Zooplankton size-structure dynamics of a lowland tropical floodplain lake

Floodplains are highly complex ecosystems representing high biodiversity and conservation values, but they are also one of the world’s most threatened ecosystems due to extensive development and anthropogenic activities. Lake Tempe is a lowland riverine floodplain lake in Sulawesi Island, Indonesia, that is subject to multiple stressors such as flow alteration, eutrophication and invasive species. In this study, the dynamics of the zooplankton community size structure was investigated in Lake Tempe. Five size-based metrics, including zooplankton mean body size, total abundance, total, biomass, and normalised biomass size spectra (NBSS) slope and intercept were assessed from net-sampled zooplankton collected monthly from March to December 2016, and the role of environmental variables in shaping these size-based metrics were also examined. Zooplankton community size structure is a useful metric as it is less labour intensive than traditional approaches, provides more data accuracy and does not require highly specialised taxonomic expertise. The zooplankton community in Lake Tempe was characterised by high density of small-sized zooplankton with an average mean body size < 500 µm equivalent spherical diameter. While the zooplankton density was characterised by the dominance of the size class 300–700 µm, there was evidence of a significant contribution of the size > 700 µm towards total biomass. Seasonal hydrological regimes, water temperature and eutrophication parameters, as indicated by total phosphorus and chlorophyll-a concentration, had a significant role in driving the variability of zooplankton community size structure in this system. The results of this study are not only a significant step in providing critical baseline information on the zooplankton assemblage of Lake Tempe, but is also a significant contribution to the overall understanding of zooplankton community structure in tropical lakes, and will aid in improving lake management plans in these regions.


