Environmental perturbation and the structure ... - Springer Link

Journal of Aquatic Ecosystem Stress and Recovery 6: 265–279, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.

265

Environmental perturbation and the structure and functioning of a tropical aquatic ecosystem G.J. Piet1,2 & J. Vijverberg1 1 Netherlands

Institute of Ecology, Centre for Limnology, Rijksstraatweg 6, 3631 AC Nieuwersluis, The Netherlands 2 Department of Fisheries, Ruhuna University, Matara, Sri Lanka Received 17 February 1998; accepted in revised form 1 December 1998

Key words: drought, pelagic minor cyprinids, foodweb, fisheries management, production, reservoir, yield

Abstract In Tissawewa, a shallow, eutrophic reservoir in southeastern Sri Lanka, the effect of a major drought on the ecosystem was studied by monitoring the size-structured fish community and its resource base. Primary production was determined as well as the production and diets of ten taxa belonging to four trophic guilds (i.e. herbivorous/ detritivorous, benthivorous, zooplanktivorous/insectivorous, piscivorous) that made up more than 98% of the total fish biomass. Two extreme states of the ecosystem were distinguished. Before the drought most primary production was generated by phytoplankton, suspended fine particulate detritus was an important food source and total fish density was high. After the drought the ecosystem was characterised by high macrophyte density, low concentration of suspended detritus and low total fish density. The availability and origin of detritus appeared to be the major factor influencing fish production in Tissawewa. The small pelagic herbivore/detritivore A. melettinus contributed the most biomass and production to the fish community before the drought. After the drought, however, biomass and production dropped considerably. In contrast, the production of the most important species in terms of fisheries yield, the exotic herbivorous/detritivorous tilapias, was hardly affected. Although the composition of their food, benthic detritus, had markedly changed. In Sri Lankan reservoirs a subsidiary fishery for pelagic minor cyprinids was suggested to increase the current yield which is based almost entirely on the exotic tilapia species. The perturbation observed in this study, however, showed that the production of pelagic species was affected particularly by the environmental changes. Exploitation of these species can, therefore, only be considered in combination with hydrological and other management measures that control the environmental conditions of the reservoirs.

Introduction As in many other areas in SE Asia, Sri Lanka has no natural lakes but many artificial reservoirs, covering a total area of around 100,000 ha. The main function of these reservoirs is irrigation of paddyfields. Fisheries are a secondary function and water management of the reservoir is dictated by the demands of the paddy farmers. Decisions on the allocation of water can have considerable consequences for the structure and functioning of the aquatic ecosystem and the exploitation of the fish resources.

Tissawewa is a typical shallow (mean depth = 1.2 m), eutrophic, lowland reservoir of about 200 ha in SE Sri Lanka, where waterlevel fluctuations are the main source of environmental variation (Piet, 1998). In 1992 a long period of extreme drought together with the pressing demands for water from the paddy farmers led the irrigation department to release all Tissawewa water. After two months Tissawewa was refilled from the upstream reservoir. These actions provided the opportunity to study the effect of reservoir drying on the structure and functioning of the ecosystem, with special reference to the fish

266 community. Two extreme states of the ecosystem were distinguished: a state before the reservoir dried up where phytoplankton dominated primary production, with high turbidity and low predator abundance; and a state after refilling the reservoir with clear water where macrophytes were the main primary producers and predator abundance was high. Before emptying, Tissawewa was representative of reservoirs in the dry zone in Sri Lanka for which the introduction of a subsidiary gillnet fishery was suggested as a means of increasing fisheries yields (De Silva & Sirisena, 1987; Sirisena & De Silva, 1989; Pet et al. 1996; Piet & Vijverberg, 1998). In the present paper the two contrasting states of the Tissawewa ecosystem are studied to determine how the changed environment affected the structural and functional properties of the Tissawewa fish community.

Table 1. Values of ecosystem parameters (mean ±95% c.l.) for the periods before (1) and after drought (2) Period 1

2

Conductivity (µS.cm−1 )

664 ± 47 518 ± 44 Secchi-depth transperancy (cm) 30 ± 2 127 ± 18 Suspended detritus (mgC.m−3 ) 7328 ± 1599 2111 ± 1813 Phytoplankton biomass 484 ± 93 143 ± 10 (mgC.m−3 ) Vegetated area (%) 10 100 Gross primary production 13034 ± 1792 6273 ± 1801 (kgC.ha−1 .yr−1 ) Fish biomass (kgC.ha−1 ) 170 ± 32 77 ± 19 Fish production 243 ± 16 128 ± 10 (kgC.ha−1 .yr−1 )

Material and methods Sampling program Sampling was conducted between April 1991 and January 1994. During this period the water level fell due to an extreme drought until the reservoir became completely dry in September and October 1992. On filling again, a different ecosystem had evolved (Table 1): Before desiccation Tissawewa (Period 1) had macrophyte beds only in the shallow, littoral zone and turbidity was high (mean Secchi-depth transparency = 30 cm) due to water column resuspension of a thick layer of detritus from the bottom. After refilling (Period 2) macrophytes were widespread and water was clearer (mean Secchi-depth transparency = 127 cm) because the detritus-layer on the bottom was gone. Because at a waterlevel below 1.5 m the fish community changes from equilibrium (Piet, 1998), the months from September 1991 until June 1992 were considered representative for the period before desiccation, while the months from December 1992 until July 1993 were representative for the period after refilling. Phytoplankton primary production was measured fortnightly from April 1991 to July 1992. Phytoplankton biomass and concentration of suspended detritus were measured from April 1991 until April 1993. Phytoplankton production was determined at five different depths half a meter apart using light and dark bottles (three replicates) filled with reservoir water from that particular depth. Incubation was from

