Context Matters: Considerations for Large-Scale Conservation

By Reed Noss

Illustration © Christophe Vorlet

I leaned against my dusty 1974 Opel in the parking lot of Sugar-creek Nature Reserve in southwestern Ohio, reviewing the day’s count. The 10 most common bird species — common grackle, northern cardinal, carolina chickadee, red-winged blackbird, American goldfinch, blue-gray gnatcatcher, European starling, indigo bunting, field sparrow, brown-headed cowbird — contrasted sharply with some of the least common: ovenbird, black-and-white warbler, cerulean warbler, yellow-throated warbler, wood thrush, scarlet tanager, pileated woodpecker. As every ornithologist familiar with this region knows, the latter species are characteristic birds of the eastern deciduous forest of North America. In an intact forest, they are common. They belong. The former group, which dominated my counts, constitutes generalists or species of the forest edge, birds that thrive in forests degraded by human activities. These species do not need the help of conservationists to survive. What might need help in Sugarcreek, it appeared, are birds of the forest interior. The only common (within the top 20) forest-interior birds in my surveys throughout the summer of 1978 were the red-eyed vireo and acadian flycatcher.

Clearly, I was dealing with a disrupted ecosystem. In today’s world, that is hardly a startling discovery. What really bothered me, however, was that Sugarcreek Reserve contained one of the larger forest patches remaining in this part of Ohio, yet the reserve’s managers apparently did not recognize the value of this forest. Here, in this 228-hectare reserve and the smaller woodlots still (at the time of my surveys) connected to it by wooded corridors, was an opportunity to maintain and restore a vestige of the great eastern deciduous forest. This forest once stretched from northern Florida to central Ontario, from the Atlantic coast to the Great Plains. It was magnificent.

Part of Sugarcreek had been farmed, and the abandoned fields were in the process of returning to forest. Osage-orange and honeylocust were spreading out from the fencerows where they had been planted long ago, while boxelder, white ash, wild black cherry, hickories, and other trees sprang up from the fields of grasses and goldenrods, which once had held corn. The forest was recovering. Yet the managers of Sugarcreek seemed bent on halting this process of succession. They were cutting down areas of regenerating forest and managing them as meadows or thickets. They were mowing insanely wide trails — up to 7 meters in some cases — apparently to accommodate large groups of visitors, but in the process they were fragmenting the forest and creating abundant edge habitat. What were they thinking?

Finally, it dawned on me. The managers were very concerned about content, in particular, how many kinds of habitats the reserve contained and the variety of experiences they offered visitors. But they ignored context, including the ecological history of the region and the nature of the landscape that now surrounded Sugarcreek Reserve. Sugarcreek, its tenuous connections to nearby woodlots notwithstanding, was a virtual island of mature and recovering forest in a sea of cornfields and rapidly proliferating suburbs. It was perhaps big enough to sustain, at least in the short term, populations of the forest-interior birds that dominated the region before Europeans arrived. That is, it might have been big enough, had it been managed for mature forest habitat. Managed instead for maximum habitat diversity and interspersion, it was probably inadequate for the species that needed the reserve most. Indeed, some of the characteristic forest birds that barely hung on at the time of my surveys, such as the black-and-white warbler, had disappeared when surveys were conducted again for the Ohio Breeding Bird Atlas, just a few years later.

In the Sugarcreek story lies a conservation paradox. Management actions undertaken at a local scale to increase diversity might have an opposite effect on a regional and ultimately global scale. This is one of the big problems with using the term “biodiversity” to mean the simple number of species in an area, without regard to their composition, relative abundances, or habitat associations. By increasing diversity locally, managers may unwittingly contribute to the local extirpation of birds and other species dependent on the mature forest that originally dominated the region. If such actions were repeated in other reserves, combined with the general loss and fragmentation of forests in the region, diversity could decline over a broad area as forest-interior species were gradually lost.

And this is not just a problem for birds. Studies have shown that the diversity of many kinds of plants and animals increases locally as weedy generalists, including many nonnative species, move into an area after disturbance by humans (1, 2). In the case of Sugarcreek, the species that benefited when managers maintained an artificial interspersion of habitats were species common throughout the agricultural and suburban landscape. Many of these species would have been rare or absent in the undisturbed forest landscape. Management increased diversity locally by maintaining populations of the early-successional and edge species that had invaded the area during or soon after agricultural development. Similarly, clearcuts often show a pulse of diversity that rivals the diversity of the mature or old-growth forests they replace. This inflated local (and usually temporary) diversity is beguiling, leading some land managers to conclude that the more they disturb the land, the better.

