Introduction

Holothurians, commonly known as sea cucumbers, are globally one of the most conspicuous organisms associated with coral reefs. These animals are exploited in bêche-de-mer fisheries, and their high value in Asian markets has encouraged global overfishing and associated declines (Conand 2004; Anderson et al. 2011; Purcell et al. 2013). Such declines are concerning as holothurian populations can take decades to recover from over-exploitation (Purcell 2010). Issues around over-exploitation of wild populations have led to research targeted at developing holothurian aquaculture (Han et al. 2016) or to approaches for improved management of their fisheries (Friedman et al. 2011; Plagányi et al. 2015). The intrinsic value of holothurians to the reef ecosystem itself, however, is less often acknowledged as a reason for concern or management. As bioturbators, holothurians offer ecosystem services that increase local productivity and may mitigate some of the impacts of climate change.

Bioturbation of sediments by holothurians releases nutrients trapped in the sediments to benthic ecosystems (Uthicke 2001a). While high densities of holothurians can reduce microalgal production (Uthicke 1999), availability of nutrients, such as ammonium released by holothurians feeding at natural densities, can enhance the growth of benthic algae (Uthicke 2001b), increasing the gross productivity of benthic reef communities (Uthicke and Klumpp 1998). This may be increasingly important as coral reefs degrade and shift to more algal-dominated systems (Hughes et al. 2003). Sediment digestion by holothurians may be responsible for up to 50% of the dissolution of calcium carbonate in reef systems (Schneider et al. 2011), an important process as the majority of calcium carbonate on coral reefs is stored in sediment (Gattuso et al. 1998). As suggested by others (Schneider et al. 2011, 2013; Wolfe et al. 2018), this process may also facilitate the growth of scleractinian corals, which are critical reef builders. Holothurian bioturbation also reduces stratification and nutrification of sediments (İşgören-Emiroğlu and Günay 2007) and can directly increase oxygen levels in the sediment (Hammond 1982). The ecological role of holothurians as bioturbators in reef environments is thus pivotal in facilitating the availability of nutrients and oxygen for other organisms. The scale of this bioturbation of sediments by holothurians, however, remains unclear for coral reefs.

Understanding the importance of holothurians to the benthic systems of coral reefs first necessitates a quantitative analysis of their bioturbation rates and an understanding of the scale at which bioturbation occurs in the system. Holothurians are unevenly distributed in reef systems (Tuya et al. 2006), which makes it difficult to determine their density at a reef scale. While previous studies have attempted to quantify the scale of bioturbation by holothurians in reef systems (Uthicke 1999; Wolfe and Byrne 2017; Hammond et al. 2020), these extrapolations were based on relatively small transect areas (100 m2) and small holothurian sample sizes (n = 12) that may not represent the wider heterogeneity of patterns found on reefs. Those studies that quantified distribution and abundance of holothurians typically used line transects or manta surveys (Uthicke and Benzie 2001a; Guzman and Guevara 2002; Friedman et al. 2011), which also have limitations.

The recent wide adoption of drones or unoccupied aerial vehicles (UAVs) (Anderson and Gaston 2013) for marine research provides a means to accurately map the distribution and behavior of many organisms in shallow aquatic environments (Raoult and Gaston 2018; Raoult et al. 2018), including holothurians that typically have high contrast against pale sediments. However, drones have never been used to assess holothurian abundance and densities, so there may be concerns such an approach would not reflect data obtained via other methods.

This study estimated the bioturbation rate of holothurians across a reef flat using drones and upscaled the estimation to an entire reef scale (minus lagoon) using geomorphic zones classified from satellite imagery. To achieve this, we assessed holothurian bioturbation rates ex situ for the dominant holothurian in Heron Reef. A proof-of-concept study was done to assess the accuracy of drones to measure holothurian abundance against traditional in-water line transect methods. Drone imagery was then digitized to determine holothurian densities in different reef geomorphic zones. Density patterns and their associated bioturbation rates were then extrapolated to the entire reef using classified satellite imagery. The methods developed here and the results from this study will elucidate the scale of effect that holothurians have on coral reefs and facilitate more accurate estimations of their ecological impacts and loss from fisheries.

