Home | Contact Us | CI Sites | Search | Site Map | Log In

Assessing the Human Dimensions of the Priority Gaps

Analyzing the human dimensions of the priority gap cells provides an understanding of the opportunities and challenges for conservation in those locations. Our analysis of the human dimensions focused on three variables: human population, land cover, and agricultural suitability. With regard to the feasibility of establishing new protected areas in the priority gaps, human population data show the number of people who might be affected, either positively or negatively, by introducing a reserve, as well as the level of pressure from human population on a new park. Land cover data provide insights on human activities and habitat conditions within each priority gap cell. Finally, data on agricultural suitability enable an assessment of the costs of making land in the priority gaps off limits to agriculture (Howard 1996, Ando et al. 1998, Pimm et al. 2001, Simpson 2004).

To evaluate the human dimensions of priority gap cells we used geographic information system technology to overlay maps of the three key variables listed above on the map of priority cells and then measured the human dimensions within each cell. The simpler analysis involves placing a map of one variable, say population, directly on top of the map of global priority gap cells and calculating a particular characteristic of the cells, such as average population density (Figure 2a). The more complex analysis involves examining combinations of variables. For example, combining maps of land cover and population density enables us to identify areas with certain desirable types of habitat that also feature population density below a certain level (Figure 2b). We used simple overlay analyses to examine human population, global land cover, and agricultural suitability (for maximized subsistence and maximized commercial scenarios) individually. We used combined overlay analyses to consider multiple human dimension variables, in various combinations, simultaneously.

Figure 2a: Method used to overlay one variable on priority gap cells

Figure 2b: Method used to overlay multiple variables on priority gap cells

Datasets Used in the Human Dimensions Analysis

In this section we review the three datasets used to assess the human dimensions of the global priority gap cells: 2002 human population, 2000 land cover, and agricultural suitability. Each dataset had the dual requirements of global coverage, to enable an evaluation of all priority gap cells, and geographic referencing (location in geographic space), to enable the systematic overlay analysis described immediately above. The datasets used were the best available at the time of analysis, providing the greatest spatial detail (highest resolution) for the most recent year possible. In addition to describing the datasets, we also discuss the limitations of each— characteristics that constrain what we can say about human dimensions in the priority gap locations.

Human population data

The human population dataset used in this analysis was developed by researchers at Oak Ridge National Laboratory (2003). Named Landscan 2002, this dataser represents a high-resolution geographically referenced estimate of global population in 2002. With a resolution of 30 arc-seconds, Landscan 2002 provides population estimates in each cell of a grid that covers the entire land surface of the earth. Cells are about 1 km² at the equator and become smaller towards the poles. This important feature of the Landscan 2002 dataset enables the estimation of population in each gap cell.

Landscan 2002 is based on census data compiled at the level of small geographic units within individual countries. These data were projected to 2002 based on demographic characteristics for each location. The greatest challenge in producing Landscan’s high-resolution, geographically referenced population estimates was allocating population estimates for particular geographic units to a global grid of 1-km² cells. To perform this allocation, Landscan researchers calculated the probable distribution of human population throughout the cells based on road locations, slope variations, land cover differences, settlement locations, and nighttime lights (Oak Ridge National Laboratory 2003). The result was a detailed estimate of the geographic distribution of human population throughout the entire world in 2002 (Figure 3).

Figure 3: Global distribution of human population, 2002 (Source data: Oak Ridge National Laboratory2003)

Global land cover data

The global land cover data used in our study was Global Land Cover (GLC) 2000, the main product of a project sponsored by the European Commission to provide critical information on the global environment at the onset of the new millennium (Fritz et al. 2002). GLC 2000’s interpretations of global satellite imagery were derived from a coordinated series of regional studies. More than 30 teams of regional specialists participated in the project, contributing to 1 of 19 areas on which they had expertise. The result was a global database that incorporates considerable local knowledge and insight. To ensure uniformity, all regional teams relied on satellite imagery from the SPOT-4 Vegetation Vega2000 dataset, and all contributors used the Food and Agriculture Organization’s Land Cover Classification System (DiGregorio & Jansen 2000). To fill in missing data, areas north of 75 degrees were mapped using Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR) data. GLC 2000 data consisted primarily of two sets of satellite imagery, each a 10-day composite of images obtained between April and October 2000. The final GLC 2000 database depicts a 30 arc-second grid of the earth’s land cover, with cells about 1 km² at the equator and slightly smaller towards the poles. Each cell is classified as 1 of 22 land cover types (Figure 4).

Agricultural suitability data

Finally, our study uses data from the recently released Global Agro-Ecological Zone (GAEZ) assessment of global agricultural suitability (Fischer et al. 2002). The GAEZ assessment examined a range of geographically referenced data for the key agricultural variables of soil, terrain, and climate. Using these data, GAEZ researchers assembled the first main component in their assessment of agricultural suitability, a land resources database for the entire world. The second main component in the assessment was a land-use database containing information on types of agricultural production systems. These systems were classified by crop-specific environmental requirements together with input and management criteria. Specifically, the land-use database consisted of 154 crop, fodder, and pasture land-use types defined for high (commercial), intermediate (commercial-subsistence), and low (subsistence) input-management or cropping scenarios. Then, by matching land-use type requirements with land resources, GAEZ researchers generated a global database of agronomically attainable yields for various crops and cropping scenarios in each cell of a global grid. The geographically referenced grid has a resolution of 5 arc-minutes — cells are about 9 km to a side at the equator and become smaller towards the poles. The database estimates each cell’s suitability for various crops or crop categories under low-, intermediate-, and high-input cropping scenarios (Figure 5).

Figure 4: Global distribution of land cover, 2000. (Source data: Joint Research Centre, European Commission 2003)

Due to the global focus of our study, we considered several crops and crop categories from the GAEZ assessment results (defined in Fischer et al. 2002), as follows:

• Cereals (barley, maize, millet, rice, rye, sorghum, and wheat)

• Cotton (used as a surrogate for fiber crops)

• Pulses (chickpea, cowpea, phaseolus bean, and soybean)

• Roots/tubers (cassava, sweet potato, and white potato)

• Oil crops (groundnut, oil palm, olive, rape, and sunflower)

• Sugar crops (sugar beet and sugar cane)

Figure 5: Global distribution of commercial agricultural suitability for cereal production. (Source data: Fischer et al. 2002)

We evaluated two cropping scenarios: (1) high input, which is generally synonymous with commercial production and consists of fully mechanized production with low labor intensity, improved yield crop varieties, and optimal applications of fertilizers and chemical disease, pest, and weed control; and (2) intermediate input, a best-case subsistence scenario assuming medium-intensive labor, improved crop varieties, and some application of fertilizer and chemical pesticides.

Using both the commercial and best-case subsistence scenarios, we calculated the maximum potential for agricultural productivity of each 9 x 9 km cell before incorporating the GAEZ results into our analysis. In the commercial scenario, for instance, if cereals were shown to be most suitable for one area and pulses for another, we based our maximum suitability calculations on cereal suitability estimates in the former area and pulse suitability estimates in the latter. The estimates we used for agricultural suitability are thus conservatively high, assuming that for each GAEZ grid cell people will know the best crop to grow and be capable of growing it.

Data limitations

Although the global datasets described above provide remarkable insights on the human dimensions of the priority gap cells, they do have certain limitations. The Landscan 2002 population dataset was based on demographic projections of earlier data to 2002 and allocations of sub-national population totals to 1-km² cells. However, the demographic projections for 2002 employed base data that often were inaccurate or inadequate, such as censuses that include major undercounts or are more than a decade old. Similarly, the data used for population allocations, such as information about road network configurations, were dated and vary widely in accuracy from one location to another.

In the case of the GLC 2000 dataset, the data suffer from a common problem associated with landcover information based on remotely sensed imagery, namely, that certain cell interpretations may be inaccurate. In addition, certain land-type distinctions significant to conservation are not captured in the dataset. For instance, GLC 2000 does not distinguish between primary forest, secondary forest, and tree crops. Similarly, it does not distinguish between intact grassland and grassland subjected to grazing. These limitations highlight the need for detailed follow-up analyses of local conditions within any given gap cell for actual conservation planning.

As with the Landscan 2002 dataset, the GAEZ assessment is based on data whose accuracy varies considerably for different parts of the world. Additionally, the assessment did not consider land degradation, an issue with growing implications for crop production. Although the GAEZ assessment included a broad range of crops and crop categories, it omitted certain perennial crops (e.g., grapes) and tree crops (e.g., coffee) because of complications associated with evaluating their productivity. Some of these crops are extremely important in many of the priority gap cells.

Figure 6a: Distribution of human population (2002) in one priority gap cell. (Source data: Oak Ridge National Laboratory 2003).

Figure 6b: Distribution of conservation-compatible land cover (2000) in one priority gap cell. (Source data: Joint Research Centre, European Comission 2003).

Figure 6c: Distribution of best-case subsistence agriculture suitability in one priority gap cell. (Source data: Fischer et al. 2002).

Finally, all three datasets suffer from the problem of coarse resolution, most notably the 5 arc-minute resolution of the GAEZ assessment. The resolutions of each are adequate for analyses of global issues, but limit the accuracy (and utility) of local evaluations.

Although the shortcomings of the datasets used in this study do not undermine our analysis, they do help define the limitations of our conclusions. We emphasize that our study is not intended to identify specific locations for new protected areas or provide recommendations for reserve design or details on local land cover and human settlement. Rather, we examined the human dimensions of priority gap cells in order to understand better the overall likelihood of achieving long-term conservation in these areas. It is absolutely essential that any specific actions to establish new reserves employ more accurate, high-resolution data for particular locations and should consult appropriate stakeholders (Eken et al. 2004). Moreover, expanded biodiversity protection should be part of nationally planned reserve systems that meet the conservation goals of host countries while contributing to global outcomes (Rodrigues et. al. 2003, 2004b).

Overview of Simple and Combined Overlays

We conducted two different types of analyses: simple and combined. In our simple overlay analyses, we considered each of the individual human dimension variables separately, namely, human population estimates for 2002, global land cover for 2000, and agricultural suitability for commercial and best-case subsistence scenarios. The analysis of human population involved two calculations of population density measured as persons/km²: (1) entire gap cells and (2) area of largest contiguous sparse population density (fewer than 10 persons/km²) per gap cell. Two measures contributed to the global land cover analysis: (1) total area of habitat compatible with conservation per gap cell and (2) area of largest contiguous habitat compatible with conservation (all compatible habitat types) per gap cell. Habitat compatible with conservation included all GLC 2000 land cover types except the following: cultivated and managed areas; mosaic (cropland/tree cover/ other natural vegetation, and cropland/shrub and/or grass cover); artificial surfaces and associated areas (including urban and other built-up areas); and water bodies. We also calculated mean agricultural suitability for each global priority gap cell under the commercial and best-case subsistence scenarios.

As Figure 6 shows, a resolution of 1 km² in the first two overlays—population density and land cover—provides a detailed view of internal gap cell composition (6a and 6b). In contrast, the 81-km² resolution of the agricultural suitability overlay (6c) shows much less detail about agricultural suitability in individual priority gap cells. However, because most of the gap cells are much larger than 81 km², that dataset still conveys a sense of the spatial variation in crop production capability.

With combined overlays, we simultaneously examine multiple variables of the human dimensions of gap cells. For instance, combined overlays enable identification of areas containing conservation-compatible habitat as well as sparse population densities. This is valuable knowledge because areas with conservation-compatible habitat and low population density are good settings for parks and other forms of biodiversity-compatible management (though some areas meeting these conditions may already be protected in some fashion, perhaps as community-conserved or indigenous lands). We conducted three types of combined overlay analyses: (1) conservation-compatible habitat combined with population density of fewer than 10 persons/km² (Figure 7); (2) conservation-compatible habitat combined with agricultural suitability; and (3) conservation-compatible habitat combined with agricultural suitability and population density.

Figure 7: Distribution of conservation-compatible land cover and sparse (fewer than 10 persons/km²) human population density in one priority gap cell.

Human Dimensions of the Priority Gap Cells

Using overlays of population density data, land cover data, and agricultural suitability data in various combinations, we analyzed the human dimensions of the priority gap cells. In the following sections discussing the results of our analysis, we focus initially on analyses of the three separate variables and then on analyses of the variables in combination with one another. Results show that the vast majority of global priority gap cells are well suited to host areas aimed at conserving biodiversity. Most gap cells had a human context with low human population density, large amounts of habitat that appears to be compatible with conservation, and low agricultural suitability. The study also found that conditions favorable for conservation in most priority gap cells held under combined analyses.

