Posted by: oikosasa | April 1, 2014

Metapopulation modeling of endangered rabbits

Can hard-to-detect individuals of an endangered and declining population allow for testing of some of the major tenets of metapopulation theory while contributing to conservation efforts? A new multi-season occupancy model combined with observations on the Lower Keys marsh rabbit may have done just that. Read the Early View paper “Testing metapopulation concepts: effects of patch characteristics and neighborhood occupancy on the dynamics of an endangered lagomorph” by Mitchell J. Eaton and co-workers. below is their summary of the paper:

The Lower Keys marsh rabbit (LKMR, Sylvilagus palustris hefneri) is an endemic species found only on a handful of islands in the lower Florida Key islands. Following decades of decline caused by habitat fragmentation and degradation, sea-level rise and high rates of mortality inflicted by non-native predators, the LKMR was listed as endangered under the U.S. Endangered Species Act in 1990. Although several of these contributors to population decline have been improved, the distribution of the marsh rabbit continues to fall. With limited dispersal abilities, this secretive species persists in isolated habitat patches found within a highly fragmented landscape, challenging managers to identify viable solutions for their conservation and recovery. Given these conditions, a better understanding of the spatial aspects of marsh rabbit population dynamics could provide important insights and contribute to management efforts. We believed that a metapopulation framework would be the most useful for describing these dynamics. We were concerned, however, that existing metapopulation models were more theoretical than practical, based on too many assumptions and insufficient for dealing with the realities of a rare and hard-to-detect species. Therefore, we developed a new, flexible multi-season occupancy model that could test the concepts and assumptions of metapopulation theory, while investigating the dynamics of this species for the ultimate purpose of making recommendations for species management and recovery.

 Since its introduction in the 1960s, and following years of model development and application to natural systems, the theoretical underpinnings of metapopulation ecology have been strongly tied to assumptions about the relationship between neighboring patches, non-habitat (‘matrix’) characteristics and the probabilities of focal-patch extinction and colonization. Much recent work in metapopulation ecology has focused on developing advanced models that allow for incorporation of more detailed biological information to explain patterns in patch dynamics. Many of these models, however, have not taken into account the realities of imperfect detection in field sampling. As a result, only incomplete information on the abundance or occupancy levels of the surrounding landscape is available when making inference on the dynamics of a focal patch. As such, metapopulation models often rely on neighboring patch characteristics (e.g., perceived quality or size) as a proxy for the existence or abundance of colonizers and assume that local colonization will increase with patch connectivity. Local extinction probability has traditionally been modeled as a function of patch size, but is also predicted to be influenced by connectivity with neighboring habitat via a ‘rescue effect’. This latter process similarly depends on assumptions about the relationship between measurable patch characteristics and neighborhood occupancy, which can be difficult to quantify without consideration of the possibilities of non-detection.

 Building on recent advancements of the use of so-called ‘autologistic’ covariate models, we have developed a new multi-season occupancy model to explicitly incorporate estimates of neighborhood occupancy when modeling the dynamics of a metapopulation. Rather than treating the status or condition of a neighboring patch as certain (i.e., as a traditional, known covariate) we consider the occupancy of neighboring patches as an unobservable variable to be estimated. Our flexible model specification allows the ‘neighborhood’ to be defined in any number of ways, permitting nearly any a priori biological hypothesis to be tested. The model allows inclusion of a gradation of neighboring patch influence on the focal patch (e.g., habitat quality, distance, etc.) using patch-specific weights, as well as the quantification of non-habitat (e.g., water bodies) within the neighborhood. Using this modeling approach, we recast many of the assumptions of metapopulation theory as hypotheses to be tested explicitly.

 Bild1

Our results supported two of the major assumptions of metapopulation theory, namely that colonization probability is positively related to the occupancy of neighboring patches and that extinction probability is negatively related to local patch size. We also found support for the rescue effect, with extinction being mitigated by higher neighborhood occupancy, and for higher colonization rates being related to larger patch size. Model selection suggested that LKMR focal patch dynamics were influenced by a neighborhood effect size of approximately 1000m. Model results also suggest that patches in coastal areas (believed to be of higher quality for the species) experienced higher turnover rates than inland patches and that disturbance from sea-level rise, storm frequency and vegetation dynamics may be further destabilizing coastal patches. We found that lower-quality inland patches, which appear to be experiencing slower species turnover dynamics, may serve as refugia and provide an important source for colonization of coastal patches following local extinctions. Our findings can help managers better understand optimal spatial habitat configuration when planning restoration activities, predicted impacts of patch-specific removal of non-native predators and where translocations of LKMR would be most effective.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: