Predator-mediated diversity of stream fish assemblages in a boreal river basin, China

Predator–prey interactions are critical for understanding species composition and community assembly; however, there is still limited research on whether and how the prey species composition or community assembly in natural communities are mediated by predators. To address this question, we performed a field investigation to examine the influence of the presence of Lutra lutra on the diversity of fish communities of the Hunchun River Basin, Jilin Province, China. Our results indicate that L. lutra, as a potential umbrella species and generalist predator in the stream ecosystem, promotes the coexistence of a vast variety of fish taxa, which emphasizes the importance of top-down control in the ecological community. We suggest that L. lutra regulates the fish community assembly likely through the stochastic process. Although this was a pilot study regarding predator–prey interactions, the results highlight the effects of predators on the prey community assembly, and emphasize the role of predators on the maintenance of biodiversity and ecosystem function. Future conservation decisions involving ecosystem biodiversity should require the inclusion of predation intensity. The inclusion of scientific research and protection of umbrella species would thus constitute an additional and important step in biodiversity conservation.

Predator-prey interactions are critical for understanding species composition and community assembly 1,2 , both of which are vital mechanisms for biodiversity maintenance [3][4][5] . The bottom-up control of prey is an essential resource constraint that influences the diversity of prey species and predator 6,7 ; however, top-down regulation theories may fare better in grasping this issue [8][9][10][11] . Prey species diversity is controlled from the top-downward due to predation, which promotes diversification or coexistence among prey species by impacting prey species niche differentiation 12 . Several previous studies of laboratory microcosms have verified the role of predators in structuring the assembly and diversity of prey communities [13][14][15] . Nevertheless, whether and how predators mediate prey species composition or community assembly in natural communities remains unclear.
The community assembly of species is a key question in community ecology and is influenced by a combination of stochastic and deterministic processes [16][17][18][19] . Predators can affect the patterns of prey coexistence, relative abundance, and diversity, therefore also influencing the relative importance of stochastic versus deterministic processes in the prey community assembly 13,20 . However, an assessment of this effect is a major challenge because the degree of predator-mediated effects is poorly known. Predators can shape the prey species composition through both consumptive effects (CEs) (i.e., prey removal by predation) and non-consumptive effects (NCEs) (i.e., prey response driven by predation risk) 21,22 , which may shift the coexistence or community assembly of prey species 23 .
The type of predator also plays a key role in the community assembly of prey species. Generalist predators are predicted to reduce α-diversity (abundance or richness) and increase β-diversity by randomly consuming prey 24 . Specialist predators, which can filter out prey species that are intolerant of them, are predicted to reduce both α-and β-diversity 24 . The different effects of generalist and specialist predators correspond to the effects of stochastic and deterministic processes in the community assembly, respectively. Predator selectivity within a given pool of prey species theoretically plays a role in the prey community assembly, as well as the degree of CEs on the total prey community size. An empirical determination of how a local community assembly from a regional species pool is influenced by a predator is therefore a fundamental goal of community ecology, because predators can shift the patterns of prey coexistence, abundance, and diversity. www.nature.com/scientificreports/ Freshwater comprises only 0.01% and 2.3% of the water on Earth and the global land surface area, respectively 19 , but the number of fish species in freshwater ecosystems is similar to that in marine systems 25 . Recent studies have shown that freshwater organisms suffer from rapid population decline and high extinction risks 26,27 . Nevertheless, the majority of conservation literature remains biased toward terrestrial organisms, with less than 20% of recent studies focusing on aquatic species 28 . Many potential threats to freshwater biodiversity have been evaluated 27 , but the effects of predators on aquatic biodiversity have received less attention, which makes it difficult to understand how aquatic species maintain diversity 29 . Streams with only apex predator (Lutra lutra) in taiga forest systems represent an ideal model that can be used to examine the influence of predators on the prey community in natural systems. In this study, we conducted a field investigation to examine the influence of the presence of L. lutra on the diversity of fish communities in streams of the Hunchun River Basin, Jilin Province, China. Our study has three objectives: (1) Does the presence of L. lutra reduce the abundance of fish and increase the richness of fish species? (2) Does the presence of L. lutra increase β-diversity of the fish community in stream ecosystems? (3) Does the presence of L. lutra mediate the community assembly of fish by increasing the importance of stochastic or deterministic processes?

