Feature

Protecting Biodiversity

Geographic strategies, computers and satellite technology can help in prioritising biodiversity protection.

Associate Professor L. Edward Harvey

Biodiversity is inextricably bound to geography. Charles Darwin and Alfred Wallace collected evidence of this during nineteenth century explorations. These and other naturalists circumnavigated the globe observing latitude and altitude gradients of plant and animal species richness in many biogeographic regions, including New Zealand.

Biogeographic knowledge of plant and animal distributions eventually led Darwin to develop the theory of evolution. Subsequent research has shown that speciation, dispersal, climate, topography, primary productivity, competition and disturbance are responsible for many spatial patterns of biodiversity, but the decrease in species richness towards the poles remains largely unexplained.

While spatial patterns of species distribution and abundance contributed to founding theories in biogeography and ecology, the focus of modern biogeography encompasses more applied research. It relies less on extended field trips to unexplored regions to study natural processes that produce spatial patterns of biodiversity. Conversely, biogeography research now commonly examines effects of the human land-use mosaic on natural ecological processes and biodiversity.

The exploratory component remains, albeit in a different form. Explorations of the spatial distribution of biodiversity now occur in computer labs, using technological advances in satellite remote sensing, and geographic information system (GIS) and image processing software. These spatial information technologies permit relatively simple analyses of a species' range size or allow one to build sophisticated statistical models to map biodiversity from taxonomic information, vegetation cover, climate and topography.

Species distribution maps remain as primary data in biogeography. However, these maps are now commonly derived remotely from satellite imagery or predicted statistically from maps of vegetation and climate, and they are stored in digital atlases or biodiversity databases.

The biogeographic "macroscope" is regularly used to inventory native vegetation and to monitor the loss of animal habitat by removal (to farmland), fragmentation (remnants form habitat islands), and ecological degradation (from introduced species, industry or recreation).

Remotely sensed imagery has revealed that more of the native New Zealand vegetation has been converted to farmland than the world average (51% compared to 37%).

Estimates of vegetation biomass from satellite imagery have also been used as a surrogate for biodiversity and to assess ecosystem health. These will be important monitoring tools in the Environmental Performance Indicators Programme operated by the Ministry for the Environment.

Geographic Approaches

Perhaps the most pressing demand for a biogeographic macroscope is in conservation management, to systematically prioritise areas for biodiversity conservation. This area-based approach relies upon a digital database of plant and animal distributions. It uses computer-intensive techniques in GIS, and software for storing, analysing and visualising spatial data. Computers are exploited to search with brute force through the database to find an optimal combination of areas that best matches a conservation goal, such as maximizing the number of species protected in the minimum number of areas.

Biological conservation will always have limited funding and resources, so priorities are needed to efficiently allocate them but still satisfy conservation goals. Historically, these priorities have been defined for endangered species rather than areas or ecosystems, even though conservation goals are geographically limited to only part of the landscape because of competition with other land uses.

This highlights the importance of geographic conservation priorities. However, these priorities must complement conservation programmes for endangered species, biodiversity assessments by Landcare Research, and more general cost-utility evaluations of natural heritage conservation projects undertaken by the Department of Conservation.

There is a growing family of computational methods to prioritise areas for conservation. In their most basic form, they determine whether some areas are irreplaceable because they contain species or ecosystems found nowhere else. More sophisticated approaches are designed to iteratively search a digital geographic database for an optimal set of areas ranked according to their conservation priority.

The conservation goal will determine the most appropriate computer algorithm to prioritise the areas. The goal may be to represent all species in a near-minimum total number of areas (the minimum-area problem), or to represent as many species as possible in a fixed number of areas (the maximum-coverage problem).

Biodiversity Hotspots

One of the first methods to prioritise areas for biodiversity conservation allocated high priority to areas with a large number of species, referred to as biodiversity hotspots. If these areas are outside reserves, they have the highest conservation priority. The rationale is that these unprotected hotspot areas form biological "gaps" in the conservation estate. Several gap analyses based on this simple measure of species richness assist conservation management in North America.

