Animals and Climate
Animals and Climate
The responses of vertebrate animals (fish, amphibians, reptiles, birds, and mammals) to climate change can be multidimensional, because their geographic distributions depend on diverse variables with complex interrelationships. Climate parameters themselves can contribute to these complexities: some species may be limited by temperature, whereas others are more sensitive to precipitation. Effective moisture, which is determined by both temperature and precipitation, may restrict other species. Snow cover, storm frequencies, and cloud cover are a few other variables that influence geographic ranges.
Seasonal extremes are more important in limiting distributions than are fluctuations in mean annual parameters. For example, the southern boundary of a “northern” species may depend on summer temperature extremes rather than extremes in winter or mean annual temperatures. If summers become warmer, populations along the southern boundary of the organism's range may be adversely affected, and the distribution may be subsequently restricted farther north. Conversely, if summers become cooler, then the species may expand its distribution farther south.
The variability of seasonal extremes is also critical. If extreme conditions occur infrequently, they may reduce population levels each time, but the populations will recover or new ones will recolonize the area when favorable conditions return. If extremes recur frequently, however, a species may be totally extirpated from an area. In this sense, the scale of climate change is also a factor. Local changes may have little effect on a species, but regional, continental, or global scales can significantly alter its distribution or ultimately lead to its extinction. Finally, the rate and magnitude of climate change must be considered. Rapid and large-magnitude changes are more difficult for animals to adapt to because of the lag in the response rate of a biotic system. Therefore, these changes will have a greater effect on animals than will slow and low-magnitude changes.
Organisms may be affected directly or indirectly by climate change. Direct effects occur when climate forces a geographic shift of a species without intermediate processes such as habitat alteration. Climate warming that causes a range shift because an organism's physiological tolerances will not permit it to inhabit the warmer environment is an example of a direct effect. An indirect effect results when a climatic change alters some variable in the environment (for example, vegetation or substrate) that then causes a distributional change. It is often difficult to determine whether climate has direct or indirect effects, but it is essential to make this distinction to understand fully an organism's potential response to climate changes, past or future.
Collared lemmings (Dicrostonyx spp.) and barren ground musk oxen (Ovibos moschatus) are two animals for which summer temperature extremes are direct limiting factors. Laboratory experiments have shown that collared lemmings can withstand fatally cold temperatures twice as long as many other rodent species, including several with larger body sizes. This adaptation allows the collared lemming to emerge on the snow surface during the winter; many other small mammals must remain below the snow to avoid being killed by severe cold temperatures. In addition to these physiological adaptations, the collared lemming's winter fur changes to white, and it grows special skin flaps on its front paws to facilitate burrowing through hard snow. Its ability to tolerate very warm temperatures, however, is low, and such temperatures are lethal to it. In the early twenty-first century the collared lemming is restricted to cold environments in the Arctic. During the Pleistocene ice ages, this lemming occurred at much lower latitudes in the United States, but it was closely associated with cold environments along the glacial front (see Figure 1). With global warming at the end of the Pleistocene, the collared lemming retreated to its present Arctic range.
The barren ground musk ox has a dense undercoat that is essentially impervious to moisture and cold air. It serves as an excellent insulator from the cold, but it also prevents effective body cooling. Therefore, the barren ground musk ox cannot tolerate warm summer temperatures. In fact, attempts to raise domestic herds in northern New England failed because the summers were too hot. Like the collared lemming, the barren ground musk ox is restricted to the Arctic in the early twenty-first century, but it also ranged to lower latitudes in the Pleistocene.
Conversely, winter temperature extremes may be important in limiting some species. For example, the northern limit of the eastern phoebe (Sayornis phoebe), a bird, is closely associated with the −4°C isotherm of January minimum temperature, suggesting that this may be a limiting factor in its distribution. Birds, like many flying insects and mammals, can respond rapidly to environmental changes; when seasonal temperatures warm or cool, volant (flying) species limited by physiological adaptations can soon change their geographic distributions. Species constrained by specific habitat requirements may not be able to respond so rapidly; in fact, considerable lag times of centuries to millennia may be involved.
