TREE vol. 8, no. 8, August 1993

I4 Doiman, A.J., Gash, J.H.C., Roberts, J. and 22 ldso, S.B., Alien, SC. and Choudhury. Plant Cell Environ. 13, 383-388 Shuttleworth, W.J. (1991) Agric. For. B.I. (1988) Agric. For. Meteoml. 43, 49-58 31 Meinzer, F.C., Grantz, D.A., and Smit, B. Meteorol. 54, 303-3 I8 23 Monteith, I.L. (1990) Agrk Fo‘or. ( I99 I ) Amt. I. Plant Physiol. 18, 329-338 15 Kim, J. and Verma, S.B. (1991) Agric, For Meteorol. 49, 155-167 32 Meinzer, F.C. and Grantz, D.A. ( I991 ) Meteorol. 57, 187-208 24 Kostner, B.M.M. etal. ( 1992) Oecologia Physiol. Plant. 83, 324-329 I6 Cohen, Y., Fuchs, M. and Green, G.C. 91,350-359 33 Monteith, J.L. (1988) /. Hydrol. 100, 57-68 ( I98 1) Plant Cell Environ. 4, 39 I-397 25 Aphaio, P.J. and Jarvis, P.G. (1991) Plant 34 Tyree, M.T. and Sperry, IS. 11988) Plant I7 Cermak, I., Demi, M. and Penka, M. CellEnviron. 14, 127-132 Physiol. 88, 574-580 (1973) Biol. Plant. 15, 171-178 26 Mott, K.A. and Parkhurst, D.F. (1991) 35 Meinzer, F.C., Goldstein, G., Neufeid. 18 Baker, I.M. and Van Bavel, C.H.M. (1987) Plant Cell Environ. 14, 509-5 I5 H.S., Grantz, D.A. and Crisosto, G.M. (1992) Plant Cell Environ. 10, 777-782 27 Davies, W.)., Mansfield, T.A. and Plant Cell Environ. 15, 471-477 I9 Grantz, D.A. ( 1990) P/ant Cell Environ. Hetherington, A.M. (1990) Plant Cell 36 lones, H.G. and Sutherland, R.A. (1991) I3,667-679 Environ. 13, 709-7 I9 Plant Cell Environ. 14, 607-612 20 Idso, S.B., Clawson, K.L. and Anderson, 28 Davies, W.J. and Zhang, 1. (1991) Annu. 37 Idso, S.B. (I 983) Agrk. Meteorol. 29, M.G. (1986) WaterResour. Res. 22, 1702-1716 Rev. Plant Physiol. Molec. Biol. 42, 55-76 213-217 21 Idso, S.B. (1987) Agric. For. Meteorol. 40, 29 Chapin, F.S. (1991) BioScience41, 29-36 38 Coilatz, G.J., Bail, LT., Grivet, C. and Berry, 105-106 30 Meinzer, F.C. and Grantz, D.A. (1990) LA. (I 99 I ) Agric. For. Meteorol. 54, 107-l 36

Adaptations and History historically through the action of George V. Lauder, Armand M. Leroi and Michael R. Rose natural selection for its current bio- logical role’14. This definition, or some version of it, is now widely accepted9J3r15-‘7, and for the sake The historical definition of adaptations has With this rise in comparative of brevity we refer to it as the his- come into wide use as comparative biol- studies has come an emphasis on torical definition of adaptations. ogists have applied methods of phylogen- the concept of adaptation as having Central to this definition, as to etic analysis to a variety of evolutionary an important historical component. many other (but not a11r8) defi- problems. Here we point out a number of Gould and Vrba’* codified this view nitions of adaptations, is the notion difficulties in applying historical methods by developing a terminology that that selection is the evolutionary to the study of adaptation, especially in distinguished between the current process that produced the trait cases where a trait has arisen but once. utility of a character and its histori- under consideration’9s20; traits that In particular, the potential complexity of cal origin. For Gould and Vrba, are considered to be adaptive thus the genetic correlations among phenotypic an adaptation was a trait whose evolve under selection for that traits, performance variables and fitness ‘historical genesis’ was due to the trait. makes inferring past patterns of selection same selective pressure as that This definition of adaptations from comparative data difficult. A given which currently maintains the trait. leads naturally to a research meth- pattern of character distribution may A trait whose origins in a clade were odology that aims to reveal the way support many alternative hypotheses of due to selective pressures other in which historical changes in selec- mechanism. While phylogenetic data are than those that currently maintain tive regimes or environments have limited in their ability to reveal evolution- it was labeled an ‘exaptation’. shaped the evolution of clades13,‘4. ary mechanisms, they have always been Gould and Vrba’s’* definition of A centerpiece of the method is the an important source of adaptive hypoth- adaptations has had considerable role