- Email: [email protected]
Tail length and female choice
TREE vol. 5, no. 11, November
1990
-., DARWIN’ FIRSTDREWATTENTIONto the extreme sexual dimorphism in tail length displayed by some birds, and suggested that this phenomenon was the result of intersexual selection. Over a century later, Andersson* performed an elegant experiment on the polygynous long-tailed widowbird, Euplectes progne, in which he showed that males with experimentally elongated tails had higher mating success than males having normal or reduced tail length. As females will beselected to respond to a character only if it varies among potential mates3, he investigated the coefficient of variation of this trait among males, and found it to be 9.4%. More recent studies on variation in tail length within and among populations of bird species are beginning to shed light on the evolutionary significanceoffemale preferencesfor these ornaments. In particular, a recent study of the pin-tailed whydah (Vidua macroura) has shown that females are exposed to far higher degrees of phenotypic variation for this character“. If tail length bears no relation to the viability of males, but confers only mating advantages5 on females, this trait would be expected to display extraordinary geographic variation between populations, as the equilibrium state of the joint evolution of female preference and male ornament is represented by a line of possible stable states, or lines of neutral equilibria6p7. Alatalo et aL8 investigated this possibility using museum skins to measure tail length of different populations of two species of whydahs (Vidua macroura and Vidua paradisea) and two species of flycatchers (Tyrannus savana and Tersiphone paradisi). They concluded that there were no indications of more geographical variation for this trait than for other characters subject only to natural selection. But tail length has not been shown to be the basis of sexual selection in any of these four species - the whydahs being promiscuous brood parasites, and the flycatchers being monogamous (in so far as any bird species can now be accorded this status!g). CraiglO has subsequently produced a similar analysis based on tail length of museum skins of Euplectesprogne itself. This is a better candidate, as it comprises three geographically separated populations: E. p. progne in South Africa, E. p. ansorgeiin Angola, Zaire and Zambia, and E. p. delamerei in Kenya, on which Andersson conducted his original experiments. Not only are these populations geographically separated, but they are M.I. Cherry is at the Dept of Terrestrial Vertebrate Biology at the South African Museum, PO Box 61, Cape Town 8000, South Africa. 9 ,990.
Flsei~er Sc~encr i’uhl,shers
Ltd
IUKI 0169~5347'90
TailLength andFemaleChoice M, I, Cherry accorded subspecific status. Speciation is predicted to be the eventual consequence of differences in ornament length between populations”, and Houde and Endler’s recent work on the guppy Poecilia reticulata’* gives credence to the idea that divergence of mating preferences represents a first step towards reproductive isolation between populations of this species. But again, differences in tail length between subspecies of E. progne appear to be no greater than for other measurements. These studies of museum skins*,1° yielded different coefficients of variation in tail length within populations: in V, paradisea, V. macroura, and both species of flycatchers, coefficients of variation in male tail length were significantly more variable (626%) than those for body-size characters (2-6%), whereas this was not the case for male V. orientalis or V. obtusa. Nor was it so in the three subspecies of E. progne, where the coefficient of variation in tail length was not even significantly greater in breeding males (S-9%) than in females (7-g%), compared with 26% for wing, bill and tarsus length. Does this mean that there is no discernible variation in tail length within populations, on which females can base their choice of mates? Such variation does appear to exist when it is studied in natural populations rather than museum skins. While it is possible to take account of allometric factors in studies of bird skins, there is no guarantee that museum collections contain a representative sample of age classes, and tail length has been shown to increase with age13-15. But two recent field studies have provided insight into other factors affecting tail length. In Jackson’s widowbird (E. jacksoni), a lekking species in which males perform an elaborate dance on small ‘courts’ that they construct on open grassland, tail length was shown statistically to be the single most important character used in mate choice16. It was positively related to a measure of body condition (comprising an index of body mass divided by tarsus length, both measures being standardized), suggesting that it may serve as an indication of male viability. Barnard4, in a field study of the pintailed whydah, was able to show that individual males differ in both the timing and the rate of growth of the four slim ornamental feathers, and females are consequently faced with a high degree of phenotypic variation $02 00