Such is the fallacy of misplaced scale. Thinking locally leads to small-minded decisions and, ultimately, to global homogenization. Land managers need to consider a bigger picture, a broader context. They need to ask, What can I do on this piece of land that will contribute most to regional and global biodiversity over the long term? In this article, I explore this question in relation to three key conservation issues: enhancing population viability, managing disturbances, and prioritizing the protection of species and natural communities. The answers are not always straightforward, but the simple process of seeking them leads us to question established orthodoxies and confront new realities.

Population Viability in a Landscape Context
Recent evidence suggests that the broader context of sites is meaningful in ways few biologists could anticipate 20 or 30 years ago. Every local area is a piece of a bigger ecological puzzle. Its importance can be appreciated only in relation to a larger whole. A natural area of seemingly marginal significance can be enormously valuable if its position and ecological role in the landscape contribute to the persistence of populations on a regional scale.

Over the last several years, my colleagues and I have studied the distribution of suitable habitats and the viability of populations of several species of carnivores — gray wolf, grizzly bear, black bear, wolverine, fisher, cougar, lynx, and others — in the Rocky Mountains of Canada and the U.S. (3, 4) Because many carnivores have demanding area requirements and occur in low densities, their viability must be considered over enormous areas. Our results, based on simulation models but validated by the fieldwork of other researchers across the region, suggest that the most vulnerable sites for carnivores are not necessarily those with the highest levels of site-specific threat (such as logging or residential development) but rather those whose continued degradation would affect nearby, high-quality areas that sustain regional populations.

The concept of population sources and sinks is germane here. A source population (or area) is one where the annual birth rate exceeds the death rate. Young animals born in sources disperse out to other areas, some of which are sinks. In a sink, deaths exceed births in a typical year, so a population can be sustained only by animals moving in from sources. Because individuals of a given species can be reliably found year to year in both sources and sinks, conducting research to differentiate the two is crucial to conservation planning and management. If a land-use plan only protected sinks while opening source areas to development, the population would quickly plummet to extinction.

Paradoxically, in some cases improving conditions in strong sinks might be nearly as crucial to regional population viability as protecting strong sources. For example, a heavily roaded area adjacent to Yellowstone National Park, where many grizzly bears are killed, may be a drain on the population within the park. The viability of the regional bear population could be enhanced by closing roads and limiting development near park boundaries, thus reducing the threat of mortality for animals dispersing from the park. Moreover, an area of low habitat quality for bears may be a key linkage between two high-quality areas whose connection is crucial because the population requires a combined acreage larger than either of the two high-quality areas alone. In this way, habitat connectivity can create a whole larger than the sum of its parts.

The point is to take a broader and more dynamic view of landscapes and populations. Most researchers and managers to date have relied on static habitat suitability models, which provide a snapshot in time of habitat conditions relative to a species’ requirements. These models can be used to predict which parts of a landscape are most favorable for a species but tell us little about how easily a particular site might be colonized or how long a population might persist there. We are now realizing the utility of dynamic population models, which predict habitat occupancy and rate of population growth of a species over time. The dynamic models begin with data on habitat quality from the static models and then build on this information by simulating the birth, dispersal, reproduction, and death of individuals throughout a study region based on what is known about the life history of the species. Thus, the dynamic model provides a population viability analysis (PVA), but unlike many PVAs, it is spatially explicit.

My student, Carlos Carroll, applied these two kinds of models to carnivores in our studies in the Rockies (5). Using the dynamic model, he found that sites that appeared highly suitable on the basis of static models might not be occupied regularly by carnivores because they are too small or isolated. This was especially the case for the lynx, a species whose populations in peripheral areas depend on dispersal from areas of more continuous distribution. Whereas the boreal zone may be occupied consistently because of its high habitat quality and continuity — and the most southerly areas in the species’ range are vacant for the opposite reason — areas in between may be occupied or unoccupied, depending on a complex combination of their habitat quality and proximity to source populations. For the lynx and other species whose habitat is patchy as a result of either natural or human factors, much potentially suitable habitat is likely to be vacant most of the time. If we want to maintain these species in the more fragmented portions of their distributions, we must assure that human-created barriers do not add to the natural barriers that limit connectivity among areas of suitable habitat.

