Heredity 84 (2000) 623±629 Received 6 March 1999, accepted 11 April 2000

Short Review The of maladaptation

BERNARD J. CRESPI* Department of Biological Science, Simon Fraser University, 8888 University Drive, Burnaby BC V5A 1S6, Canada

This review contains a description of a research program for using studies of and teleonomy; (2) deter- the study of maladaptation, de®ned here in terms of deviation mination of the causes of the deviation, using analyses of from adaptive peaks. Maladaptation has many genetic causes, genetics, development, or other methods. Conditions for including , inbreeding, drift, gene ¯ow, heterozygote unambiguously identifying maladaptation are considerably advantage and pleiotropy. Degrees of maladaptation are more stringent than those for demonstrating and determined by genetic architecture and the relationship remarkably few studies have clearly identi®ed and character- between the rates of selective, environmental change and the ised maladaptative traits. A thorough understanding of the nature and extent of genetic responses to selection. The nature of phenotypic variation will never be achieved without empirical analysis of maladaptation requires: (1) recognition of an analysis of the scope and usual causes of maladaptation. putative maladaptation, using methods from , teleonomy, development and genetics, followed by an assess- Keywords: adaptation, adaptive peaks, evolution, maladap- ment of the nature and degree of deviation from adaptation, tation.

`But by far the most important consideration is that the chief `maladaptation' and discussed the usefulness of these de®ni- part of the organization of every being is simply due to tions for addressing various questions. Secondly, I have inheritance; and consequently, though each being assuredly is discussed the causes of maladaptation at di€erent levels in the well ®tted for its place in nature, many structures now have no hierarchy of biological information (Arnold, 1992). Thirdly, I direct relation to the habits of life of each species'. Darwin have focused on methods for identifying and analysing (1859; p. 199). maladaptation, drawing on approaches that have developed independently in di€erent ®elds and seeking to reconcile the often-acrimonious opinions of geneticists and ecologists Introduction regarding the usefulness of their research methods. The study of phenotypic adaptation and its genetic basis is central to evolutionary . The term `adaptation' has accumulated myriad de®nitions (reviews in Reeve & Sher- De®ning adaptation and maladaptation man, 1993; Rose & Lauder, 1996) but are forms Our primary criterion for choosing a de®nition should be its of traits that are always construed as the result of natural utility for answering questions of interest. De®nitions of selection, whereas individuals with variant traits that are less adaptation can be categorized into four main types: (1) well ®t to the environment exhibit lower reproductive success. teleonomic; (2) phylogenetic; (3) population genetic; (4) quan- However, the ®t of traits to environments epitomized in the titative genetic. term adaptation (`®t for', from the Greek `ad aptos') can Teleonomic de®nitions of adaptation developed in evolu- never be perfect, partly because organisms are always tionary and behavioural ecology (Thornhill, 1990) and focus adapted to at least one generation in the past. Thus, some on the functional design of phenotypic traits ± how they have degree of deviation from the maximal possible degree of been `designed' by the blind watchmaker of selection to adaptation is always expected. Such deviations have been `function' in some environmental context. These de®nitions analysed under a variety of rubrics, using the tools of emphasize the selective maintenance of traits, and involve population and quantitative genetics, developmental biology, identi®cation and quanti®cation of the ®t between form and behavioural and evolutionary ecology. The purpose of and function (Reeve & Sherman, 1993). The implementation this review is to synthesize perspectives and information from of teleonomic de®nitions into a research program requires these disparate disciplines and to analyse the nature and speci®cation of a strategy set (a range of possible trait forms), causes of maladaptation. Firstly, I have categorized de®ni- an application of some ®tness criterion (what is maximized, tions of `adaptation', translated them into de®nitions of such as performance of some task or a component of ®tness) and deliniation of constraints (®xed parameters that bound *Correspondence. Tel.: +1 604 291 3533, Fax: +1 604 291 3496, the analyses and link the strategy set with the ®tness E-mail: [email protected] criterion). This research program is based on the premise

