STRUCTURAL VERSUS FUNCTIONAL HABITAT CONNECTIVITY MEASURES TO EXPLAIN BIRD DIVERSITY IN FRAGMENTED ORCHARDS
ABSTRACT Habitat connectivity plays a paramount role in the biodiversity of fragmented landscapes. Commonly, connectivity is measured using simple structural metrics, e.g. Euclidean distances between habitat patches. Recently, functional measures such as cost-distance metrics have been proposed. Cost-distance metrics account for behavioural aspects of investigated organisms. They weight the habitats of the investigated landscape according to specific cost values, and model the optimal dispersal corridor according to these values. This study investigated i) if structural or functional connectivity measures explain biodiversity in a focal habitat better and ii) if the appropriateness of the measure differs between patch and landscape scale. We mapped the landscapes within a 500 m radius around 30 fragmented traditional orchards (focal patch). Connectivity measures were based on either Euclidean distances (structural) or cost-distances (functional) to other suitable habitat patches. Birds were used as biodiversity indicators. For analysis, we calculated species richness and total abundance of all species with a preference for woody habitats. In addition, abundances of four wood-preferring bird species were also examined individually. Linear models were created using stepwise forward selection. The relative performance of structural and functional connectivity measures was scale dependent. Structural metrics explained more variance at the patch scale whereas functional metrics explained more variance at the landscape scale. We conclude that simple structural metrics can be used to investigate local or small-scale effects on bird diversity but that investigations of landscape scale connectivity should consider behavioural aspects by using more complex functional metrics. The comparison between group and single species showed that not all individual species behave similarly to group results. Whilst the use of organism groups must be treated with caution, it is certainly worthy of future study.
SourceAvailable from: Angela Lausch[Show abstract] [Hide abstract]
ABSTRACT: For quantifying and modelling of landscape patterns, the patch matrix model (PMM) and the gradient model (GM) are fundamental concepts of landscape ecology. While the PMM model has been the backbone for our advances in landscape ecology, it may also hamper truly universal insights into process–pattern relationships. The PMM describes landscape structures as a mosaic of discretely delineated homogenous areas. This requires simplifications and assumptions which may even result in errors which propagate through subsequent analyses and may reduce our ability to understand effects of landscape structure on ecological processes. Alternative approaches to represent landscape structure should therefore be evaluated. The GM represents continuous surface characteristics without arbitrary vegetation or land-use classification and therefore does not require delineation of discrete areas with sharp boundaries. The GM therefore lends itself to be a more realistic representation of a particular surface characteristic. In the paper PMM and GM are compared regarding their prospects and limitations. Suggestions are made regarding the potential use and implementation of both approaches for process–pattern analysis. The ecological and anthropogenic process itself and its characteristics under investigation is decisive for: (i) the selection of discrete and/or continuous indicators, (ii) the type of the quantitative pattern analysis approach to be used (PMM/GM) and (iii) the data and the scale required in the analysis. Process characteristics and their effects on pattern characteristics in space and time are decisive for the applicability of the PMM or of the GM approach. A low hemeroby (high naturalness and low human pressure on landscapes) allows for high internal-heterogeneity in space and over time within patterns. Such landscapes can be captured with the GM approach. A high hemeroby reduces heterogeneity in space and time within patterns. For such landscapes we recommend the PMM model.Ecological Modelling 01/2015; 295(1):31–41. DOI:10.1016/j.ecolmodel.2014.08.018 · 2.33 Impact Factor
[Show abstract] [Hide abstract]
ABSTRACT: Population genetics theory predicts loss in genetic variability because of drift and inbreeding in isolated plant populations; however, it has been argued that long-distance pollination and seed dispersal may be able to maintain gene flow, even in highly fragmented landscapes. We tested how historical effective population size, historical migration and contemporary landscape structure, such as forest cover, patch isolation and matrix resistance, affect genetic variability and differentiation of seedlings in a tropical palm (Euterpe edulis) in a human-modified rainforest. We sampled 16 sites within five landscapes in the Brazilian Atlantic forest and assessed genetic variability and differentiation using eight microsatellite loci. Using a model selection approach, none of the covariates explained the variation observed in inbreeding coefficients among populations. The variation in genetic diversity among sites was best explained by historical effective population size. Allelic richness was best explained by historical effective population size and matrix resistance, whereas genetic differentiation was explained by matrix resistance. Coalescence analysis revealed high historical migration between sites within landscapes and constant historical population sizes, showing that the genetic differentiation is most likely due to recent changes caused by habitat loss and fragmentation. Overall, recent landscape changes have a greater influence on among-population genetic variation than historical gene flow process. As immediate restoration actions in landscapes with low forest amount, the development of more permeable matrices to allow the movement of pollinators and seed dispersers may be an effective strategy to maintain microevolutionary processes.Heredity advance online publication, 15 April 2015; doi:10.1038/hdy.2015.30.Heredity 04/2015; DOI:10.1038/hdy.2015.30 · 3.80 Impact Factor