9.00 a.m. to 12.00 noon. Incubated bottles were fixed immediately with Winkler reagent and the acidified bottles were titrated within three hours after collection. Gross and net photosynthetic rates were calculated following Vollenweider (1974). Phytoplankton biomass was determined from Chlorophyll-a measurements at the same depths as phytoplankton production. Water samples were filtered in the field through glass fibre filters using a pressure syringe. Chlorophylla was extracted following Moed and Hallegraeff (1978). The concentration of suspended particulate organic matter (POM) was calculated by subtracting the phytoplankton concentration from the seston concentration determined following Golterman et al. (1978). Measurements which were not in mg C.m −3 , such as Chlorophyll-a (mg.m−3) and seston (mg dry weight.m−3), were converted as follows: 1 g chlorophyl-a = 25 g carbon (Bailey-Watts, 1974), 1 g dry weight = 0.5 g carbon (Winberg et al., 1971) and 10 g fresh weight = 1 g dry weight (Winberg et al., 1971). Biomass is expressed in kg C.ha−1 and production (gross for primary producers, net for primary and secondary consumers) in kg C.ha−1 .yr−1. Fish were sampled monthly from September 1991 to January 1994. Each sampling occasion consisted of setting 120 m of multimesh (12.5–90 mm stretched mesh) gillnets in four habitats (Pet & Piet, 1993), twice during daytime and twice at night. Two active gears were also used: bottom trawl and castnet. These gears ensured the entire size range (>3 cm total length) of the available species was sampled.

267 Table 2. Number of fish caught (C), used for gut content analysis (G) and for determination (D) of the length–weight relationship and Gonado-somatic index per species. The twoletter species abbreviations are used throughout the document C

Species

A. melettinus O. mossambicus O. niloticus B. chola B. dorsalis B. sarana H. gaimardi R. daniconius G. giuris Mystus spp.

AM OM ON BC BD BS HG RD GG MY

Gillnet

Active

99429 12782 2048 13701 14096 4971 2858 12660 1849 4942

143437 7020 639 1020 230 369 20047 10411 1876 5535

G

D

2912 2706 636 3412 2586 1340 1229 2649 1289 1733

1667 3134 823 1917 2922 1618 152 1541 243 365

Estimating production of primary invertebrate consumers Although production of primary invertebrate consumers in Tissawewa was not measured in the present study, it can be assessed from the consumption of the fish assemblage assuming fish feed on invertebrates with an ecological efficiency (i.e. ratio of the food requirements of one trophic level to the production of the previous level) of 40% (Vijverberg et al., 1990). For estimation of the consumption of the primary invertebrate consumers an ecological efficiency of 70% (Winberg, 1980) was used. Furthermore, all invertebrates were assumed to feed on phytoplankton and detritus in the same ratio as was observed for herbivorous fish. Fish stock assessment and production

The most abundant fish species in Tissawewa are the cyprinids Amblypharyngodon melettinus (Valenciennes), Barbus chola (Hamilton), Barbus dorsalis (Jerdon), Barbus sarana (Hamilton) and Rasbora daniconius (Hamilton), one goby Glossogobius giuris (Hamilton), one halfbeak Hyporamphus gaimardi (Valenciennes), two catfishes Mystus gulio (Hamilton) and Mystus vittatus (Bloch), which for practical reason were treated as Mystus spp., and two introduced exotic tilapias Oreochromis mossambicus (Peters) and Oreochromis niloticus (Linnaeus). These species together made up more than 98% of the biomass as determined from experimental gillnet catches and bottom trawl surveys (Piet, 1996). In order to distinguish size-specific differences in feeding behaviour the total size range was divided into six size-classes: e.g. size-class 1 (3.0–4.4 cm total length), size-class 2 (4.5–6.9 cm), size-class 3 (7.0–9.9 cm), size-class 4 (10.0–13.9 cm), size-class 5 (14.0– 18.9 cm) and size-class 6 (≥19.0 cm). For analysis of gut contents of the ten most abundant species, each month a maximum of 10 specimens per species, sizeclass, gear type, station and time of day were pooled. A maximum of twenty-four specimens per 0.5-cmclass per month of the six most abundant species were measured, weighed and dissected to ascertain gonad weight and sex. In total 359,920 fish were caught of which 20,492 fish were used for gut content analysis and 14,382 fish were dissected and used to develop the length–weight relationship and gonado-somatic index (Table 2).

To determine the fish assemblage composition, the length-frequency distributions of the experimental gillnet catches were corrected for fishing effort and gillnet selectivity. Gillnet selectivity was estimated using the extended Holt model (Pet et al., 1995a). The standing stock biomass of O. mossambicus, over the full size-range, was calculated to be 128 kg.ha−1 (Pet et al., 1995b). Standing stock biomass was calculated per month for all other species/size-classes on the basis of species/size-class composition in experimental gillnet catches and the estimated standing stock biomass of O. mossambicus. Production of each population for a given time interval is the sum of the growth increment of all individuals of that population. This growth increment includes somatic tissue and gonads. Monthly growth of somatic tissue was calculated from the sex-dependent growth curves and length-weight relationships per period of each species. This monthly growth was calculated using the von Bertalanffy growth curve (von Bertalanffy, 1934) and the parameters K and L∞ (were partly derived from Pet et al. (1995b) and partly estimated on the basis of Length-Frequency Distribution Analysis (LFDA). For the sex-dependent growth curves, biomass increase per sex was calculated from a size-dependent sex-ratio per species (Pet et al., 1995b). For females the lengthweight relationship was corrected for a size-dependent gonad weight while gonad production was estimated from the decreases in gonad weight when spawning, which was assumed to occur twice a year (Pet et al., 1995b).