given to measuring organismal esesand will continue to be so. influence on other workers who have performance. The method, sum- elaborated on the historical defi- marized in Box I, consists of four The last five years have witnessed nition of adaptation. For example, steps. First, historical changes in a virtual explosion in comparative Coddingtonr3 defined an adap- environmental circumstances or in studies of organismal design. Fueled tation in cladistic terms as an a ‘selective regime’r4 that might in part by the increasing accept- ‘apomorphic function promoted by have plausibly influenced the evol- ance of cladistic approaches to in- natural selection’. Furthermore, the ution of a clade are identified. It is ferring relationships among taxa, elaboration of an historical dimen- assumed that environmental charac- comparative biologists have begun sion to the concept of adaptation teristics relevant to fitness can be to attack a wide range of evolution- has led to another definitional ex- mapped onto a phylogenetic tree ary problems. Sexual dimorphism, pansion, namely, that the only much like any other character. life history, ecology, muscle function traits that should be called adap- Second, the performance attributes and insect-plant coevolution are tations are those which confer en- that might have possibly conferred but a few of the many issues that hanced performance in the current increased fitness under a given have been investigated by com- environment relative to the traits selective regime are identified. parative phylogenetic methods’-“. that preceded them evolutionarily: Third, the historical changes in or- ‘To constitute an adaptation, a ganismal traits (be they behavioral, George Lauder, Armand Leroi and Michael Rose are character must be shown to pro- morphological or physiological) that at the Dept of Ecology and Evolutionary Biology, vide current utility to the organism might have plausibly influenced such University of California, Irvine, CA 92717, USA. and to have been generated performance traits are identified.

294 0 1993, Else&v Science Publishers Ltd TREE vol. 8, no. 8, August 7993

And fourth, the relative perform- ance of these organismal traits in ancestral (plesiomorphic) and derived (apomorphic) states is measured. If the derived trait has a superior performance relative to its ancestor, and arose after historical change in the ‘selective regime’, then, and only then, is the trait an adaptation (Box I). The historical method for ana- lysing characters outlined in Box 1 promises two key advances. First, under this methodology, a trait need not have evolved numerous (phylo- genetically independent) times to be deemed an adaptation. Instead, even a trait that has evolved only once might be analysed for en- vironmental and performance cor- relates and pass the requisite tests appropriate to be labelled an adap- tation (Box 1). Second, the com- parative phylogenetic method sug- gests that one will be able to inventory characters within a clade the trait under consideration should circumstance E’), but also of A’ due to determine which traits are adap- not be genetically correlated with to the positive genetic correlation tations, which are disaptations and others that might influence fitness, of A and B. A’ would be identified which are nonaptations without and that fitness should be deter- as an adaptation given the phylo- having to measure directly the ac- mined by the performance variable genetic pattern shown in Case 2 tion of selection on each trait (Box 1). under consideration. (Box 21, under the historical defini- Our purpose here is to examine Box 2 (Case 1) illustrates these tion of adaptation. But in fact selec- the mechanistic assumptions that relations, and how they might ap- tion would not have operated di- underly the historical approach to pear on a phylogeny. In this rectly for character A’, and thus it adaptations, to illustrate potential example, the common ancestor of should not be classified as an ad- difficulties with inferring the past the three species has a particular aptation. The phylogenetic pattern action of selection solely from trait (A), performance (a), and is in