Birds, the best studied of all taxa, provide many other examples of the importance of considering landscape context and source-sink dynamics in conservation. In intensive agricultural regions such as southern Illinois and southern Ontario, most of the remaining patches of forest, and indeed the entire landscape, appear to be sinks for certain species of birds, such as the ovenbird and wood thrush. Predation by edge-loving predators, such as opossums, house cats, raccoons, and crows, and nest parasitism by cowbirds are usually the primary causes of reproductive failure. Why, then, are ovenbirds and wood thrushes still found in small forest patches in these regions year after year? Although proof has been hard to come by, it seems likely that the forest-interior birds that remain in these landscapes are dispersing in from other regions that are still dominated by forest. These regions, such as the Ozark Mountains and central Ontario, are population sources. The excess young birds move out, looking for new areas to raise a family, and wind up in sinks such as southern Illinois and southern Ontario. The only plausible way to change such regions from sinks back to sources is through massive reforestation.

Disturbance Mosaics Need Room to Move
One of my favorite ecosystems in North America is the longleaf pine forest that once dominated the southeastern coastal plain of the United States but has been reduced by 97 percent or more since European settlement. This forest (which is really a savanna — a grassland with scattered trees) was sustained by lightning-ignited fires that returned every two to five years or so. Today, managers of an isolated longleaf pine reserve might conclude that since lightning has not set a fire in their reserve for 15 years, then that must be the natural fire frequency, and there is no need to take management action.

Yet, if they were to step back and take in the bigger picture, they might draw a very different conclusion. The probability of a fire igniting in any single, small reserve is far smaller than the probability of an ignition somewhere across a vast landscape. In fact, in the unfragmented landscape that existed before Europeans arrived, a single ignition might have resulted in a fire that spread over many thousands of acres. Now, barriers such as roads, clearcuts, and urban areas prevent that natural spread — and hence the frequency — of fires.

All of the remaining stands of longleaf pine are fragments of varying size. Beginning in some of the more forward-thinking agencies, such as Florida State Parks, managers now regularly use prescribed burns to simulate the natural fire regime. Species that depend on the open-canopied stands of longleaf pine, such as the endangered red-cockaded woodpecker, have benefited from this management. Yet, it took managers a long time to recognize the broader spatial and temporal context of the remaining stands of longleaf pine and other fire-dependent communities. The general public does not yet see the bigger picture. Prescribed burning is now threatened in Florida and other regions because of public concerns about air quality and safety issues.

The need to consider context begins with the realization that, at one scale or another, all disturbances are patchy. In many kinds of ecosystems, natural disturbance and recovery processes create a shifting mosaic of successional stages across the landscape. In forests, newly disturbed sites, old-growth patches, and many stages in between coexist across a broad region. Over a span of time, the proportion of habitats in various stages may not vary much, although the actual locations of these stages will change relatively frequently as new disturbances arise and previously disturbed areas recover. In this disturbance mosaic, habitats for a range of native species that depend on different stages are maintained over time.

Disturbance mosaics, however, cannot operate in small areas. In a tiny reserve, a single intense fire or windstorm may eliminate all the old growth at once, leaving no refuge from which old-growth species can recolonize the disturbed area over time. Similarly, any earlier successional stage could disappear simply from lack of disturbance, and all the sun-loving species could be shaded out. Managers must not only consider the natural distribution of stages but also try to maintain some semblance of that mosaic through time and across the landscape as a whole, rather than within individual reserves or sites. In addition, given the broader historical and regional context of the landscape, managers should emphasize the development of those stages — typically old growth — that have declined the most since European settlement.

A recent study of fire history over the last 3,000 years in the Oregon Coast Range, combined with a simulation of changes in forest age classes over this period, determined that variability in the distribution of age classes over time was much greater in small areas — for example, the scale of an individual national forest or, smaller yet, a reserve within a national forest — than across the region as a whole (6). In a small reserve, old-growth or late-successional forest could easily disappear by chance, eliminated by a single fire. Across the entire region, however, the simulated proportion of old-growth forest varied mostly between 25 percent and 75 percent and averaged 46 percent, assuring dependable late-successional habitat conditions for such species as the northern spotted owl. When the proportion of old growth declines to a small level (currently about 10 percent) and persists at that level for a long time, the survival of associated species is jeopardized and their recovery is precluded. How much old growth is needed to restore viable populations of associated species is not known with certainty and would depend on the spatial arrangement of stands as well as their total area. Nevertheless, a prudent management goal would be to grow more old growth, up to at least 25 percent of the region, to bring it within the historic range of variability.