268 Diet composition and consumption To study the size-dependent diet composition of a species, the contents of the stomach or anterior 1/3 part of the gut were analysed. The relative biovolumes of food items in gut/stomach were estimated using the points method (Hynes, 1950) with a microscope. The following food items were distinguished: Fish, insects (both terrestrial and aquatic but not benthic), microcrustacean zooplankton (cladocerans, cyclopoid and calanoid copepods), gastropods, fish eggs, chironomids, micro-benthos (mainly ostracods and some benthic cladocerans), macrophytes, epiphytic filamentous algae, phytoplankton and unrecognisable matter (digested matter and/or detritus). When a fish was found in the gut, the species and size-class were determined if possible. Because detritus can be an important food source in tropical fish assemblages (Bowen, 1979, 1980, 1981) it is important to distinguish between ingested detritus and digested matter. In Tissawewa four potential detritivores were distinguished (Piet, 1996): A. melettinus, O. mossambicus, O. niloticus and B. chola. For the non-detritivores the proportion of digested matter was closely correlated with body length (Piet, 1996). Assuming for detritivores a similar digestion rate to non-detritivores, the proportion of detritus in the guts of the detritivores could be determined by subtracting the calculated proportion of digested matter (Piet, 1996) from the observed proportion of unrecognisable matter. Food consumption of each fish species was estimated from the production and the diet composition per size-class assuming a conversion efficiency (i.e. ratio of production of a consumer to consumption of its food) depending on food characteristics rather than consumer characteristics. Brett and Groves (1979) found conversion efficiencies of 20% for herbivorous fish and 29% for carnivorous fish. In the present study these conversion efficiencies were used for food items of plant and animal origin, respectively. Results Primary production In Tissawewa primary production was generated predominantly by phytoplankton and macrophytes. Biomasses of these primary producers differed markedly before (Period 1) and after (Period 2) the desiccation of the reservoir. Average phytoplankton

biomass was 484 kg C.ha−1 during Period 1 and only 143 kg C.ha−1 during Period 2. From the phytoplankton production-biomass relationship established during Period 1 and the mean phytoplankton biomasses during the respective periods, average gross production rates were calculated of 13034 kg C.ha−1 .yr−1 during Period 1 and 6273 kg C.ha−1 .yr−1 during Period 2 (Figure 1). Respiration was on average 54% of the gross production. In contrast to phytoplankton, the production of macrophytes, judging from the area of the reservoir covered with vegetation, was markedly higher after refilling. During Period 1 vegetation covered 30% of the reservoir when full and decreased with decreasing waterlevel until it was negligible below a water depth of 2.5 m. On average about 10% of the reservoir area during Period 1 was covered with macrophyte beds, whereas during Period 2 the reservoir was completely covered. Diet composition Based on the feeding behaviour throughout the entire sampling period we divided the species of the Tissawewa fish assemblage into four guilds: The herbivorous/detritivorous guild consisting of A. melettinus, O. mossambicus and O. niloticus, the benthivorous guild consisting of the barbs B. chola, B. dorsalis and B. sarana, the zooplanktivorous/insectivorous guild consisting of H. gaimardi and R. daniconius and the piscivores consisting of G. giuris and Mystus spp. (Figure 2). Within these guilds we observed considerable differences in size-specific diet composition of a species between periods. The diet of the herbivorous/detritivorous species during Period 1 consisted mainly of detritus and to a lesser extent of phytoplankton and macrophytes. After the drought the proportion of detritus in the diet of the tilapias was lower, whereas the proportions of macrophytes and epiphytic algae were higher. For the largest size-classes of A. melettinus no differences between periods were observed but the smallest sizeclass displayed a larger proportion of zooplankton in its diet in Period 2. During Period 1 the main food items eaten by the barbs B. chola, B. dorsalis and B. sarana were benthic food items such as chironomids, gastropods and micro-benthos. Although the diets of the barbs were qualitatively similar throughout the entire sampling period, the quantities of the different food items differed considerably, both between species and between periods. Comparison of the most important

269

Figure 1. Relationship for Tissawewa between phytoplankton biomass (B) and production (P) established from 32 sampling occasions during Period 1. Indicated are the mean phytoplankton biomasses and the estimated production for the periods before (1) and after the drought (2).

size-classes, in terms of biomass, shows B. chola fed mainly on chironomid larvae and micro-benthos. The third most important food item differed between periods: detritus during Period 1 and fish eggs during Period 2. B. dorsalis fed on chironomid larvae with the second most important food item during Period 1, gastropods, being replaced by insects and microbenthos in Period 2. The biggest difference between the periods was observed for B. sarana where the principal food item during Period 1, gastropods, was largely replaced by insects in Period 2. Before the drought the zooplanktivorous/insectivorous species fed almost entirely on zooplankton, especially for R. daniconius, with an increasing proportion of insects with increasing size. During Period 2 zooplankton was replaced by insects, especially for H. gaimardi, although a high proportion of macrophytes was observed in the gut. The piscivores exhibited some distinct size-related changes in their diet. With increasing size both species

switched from zooplankton to invertebrates to fish. Overall, the most important prey fish was A. melettinus but this was dependent on predator size and abundance of A. melettinus. The main alternative prey species were the carnivorous pelagics H. gaimardi and to a lesser extent R. daniconius. During Period 2 the proportion of fish in the diet of both species was lower than in Period 1 and was replaced by insects, chironomid larvae and (only for G. giuris) shrimps. Composition and production of fish assemblage During Period 1 the standing stock biomass was dominated by A. melettinus, with a biomass almost equal to that of all other species combined (Figure 3). Other abundant species in terms of biomass were the cyprinids B. dorsalis, B. chola, R. daniconius and the tilapia O. mossambicus. The standing stock biomass of the entire fish assemblage during Period 1 (170.4 kg C.ha−1 ) was markedly higher than during

270

Figure 2. Gut contents (%) on basis of biovolume per species, per size-class and per period. Species are grouped together according to their guild. Only the most important food items were included.

271

Figure 2. Continued.

272

Figure 3. Standing stock biomass and production per species and per size-class for Period 1 (before the drought) and Period 2 (after the drought). For species abbreviations see Table 2. Note the multiplication factor (5×) for biomass and production of A. melettinus (AM) during Period 1.