Analyses of separate variables

Population

Human population is generally considered to be at the core of the current decline of biodiversity. As various studies have established, the presence of humans usually impacts biodiversity adversely (Balmford et al. 2001, Harcourt et al. 2001, Gorenflo 2002, Parks & Harcourt 2002). But researchers have yet to define a threshold of population density beyond which human presence definitely begins to compromise biodiversity. Several important protected areas, for example, are located near cities. In contrast, areas with lower population densities can have adverse impacts on biodiversity, depending on the human activities involved. For example, researchers have shown that many extractive industries, such as hunting, can significantly impact biodiversity even in areas low in human population (Robinson & Bennett 2000). Similarly, commercial agriculture in the sparsely settled Cerrado of eastern Brazil has converted broad tracts of natural habitat and greatly reduced biodiversity (Gorenflo et al. 2006). But generally speaking, biodiversity tends to fare better in areas with low population densities.

We examined human presence in priority gap cells to determine the potential level of pressure from the surrounding population on new conservation areas. We also wanted to determine how many people might be affected, either positively or negatively, by new protected areas. We found that, based on average population densities in 2002, the overall population in priority gap cells does not tend to be sparse—with only about 16 percent of the cells registering overall densities of fewer than 10 persons/km² (Figure 8a). However, the vast majority of gap cells contained large tracts of land populated at a density of fewer than 10 persons/km² (Figure 8b). Large-scale human impacts are unlikely at density levels this sparse (see Gorenflo in press, Sanderson et al. 2002). Most cells appear to contain areas of low population density where new protected areas could potentially be designed to avoid existing residents and be managed primarily for biodiversity conservation (e.g., IUCN categories I and II). Places with higher densities of people will require less restrictive IUCN categories that rely on zoning to balance conservation and livelihood objectives. Other approaches, such as conservation concessions and private reserve areas, may also be appropriate in more densely populated settings (Brandon 2002, UNESCO/ MAB Secretariat 2002, Merkl et al. 2003).

Figure 8a: Average population density per gap cell.

Figure 8b: Areas with densities of fewer than 10 persons/km².

Land cover and use

To complement our analysis of population density, we analyzed the land cover that characterized each priority gap cell in 2000. Compared to human population, land cover data more directly indicate whether the conditions in priority gap cells are compatible with conservation — showing the actual impacts of many human activities. In particular, land cover data reveal the extent of habitat conversion, the leading threat to biodiversity, and thus are good indicators of conservation potential in priority gap cells.

Of the 22 possible types of land cover used in the GLC 2000 project, we considered four — those showing evidence of human modification or use, including agriculture and artificial surfaces — to be incompatible with conservation. This categorization yields conservative estimates of conservation-compatible habitat in each gap cell. Many species can tolerate the disturbed habitats that our evaluation excludes, particularly mosaics of natural habitat and agriculture (Sanderson et al. 2003). This conservative quality helps to compensate for overestimates of conservation-compatible habitat due to certain inadequacies in the GLC 2000 data, such as the lack of distinction between tree crops and natural forest.

Using the above assumptions about habitat compatibility for each cell, we examined the GLC 2000 data both for the total amount of land cover compatible with conservation and for contiguous tracts of conservation-compatible land cover. We found that most cells contain large amounts of conservation-compatible habitat — nearly 91 percent of the cells contained a total of more than 10,000 ha, with more than 59 percent containing a total of more than 100,000 ha (Figure 9a). However, substantial habitat alteration is evident in many gap cells despite the presence of large amounts of conservation-compatible habitat. In more than one-fifth of the cells, at least 70 percent of the habitat had been converted to some form of human use by 2000, while 36 percent of the cells had lost at least half of their natural habitat (Figure 9b).

Figure 9a: Total area of conservation-compatible habitat.

Figure 9b: Percentages of total area of gap cells compatible with conservation.

Figure 9c: Largest tract of contiguous conservation-compatible habitat per gap cell.

Examining the total conservation-compatible area in a priority gap cell provides a rough sense of the feasibility of establishing a new protected area there. But the total compatible area may consist of fragmented pieces interspersed with converted habitat, which will not support species that need large areas of uninterrupted habitat to survive. To address this need for connectivity, we calculated the size of the largest contiguous tract of conservation-compatible land in each gap cell. Results of these calculations indicate that more than 83 percent of the cells contain contiguous tracts of compatible land in excess of 10,000 ha, while nearly half (47.2%) contain contiguous tracts of compatible land larger than 100,000 ha (Figure 9c). Different species have vastly different habitat requirements (Sanderson et al. 2003), so there is no uniform standard for the ideal size of a reserve. However, research has shown that tracts of land spanning 10,000 ha or more tend to provide the conditions necessary for many species to survive (Terborgh & van Schaik 1997).

