Habitat monitoring is an important tool for assessing the threat and conservation status of species and protected areas. This can be done at global and regional scales, where data are available. Conservation International (CI) uses habitat-based indicators, among other indicators, to assess the impacts of its conservation work. Donors for conservation projects are also requiring monitoring indicators to assess the impacts of their conservation investments. The UN Convention on Biological Diversity also has developed a list of indicators for monitoring to assess progress towards the 2010 target. Another major global assessment of species threat is the IUCN Red List. Threat levels estimated by the IUCN are dependent on criteria that include habitat extent, fragmentation and rate of change.
To View Data Access Click Here
The lists of indicators prioritized by these and other related consortia, such as the Conservation Measures Partnership and the Cambridge Group, all reinforce the key roles of habitat extent, fragmentation and rate of change in threat assessments. This is in part because of 1) the importance of habitats themselves, both as assemblages of species and as performers of ecological processes, 2) the relationships among a species’ habitat extent, resource availability and potential population, and 3) the relatively low cost and high feasibility of habitat monitoring with satellite data.
Forest monitoring is critically important to assessments of biodiversity and conservation status because of the high levels of diversity and endemism in forests. This is especially so in the tropics and non-tropical biodiversity hot spots. In these areas, rates of forest clearance and fragmentation are rapid. They have direct implications for the threat and conservation status of species and protected areas. Monitoring forest change is also important for global climate because of both deforestation emissions and altered land-atmosphere exchanges of energy, water and carbon.
CI and colleagues have begun monitoring forests although have the goal of monitoring other and more precise habitat types. This is partly possible globally for cover and fragmentation. However, changes in extent in most places must be monitored at higher spatial resolutions. The regional results presented here are part of an effort to provide a high-resolution baseline of cover, fragmentation and change for the ~1990 to ~2000 time period. This corresponds to the baseline for the indicators of CI and the CBD. They also complement the many high-quality in-country maps of forest and other habitat types, which are far more abundant than in-country products of habitat change. For selected areas the baselines have been updated to ~2005.
CI and partners are conducting a series of forest-monitoring studies to estimate changes over time. Because of the small scale of most changes, this must be done at a finer resolution than that of a global analysis of cover. We rely largely on the 30-m resolution data from Landsat, especially the global archives made available for free from NASA’s Geocover program. For all areas listed we have created maps of forest cover and change from ~1990 to ~2000; for some areas additional data from ~1986, ~1975 or earlier are included, and for selected areas the forest cover and change maps have been updated to ~2005.
All CI maps have been produced with a common methodology. Some improvements are anticipated; however there are some key aspects common to all. This is also true for those produced by partners and colleagues. CI’s methodology and a discussion of other methods used for the products listed here is provided below. In most cases the maps were produced via collaboration between CI or other US-based researchers and in-country partners.
Data and Methods
The forest cover and change monitoring approach involves two or three epochs of relatively high resolution satellite imagery, such as Landsat data. Forest cover and change are mapped by analyzing imagery from circa 1990, 2000, and 2005 and are mapped in a single process using imagery from two dates at once (1990-2000 or 2000-2005). Forest is defined as closed canopy, mature natural forest. If Landsat imagery are used, these data are generally acquired at no-cost; the Geocover c.2000 Landsat image usually provides the base image as this product has been ortho-rectified and has the highest locational fidelity. The two other epochs are registered to the Geocover and two classifications of forest cover and change for the two periods generated.
A supervised classification approach is employed. Changes are directly classified within multi-temporal images and numerous sub-classes created for each final class. Multiple iterations are run using maximum likelihood classification or See5 [through the Classification and Regression Tree (CART) classification interface] and sub-classes from the final iteration are merged. Analysts delineate training sites for each land cover or change class, based on visual interpretation and by referring to ground reference data and high-resolution imagery, such as Quickbird, available on Google Earth. For 3-date classifications, a matrix is generated to highlight class overlap, and recoding, by referencing the input images, is then performed where necessary to yield the final 3-date classification. Validation of each final product is performed using available aerial photographs and satellite imagery available through Google Earth. The resultant data represent a series of regional or national level studies. Each one provides a complete estimate of forest cover and change in both over time. The final maps are filtered so that patches and clearings as small as 2 hectares are reported. In some cases, sub-classes are included, such as Spiny Forest and Woodland and Mangrove in Madagascar. For each of the nine areas listed below, a zip file is provided containing a summary document, digital map, and graphics
CABS and partners have a goal to complete a precise baseline estimate of forest cover, fragmentation and change for the entire tropics, country-by country. These have been created with in-country partners, and links to all partners and sites with similar products produced by colleagues are provided below. A map of current or completed projects is provided below (global progress map)
These data can provide baseline rates of change for various assessments, including the Convention on Biological Diversity. They also form a basis for analysis of threats and conservation status of all protected areas, other priority areas, and forest-obligate species that have estimates of ranges.
Related Analysis of Threat and Conservation
For each of these maps of forest cover and change, standard analyses are conducted to summarize the threat and conservation status of individual species and protected areas. These include data from the IUCN World Database on Protected Areas and global assessments of the IUCN Species Specialist Groups, the Red List consortium and Birdlife International.
We believe it is possible and cost effective to conduct high-resolution monitoring of all tropical forests and other forested hot spots on a five-year basis. This however requires delivery of a low-cost, global ~2005 data set based on Landsat and other data, and an ensured continuation of such programs. For complete forest coverage, and for most other habitats, monitoring can now be conducted at 250m or coarser on a yearly basis. This would use newer versions of MODIS and SPOT VEGETATION data than those used in the global products analyzed here. Raw data are available, but coordination among implementing agencies and laboratories must be directed towards a new era of near-real time monitoring habitat change.
List of Projects and Partners
1. Brazilian Amazon (INPE, NASA LBA)
2. Brazilian coastal forest (S.O.S. Mata Atlantica, INPE, CI)
3. Burma (SI, CI)
4. CARPE landscapes (UMD, CI)
5. Central America Biodiversity Corridor (U.S. AID, NASA, SERVIR)
6. Kenya, Coastal Forests (Sokoine University of Agriculture, CI
7. Tanzania, Coastal Forests (Sokoine University of Agriculture, CI)
8. Tanzania, Eastern Arc mountains (Sokoine University of Agriculture, Forest and Beekeeping Division of the Ministry of Natural Resources and Tourism))
9. Liberia and Guinea (CI)
10. Madagascar (CI)
11. Meso America (ECOSUR, BTFS, WCS, CI)
12. Non-Brazilian Amazon and Andes (Bolivia, Peru, Ecuador, Colombia, Venezuela)
13. North-east Philippines (CI)
14. Papua New Guinea (CI)
15. Paraguayan forest and woodlands (UMD, Guyra Paraguay, CI)
16. Sichuan Alps (CI)
17. Sumatra (WRI, CI, UMD)
Forest cover and change data for each of the regions below can be accessed here. Associated data descriptions and publications are contained in each zip file on the Data Access page.
China c.1990-c.2000 (Sichuan Province)
Mexico c.1990-c.2000-c.2007 (5 southern states)
Philippines c.1990-c.2000 (selected corridors)
Sumatra, Indonesia c.1990-c.2000
Tanzania (Eastern Arc Mountains) c.1970-c.2000
Tanzania (Eastern Arc Mountains) c.1990-c.200
Tanzania (Coastal Forests) c.1990-c.2000