Period 2 where it was only 77.4 kg C.ha−1 (Table 1). The species most affected by the environmental changes after the drought was A. melettinus, for which the biomass decreased from 77.1 kg C.ha −1 to 5.2 kg C.ha−1 between Periods (Figure 3). For some species (e.g. the piscivores G. giuris and Mystus spp., the tilapia O. niloticus, and the estuarine invader H. gaimardi) biomass increased after reservoir refilling in spite of the considerably lower total standing stock biomass. Total fish production is the sum of the production of somatic tissue and gonads. Growth of somatic tissue was calculated for each species per period from the growth curve and the length–weight relationship

(Table 3). Gonad production was determined from the calculated size-specific proportion of the body weight occupied by eggs at the time of spawning (Table 4). Compared to the change in standing stock biomass, the change in production from 243 kg C.ha−1 .yr−1 to 128 kg C.ha−1 .yr−1 , was relatively small (Figure 3). Possible reasons for this are that after the drought: (1) P/B ratios of the benthivorous and piscivorous species were higher, and (2) a higher proportion of the biomass consisted of species other than A. melettinus which had a relatively low P/B-ratio. Before desiccation of the reservoir, A. melettinus was the most important species in terms of production, representing 45% of the total fish production. In spite of its

273 Table 3. Calculated parameters used for determination of somatic growth. K and L∞ (of species marked with ∗ were calculated by Pet et al. (1995) for females (f) and males (m). The length–weight relationship is described by ln(W ) = a + b∗ln(L) with total length (L) in millimeters and weight (W ) in grams fresh weight. For species abbreviations see Table 2. Period 1 refers to the period before the drought and period 2 to after the drought

Species f AM∗ OM∗ ON BC∗ BD∗ BS∗ HG RD∗ GG MY

Von Bertalanffy growth curve K L∞ m f

1.50 0.75

1.60 0.65

8.0 26.0

1.10 0.95 0.45

12.0 19.0 26.0

1.20

11.0

0.61 1.00 0.85 0.40

7.0 30.0 30.0

0.21 1.10

m

10.5 17.0 22.0 19.7

0.59 0.29

relatively low standing stock biomass, the fast growth of the tilapia O. mossambicus made it the second most important species in terms of production. After reservoir refilling the main producers of fish biomass were B. dorsalis, G. giuris, O. mossambicus and R. daniconius. The food web The major food items sustaining fish production during Period 1 were detritus (46% by weight) and phytoplankton (20%) (Table 5). Other plant material such as macrophytes and epiphytic algae were not important food sources during this period, sustaining only about 3% of the total fish production. Zooplankton and chironomid larvae were the main animal food sources sustaining respectively 11% and 6% of the fish production. More than 70% of the production of the barbs B. chola, B. dorsalis and B. sarana came from benthic food items such as chironomid larvae (30%), gastropods (19%) and microbenthos (18%). The production of the piscivores G. giuris and Mystus spp. was sustained primarily by prey fish (55%, mainly A. melettinus), while production of R. daniconius and H. gaimardi was mainly based on zooplankton (76%) and to a lesser extent on insects (18%). In contrast to Period 1, the importance of plant material in sustaining fish production was markedly lower during Period 2. Macrophytes and epiphytic algae were the main food sources sustaining 17% of

10.0 30.2 22.3

Length–weight relationship Period 1 Period 2 a b a b −11.97 −10.79 −10.86 −11.74 −11.89 −11.21 −12.76 −11.68 −11.39 −11.09

3.11 2.97 3.00 3.14 3.13 2.99 3.02 3.02 2.94 2.95

−11.97 −10.36 −11.11 −11.20 −11.28 −11.80 −12.76 −11.69 −11.39 −11.09

3.10 2.89 3.06 3.02 3.01 3.14 3.02 3.03 2.94 2.95

the fish production while detritus and phytoplankton sustained only 7% and 6%, respectively (Table 5). Consequently, the relative importance of the animal food sources was higher after refilling of the reservoir. Chironomid larvae and insects were the most important animal food sources sustaining respectively 23% and 19% of the fish production. After the drought these two food items were the main replacement for food items such as the gastropods for the barbs, fish and zooplankton for the piscivores and zooplankton for the zooplanktivores/insectivores. Consumption of aquatic insects increased during Period 2. Only 46% of the insects consumed in this period by the entire fish assemblage were from terrestrial origin as compared to 72% during Period 1. From the estimates of primary production, consumption and production of the fish assemblage, together with data on commercial fisheries, a foodweb for the Tissawewa ecosystem was constructed for the periods before and after the drought (Figure 4). Comparison of the carbon fluxes in these foodwebs clearly shows the impact of environmental change on the Tissawewa food web. Primary production and production of detritus were markedly lower after the drought. Consequently production of primary consumers (herbivorous fish and invertebrates) was lower during Period 2. This was mainly caused by a large decrease in the production of herbivorous fish whereas the production of herbivorous invertebrates was hardly affected.

Figure 4. Overview of the major pathways of energy flow in the Tissawewa foodweb. Production (kg C.ha−1 .yr−1 ) is given for primary producers as well as primary and secondary consumers. Indicated are the relative proportions of the primary producers partaking in a specific pathway. The fish species involved are indicated for each pathway. For species abbreviations see Table 2. Values of production or relative proportion are indicated above for Period 1, below for Period 2.

