Connectivity Measurement in

NRS 534: Ecology of Fragmental Yuyu Zhou Spring 2005 Overview Connectivity refers to the degree to which a landscape facilitates or impedes ecological flows such as the movement of among patches and therefore the rate of movement among local populations in a metapopulation (McGarigal et al. v3). Although connectivity is important in , there is no precise definition and it is difficult to quantify and imply because of the difference that it is processed as an independent or dependent variable, and structural or functional variable. Goodwin gave a good review about the connectivity research. According to this paper, studies of landscape connectivity are divided between those that seek to understand how landscape structure and movement behavior interact to dictate connectivity and those that seek to evaluate the impact of connectivity on other ecological processes (Goodwin 2003). With the limitation of the knowledge and method, most present researches process connectivity as independent variable. Therefore, connectivity is mainly discussed as an independent variable in this paper. Furthermore, according to another standard, connectivity measurements can be divided into direct and indirect ones. Direct measurements must incorporate a measure of some aspect of movement through the landscape. For the direct method, connectivity can be measured as the probability of movement between two source patches, dispersal success and search time. Landscape indices that characterize the spatial pattern of a landscape are common indirect method. In the following parts, the data, methods (independent or dependent) and tools for the connectivity measurement will be discussed.

According to landscape ecology, it is founded on the notion that environmental patterns influence ecological processes. Landscape pattern analysis included 4 basic types of spatial data: spatial point patterns, linear network patterns, surface patterns, and categorical map patterns (McGarigal et al.). Connectivity is often measured as a landscape pattern metrics, which refers exclusively to indices developed for categorical maps. Categorical map patterns represent data in which the system property of interest is represented as a mosaic of discrete patches. Land cover/use data from remote sensing and GIS are important source in the connectivity measurement. Information extraction from remote sensing data is a key issue in environmental monitoring and many landscape ecological applications. From traditional pixel-based to new object- based classification methods, remote sensing can provide multi-scale land cover for the landscape ecological applications. Connectivity can also be measured using other landscape patterns. Goossens et al analyzed the network connectivity using the SPOT data (Goossens et al. 1991). Graph theory was also used to research the landscape connectivity (Cantwell et al. 1993, Bunn et al. 2000). The data used in other patterns can also be derived from the categorical map.

As discussed above, connectivity was mainly measured as an independent variable. As an independent variable, connectivity can be measured as structural or functional. There are several researches that treated it as a structural variable. Yue et al used remote sensing derived classification to calculate the patch connectivity index (Yue et al. 2004). Goossens used edge-enhanced SPOT images for connectivity analysis (Goossens et al. 1991). Cantwell et al used graph theory to analyze the connectivity for several common elements such as fields, roads and so on (Cantwell et al. 1993). As a functional variable, the connectivity measurements depend on the research problems. Bunn et al used graph-theoretic approach to compare the connectivity for two focal (Bunn et al. 2000). Goodwin et al measured the connectivity in six different ways based on movements of individual (Goodwin et al. 2002a, 2002b). D’Eon et al presents a unique method for measuring connectivity between patches as a function of organism vagility (D’Eon et al, 2002). Van Langevelde used different threshold distances to calculate the degree of habitat connectivity (Van Langevelde 2000).

We discussed the connectivity measurements with the objective to evaluate its impact on other ecological process above. There are also connectivity measurements to seek to understand how landscape structure and movement behavior interact. Modeling methods are used to make such researches. There are patch model, grid based model, vector based model for connectivity research. Schippers et al. used a GIS-based model to estimate connectivity (Schippers et al. 1996).

FRAGSTATS is powerful tool in the calculation of connectivity metrics, and it provides many metrics about the connectivity. In FRAGSTATS, connectance index that is mainly used as a functional metrics and patch cohesion index that is mainly used as physical connectedness of the patches are two important connectivity metrics (McGarigal et al. v3). There are also some other related metrics such as isolation, and they can also be used as an accessorial metrics for the measurement of connectivity.

Connectivity as an important element of landscape structure is difficult to define precisely and quantified and applied in practice. Much work needs to do on the research of connectivity. Landscape connectivity must be accessed at the scale of the interaction between an organism and the landscape. Before the measurements were carried out, the research problem should be analyzed carefully because it determined what connectivity measurement should be performed and what data and scale should be chosen. Studies using connectivity as independent variable and dependent variable should be combined to make the multiple measurements of connectivity. Furthermore, the limitation of connectivity measurements in research should also be considered.

Annotated Bibliography

1. Bunn, A. G., D. L. Urban, and T. Keitt. 2000. Landscape connectivity: A conservation application of graph theory. Journal of Environmental Management. 59(4): 265-278. This paper used graph-theoretic approach to compare the connectivity for two focal species. It indicated using edge-thresholding operation in this method that mink perceived this landscape as connected, while prothonotory warblers did not according to their different dispersal capabilities. The node-removal operation in this method allowed ecologist to determine patch function in reference to both species. The graph can provide a powerful visualization of connectivity with dispersal estimates for the species in the landscape management.

2. Cantwell, M. D., and Forman, R.T.T. 1993. Landscape graphs: Ecological modeling with graph theory to detect configurations common to diverse landscapes. Landscape Ecology. 8(4): 239-255. This paper used landscape graphs approach to identify common configurations within landscapes, to examine the connectivity of elements in landscapes and to illustrate potential links with other modeling approaches. Many of the identified patterns using graph theory were useful for ecological understanding as well as landscape planning and management. Matrices could be constructed from landscape graphs to more fully understand the connectivity. It could be concluded in this paper that the graph construction method was a way to describe landscape structure, and to model landscape element size, connectivity and direction of flow.

