Monitoring Strategy for Analyzing the Impacts of Coastal Finfish Aquaculture on Seafloor Sediments in Indonesia

Study Area

Pegametan Bay, the target site investigated in this study, is located northwest of Bali Island, Indonesia (8° 8.55’ S and 114° 35.30’ E). The bay covers approximately 8.5 km² area and belongs to a coral fringing reef system that extends to the Bali Sea [6]. Multi-branched tidal channels with water depths approximately ranging between 5 m and 25 m surround the higher flats that are covered with water even at low tide. An outer reef barrier largely shields the inner reef area from the waves (Figure 1). Hydrodynamics are primarily driven by mixed tides ranging 1-1.8 m. In general, tidal currents in the channels are weakly developed. The channel to the west is wider and shallower, and the current velocities are extremely small, as they end on the coastline. Current velocities are mostly higher in the eastern channel than those in the western channel, as the former is narrower and open to the Bali Sea at both ends.

Click to enlarge

Three main depositional environments were observed in the bay. Mud predominates in the deeper channel beds, and medium- to coarse-grained carbonate sands are common in the higher flats. The seabed consists of muddy to sandy carbonate sediments with coarser coral debris. Muds often have a plastic consistency with low permeability resulting from consolidation processes that inhibit the flux of oxygen from seawater into the seabed. As sediment pores are poorly interconnected in muddy deposits, the free oxygen bound in the pore water is rapidly biologically respirated. Low redox potentials have been observed in seafloor sediments, while tidal currents and bioturbation in the bay are too weak to provide effective aeration of the sediments and oxic respiration systems. Moreover, oxygen deficiency is reinforced when organic waste settles onto the seafloor, as the oxygen demand for bacterial degradation of matter increases.

Marine Finfish Aquaculture Facilities

The northwest of Bali is a typical marine finfish aquaculture site for the cultivation of high-value commodities in operation in Indonesia. Most cultivated species are Asian seabass [7] and humpback grouper [8]. Finfish mariculture using floating net cages has been practiced in Pegametan Bay since 2001 and has grown to 30 farms in 2015. Figure 2 shows the locations of these fish farms. There are 28 traditional fish farms with assemblies of up to 380 rectangular net cages sized 3 m × 3 m × 3 m each. The nets are hooked in wooden rafts kept afloat by plastic drums, which ensure the buoyancy of the farming constructions. The stocking density of traditional farms used mainly for the cultivation of groupers is approximately 10–20 kg/m3. Additionally, there are two farm clusters in the eastern channel with 7–8 circular high density polyethylene (HDPE) floating units relatively close to each other used as nursery and for on-growing (see farms 21 and 30 in Figure 2). The cages are 20 m in diameter and have cage depths of approximately 6–7 m. The latter are used mainly for the cultivation of Asian seabass with stocking densities of approximately 25–30 kg/m3. Larger farms use formulated feed (pellets) as a diet, whereas small-and medium-sized farms use a mix of trash fish and pellets for feeding. Suitable areas for marine finfish farming are shown in Figure 2 (in green). Notably, most of the fish farms are well sited, reflecting the experience of fish farmers and planners [4].

Click to enlarge

Sampling Locations

Analysis of benthic impacts due to fish farming was performed in November 2015 and January 2016. The 12 largest farms in operation in 2015, which accounted for approximately 80% of the fish production, were selected for these assessments. This included farm 2, 11, 13, 15, 16, and 18 in the Western Channel and farms 20, 21, 23, 27, 28, and 30 in the Eastern Channel (Figure 2). Among the monitored fish farms, only Farm 11 was located outside areas suitable for marine finfish aquaculture activity. Table 1 provides information on sampled fish farms, sampling locations, and the number of sediment samples collected underneath the sampled fish farms. Three to five sediment samples per farm were collected from the traditional farms. Overall, 42 sediment samples were collected from underneath fish farms. The distance between sampling locations under the traditional fish farms was set to approximately 20–30 m. One sample was collected centrally below most circular cages of the farm clusters numbered 21 and 30 (see Figure 2). Peak

current velocities and water depths at sampling locations were obtained from three-dimensional model simulations covering a spring per neap tidal cycle from 6–20 Jan 2008 (two weeks) with tidal ranges varying approximately 0.7-1.8 m [9]. Water depths at farming locations are generally lower in the Eastern Channel, and the modeled current velocities exhibit significant differences. Peak current velocities in the Eastern Channel range approximately 0.09–0.22 m/s, whereas those in the Western Channel are ≤ 0.03 m/s.