this case is misleading as to the comparative data on traits, environ- subject to environment E. In the underlying mechanistic cause of ments and performance, and to in- course of cladogenesis, a change character evolution. dicate the value that such compara- in environmental circumstance (E’) A second situation in which a tive data still have in suggesting imposes a novel selection pressure phylogenetic pattern might be mis- mechanistic hypotheses and in test- for performance attribute a’ and leading is shown in Case 3 of Box 2. ing proposed scenarios. hence trait A’. Trait A’, then, is an If a single trait A is positively adaptation because its evolution is genetically correlated with two dif- The historical approach to adaptation: directly due to its superior perform- ferent performance variables (for genetic and environmental complexities ance in E’ relative to the ancestral example, locomotor endurance and Identifying adaptations using the condition (A). feeding rate), then fixation of trait method outlined in Box 1 requires But the relationship among traits, A’ may result from the fitness conse- that assumptions be made concern- performance and fitness in organ- quences of performance trait p in ing the genetic relationships among isms is rarely as simple as this2’-23. environment E’. Given the phylo- organismal traits, performance and When it is not, then application of genetic pattern produced by this fitness, as for most clades we have historical methodology to discover pattern of correlations, the histori- few data on the genetic basis of adaptations may lead to a false cal definition of adaptation would these features’6*20. First, there must conclusion. One possible way in show trait A’ to be an adaptation be positive genetic correlations which problems might arise is illus- for o’ whereas in fact it would not among the organismal trait, the trated in Case 2 of Box 2. If a par- be, since selection did not favor performance variable and fitness ticular trait under analysis, A, is performance trait a’. Rather, A’ in the taxa and environments con- positively genetically correlated evolved due to its correlation with sidered in the historical analysis. with another trait B, but is uncor- a second performance variable. Second, the genetic relationships related genetically with the per- Although in Box 2 we have sep- among the trait, performance vari- formance variable a, then selection arated the fitness of a trait from able and fitness must be simple. for trait B’ might result in the fix- measurement of its performance, in Implicit in the historical definition ation not just of B’ (due to increased practice the historical methodology of adaptations is the notion that fitness in the new environmental for uncovering adaptive traits treats 295 TREE vol. 8, no. 8, August 1993

these two attributes as largely syn- was fixed in the population, and alternative character states, and onymous: there is seldom a way to its phenotypic attributes correctly mapping them onto a phylogeny. measure fitness directly in most of identified as apomorphies in a But how much do such codings the taxa that have been analysed cladistic analysis, an analysis of the reveal about the selective forces using historical methods, and per- evolution of burst running speed acting upon a trait? Examples of formance is used as a surrogate for in this lineage would show per- selective regimes that have been fitness. However, the problem that formance to have manifestly de- used for historical analysis in- arises when performance is equated creased, thus making it a ‘disap- clude ‘terrestrial’, ‘arboreal’ and with fitness can be seen in the fol- tation’. One reply to this scenario ‘bird-pollinated”4. These selective lowing example. Consider an allele might be that since the historical regimes incorporate many environ- a which, although conferring in- definition of adaptation always ref- mental features. A transition from a creased performance on some per- erences a particular performance terrestrial to an arboreal selective formance trait, say, burst running variable for a trait, the above regime, for example, might be used speed, has many deleterious