Agricultural suitability

We examined the agricultural suitability of each priority gap cell to evaluate the general impact of restricting agriculture in newly established protected areas. Agriculture is the leading cause of natural habitat conversion worldwide (Wood et al. 2000). With steady growth in population and food demand predicted to continue until the middle of this century (Tilman et al. 2001), the pressure to convert pristine land to agriculture will increase dramatically. Such pressure undoubtedlywill affect existing protected areas as well as any new reserves in priority gap cells or elsewhere. An analysis of agricultural suitability shows whether new reserves would incorporate land with high agriculture potential (Brandon 1995, Wood 1995).

To examine agricultural suitability in the gap cells, we analyzed best-case scenarios for both subsistence and commercial production. Our analysis of subsistence production scenarios indicates that the majority of priority gap cells are, at best, marginally suited for agriculture. The mean agricultural suitability of 52.3 percent of the cells falls into one of the three lowest suitability categories defined by the GAEZ (unsuitable, very marginal, and marginal) (Figure 10a). This means that productivity levels in these cells are 25 percent (the highest productivity possible in the marginal suitability category) or less than the maximum observed global levels for the crop categories concerned. In contrast, not a single cell featured a mean agricultural suitability in the most productive category (very high), and the mean suitability of only 2.7 percent of the cells reached the second most productive category (high). These findings suggest that most of the priority gap cells would have low yields if used for subsistence agriculture. In areas with such limited agricultural potential, support to local residents through strategic investments in conservation-compatible activities, incentives for local management, and paying for ecosystem services are likely to provide better and more sustainable options than subsistence agriculture (Kremen et al. 2000, Ferraro 2002, Sanderson et al. 2003, Eken et al. 2004).

Under a commercial agriculture scenario, the mean agricultural suitability of gap cells increases only slightly over its subsistence counterpart. Our analysis of commercial production indicated that the mean agricultural suitability of nearly 47 percent of the priority gap cells falls in one of the three lowest categories (Figure 10b). However, three gap cells, or 0.2 percent, registered very high mean agricultural suitability, while 7.6 percent registered high mean suitability. Pending more detailed analyses based on local data, including production costs, market access, and crop prices, expanding the global network of reserves in areas with high agricultural potential likely would pose real challenges to conservation planning and implementation.

Figure 10a: Mean agricultural suitability per priority gap cell, subsistence agriculture scenario.

Figure 10b: Mean agricultural suitability per priority gap cell, commercial agriculture scenario.

Our assessment of agricultural suitability indicates that most priority gap cells are not well suited for crop production, even for growing the highest yield crops. This is not to say that the gap cells will not be under pressure in the future — in many of the tropical countries where gap cells are located, high poverty rates and an increasing demand for food will generate pressure for agricultural conversion even on unproductive land. Our analysis simply indicates that the agricultural yields in the gap cells would be limited, and thus the lost agricultural yields from establishing protected areas in most of the cells would be low relative to other areas more suited to agriculture.

Combined analyses

Although evaluating the three variables of population density, land cover, and agricultural suitability individually is valuable, analyzing them in combination with one another provides greater insights on overall feasibility of conservation in these areas. We analyzed these variables in three different combinations: (1) human population and conservation-compatible habitat, (2) agricultural suitability of conservation-compatible habitat, under both agricultural production scenarios, and (3) human population, conservation-compatible habitat, and agricultural suitability, again under both agricultural scenarios.

Combined analysis I: Human population and contiguous tracts of conservation-compatible habitat

Our combined analysis of population density and conservation-compatible habitat showed that the majority of priority gap cells feature large tracts of land that are simultaneously compatible with conservation and have low population densities, the latter defined as fewer than 10 persons/km² (Figure 11). Of the total number of priority gap cells, contiguous tracts of at least 10,000 ha with low population densities are present in 61.1 percent of the gap cells, while contiguous tracts of between 1,000 and 10,000 ha with low population density occur in another 25.4 percent of the cells.

Figure 11: Largest contiguous conservation-compatible tracts with population density of fewer than 10 persons/km².

