Introduction
In the current era of the Anthropocene, river ecosystems worldwide are extensively modified by human activities (Cid et al. , Dudgeon et al. 2006). Land-use practices are one of the main threads that leads to the loss of freshwater biodiversity (Dudgeon 2019). Besides the direct negative effect on overall abundance and richness of organisms, indirect effect may co-occur by altering the linkages among habitats and communities (Foley et al. 2005). Rivers and their adjacent terrestrial zones are tightly linked via multiple fluxes and pathways (Polis and Hurd 1996, Baxter et al. 2005, Bartels et al. 2012). In this study, we focus on the animal-mediated flux from land-to-water, which is the one of terrestrial invertebrates falling into water bodies and being eaten by fish. This flux can be pronounced and may provide up to 50% of the annual energy budget of fish individuals (Nakano and Murakami 2001, Baxter et al. 2005). While it has been shown that even the nutrient pool of a lake can substantially be subsidized by terrestrial phosphorus excreted by fish feeding on surface insects (Mehner et al. 2005), these fluxes are generally assumed to be strongest in aquatic systems with extended riparian zones, especially in streams with a high canopy cover (Edwards and Huryn 1996, Kawaguchi and Nakano 2001). Although this animal-mediated flux between adjacent ecosystems is increasingly recognized, little is known on how anthropogenic perturbations may influence the strength of such exchanges (but see Larsen et al. 2016). For example, the degradation of riparian zones, such as from livestock grazing (Saunders and Fausch 2012) or deforestation (Kawaguchi and Nakano 2001), significantly decreased the reliance of fish on terrestrial insects. However, another aspect of land use with a strong impact on our river ecosystems and the potential to alter food web dynamics is the one from mining activities.
Lignite (brown coal) mining activities on aquatic systems is influencing the chemical, physical, biological, and ecological properties of the system (Gray 1997, Byrne et al. 2012), and iron is one of the key contaminants involved. Due to the oxidation of ferrous sulphide metals (e.g. pyrite and marcasite) during mining or related to the progressive groundwater resurgence in mine reclamation areas, large amounts of iron contaminate groundwater and adjacent aquatic systems. The well-oxygenated and neutral pH conditions in lotic systems will evoke a flocculation of iron (III) oxides, leading to browning and a turbid environment (Gray 1997, Byrne et al. 2012; Figure 1). Besides potential toxic effects via direct metal uptake (Vuori 1995), river organisms suffer from the formation of precipitates covering gills and eggs (Andersson and Nyberg 1984, Gerhardt 1992). However, most dominantly, iron precipitates alter the habitat availability for stream organisms as the fine iron sediments bury hard substrates, periphyton, organisms, and clog interstitial spaces of the benthic habitat (Letterman and Mitsch 1978, McKnight and Feder 1984). The communities of benthic invertebrates in river sections characterized by high iron oxide concentrations are therefore typically found to be of lower abundance, species biomass, and diversity (Rasmussen and Lindegaard 1988, Vuori 1995, Maret et al. 2003, Cadmus et al. 2016). In turn, this could have consequences for the food availability of fish inhabiting the river sections of high iron concentrations, potentially limiting the trophic transfer and the energy available for fish production. Furthermore, we suggest that iron has the potential to restructure the lotic food webs and rearrange the connections to the terrestrial systems. It can be expected that fish inhabiting river sections where benthic invertebrates as food is scarce rely to a higher degree on terrestrial invertebrate food resources.
A straight-forward approach to estimate the contribution of terrestrial insects to the diet of fish is to simply look at the gut content (Manko 2016). However, the identification of different prey items can be problematic in cyprinid fishes that have pharyngeal teeth that crush the food. Here, dietary studies using DNA metabarcoding can provide a high resolution of resource use (Amundsen and Sánchez-Hernández 2019). Nonetheless, DNA metabarcoding of gut content analyses only allows a snap-shot of the recently ingested prey. In addition, stable isotopes can provide an integrated estimate of resource use over longer time periods (Grey 2006, Boecklen et al. 2011). In the context of identifying the contribution of terrestrial insects to the diet of fish, stable isotopes of hydrogen (δ2H) are particularly useful due to the strong divergence between aquatic and terrestrial endmembers (Doucett et al. 2007, Vander Zanden et al. 2016).
Here, we studied the cumulative effects of iron (III) oxides on a lotic food web and the linkages to the terrestrial surroundings. Specifically, we predict that in river sections of high iron concentrations (i.e. upstream of the dam), less benthic invertebrates are present and fishes feed on terrestrial insects to a higher degree compared to fish individuals found in river sections of lower iron concentrations (i.e. downstream of the dam, following water remediation). This will be indicated by a higher number of terrestrial species found in the gut content (identified by their DNA-sequences using metabarcoding techniques), and further, by a higher terrestrial index calculated from hydrogen stable isotope values demonstrating the longer-term resource use. In addition, we predict that low abundance of benthic invertebrates upstream of the dam will decrease the general fish abundance, and hence the condition factor of the top predator (piscivorous pike Esox lucius ).