Methods

Study site

Field surveys were conducted at Heron Reef on the southern end of the Great Barrier Reef (S 23.4423°, E 151.9107°) in September 2016 and February 2019 (Fig. 1). Heron Reef falls under different management zones including a Conservation Park, a Marine National Park, a Public Appreciation Area and a Scientific Research Zone (GBRMPA 2003). Unlike the northern two-thirds of the Great Barrier Reef, this reef was relatively unscathed in the 2016 and 2017 global bleaching events (Hughes et al. 2017, 2018), and the condition of the reef was generally considered healthy at the time of sampling. Holothurian densities inhabiting Heron Reef are considered representative of healthy reef environments in the Pacific, given the reef’s protected status and the lack of broad-scale impacts on this reef.

Fig. 1
figure 1

Map of Heron Reef a, b showing inner and outer reef flat geomorphic zones. c shows the placement of the eight 30-m-long snorkel and drone transects used for the proof of concept study c and the drone survey site (red outline) with the seven virtual transects used to speed digitizing. d shows imagery from one of the proof of concept transects. e shows holothurians digitized from the drone imagery

Full size image

Bioturbation experiment

Holothuria atra is the most common species of holothurian on Heron Reef flat (Williamson et al. 2017). This species is often found adjacent to reef bommies, on open sediment in shallow reef flat environments (Raoult et al. 2016) and in the lagoon (Madin et al. 2019). It is considered responsible for a substantial portion of bioturbation across Heron Reef. Feeding rates of H. atra are considered fairly constant over days to seasons (Uthicke 1999; Mangion et al. 2004). To assess the productivity of H. atra, a 24-h feeding experiment was run using flow-through outdoor aquaria at Heron Island Research Station. Individual flow-through aquaria (300 × 300 × 300 mm) were set up to receive a constant flow rate (1 L min−1) of sand-filtered seawater pumped directly from the adjacent reef. Twenty-seven holothurians were then collected from various locations on the inner to outer reef flat from the southern side of the island within the Scientific Research Zone, along with ~ 2 kg of the sediment on which they resided. This area is representative of similar habitat, which comprises 58% of Heron Reef (Fig. 1). Individuals were gently transferred from the reef to the aquaria in buckets filled with seawater to reduce stress that may affect feeding rates. Collectors were careful not to handle the holothurians more than necessary. Each 2 kg of sediment was carefully placed in an aquarium so that the upper surface of the sediment remained as upright as possible and allowed to settle for 10 min. After this, the holothurian associated with that sediment was carefully added. A light shade cloth was placed over all aquaria to mimic light penetration at their natural depth.

Individuals were held for 24 h and their fecal pellets collected every three hours as per Uthicke et al. (1999). Pellets were dried at 60 °C for 24 h and then weighed to the nearest milligram. The total amount of dry fecal matter produced after 24 h was assessed by adding all fecal collections per individual and averaging the data.

Proof-of-concept drone-based aerial holothurian counts

A proof-of-concept study was done to validate the use of drone imagery for holothurian counts. To assess the difference between holothurian counts acquired in water via snorkel compared to drone imagery, eight 30-m transect tapes approximately 50 m apart were deployed. Tapes were oriented perpendicular to the observed geomorphic zonation, with four validation transects each in the inner and outer reef flat zones (Fig. 1c). Two observers snorkelled along each validation transect and counted the number of holothurians within a one-meter distance either side of each transect tape. The observers then changed sides and travelled back down the transect, repeating the counts. The average of the two counts per observer was used.

Using a pre-determined flight path, we flew a DJI Phantom 4 Pro with a standard RGB camera over the survey area containing the eight validation transect tapes. We used a flight altitude of 20 m to ensure each transect tape was visible, and to achieve the spatial detail required to identify as many holothurians as possible. We used an overlap of 85% and sidelap between flight lines of 75%, taking care to fly at low tide in the afternoon to avoid sunglint and specular reflection at the water’s surface (Joyce et al. 2019).

