Nearly all of what was historically tallgrass prairie has been lost due to conversion to agriculture and increasing urbanization. This thesis focuses on strategies to restore native species diversity to working (i.e., agricultural) landscapes in the American Midwest. The practicality and efficacy of restoration of prairie function to working lands has been demonstrated for some taxa in the short-term (i.e., years 1-5), but the effects of continuing consistent management in the medium- and long-term are less studied. I focus here on butterflies and floral resources due to (1) the well-known sensitivity of butterflies to habitat changes (2) the likely concordant response between insect pollinators and floral resources due to their tight ecological relationship and (3) the ecosystem services provided by pollinators and the need to manage lands for such a service. Taken together, this body of work seeks to integrate empirical and modeling approaches to create a more holistic understanding of how grassland restoration strategies affect insect pollinators.

Management strategies are frequently recommended on the basis of only a few years of field sampling, after which, research often ceases and is replaced by opportunistic sampling by managers without a formal experimental framework. The first chapter of this thesis examines the effect of a continuous decade of consistent management with pyric-herbivory (i.e., various combinations of fire and grazing) on butterfly and floral resource plant communities. The management methods being considered here are burn-only (i.e., no grazing with the whole site burned once every three years) graze and burn (i.e., cattle grazing the entire site with a full-site burn once every three years), and patch-burn grazing (i.e., cattle grazing the entire site with a burn on one third of the site every year). We find that not only do butterfly and floral resource communities vary in terms of abundance, species richness, and diversity among management strategies; they do so in a way that is not consistent with what is typically observed in studies conducted over a shorter time period (as will be more fully discussed in that chapter). Additionally, while the community composition of butterflies varies significantly among management types in three of the five study years, floral resource community composition does not vary significantly in any study year. This may be indicative of an effect of management on larval (i.e., caterpillar) resources that is not as impactful on the nectar resources upon which adult butterflies depend. The community composition variation in butterflies appears to be primarily driven by the hyper-abundance of particular families of butterflies in each of the three management types. These results demonstrate that short-term responses may not match long-term responses and thus indicate that field studies should take place–where feasible–over more sampling seasons such that management recommendations are more fully informed.

When restoring prairie function to working landscapes, one must plan for the effects of anthropogenic climate change. In the second chapter of this thesis, I focus on the potential changes in bioclimatic suitability to plant species included in a restoration seed-mix used to jump-start the reintroduction of a particular community of native plants. Using Species Distribution Modeling (SDM), I correlate occurrence records of a particular species with a suite of climatic variables to predict where suitable bioclimatic conditions might be in the future. Given that SDMs can function well on publicly available data and are relatively intuitive in terms of how they infer changes to predicted distribution, they can be a powerful tool. However, the sheer number of distribution maps generated (one per species per climatic scenario) can be unwieldy in the context of restoration of entire plant communities. We sought therefore to assess whether modeled species exhibited conserved responses to climate change within functional groups. Our results indicate that for forbs, warm-season grasses, and legumes, species within a functional group tend to respond similarly to one another. Cool-season grasses, on the other hand, did not vary in a synchronous fashion, which may be indicative of more variable suitable conditions among species within this group. In addition, some functional groups tended to be much more sensitive (i.e., they demonstrated substantial changes between currently suitable areas and those predicted to be suitable in the future) to climate change conditions than others. Given that forbs tended to exhibit the most dramatic response to climate change, and other groups tended to exhibit more similar current and future distributions, managers interested in restoring flowering plant communities should consider either more southerly ecotypes or congeners to species of interest that are more tolerant of warmer and relatively drier conditions. One important caveat to the use of SDMs in restoration, however, is that such models are best applied at the continental scale (due to problematic assumptions of the role of microhabitat at small spatial scales), whereas land managers are likely interested in the regional or local spatial scale. More spatially precise estimates of the impact of climate change should consider microhabitat, with the SDM outputs presented here and elsewhere informing the areas of interest.

Restoration of grassland function must occur in the immediate future if threatened prairie taxa are to be conserved. Such conservation may most profitably focus on increasing the ecological value of agricultural land because such lands occupy a majority of much of the American Midwest. The first chapter of this thesis indicates that the traditional length of field studies may be insufficient in capturing the full consequences of management for butterflies and floral resource. Additionally, single metrics for community response, though informative, are unlikely to be enough in quantifying the full scope of ecologically-meaningful community response and multivariate community composition methods must be included. The second chapter of this thesis suggests that plant species within a functional group generally demonstrate conserved responses to climate change and that functional groups respond differently to potential future conditions. This also serves as evidence that modeling approaches are a useful complement to field methods and can provide an added dimension in creating restoration plans effective in the short-and long-terms.

Collectively, the insights presented in this work demonstrate the need for restoration strategy to include many different approaches and emphasize ways in which management can be more likely to be successful in the long term.