for this character. By correcting for seasonal variation in the ornaments of males caught at different times during the breeding season, she showed (using average growth rates) that mid-season coefficients of variation were as high as 18% and 55% for two different populations studied, compared with coefficients of variation of only 2-4% for tarsus, wing and unornamented tail feathers. But growth rates of multiply-trapped males actually varied considerably, despite the fact that all males were supplied ad libitum with commercial seed, suggesting that differential growth rates are not determined solely by varying food supply. Males that grew their long tails early were first to acquire call sites, but those with the longest tales were not the most successful at obtaining copulations 17. So, in one species at least, maintenance of an ornament over a long period may be more important in determining mating success than the maximum length attained. Experimental manipulation of tail length has been attempted in only two other species: the monogamous swallow Hirundo rustica’* and the promiscuous shaft-tailed whydah V. regialg. Both studies showed that females preferred longer tails, but only in the former case was tail length manipulated within the range of natural variation forthis trait. Both yielded lower coefficients of variation: 5.7% for V. regia, and 9.6% for H. rustica. These figures, and the value of 9.4% for E. progne, have not been corrected for seasonal variation, and therefore may well be underestimates of the phenotypic variation to which females are exposed. Returning to the question of extraordinary differences between populations predicted by the lines of neutral equilibria, Barnard4 argues that a comparison of corrected ornament lengths between her two populations of pin-tailed whydahs is not possible because the mid-season date chosen for standardization purposes differed. A really definitive study on the role of ornament length in the evolution of sexual selection remains to be done. By monitoring different populations of the same species freely supplied with food, it should be possible to determine whether male ornament length evolves to a single point, or a set of points representing neutral equilibria. In the guppy Poecilia reticulata, female preference for orange colouration varies among populations, depending on the extent to which this 349
TREE vol. 5, no. II, November
trait is expressed in males12. Similarly, if male tail length differs among bird populations, females could have a preference for long tails that varies in accordance with the expression of the male trait. But female preferences may not simply be for males with the longest tails; rather, they may be for males with fast-growing ornaments, or for males that can maintain the ornaments over a period of time. References 1 Darwin, C. (1871) The DescentofMan, and Selection in Relation to Sex, John Murray
2 Andersson, M. (1982) Nature 299, 818-820 3 Searcy, W.A. (1982) Annu. Rev. Ecol. Syst. 13,57-86 4 Barnard, P. Biol. J. Linn. Sot. 40 (in press) 5 Fisher. R.A. (1930) The Genetical Theory of Nat&a/ Selection, Clarendon Press 6 Lande, R. (1981) Proc. Nat/ Acad. Sci. USA 78,3712-3725 7 Lande, R. (1982) Evolution 30, 253-269 8 Alatalo, R. V., Hogland, J. and Lundberg, A. (1988) Biol. J. Linn. Sot. 34, 363-374 9 Harvey, P.H. and May, R.M. (1989) Nature 337,508-509
TheDecline oftheNewfoundland Crossbill Stuart L. Pimm AN ENDEMIC SUBSPECIES of the red crossbill, Loxia curvirosfra percna, underwent a dramatic decline on the island of Newfoundland (Canada) in the early 1970s. Its numbers have been reduced to the point where its extinction is quite probable, although it may survive on some nearby smaller islands. In contrast, the white-winged crossbill (Loxia leucoptera) appears to have maintained its numbers. Almost certainly, the cause of the red crossbill’s decline was the introduction of the red squirrel (Tamiasciurus hudsonicus) to Newfoundland. The introduction of alien species as the cause of the demise of native species is all too familiar (see Refs 1 and 2). What is special about Benkman’s work on this decline3, and on crossbills in general-, is that it provides insight into why island species are so vulnerable to introduced species. The circumstantial evidence implicating the red squirrel in the decline of the red crossbill is quite strong. First, the timing fits. Red squirrels were introduced to Newfoundland in 1963 and 1964 by the Newfoundland Wildlife Service, to provide prey for the native pine marten (Martes americana) which had been trapped to low numbers. The
Stuart Pimm is at the Dept of Zoology and Graduate Program in Ecology, The University of Tennessee, Knoxville, TN 37996, USA, and currently at The Centre for Population Biology, Imperial College at Silwood Park, UK. 350
squirrels quickly achieved high densities, and are now perhaps twice as dense in forests on Newfoundland as they are on the mainland. Second, the current distributions are nonoverlapping; the squirrels have not reached the small offshore islands, the only places where the crossbills are now known to occur. Third, the mechanism of competition for food seems plausible. Red crossbills once occupied, and red squirrels now occupy, black spruce (Picea mariana) forests in Newfoundland. In contrast, the white-winged crossbill feeds more on white spruce (Picea glauca) and tamarack (Larix laricina). Interestingly, red crossbills in mainland areas typically feed on various pines, rather than spruces, but pines are relatively uncommon in Newfoundland. While examples of introduced species causing extinctions of native species are common, examples providing convincing evidence for competition (particularly competition between species from such different the evitaxa) are few le7. Certainly, dence for competitive replacement that Benkman presents is indirect, as it almost certainly must be for changes on such a large spatial scale. (The evidence does not preclude a mechanism involving a shared predator or parasite; conceivably, the squirrel and the crossbill could share a pathogen to which the squirrels but not the crossbills are relatively immune.) Island communities are often damaged by introduced species’.