A more recent geographic approach to prioritising areas for conservation considers the difference or variety of species in an area, not just the number of species. The rationale is that two areas may have the same number of species, but differ substantially in their diversity or composition of species. Several specialised GIS software packages use this approach to map biodiversity and prioritise areas for biodiversity protection.

One of the most comprehensive and widely used was written by Dr Paul Williams of London's Natural History Museum. His WORLDMAP software is applicable to any group of organisms and region, and designed to map several measures of biodiversity and interactively prioritise areas using several innovative strategies.

Bird Diversity

New Zealand supports about 28,800 native animal species, but only about 50% have been identified. Furthermore, few systematic surveys of geographic patterns of biodiversity have ever been undertaken.

The most geographically comprehensive field survey of birds in New Zealand was conducted by 800 volunteers in the Ornithological Society from 1969 to 1979. Maps of the presence or absence of almost 150 bird taxa (species and subspecies) in 10,000 yard grid square areas were published in The Atlas of Bird Distribution in New Zealand.

The range-size rarity of 97 native bird species is represented by the grid cells; the darker the cells, the more rare species within that area. The graphs illustrate the average number of bird species in each row (latitude) and column (longitude).

We can use WORLDMAP to map the range-size rarity of native land and freshwater bird species, selecting 97 species (excluding seabirds, migrants, human-introduced species and offshore species). This biogeographic version of rarity is calculated from each species' area of occupancy (i.e., the number of grid cells within the map area), not their local population size.

Species richness tends to decrease toward southern latitudes, and from west to east. Richness is also high in many coastal areas because these areas have habitat for both coastal and interior forest bird species. Hotspots of bird richness can be identified in Northland, south Westland, and northwest Nelson, and coincide with regions of high plant diversity.

In this simple example, the objective is to prioritise areas outside the existing conservation estate, defined as grid cells with a reserve (National Park, World Heritage Area, or Forest Park) which covers more than half the 10,000 x 10,000 yard cell area. There were 709 of these reserve grid cells. Eight native bird species were not observed in the reserve grid cells when the bird atlas was compiled. Which areas outside the reserves should have high conservation priority?

The simplest geographic strategy is to map richness hotspots, a concept originally used for global regions with exceptional species richness. This strategy is implemented by overlaying distribution maps for all 97 bird species and then ranking grid cells outside the reserve cells by their species richness.

Alternatively, we can rank the grid cells by their range-size rarity, and then define areas with the most rare species as rarity hotspots. Hotspots of species richness and range-size rarity do not necessarily coincide.

Computer Searches

The most computer-intensive algorithms step through simple rules to iteratively search for a set of areas that optimally satisfy a conservation goal. The greedy area set algorithm finds the smallest set of grid cell areas that contains all species, maximising the species richness complement at each step in the area selection.

In other words, first find the area with the highest species richness. Then find the area with the next highest richness after excluding species already occurring in the first area. Repeat this until all species are represented in the greedy area set.

When we include the existing 709 reserve grid cells, only four additional areas are required in the greedy area set to ensure all bird species are represented in at least one grid cell area. These are coastal areas in the Foxton, Mt. Allen, West Whanganui, and the Low Plains Ecological Districts.

The near-maximum-coverage algorithm is a highly effective method to prioritise areas outside an existing reserve network. It uses the species complementarity concept, species richness, and range-size rarity to find a set of areas that represents a near-maximum number of species.