Sediment samples were also collected at 14 reference locations (7 in each channel), as shown in Figure 2. Samples were collected to obtain representative information on undisturbed sediment conditions to assess the impacts of fish farms on the seafloor. Sampling locations were carefully selected in the open spaces between the fish farms to reflect conditions unaffected by farming. The minimum distance from a reference location to the nearest fish farm was chosen as 70 m to minimize potential farming influence [10].

Sampling Methodology and Analysis

In this study, the effectiveness of several sediment indicators was investigated for assessing the effects of fish farms on seafloor sediments. Sediment samples were analyzed for total organic matter (TOM; % of dry weight), particulate organic nitrogen (PON; mg/g), Eh (redox potential; mV), and total dissolved sulfide (TDS; mg/l). TDS, Eh, and pH in the pore water were measured on-site, whereas concentrations of TOM and PON were determined in the laboratory. The sampling methodology was as follows: At each sampling location, the divers initially documented visual evidence of benthic impacts through video recordings. The focus was on the sediment characteristics, evidence of bioturbation, and the presence of corals and algae. At the sampling locations, divers slowly pushed a PVC pipe of 8 cm diameter and 20 cm length approximately 15 cm into the seabed. The pipe was carefully closed with caps at both ends and brought vertically to the free surface. At the fish farms, the upper cap of the pipe was removed, and the overlying water slowly drained and purged. The smell and appearance of the sediment surface were recorded. Then, metrologically prepared probes recording TDS and temperature (ATM Germany H2S-sensor 15091001), pH (Meinsberg Germany EGA 142/PT1000), and Eh (WTW Germany SenTix® PtR) were introduced into the water saturated sediment core and fixed at approximately 3 cm below the sample surface until stable values were obtained. This depth was chosen to ensure that the invasive measurement in the sensitive surface layers was not disturbed by leaching of the oxidized surface water films. Then, the probes were cleaned and left for approximately 20 min to recover in a recipient with fresh seawater. Meanwhile, the upper 1 cm sediment layer was sampled at undisturbed sections of the core to determine the recent composition of the seabed sediment. Samples were then applied to a thin layer on a labeled PE film and exposed to the sun for approximately 15–20 min for drying. Next, the samples were tightly packed and shipped to Germany. In the laboratory of Kiel University, the samples were desalinated, dried, and homogenized. Each sample was split into three subsamples and the results averaged to compensate for possible homogenization errors. PON was determined by combustion/gas chromatography with a carbon nitrogen elemental analyzer (CN-Analyzer Euro EA, HekaTech Germany), whereas TOM was determined by loss of ignition (LOI) at 550°C [11] using a muffle furnace (Heraeus Germany TM M110T).

Visual Observation

The primary visual inspection of fish farming impacts at the site in Bali covered all the sampling locations. The presentation of the results of the study has been restricted to the samples collected at one traditional fish farm, that is, Farm 18 and at one of the farm clusters, that is, Farm 21 (Figure 2) for conciseness. The observations made under these fish farms and at nearby reference locations reflect the conditions of the aquaculture site reasonably well. Farm 18, on the Western Channel, consisted of 168 rectangular cages with 3 m cage depths. The estimated stocking density in 2015 was approximately 10–20 kg/m³. Farm Cluster 21, on the Eastern Channel, utilizes seven large HDPE cages with cage depths of up to 7 m. It is one of the largest producing farms in the bay, with a stocking density of approximately 25–30 kg/m³. Figure 3 provides an overview of the conditions underneath the selected fish farms and nearby reference locations. Figure 3(a) shows the seabed at reference location 12 at a water depth of approximately 17 m (see Figure 2). Seafloor sediments are characterized by bioturbation. Figure 3(b) shows the sediment core from the same area, consisting of gray carbonate sand with increasing amounts of silt towards the surface. For comparison, a sediment core collected beneath Farm 18 at a water depth of approximately 18.6 m is shown in Figure 3(c). The sediments underneath this farm evidently have very similar properties in terms of grain size, consistency, and color association than those at the neighboring reference location. No farming impact, such as organic accumulation, was observed in the samples collected from underneath the farm.

