceased pumping and closed its osculum after 70 mins. E8 kept its osculum closed. E10 closed its osculum after <1 min, and E9 kept its osculum open even after the second dose (although maximal excurrent speed and pumping rate decreased by 23% and 34%, respectively).

After the third sediment dose, all individuals except E9 closed their oscula and ceased pumping until nightfall. Pumping ceased through the night, in concordance with the diel pumping pattern observed before treatment. E9 kept its oscula open although excurrent speed and pumping rates were reduced by 39% and 55% respectively. Approximately 48 h after the first sediment treatment, E8 and E9 exhibited excurrent speeds and pumping rates similar to pre-treatment levels, and the diel pattern was re-established with maximal excurrent speeds at peak light intensity. A similar pattern was observed in E8, although excurrent speeds did not return to pre-sediment levels. Explant 10 did not restart pumping or reopen its osculum for at least 72 h post-exposure to sediment.