274

275 Table 4. Proportion (%) of the body weight per year spent as gonads per species and per size-class as determined during Period 1. For species abbreviations see Table 2. Size-classes define total length in cm as follows: size-class 2 (4.5–6.9), size-class 3 (7.0–9.9), size-class 4 (10.0–13.9), size-class 5 (14.0–18.9) and size-class 6 (≥19.0) Species


Size-class 4 5

2

3

11.5

17.8 0.8

1.1 0.1 13.4 9.8 2.4 13.0 17.1 1.2 1.2

4.4 6.8 16.0 25.9

32.1

6

0.9 0.7 15.3 17.0 7.4 8.5

2.5

7.3

3.7 3.7

3.6 3.6

Table 5. Calculated consumption (kgC.ha−1 .yr−1 ) by fish species before (Period 1) and after the drought (Period 2) Food items


Period 1 Fish 0 1 Adult insects 0 3 Zooplankton 8 10 Gastropods 0 0 Shrimps 0 0 Chironomid larvae 0 1 Fish eggs 1 0 Microbenthos 0 0 Macrophytes 2 21 Phytoplankton 155 67 Detritus 374 103 Period 2 Fish Adult insects Zooplankton Gastropods Shrimps Chironomid larvae Fish eggs Microbenthos Macrophytes Phytoplankton Detritus

0 0 2 0 0 0 0 0 0 7 21

0 0 0 2 0 10 0 3 0 0 0 19 0 4 0 17 4 3 4 0 9 20

0 0 5 2 4 1 0 0 0 0 8 1 1 1 1 0 38 21 23 6 10 0

0 3 2 1 0 10 4 7 1 0 3

0 1 0 6 3 1 12 1 10 28 11 0 0 0 0 45 1 0 6 0 0 22 0 0 1 3 0 0 0 0 0 0 0

0 13 3 4 0 62 7 17 4 0 0

0 7 1 5 0 6 0 2 7 1 0

0 4 1 0 0 1 0 0 5 0 0

0 18 71 0 0 3 0 0 2 0 0

0 34 17 0 0 6 0 1 15 1 0

7 0 1 0 0 1 0 0 0 0 0

34 22 1 0 13 10 0 1 0 0 0

7 2 5 0 0 1 0 0 0 0 0

2 13 8 0 0 19 0 1 1 1 0

To evaluate the impact of the environmental perturbation four major pathways were distinguished in the Tissawewa foodweb: (1) pelagic pathway via invertebrates as primary consumers, (2) pelagic pathway via fish (A. melettinus) as primary consumers, (3) benthic pathway via invertebrates as primary

consumers, and (4) benthic pathway via fish (O. mossambicus and O. niloticus) as primary consumers. The impact of the environmental changes follows not only from the difference in the absolute amount of energy flowing through the various pathways but also from the changes in the relative contributions of these pathways to the total energy flow through the food web. These quantitative changes also affected some of the pathways qualitatively. Before the drought the benthic and pelagic pathways were about equally important as were the fish and invertebrate pathways. In the pelagic zone most of the production was consumed by the fish while in the benthic zone most production was consumed by the invertebrates. After the drought most of the production flowed through the benthic invertebrate pathway to the detriment of the pelagic fish pathway. The pelagic fish pathway was mainly affected by the decrease of the concentration of suspended detritus and phytoplankton together with the high predation pressure on A. melettinus. In contrast, the other herbivorous/detritivorous fish, the exotic tilapias sustaining the benthic fish pathway, were hardly affected. The benthic invertebrate pathway was the only pathway through which the same amount of energy flowed as before the drought.

Discussion Detritus For a long time it has been recognised that nonliving organic matter is very important in the structure and functioning of aquatic ecosystems (Mann, 1988). Detailed studies of many lake ecosystems indicate that a majority of energy flows are utilizing non-planktonic organic carbon and that most of the heterotrophic metabolism is microbial and benthic rather than planktonic (Wetzel, 1995). Under normal circumstances, comparable to the situation before desiccation of the reservoir (Piet, 1998), detritus was by far the most important food source for the fish assemblage. When determining the proportion of detritus in the diet from gut contents alone it should be realised that: (1) it is difficult to distinguish detritus from formerly living matter digested in the gut of the fish, and (2) the term detritus covers a broad range of dead organic matter from different origins. The first argument has been accounted for by the correction for the proportion of digested matter in the gut. Although the proportion of

276 detritus might still be slightly overestimated because digestion rates of herbivorous species are higher than those of non-herbivorous species (Fänge et al., 1979). The second argument is illustrated by differences in the origin of detritus consumed by the various detritivorous species. A. melettinus is a filter feeder, feeding on phytoplankton and suspended detritus in the watercolumn (Bitterlich, 1985; Pethiyagoda, 1991), whereas the tilapias prefer surface-grazing to filterfeeding (Dempster et al., 1993) and feed mainly on the benthic detrital aggregate (Bowen, 1979, 1980, 1981; Schiemer & Duncan, 1987). These differences in feeding behaviour are corroborated by the spatial distribution of the species; the tilapias are found close to the bottom when feeding while A. melettinus occupies the water column (Piet, 1996; Piet & Guruge, 1998). Before desiccation of the reservoir there was a thick layer of fine particulate detritus on the bottom which disappeared when the reservoir was dry (pers. observations). Based on the relatively high gross production by phytoplankton, and relatively small area covered with macrophytes, the autochthonous part of this detritus probably originated mainly from phytoplankton. When the reservoir was dry the organic material on the bottom decomposed, which is to be expected under well-aerated conditions and prevailing high surface temperatures (Kalk et al., 1979), and was blown away. Therefore, after refilling of the reservoir, accumulation of detritus had to recommence. During this process, the shallowness of the reservoir and the relatively low turbidity allowed access light to reach the bottom, thus promoting macrophyte growth (Blazka et al., 1980). The presence of macrophytes which covered the entire reservoir area limited phytoplankton growth (Wetzel et al., 1972) and, therefore, a considerably higher proportion of the autochthonous detritus formed during Period 2 was from macrophyte origin. Through time this will only be amplified because the decomposition rate of fine particulate organic matter is markedly higher than that of particulate leaf organic matter (Anderson, 1987). The abundant vegetation affects the availability of suspended detritus in the pelagic zone in a three ways: (1) by decreasing turbulence in the water column, (2) because macrophytes often inhibit algal production, and (3) particulate leaf organic matter is suspended less easily than fine particulate organic matter from algal origin. These arguments explain the observed changes in feeding behaviour and density of the main producers of fish biomass, that of the herbivor-