Introduction
Riverine floodplains are some of the most biologically productive ecosystems, providing various ecosystem goods and services, from water supply and food provision and habitats for biodiversity, to other societal benefits such as flood control, transportation and recreation (Tockner et al. 2010, Watkins et al. 2013, Keddy et al. 2009).Hydrological exchange in the riverine floodplain facilitates the exchange processes of suspended solids, organic matter and nutrients between riverine and terrestrial environments and the floodplain (Junk et al. 1989, Tockner et al. 1999).These complex interchanges between the aquatic and terrestrial habitats drive the biotic community structure (Górski et al. 2013) and support high biodiversity (Tockner and Stanford 2002, Thomaz et al. 2007, Deksne and Škute 2011).However, under increasing urbanisation and anthropogenic activities, the pressure on floodplains continues to grow at an unprecedented rate (Keddy et al. 2009, Monk et al. 2019).
Despite their importance, the majority of floodplain ecosystems globally have been lost, degraded or strongly modified by human activities which affect the ecological integrity and ecosystem services delivery of floodplain ecosystems (Monk et al. 2019, Chaparro et al. 2019, Reis et al. 2017).Consequently, many riverine floodplain ecosystems are now 92 Page 2 of 16 experiencing reduced flow, fragmented habitats and water quality deterioration (Tockner et al. 2010, Tockner andStanford 2002).Indonesia is no exception, with many of its freshwater ecosystems heavily degraded due to increasing land use activities, hydrological alteration, and the rapid spread of non-native species.Many Indonesian lakes are suffering from excessive nutrient loading, flow modification and the rapid spread of invasive species (KLHK 2018, KLHRI 2011 ).
The accelerated degradation of critical lake ecosystems and the potential impact on the local economy has led the Indonesian central government to release a list of 15 major lakes as part of their national priority plan.Lake Tempe is one of these priority lakes and is a riverine floodplain system located in South Sulawesi Province of Sulawesi Island, Indonesia.It is known as one of the most productive lakes in Indonesia, with an annual fisheries yield of 58,400 tonnes recorded in 1945; thus, it is an important economic source for the region.However, due to over-exploitation and water quality deterioration, among other factors, fisheries productivity has declined significantly.Moreover, intensive urban development in the region, particularly within the watershed, has been identified to be significant stressor in the degradation of the Lake Tempe ecosystem; this includes deforestation along the catchment, increasing agricultural runoff, and intensive aquaculture activities (Harsono 2016).Due to its status as a provincial and national priority lake, there has been a recent focus on further research to understand the ecological and economical values of the lakes to support the Lake Tempe ecosystem rehabilitation programme.This has led to many studies and assessments focused on ecological assembly and the economic importance of the lake, as well as on the impact of anthropogenic activities on the lake's biotic and abiotic components.These include studies on hydrological regimes (Setiawan and Wibowo 2013), water quality (Jasalesmana et al. 2014), fish biodiversity and community structure (Dina et al. 2019, Nasution 2015), and aquatic macrophytes (Nugraha et al. 2019).Given that floodplain ecosystems, such as Lake Tempe, are highly influenced by temporal and seasonal changes of physical, chemical and biological processes, it is important to understand how these seasonal variations of abiotic processes affect the biotic components, and thus the ecosystem structure in the lake.The Lake Tempe rehabilitation plan is currently focused on indicators such as the lake water level, chemical characteristics and fish community; other biotic indicators, including plankton communities, have been neglected and not included as an important component of water quality assessment.
Zooplankton play an important role in aquatic ecosystems, facilitating energy transfer from lower to higher trophic levels in the aquatic food webs, and are sensitive to environmental changes (Van Egeren et al. 2011, Jeppesen et al. 2011, Mimouni et al. 2015).There is evidence of a well-established relationship between multiple environmental drivers and the structure of zooplankton communities (Dodson 1992, Pinel-Alloul et al. 1999).The sensitivity of zooplankton to both bottom-up (i.e., water chemistry and control by phytoplankton) and top-down pressures (i.e., predation by macro-invertebrates and fish) are well studied (Mimouni et al. 2015, Patoine et al. 2002, Van Egeren et al. 2011).Zooplankton studies in Indonesian freshwater ecosystems have mainly focused on taxonomic diversity, and monitoring studies have poor biological inventories due to the general lack of taxonomic studies of freshwater zooplankton.In particular, little is known about the zooplankton community structure in Lake Tempe.Our previous study on species composition and diversity of zooplankton in Lake Tempe found that the zooplankton in the lake are typically characterised by the dominance of rotifers, with a significantly lower number of daphniid species (Toruan and Setiawan 2017).
Zooplankton community structure has traditionally been assessed using a taxon-based approach, involving the analysis of individual taxa abundance and community species richness in response to environmental heterogeneity.However, as energy transfers through the food chain, predator-prey relationships and feeding behaviours in the ecosystems are more dependent on body size than individual species (Dickie et al. 1987, Hébert et al. 2016), and a size-based approach would provide a functional perspective of the community structure and ecological interactions as compared with individual species (Woodward et al. 2005, Litchman et al. 2013).The use of body size as a metric to characterise ecological organisation is based on empirical evidence that many fundamental physiological and ecological processes in aquatic ecosystems, such as growth, respiration and production, are allometrically correlated to body size (Schmidt et al. 2006, Litchman et al. 2013, Peters 1992).Community-level studies using the body-size approach focus on the distribution of individual biomass among size classes, regardless of the variation in component species (Dickie et al. 1987).As body size is a key element determining the dynamics of community organisation and ecological networks (Schmidt et al. 2006, Peters 1999), deviations in community size structure in response to perturbations may affect ecosystem function.As a proxy to understand the ecological shift of the zooplankton community, the sensitivity of the body-size metric in response to multiple environmental stressors has been tested in many studies.This has been shown through the effect of top-down (i.e., predation) and bottom-up (i.e., eutrophication) factors on zooplankton size distribution (Braun et al. 2021, Li et al. 2022, Mines et al. 2013, Ye et al. 2013), changes in zooplankton body size along urbanisation gradients (Brans et al. 2017, Toruan et al. 2021) and the influence of land-use changes and forest fires on zooplankton community size structure (Patoine et al. 2002, Ghadouani et al. 2006).
The overall objective of this study is to examine the temporal and spatial dynamics of zooplankton in Lake Tempe, Indonesia.Water quality indicators and zooplankton abundance and biomass size spectra data from three sampling regions (Northern, Central and Southern) in Lake Tempe were collected from March to December 2016, and laser optical plankton counter (LOPC) analyses were performed on preserved samples.The specific aims are as follows: (1) to describe the spatial and temporal patterns of zooplankton abundance, biomass and size structure in Lake Tempe, and (2) to assess which environmental variables drive the dynamics of zooplankton community size structure in Lake Tempe.