pleio- example is, indeed, truly a disap- to analyse the possible adaptive tropic effects on fitness-related tation no matter the increase in the nature of a derived morphological traits: fecundity, survivorship and population’s mean fitness. But this trait in the limb by the methods so on. Such an allele could be re- produces the awkward situation of of Box I. But the derived feature placed by another, call it a’, which labeling some of the phenotypic of limb morphology may have confers feeble performance but properties of a’ ‘adaptations’ and evolved in response to selection for beneficial pleiotropic effects on others ‘disaptations’, where what vertical climbing ability, the ability fitness-related traits, when this matters is the relative net fitness of to walk on branches, the ability to latter allele always has a higher net the alleles in question. feed in trees, or the ability to carry fitness than the former. The traits Use of the historical definition of young along branches. Each of associated with the invading allele adaptations also requires that the these different possible selective would clearly be adaptations in phylogenetic distribution of selec- factors is subsumed under the gen- the sense that their increase in fre- tive forces be identified. This is eral heading ‘arboreal selective quency was due to natural selec- done by coding different environ- regime’. In addition, as reflected in tion for that allele. Yet once a’ ments (or selection regimes’) as Case 2 of Box 2, the derived limb

296 TREE vol. 8, no. 8, August 1993

condition could have evolved as a The evolutionary inconstancy of 1137-l 156 correlated response of selection for, the genetic background bears on our 3 Hoglund, I. (I 989) Am. Nat 134, say, body size, in arboreal habitats. ability to use mutational or pheno- 72-87 4 Basoio, A. ( 1990) Science 250, SOS-8 IO Were this in fact true, then the typic manipulations to recover the 5 Bjorklund, M. (1990) I. Evol. Biol 3, derived limb trait would not be fitness and performance properties 171-183 an adaptation, no matter what the of ancestral traits. If manipulations 6 Farrell, B. and Mitter, C. (I 9901 Evolutjon relative performance tests show. designed to simulate the ancestral 44, 1389-l 403 condition of a trait are to yield 7 Harvey, P.H., Read, A.F. and Promislow, Defining the selective regime D.E. ( 1990) Oxt: Serv. Evol. Bio/. 6, 13-3 I broadly (e.g. as ‘arboreal’) precludes meaningful performance or fitness 8 Lauder, G.V. (19901 Annu. Rev. Eco/. Syst. recognition of specific selection estimates, then those manipu- 21,317-340 factors that may have been causally lations should be done against the 9 Brooks, D.R. and McLennan, D.A. (I991 I responsible for the evolution of a ancestral genetic backgound. While Phylogeny, Ecology and Behavior, University of Chicago Press trait. phylogenetic methods enable us to 10 Farrell, B., Dussourd, D. and Mitter. C. The many environmental factors, reconstruct the history of particular (1991) Am. Nat. 138,881-900 and thus selection forces, that traits and possibly even certain 1 I Harvey, P.H. and Purvis, A. (I991 I Nature might have influenced the evol- features of this background9,27, they 351,619-624 do not enable us to reconstruct the 12 Gould, S.j. and Vrba, E.S. (19821 ution of a trait also cause difficulty Paleobiology 8, 4-I 5 when it is necessary to formulate a myriad genetic and phenotypic in- I3 Coddington, LA. (1988) Cladistics 4, 3-22 performance test of derived and teractions and correlations that de- 14 Baum. D.A. and Larson, A. (1991) Syst. ancestral traits. Choice of the test termine the fitness of a trait. If such Zoo/. 40, l-18 must be matched to the hypoth- correlations among traits cannot be 15 Greene, H.W. ( 1986) Fieldiana. Zoo/. 3 I, l-12 esized selective regime. But per- recovered in extant taxa, then we 16 Mishler, B.D. (1988) in P/ant formance tests by their very nature will be unable to infer accurately Reproductive Ecology (Doust, I.L. and evaluate a specific operational hy- the relative fitness of ancestral and Doust, L.L., eds). pp. 285-306. Oxford pothesis, while the selective regime derived characters (Box 2). University Press might be only broadly defined. 