Click to enlarge

In contrast, the seabed beneath the circular cages of Farm Cluster 21 showed a smooth, undulating structure with no bioturbation features. Vast areas beneath the cages were found to be completely covered with a black mud layer overlaid with shimmering bright mats of sulfur bacteria (Figure 3d). Gas bubbles of H2S escaped from the seabed when agitated by the diver. Figure 3(e) depicts one of the sediment cores collected from beneath the farm cluster. The muddy material is black due to the presence of iron sulphites resulting from anaerobic bacterial desulfation. Similar seafloor sediment conditions were only observed in Farm Cluster 30. Figure 3(f) shows the sediment core collected at the reference location 7 approximately 150 m south of Farm Cluster 21 (Figure 2). The silty sediment was plastic and light gray. Similar to the other samples obtained at reference locations, there was no visible evidence of impacts due to fish farming.

Quantitative Assessment

Next, we analyzed the impacts of fish farms on seafloor sediments based on the measured sediment quality indicators. Figure 4(a) shows the variation of TOM with PON in the seafloor sampled sediments. TOM concentrations ranged 4.6–11.8% at reference locations and 4.2–18.8% underneath the fish farms. The levels of organic matter in sediments gradually increased from the low values observed at the reference locations and beneath traditional nonaffected fish farms (see Section A in Figure 4a), to the highest values at the farm clusters 21 and 30 (see Section B in Figure 4a). Notably, TOM concentrations under most of the traditional fish farms are comparable to those determined at reference locations. Based on the statistical analysis results and considering the values observed at those farms exhibiting deterioration, a threshold value was defined for TOM of approximately 12%. Details of the estimation of this critical value are provided in Section 4.2. The dashed red line in Figure 4(a) indicates the limiting value to distinguish between contaminated and uncontaminated sediment conditions. PON concentrations in sediments ranged 0.4–1.3 mg/g in pristine areas and 0.6– 7.1 mg/g below fish farms (Figure 4a). PON concentrations up to approximately 1.5mg/g at traditional fish farms are not different from pristine reference areas. At most fish farms, the PON concentrations were slightly higher compared to those at reference locations (see section A in Figure 4a). Although biodegradation of waste was occurring at the seafloor underneath these farms, the PON levels had not yet reached levels measured in the pristine areas. For the sediment conditions underneath the circular cages (farm clusters 21 and 30), the null hypothesis was rejected in the U-tests (Table 2). Notably, PON concentrations at both farm clusters significantly deviated from those characterizing pristine conditions. The threshold value in terms of PON for the site in Bali was defined in the order of 1.5 mg/g (see dashed red line in Figure 4a, and Section 4.2). An analysis of the variation of pH with PON in all the sediment samples was also conducted (see sections A and C of Figure 4a). The pH values measured in the sediments varied between 7.5 and 8.1 at reference locations and from 6.9-8.2 underneath fish farms. A strong correlation existed between pH and PON, as shown in Figure 4(a). Decreasing pH levels in organic muds underneath the severely affecting farm clusters 21 and 30 indicated the increasing acidity in the seafloor sediment due to increasing levels of H2S.

Figure 4(b) shows the variation of TOM with TDS in the sediment samples. The data array of measured TDS values ranged 0.0–1.9 mg/l at undisturbed locations and 0.0–78.4 mg/l at fish farms. In general, TDS values at reference locations and beneath traditional fish farms were very low (section B in Figure 4b). In contrast, below farm clusters 21 and 30, the TDS values were significantly high. The high variance of the sulfite contents in the highly impacted seabed sediments were attributed to the non-homogeneous geochemical conditions. High TDS values obtained for these farms were related to increased TOM levels (see section A in Figure 4b). However, the opposite did not always hold true. Notably, although the TOM values in the sediments at Farm 11 were also elevated, TDS concentrations at this fish farm were only slightly increased. Based on the measured values, the transition from undisturbed to impacted seafloor sediment conditions occurred from approximately 2 to 5 mg/l, as shown in Figure 4(b).