3. D'Eon, R. G., Glenn, S. M., Parfitt, I., and Fortin, M.-J. 2002. Landscape connectivity as a function of scale and organism vagility in a real forested landscape. Conservation Ecology. 6(2): 10. This paper used the slope of the linear regression model between the critical distance and the logarithmic transformation of the corresponding Distance to edge as a measure of overall landscape connectivity for a given patch type and landscape. It indicated that the distance to edge and amount of area within clusters were most in the connectivity measurements. This approach was used to assess connectivity between harvest, old- growth, and recent wildfire patches in a real forested landscape. It was found that harvest patches had the largest connectivity and wildfire patches had the smallest connectivity among three types of patches.

4. Goodwin, B.J. and Fahrig, L. 2002a. Effect of landscape structure on the movement behavior of a specialized goldenrod beetle, Trirhabda borealis. Canadian Journal of Zoology. 80(1):24-35. This paper reported the influence of different patch and edge types on the local movement behavior of adult goldenrod beetles using a series of field experiments. They hardly move in a goldenrod patch, and sustain relatively slow, directed movements in a cut patch, and have many brief bursts of fast meandering movements in a netting patch. It indicated that the dispersal ability for beetles depend on landscape structure. Connectivity, which accurately reflects the impact of landscape structure on movement behavior, would be better than dispersal ability in the application of landscape ecology. However, the dispersal ability would be useful in the measurement of connectivity.

5. Goodwin, B.J. and Fahrig, L. 2002b. How does landscape structure influence landscape connectivity? Oikos. 99(3):552-570. This paper constructed simulation and experimental landscapes using the information of movement behavior of beetle in three kinds of patches to access the different aspects of landscape structure on landscape connectivity. Goodwin et al used six methods to measure the connectivity: transition probabilities between habitat patches, transition probabilities between habitat cells, mean number of habitat patches visited per individual, mean number of habitat cells visited per individual, habitat patch immigration, and habitat cell immigration. It indicated from the paper that the landscape connectivity was poorly defined and different results could be acquired for the same landscape using the different methods. It also concluded that connectivity decreased with the increase of inter-patch distance, and that habitat elements had more influence than matrix elements given that matrix patches differed only in their influence on movement behavior. This paper also suggested that cell immigration was a good measure of landscape connectivity.

6. Goodwin, B.J. 2003. Is landscape connectivity a dependent or independent variable? Landscape ecology. 18: 687-699. Goodwin gave a good review of the research in landscape connectivity from 1985 to 2000. He concluded that connectivity could be researched as an independent variable or dependent variable. To understand how landscape structure and movement behavior interact, connectivity should be treated as a dependent variable, while to evaluate the impact of connectivity on other ecological processes, it should be treated as an independent variable. Most of previous research treated connectivity as independent variable. Goodwin suggested it should be treated as a dependent variable based on the model methods combined with previous methods to make the multiple measurements in the future research.

7. Goossens, R., D'Haluin, E., Larnoe, G. 1991. Satellite image interpretation (SPOT) for the survey of the ecological infrastructure in a small scale landscape (Kempenland, Belgium). Landscape ecology. 5(3): 175-182. This paper used SPOT imagery based on edge enhancement techniques to map the ecological infrastructure for the further research by ecologist. A network connectivity analysis was performed on one test-site, and the result using general filter was similar with that from the topographical map. This paper indicated the advantage of satellite images over maps was the timely and repetitive acquisition for determination landscape changes and trends.

8. McGarigal, K., et al. FRAGSTATS User Guidelines. Version 3. FRAGSTATS is a spatial pattern analysis program for categorical maps. It can compute three groups of metrics: patch level, class level, and landscape level. And it can use many data source as input. This user guideline not only gives the algorithms of all metrics, but also explains the background of these metrics. Furthermore, this user guideline discusses many concepts in the landscape ecology. It is a useful tool in the landscape pattern research.

9. Schippers, P., Verboom, J., Knaapen, J. P. and van Apeldoorn, R. C. 1996. Dispersal and habitat connectivity in complex heterogeneous landscapes: an analysis with a GIS- based random walk model. Ecography. 19:97-106. This paper derived and evaluated a new model GRIDWALK, which estimated the connectivity of the landscape between existing (occupied) or potential (extinct or otherwise unoccupied) populations using the badger population as an example. This model could give insight into the dispersal pattern and revealed the connectivity network. The model could also be applied to a wide range of animal species whose dispersal pattern was determined by landscape characteristics.

10.Van Langevelde, F. 2000. Scale of habitat connectivity and colonization in fragmented nuthatch populations. Ecography. 23: 614-622. This paper used different threshold distances to calculate the degree of habitat connectivity based on graph theory to investigate effects of habitat connectivity measured at different spatial scales. When an effect of the degree of habitat connectivity on colonization by nuthatches could be demonstrated, the spatial scale at which colonization could be best explained by the degree of connectivity of the patches was identified. It also suggested in this paper that the probability that organisms move between patches may be used to replace the distance to provide a refined measure of the degree of connectivity of individual patches.

11. Yue, T., Xu, and B., Liu, J. 2004. A patch connectivity index and its change in relation to new wetland at the Yellow River Delta. Int. J. Remote Sensing. 25(21): 4617- 4628. This paper introduced a patch connectivity index and analyzed the relationship between patch connectivity, ecotope diversity and human impact intensity. The patch connectivity has negative correlation with ecotope diversity and human impact intensity. However, it was confused that patch connectivity has negative correlation with ecotope diversity. For this paper, the conclusion should be paid more attention because the correlation coefficient was obtained only from three years.