«109 Sustainable Development and Planning IV, Vol. 1 Prioritizing habitat patches for conservation in fragmented landscapes/townscapes using ...»
Sustainable Development and Planning IV, Vol. 1
Prioritizing habitat patches for conservation in
fragmented landscapes/townscapes using
network-based models and analyses
Stockholm Resilience Centre & Dept. of Systems Ecology,
Stockholm University, Sweden
To preserve biodiversity in highly fragmented landscapes, it is crucial to
conserve and possibly improve the connectivity among remaining habitat
patches. However, multiple users are competing for a limited amount of land, and conservation efforts subsequently need to be efficiently directed to maximize biodiversity given limited resources. In terms of connectivity, this could be expressed as to which habitat patches do we really need to preserve, and which patches could we lose without facing any significant negative effects on connectivity? Network-based models of fragmented landscapes provide for comprehensive visualizations and analyses of landscape connectivity that could help in prioritizing habitat patches for conservation. This is especially valuable in a planning context where many different types of agents are typically involved, thus emphasizing the importance of being able to present key ecological implications of different spatially explicit habitat configurations in a easily understandable way.
Here, three different aspects of landscape connectivity are presented, all suggested as being particularly suitable for network-based modeling approaches.
These are: (i) estimating to what degree the landscapes spatial configuration enables re-colonization following local extinctions; (ii) identifying clusters of patches that together form sufficient habitat; and (iii) identifying key-stone patches that are crucial in providing connectivity.
Keywords: habitat patches, land use planning, network models, landscape fragmentation, graphs.
WIT Transactions on Ecology and the Environment, Vol 120, © 2009 WIT Press www.witpress.com, ISSN 1743-3541 (on-line) doi:10.2495/SDP090111 110 Sustainable Development and Planning IV, Vol. 1 1 Introduction Biodiversity and species persistence is highly dependent on access to high quality habitats. Societal and economic development, on the other hand, often leads to the reduction of both the amount and the quality of natural habitats. The amount of natural habitat decreases as more land is converted to, for example, agricultural production or housing development. Also, the habitat quality can be reduced when, for example, a previously relatively undisturbed forest fragment is converted to modern monoculture forestry, or to a recreational park. In addition, the reduction of natural habitat increases the level of fragmentation among remnant habitat patches. A patch represents a spatially contiguous area of land with a biophysical composition which potentially suits as adequate habitat for a focal species. An increased level of fragmentation implies that the connectivity among remnant habitat patches is decreased. Connectivity is, on a general level, the degree to which the spatial pattern of habitat patches in the landscape facilitates or impedes the movement of organisms (Taylor et al ).
Decreased possibilities for species movements can, for example, result in local species extinctions (Levin ) and thus in biodiversity reductions. Hence, it is not only the biophysical composition (i.e. the quality) and the geographical area of a given habitat patch as such that define its ecological value, but its level of connectivity in relation to other habitat patches in a landscape is also important.
An important arena for decision-making about various land uses that affects the amount, quality and connectivity of species habitat are the planning processes at the municipality level. Here, as well as in many other contexts where different usages of land are decided, multiple users with different interests, knowledge and perceptions are competing for a limited amount of land.
In such settings, conservation efforts subsequently need to be efficiently directed to maximize biodiversity given limited resources. Since land uses are decided in a competitive context where different interests are to be weighted against each other in the planning process, tools that provide the various decision-makers and stakeholders with adequate assessment of the ecological importance of different habitat patches are crucial. This kind of tool would help in making informed decisions on which habitat patches to preserve in maximizing various ecological values, given that there are other competing land uses, and which patches could be converted to other purposes without any severe ecological consequences.
Furthermore, it is not only the sheer existence of such tools that is important.
Firstly, the actors that, to some degree, represent the interest of preserving/creating habitat would benefit from being able to use the tools themselves without the need for constant support from external experts. This is in part a result of the dynamic nature of the planning process itself where different scenarios of land uses are constantly being presented and contested. In such situations, it is highly beneficial if the conservation agents have access to tools that they can use themselves to evaluate different land use scenarios more or less in real time instead of having to solely rely on assessments conducted by external experts on a consultation basis. Therefore, ease of use is of central value for any such tool. Secondly, it is important that these tools provide results that WIT Transactions on Ecology and the Environment, Vol 120, © 2009 WIT Press www.witpress.com, ISSN 1743-3541 (on-line) Sustainable Development and Planning IV, Vol. 1 are fairly easily comprehensible by non-experts and lay men. If the results from ecological assessment are hard to understand and interpret, it might be hard to gain acceptance and support for different habitat conservation schemes among other actors involved in the planning process.
A fairly recent innovation in studying various aspects of landscape connectivity resulting from habitat fragmentation is the network-centric modeling approach (Keitt et al , Cantwell et al , Urban and Keitt ). This modeling approach (which is often called graph-theoretical) has also been specifically suggested as very useful in prioritizing and ranking the importance of different habitat patches from a connectivity perspective e.g. Pascual-Hortal and Saura . Furthermore, the modeling approach as such does not rely on large quantities of detailed ecological data; rather it can quickly provide a coarsegrain analysis of an entire landscape using fairly low amounts of data. The network modeling approach thus provides for a favorable trade-off between how well the model portrays reality and the data it requires to do so (Calabrese and Fagan ). Hence, this modeling approach fulfils the different aspects suggested as important for an effective patch prioritizing tool for use in planning as presented here, and is therefore a promising candidate for further development and possibly also for deployment in real-world planning situations.