Figure 4(c) shows the variation in TOM with Eh for the sampled sediments. The Eh values at the reference locations ranged from -7 mV to -302 mV and at the sampled fish farms from -28 mV to -386 mV. The measured values of these two populations overlapped over wide ranges. In general, values were highly non-homogeneous across channels, and the correlation of TOM with Eh was extremely weak. Nevertheless, the highest TOM values coincided reasonably well with the lowest measured Eh values at the most impacted farm clusters 21 and 30. The limiting value that characterized undisturbed conditions was considered as -302 mV, as it was the lowest redox value measured at reference locations (see dashed red line in Figure 4c). Notably, the Eh values measured in some of the samples collected underneath farms 18, 20, and 27 were below this limit. Similar to the observations for TDS, it was difficult to detect sediment degradation in the early stages through Eh measurements.

Click to enlarge

(a)

Click to enlarge

(b)

(c) Figure 4: Measured sediment parameters at fish farms and reference locations in Pegametan Bay. (a) TOM versus PON and pH versus PON; (b) TOM versus TDS; and (c) TOM versus Eh. Note: n is the size of each random sample; emphasized red lines indicate the defined/chosen threshold values.

Best Suited Sediment Quality Indicator(s) for Remote Sites in Southeast Asia

Among the four analyzed sediment indicators, TOM and PON were found to be the best suited for monitoring benthic impacts at monitoring benthic impacts at the site studied. Both parameters showed a gradual increase from low concentrations at pristine locations to high values at the most impacted fish farms in the selected site in Bali (Figure 4a). Sediment sampling can be easily accomplished with grab samples from boats or directly from fish farms. Determination of TOM can be performed in muffle furnaces by loss of ignition, which is inexpensive and less error- prone in terms of preparation. Similar to TOM, elevated PON concentrations in seabed sediments provided reliable evidence of incomplete bacterial degradation of waste matter. Nitrogen is deficient in oligotrophic systems and is an essential component of organic wastes released from fish farms. However, carbonate sediment preparation for the CN analysis of PON requires caution and skill.

The geochemical parameters, TDS and Eh, also proved to be effective indicators of organic enrichment due to fish farming. However, these could not resolve the transition from undisturbed to impacted sediment conditions. Analysis of measurements in some fish farms had shown that organic enrichment occurred soon after minor traces of TDS were detected in sediments. Therefore, TDS could not detect sediment deterioration in the early stages. Similarly, Eh was found to provide meaningful information on the oxygen status in sediments beneath fish farms, but the values highly varied and were usually negative in muddy depositional environments. In addition, the variation of Eh values in pristine areas, as observed in this study, hindered the identification of emerging impacts due to fish farming through U-tests. Moreover, attention should be given to the measurements and analyses of the TDS and Eh. As direct measurements on the seafloor are impractical, samples brought to the free surface would partially exhibit gas losses. Outgas flow and unnoticed leaching of oxygenated surface water into the sediment subsurface can lead to biases, such as increases in Eh and decreases in TDS.

Site-specific Threshold Values for PON and TOM

The estimation of threshold values using PON and TOM data has been investigated in many studies. The majority of these have primarily covered aquaculture sites in subtropical to subpolar climate zones [8, 13, 14, 15, 16, 17, 18]. Various values have been reported in these literatures, reflecting the different environments and cultivated species. However, information on the critical values of sediment parameters applicable to tropical environments is limited. In this study, we attempted to estimate the threshold values of TOM and PON at the site in Bali. This was achieved by combining the results of the non-parametric U-tests with the analysis of the measured values at the fish farms experiencing initial signs of sediment deterioration (Figure 4a). Focus was laid on the conditions observed at fish farms 11 and 23. As shown in Table 2, the null hypothesis was confirmed for TDS and Eh, but rejected for TOM and PON at Farm 11. For Farm 23, the null hypothesis was rejected only for the TOM. By analyzing the range of values observed at these fish farms, threshold values were defined for TOM of approximately 12% and PON of approximately 1.5 mg/g (Figure 4a). Although these values have been applied to provide a preliminary assessment of the seafloor sediment conditions, they should be considered as site-specific.