Are All Species Created Equal?
I am continually surprised by conservation assessments and management plans that, explicitly or implicitly, give all species equal weight. As living creatures, the European starlings and common grackles that I observed at Sugarcreek are every bit as respectable as the cerulean warblers and wood thrushes. Similarly, house flies and common dandelions are as admirable in the evolutionary scheme of things as the San Diego fairy shrimp or Ka’u silversword. But the former are not of conservation concern, whereas the latter are.

Perhaps all species are created equal, but treating all species as equal in any given place or time would be a disaster. As Jared Diamond noted nearly three decades ago, “Species must be weighted, not just counted…[as] the question is not which refuge system contains more total species, but which contains more species that would be doomed to extinction in the absence of refuges” (7).

Although Diamond’s advice is slowly sinking in, it has been an uphill battle. Consider diversity indices. In the 1960s and 1970s, these indices were common in the ecological literature and are sometimes still used today in environmental impact assessments. The simplest measure of diversity is species richness, the number of species in a defined area or sample. More complex indices usually combine richness with a measure of the evenness of abundances, with evenness peaking when all species in a sample have identical numbers of individuals. All these measures hide information on the identity of species, yet they have been used to place higher value on communities with higher richness or diversity, regardless of species composition. Thus, a species-rich community dominated by nonnative species would be favored over a community dominated by natives, but with fewer species. Although few conservationists make this error today, some people who propose engineering projects, timber sales, and other developments justify their projects on the basis that they will increase biodiversity, meaning the number of species locally, regardless of identity and conservation status. They create the same fallacy of misplaced scale that I saw at Sugarcreek, and some decision-makers still fall for it.

A useful tool for weighting species according to their conservation value is the “G/S” ranking system developed by The Nature Conservancy and now managed by a spin-off organization, NatureServe (see This system tracks species according to their level of imperilment, based on rarity and other measures of vulnerability to extinction. The highest ranked species (G1) are considered critically imperiled globally. Most of these species are narrow endemics, found only within small areas and in small populations. One shopping mall could wipe them off the face of the earth.

At the opposite end of the spectrum, G5 species are demonstrably widespread, abundant, and secure. Species are also ranked on national and subnational (e.g., state) scales, and the system is similarly applied to ecological communities. A species or community that is found in only one site in a state, but is common elsewhere, might be ranked G5S1. A species or community with this rank usually would not warrant as much concern as one ranked G1S1 or G2S1, but it would still be worth protecting, especially if it were a disjunct population that is genetically distinct from the rest of its species.

Today, conservationists and managers are more concerned with entire ecosystems than with single species, and rightly so. But here again, simplistic notions of diversity are foolhardy. Maximizing the diversity of ecosystems in an area through management makes no sense. The historical as well as the spatial context of ecosystems must be considered. Ecosystems in the United States that have suffered dramatic declines since European settlement — for example, longleaf pine and coastal redwood forests, tallgrass prairie, and seagrass meadows — deserve greater attention than ecosystems that are still relatively common, and arguably more than those ecosystems that have always been rare. Treating different species and ecosystems as nonequals might smack against our egalitarian ideals, but it is the only efficient way to confront the extinction crisis.


I learned some valuable lessons in my studies at Sugarcreek, lessons that have been re-inforced by other researchers, in many regions, over the intervening years. The task ahead is to translate the research into management practice, and this is basically a matter of education. Aldo Leopold mused that “one of the penalties of an ecological education is that one lives alone in a world of wounds,” (8) but there are also many rewards of such education. We may see the wounds when we look at the big picture, but we also see how wounds can be healed. We see the potential for a fully restored and vibrant landscape, a nature that is whole. With a better understanding of ecology combined with a genuine conservation ethic, managers will make the right decisions.


In the 1960s, ecologists Robert MacArthur and Edward Wilson introduced their theory of island biogeography, which suggested that the number of species on an island or island-like habitat represents a balance between colonization by new species and extinctions of existing species on the island. These processes, in turn, are determined by how close the island is to the mainland (or to a large patch of habitat) and by the size of the island, respectively. Big islands near a source of colonists are predicted to be the most diverse. Although empirical support for MacArthur and Wilson’s theory is mixed, and many other factors come into play in determining diversity, the theory was historically pivotal in that it got many scientists and conservationists thinking seriously about the effects of habitat area and isolation on population persistence and species diversity.

In the early 1980s, the field of landscape ecology was introduced to North America. Richard Forman and others discussed the factors that determine species distributions and abundances across areas many miles wide. They pointed out that the pattern of habitat patches and corridors — as well as the nature of the surrounding landscape matrix — could have a substantial effect on species composition and ecological processes. That now seems so obvious. Yet thinking about habitat at such broad scales differed strikingly from the scale at which wildlife habitat was usually evaluated in those days and often still is today.