In this work we will further examine the potential of the network modeling approach by specifically studying how it can be used to study three important and different, although not mutually exclusive, aspects of landscapes connectivity. These aspects are: (1) large-scale connectivity which is important, for example, in metapopulation dynamics; (2) local connectivity, which is important in order to provide sufficient home ranges in highly fragmented areas;
and (3) identification of critical patches that much more than others provide for critical connections and dispersal pathways in the landscape (e.g. stepping stones). The study is based on a review of a selected sample of the emerging literature applying network based modeling in studying landscape ecology and/or metapopulation dynamics.
2 The network model of landscapes The network modeling approach rests on the basic abstraction of a landscape as consisting of scattered patches of habitat that are laid out in a landscape matrix which, to a varying degree, is inhospitable for the focal species, and that the connectivity between any pair of habitat patches depends on the effective distance separating them. If the effective distance is below the maximum dispersal distance of the focal species, that particular pair of patches is considered as connected in terms of possibilities for direct dispersal. In network terminology, habitat patches are the nodes in the network, and all pairs of patches that are separated by an effective distance below the maximum dispersal distance of the focal species are connected by links (Fig. 1).
The effective distance is dependent on the geographical distance separating the patches and the permeability of the different types of land making up the matrix between the different patches. If the matrix is very hard to cross, either WIT Transactions on Ecology and the Environment, Vol 120, © 2009 WIT Press www.witpress.com, ISSN 1743-3541 (on-line) 112 Sustainable Development and Planning IV, Vol. 1 from an energetic and/or risk of mortality due to the predation perspective, the permeability is considered low, and two patches geographically close to each other would be modeled as separated by a longer effective distance. In a GIS, cost-distance analyses can be applied to calculate the effective distances separating patches e.g. Bunn et al , Verbeylen et al .
Figure 1: The habitat patches A-D are located in the landscape, and R is the maximum dispersal distance of the focal species. All patches separated with a distance less than R are considered as connected (right side).
Furthermore, a central assumption is that a species can disperse along paths consisting of several links. Thus, species are assumed to be able to move between patches that are not directly, but indirectly connected through network paths (e.g. from B to C via A in Fig. 1). Hence, a species would be able to disperse throughout large areas of the landscape by moving from patch to patch (assuming that such paths exist). In other words, the network model merges species dispersal processes with spatial patterns of habitat patches in one single model.
The network model of the fragmented landscape provides, however, only for the basic data structure which can be further analyzed using various methods and metrics from research fields such as social network analysis and/or graph theory (Bodin and Norberg ). This fairly simple but still relevant abstraction of the complex patterns of connectivity among habitat patches makes connectivity analyses of fragmented landscapes easier to conduct, and many more or less standardized and easily assessable tools and methods are available off-the-shelf (see e.g. Bodin and Norberg ).
3 Three ecological aspects related to landscape fragmentation
3.1 Overall connectivity and species persistence Several decades ago, Levin  introduced the concept of metapopulation. A metapopulation is a set of spatially separated local populations which may, individually, undergo local extinctions, but where the aggregated population is maintained by re-colonization events resulting from dispersing organisms.
WIT Transactions on Ecology and the Environment, Vol 120, © 2009 WIT Press www.witpress.com, ISSN 1743-3541 (on-line) Sustainable Development and Planning IV, Vol. 1 Hence, upholding a high level of landscape connectivity is central in order to provide for enough dispersion to prevent overall population declines induced by the occurrence of local extinctions. Building on earlier work, Hanski and Ovaskainen  introduced a spatially explicit metapopulation model that incorporates the flux of dispersing organisms between individual patches at the landscape level. Their model uses spatially explicit information from a real landscape to access the capacity of a specific metapopulation to actually persist in that particular landscape. This research shows, for example, that metapopulation dynamics is very important to take into account when designing natural reserves. A key design issue is therefore to make sure that the level of connectivity among reserves is high enough given species dispersal capabilities (e.g. Moilanen and Cabeza ). If reserves are too geographically separated, local extinctions are to be expected (Cabeza and Moilanen ).
For this kind of study and assessment, network-based models of landscape have great potential (alongside other modeling approaches). Here, the possible existence of network components is of key interest. A network component is an isolated subnet confined within a larger network. By definition, no links exist between nodes of different components. Thus, if a network model describing a particular landscape consists of two or more network components, a dispersing species cannot traverse the entire landscape in moving from patch to patch (Keitt et al , Urban and Keitt ). The ecological interpretation of a network consisting of several network components is that species can only disperse to patches belonging to the same component, and are thereby isolated from other network components in the landscape. Thus, a network component would correspond to a single metapopulation which in turn may consist of one or several individual local populations confined to patches within that particular network component. If a network component is small, the entire metapopulation would be small only consisting of a very limited number and/or sizes of local populations and, therefore, far less persistent than a larger metapopulation. From a conservation perspective, it would consequently be important to aim for a spatial configuration of the habitat patches in the landscape that would minimize the number of network components. This is in line with findings derived using other modeling approaches which have led to the suggestion that natural reserves should be clustered together spatially (Cabeza et al ).