Ó 2000 The Genetical Society of Great Britain. 623 624 B. J. CRESPI

that a history of natural selection leads to forms of traits subject to unfavourable environmental change, or (2) the (adaptations) that are optimal in a given environmental degree to which the ®tness of the current genotype lags be- context, within the range of the strategy set. Quanti®cation of hind the optimal genotype in a changing environment (see current selection is not necessary under this program, because also Kirkpatrick, 1996). Gillespie (1991; p. 63 and p. 305) current selection need not re¯ect the selective pressures concludes that such loads are often heavy as populations are 1during the trait's evolution (Thornhill, 1990). This optimiza- usually far from allelic equilibrium, apparently; this is because tion approach to adaptation has been useful in characterising genotypic adaptive peaks outrun responses to selection. the nature and causes of associations between traits, and Load-based metrics of maladaptation require estimating the between traits and environments, especially for aspects of relation between alleles or genotypes and ®tness, and where behaviour and life history. feasible they provide a strong link from maladaptation to its Under a teleonomic research program, `maladaptation' can causes. be de®ned as prevalence in a population of a `strategy' (a form Quantitative geneticists normally discuss adaptation in the of a phenotype), that does not lead to the highest relative context of phenotypic adaptive topologies, where local and ®tness of the strategies in the allowed set. This viewpoint has global peaks represent optimal population states (Schluter & often been dicult to implement, due to the diculties 2Nychka, 1994). For quantitative traits, approaches to these involved in de®ning a complete and accurate set of strategies peaks are governed by the form and strength of multivariate and constraints, the complex selective pressures on many traits selection and response to selection which can be predicted (at and the challenge of measuring ®tness in an evolutionari- least in the short term) using the genetic variance-covariance ly meaningful way (Lewontin, 1979). Moreover, if a formerly matrix (Shaw et al., 1995). By this perspective, maladaptation neutral trait (e.g. red vs. yellow ¯owers) comes under selection, can be quanti®ed as the distance of a population from the such that plants with red ¯owers have higher reproductive nearest adaptive peak (Loeschcke, 1987; Bjorklund, 1996). success (e.g. their pollen fertilizes more ovules), then red This distance is largely a function of the degree to which the ¯owers will be called an adaptation under some teleonomic population does not exactly track the vector of directional de®nitions (e.g. Reeve & Sherman, 1993), even before there has selection, as a result of aspects of genetic response to selection been any genetic response to selection. that prevent the largest possible selection-driven step up- Phylogenetic de®nitions of adaptation require the use of a wards. Stabilizing selection serves to de®ne the optimal, phylogeny to infer the origin of a trait, inference of the `adapted' state or peak, though a population may spread `selective regime' under which the trait arose, and an analysis more or less widely down its sides. As we can measure of performance of the trait under its ancestral and current selection in the ®eld, construct adaptive surfaces and estimate selective regime (Baum & Larson, 1991 and references G in the ®eld or laboratory, this measure of maladaptation is therein). If the trait arose under its current selective regime feasible to employ. Moreover, it provides a bridge between and exhibits higher performance than its antecedent, then it is genetic mechanisms of microevolutionary change, notably considered an adaptation. This de®nition focuses on joint additive genetic variance, pleiotropy, and linkage disequilib- analyses of the origin and maintenance of traits, under the rium, and aspects of ecology, as depicted in adaptive presumption that traits changing in function over the speci®c landscapes. To the extent that adaptive topologies move time period considered should be categorized di€erently from across generations, or populations are displaced downhill by traits that do not. Baum & Larson (1991) provide explicit genetic happenstance, populations will be o€ their peaks and criteria for identifying maladaptation under this phylogenetic thus, to some quanti®able degree, maladapted. perspective; a trait is maladaptive (a.k.a. `disadapted', in their I have purposefully avoided the term `constraint' in the lexicon) if it exhibits lower utility (performance at some task) exposition above as most authors use it in a heuristic and than its antecedent state, in its selective regime (environmen- general sense to refer to deviations from some expected course tal context). This research programme for categorizing traits of evolution (e.g. Maynard Smith et al., 1985; Antonovics & has yet to take hold, apparently due to uncertainties of 3van Tienderen, 1991; Pigliucci & Kaplan, 2000). My goal is to inferring ancestral states and selective regimes (Leroi et al., describe a research program for analysing a speci®c type of 1994) and a stronger focus on the importance of testing such constraint, one that can be estimated and dissected statistically for adaptation via quantifying convergence empirically. To do so, one must draw insight into maladap- (Doughty, 1996). tation from each of the four perspectives above (Stearns, Population and quantitative genetic perspectives on adap- 1984). tation involve: (1) relating alleles and genotypes to phenotypes and ®tness in current populations; or (2) quantifying current Causes of maladaptation selection, and expected or observed responses to selection, on single-locus and polygenic traits. For population geneticists, The ultimate causes of maladaptation are aspects of genetic adaptation involves gene substitutions driven by selection, or systems in relation to changes in environments. These include the maintenance of variation via selection. In the case of gene such processes as mutation, drift, inbreeding, selection, pleio- substitution, the degree of adaptation and maladaptation can tropy, linkage disequilibrium, heterozygote advantage and be quanti®ed as the `substitutional load' or the `lag load' (e.g. gene ¯ow. Most are maladaptive or non-adaptive Maynard Smith, 1976), measures of: (1) the reproductive because their e€ects are independent of adaptive signi®cance excess required to prevent extinction of a small population and traits of organisms are normally reasonably well adapted