Orthomosaics of the region were created using Pix 4D from the resultant photos. Mosaics of the eight transect tapes were then manually digitized with a 1-m buffer either side of the transect to extract imagery co-incident with the in-water counts. Three observers then independently and manually digitized the number of holothurians within each image transect. The difference (if any) between in-water counts and drone counts was then assessed with a linear mixed model using the lme4 package (Bates et al. 2014) in R V. 3.4.4 (Team 2013). The model was designed with holothurian density (individuals per m2) as the response variable, the method (drone or in-water) as the independent variable with an interaction with geomorphic zone and transect nested in zone as a random factor. Including observer as a random factor would have been ideal (accounting for between-observer variation), but as there were only two in-water observers, this was not possible. Moreover, some variation between observers for both methods is to be expected and should not prevent comparison between the two methods. If any significant effects were identified, the scale of the effect was determined using least-square means using the package emmeans (Lenth et al. 2018).

Estimating holothurian density across a large ‘reef-scale’ drone survey

The proof of concept showed that holothurians could be counted using drone-based aerial surveys to provide an (albeit conservative) estimate of abundance (see results). To apply the method over a larger area, an Aeronavics Bot Solo drone fitted with a Sony a7R DSLR camera (36 megapixel, pancake lens) was used to survey a typical cross-reef study site on the southern reef at Heron Island at an altitude of 60 m (Fig. 1c). Pix 4D was used to mosaic drone images and create an orthomosaic covering an area of 2.73 ha with a ground sampling distance (pixel size) of 0.8 cm.

The study site incorporated two geomorphic zones: the inner reef flat and the outer reef flat. The geomorphic zones were defined based on the classification of Dove satellite imagery acquired in January 2018, available in the Allen Coral Atlas (Kennedy et al. 2020). Given that geomorphic zones are defined primarily by their location and level of exposure to physical processes (Hopley et al. 2007), it is unlikely that the zones have changed in the 15 months between our 2016 drone survey and the capture of the satellite imagery used to define them here. We were unable to sample the lagoon or reef crest zones as they were beyond the range of the drone from the island and we did not have access to a boat at the time of survey.

ArcGIS 10.5.1 was used to digitize each visible holothurian in the orthomosaic. This was manual process is time-consuming, so we also assessed whether accurate counts could be achieved by digitizing only a subset of the imagery. To do this, “virtual transects” were placed across the drone orthomosaic. The study area was divided into seven 2-m-wide ‘virtual transects’ 10 m apart running the length of the study area (~ 350 m) (Fig. 1c). The number of holothurians within each geomorphic zone along these transects was calculated and divided by the area of each geomorphic zone in the transect to give a density of holothurians per m2 in each geomorphic zone. These results were compared to those obtained by manually digitizing holothurians across the entire study area, and the difference was negligible (Supplementary 1).

Upscaling holothurian densities and bioturbation rates across Heron Island Reef

Holothurian densities recorded in inner and outer reef flat geomorphic zones of the ‘reef-scale’ drone study area were upscaled using the total area of these zones across Heron Reef (Kennedy et al. 2020). This accounted for a total area of 1682 ha, or 57% of Heron Reef. Bioturbation rates based on the 24-h experiments for H. atra were calculated for densities of holothurians in each of the two geomorphic zones within the entire reef using the following formula:

$$ B = \sum \left( {A_{n} \times D_{n} \times P} \right) $$

where \(B\) represents the total holothurian bioturbation rate across Heron Reef in kg per year, \(A\) the area of a geomorphic zone \(n\) in m2, \(D\) the density of holothurians per m2 in geomorphic zone \(n\), and \(P\) the mean annual holothurian bioturbation rate in kg year−1. To account for any biases identified with drone counts relative to in-water counts, the densities of holothurians identified in the drone survey were scaled using the mean differences identified with the linear mixed model. To produce a conservative estimate for total bioturbation rates that includes the uncertainty identified in most of these values, total bioturbation across the reef was estimated in a Monte-Carlo-Markov-Chain framework using a custom R script (Supplementary 2) with 106 runs. These measures of uncertainty included standard deviations around the mean counts of holothurians in the larger survey (estimated from the coefficient of variation of the in-water counts), the standard deviation in bioturbation rates from the feeding experiments, and the standard deviation around the mean difference between drone and in-water counts for the reef flat.

Results

Bioturbation experiment

H. atra produced 38.24 ± 18.82 g (mean ± SD) over the 24-h period. Upscaling to an annual production of dry fecal matter per individual, one H. atra on Heron Reef was estimated (mean ± SD) to produce 13.96 ± 6.87 kg yr−1.