Study site
Lake Tempe is a riverine floodplain system located in the South Sulawesi Province of Sulawesi Island, Indonesia (Fig. 1).The lake system was formed along the Wallanea-Cendranae River and is connected to three main riverine systems, namely Wallanae-Cendranae in the south, Bila in the north and Batu-batu in the west.A total of 23 rivers drain into Lake Tempe, with only one outlet at the River Cendranae.The total catchment area of the lake is 3288 km 2 , which is divided into three main subcatchments: Bila (1667 km 2 ), Sidendreng (739 km 2 ) and Batu-batu (733 km 2 ).The total lake surface area is 478 km 2 ; the predominant land use in the catchment is for irrigated rice and agricultural fields (Fig. 1a), resulting in significant nutrient inputs into the lake.Human activities surrounding the lake have also increased substantially, contributing to the increase of domestic sewage being discharged directly to the lake.Nutrient enrichment and siltation in the lake has increased, mainly since the 1990s, due to land use changes in the catchment (Jasalesmana et al. 2014, Setiawan andWibowo 2013).Most areas of Lake Tempe are very shallow, with water levels fluctuating seasonally, resulting in some areas being temporarily dry during the dry period.The mean and maximum depth of the lake also varies seasonally: the minimum depth was 1 m, while the maximum depth was 5.3 m, which was observed during the flood periods.
Macrophyte coverage was estimated at 40%, of which water hyacinth (Eichhornia crassipes) and water spinach (Ipoemea aquatia) were the predominant species.The fish community in Lake Tempe was mostly omnivorous and herbivorous.During our study period from January to December 2016, silver barb fish (Barbonymus gonionatus) was the most dominant fish in the lake followed by golden tank goby fish (Glossogobius aureus) and tank goby fish (Glossogobius giuris).

Field sampling
Field sampling was conducted monthly from March to December 2016 in three sampling regions of Lake Tempe (Southern, Central and Northern; Fig. 1b), with three sampling points in each area.The southern region is characterised by massive macrophyte beds and is shallow and temporarily dry during the dry period; the central region is the deepest part of the lake, permanently inundated and a fisheries conservation area, while the northern region is an open water area, permanently inundated and is the inlet area from two riverine systems.Rainfall data were collected from rain gauges located within the catchment (Matajang, Ugi, Betao, Watang Kalola, Lapajung, Wala, Baruku, Sengkang, Malanroe and Tandru Tedong); the network of stations were located 11-45 km from Lake Tempe.
Zooplankton were sampled using a 56 µm mesh size plankton net (length 60 cm, diameter 30 cm) by vertical tow from just above the sediment bed to the surface, preserved in 4% sugar-buffered formaldehyde and transported to laboratory for further analysis; where water depth was less than 1 m, 20 L of water was filtered using the same plankton net.Due to administrative and logistical issues, samples could not be collected in January and February 2016; thus, our data only represents a 10 month period, instead of a full 12 months.
Water temperature, dissolved oxygen (DO), pH, electric conductivity and water turbidity were measured in situ at the subsurface (approximately 10-20 cm below the surface) with a Horiba U-51 Water Quality Checker (HORIBA Advanced Techno Co., Ltd., Kyoto, Japan).Water depth was measured in metres, while water transparency was measured as Secchi depth, also in metres.Water samples for nutrient analysis, including total nitrogen (TN), total phosphorus (TP) and chlorophyll-a (chla), were collected using a Van Dorn sampler (Eijkelkamp Soil & Water, Giesbeek, the Netherlands) from the subsurface at each sampling location.Following standard methods (APHA, 2013): TN was determined by a spectrophotometric method involving persulfate digestion; TP was determined by a spectrophotometric method with ascorbic acid; and, total chl-a was determined by a spectrophotometric method involving acetone extraction.All analysis for water quality parameters were conducted by the Hydro-chemistry Laboratory at the Research Centre for Limnology, part of the National Research and Innovation Agency (BRIN), Republic of Indonesia (formerly known as the Indonesian Institute of Sciences, LIPI).Macrophyte coverage and fish community were recorded as part of the field survey campaign and were reported separately from this study.Macrophyte data are as reported in Kurniawan et al. (2017), while fish community data are available in Dina et al. (2019).