ous/detritivorous fish. The availability of suspended fine particulate organic matter, the main food source of A. melettinus, decreased while the composition of the bottom deposits, the main food source of the tilapias changed from mainly small particles of algal origin to larger particles of macrophyte origin. The apparent change in consumption by the tilapias from mainly detritus to macrophytes therefore represents a change in composition of their food source rather than a change in feeding behaviour of the fish. Exploitation of fish production The commercial catches were estimated at 24.2 kg C.ha−1 .yr−1 for Period 1 and 16.7 for Period 2 (Pet et al., 1995c). Before the drought, the utilisation of the biological fish production in terms of fish yield was 10.0%, a value equal to the utilisation efficiency (i.e. ratio of fish yield to fish production) which, according to Blazka et al. (1980), can be obtained from lakes and reservoirs. After refilling of the reservoir an even higher utilisation efficiency of 13.0% was achieved. Tilapia was by far the most important species in the commercial catches; during Period 1, 72% of the total catch consisted of tilapia while during Period 2 this was 95% (Pet et al., 1995c). In Sri Lankan reservoirs, the efficiency with which commercial fisheries exploit the fish assemblage can be increased considerably by a subsidiary gillnet fishery aimed specifically at the currently untapped resource of small pelagic species (De Silva & Sirisena, 1987; Sirisena & De Silva, 1989; Pet et al., 1996; Piet & Vijverberg, 1998). The findings in the present study that the production of pelagic species was particularly affected by the perturbation shows that these kind of fisheries management measures can only be successfull in combination with hydrological management of the reservoir. Moreover, because most (90%) of the production of higher aquatic plants and attached periphytic communities never enters a metazoan tract and, consequently, enters the detrital pool (Wetzel, 1995), any shift in primary production from phytoplankton to macrophytes will result in a decreased production of pelagic species. The persistence of notably the benthic species in the “new” environment after the reservoir refilling is not surprising. The ‘new’ environment resembles the littoral zone with its macrophyte beds, available in the lowland reservoirs during the wet season, which in turn resembles the natural marshes formed during the wet season (Piet, 1998). The littoral zone, however, is

277 generally avoided by the pelagic species (Piet, 1996; Pet & Piet, 1993). Production estimates and potential for exploitation Gross primary production and respiration of the phytoplankton were estimated during Period 1 at 13034 kg C.ha−1 .yr−1 and 7091 kg C.ha−1 .yr−1 respectively, resulting in a net primary production of 5943 kg C.ha−1 .yr−1 or 46% of the gross primary production. This gross primary production is roughly in the same range as that reported for other shallow eutrophic tropical waterbodies such as Lake George, with 19726 kg C.ha−1 .yr−1 (Ganf, 1972), or Lake Chad with 5500 kg C.ha−1 .yr−1 (Carmouze, 1983). The observed respiration loss of 54% for Tissawewa is high compared to the range of 21–37% found in other studies, predominantly in temperate regions (Vollenweider & Nauwerk, 1961; Lewis, 1974; Capblancq, 1982; Sondergaard et al., 1985; Garnier & Mourelatos, 1991). With respect to the production of invertebrate primary consumers in Tissawewa, only zooplankton production can be compared to that in other shallow, eutrophic waterbodies in the tropics. In Lake George (Burgis, 1974; Burgis et al., 1973) and Lake Chad (Carmouze et al., 1983) zooplankton production was estimated at 216 kg C.ha−1 .yr−1 and 237 kg C.ha−1 .yr−1 resulting in an approximate transfer efficiency of respectively 1.1% and 4.3%. The transfer efficiencies for zooplankton in Tissawewa of 2.4% (Period 1) and 1.6% (Period 2) are within the range observed for these lakes. The conditions before recession of the reservoir (Period 1) are representative for the conditions in most reservoirs in the dry zone in Sri Lanka (De Silva & Sirisena, 1987) and possibly other countries in the region. During this period the total fish biomass was estimated at 171 kg C.ha−1 and the total fish production at 243 kg C.ha−1 .yr−1 . Fish production was 1.9% of the gross primary production. This is a high transfer efficiency (i.e. ratio of production of a particular step of the trophic chain to the gross primary production) compared to transfer efficiencies reported for other lakes and reservoirs in both the tropics and the temperate zone, ranging from 0.2% to 1.6% (Blazka et al., 1980). Fish production in Tissawewa was also markedly higher than the fish production of 157 kg C.ha−1 .yr−1 calculated according to the relationship between annual net phytoplankton production and fish production observed for mainly temperate waterbodies (Downing et al., 1990).

Compared to the waterbodies in the temperate zone the high transfer efficiency of fish production in this tropical reservoir may be due to the short food chain: primary producers made up 71% of the total production consumed by the fish assemblage, various invertebrate primary consumers 26%, while fish, consisting mainly of primary consumers, made up a meagre 2%. Comparison, in terms of fish production, of Tissawewa with other tropical waterbodies is difficult because contrary to the present study, the estimated fish production of most other tropical communities is based on commercial catches only. These generally exclude the smaller sized species and therefore underestimate total fish production. In Tissawewa about 90% of the total fish production is realised by size-classes 1–4 which consist of fish smaller than the gillnet selection range of the commercial fishery. Thus, the observation that in most of the tropical reservoirs in SE Asia the fish production (i.e. fish yield) is lower than would be expected on the basis of the high primary production (Henderson, 1979; FAO, 1980) is more representative of the low efficiency with which commercial fisheries exploit the entire fish assemblage than a true estimate of biological fish production.