Proof-of-concept of drone-based aerial holothurian counts

A total of 29 paired in-water counts by two observers along the same transect at the same time were conducted. The highest number of holothurians counted on any transect was 37, and the lowest was zero. Paired in-water counts showed good, but not perfect, alignment between observers, with less than 8% discrepancy between counts (mean difference between observers 1.4).

When comparing the two methods (in-water and drone), densities of holothurians counted from drone imagery were significantly lower (df = 173, F = 25.8, p < 0.001) by 0.07 ± 0.01 (estimate ± S.E.) holothurians per m2 than those in-water (Fig. 2). Both methods showed that the inner reef flat had significantly higher holothurian densities (df = 16, F = 9.2, p = 0.023), approximately three times more than those measured in the outer reef flat. Tukey’s HSD post hoc tests found significant differences between drone and in-water densities counted in the inner reef flat (estimate =  − 0.13 ± 0.02, df = 73, t =  − 6.4, p < 0.001) but not between both methods in the outer reef flat (df = 73, t =  − 0.8, p = 0.82). The marginal R2 for this model was 0.55, with the conditional R2 (including the variance explained by the random factor) of 0.87, suggesting our model explained nearly 90% of the variation in our data.

Fig. 2
figure 2

Comparison of snorkeler-based transects to drone counts across the same transects

Full size image

Estimates of holothurian density across a large ‘reef scale’ drone survey

The total area surveyed in the reef scale drone orthomosaic was 27,348 m2. The spatial pattern of holothurian density was similar in the larger drone survey area to that found in the proof-of-concept study and the paired in-water survey transects (Fig. 2), with holothurian densities approximately 40% higher in the inner reef (0.2 per m2) relative to the outer reef (0.14 per m2) (Table 1).

Table 1 Number of holothurians counted via drone and snorkel surveys in each of the geomorphic zones within the study area
Full size table

Upscaling holothurian densities and bioturbation rates across Heron Island Reef

The total area of the two geomorphic zones across Heron Reef was 16.8 km2, as determined from satellite imagery (Table 1). Excluding the reef lagoon that was not surveyed, the mean rate of total holothurian bioturbation across Heron Reef as determined from Monte-Carlo-Markov-Chain was estimated at over 63,970 ± 4168 metric tons per year (mean ± S.E) (Fig. 3).

Fig. 3
figure 3

Posterior density distribution of modeled total bioturbation per year by holothurians across the inner and outer reef geomorphic zones of Heron Reef

Full size image

Discussion

This research shows that bioturbation by holothurians is a substantial contributor to sediment reworking on Heron Island Reef. Excluding the reef lagoon where holothurians also occur, holothurians were found to produce a conservative estimate of over 64,000 metric tonnes of bioturbated sediment per year across Heron Reef, or approximately 3800 tonnes km−2 y−1. Since holothurian densities are very sensitive to overfishing and population recovery is slow (Uthicke and Benzie 2001a; Uthicke et al. 2004), overexploitation of holothurians is likely to have long-term effects on coral reef sediment communities and the amount of organic carbon available in the water column for nearby organisms. In the current context of anthropogenic pressures on coral reefs (Bellwood et al. 2019), use of remote sensing techniques offers a means to rapidly assess densities of holothurians in shallow reef habitats to facilitate more accurate and targeted management decisions.

Measured bioturbation rates

Our results suggest the total amount of sediment bioturbated by holothurians is over 3800 tonnes km−2 y−1. This value is lower than the 4600 tonnes km−2 y−1 estimated to be bioturbated by H. atra by Uthicke (1999). Uthicke (1999) and others (Yamanouchi 1939) documented bioturbation rates of 67 and 86 g day−1 per individual for H. atra, respectively. Our study, and that of Klinger et al. (1994), documented bioturbation rates of 38 and 11 g day−1 per individual, respectively. Differences in these rates could be due to a selection of larger H. atra in the Uthicke (1999) and Yamanouchi (1939) studies, but this is difficult to tell as accurately weighing and measuring holothurians is problematic due to their ability to extend their bodies and hold varying amounts of fluid. Alternatively, our lower bioturbation rate may have been due to differences in our experimental design. While other studies often allow an acclimation period of approximately 4 h in aquaria prior to the start of their bioturbation experiments for the animals to settle (e.g., Uthicke 1999), our measure of bioturbation started as soon as the animals entered the aquaria. We did not run the experiment over longer timescales due to concerns that the holothurians would consume all the palatable sediment and adjust their feeding rates accordingly. However, this could have caused a reduction in bioturbation in the first couple of hours. Our study thus provides a very conservative bioturbation rate for H. atra over the 24-h experiment.