17 Harvey, P.H. and Pagel, M D. ( I99 I ) The Comparative Method in Evolutionary Given the selective regime ‘ar- The role of comparative methods Biology, Oxford University Press boreality’, then which of the many Despite the limitations noted 18 Reeve, H.K. and Sherman. P.W. (1993) possible performance tests should above concerning the ability of 0. Rev. Biol. 68, l-32 be conducted on ingroup and out- comparative methods to reveal 19 Sober, E. (1984) The Nature ofSelection: Evolutionary Theory in Philosophical Focus, group taxa? If we test individuals adaptive processes, it is possible MIT Press in each taxon for their maximum to analyse mechanisms of adap- 20 Brandon, R.N. (1990) Adaptation and speed along a tree branch, and tation by directly studying the gen- Environment, Princeton University Press the actual selective force that gave etic relations and selective forces 21 Arnold, S.I. (1983) Am. Zoo/. 23, rise to the derived morphological that influence the evolution of ex- 347-361 22 Arnold, SJ. ( 1986) in Predator-Prey trait selected for increased vertical tant replicate populations2’~22~28-32. Relationships: Perspectives and Approaches climbing ability, then the results of Also, multiple evolutionary origins from the Study of Lower Vertebrates (Feder, our performance test will be irrel- of traits in correlation with an en- M.E. and Lauder, G.V., eds), pp. 157-l 79, evant to the question of adaptation vironmental state, long held as University of Chicago Press 23 Service, P.M., Hutchinson. E.W. and in response to the actual selection evidence of adaptation33r34 remain Rose, M.R. ( 1988) Evolution 39, 708-7 I6 mechanism. strong evidence that selection has 24 Clarke, G.M. and McKenzie, I.A. (1987) But it is not only the number of had a similar effect across lineages Nature 325, 345-346 possible performance attributes in- despite differences in genetic back- 25 Dykhuizen, D.E., Dean, A.M. and Hart\, fluencing fitness nor the number of ground. The comparative method D.L. ( 1987) Genetics I 15. 25-3 I 26 Lenski, R. (1988) Evolution 42, environmental factors which com- does have unique and valuable 433-440 pose a selective regime that make roles to play in the study of evol- 27 Maddison. W.P. and Maddison, D.R. reconstructing the causes of evol- ution Among other things, phylo- ( 1992) MacClade: Analysis of Phylogeny and ution so difficult. It is also that the genetic analyses can assist with the Character Evolution, Sinauer 28 Reznick, D.N. and Endler, ].A. (1982) parts of an environment interact to falsifying of adaptive hypotheses Evo/ution 36. 160-177 determine its fitness function, and which make explicit historical pre- 29 Endler, I. ( 1986) Natural Selection in the that these interactions are likely dictions of the order in which a Wild, Princeton University Press to be of daunting complexity and clade acquires evolutionary novel- 30 Reznick, D.N. and Bryga, H. (1987) subtlety. This is apparent when we ties2*4~9~‘0~35. Comparative patterns Evolution 4 I, 1370-l 385 31 Reznick, D.N. (1989) Evolution 43, consider that one of the most im- have always been an important 1285-1297 portant components of the selec- source of adaptive hypotheses36p37; 32 lones, R., Culver, D.C. and Kane, T.C. tive regime is the organism itself. they will continue to be so, even (1992) Evolution 46, 353-365 The average physiology, behavior given the difficulties of inferring 33 Futuyma. D.I. (1986) Evolutionary and morphology of a population Biology, Sinauer evolutionary mechanisms. 34 Ridley, M. (1992) Trends Ecol. Evol. 7. 37 form a unique background against 35 Sillen-Tullberg. B. II9881 Evolution 42, which novel traits are tested by 293-305 selection - a background, further- References 36 Williams, G.C. ( 19661 Adaptation and more, which itself evolves, often so I Wiley, E.O. (1981) fhylogenerics: The Natural Selection, Princeton University Theory and Practice of Phylogenetic Press as to alter profoundly the fitness Systematics, Wiley Interscience 37 Stearns, S.C. (1992) The Evolution of Life value of any given trait24-26. 2 Donoghue, M.1. (1989) Evolution 43, Histories, Oxford University Press