Zooplankton data collection
Data for zooplankton particle size, abundance and biomass were collected using a lab circulator laser optical plankton counter (LOPC; Rolls-Royce Naval Marine, Peterborough, Canada).The LOPC counts and estimates the size of zooplankton particles as equivalent spherical diameter (ESD) (Herman et al. 2004).Although the LOPC can detect particle sizes from 100 to 4500 µm ESD, for this study only particle size biomass data for the sizes between 300 and 2000 µm ESD were used, as we found that air bubbles in the lab circulator LOPC prevented accurate counts for particle sizes < 300 µm and there were extremely low counts for zooplankton larger than 2000 µm ESD in all samples that were analysed.The LOPC data were processed using the LOPC data processing software developed by Herman et al. (2004).Using an LOPC to count abundance and estimate zooplankton biomass has several advantages, including (1) it is a robust and quick method, which allows for a large number of samples to be processed, supporting large-scale analysis, and (2) little to no taxonomic background knowledge is required prior to sample processing.However, a LOPC may not be ideal to directly process samples from shallow water bodies, and in these cases a LOPC equipped with a sample circulator is required.Additionally, in previous studies it has been found that the presence of small bubbles in samples can contaminate LOPC data below 300 µm (Toruan et al. 2021, Finlay et al. 2007b).

Zooplankton size spectra
Zooplankton community structure was quantified using sizebased metrics including zooplankton mean body size (ESD, µm), zooplankton abundance (as individuals, ind L −1 ), zooplankton biomass (mg L −1 ) and the normalised biomass size spectra (NBSS).The mean body size here refers to the weighted mean body size by abundance measured as ESD, as measured by the LOPC.Biomass was calculated using the equation as in Finlay et al. (2007a), which assumes that the biovolume of zooplankton is equal to the volume of an ellipsoid: where f is the length to width ratio.The aggregate abundance and biomass for each sample were determined per volume of water sampled.Furthermore, the biomass size spectrum was determined by binning the size biomass data for particle size between 300 and 2000 µm ESD into a series of logarithmically equal size intervals which resulted in 26 bins.These biomass size spectra were then normalised by dividing the biomass in each bin size by the width of the bin.

Statistical analyses
Zooplankton size and biomass distribution were plotted as histograms for the sizes between 300 and 2000 µm ESD, with bin intervals for every 105 µm ESD.The LOPC processing software produced 15 µm bins, which were used to plot the NBSS; however, for the purpose of displaying the size and biomass distribution more clearly, 105 µm bin intervals were used.
A one-way analysis of variance (ANOVA) was performed to examine the spatial and temporal differences among environmental variables, zooplankton mean size, abundance, biomass, NBSS slope and NBSS intercept.Fisher's least significant difference (LSD) post hoc testing was then performed to determine the level of differences.
Ordination using principal component analysis (PCA) was used to summarise and display the variability of the environmental factors in Lake Tempe and to identify the environmental variables that contribute to the variability in zooplankton size-spectra metrics; PCA biplots were plotted using R (version 4.2.0) with packages factoextra (version 1.0.7),ggplot2 (version 3.4.1)and ggpubr (version 0.60) (R Core Team, 2021).Multiple linear regression models, performed in SigmaPlot (version 14.0, Systat Software, Inc, Chicago, USA), were then used to verify the relationship between zooplankton size-spectra metrics (e.g., mean body size, total abundance, total biomass, NBSS slope and NBSS intercept) and environmental variables, with the explanatory variables including water temperature, water transparency, TP, TN and total chl-a.All data, except zooplankton mean body size, NBSS slope and NBSS intercepts, were log-base 10 (log 10 ) transformed prior to regression analysis to meet the normal distribution based on the assumption of equal variance and normality of residuals of the data.All ANOVA and regression analysis were considered significant when p < 0.05.

Hydrological characteristics
The Lake Tempe floodplains are characterised by shallow waterbodies and varied water levels throughout the year depending on rainfall events (Fig. 2a).From 2000 to 2010, mean monthly water levels at Lake Tempe ranged from 3.26 to 5.37 m above sea level (ASL) with the average being 4.3 m ASL.During the study period, March to December 2016, the water depth ranged from 3.25 to 8.67 m ASL, with an average of 5.59 m ASL (Fig. 2b).Most areas of Lake Tempe were very shallow, with water levels fluctuating seasonally, resulting in some areas being temporarily dry during the dry period.The rainy season within the catchment was observed in November; however, the most significant flow from the main river into the lake was observed in December, resulting in the observed increase in lake water level.During May-July, the floodplain was fully inundated, while the dry period started in 92 Page 6 of 16 August and lasted through to October, when the water level was its lowest.

