Incorporating anthropogenic and land cover change into conservation planning

Incorporating dynamic scenarios of anthropogenic and natural land cover change processes into systematic conservation planning
David Troupin and Yohay carmel.

The loss of biodiversity is the subject of worldwide concern. Protected areas constitute a central tool in this effort to reverse this trend. Systematic conservation planning is a framework that aims to optimally locate, prioritize, and design conservation area networks, in which biodiversity is well-represented, protected, and sustainable. An often-noted shortcoming of most conservation planning studies is that they do not account for ecological and anthropogenic processes or future uncertainty. To date most studies still rely on static biodiversity patterns and ignore processes. The goal of my thesis was to address this gap and to develop an approach for selecting protected areas in a network that will improve biodiversity representation and persistence over time, by explicitly incorporating ecological and anthropogenic processes and future uncertainty. The case-study focused on Israel’s Mediterranean region. The study species were 87 breeding bird species. The classes of a land cover map produced by integrating data from several sources served as a proxy for the study species’ habitats. Three ornithologists ranked each species’ habitat associations to each land cover class.
An assessment of the existing protected areas revealed that they provided adequate coverage to only 23% of the study species, clearly demonstrating the need for action. I compared three strategies for expanding the protection of the study species based on their present-day distributions: (1) expanding the PA system by focusing on the protection of remaining natural habitats; (2) complementing the PA system by improving the conservation value of agricultural habitats; and (3) a combination of the two approaches. Of the three strategies, expanding the existing protected area system based on the combined strategy was the most beneficial since it provided greater coverage to the target species’ habitats, and resulted in a larger, more compact, and less patchy conservation area network.
Subsequently I modeled two of the study area’s major land cover change processes: urban development (resulting directly from human activities) and Mediterranean vegetation dynamics (an ecological process). I simulated the scenarios for these processes 60 years into the future using DINAMICA-EGO, a cellular-automata simulation model. I constructed alternative scenarios for each process based on scientific literature and expert opinion. For urban development the two scenarios of regulated and unregulated development differed in the rate of development and its spatial distribution. Compared to the regulated development scenario, the unregulated development scenario resulted in larger areas of built-up land particularly in the central region of the study area. For example, assuming low development levels, after 20 years the total amount of built-up land in the unregulated development scenario was expected to be 980 km2, while in the regulated development scenario the corresponding value was expected to be 906 km2. Over time and under higher development rates these differences became larger. The unregulated development scenario also resulted in a more fragmented pattern of built-up patches, with a larger number of relatively small and scattered patches. For vegetation dynamics the two scenarios were the continuation of observed past trends (corresponding to moderate climate change) and altered succession patterns and fire regime (corresponding to severe climate change). The severe climate change scenario resulted in a more even distribution of the different vegetation formations while in the moderate climate change scenario, the dense tree formation became the dominant formation (>50% of the natural vegetation formations) after 20 years. For each scenario I then mapped the distributions of available habitats for each study species. The results indicate that for the study species the combination of the unregulated development and the moderate climate change is the least favorable scenario as the loss of suitable available habitats was highest in this scenario. Following this, for each scenario I used the distributions of available habitat for each species as input and ran the conservation planning software MARXAN in order to identify the conservation priority areas for the given scenario.
There are several approaches for using information from scenarios in planning. To date, only few studies have compared the utility of such approaches in designing conservation area portfolios that are robust to uncertainty. I examined whether the use of scenarios improves the robustness of the selected conservation areas to future changes and uncertainty. I compared between three approaches: (1) considering a range of scenarios; (2) assuming the realization of specific scenarios, e.g., the ‘worst-case’ scenario; and (3) a reference strategy based only on current distributions.
The performance of the different conservation area networks was assessed based on the number of species for which representation targets were met in each scenario. The portfolio based on all scenarios consistently performed better than the other portfolios, and was more robust to errors, i.e., a larger number of species were protected when the assumed specific scenario did not occur. On average, using the portfolio based on all scenarios resulted in the obtainment of representation targets for five additional species compared with the portfolio based on the current distributions (approximately 33 species versus 28 species, respectively). The differences in the number of species for which representation targets were met between the portfolio based on all scenarios and the portfolios that were based on a specific scenario were smaller but consistent – the portfolio based on all scenarios was expected to achieve the representation target for 1-3 additional species.
These results highlight the importance of considering a meaningful and wide range of scenarios, rather than assuming the certainty of specific scenarios or that the ‘worst-case’ scenario approach will perform well under a wide range of scenarios.

View full thesis here