The combined effects of multiple stressors on freshwater biota are still poorly understood. Recent studies on benthic invertebrates have shown that complex interactions frequently occur, leading to often unexpected responses. Recent developments in DNA metabarcoding have revealed that responses typically inferred for benthic invertebrates are often underestimated due to a lack of taxonomic resolution. Furthermore, stressor impacts on important biological traits such as biomass and size have rarely been addressed.

The two latter aspects can now be potentially overcome by coupling DNA-based identification with optical methods and machine learning. The innovative combination of approaches has the potential to significantly advance multiple stressor research by matching changes in species and OTU (operational taxonomic units) diversity with their biological traits. Project A08 will investigate multiple stressor impacts on benthic invertebrate communities and taxon-specific traits and link these to changes in organic matter (CPOM) decomposition as an essential ecosystem function. The project has three goals and workpackages (WPs): In WP1 we will assess individual and combined effects of increased temperature, salinity and reduced flow on the invertebrate community by building on results of the three replicated ExStream field experiments coordinated by project Z02.

WP1 will utilise automatic image-based taxon recognition and biomass estimation to also register changes in size and biomass per taxon. While we expect a reduction of species richness upon stressor application, responses are predicted to be largely taxon specific. Furthermore, we expect sensitive benthic invertebrate taxa (e.g. Ephemeroptera, Plecoptera, Trichoptera, ‘EPT’) to decrease in abundance but also overproportionately in biomass, as a higher proportion of the energy budget has to be allocated to physiologically coping with the stressor (homeostasis, higher basal metabolism), while a smaller proportion of the energy budget remains for somatic growth. In WP2 we will analyse all counted and measured benthic invertebrates obtained from WP1 using DNA metabarcoding to maximise response pattern detection at species/OTU level.

We expect a much greater diversity of responses due to improved resolution especially for smaller taxa such as Chironomidae. Based on the Asymmetric Response Concept (ARC), we predict community recovery after stress release to be largely stochastic, yet community composition to be significantly influenced by the endpoints of communities under stress, e.g. due to priority effects. WP3 will assess degradation of CPOM degradation under multiple stressor increase and release and link these important ecosystem function measures to the measured community change parameters obtained in WPs 1 and 2. Through this link we can identify stressor scenarios and taxa associated to ecosystem function decline.

Iris Madge Pimentel (University of Duisburg-Essen)

Individual and combined stressor effects on freshwater invertebrate communities and an associated ecosystem function

Streams are increasingly exposed to multiple human-induced stressors which reduce biodiversity and alter the composition of species communities. An understanding of the effects that these stressors have on species communities and how the communities recover after removal of the stressors is crucial to pave the way for the success of stream restoration projects. Focusing on invertebrate communities in streams, this project studies both, the effects of and the recovery from three common human stressors: temperature increase, salinization and flow velocity reduction. In addition, the project explores the combination of two complementary approaches for community analyses: DNA metabarcoding and automated image-based species identification with deep-learning methods.

Freshwater invertebrates substantially contribute to leaflitter decomposition and thereby to the provisioning of nutrients within the food web of rivers. Unfortunately, stream restoration actions often entail only slow and incomplete reassembly of the natural invertebrate communities. This makes evident that processes of community degradation and recovery are not symmetric, as the recolonization of restored ecosystems depends on factors that are hard to predict. These factors include differential dispersal abilities of species, potential dispersal barriers, and the persistence of populations which may compete with or facilitate the establishment of former community members. So, how are stream invertebrate communities and their associated ecosystem functions such as leaflitter decomposition altered by common human stressors? And in how far does their recovery differ depending on the stressor(s) they were exposed to?

The ExStream system is exceptionally suitable to answer these questions. Water is redirected from the stream through multiple mesocosms. This yields near-natural conditions in the mesocosms, enables sufficient replication for statistic inferences and allows for migration of invertebrates. We manipulate three variables of interest, increased temperature, salinization and reduced flow velocity. This allows us to analyze the degradation process of invertebrate communities by single and multiple stressors. In addition, we can explore how this affects ecosystem functioning, exemplified on leaflitter decomposition. To allow for recovery, we reset all environmental variables to natural conditions and study community reassembly upon stressor release.

Using morphological identification of invertebrates for community analysis is time-consuming and prone to misidentification of morphologically similar species. In contrast, DNA metabarcoding is a much faster approach and can even differentiate between such cryptic species by taking advantage of differences in DNA sequences in a specific gene region. However, it does not give data on abundances, although this is a crucial information for diversity measures. Here, the project explores the combination of DNA metabarcoding with automated image-based species identification to retrieve both comprehensive species lists and abundance data. First, convolutional neural networks which are trained on freshwater macroinvertebrates identified by experts are used to identify specimens to the lowest taxonomic level possible. This generates abundance data for species that can be recognized by the algorithm. Second, DNA metabarcoding is applied to retrieve a comprehensive taxa list including also morphologically similar species.

Contact: iris.madge-pimentel@uni-due.de

First Supervisor: Prof. Dr. Florian Leese (University of Duisburg-Essen, Aquatic Ecosystem Research)
Second Supervisor: Dr. Jeremy Piggott (Trinity College Dublin (Ireland), Zoology)
Mentor: Dr. Thomas Ehlert (Bundesamt für Naturschutz)