Environmental variables
Most of the variability in the environmental conditions in the lake were characterised by variations in mean water depth and transparency, chl-a, TP, TN, pH and DO, as shown by the PCA ordination (63% of the total variation; Fig. 3).Water depth varied between 1.79-3.15and 1.75-2.35m during the wet and dry periods, respectively (Fig. 2b).Similarly, water transparency, measured as Secchi depth, was slightly lower in the dry period than in the wet period (Fig. 2b).The water temperature in Lake Tempe was slightly warmer during the dry period, with the highest mean temperature of 30.6 °C being recorded during September-November (Table 1).Lake Tempe is considered eutrophic, with seasonal variabilities in nutrient concentrations such as TP, TN and chl-a (Fig. 4).Early in the rainy period, TN ranged between 500 and 600 µg L −1 , and with increasing water levels associated with the intense rainy period, increased to 700-800 µg L −1 .TN increased during the dry period to a concentration of 2000 µg L −1 .Similarly, TP concentrations varied across the sampling months, with lower concentrations observed during the early wet months and increased as the lake become   during the high water level period in June-August; TP was lower in the wet period compared with in the dry period, with an average concentration of 100 µg L −1 and 146 µg L −1 in the wet and dry periods, respectively.Regarding chl-a, the concentrations temporarily fluctuated, ranging from the lowest level (3.8 µg L −1 ) to its highest levels (~46 µg L −1 ) during the wet and dry periods, respectively.There was no spatial variation in chl-a concentration (F 2,48 = 0.255, p = 0.776), TP (F 2,48 = 0.25, p = 0.793) or TN (F 2,48 = 0.052, p = 0.979).
Table 1 Environmental variables during the wet and dry periods in Lake Tempe.Data are presented as mean values with the range (minimum to maximum) in parentheses; the wet period is from November to July and the dry period is from August to October.

Environmental variables
Wet period mean (min and max) Dry period mean (min and max) All months mean (min and max)

Zooplankton community size structure
The zooplankton community in Lake Tempe was characterised by a high density of small-sized zooplankton within the size range of 300-700 µm and an extremely low density of zooplankton larger than 700 µm in all sampling months.Ordination using PCA shows that more than 70% of the total variation of density distribution was shown by zooplankton sizes between 300 and 700 µm ESD, as these size classes (i.e., size < 700 µm) are oriented along the first axis (PC1) of the PCA, with longer vectors than the larger size classes.(Fig. 5a).Similarly, small-sized zooplankton also contributed significantly to the total aggregate biomass (Fig. 5b); however, the sizes between 700 and 1140 µm had a notable contribution towards the total biomass, as these size classes oriented along the first axis (PC1) of the PCA, with longer vectors than the other large size classes (Fig. 5b).
The overall summary of the zooplankton community size structure in Lake Tempe is shown in Fig. 6, where the overall mean body size is smaller than 500 µm ESD.There were significant differences between sampling months for zooplankton mean size (F 9,48 = 4.69, p < 0.001), total abundance (F 9,48 = 5.591, p < 0.001), total biomass (F 9,48 = 3.986, p = 0.01), NBSS slope (F 9,48 = 4.36, p < 0.001) and NBSS intercept (F 9,48 = 9.725, p < 0.001).The largest mean body size was observed in the dry period while the smallest mean body size was observed in the wet period.Total zooplankton density ranged from 49 to 393 individuals per litre (ind L −1 ), with lower densities being observed in the wet period, especially in March, April and May, and fluctuated as the lake become fully inundated.The highest densities were observed in July and December, which corresponded with high and low water level periods, respectively.Similarly, total zooplankton biomass was low during March-May and increased substantially from June to December.The highest total biomass was observed in November at 20.82 mg L −1 .When the zooplankton community was classified into four size classes (i.e., 300-500, 501-750, 751-1000 and 1001-2000 µm), the zooplankton community in Lake Tempe was dominated by smallest size group (Fig. 7).The 300-500 µm size class contributed to more than 60% of the total density followed by the 501-750 µm size fraction, in all sampling months.The size classes larger than 751 µm contributed less than 10% to the total density.In contrast, biomass of the 300-500 and 501-750 µm size classes contributed to 60-80% of the total biomass, while the 1001-2000 µm size class was about 20-30% of the total biomass.In this study, there were no spatial differences observed in zooplankton mean body size (F 2,48 = 0.0668, p = 0.429), abundance (F 2,48 = 0.355, p = 0.703) or biomass (F 2,48 = 1.183, p = 0.316).
Complete results of the NBSS analyses based on zooplankton biomass data in each 15 µm bin size for the size fraction from 300 to 2000 µm ESD are shown in Fig. 8, while Table 2 summarises the spatial and temporal NBSS slope and NBSS intercept.The mean NBSS slope ranged from −1.06 to −1.73, and slopes in this range (i.e., greater than −1) indicate the dominant contribution of small-sized zooplankton towards total biomass.