1990
10 Craig, A.J.F.K. (1989) S. Afr. J. SC;. 85, 523-524 11 Arnold, S.J (1985) Experientia 41, 12961310 12 Houde, A.E. and Endler, J.A. (1990) Science 23 1,1405-l 408 13 Verheyen, R. (1956) Gerfaur 46, 65-74 14 Smith, L.H. (1965) Science 147. 510-513’ 15 Smith, L.H. (1982) Aust. J. Wild/. Res. 9,31 l-330 16 Andersson, S. (1989) Behav. Ecol. Sociobiol. 25, 403-410 17 Barnard, P. (1989) Ostrich 60, 103-117 18 Melter, A.P. (1988) Nature 332,64@-642 19 Barnard, P. (1990) Anim. Behav. 39, 652-656
There are two main reasons for this. First, as is well known, they are typically species poor, and so may be more likely to be invaded than species-rich communities. The more introduced species, the more likely that one of those species will do harm. Second, and more controversially, the chances that any one introduced species will cause marked changes in the abundance of native species is greater on islands than elsewhere’J. Islands often lack generalized herbivores or carnivores, which when introduced without their respective predators and diseases may be expected to wreak the same havoc on lower trophic levels as they would in continental communities if those predators and diseases were removed. Also, island herbivores and carnivores might be more trophically specialized, and so more vulnerable to the loss of their food resources, than trophically more generalized mainland species that can switch to other resources. This is a difficult idea to test, but Benkman’s results provide a compelling explanation of how the crossbills might have evolved to be vulnerable. Benkman argues that the historical absence of red squirrels from Newfoundland has had a direct evolutionary effect on the cones of the black spruce, and a secondary evolutionary effect on the crossbills. The key morphological variable is the thickness of the cone scales, because this correlates with the time required by crossbills to remove seeds from closed cones. Benkman suspected that the black spruce cone scales would be thinner on Newfoundland, because of the absence of the squirrels. If so, seeds in the closed cones would be more accessible to crossbills on Newfoundland than on the mainland.
1990
-., DARWIN’ FIRSTDREWATTENTIONto the extreme sexual dimorphism in tail length displayed by some birds, and suggested that this phenomenon was the result of intersexual selection. Over a century later, Andersson* performed an elegant experiment on the polygynous long-tailed widowbird, Euplectes progne, in which he showed that males with experimentally elongated tails had higher mating success than males having normal or reduced tail length. As females will beselected to respond to a character only if it varies among potential mates3, he investigated the coefficient of variation of this trait among males, and found it to be 9.4%. More recent studies on variation in tail length within and among populations of bird species are beginning to shed light on the evolutionary significanceoffemale preferencesfor these ornaments. In particular, a recent study of the pin-tailed whydah (Vidua macroura) has shown that females are exposed to far higher degrees of phenotypic variation for this character“. If tail length bears no relation to the viability of males, but confers only mating advantages5 on females, this trait would be expected to display extraordinary geographic variation between populations, as the equilibrium state of the joint evolution of female preference and male ornament is represented by a line of possible stable states, or lines of neutral equilibria6p7. Alatalo et aL8 investigated this possibility using museum skins to measure tail length of different populations of two species of whydahs (Vidua macroura and Vidua paradisea) and two species of flycatchers (Tyrannus savana and Tersiphone paradisi). They concluded that there were no indications of more geographical variation for this trait than for other characters subject only to natural selection. But tail length has not been shown to be the basis of sexual selection in any of these four species - the whydahs being promiscuous brood parasites, and the flycatchers being monogamous (in so far as any bird species can now be accorded this status!g). CraiglO has subsequently produced a similar analysis based on tail length of museum skins of Euplectesprogne itself. This is a better candidate, as it comprises three geographically separated populations: E. p. progne in South Africa, E. p. ansorgeiin Angola, Zaire and Zambia, and E. p. delamerei in Kenya, on which Andersson conducted his original experiments. Not only are these populations geographically separated, but they are M.I. Cherry is at the Dept of Terrestrial Vertebrate Biology at the South African Museum, PO Box 61, Cape Town 8000, South Africa. 9 ,990.