Moisture can also have direct effects on the distributions of species. Laboratory and field studies indicate that the eastern chipmunk (Tamias striatus) requires more water than any of the western chipmunks. As a consequence, this species is restricted to moist forest environments in eastern North America. If climates change so that environments become moister or drier, this species may expand or contract its range accordingly. The moister climates of the Pleistocene allowed the eastern chipmunk to extend its distribution westward, but with late Quaternary warming and drying, by 8,000 years ago it was restricted to its modern distribution (see Figure 2). In fact, water balance and water stress may be important factors in determining the times of hibernation for many ground-dwelling squirrels.
Moose (Alces) and other members of the deer family (the cervids) can be limited locally by snow depth and snow crusting; biologists refer to this as “enforced yarding.” When snow depth is greater than 75 centimeters and the snow is heavily crusted, moose may be restricted to localized feeding areas (yards). Snow may also be important on a regional scale under certain climatic conditions. The long
Climate can affect the life stages of a species in different ways. Like those of many cold-blooded (ectothermic) vertebrates, the eggs of the European pond tortoise (Emys orbicularis) are laid in light soil exposed to the warmth of the sun, dependent on ambient temperatures for incubation. The eggs of the pond tortoise will hatch only in the geographic area within the 18°C July isotherm. The long-lived adult tortoises may wander some distance north of this isotherm, but reproduction is limited to the area below it. During the warmer Pleistocene interglacials, the pond tortoise extended its distribution into England; the large numbers of their fossil remains indicate that they represent breeding populations and not wandering vagrants.
For the loggerhead sea turtle (Caretta caretta), soil temperatures of sediments enclosing the eggs may be important in determining the sex ratio of the hatchlings. When temperatures are elevated above the optimum, the sex ratio may be skewed toward more females. Field studies in a rookery (nesting site) at Cape Canaveral, Florida, have shown that almost 90 percent of the hatchlings for more than 5 years have been females. These female-biased sex ratios pose some interesting theoretical challenges for conservation of this species in light of possible future global warming.
Habitat alterations, especially vegetational changes, are probably the most common indirect effect of climate on animals. For example, tree squirrels (Sciurus spp.) require forested or woodland habitat, whereas black-tailed prairie dogs (Cynomys ludovicianus) inhabit shortgrass prairie. Woodland and grassland habitats are limited by precipitation. If climates become drier, then shortgrass prairie will expand and forests will contract; if climates become moister, the reverse will occur. Black-tailed prairie dogs will alter their range in accordance with the distribution of shortgrass prairie, and tree squirrels will follow the distribution of woodland.
Correlations between vegetational associations and animal species may not always indicate that a particular species is dependent on a certain vegetational type. Today in England, the beetle Calosoma inquisitor preys almost exclusively on caterpillars that defoliate oak trees, so its distribution is strongly correlated with the distribution of oaks. After the glacial ice retreated, however, Calosoma reinvaded England before any oaks arrived. This suggests that the beetles may have been directly affected by temperature changes and that they are not as highly coevolved with oaks as suggested by their modern interactions.
Substrate is an important limiting factor, especially for fossorial (burrowing) species. For gophers, soil type and texture may be more critical than vegetation. In the United States, there are 3 different genera of gophers (Geomys, Thomomys, and Cratogeomys) and each genus has specific and different preferences for soil type and soil depth. Thomomys and Cratogeomys inhabit more xeric (drier) environments with shallow soils; Geomys prefers moister habitats with better-developed soils. Climates that influence the formation of different soil types will select for different gophers. Changes in the stratigraphic distribution and abundance of remains of these 3 gophers in deposits of a central Texas cave document decreasing soil depth on the Edwards Plateau throughout the Holocene as a result of erosion.
Substrate can also be important for fish. The blue sucker (Cycleptus elongatus) inhabits deep, swift channels of large streams with a bottom of sand, gravel, or rock. The river carpsucker (Carpiodes carpio) prefers quiet, silt-bottomed pools, backwaters, and oxbows of large streams with moderate or low gradients. Stream habitats are determined by a variety of variables, but precipitation patterns in drainage basins can influence stream regimes and thus determine the types of fish that inhabit them. Fish remains from archeological sites along the Illinois River record the physical evolution of the river's floodplain during the warm middle Holocene.
Biological interactions also determine distributions, but these interactions may be dependent on climate. Competition may limit the distribution of one species while allowing another species to expand. In the southern plains of the United States, the cotton rat (Sigmodon hispidus) appears to be expanding its distribution at the expense of the prairie vole (Microtus ochrogaster). Laboratory and field observations indicate that the cotton rat will aggressively displace the prairie vole in all areas where they come into contact, except where vegetational cover is thick—a climatically dependent variable.
Predation is another biological interaction with important implications for controlling distribution. For white-tailed deer (Odocoileus virginianus), snow depth may be a critical factor. Wildlife biologists have found that wolf predation on white-tailed deer in Minnesota is more effective in deeper snow because the snow makes it harder for the deer to escape from wolves, which can run on top of the snow; however, the long legs of moose give them an advantage in deep snow.
Parasites and disease can be effective limiting factors, and their prevalence can be dependent on climatic conditions. Anthrax is a fibril disease caused by the bacterium Bacillus anthracis, which adversely affects almost all mammals, including humans. Under dry conditions the bacterium forms highly resistant spores that may persist and retain their virulence in contaminated soils for years. Dry, windy environments favor the spread of this disease. Because of these environmental conditions in the northern plains of North America, anthrax has periodically threatened the survival of bison in Canadian wildlife parks.
The meningeal worm Parelaphostrongylus frequently infects white-tailed deer, which can tolerate this nematode. However, this worm can be fatal to other cervids such as moose, elk (Cervus elaphus), and caribou (Rangifer tarandus). When white-tailed deer invade an area inhabited by other cervids, this parasite can cause reduction or even elimination of the other cervid populations. The ranges of these cervids may be reduced by this parasite when environmental conditions, especially warm winters, allow white-tailed deer to invade their ranges; it appears that the presence of this parasite in Minnesota deer populations may limit the southern distribution of moose.
Animals may escape climatic effects through ethological (behavioral) adaptations that allow them to inhabit environments normally inhospitable to them. Many northern rodents avoid severe winter temperatures by living under the snow cover, which buffers the ambient temperature. The brown lemming (Lemmus sibiricus) lives in the same Arctic environment as the collared lemming, but it is not physiologically adapted to survive for long periods in extremely cold temperatures. Instead, it stays under the snow and requires a continuous snow cover. Pikas (Ochotona spp.) have a high body temperature and low upper lethal temperature, so they are very sensitive to warm summer temperatures. They occupy alpine talus (broken rock) patches, which offer cool, protected microenvironments. At lower elevations pikas may curtail their foraging activities during midday, and they may reduce their foraging distance from the talus to further avoid heat. These behavioral traits allow pikas to extend their distributions beyond their optimal range.
Hibernation and estivation are behavioral and physiological mechanisms that provide protection from environmental conditions. Hibernation is a period of inactivity during the winter season; estivation is a period of inactivity during the summer months. Significant changes in physiological processes, known as torpor, occur during these times of inactivity. In estivation, body temperature cannot be lowered below a certain level, but during hibernation body temperature may be reduced to the level of the surrounding environment. Hibernation and estivation are controlled by a variety of climatically dependent factors, especially food supply and availability of water.
Some species—such as the cactus mouse (Peromyscus eremicus), an inhabitant of the deserts of the southwestern United States and northern Mexico—have discontinuous inactive periods during both summer and winter. Other species—such as the 13-lined ground squirrel (Spermophilus tridecemlineatus)—have inactive periods for nearly continuous periods as long as 250 days. Species such as the California pocket mouse (Perognathus californicus) have the ability to lower their body temperature to near that of the surrounding environment in daily torpor cycles rather than seasonal ones. The length of the daily torpor is related to its food supply—abundance of seeds—which fluctuates with precipitation patterns.
The time of hibernation (denning period) for black bears (Ursus americanus) is strongly correlated with climatic conditions. In the southern United States, black bears may hibernate for a month or not at all, whereas in Alaska the hibernation will last for as long as 6 months. Hibernation time also increases with altitude. Changes in hibernation time (climate) may not only affect the distribution of black bears, but it may be an important factor in determining body size. Short hibernation periods allow for greater foraging and hence larger body size; longer hibernation reduces foraging and selects for smaller size. Vegetational quality may also be a critical factor.
It is interesting to consider how aggregates of species might respond to the same climatic change. One model would suggest that large groups of species (communities) would respond as a unit and shift their geographic distribution at the same time, in the same direction, and at the same rate. This community unit model suggests that these species are tightly linked and highly coevolved assemblages. In an individualistic model, each species responds to environmental changes in accordance with their own tolerance limits. Species that occur together may migrate in different directions, at different rates, and at different times after a climate change; this individualistic response will result in new associations of species, whereas a community unit response will result in an association of the same species in a different location.
These two models have very different implications for understanding and planning for responses to climate change by communities, biomes, and ecosystems. The community unit model would predict that communities are stable in composition and that they persist as definable entities over long periods of geologic time. This model therefore permits the use of analogies with modern communities for paleoenvironmental reconstructions and for future predictions.
An individualistic model treats communities as loosely organized collections of species in which coexistence depends on the tolerance limits of individual species distributed along environmental gradients. Species associations may thus appear ephemeral in geologic time, and the species composition of modern communities may not be pertinent for reconstructing past communities or assembling future ones. This model would also suggest that it is imperative to have systematic connections between protected areas to facilitate species migration.
Paleoecological studies demonstrate that late Quaternary climatic changes did not cause simple latitudinal or altitudinal shifts in communities. Plants and animals responded as individual species to these changes. As a result, some Pleistocene communities do not have any modern analogs. We might therefore expect communities responding to future climate changes to have compositions quite different from those of today.
Observable Impacts of Climate Change Today.
There are observable impacts of climate change on terrestrial ecosystems, including changes in the timing of the onset, completion, and length of the growing season, phenology, primary production, and species distributions and diversity (Walther, 2002; Parmesan, 2003). Because species differ greatly in their life-history strategies, physiological tolerances, and dispersal abilities, there is high variability in detecting species responses to climate change. As highlighted above, many animals have evolved powerful mechanisms to regulate their physiology, thereby avoiding some of the direct influences of climate change, and instead interact with climate change through indirect pathways involving their food source, habitat, and predators (Schneider, 1996). Consequently, most distributional studies incorporate integrated measures of direct and indirect influences to changes in the climate environment and tend to focus on animals, whereas phenological studies, which incorporate measures of direct influences, focus on plants and insects. Although most studies tend to separate distributional and phenological effects of climate change, it is important to keep in mind that the two are not independent and interact with other changing variables to determining impacts to species (Parmesan, 2006). In addition, most of the observed species responses have described changes in species phenologies (Parmesan, 2006).
Using modeled climatic variables and observed species data, Root (2005) shows that human activities have contributed significantly to temperature changes and that human-changed temperatures are associated with discernible changes in animal traits (IPCC, 2007). Evidence from two meta-analyses (143 studies, Root, 2003; 1,700 species, Parmesan, 2003) and a synthesis (866 studies, Parmesan, 2006) on species from a broad array of taxa suggests that there is a significant impact from recent climatic warming in the form of long-term, large-scale alteration of animal populations including changes in distribution (Root, 2006, 2003; Parmesan, 2003). [See Meta-analysis.] Parmesan (2006) describes three types of studies documenting shifts in species ranges: (1) those that measure the an entire species range, (2) those that infer large-scale range shifts from observations across small sections of the species’ range, and (3) those that infer large-scale range shifts from small-scale change in species abundances within a local community. Although very few studies have been conducted at a scale that encompasses an entire species’ range [amphibians (Pounds, 1999, 2006), pikas, (Beever, 2003) birds (Dunn, 1999), and butterflies (Parmesan, 2006, 1996)], there is a growing body of evidence that has inferred large shifts in species range across a very broad array of taxa. Nearly 60 percent of the 1,598 species studies exhibited shifts in their distributions and/or phenologies over the 20- and 140-year time frame (Parmesan, 2003). Field-based analyses of phenological responses of a wide variety of different species have reported shifts as great as 5.1 days per decade (Root, 2003), with an average of 2.3 days per decade across all species (Parmesan, 2003).
Asynchronous Responses and Their Implication.
