UC Berkeley

Metapopulations occupy spatially divided habitats and understanding how that fragmentation affects their survival, growth, dispersal, and persistence is critical to their conservation. Researchers in many sub-fields of ecology and evolutionary biology test hypotheses relating to metapopulation dynamics and landscape spatial structure. Key aspects of these hypotheses are sometimes (a) large numbers of subpopulations and dispersal corridors and (b) their positions relative to each other. Comparing such spatial hypotheses using traditional lab equipment and methods is impractical, unwieldy, expensive, or impossible.

I invented the Metapopulation Microcosm Plate (MMP) to overcome these drawbacks. This device resembles a 96-well microtiter plate; the 96 wells represent habitat patches and they are connected by dispersal corridors that can be modified in their spatial position to create various artificial landscapes, with hundreds of non-intersecting dispersal corridors of varying lengths. The device can be filled with nutrient broth and used to culture microbial metapopulations.

In Chapter One, I first demonstrate that bacterial travel time is significantly faster through MMP dispersal corridors that are shorter, but is unaffected by corridor vertical position within the plate. Thus, MMPs satisfy the necessary assumptions for use in metapopulation experiments. Furthermore, travel time by bacteria with fully functional flagella was significantly faster than that of bacteria with disabled flagella, indicating that the bacteria actively swim through the corridors, rather than traveling by simple diffusion. Thus, MMPs can test hypotheses that account for behavioral responses. MMPs can be used to test many spatial hypotheses that have previously been prohibitively difficult to test. Further, by incorporating individual behavioral responses to within-patch conditions, MMPs incorporate greater realism than do directed pipetting or other artificial dispersal methods.

In Chapter Two, I used MMPs to explore how recolonization and recovery after subpopulation extinction differs in metapopulations in which the dispersal corridors have different spatial arrangements. Some metapopulations have corridors spread relatively evenly through space in a homogeneous arrangement such that most subpopulations are connected to a few neighbors, while others have corridors clustered in a heterogeneous arrangement, creating a few highly connected subpopulations and leaving most subpopulations with only one or two neighbors. Graph theory and empirical data from other biological and non-biological networks suggest that heterogeneous metapopulations should be the most robust to subpopulation extinction. Here, I compared the recovery of metapopulations with homogeneous and heterogeneous corridor arrangements following small, medium, and large subpopulation extinction events. I found that while metapopulations with heterogeneous corridor arrangements had the fastest rates of recovery following extinction events of all sizes and had the shortest absolute time to recovery following medium-sized extinction events, metapopulations with homogeneous corridor arrangements had the shortest time to recovery following the smallest extinction events.

Finally, for Chapter Three I conducted an experiment to test whether metapopulations with heterogeneous corridor arrangements recover more slowly from extinctions targeted at high connectivity subpopulations than random extinctions in low connectivity subpopulations. Simulations of the World Wide Web and other heterogeneous networks have demonstrated that, while they are very robust to random loss of nodes, targeted attacks on highly connected nodes can lead to failure of the entire network. Based on these simulations, I predicted that metapopulations with heterogeneous corridors would recover fastest when extinctions occurred in low connectivity wells, regardless of extinction event size. Unlike in theoretical networks, however, the corridor arrangements of metapopulations cultured in MMPs cannot be completely homogeneous, because wells on the edge will be slightly less connected than those in the center. However, I predicted that a small deviation in connectivity would be unimportant and that recovery in metapopulations with homogeneous corridors would not be affected by whether extinctions were in low connectivity or high connectivity wells. Instead, I found that, at both low and medium levels of extinction targeted at highly connected subpopulations, both heterogeneous and homogeneous metapopulations recovered more quickly when those extinctions were targeted at high connectivity wells, but that when many subpopulations went extinct, all metapopulations recovered fastest when those extinctions were in low connectivity wells.

This work demonstrates that MMPs can be used to test the assumptions of metapopulation theory, especially those involving large numbers of subpopulations and dispersal corridors. I have shown that metapopulations with heterogeneous corridor arrangements have the fastest rates of recovery from subpopulation extinction, but that that faster rate only translates to a shorter absolute time of recovery after larger extinction events. Furthermore, when smaller numbers of subpopulations go extinct, metapopulations recover more quickly when those extinctions are targeted at high connectivity subpopulations, but when large numbers of subpopulations go extinct, recovery is faster when low connectivity subpopulations are targeted. This suggests that dispersal corridors that are clustered in space may help to alleviate the effects of habitat fragmentation in some circumstances, but exacerbate them in others.