Flsei~er Sc~encr i’uhl,shers
Ltd
IUKI 0169~5347'90
TailLength andFemaleChoice M, I, Cherry accorded subspecific status. Speciation is predicted to be the eventual consequence of differences in ornament length between populations”, and Houde and Endler’s recent work on the guppy Poecilia reticulata’* gives credence to the idea that divergence of mating preferences represents a first step towards reproductive isolation between populations of this species. But again, differences in tail length between subspecies of E. progne appear to be no greater than for other measurements. These studies of museum skins*,1° yielded different coefficients of variation in tail length within populations: in V, paradisea, V. macroura, and both species of flycatchers, coefficients of variation in male tail length were significantly more variable (626%) than those for body-size characters (2-6%), whereas this was not the case for male V. orientalis or V. obtusa. Nor was it so in the three subspecies of E. progne, where the coefficient of variation in tail length was not even significantly greater in breeding males (S-9%) than in females (7-g%), compared with 26% for wing, bill and tarsus length. Does this mean that there is no discernible variation in tail length within populations, on which females can base their choice of mates? Such variation does appear to exist when it is studied in natural populations rather than museum skins. While it is possible to take account of allometric factors in studies of bird skins, there is no guarantee that museum collections contain a representative sample of age classes, and tail length has been shown to increase with age13-15. But two recent field studies have provided insight into other factors affecting tail length. In Jackson’s widowbird (E. jacksoni), a lekking species in which males perform an elaborate dance on small ‘courts’ that they construct on open grassland, tail length was shown statistically to be the single most important character used in mate choice16. It was positively related to a measure of body condition (comprising an index of body mass divided by tarsus length, both measures being standardized), suggesting that it may serve as an indication of male viability. Barnard4, in a field study of the pintailed whydah, was able to show that individual males differ in both the timing and the rate of growth of the four slim ornamental feathers, and females are consequently faced with a high degree of phenotypic variation $02 00
for this character. By correcting for seasonal variation in the ornaments of males caught at different times during the breeding season, she showed (using average growth rates) that mid-season coefficients of variation were as high as 18% and 55% for two different populations studied, compared with coefficients of variation of only 2-4% for tarsus, wing and unornamented tail feathers. But growth rates of multiply-trapped males actually varied considerably, despite the fact that all males were supplied ad libitum with commercial seed, suggesting that differential growth rates are not determined solely by varying food supply. Males that grew their long tails early were first to acquire call sites, but those with the longest tales were not the most successful at obtaining copulations 17. So, in one species at least, maintenance of an ornament over a long period may be more important in determining mating success than the maximum length attained. Experimental manipulation of tail length has been attempted in only two other species: the monogamous swallow Hirundo rustica’* and the promiscuous shaft-tailed whydah V. regialg. Both studies showed that females preferred longer tails, but only in the former case was tail length manipulated within the range of natural variation forthis trait. Both yielded lower coefficients of variation: 5.7% for V. regia, and 9.6% for H. rustica. These figures, and the value of 9.4% for E. progne, have not been corrected for seasonal variation, and therefore may well be underestimates of the phenotypic variation to which females are exposed. Returning to the question of extraordinary differences between populations predicted by the lines of neutral equilibria, Barnard4 argues that a comparison of corrected ornament lengths between her two populations of pin-tailed whydahs is not possible because the mid-season date chosen for standardization purposes differed. A really definitive study on the role of ornament length in the evolution of sexual selection remains to be done. By monitoring different populations of the same species freely supplied with food, it should be possible to determine whether male ornament length evolves to a single point, or a set of points representing neutral equilibria. In the guppy Poecilia reticulata, female preference for orange colouration varies among populations, depending on the extent to which this 349
TREE vol. 5, no. II, November
trait is expressed in males12. Similarly, if male tail length differs among bird populations, females could have a preference for long tails that varies in accordance with the expression of the male trait. But female preferences may not simply be for males with the longest tails; rather, they may be for males with fast-growing ornaments, or for males that can maintain the ornaments over a period of time. References 1 Darwin, C. (1871) The DescentofMan, and Selection in Relation to Sex, John Murray