Environmental drivers of zooplankton size spectra
Variability in zooplankton community size structure was related to eutrophication parameters including TP, TN, chl-a and Secchi depth.Multiple linear regression analysis (Table 3) showed that total biomass was strongly correlated with chl-a concentration (p < 0.001, R 2 = 0.40).TN and chl-a were positively significantly correlated with total zooplankton abundance (p < 0.01, R 2 = 0.54), while mean body size was positively significantly correlated with water temperature (p < 0.01, R 2 = 0.21).The NBSS parameter showed a significant correlation with environmental variables, with the NBSS slope being significantly correlated with TP, chla, water temperature and Secchi depth (p < 0.01, R 2 = 0.48; Table 3).Conversely, only TP and water temperature showed a significant correlation with NBSS intercept.

Discussion
We explored the main ecological factors associated with zooplankton community size structure in a tropical riverine-floodplain lake.Using zooplankton body size metrics, we were able to detect the variation in zooplankton mean body size, abundance and biomass size spectra in Lake Tempe (Fig. 6).Most of the variability in zooplankton abundance and biomass was seen in the smaller size fractions (e.g., sizes 300-500 µm) (Fig. 7).The seasonal hydrological regime and bottom-up drivers, such as nutrients parameters, are most important drivers of zooplankton size structure in the lake (Table 3).
Previous studies have indicated that zooplankton community structure in river-floodplain systems is strongly associated with habitat diversity and hydrological regimes (Golec-Fialek et al. 2021, Sharma 2005).Such association is caused by the dynamics of flood events, which is the driving factor for spatial and temporal variations of community structure in riverine lake systems (Junk et al. 1989, Toruan  and Setiawan 2017).Changes in hydrological phases, from low-water to high-water periods in riverine-floodplain lakes are characterised by distinct biotic assemblages, which are related to different ecological mechanisms in each phase.For example, environmental variability following flooding and non-flooding events have a significant role in shifting zooplankton assemblages from crustacean-dominated to rotiferdominated communities (Chaparro et al. 2019, Toruan andSetiawan 2017).
Our results indicate that small-sized zooplankton within the size range 300-500 µm are dominant in Lake Tempe (Fig. 7a), which was positively correlated with eutrophication parameters including TP and chl-a concentration (Table 3).Zooplankton biomass and size diversity in lake ecosystems is driven by environmental variables including eutrophication (Sprules and Munawar 1986), lake acidification (Pinel-Alloul et al. 1990), water transparency (Stemberger and Miller 2003), biotope and ecosystems size (Gaedke, 1992), as well as biotic stressors, such as cyanobacterial blooms (Ghadouani et al. 2006), invasive species (Mines et al. 2013) and fish predation (Yurista et al. 2014).The positive association of abundance and biomass of small-sized zooplankton to chl-a concentration indicates that the increase in resource availability, such as phytoplankton biomass as indicated by total chl-a concentration, may be related to the increasing aggregate properties of the zooplankton community (biomass and density).Comparable results were found in previous studies in which low biomass and small average size of zooplankton are common in eutrophic lakes and during cyanobacterial blooms (Sprules and Munawar 1986, Gaedke et al. 2004, Ghadouani et al. 2006).Larger-sized zooplankton, such as the large filter feeding Daphnia, are usually absent with the occurrence of cyanobacterial blooms (Ghadouani et al. 2006, Ersoy et al. 2017); in this study, we also found that large size cladocerans (size range 1001-2000 µm), such as daphniids were absent from Lake Tempe.This finding is also consistent with our previous findings on zooplankton community structure in Lake Tempe, in which we identified that smaller cladocerans from families Bosminidae and Chydoridae were the most dominant, along with small cyclopoid copepods (Toruan and Setiawan 2017).
The contribution of small-sized zooplankton towards the total biomass and abundance of zooplankton in Lake Tempe was also reflected by the steeper NBSS slopes, where the slopes ranged from −1.06 to −1.73 (Fig. 8).The theoretical value of NBSS in stable freshwater ecosystems is close to −1 (Sprules andMunawar 1986, Sprules et al. 2016), and the deviation from this value indicates  2 92 Page 12 of 16 a shift in zooplankton community structure (Gaedke et al. 2004, de Eyto andIrvine 2007).Steeper slopes indicate high productivity but low energy transfer efficiency to higher trophic levels, resulting in a higher biomass concentration in the smaller-sized organisms, while a moderate slope indicates low productivity with high energy transfer efficiency, and thus a higher biomass of the largersized organisms (Zhou 2006).