In 1986, Larry Harris and I published a paper in which we urged conservationists to expand their thinking from the content of natural areas to the context of the landscape in which they are embedded. We noted that areas usually are selected for protection based on their contents — the species and communities they contain. Often the interest is in the scenic features or recreational potential of an area or a lack of conflicts with resource production. Sometimes it is just plain opportunism — the land is available for a reasonable price. Even when more scientific criteria are applied to selecting reserves, for example, representing a particular plant community or protecting a population of a rare species, the focus is on the piece of land or water itself, not the landscape that surrounds it. Hence, the native biodiversity of many reserves degrades over time as area-demanding species drop out and external influences increase.

Conservation planning today, in contrast to just a decade or two ago, is usually pursued on a regional scale. The Wildlands Project, formed in 1991, pioneered the development of regional reserve networks, drawing on preliminary designs I and a few others developed during the 1980s. Ultimately, the Wildlands Project plans to extend such designs across North America and beyond. The U.S. Gap Analysis Project, initiated in the early 1990s by J. Michael Scott and Blair Csuti and now housed in the U.S. Geological Survey, set out to evaluate how well biodiversity is represented in existing reserves across the country. The inevitable answer was — not too well. By the mid-1990s, The Nature Conservancy, World Wildlife Fund, Conservation International, the Wildlife Conservation Society, and many other groups had embraced ecoregional planning or similar large-scale efforts. Meanwhile, Australian biologists, notably Robert Pressey and colleagues, developed sophisticated computer algorithms that made the selection of new reserves to complement existing ones much more efficient. Given explicit goals and reasonably good databases on biodiversity surrogates, these algorithms can answer the perennial question — how much is enough? — much more reliably than in the past. Finally, we are beginning to see the big picture.

Forman, R.T.T. and M. Godron. 1986. Landscape Ecology. Wiley, New York, NY.

MacArthur, R.H. and E.O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, NJ.

Noss, R.F. and L.D. Harris. 1986. Nodes, networks, and MUM’s: preserving diversity at all scales. Environmental Management 10:299-309.


1. The integrity of any piece of land or water is ultimately dependent on the integrity of the landscape that surrounds it.

2. Many species require areas well outside protected area boundaries. Whenever possible, work with regional planners to assure that key pieces of habitat are functionally connected, through habitat corridors or low-intensity intervening land use.

3. Ecological history is highly informative. Which habitats and species have prospered and which have declined as the extent and intensity of human activity increased?

4. Bigger scales are ultimately more meaningful for conservation of global biodiversity. Recognize that management may increase the number of species on one spatial scale while decreasing it on others.

5. Do not treat all species as equal. Focus attention on those most vulnerable to extinction as a result of human actions, those that have declined most, and those that are ecologically significant (including exotics and other opportunistic species that threaten native biodiversity).

6. Limit human activities around and within reserves and other natural areas. Surround sensitive areas of any size with transitional or buffer zones where the intensity of human uses increases with greater distance from the sensitive area.

Literature Cited

Noss, R.F. 1983. A regional landscape approach to maintain diversity. BioScience 33:700-706.

Mooney, H.A. 1988. Lessons from Mediterranean-climate regions. In Wilson, E.O. ed. Biodiversity. National Academy Press, Washington DC.

Carroll, C., R.F. Noss, and P.C. Paquet. 2001. Carnivores as focal species for conservation planning in the Rocky Mountain region. Ecological Applications 11:961-980.

Noss, R.F. et al. In Press. A multi-criteria assessment of the irreplaceability and vulnerability of sites in the Greater Yellowstone Ecosystem. Conservation

Carroll, C. 2000. Spatial modeling of carnivore distribution and viability. Dissertation. Oregon State University, Corvallis, OR.

Wimberly, M.C. et al. 2000. Simulating historical variability in the amount of old forests in the Oregon Coast Range. Conservation Biology 14:167-180.

Diamond, J.M. 1976. Island biogeography and conservation: Strategy and limitations. Science 193:1027-1029.

Leopold, A. 1993. Round River. Oxford University Press, New York.

About the Author

Reed Noss, PhD is Chief Scientist for The Wildlands Project and Distinguished Research Professor of Biology at the University of Central Florida. He is a former editor of Conservation Biology (1993-1997) and Past President of the Society for Conservation Biology (1999-2001).