Environmental parameters such as in water temperature between studies could also partly explain differences in bioturbation rates. Mean seawater temperatures were reported as 27 °C for Uthicke (1999)’s Lizard Island study and 29 °C for Yamanouchi’s (1939) Palao Island study but 24 °C at Heron Island for this study and Klinger et al.’s (1994) research. Water temperature affects holothurian metabolism and overall performance, including bioturbation rates (Fraser et al. 2004; Wheeling et al. 2007; Schiell and Knott 2010). Differences in bioturbation rates between studies could also be caused by differences in organic matter in the sediments that make the sediments more or less palatable (Hammond et al. 2020). Rates of bioturbation do not only differ within studies on H. atra but also between species and over seasons (Wolfe and Byrne 2017) and potentially between years (Schiell and Knott 2010). Such variability in rates of bioturbation means that it is difficult to make a ubiquitous and conclusive statement on the amount of sediment reworked by holothurians across coral reefs. We advocate that studies should continue to assess bioturbation on a reef to reef and species-specific basis to account for such variability. Regardless of the mechanism, we are confident in our bioturbation rates for H. atra in this study due to the high number of individuals assessed and the continual monitoring of feeding rates for 24 h. Bioturbation rates from this study could be considered conservative and highlight the necessity for assessing rates at specific sites before extrapolation.

Monitoring holothurian populations with drones

Assessing the abundance and diversity of holothurians from drone imagery produced results lower than those recorded in situ with traditional methods. If the average in-water transect counts were upscaled, the total count of holothurians would have been 5,132,738 across the inner and outer reef flats at Heron Reef (Table 1). This is 41% higher than the total counted using the drone method (3,033,733 individuals, noting that these counts are not directly comparable as the reef-scale drone survey was conducted at a different time to the in-water transects).

The lower detection of holothurians by drone is contrary to other studies assessing count data of fauna via drones that generally find drones to detect higher numbers of organisms (Hodgson et al. 2018). However, drones cannot survey under coral or rock overhangs like a snorkeler can and the proclivity of some holothurians to coat themselves in sediment also hinders their detection in drone imagery. These factors likely led to the under-counts of holothurians by drones relative to snorkelers in the proof of concept transects. Distortion effects of the water column can also make visibility challenging in drone imagery but can be mitigated by flying at low altitude, using polarizing filters, choosing the lowest possible tide and selecting calm weather conditions to avoid ripples. Nevertheless, in shallow reefs, drones can cover much larger areas than traditional in-water monitoring and produce data that are re-examinable (Joyce et al. 2019). Holothurian monitoring programs could use this approach to produce assessments covering larger areas at a faster rate than traditional approaches. provided the under-estimation of holothurian abundance from drones is accounted for as we have here using a Monte-Carlo-Markov-Chain framework to incorporate measures of uncertainty in the detection of holothurians from drone imagery and the bioturbation rate. The most time-consuming element of the method is the manual digitization of each holothurian from the imagery. Our ‘virtual transects’ yielded almost identical densities to those calculated by digitizing holothurians across the entire area and substantially reduced the time required to digitize (Supplementary 1). This method is thus recommended for future studies though care should be taken to ensure sufficient ‘virtual transects’ are used. There is also the possibility of automating counts. For example, species identification and distribution assessments could be undertaken using machine learning (Dujon and Schofield 2019; Lyons et al. 2019). This would substantially accelerate the processing time and allow coverage of even larger areas. Moreover, with continual improvements in flight endurance and camera resolution (Crutsinger et al. 2016), we predict that the areas that drones can survey for holothurian monitoring will increase, allowing larger and more rapid mapping of holothurian densities.