The rules for the near-maximum-coverage algorithm are summarised below to illustrate the computational complexity of this strategy (the rarest species is the one with the fewest grid-cell records):

1. select all areas with species that have single records
2. repeat rules a to e until all species are represented:
   a) select grid cells with the greatest complementary richness in just the rarest species. If there are ties, then,
   b) select areas among ties with the greatest complementary richness in the next rarest species. If there are persistent ties, then,
   c) select areas among ties with the greatest complementary richness in the next-next-rarest species and so on. If there are persistent ties, then,
   d) select areas among ties with the greatest complementary richness in the next-next-next-rarest species and so on.
   If persistent ties remain, or no next-r next-next-r next-next-next-rarest species, then,
   e) select areas among persistent ties with the lowest grid-cell number (this is to ensure repeatability in tests),
3. reject any areas that in hindsight are redundant, or unnecessary to represent all species
4. repeat steps 1-3 to represent every species at least once, twice and so on, until the required number of areas is reached
5. re-order areas by complementary richness
6. choose the first n areas from the re-ordered list of areas

This gives us a set of identified cells which meet the near-maximum coverage rules. If I arbitrarily constrain the algorithm to select 20 areas, and attempt to ensure that all species are represented in at least four areas, we end up with a listing of grid cells identifying the highest priority areas. Requiring species to be represented in at least four areas does introduce some redundancy, but it also ensures that all species are represented in more than one prioritised area.

Working through the algorithm, we find that we can encompass all 97 bird species by adding four grid cell areas to the 709 reserve grid cells: Mt Allen, Foxton, Low Plains and West Whanganui.

The top six of a near-maximum-coverage set of 20 grid cells selected to be outside the conservation estate and constrained to ensure that all species occur in at least four grid cells.
Each grid cell adds unrepresented species to the cumulative coverage of the 89 species already found within the 709 reserve grid cells (91.75% of the total 97 native species).

Native bird species in the reserve network

The maximum species richness of 50 species occurs in a coastal Punakaiki Ecological District cell. The six numbered cells outside the reserve network are the highest priority areas in a near-maximum-coverage set.

Priorities

Approaches to setting conservation priorities have expanded from crisis responses to save individual endangered species to a concern for the whole of biological diversity. It is inevitable that the conservation of all species will not be possible, so it is important to prioritise areas for conservation on the basis of spatial patterns of biodiversity by exploiting geographic information technologies.

Software such as WORLDMAP can guide conservation priorities with a species representation goal and a network of conservation management areas capable of satisfying that goal. However, the challenge for maximising biodiversity protection will be to incorporate these geographic priorities into land use planning and management.

For example, relationships between land-use and bird species composition can be examined for prioritised areas using the NZ Land Resource Inventory. Looking closely at the Foxton Ecological District (previously identified as a high priority area), we can see that the vegetation cover is predominantly pasture. Nonetheless, the relatively small areas of dune vegetation, coastal forest, and wetlands are important habitat for 50 native bird species.

The use of GIS and priority-areas analysis is growing rapidly in the conservation management of many countries. In New Zealand, priorities are needed to efficiently allocate limited resources but still satisfy conservation goals. Competition with other land uses limits these goals to only part of the landscape, so it will be necessary to identify areas with high conservation priority.

Areas with high biodiversity that fall outside the reserve network are obvious geographic candidates. Likewise, areas with unique species composition within reserves deserve high priority. A systematic geographic approach using a macroscopic toolbox of computing tools for spatial analysis of biodiversity will provide guidance when prioritising conservation resources.

The availability of national digital databases of land use, vegetation cover, and species distributions, and geographic information system software provide an opportunity to develop geographic conservation priorities in New Zealand.

Dr Harvey has published maps, statistical analyses, and colour graphics on spatial patterns of NZ bird diversity at http://www.rem.sfu.ca/gis/Projects/Eh/Nzbirds/

WORLDMAP software and information on priority areas analysis are available from the Biogeography and Conservation Lab (Natural History Museum, London) at http://www.nhm.ac.uk/science/projects/worldmap

Landcare Research (http://www.landcare.cri.nz) and the Department of Conservation (http://www.doc.govt.nz) maintain the most extensive online information about spatial patterns of New Zealand biodiversity.

Associate Professor L. Edward Harvey is in the School of Resource and Environmental Management at Simon Fraser University in Canada.