With continued degradation of ecosystems, we need to know how to restore biodiversity --- both for conservation and to ensure the provision of essential services provided by nature. To manage and restore diversity in human-modified systems, however, we need to understand the mechanisms that originally maintained biodiversity. A fundamental and widely supported theory of biodiversity is the idea that diversity begets biodiversity (i.e., environmental heterogeneity, disturbance, biodiversity itself). These processes contribute to turnover of species through space and time and subsequent heterogeneity of community composition ($\beta$-diversity) --- primary determinants of the total species richness supported by a landscape. Communities are being homogenized as human actions such as habitat conversion, land management practices and invasive species disrupt the processes maintaining diversity.

In this dissertation, I examine the assembly of plant-pollinator communities in a variety of landscapes through time and space to better understand how environmental, disturbance and interaction diversity sustains biodiversity. I focus on mutualistic communities because they are influential biological interactions for the generation and maintenance of biodiversity. Plant-pollinator mutualisms are also particularly important for service provision. Pollination systems, however, are under increasing anthropogenic threats. Understanding how to maintain plant-pollinator community biodiversity is this both timely and imperative.

I first investigate the capacity of environmental, disturbance and interaction diversity to sustain biodiversity in a system in Yosemite National Park where nature still drove these processes. In frequent fire forests in Yosemite National Park, California, I found that fire diversity is important for the maintenance of flowering plant and pollinator diversity, and shifts towards lower diversity fire regimes will negatively influence the long-term species richness of these communities. Changing climate and fire suppression are eroding fire diversity and thus homogenizing communities, and thus we must explore management practices that can maintain fire diversity. In these systems, fire diversity is promoted directly through prescribed fires with varied burn conditions and allowing wildfires to burn. These management strategies are already recommended, and my results affirm that their usage should continue and expand.

In Yosemite, I was able to examine the mechanisms sustaining diversity in a natural system and make recommendations for maintaining those processes. When a landscape is already degraded, however, we must determine what restoration efforts are able to reassemble functional communities of interacting organisms. This is often the case in agricultural landscapes where widespread conversion of natural ecosystems to agriculture, combined with intensification of farming practices, has led to the homogenization of biological communities. In Chapter 2, I use a long-term pollinator survey data from the intensively managed agricultural landscape of the Central Valley of California to show that on-farm habitat restoration in the form of native plant ``hedgerows,'' when replicated across a landscape, can re-establish community spatial turnover. I also determined that the mechanism promoting community spatial heterogeneity was the successional dynamics of hedgerow communities promoted the assembly of phenotypically diverse communities, leading to the accumulation of differences in community composition between sites over time. This work elucidates the drivers of spatial and temporal diversity while also validating the role of small-scale restorations such as floral-enhancements for conserving biodiversity and promoting ecosystem services in agricultural areas.

To fully understand the mechanisms maintaining communities we must also combine our understanding of the ecological processes enabling their persistence with the evolutionary processes that assembled those communities. Coevolution is a key process producing and maintaining complex networks of interacting species. In Chapter 3, I use a theoretical approach to determine whether the structure of interactions varied depending on the community's evolutionary history. I found that coevolution leaves a weak signal on interaction patterns. Our results suggest that determining whether assembly processes structure interactions within a community requires a synthetic approach, combining data about the biogeographic history of the interacting lineages and their evolution.