Studies are beginning to document the asynchronous nature in the response of animal species directly to climatic change and ponder the long-term implications of these mismatches that may alter competitive interactions, food source availability, and habitat availability at multiple life-history stages. The most vivid example of this to date is in migratory birds. For example, the timing of arrival at breeding territories and overwintering grounds for migratory birds is an important determinant of reproductive success, survivorship, and fitness. Climate variability on interannual and longer time scales can alter phenology and ranges of migratory birds by influencing the time of arrival and/or the time of departure. The earlier onset of spring has consequences for the timing of migration and breeding in birds that evolved to match peak food availability (Visser, 2006). We might expect that the timing of migration would track temporal shifts in food availability caused by changes in climate and the advancement of spring. In a continental-scale study of bird phenology that covered the entire United States and Canadian breeding range of a tree swallow (Tachycineta bicolor) from 1959 to 1991, Dunn and Winkler documented a 9-day advancement of laying date that correlated with the changes in May temperatures (Winkler, 2002; Dunn, 1999). In a study of the first arrival dates of 103 migrant bird species (long-distant and very long-distant migrants) in the Northeast during the period 1951–1993 compared to 1903–1950, all migrating species arrived significantly earlier, but the birds wintering in the southern United States arrived on average 13 days earlier, whereas birds wintering in South America arrived 4 days earlier (Butler, 2003). In a reversal of arrival order for short- and long-distance passerines, Jonzen (2006) showed that long-distance migrants have advanced their spring arrival into Scandinavia more than short-distance migrants, based on data from 1980 to 2004. Similarly, in a 42-year analysis of 65 species of migratory birds through Western Europe, researchers found autumn migration of birds wintering south of the Sahara had advanced, whereas migrants wintering north of the Sahara delayed autumn migration (Jenni, 2003). Finally, a study that combined analysis of spring arrival and departure dates of 20 trans-Saharan migratory bird species to the United Kingdom found an 8-day advance in the arrival and the departure time to the breeding grounds, but no change in the residence time. The timing of arrival advanced in relation to increasing winter temperatures in sub-Saharan Africa, whereas the timing of departure advanced in response to elevated summer temperatures in their breeding ground (Cotton, 2003). These asynchronous changes will likely impact the long-term reproductive success of the migratory birds, but at this time it is not clear how. Without an understanding of how these asynchronous changes correlate with phenology of the food resource, it is difficult to discern what the long-term consequences might be (Visser, 2005). As these studies suggest, when spring migration phenology changes, migrants may be showing a direct response to trends in weather or climatic patterns on the wintering ground and/or along the migration route or there may be indirect microevolutionary responses to the selection pressures for earlier breeding (Jonzen, 2006). A climate change signature is apparent in the advancement of spring migration phenology (Root, 2003) but the indirect effects may be more important than the direct effects of climate in determining the impact on species persistence and diversity. Indeed, there is no reason to expect migrants and their respective food sources to shift their phenologies at the same rate.
Differential shifts will lead to mistimed reproduction in many species. There may be significant consequences of such mistiming if animal populations are unable to adapt (Visser, 2004). Phenological shifts in migration timing in response to climate change may lead to the failure of animal species to breed at the time of abundant food supply (Visser, 2005, 2006; Stenseth, 2002) and, therefore, may have implications for population success if the shift is not synchronous with food supply availability or if competitive interactions between species are exacerbated. Understanding where climate change-induced mistiming will occur and their underlying mechanisms will be critical in assessing the impact of climate change on the success of animal species (Visser, 2005). The responses will not be uniform across animal species ranges and are thus likely to be highly complex and dependent on species-specific traits, characteristics of local microhabitats, and aspects of local microclimates.