Acknowledgements The present study is part of the combined RU/ WOTRO/NIOO-CL/WAU-DFCF Research Project on Inland Fisheries. The Department of Zoology of the Ruhuna University (RU) in Matara, Sri Lanka, is acknowledged for their hospitality and cooperation. The project was financially supported by grant W84313 of the Netherlands Foundation for the Advancement of Tropical Research (WOTRO), the Centre for Limnology (CL) of the Netherlands Institute of Ecology (NIE) and the Department of Fish Culture and Fisheries (DFCF) of Wageningen Agricultural University (WAU). The authors thank P.B. Amarasinghe for providing us with the primary production estimates and D.N. Ariyasena, K.A.D.W. Senaratne and S.W. Kotagama for the observations on the fisheating birds. We also wish to thank E.A. Huisman (WAU-DFCF) for critically reviewing the manuscript and D.A.K. Gunasekara, N.W. Shanta, W. Sanath, W. Rohana, N.Y. Samaraweera, J.W. Jayasuriya, M.N.T. Munideewa, J.S. Pet, J.W.M. Wijsman, C. Soede, G.J.M. Gevers and M.M. Fuchs for their assistance in the field.

278 References Anderson, J.M., 1987. Production and decomposition in aquatic ecosystems and implications for aquaculture. In: D.J.W. Moriarty & R.S.V. Pullin (eds), Detritus and Microbial Ecology in Aquaculture. ICLARM Conference Proceedings, Vol. 14: 123–147. Bailey-Watts, A.E., 1974. The algal plankton of Loch Leven, Kinross. Proc. R. Soc. Edinb. (B) 74: 135–156. Bertalanffy, L. von, 1934. Untersuchungen uber die gesetzlichkeiten des wachstums. Allgemeine grundlagen der theorie. Roux d’Arch. Entwicklungsmech. Org. 131: 613–653. Bitterlich, G., 1985. The nutrition of stomachless phytoplankton fish in comparison with Tilapia. Hydrobiologia 121: 173–179. Blazka, P., T. Backiel & F.B. Taub, 1980. Trophic relationships and efficiencies. In: E.D. Le Cren & R.H. Lowe-McConnell (eds), The Functioning of Freshwater Ecosystems. Cambridge University Press, London: 393–410. Bowen, S.H., 1979. A nutritional constraint in detrivory by fishes: The stunted population of Sarotherodon mossambicus in Lake Sibaya, South Africa. Ecol. Monogr. 49: 17–31. Bowen, S.H., 1980. Detrital nonprotein amino acids are the key to rapid growth of Tilapia in Lake Valencia, Venezuela. Science 207: 1216–1218. Bowen, S.H., 1981. Digestion and assimilation of periphytic detrital aggregate by Tilapia mossambica. Trans. Amer. Fish. Soc. 110: 239–245. Brett, J.R. & T.D.D. Groves, 1979. Physiological energetics. In: W.S. Hoar, D.J. Randall & J.R. Brett (ed.), Fish Physiology, Volume VIII, Bioenergetics and Growth. Academic Press, New York. Burgis M.J., J.P.E.C. Darlington, I.G. Dunn, G.G. Ganf, J.J.G. Gwahaba & L.M.M. McGowan, 1973. The biomass and distribution of organisms in Lake George, Uganda. Proc. R. Soc. Lond. 184: 271–298. Burgis, M.J., 1974. Revised estimates for the biomass and production of zooplankton in Lake George, Uganda. Freshwat. Biol. 4: 535–541. Capblancq, J., 1982. Phytoplankton et production primaire. In: R. Pourriot, J. Capblancq, P. Champ & J. Meyer (eds), Ecologie des eaux continentales. Masson, Paris: 1–48. Carmouze, J.P., J.R. Durand & C. Leveque, 1983. Lake Chad: Ecology and Productivity of a Shallow Tropical Ecosystem. Monogaphiae Biologicae 53. Dr W. Junk Publishers, The Hague, Boston. Dempster, P.W., M.C.M. Beveridge & D.J. Baird, 1993. Herbivory in the tilapia Oreochromis niloticus – a comparison of feeding rates on phytoplankton and periphyton. J. Fish Biol. 43: 385– 392. De Silva, S.S., 1985. Status of the cichlid Sarotherodon mossambicus (Peters) in the reservoir fishery of Sri Lanka: A management strategy and ecological implications. Aquaculture and Fisheries Management 1: 91–102. De Silva, S.S., 1988. Reservoirs in Sri Lanka and Their Fisheries. FAO Fisheries Technical Paper 298. 128 pp. De Silva, S.S. & H.K.G. Sirisena, 1987. New fish resources in Sri Lanka: feasibility of introduction of a subsidiary gillnet fishery for minor cyprinids. Fish Res. 6: 17–34. Downing, J.A., C. Plante & S. Lalonde, 1990. Fish production correlated with primary production, not the morphoedaphic index. Can. J. Fish. Aquat. Sci. 47: 1929–1936. Fänge, R. & D. Grove, 1979. Digestion. In: W. S. Hoar, D.J. Randall & J.R. Brett (eds), Fish Physiology, Vol. VIII. Academic Press, New York: 161–241.

FAO, 1980. Comparative studies on freshwater fisheries. Report of a workshop held at the Istituto Italiano di Idrobiologia. Pallanza, Italy. FAO Fish. Techn. Pap. 198: 1–46. Fernando, C.H., 1971. The role of introduced fish in fish production in Ceylon’s freshwaters. In: E. Duffey & A.S. Watt (eds), The Scientific Management of Animal and Plant Communities for Conservation. Blackwell, Oxford: 285–310. Fernando, C.H., 1984. Reservoirs and lakes of southeast Asia (Oriental region). In: F. Taub (ed.), Lakes and Reservoirs. Elsevier, The Netherlands: 411–446. Fernando, C.H. & J. Holcik, 1982. The nature of fish: A factor influencing the fishery potential and yields of tropical lakes and reservoirs. Hydrobiologia 97: 127–140. Fernando, C.H. & J. Holcik, 1991. Fish in reservoirs. Int. Revue Ges. Hydrobiol. 76: 149–167. Ganf, G.G., 1972. The regulation of net primary production in Lake George, Uganda, East Africa. In: Z. Kayak & A. HillbrichtIlkowska (eds), Productivity Problems of Fresh Waters. Polish Scientific Publishers, Warsaw and Krakow: 693–708. Garnier, J. & S. Mourelatos, 1991. Contribution of grazing in phytoplankton overall losses in a shallow French lake. Freshwater Biology 25: 515–523. Golterman, H.L., R.S. Clymo & M.A.M. Ohnstad, 1978. Methods for Physical and Chemical Analysis of Fresh Waters. IBP Handbook 8, 2nd edition. Blackwell Scientific Publ., Oxford. Henderson, H.F., 1979. The significance of limnology in the development of fisheries in man-made lakes and river basins. Proc. Int. Conf. on Kainji Lake and River Basins Development in Africa 1: 135–145. Hynes, H.B.N., 1950. The food of fresh-water sticklebacks (Gasterosteus aculeatus and Pygosteus pungitius) with a review of methods used in studies of the food of fishes. J. Anim. Ecol. 19: 36–58. Kalk, M., A.J. McLachlan & C. Howard-Williams, 1979. Lake Chilwa, Studies of Change in a Tropical Ecosystem, Monographiae Biologicae Volume 35. Dr W. Junk Publishers, The Hague. 462 pp. Lewis, W.M., 1974. Primary production in the plankton community of a tropical lake. Ecological Monographs 44: 377–409. Mann, K.H., 1988. Production and use of detritus in various freshwater estuarine, and coastal marine ecosystems. Limnol. Oceanogr. 33: 910–930. Moed, J.R. & G.M. Hallegraeff, 1978. Some problems in the estimation of chlorophyll-a and phaeopigments from pre- and postacidification spectrophotometric measurements. Int. Revue ges. Hydrobiol. 63: 787–800. Pet, J.S. & G.J. Piet, 1993. The consequences of habitat occupation and habitat overlap of the introduced tilapia Oreochromis mossambicus (Peters) and indigenous fish species for fishery management in a Sri Lankan reservoir. J. Fish Biol. 43: 193–208. Pet, J.S., G.J.M. Gevers, W.L.T. Van Densen & J. Vijverberg, 1996. Management options for a more complete utilization of the biological fish production in Sri Lankan reservoirs. Ecology of Freshwater Fish 5: 1–14. Pethiyagoda, R., 1991. Freshwater Fishes of Sri Lanka. Wildlife Heritage Trust of Sri Lanka. Piet, G.J., 1996. On the Ecology of a Tropical Fish Community. Thesis, Wageningen Agricultural University. 189 pp. Piet, G.J., 1998. The impact of environmental perturbation on a tropical fish community. Can. J. Fish. Aquat. Sci. 55: 1842–1853. Piet, G.J. & W.A.H.P. Guruge, 1998. Diel variation in feeding and vertical distribution of ten co-occurring fish species: Consequences for resource partitioning. Env. Biol. Fish. 51: 67–86.

279 Piet, G.J. & J. Vijverberg, 1998. An ecosystem perspective for the management of a tropical reservoir fishery. Internat. Rev. Hydrobiol. 83: 103–112. Schiemer F. & A. Duncan, 1987. The significance of the ecosystem approach for reservoir management. In: S.S. De Silva (ed.), Proc. Regional Workshop on Reservoir Fisheries Management and Development. Kathmandu, Nepal, 22–29 Nov. 1987: 183–194. Sirisena, H.K.G. & S.S. De Silva, 1989. New fish resources of reservoirs in Sri Lanka. II. Further studies on a gillnet fishery for minor cyprinids. Fish. Res. 7: 17–29. Sondergaard, M., B. Riemann & N.O.G. Jorgenson, 1985. Extracellular organic carbon during a diatom bloom in Lake Mosso: Molecular weight fractionation. Freshwater Biology 12: 313– 320. Vijverberg, J., M. Boersma, W.L.T. Densen, W. van, Hoogenboezem, E.H.R.R. Lammens, & W.M. Mooij, 1990. Seasonal variation in the interactions between piscivorous fish, planktivorous fish and zooplankton in a shallow eutrophic lake. Hydrobiologia 207: 279–286. Vollenweider, R.A., 1974. A Manual on Methods for Measuring Primary Production in Aquatic Environments. IBP Handbook 12. Blackwell Scientific Publ., Oxford.

Vollenweider, R.A. & A. Nauwerck, 1961. Some observations on the C14-method for measuring primary production. Verhandlungen der internationalen vereinigung fur theoretische und angewandte limnologie 14: 134–139. Wetzel, R.G., 1995. Death, detritus, and energy flow in aquatic ecosystems. Freshwater Biology 33: 83–89. Wetzel, R.G., 1990. Reservoir ecosystems: Conclusions and speculations. In: K.W. Thornton, B.L. Kimmel & F.E. Payne (eds), Reservoir Limnology: Ecological Perspectives. John Wiley and Sons, New York: 227–238. Wetzel, R.G. & B.A. Manny, 1972. Decomposition of dissolved organic carbon and nitrogen compounds from leaves in an experimental hard-water stream. Limnology and Oceanography 17: 927–931. Winberg, G.G. et al., 1971. Methods for calculating productivity. In: W.T. Edmondson & G.G. Winberg (eds), Secondary Productivity in Fresh Waters. Blackwell Scientific Publications Oxford: 296– 317. Winberg, G.G., 1980. General characteristics of freshwater ecosystems based on Soviet IBP studies. In: E.D. Le Cren & R.H. Lowe-McConnell (eds), The Functioning of Freshwater Ecosystems. Cambridge University Press, London: 481–492.