2 Andersson, M. (1982) Nature 299, 818-820 3 Searcy, W.A. (1982) Annu. Rev. Ecol. Syst. 13,57-86 4 Barnard, P. Biol. J. Linn. Sot. 40 (in press) 5 Fisher. R.A. (1930) The Genetical Theory of Nat&a/ Selection, Clarendon Press 6 Lande, R. (1981) Proc. Nat/ Acad. Sci. USA 78,3712-3725 7 Lande, R. (1982) Evolution 30, 253-269 8 Alatalo, R. V., Hogland, J. and Lundberg, A. (1988) Biol. J. Linn. Sot. 34, 363-374 9 Harvey, P.H. and May, R.M. (1989) Nature 337,508-509
TheDecline oftheNewfoundland Crossbill Stuart L. Pimm AN ENDEMIC SUBSPECIES of the red crossbill, Loxia curvirosfra percna, underwent a dramatic decline on the island of Newfoundland (Canada) in the early 1970s. Its numbers have been reduced to the point where its extinction is quite probable, although it may survive on some nearby smaller islands. In contrast, the white-winged crossbill (Loxia leucoptera) appears to have maintained its numbers. Almost certainly, the cause of the red crossbill’s decline was the introduction of the red squirrel (Tamiasciurus hudsonicus) to Newfoundland. The introduction of alien species as the cause of the demise of native species is all too familiar (see Refs 1 and 2). What is special about Benkman’s work on this decline3, and on crossbills in general-, is that it provides insight into why island species are so vulnerable to introduced species. The circumstantial evidence implicating the red squirrel in the decline of the red crossbill is quite strong. First, the timing fits. Red squirrels were introduced to Newfoundland in 1963 and 1964 by the Newfoundland Wildlife Service, to provide prey for the native pine marten (Martes americana) which had been trapped to low numbers. The
Stuart Pimm is at the Dept of Zoology and Graduate Program in Ecology, The University of Tennessee, Knoxville, TN 37996, USA, and currently at The Centre for Population Biology, Imperial College at Silwood Park, UK. 350
squirrels quickly achieved high densities, and are now perhaps twice as dense in forests on Newfoundland as they are on the mainland. Second, the current distributions are nonoverlapping; the squirrels have not reached the small offshore islands, the only places where the crossbills are now known to occur. Third, the mechanism of competition for food seems plausible. Red crossbills once occupied, and red squirrels now occupy, black spruce (Picea mariana) forests in Newfoundland. In contrast, the white-winged crossbill feeds more on white spruce (Picea glauca) and tamarack (Larix laricina). Interestingly, red crossbills in mainland areas typically feed on various pines, rather than spruces, but pines are relatively uncommon in Newfoundland. While examples of introduced species causing extinctions of native species are common, examples providing convincing evidence for competition (particularly competition between species from such different the evitaxa) are few le7. Certainly, dence for competitive replacement that Benkman presents is indirect, as it almost certainly must be for changes on such a large spatial scale. (The evidence does not preclude a mechanism involving a shared predator or parasite; conceivably, the squirrel and the crossbill could share a pathogen to which the squirrels but not the crossbills are relatively immune.) Island communities are often damaged by introduced species’.
1990
10 Craig, A.J.F.K. (1989) S. Afr. J. SC;. 85, 523-524 11 Arnold, S.J (1985) Experientia 41, 12961310 12 Houde, A.E. and Endler, J.A. (1990) Science 23 1,1405-l 408 13 Verheyen, R. (1956) Gerfaur 46, 65-74 14 Smith, L.H. (1965) Science 147. 510-513’ 15 Smith, L.H. (1982) Aust. J. Wild/. Res. 9,31 l-330 16 Andersson, S. (1989) Behav. Ecol. Sociobiol. 25, 403-410 17 Barnard, P. (1989) Ostrich 60, 103-117 18 Melter, A.P. (1988) Nature 332,64@-642 19 Barnard, P. (1990) Anim. Behav. 39, 652-656
There are two main reasons for this. First, as is well known, they are typically species poor, and so may be more likely to be invaded than species-rich communities. The more introduced species, the more likely that one of those species will do harm. Second, and more controversially, the chances that any one introduced species will cause marked changes in the abundance of native species is greater on islands than elsewhere’J. Islands often lack generalized herbivores or carnivores, which when introduced without their respective predators and diseases may be expected to wreak the same havoc on lower trophic levels as they would in continental communities if those predators and diseases were removed. Also, island herbivores and carnivores might be more trophically specialized, and so more vulnerable to the loss of their food resources, than trophically more generalized mainland species that can switch to other resources. This is a difficult idea to test, but Benkman’s results provide a compelling explanation of how the crossbills might have evolved to be vulnerable. Benkman argues that the historical absence of red squirrels from Newfoundland has had a direct evolutionary effect on the cones of the black spruce, and a secondary evolutionary effect on the crossbills. The key morphological variable is the thickness of the cone scales, because this correlates with the time required by crossbills to remove seeds from closed cones. Benkman suspected that the black spruce cone scales would be thinner on Newfoundland, because of the absence of the squirrels. If so, seeds in the closed cones would be more accessible to crossbills on Newfoundland than on the mainland.