While this study was not explicitly designed to test the link between the fish population and zooplankton community structure, Lake Tempe has been impacted by long-term intensive fisheries activities, including over-harvesting and fish introduction.In fact, the economic revenue from fisheries activities is a significant contribution to the local and provincial economy.Fish over-harvesting has been linked to a significant decline in fish stock, and fish restocking in the lake was implemented as early as 1923.The lake has since been stocked with commercial fish which are mostly nonnative to the lake.A recent study showed that this lake is dominated by a high abundance of omnivorous and planktivorous fish such as silver barb fish (Barbonymus gonionatus), golden tank goby fish (Glossogobius aureus), tank goby fish (Glossogobius giuris) and Nile tilapia (Oreochromis niloticus) (Dina et al 2019).In another study, we directly tested the link between fish and zooplankton community structure in Lake Tempe (Toruan et al. 2022) and found that there was no direct link between fish density and zooplankton community structure in Lake Tempe; this may be linked to longterm fish predation pressure (Hambright 2008, Quintana et al. 2015).Zooplankton community size structure can be influenced by fish predation.Size-selective predation by fish has been a key influence to control the size diversity of zooplankton (Dodson 1974, Ersoy et al. 2017).In addition, fish predation is also responsible for the absence of large-bodied zooplankton from lake ecosystems.Intense predation by planktivorous fish may cause a shift in the zooplankton community structure, from large-sized to small-sized dominated; thus, a high density of planktivorous fish could also have a positive effect for smaller crustacean communities (Jeppesen et al. 1996, Jeppesen et al. 1997, Zhang et al. 2013).Zooplankton are critical part of freshwater ecosystems, connecting the lower and higher trophic levels, thus its distribution influences the entire ecosystem and interactions with other planktonic organisms (Pinel-Alloul et al. 2021, Pinel-Alloul et al. 1988).Zooplankton abundance and biomass have a strong effect on the biomass of other planktonic organisms (i.e., phytoplankton population by consumption) and fish populations (by being consumed) (Vanni 2002).
Several studies and experiments have documented that, in lakes with high levels of planktivorous fish, the zooplankton community is often dominated by small cladocerans, copepods and rotifers (Bonecker et al. 2011, Jeppesen et al. 1997, Liu et al. 2020).The absence of large cladocerans, such as daphniids, in our previous study (Toruan and Setiawan 2017) was consistent with the findings of this study in which density and biomass of zooplankton larger than 700 µm are significantly low.These results could be linked to the reduced fish yield in Lake Tempe in the previous years and the current declining trends.The low abundance of zooplankton usually co-occurred with low fish recruitment, especially during the fish spawning period.The high abundance of prey during the spawning and early life stages will increase the recruitment success (Platt et al. 2003).The low density of zooplankton observed in Lake Tempe may indicate that there might not be enough food resources available during the recruitment period and, in the long run, this may result in the declining fish productivity.
The use of zooplankton as an ecological indicator for monitoring environmental changes is significant, as they are sensitive to environmental perturbations, as confirmed by the results of this study (Table 3).Zooplankton communities demonstrate fast responses when disturbances occur, including eutrophication, acidification, predation and hydrological changes (García-Chicote et al. 2019, Beaver et al. 2013, Yin et al. 2022).As such, changes in zooplankton community structure can be used as a biological quality indicator for lake water quality monitoring and conservation plans (Jeppesen et al. 2011, Pinel-Alloul et al. 2021).With the increasing urgency of the Indonesian government to prioritise the rehabilitation of ecosystems of critical importance to their people and economy, the results of this study are applicable to more than just Lake Tempe.The other Indonesian priority lakes not only share similarities in their anthropogenic disturbances, but also their geographic region.As such, the results of this study can be used to inform the lake managers to include zooplankton community structure as a biological quality indicator in their monitoring, management and rehabilitation plans, not only for Lake Tempe, but other Indonesian national priority lakes.