Assessment of the Operating Conditions of the Selected Site in Bali

From the analysis results, three groups of farms were identified based on their impact on the seafloor sediments in Pegametan Bay, that is, fish farms operating well within carrying capacity, farm clusters drastically impacting the seafloor sediments, and farms exhibiting initial signs of sediment quality deterioration. It was observed that underneath most of the traditional fish farms with cage depths ≤ 3 m and operating with stocking densities ≤ 20 kg/m3, the measured values of all sediment indicators were comparable to those observed at pristine locations. This included farms 2, 13, 15, 16, and 18 in the Western Channel and farms 20, 27, and 28 in the Eastern Channel (Figure 2). Although current velocities at these fish farms were too low to displace settled wasted particles, there were no signs of accumulation of organic matter. Thus, settled organic waste was evidently completely biodegraded, and production was below the carrying capacity of fish farms.

The second group of farms comprised farm clusters 21 and 30, which had caused major damage to the seafloor. Both farm clusters consisted of several circular cages with approximately 6–7 m cage depths and operated at stocking densities of approximately 25–30 kg/m3. The levels of TOM, PON, and TDS measured underneath the circular cages were significantly enhanced, and the redox potential was significantly lower than that at the pristine locations. Denitrification and desulfation were the predominant bacterial respiration systems, and the pH in sediments was lower than that in undisturbed locations. While nitrate reduction ranks just behind oxygen respiration in terms of its energy yield [19], bacterial sulfate reduction was significantly less efficient. Hence, the degradation slowed down and waste accumulated. Organic enrichment was observed beneath both farm clusters (Figure 3e), which revealed an exceeded carrying capacity.

The third group of farms comprised traditional fish farms 11 and 23. Both farms employed rectangular cages with 3 m cage depths and operated at stocking densities of 10–20 kg/ m3. At Farm 11, the null hypothesis was rejected for TOM and PON, but not for TDS and Eh (Table 2). At Farm 23, the null hypothesis was confirmed for all sediment parameters except PON. The statistical analysis results indicated a poor ecological balance between the deposition of organic waste and its bacterial degradation beneath these farms. This may be because both farms had been operating just at or beyond the limits of carrying capacity. At first, the deteriorating seafloor conditions under the farms was unexpected, as cage depths and stocking densities were similar to the remaining traditional farms at the site in Bali. However, besides being the third largest farm in the site, Farm 11 is located outside suitable areas and relatively close to the shore [4]. Although flushing rates were comparable to those of other farms in the Western Channel, parts of Farm 11 might be exposed to even less flushing. Moreover, Farm 23 was characterized by the lowest flow Reynolds number (Re) among all fish farms in the Eastern Channel [5]. The near-field carrying capacity is directly related to Re, which could explain the deterioration tendencies observed in this case. Furthermore, Farm 23 is very close to the central coral reef platform separating the tidal channels, and just at the border of the suitable areas (Figure 2). Hence, like Farm 11, parts of this farm might be subjected to very small flushing rates.

Adoption of the Results for Estimating Carrying Capacity at Farm Level

The results of the monitoring strategy proved to be well suited for optimizing the PCC method developed by Mayerle, et al. [5]. This method predicts the maximum allowable fish production of individual fish farms. A non-dimensional relationship was derived between the ratio of deposited to emitted organic matter, the hydrodynamic character of the fish farm location, and the characteristic settling velocity of the particulates. The maximum fish production and allowable cage depths could be determined by limiting the carbon deposition beneath the farms to ensure environmental sustainability. To account for the simplifications and assumptions made in the derivation of the relationship and the uncertainties in the selection of the input data used in the predictions, a site-specific correction coefficient was introduced. This coefficient was determined from on-site analysis of sediment quality underneath fish farms exhibiting initial signs of deterioration. This is typically the case for farms 11 and 23 in Pegametan Bay, as shown in Section 4.3. The correction coefficient could be determined by comparing the PCC predictions with the threshold value of the carbon load on the seafloor at these farming locations. Thus, the accuracy of predicting near-field carrying capacity could be improved. Details on the PCC method and estimation of the correction coefficient are available in Mayerle, et al. [5].