Current rates of species loss triggered numerous attempts to protect and conserve biodiversity. Species conservation, however, requires species identification skills, a competence obtained through intensive training and experience. Field researchers, land managers, educators, civil servants, and the interested public would greatly benefit from accessible, up-to-date tools automating the process of species identification. Currently, relevant technologies, such as digital cameras, mobile devices, and remote access to databases, are ubiquitously available, accompanied by significant advances in image processing and pattern recognition. The idea of automated species identification is approaching reality. We review the technical status quo on computer vision approaches for plant species identification, highlight the main research challenges to overcome in providing applicable tools, and conclude with a discussion of open and future research thrusts.

Synthesis The interplay between bottom‐up and top‐down effects is certainly a general manifestation of any changes in both species abundances and diversity. Summary variables, such as species numbers, diversity indices or lumped species abundances provide too limited information about highly complex ecosystems. In contrast, species by species analyses of ecological communities comprising hundreds of species are inevitably only snapshot‐like and lack generality in explaining processes within communities. Our synthesis, based on species matrices of functional groups of all trophic levels, simplifies community complexity to a manageable degree while retaining full species‐specific information. Taking into account plant species richness, plant biomass, soil properties and relevant spatial scales, we decompose variance of abundance in consumer functional groups to determine the direction and the magnitude of community controlling processes. After decades of intensive research, the relative importance of top–down and bottom–up control for structuring ecological communities is still a particularly disputed issue among ecologists. In our study, we determine the relative role of bottom–up and top–down forces in structuring the composition of 13 arthropod functional groups (FG) comprising different trophic consumer levels. Based on species‐specific plant biomass and arthropod abundance data from 50 plots of a grassland biodiversity experiment, we quantified the proportions of bottom–up and top–down forces on consumer FG composition while taking into account direct and indirect effects of plant diversity, functional diversity, community biomass, soil properties and spatial arrangement of these plots. Variance partitioning using partial redundancy analysis explained 21–44% of total variation in arthropod functional group composition. Plant‐mediated bottom–up forces accounted for the major part of the explainable variation within the composition of all FGs. Predator‐mediated top–down forces, however, were much weaker, yet influenced the majority of consumer FGs. Plant functional group composition, notably legume composition, had the most important impact on virtually all consumer FGs. Compared to plant species richness and plant functional group richness, plant community biomass explained a much higher proportion of variation in consumer community composition.

Summary1. We studied the theoretical prediction that a loss of plant species richness has a strong impact on community interactions among all trophic levels and tested whether decreased plant species diversity results in a less complex structure and reduced interactions in ecological networks. 2. Using plant species-specific biomass and arthropod abundance data from experimental grassland plots (Jena Experiment), we constructed multitrophic functional group interaction webs to compare communities based on 4 and 16 plant species. 427 insect and spider species were classified into 13 functional groups. These functional groups represent the nodes of ecological networks. Direct and indirect interactions among them were assessed using partial Mantel tests. Interaction web complexity was quantified using three measures of network structure: connectance, interaction diversity and interaction strength. 3. Compared with high plant diversity plots, interaction webs based on low plant diversity plots showed reduced complexity in terms of total connectance, interaction diversity and mean interaction strength. Plant diversity effects obviously cascade up the food web and modify interactions across all trophic levels. The strongest effects occurred in interactions between adjacent trophic levels (i.e. predominantly trophic interactions), while significant interactions among plant and carnivore functional groups, as well as horizontal interactions (i.e. interactions between functional groups of the same trophic level), showed rather inconsistent responses and were generally rarer. 4. Reduced interaction diversity has the potential to decrease and destabilize ecosystem processes. Therefore, we conclude that the loss of basal producer species leads to more simple structured, less and more loosely connected species assemblages, which in turn are very likely to decrease ecosystem functioning, community robustness and tolerance to disturbance. Our results suggest that the functioning of the entire ecological community is critically linked to the diversity of its component plants species.

Changes to primary producer diversity can cascade up to consumers and affect ecosystem processes. Although the effect of producer diversity on higher trophic groups have been studied, these studies often quantify taxonomy-based measures of biodiversity, like species richness, which do not necessarily reflect the functioning of these communities. In this study, we assess how plant species richness affects the functional composition and diversity of higher trophic levels and discuss how this might affect ecosystem processes, such as herbivory, predation and decomposition. Based on six different consumer traits, we examined the functional composition of arthropod communities sampled in experimental plots that differed in plant species richness. The two components we focused on were functional variation in the consumer community structure (functional structure) and functional diversity, expressed as functional richness, evenness and divergence. We found a consistent positive effect of plant species richness on the functional richness of herbivores, carnivores, and omnivores, but not decomposers, and contrasting patterns for functional evenness and divergence. Increasing plant species richness shifted the omnivore community to more predatory and less mobile species, and the herbivore community to more specialized and smaller species. This was accompanied by a shift towards more species occurring in the vegetation than in the ground layer. Our study shows that plant species richness strongly affects the functional structure and diversity of aboveground arthropod communities. The observed shifts in body size (herbivores), specialization (herbivores), and feeding mode (omnivores) together with changes in the functional diversity may underlie previously observed increases in herbivory and predation in plant communities of higher diversity

Background Deep learning algorithms for automated plant identification need large quantities of precisely labelled images in order to produce reliable classification results. Here, we explore what kind of perspectives and their combinations contain more characteristic information and therefore allow for higher identification accuracy. Results We developed an image-capturing scheme to create observations of flowering plants. Each observation comprises five in-situ images of the same individual from predefined perspectives (entire plant, flower frontal- and lateral view, leaf top- and back side view). We collected a completely balanced dataset comprising 100 observations for each of 101 species with an emphasis on groups of conspecific and visually similar species including twelve Poaceae species. We used this dataset to train convolutional neural networks and determine the prediction accuracy for each single perspective and their combinations via score level fusion. Top-1 accuracies ranged between 77% (entire plant) and 97% (fusion of all perspectives) when averaged across species. Flower frontal view achieved the highest accuracy (88%). Fusing flower frontal, flower lateral and leaf top views yields the most reasonable compromise with respect to acquisition effort and accuracy (96%). The perspective achieving the highest accuracy was species dependent. Conclusions We argue that image databases of herbaceous plants would benefit from multi organ observations, comprising at least the front and lateral perspective of flowers and the leaf top view. Electronic supplementary material The online version of this article (10.1186/s13007-019-0462-4) contains supplementary material, which is available to authorized users.