Conclusions
This study presents the temporal and spatial zooplankton community size structure and their relationship with environmental variables in Lake Tempe from March to December 2016.The zooplankton community in Lake Tempe was characterised by a high abundance of small-size zooplankton with overall mean body sizes smaller than 500 µm ESD.Despite no spatial variation in community size structure, we found temporal changes in community size structure.Eutrophication parameters, such as TP and chl-a, along with water temperature, were the main environmental variables driving the variability of zooplankton community structure in Lake Tempe.The data collected in this study on zooplankton community size structure and their relationship with environmental variables provide a significant addition to the growing body of knowledge and understanding of assessing zooplankton community changes by size and biomass distribution.Moreover, the results of this study will be critical in the development of future management plans for the regionally significant Lake Tempe.In the broader context of lake management, the information obtained through this study provides a significant contribution towards the management of priority lakes in Indonesia, in which the results presented here are transferable to other similar lakes in the region.

Fig. 1
Fig. 1 Maps of Lake Tempe showing a land-use classification in the Lake Tempe catchment and b sampling zones on a bathymetric map of the lake fully inundated.The TP concentration ranged from 40 to 70 µg L −1 during March-May, and from 130 to 140 µg L − 1

Fig. 2
Fig. 2 Monthly variability of the hydrological characteristics of Lake Tempe.a Rainfall (left axis) data were collected from the nearest rain gauge station in the Lake Tempe catchment, and mean monthly water level (right axis) data are from Setiawan (unpublished data), both are shown as a 10 year average from 2000 to 2012; mean monthly water

Fig. 3
Fig. 3 Principal component analysis (PCA) ordination of the eight environmental variables examined in Lake Tempe from March to December 2016.A visual representation of the contribution of the control variables to the principal axes is shown by the colour bar (numbers as percentage, calculated using R package factoextra)

Fig. 4
Fig. 4 Temporal variability mean (± standard error) of chlorophyll-a, total phosphorus (TP) and total nitrogen (TN) concentrations in Lake Tempe from March to December 2016 (a-c), and the spatial mean

Fig. 5
Fig. 5 Biplots of principal component analysis (PCA) showing a the zooplankton size density and b biomass distribution of each of the 105 µm bin size intervals.For each vector shown, the X prefix denotes greater than, e.g., X510 refers to sizes greater than 510 µm.Contributions (%) of each of the control variables to the principal axes is shown by the colour bar

Fig. 6
Fig. 6 Temporal (a-e) and spatial (f-j) variation of zooplankton community size structure.Parameters shown are: a, f mean body size; b, g total abundance (aggregate); c, h total biomass (aggregate); d, i NBSS slope and e, j NBSS intercept.Temporal variation was determined based on monthly data averaged from the three sampling areas,

Fig. 7
Fig. 7 Relative abundance (a) and biomass (b) of three different size classes of zooplankton plotted with total abundance and biomass of all sizes combined.Please note the relative abundance and biomass are shown on the primary axis (stacked bars), while total and abun-

Fig. 8
Fig. 8 Monthly zooplankton normalised biomass size spectra (NBSS) of Lake Tempe.The black dots represent the normalised biomass in each size class, while the regression line indicates the slope of the NBSS.A summary of the slopes and intercepts is presented in Table 2

Table 2
Slopes and intercepts of the normalised biomass size spectra (NBSS) of net collected zooplankton in all sampling regions in Lake Tempe from March to December 2016.

Table 3
Results of multiple linear regression models showing the correlation between environmental variables, chlorophyll-a (chl-a), total phosphorus (TP), total nitrogen (TN), temperature (Temp), Secchi depth and zooplankton size spectra metrics, including normalised biomass size spectra (NBSS).Values indicate the coefficient of correlation where significant correlation (p < 0.05) are marked with asterisks (*; exact p-values included).R 2 values indicate the percentage of variation explained