The overall aim of this study was to deepen the knowledge of multi-habitat networks. A broad array of basic aspects of the studied plant-pollinator network at Skalasand on the island of Fårö, Gotland was covered and compared witha similar study at the island of Gotska Sandön. The species composition differs considerably between the two networks. The Skalasand network has a richer faunaof insects belonging to the groups Syrphidaeand Apiformes, while Muscidae-flies constitute a considerable part of the Gotska Sandön network. However, the networks also have resembling modules. The modules are always established around one or a few plant species actingas hubs while animals visiting several habitats work as connectors linking the different modules. The Syrphidae and Apiformes are the two taxonomic groups that are the most generalistic when it comes to interacting in different habitats.

There may be several mechanisms that drive observed interactions between plants and pollinators in an ecosystem, many of which may involve trait matching or trait complementarity. Hence a model of insect species activity on plant species should be represented as a mixture of these linkage rules. Unfortunately, ecologists do not always know how many, or even which, traits are the main contributors to the observed interactions. This thesis proposes the Latent Dirichlet Allocation (LDA) model from artificial intelligence for modelling the observed interactions in an ecosystem as a finite mixture of (latent) interaction groups in which plant and pollinator pairs that share common linkage rules are placed in the same interaction group. Several model selection criteria are explored for estimating how many interaction groups best describe the observed interactions. This thesis also introduces a new model selection score called ``penalized perplexity". The performance of the model selection criteria, and of LDA in general, are evaluated through a comprehensive simulation study that consider networks of various size along with varying levels of nesting and numbers of interaction groups. Results of the simulation study suggest that LDA works well on networks with mild-to-no nesting, but loses accuracy with increased nestedness. Further, the penalized perplexity tended to outperform the other model selection criteria in identifying the correct number of interaction groups used to simulate the data. Finally, LDA was demonstrated on a real network, the results of which provided insights into the functional roles of pollinator species in the study region.

Traditionally managed temperate semi-natural meadows are generally considered as one of the most diverse habitats worldwide. During the past decades, management changes and abandonment of meadows pose serious threats to communities of both flowering plants and pollinators. However, the extent to which are pollinators affected by changes in floral resources still remain unknown. To investigate the effect of floral resources on visitation specialization, species richness and abundance of hoverflies, pollination networks from 13 meadows in the Železné hory PLA were constructed and analysed. Moreover, patterns were recorded on two spatial scales, in transects and in the whole meadows. To specify the specialization of hoverfly pollinators, unvisited plant species and all resources present on a site (transect/meadow), offered by plants as floral abundance and nectar rewards, was added in the networks. Relationships between plant species richness, floral abundance and nectar sugar production were positively correlated with visitation specialization of hoverflies on the meadow scale. Species richness and abundance of hoverflies was affected by floral resources on the transect scale, unlike the specialization of hoverflies. The usage of extended network indices may act as an improvement towards the better...

Plant-pollinator interaction assembly can be represented as so called pollination network. Using these networks, assembly of plant-pollinator communities and geographic patterns in pollination networks can be studied. During last years, a great progress in describing patterns in structure of plant-pollinator interaction webs and in describing their spatial variability was made. However, these results are mostly based on data from previous studies, which are not seamlessly comparable due to their nonuniformity. My thesis aims at the methodological concepts of study on pollination networks. Based on meta- analysis of previously published data and on data from four years detailed study of plant-pollination system from one locality, I show several methodological problems caused by dissimilarities in sam- pling method of current pollination-network data. These dissimilarities could be responsible for some observed patterns in the assembly of pollination networks. Thus, I propose a uniform method for observing pollination networks suitable for macroecolo- gical meta-analyses (pollination network samples). I tested this method in the field and I described 49 pollination networks from the central and eastern Europe by using such method. In these ne- tworks, no nonrandom structure character such as...