Materials and methods
Study area. The field study was conducted in the Northeast Tiger and Leopard National Park (42° 31′ 06″-44° 14′ 49″ N, 129° 5′ 0″-131° 18′ 48″ E) in the Hunchun River Basin (HRB), Jilin Province, China (Fig. 1). Fish assemblages from the streams were sampled to test the impacts of the presence or absence of L. lutra on fish biodiversity. We established transects along 10 streams in the southern area of the HRB (Fig. 1). Each stream was approximately 6 km long. The presence or absence of L. lutra in each stream was determined by (1) the presence www.nature.com/scientificreports/ or absence of L. lutra fecal samples collected monthly along each transect from June to September 2021 and (2) camera trapping data from the Tiger-Leopard Observation Network in China 30 , both of methods are widely used in the population investigation of L. lutra 31-33 . Data collection. We selected 4-6 sample sites for each transect from August to September 2021 and used the collected samples to investigate the compositions of the fish populations at these sites. Each sampling site covered an area with an upstream-downstream length of 200 m with a spacing of least 800 m. We measured the width, depth, flow, bottom quality of the stream, and vegetation cover (50 m around the stream) at 100-m intervals to obtain three sets of data regarding the stream environment at each sample site. Electrofishing was applied to collect fish at each sample site over a 0.5-h period. Each sample species was identified by referring to the relevant reference books and Fish Base Search (www. fishb ase. se/ home. htm) 34 . For each sample site, we recorded the number of each fish species after identification, and then measured the maximum length and weight of the dominant species (30% of the total number). All samples were released after data collection. The diversity of the fish assemblages was specifically represented by the species richness, Shannon-Wiener index, species evenness, catch per unit effort (CPUE, in 1-h period), and the CPUE of the dominant species. The Shannon-Wiener indices were calculated for each sample site according to (1): where H is the Shannon-Wiener index, xi represents the proportion of the i-th species to the total, and n is the number of fish species. The Shannon-Wiener index was calculated using the Past4 Project 1.0.0.0 (Past 4-the Past of the Future-Natural History Museum (uio.no)).
The species evenness was calculated according to (2): where E is species evenness, H is the Shannon-Wiener index, H max is the maximum Shannon-Wiener index to the total. The species evenness indices were also calculated using the Past4 Project 1.0.0.0 (Past 4-the Past of the Future-Natural History Museum (uio.no)).
We also calculated the β-diversity index to reflect the impact of L. lutra on the fish communities in each sample sites according to (3): where x and y are the abundances of fish species in the first and second fish communities, respectively, and z is the minimum abundance of the common species to both communities. The β-diversity indices were calculated using the R package functions 'vegan' and 'reshape2' .
The sampled fish species were assigned to functional categories based on diet (Nutritional Functional Group), morphological traits (Morphological Functional Group), habitat (Habitat Functional Group), and behavior (Reproductive Functional Group), as described in Fish Base Search (www. fishb ase. se/ home. htm).

Statistical analyses.
Linear mixed models (LMMs) were fitted to model the influence of L. lutra on the fish communities. The Shannon-Wiener index, species evenness, species richness, CPUE, β-diversity, abundance of the fish functional groups, and the CPUE of the dominant species were the response variables. The habitat variables of the streams, including the width, depth, flow, bottom quality of the stream, and vegetation cover were considered as random effects in the models to distinguish between the effects of non-independent variables. The presence (1)

