- Email: [email protected]
How effective is interdemic selection?
CORRESPONDENCE all publicity to the contrary, global environmental damage is, and has always been, inversely related to population growth.’ I am not sure where Chichilnisky has seen ‘all the publicity’ about high population growth in the North. To the contrary, I have seen abundant publicity about the low rates of population growth -with fertility rates below replacement level - in most European countries. It seems Chichilnisky is confusing population growth with population density. Elsewhere in her article she states that ‘[clontrary to popular wisdom, the less populated regions cause the most environmental damage.’ This statement contradicts her claim in Box 1 and suggests incorrectly that the lesser-developed countries, where population growth is generally high but population density still low in many cases, cause the most damage. In fact, if we accept her linkage of environmental damage to excessive resource consumption, then some of the most heavily populated regions (e.g. Europe and Japan) cause the most damage. The United States is something of an anomaly, in that it probably causes the most global damage per capita but has relatively low population density and fertility. It is also worth considering alternative measures of environmental damage. I don’t want to let the ‘North’ off the hook, but tropical countries with their much higher species density and endemism may experience much greater damage per unit area from human activities than higher-latitude regions. If species extinctions are a measure of environmental damage, then the landless farmer in Brazil slashing and burning 10 ha of primary tropical forest may be more destructive than the rich developer in the United States destroying 10 000 ha of second-growth temperate forest. Remember, direct habitat alteration is still the greatest cause of biodiversity losses. Virtually all economies, political systems, and cultures are degrading the natural environment today; to what extent has to do with population, level of consumption, and efficiency of resource production, as Paul Ehrlich and colleagues pointed out long ago2. It may be politically correct to view the North as ‘bad’ and the South as ‘good’ (which I’m sure Chichilnisky did not intend to suggest); but we all must share the blame and collectively try to find solutions. Despite some loose terminology, Chichilnisky’s article brings us a long way toward solutions.
Reed F. Noss 7310NW Acorn Ridge, Corvallis, OR 97330, USA References 1 Chichilnisky, G. (1996) Trends Ecol. Evol. 11, 135-140 2 Ehrlich, P.R., Ehrlich, A.H. and Holdren, J.P. (1970) Ecoscience: Population, Resources, Environment, W.H. Freeman
Reply from G. Chichilnisky Reed Noss ‘North’ and developing terminology. Prices and
298
calls attention to my use of the terms ‘South’ to describe industrial and countries. This is completely standard Indeed, in my section on Resource Overconsumption I define these terms:
0
1996, Elsevier Science Ltd
Year
??Industrial, 0
Developing,
fossil fuels fossil fuels
Industrial, Developing,
beef and veal beef and veal
Fig. 1.Consumption of fossil fuels and meat in industrial and developing countries, 1961-1990.
‘[O]n the whole, the situation today can be described as the industrial countries (the North) overconsuming environmental resources which are overextracted in the developing countries (the South)‘. Figure 1 illustrates this for fossil fuels. The fact is that most industrial countries are in the North and most developing countries are in the South. This is why one uses North and South to describe industrial and developing countries. The coincidence is, of course, not perfect. Australia is in the South and the World Bank ranks it today as the richest nation in the world using a new valuation which values natural resources highly. The Russian Federation is in the North and some parts of it qualify as developing countries. North and South are terms that are typically used in economics and political sciences to describe industrial and developing countries, without a tight connection to geography. Noss’s second point refers to my sentence in Box 1: ‘despite all publicity to the contrary, global environmental damage is, and has always been, inversely related to population growth’. He misread this sentence, thinking that I was mentioning publicity about ‘high population growth in the North’. I was not-just the opposite. The North’s population growth is the lowest, as I mention in my article. The South’s is the highest. My point is that much publicity goes to imply that environmental damage will disappear if only population growth in developing nations would decrease. This is incorrect today, and also in historical terms. If humans were to disappear, of course, the problem would not be there. Indeed, with the 50% decline some observe in male sperm, the ultimate solution may be within our reach. But in reality the areas of the world with lower rates of population growth are causing today, and have caused historically, the most damage to the
global environment. I refer specifically to global emissions of greenhouse gases and CFCs, and to global destruction of biodiversity. These are the two main global environmental problems, according to the United Nations Earth Summit, and according to all major international agencies that are concerned with the environment. The North emits most CFCs and greenhouse gases and has the smallest remaining biodiversity. Figure 1 supports the facts mentioned here, as does the WRI, UNEP, UNDP report World Resources: People and the Environment 1995 cited in my TREE article.
Graciela
Chichilnisky
Project on Information and Resources, Dept of Statistics, Columbia University, New York, NY 10027, USA
How effective is interdemic selection? In their recent article, Harrison and Hastings’ conclude that ‘adaptive evolution is unlikely to occur by interdemic selection’. Their inference is based on models that indicate that the potential for interdemic selection is, at best, limited. Experimental studies of group selection stand in stark contrast to these theoretical results. Compared to individual selection experiments, there have been relatively few interdemic selection experiment+8. Nevertheless, these experiments have clearly demonstrated that experimental systems respond rapidly to interdemic selection. A diverse range of organisms respond to group selection [for example, plants (Arabidopsi+, TREE wol. II,
no.
7 July 1996
CORRESPONDENCE insects (Jribo/ium)2,3-7 and vertebrates (chickens)s], and a range of different forms of interdemic selection have been investigated that include mild4-6 and intense2,3 selection by differential extinction and recolonization, as well as interdemic selection by differential migration7. Traits examined include those that appear to be properties associated solely with the individual (leaf area4, migratory behavior3.5, egg production8), and also traits that are only expressed at the population level (population size23587). In every case, these experiments have resulted in rapid responses to group selection. In those experiments where individual selection was also applied3,4, the response to interdemic selection was greater than the response to individual selection. We have an inconsistency. Theory predicts that interdemic selection is ineffective, and experiments demonstrate that it is not only effective but in many cases it is more effective than individual selection. When theory and experiment are in conflict, it is appropriate to question what is missing from the theory. The answer was suggested independently by two different investigators nearly 20 years ago. In a review, Wade9 suggested that one of the main differences between his experimental system and models of group selection is that the experimental system exhibited extensive gene interaction (epistasis and genetically based interactions among individuals), whereas models of group selection all assumed additivity. Griffinglo,ll modeled selection for yield in crop plants. He found that if he included ‘associate effects’, that is, genetically based interactions among individuals, these would evolve most effectively by interdemic selection. Thus, what is missing from models of interdemic selection is gene interaction, perhaps the most important omission being genetically based interactions among individuals. Experimental evidence emphasizes the importance of genetically based interactions among individuals in determining phenotypes. Goodnight observed a positive response to group selection and a negative response to individual selection for leaf area. Muir8, while performing interdemic selection for egg production among cages of chickens, found that the group-selected breed had higher egg production relative to commercial (i.e. individually selected) breeds when housed in groups, but relatively lower egg production when housed singly. This indicates that interdemic selection was acting on the social interactions among the chickens. These results support Griffing’slo,ll conclusion that interdemic selection will be far more effective than individual selection when there are competitive interactions. The models cited by Harrison and Hastings apparently do not agree with a body of empirical work that has been developed for over 20 years. It is perhaps time, however, that these experiments were taken seriously, and that evolutionary biologists acknowledge the importance and novel consequences of genetically based interactions among individuals.
Charles J. Goodnight Lorl Stevens Dept
of Biology, University of Vermont, 115 Marsh Life Science Building, Burlington, VT 05405-0086, USA TREE 001.
II,
no.
7 July
1996
References
size or longevity4,5. Thus, even the demographic
Harrison, S. and Hastings, A. (1996)
Trends Ecol. conditions for classic extinction-driven
Eve/.11,180-183 Wade, M.J. Craig, D.M. Goodnight, 545-558 Goodnight,
(1977) Evolution 31, 134-153 (1982) Evolution 36,271-282 C.J. (1985) Evolution 39, C.J. (1990)
Evolution 44,
1614-1624 6 Goodnight, C.J. (1990) Evolution 44, 1625-1636 7 Wade, M.J. and Goodnight. C.J. (1991) Science 253,1015-1018 8 Muir, W.M. PoultrySci. (in press) 9 Wade, M.J. (1978) Q. Rev. Biol. 53,101-114 10 Griffing, B. (1967) Aust. J. Biol. Sci. 20, 127-139 li Griffing, B. (1977) in Proceedings of the International Congress on Quantitative Genetics, August 16-21, 1976(Pollak, E., Kempthorne, 0. and Bailey, T.B., eds), pp. 413-434, Iowa State University Press I2 Goodnight, C.J. (1991) Am. Nat. 138, 342-354
Reply from S. Harrison and A. Hastings We are aware of the experimental work on Jribolium and Arabidopsis to which Goodnight and Stevens allude, and of similar work on Drosophilal, though we were regrettably unaware of the chicken study. These experiments do little to affect the conclusion that the conditions for interdemic selection are improbable in nature. Many others (e.g. Refs 2,3) have already argued on strictly theoretical grounds that interdemic selection is unlikely because of the population structure it requires-that is, small population sizes and low gene flow are required so that populations can diverge through drift, yet also enough gene flow or recolonization is needed to allow fitter populations to prevail through selective extinction or migration. That such a delicate balance may possibly be acheived in artificial experimental settings does not make it likely in natural ones. The footnotes we added to this well-established argument were based more on ecological considerations than on population genetic models. First, we argued that most natural systems characterized by frequent population turnover involve ‘weedy’ species with excellent powers of dispersal, hence little propensity to form genetically distinct local populations. Second, we noted that a long-term effect of population turnover is to deplete total genetic variation at the metapopulation level, thus reducing the potential for strong selective differences among local populations. For these two reasons we find it hard to envisage where in nature we will find metapopulations with substantial genetic differentiation among local populations as well as substantial turnover. Third, we argued that there are few good empirical examples of classic metapopulation dynamics: most spatially structured population systems that have been studied show too much migration, too little migration, or too much inequality in population
interdemic selection may be rare; of course, also required are high genetic variation among populations and appreciable selection on this variation. We are not saying this is impossible, only that nature may find it hard to keep all these balls in the air at once. Our purpose was not to add another voice to the levels of selection debate, however. We wrote our essay because ecologists are currently fascinated with the idea of metapopulation dynamics, and are aware that theory suggests that population subdivision and turnover have evolutionary implications as well. We attempted to review these implications and to direct attention to those that may be most important. We concluded that interdemic selection and ‘shifting balance’ scenarios, in which populations diverge randomly, are unlikely because of the narrow or even self-contradictory conditions of population structure that they imply. However, spatially varying natural selection in structured populations, and selection on traits directly involved in extinction and colonization (e.g. the evolution of migration rates), are examples of subjects on which recent progress has been made6-9 and that merit continued attention from theorists and empiricists. Finally, we agree with Goodnight and Stevens that the potential for interdemic selection is enhanced by interactions between the fitness value of traits and their population contexts; a point about which theorists have long been awareg. We know of no experimental work that demonstrates such effects (though we remain to be enlightened by chickens), and there is a general dearth of evidence on the existence or importance of fitness interactions in nature.
Susan Harrison Alan Hastings Division of Environmental University of California, Davis, CA 95616, USA
Studies,
References 1 Katz, A.J. and Young, S.S.Y. (1975) Genetics 81, 163-175 2 Maynard Smith, J. (1976) Q. Rev. Biol. 51, 277-283 3 Slatkin, M. (1987) Science 236, 787-792 4 Hastings, A.M. and Harrison, S. (1994) Annu. Rev. Ecol. Syst. 25, 167-188 5 Harrison, S. and Taylor, A.D. in Metapopulation Biology, Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press) 6 Olivieri, I., Couvet, D. and Gouyon, P.H. (1990) Trends Ecol. Evol. 5, 207-210 7 Hastings, A. (1991) Biol. J. Linn. Sot. 42, 57-71 8 Gyllenberg, M., Hanski, I. and Hastings, A. in Metapopulation Biology, Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press) 9 Barton, N.H. and Whitlock, M.C. in Metapopulation Biology Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press)
0 1996, Elsevier Science Ltd
299
Reed F. Noss 7310NW Acorn Ridge, Corvallis, OR 97330, USA References 1 Chichilnisky, G. (1996) Trends Ecol. Evol. 11, 135-140 2 Ehrlich, P.R., Ehrlich, A.H. and Holdren, J.P. (1970) Ecoscience: Population, Resources, Environment, W.H. Freeman
Reply from G. Chichilnisky Reed Noss ‘North’ and developing terminology. Prices and
298
calls attention to my use of the terms ‘South’ to describe industrial and countries. This is completely standard Indeed, in my section on Resource Overconsumption I define these terms:
0
1996, Elsevier Science Ltd
Year
??Industrial, 0
Developing,
fossil fuels fossil fuels
Industrial, Developing,
beef and veal beef and veal
Fig. 1.Consumption of fossil fuels and meat in industrial and developing countries, 1961-1990.
‘[O]n the whole, the situation today can be described as the industrial countries (the North) overconsuming environmental resources which are overextracted in the developing countries (the South)‘. Figure 1 illustrates this for fossil fuels. The fact is that most industrial countries are in the North and most developing countries are in the South. This is why one uses North and South to describe industrial and developing countries. The coincidence is, of course, not perfect. Australia is in the South and the World Bank ranks it today as the richest nation in the world using a new valuation which values natural resources highly. The Russian Federation is in the North and some parts of it qualify as developing countries. North and South are terms that are typically used in economics and political sciences to describe industrial and developing countries, without a tight connection to geography. Noss’s second point refers to my sentence in Box 1: ‘despite all publicity to the contrary, global environmental damage is, and has always been, inversely related to population growth’. He misread this sentence, thinking that I was mentioning publicity about ‘high population growth in the North’. I was not-just the opposite. The North’s population growth is the lowest, as I mention in my article. The South’s is the highest. My point is that much publicity goes to imply that environmental damage will disappear if only population growth in developing nations would decrease. This is incorrect today, and also in historical terms. If humans were to disappear, of course, the problem would not be there. Indeed, with the 50% decline some observe in male sperm, the ultimate solution may be within our reach. But in reality the areas of the world with lower rates of population growth are causing today, and have caused historically, the most damage to the
global environment. I refer specifically to global emissions of greenhouse gases and CFCs, and to global destruction of biodiversity. These are the two main global environmental problems, according to the United Nations Earth Summit, and according to all major international agencies that are concerned with the environment. The North emits most CFCs and greenhouse gases and has the smallest remaining biodiversity. Figure 1 supports the facts mentioned here, as does the WRI, UNEP, UNDP report World Resources: People and the Environment 1995 cited in my TREE article.
Graciela
Chichilnisky
Project on Information and Resources, Dept of Statistics, Columbia University, New York, NY 10027, USA
How effective is interdemic selection? In their recent article, Harrison and Hastings’ conclude that ‘adaptive evolution is unlikely to occur by interdemic selection’. Their inference is based on models that indicate that the potential for interdemic selection is, at best, limited. Experimental studies of group selection stand in stark contrast to these theoretical results. Compared to individual selection experiments, there have been relatively few interdemic selection experiment+8. Nevertheless, these experiments have clearly demonstrated that experimental systems respond rapidly to interdemic selection. A diverse range of organisms respond to group selection [for example, plants (Arabidopsi+, TREE wol. II,
no.
7 July 1996
CORRESPONDENCE insects (Jribo/ium)2,3-7 and vertebrates (chickens)s], and a range of different forms of interdemic selection have been investigated that include mild4-6 and intense2,3 selection by differential extinction and recolonization, as well as interdemic selection by differential migration7. Traits examined include those that appear to be properties associated solely with the individual (leaf area4, migratory behavior3.5, egg production8), and also traits that are only expressed at the population level (population size23587). In every case, these experiments have resulted in rapid responses to group selection. In those experiments where individual selection was also applied3,4, the response to interdemic selection was greater than the response to individual selection. We have an inconsistency. Theory predicts that interdemic selection is ineffective, and experiments demonstrate that it is not only effective but in many cases it is more effective than individual selection. When theory and experiment are in conflict, it is appropriate to question what is missing from the theory. The answer was suggested independently by two different investigators nearly 20 years ago. In a review, Wade9 suggested that one of the main differences between his experimental system and models of group selection is that the experimental system exhibited extensive gene interaction (epistasis and genetically based interactions among individuals), whereas models of group selection all assumed additivity. Griffinglo,ll modeled selection for yield in crop plants. He found that if he included ‘associate effects’, that is, genetically based interactions among individuals, these would evolve most effectively by interdemic selection. Thus, what is missing from models of interdemic selection is gene interaction, perhaps the most important omission being genetically based interactions among individuals. Experimental evidence emphasizes the importance of genetically based interactions among individuals in determining phenotypes. Goodnight observed a positive response to group selection and a negative response to individual selection for leaf area. Muir8, while performing interdemic selection for egg production among cages of chickens, found that the group-selected breed had higher egg production relative to commercial (i.e. individually selected) breeds when housed in groups, but relatively lower egg production when housed singly. This indicates that interdemic selection was acting on the social interactions among the chickens. These results support Griffing’slo,ll conclusion that interdemic selection will be far more effective than individual selection when there are competitive interactions. The models cited by Harrison and Hastings apparently do not agree with a body of empirical work that has been developed for over 20 years. It is perhaps time, however, that these experiments were taken seriously, and that evolutionary biologists acknowledge the importance and novel consequences of genetically based interactions among individuals.
Charles J. Goodnight Lorl Stevens Dept
of Biology, University of Vermont, 115 Marsh Life Science Building, Burlington, VT 05405-0086, USA TREE 001.
II,
no.
7 July
1996
References
size or longevity4,5. Thus, even the demographic
Harrison, S. and Hastings, A. (1996)
Trends Ecol. conditions for classic extinction-driven
Eve/.11,180-183 Wade, M.J. Craig, D.M. Goodnight, 545-558 Goodnight,
(1977) Evolution 31, 134-153 (1982) Evolution 36,271-282 C.J. (1985) Evolution 39, C.J. (1990)
Evolution 44,
1614-1624 6 Goodnight, C.J. (1990) Evolution 44, 1625-1636 7 Wade, M.J. and Goodnight. C.J. (1991) Science 253,1015-1018 8 Muir, W.M. PoultrySci. (in press) 9 Wade, M.J. (1978) Q. Rev. Biol. 53,101-114 10 Griffing, B. (1967) Aust. J. Biol. Sci. 20, 127-139 li Griffing, B. (1977) in Proceedings of the International Congress on Quantitative Genetics, August 16-21, 1976(Pollak, E., Kempthorne, 0. and Bailey, T.B., eds), pp. 413-434, Iowa State University Press I2 Goodnight, C.J. (1991) Am. Nat. 138, 342-354
Reply from S. Harrison and A. Hastings We are aware of the experimental work on Jribolium and Arabidopsis to which Goodnight and Stevens allude, and of similar work on Drosophilal, though we were regrettably unaware of the chicken study. These experiments do little to affect the conclusion that the conditions for interdemic selection are improbable in nature. Many others (e.g. Refs 2,3) have already argued on strictly theoretical grounds that interdemic selection is unlikely because of the population structure it requires-that is, small population sizes and low gene flow are required so that populations can diverge through drift, yet also enough gene flow or recolonization is needed to allow fitter populations to prevail through selective extinction or migration. That such a delicate balance may possibly be acheived in artificial experimental settings does not make it likely in natural ones. The footnotes we added to this well-established argument were based more on ecological considerations than on population genetic models. First, we argued that most natural systems characterized by frequent population turnover involve ‘weedy’ species with excellent powers of dispersal, hence little propensity to form genetically distinct local populations. Second, we noted that a long-term effect of population turnover is to deplete total genetic variation at the metapopulation level, thus reducing the potential for strong selective differences among local populations. For these two reasons we find it hard to envisage where in nature we will find metapopulations with substantial genetic differentiation among local populations as well as substantial turnover. Third, we argued that there are few good empirical examples of classic metapopulation dynamics: most spatially structured population systems that have been studied show too much migration, too little migration, or too much inequality in population
interdemic selection may be rare; of course, also required are high genetic variation among populations and appreciable selection on this variation. We are not saying this is impossible, only that nature may find it hard to keep all these balls in the air at once. Our purpose was not to add another voice to the levels of selection debate, however. We wrote our essay because ecologists are currently fascinated with the idea of metapopulation dynamics, and are aware that theory suggests that population subdivision and turnover have evolutionary implications as well. We attempted to review these implications and to direct attention to those that may be most important. We concluded that interdemic selection and ‘shifting balance’ scenarios, in which populations diverge randomly, are unlikely because of the narrow or even self-contradictory conditions of population structure that they imply. However, spatially varying natural selection in structured populations, and selection on traits directly involved in extinction and colonization (e.g. the evolution of migration rates), are examples of subjects on which recent progress has been made6-9 and that merit continued attention from theorists and empiricists. Finally, we agree with Goodnight and Stevens that the potential for interdemic selection is enhanced by interactions between the fitness value of traits and their population contexts; a point about which theorists have long been awareg. We know of no experimental work that demonstrates such effects (though we remain to be enlightened by chickens), and there is a general dearth of evidence on the existence or importance of fitness interactions in nature.
Susan Harrison Alan Hastings Division of Environmental University of California, Davis, CA 95616, USA
Studies,
References 1 Katz, A.J. and Young, S.S.Y. (1975) Genetics 81, 163-175 2 Maynard Smith, J. (1976) Q. Rev. Biol. 51, 277-283 3 Slatkin, M. (1987) Science 236, 787-792 4 Hastings, A.M. and Harrison, S. (1994) Annu. Rev. Ecol. Syst. 25, 167-188 5 Harrison, S. and Taylor, A.D. in Metapopulation Biology, Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press) 6 Olivieri, I., Couvet, D. and Gouyon, P.H. (1990) Trends Ecol. Evol. 5, 207-210 7 Hastings, A. (1991) Biol. J. Linn. Sot. 42, 57-71 8 Gyllenberg, M., Hanski, I. and Hastings, A. in Metapopulation Biology, Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press) 9 Barton, N.H. and Whitlock, M.C. in Metapopulation Biology Ecology, Genetics and Evolution (Hanski, I. and Gilpin, M.E., eds), Academic Press (in press)
0 1996, Elsevier Science Ltd
299