Begun, M., J. L. Harper, and C. R. Townsend. Ecology: Individuals, Populations and Communities. Sunderland, Mass.: Sinauer, 1986.Beever, E. A., P. F. Brussard, and J. Berger. “Patterns of Apparent Extirpation among Isolated Populations of Pikas (Ochotona princeps) in the Great Basin.” Journal of Mammalogy 84 (2003): 37–54.Butler, C. J. “The Disproportionate Effect of Global Warming on the Arrival Dates of Short-Distance Migratory Birds in North America.” Ibis 145, no. 3 (2003): 484–495.Chepko-Sade, B. D., and Z. T. Halpin, eds. Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics. Chicago: University of Chicago Press, 1987.Cotton, P. A. “Avian Migration Phenology and Global Climate Change.” Proceedings of the National Academy of Sciences of the United States of America 100, no. 21 (2003): 12219–12222.Dunn, P. O., and D. W. Winkler. “Climate Change Has Affected the Breeding Date of Tree Swallows throughout North America.” Proceedings of the Royal Society Biological Sciences Series B 266 (1999): 2487–2490.Gee, J. H. R., and P. S. Giller, eds. Organization of Communities; Past and Present. Oxford: Blackwell Scientific, 1986.Graham, R. W., and E. C. Grimm. “Effects of Global Climate Change on the Patterns of Terrestrial Biological Communities.” Trends in Ecology and Evolution 5 (1990): 289–292.Hazard, E. B. The Mammals of Minnesota. Minneapolis: University of Minnesota Press, 1982.Intergovernmental Panel on Climate Change. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by S. Solomon et al., 996 pp. Cambridge, UK: Cambridge University Press, 2007.Jenni, L., and M. Kery. “Timing of Autumn Bird Migration under Climate Change: Advances in Long-Distance Migrants, Delays in Short-Distance Migrants.” Proceedings of the Royal Society Biological Sciences Series B 270 (2003): 1467–1471.Jonzen, N., A. Linden, et al. “Rapid Advance of Spring Arrival Dates in Long-Distance Migratory Birds.” Science 312 (2006): 1959–1961.Mrosovsky, N., and J. Provancha. “Sex Ratio of Hatchling Loggerhead Turtles: Data and Estimates from a 5-Year Study.” Canadian Journal of Zoology 70 (1992): 530–538.Murie, J. O., and G. R. Michener. The Biology of Ground-dwelling Squirrels. Lincoln, Neb.: University of Nebraska Press, 1984.Parmesan, C., N. Ryrholm, et al. “Poleward Shifts in Geographical Ranges of Butterfly Species Associated with Regional Warming.” Nature (London) 399 (1999): 579–583.Parmesan, C., and G. Yohe. “A Globally Coherent fingerprint of Climate Change Impacts across Natural Systems.” Nature 421 (2003): 37–42.Parmesan, C.. “Ecological and Evolutionary Responses to Recent Climate Change.” Annual Review of Ecology Evolution and Systematics 37 (2006): 637–669.Peters, R. L., and T. E. Lovejoy, eds. Global Warming and Biological Diversity. New Haven, Conn.: Yale University Press, 1992.Pounds, J. A., M. P. L. Fogden, et al. “Biological Response to Climate Change on a Tropical Mountain.” Nature 398 (1999): 611–615.Pounds, J. A., M. R. Bustamante, et al. “Widespread Amphibian Extinctions from Epidemic Disease Driven by Global Warming.” Nature 439 (2006): 161–167.Root, T. L. “Effects of Global Climate Change on North American Birds and Their Communities.” In Biotic Interactions and Global Change, edited by P. Kareiva, J. Kingsolver, and R. Huey, pp. 280–292. Sunderland, Mass.: Sinauer, 1992.Root, T. L., et al. “Fingerprints of Global Warming on Wild Animals and Plants.” Nature 421 (2003): 57–60.Root, T. L., et al. “Human-modified Temperatures Induce Species Changes: Joint Attribution.” Proceedings of the National Academy of Sciences 102, no. 21 (2005):7465–7469.Schneider, S. H., and T. L. Root. “Ecological Implications of Climate Change Will Include Surprises.” Biodiversity and Conservation 5, no. 9 (1996): 1109–1119.Stenseth, N. C., and A. Mysterud. “Climate, Changing Phenology, and Other Life History Traits: Nonlinearity and Match–Mismatch to the Environment.” Proceedings of the National Academy of Sciences of the United States of America 99, no. 21 (2002): 13379–13381.Visser, M. E., and C. Both. “Shifts in Phenology due to Global Climate Change: The Need for a Yardstick.” Proceedings of the Royal Society Biological Sciences Series B 272 (2005): 2561–2569.Visser, M. E., L. J. M. Holleman, et al. “Shifts in Caterpillar Biomass Phenology due to Climate Change and Its Impact on the Breeding Biology of an Insectivorous Bird.” Oecologia (Berlin) 147 (2006): 164–172.Walther, G, et al. Ecological Responses to Recent Climate Change. Nature 41 (2002): 389–395.Winkler, D. W., P. O. Dunn, et al. “Predicting the Effects of Climate Change on Avian Life-History Traits.” Proceedings of the National Academy of Sciences of the United States of America 99, no. 21 (2002): 13595–13599.
Russell W. Graham; revised and updated by Rebecca Shaw