Combined analysis II: Conservation-compatible habitat and agricultural suitability

A combined analysis of conservation-compatible habitat and agricultural suitability in the priority gaps allows us to estimate the cost of keeping such habitat out of agricultural production. In general, we found that contiguous tracts of conservation-compatible land in excess of 10,000 ha in priority gap cells have low agricultural suitability. For the subsistence agriculture scenario, mean agricultural suitability of only 2.2 percent of these large tracts of conservation-compatible land fall in the top two productive categories. In contrast, 56.5 percent of the large tracts registered mean suitability in the least productive three categories (Figure 12a). The results are similar under a commercial agriculture scenario, in which intensive agriculture is conducted so as to yield the highest possible crop production. Under this scenario, mean suitability for large contiguous tracts of conservation-compatible land was great for relatively few cells—0.2 percent registered very high and 5.7 percent registered high mean suitability. In contrast, mean agricultural suitability in more than half (52.7%) of the large tracts of conservation-compatible land was in the three lowest categories of productivity (Figure 12b).

Combined analysis III: Human population, conservation-compatible habitat, and agricultural suitability

Under the subsistence scenario of agricultural production, contiguous tracts of conservation-compatible land larger than 10,000 ha with sparse human settlement had low agricultural suitability (Figure 13a). Only 2.1 percent of the tracts with these characteristics had mean suitability measures in the top two suitability categories while nearly 65 percent had mean agricultural suitability in the three lowest categories. Under the commercial cropping scenario, less than 6 percent of gap cells with large contiguous conservation-compatible tracts and low population density had mean agricultural suitability measures in the top two suitability categories (Figure 13b). The mean agricultural suitability measures of nearly two-thirds (65.8%) of these gap cells fell within the three lowest suitability categories under the commercial scenario.

Geographical patterns in the human dimensions analysis

After identifying the gap cells where the human dimensions are most favorable to some type of biodiversity conservation, we analyzed their geographic distribution. In the global gap analysis, researchers found that priority gap cells are disproportionately located in the tropics, on islands, and in mountains (Rodrigues et al. 2004b). We looked for geographic patterns in the following three categories of gap cells: (1) those with large contiguous tracts (10,000 ha or greater) of conservation-compatible land, (2) those with large contiguous tracts of conservation-compatible land with low human population density (fewer than 10 persons/km²), and (3) those with large contiguous tracts of conservation-compatible land with low human population density and low agricultural suitability (mean agricultural suitability in the three lowest categories), under both subsistence and commercial cropping scenarios.

Figure 12a: Agricultural suitability of largest contiguous conservation-compatible tracts (>10,000 ha), subsistence scenario.

Figure 12b: Agricultural suitability of largest contiguous conservation-compatible tracts (>10,000 ha), commercial scenario.

Figure 13a: Agricultural suitability of largest contiguous conservation-compatible tracts (>10,000 ha), with fewer than 10 persons/km², subsistence scenario.

Figure 13b: Agricultural suitability of largest contiguous conservation-compatible tracts (>10,000 ha), with fewer than 10 persons/km², commercial scenario.

The largest contiguous tracts of conservation-compatible land are distributed much the same as the priority gap cells themselves—that is, they are concentrated in the tropics, (Figure 14a; see also Figure 1). This similarity is to be expected because more than 83 percent of the priority gap cells contain contiguous tracts of compatible habitat 10,000 ha or larger. The only notable difference is that large tracts of conservation-compatible habitat are not as prevalent in the northern and central Andes and in the central Philippines as are the priority gap cells in general.

With the additional constraint of low population density, the distribution of conservation-compatible land in the gap cells changes to exclude part or all of some islands, such as Hispaniola, Puerto Rico, Jamaica, Sri Lanka, and parts of Indonesia and the Philippines (Figure 14b). This is a consequence of islands in general having higher population densities than mainland areas due to less land area for human settlement. The constraint of low population density also removes some gap cells located on the coast, areas that are often densely populated.

Adding the final constraint of low agricultural suitability causes more island territory, selected portions of the Andes, central Mexico, central and southern Brazil, and west and east Africa to be excluded from the map (Figures 14c and 14d). Our analysis thus shows that the greatest opportunities to expand the current global network of biodiversity protection to fill priority conservation gaps occur on larger land masses away from the coasts, while the gaps that will be most challenging to fill occur on islands and in coastal locations.

Figure 14a: Priority conservation gap cells with large contiguous tracts of conservation-compatible land (in red).

Figure 14b: Priority conservation gap cells with large contiguous tracts of conservation-compatible land characterized by low population density (in red).

Figure 14c: Priority conservation gap cells with large contiguous tracts of conservation-compatible land characterized by low population density and low subsistence agriculture suitability (in red).

Figure 14d: Priority conservation gap cells with large contiguous tracts of conservation-compatible land characterized by low population density and low commercial agriculture suitability (in red).

< previous section  < index >  next section >