Discussion
In this study, L. lutra is investigated as a potential umbrella species in a stream ecosystem [35][36][37] , whose presence naturally promotes the coexistence of a large number of fish taxa. The interaction of L. lutra and the fish community is an empirical example of the role of top predators in regulating prey biodiversity, and underscores the importance of top-down control in ecological communities. As a generalist predator, L. lutra was associated with a reduced abundance but enhanced richness of fish assemblages, both of which could enhance the fish community diversity. The β-diversity was also found to increase in the presence of L. lutra due to the enhanced fish species richness. We suggest that L. lutra likely regulated the fish community assembly through the stochastic process. Predator effects on the biomass and biodiversity of prey species are well understood 38 and also verified in this study. Differing from theoretical and empirical observations 20,24 , L. lutra as a generalist predator was found in this study to enhance the α-diversity of fish assemblages (e.g., Shannon-Wiener index and species richness). This indicates that the presence of L. lutra not only influenced the number of fish species, but also changed the uniformity of fish species in the community. The diet of L. lutra in this region mainly comprises stream fishes, of which Phoxinus, Cottus, and Cobitis make the highest contribution (unpublished data); these were also the dominant fish species of the stream ecosystem in this region. Therefore, when the stream fishes are under the predation of L. lutra, the number of the most dominant species (Phoxinus) decreases, whereas the uniformity of the fish species in the community would be enhanced, thus resulting in a higher Shannon-Wiener index. Compared with specialist predators, generalist predators may not completely remove a prey species at the local scale through CEs; as a result, generalist predators may alter the interspecific competition between prey species through NCEs and enhance the species richness of the community 39,40 .
Although our study does not explicitly elucidate the mechanisms of stochastic process in shaping the fish community assembly, we speculate that the increasing β-diversity of the fish community through stochastic process primarily results from the presence of L. lutra in the studied streams. Predators can either strengthen stochastic process and increase β-diversity by reducing the population size of all prey populations, or they can strengthen deterministic process leading to species extinction, or both processes can simultaneously occur in structuring the community assembly 20,41,42 . In our study, the presence of L. lutra was associated with a reduced abundance of dominant fish species and an overall abundance of fish, but clearly showed positive impacts on several rare fish species in the study area (e.g., S. malma, O. masou, L. reissneri). Both of these processes could  www.nature.com/scientificreports/ enhance the overall fish richness and species heterogeneity in different rivers, which may contribute to higher β-diversity. The habitat variables of the streams also showed no impact on fish community diversity. These lines of evidence support that the stochastic process predominates in the community assembly of fish species. To www.nature.com/scientificreports/ verify the coexistence mechanisms of the fish species, further studies should be designed in association with the phenotypic plasticity 43 and physiological response 44 in response to CEs or NCEs. Functional groups are sensitive to changes in species richness, especially the functional groups of species at higher trophic levels 45,46 . Our results indicate that the top-down effects of L. lutra could enhance the functional diversity of stream fishes, which may be due to the regulation of the predator on the prey species richness 39 . In this study, three rare species (e.g., S. malma. O. masou, L. reissneri) provided four unique functional groups, such as parasitism of the nutritional functional group, a flat and strong spindle of the morphological functional group, and a no-protagonist sinking egg of the reproductive functional group, all of which increased the functional groups owing to the higher species richness. Rare species often provide unique functional contributions, thus the loss of functional diversity may constrain the community and ecosystem processes 47 .

Conclusion
Our results show that the presence of L. lutra influenced the variations of taxonomic and functional diversity of fishes in streams of a taiga forest system. This work suggests that this predator can increase the importance of stochastic process in shaping the fish community assembly by reducing the size of all prey populations. This study also provides empirical field results that emphasizes the role of predators in maintaining biodiversity and ecosystem function. However, both of the interactions of different fish species and unique habitat selection of carnivorous fish species may simultaneously impact the fish community, except the impacts of L. lutra. As a pilot and preliminary research, our results help the understanding of predation on prey communities. If the impacts of multi-predator interactions on prey communities needed to be investigate, greater sampling efforts will be possible in the future. Future conservation measures involving biodiversity in ecosystems should require the inclusion of predation strength. The inclusion of scientific research and protection of umbrella species would thus constitute an additional and important step in biodiversity conservation.