While drone and satellite imagery robustly mapped H. atra density at a reef scale in this study, the biology, demography and ecology of the organisms to be mapped are important considerations in the efficacy of this method. The distribution of holothurians is typically linked with natural sediment features; however, the abundance and movement of holothurians can also be influenced by the life stage of the animal, sediment quantity and quality, light intensity, water temperature and depth (Sloan and von Bodungen 1980; Uthicke and Karez 1999; Dong et al. 2011; Morgan 2011; Navarro et al. 2013; Domínguez-Godino and González-Wangüemert 2020). As such, holothurians may be patchily distributed at scales finer than the geomorphic zones reported in this study and patchiness may be species-specific (Klinger et al. 1994). Knowledge of the ecology and biology of the organism to be mapped is thus important to the choice of scale and interpretation of drone-based population assessments. Drones give the ability to capture continuous high spatial and temporal resolution data over much larger areas than in-water methods and also enable finer-scale distributions of organisms to be monitored over time.

Implications for fisheries management

While H. atra, is the most common species on Heron Reef (Williamson et al. 2017), it is only a low value species in the bêche-de-mer fisheries (Purcell et al. 2010; Eriksson and Byrne 2015). The mapping method presented in this research could easily be translated to higher value commercial species where they occur in shallow reefs. H. leucospilota, H. edulis, S. hermanni and S. variegatus were also able to be identified in the imagery in this study. Many of these species are targeted for traditional fisheries in the Pacific (Drumm and Loneragan 2005; Friedman et al. 2011), with deeper-water species such as H. whitmaei and H. scabra more often targeted by commercial fisheries on the Great Barrier Reef and Pacific islands (Uthicke and Benzie 2001a, b). Our study suggests that if bêche-de-mer fisheries target only a few species, leaving dominant species such as H. atra present on reefs, the repercussions on coral reef ecosystems may not be as serious as the impacts of a broader, indiscriminate fishery. However the differing rates of bioturbation between holothurian species must be taken into account. For example, less abundant species targeted by fisheries such as Thelenota ananas are much larger (3–6 kg) than H. atra (< 100 g) (Purcell et al. 2016) and thus likely to bioturbate comparatively more. Studies should determine broader patterns of bioturbation in holothurians and relate this to the biology and ecology of each species. Changes in total bioturbation from holothurians in a reef could then be estimated and modeled as a result of specific species and population declines.

Implications to ecosystem functioning and reef health

The link between healthy reef systems and bioturbation rates of holothurians is largely unknown. In the context of holothurian overharvesting through bêche-de-mer fisheries, which can reduce holothurian abundances to a quarter of initial numbers for over 50 years (Holland 1994), associated bioturbation of benthic systems without holothurians could decline to less than a quarter of the levels found in a healthy reef systems. Direct extrapolation between total bioturbation, as determined in this study, and the associated benefits of bioturbation (e.g., algal productivity) is, however, not possible from currently available research. To our knowledge, no study has directly examined the link between holothurian bioturbation rates and flow-on benefits to ecosystems. There is evidence that a localized loss of holothurians causes a 63% reduction in O2 sediment penetration (Lee et al. 2017), but it is not clear how bioturbation rates relate to this. Future studies should aim to directly link bioturbation rates to ecosystem flow-on benefits to more accurately predict the effects of declining holothurian populations on coral reefs.

It is well accepted that coral reefs are projected to continue suffering substantial losses of reef structure and functionality from changing ocean conditions, including the dissolution of calcium carbonate due to ocean acidification (Albright et al. 2016; Doney et al. 2009; Johnson et al. 2014; Kornder et al. 2018; Shaw et al. 2015). The positive effects of biogenic buffering on carbonate chemistry by one species of holothurian has recently been documented (Wolfe et al. 2018). The mass of bioturbated sediments produced by holothurians on a reef scale, conservatively estimated here to be slightly higher than the mass of five Eiffel Towers (Castellaro et al. 2016) per year on Heron Reef, highlights the scale of the effect that these organisms may have as biogenic buffers against increasing dissolution of calcium carbonate. We advocate that the functional role of holothurians on coral reefs is highly likely to be more substantial than previously thought and that greater attention needs to be directed to their management and ecology, particularly in relation to overharvesting on reefs already compromised in resilience.