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629. EVOLUTION OF MALADAPTATION 625

(Orr, 1998). Maladaptation may also be caused by a lack of Methods to identify and analyse maladaptation sucient genotypic variation for phenotypes to maximally respond to selection. Such paucity of variation may be due to Our maladaptationist research program has two main com- drift, inbreeding, past directional selection, or a low rate of ponents: (1) recognition and quanti®cation of putative mutation. For example, drift and inbreeding remove popula- maladaptation; and (2) determination of maladaptation's tions from genotypic adaptive peaks and thus may lead to causes. As described below, putative maladaptation can be phenotypic maladaptation. Although years of arti®cial and recognized using information from phylogenetics, develop- strong selection experiments on single traits attest to high mental processes, teleonomy and optimality theory and levels of variation maintained in the short term for most traits genetics. Deviations from adaptation can then be quanti®ed and species, the relevance of these experiments to nature directly using studies of selection and selection response, or remains uncertain (Harshman & Ho€mann 2000). This is in quanti®ed indirectly via optimality studies, each of which can part because natural and arti®cial selection may often di€er provide information about directions and distances from profoundly in their targets, strength and consequences. Pleio- adaptive peaks. Finally, to demonstrate maladaptation and tropy, whereby genes a€ect multiple traits, is considered a exclude adaptive alternative hypotheses, it is essential to near-universal mode of gene action and it may facilitate determine its causes using information from population maladaptation by impeding the joint optimization of several genetics, quantitative genetics, developmental mechanisms, traits (Barton, 1990). Linkage disequilibrium, due to linkage, or other approaches. drift, or selection, similarly leads to more-or-less severe Methods from teleonomics and optimality, phylogenetics, limitations of genetic e€ects on phenotypes. Heterozygote genetics, and development play di€erent roles in the analysis of advantage provides a third example of intrinsic properties of maladaptation. Teleonomic and optimality approaches usually genetic systems that cause deviations from maximal popula- make quantitative predictions regarding phenotypes and these tion adaptation. Finally, gene ¯ow between di€erently adapted predictions usually more or less succeed and more or less fail. populations may also lead to maladaptation, the degree of When faced with deviation from optimality predictions we can which will depend on migration rates and selection intensities either hold tight to our optimality paradigm, inspect the nature (Slatkin, 1985). of the deviations and question our strategy set, constraints, Quantitative-genetic methods allow estimation of both and ®tness criterion, or we can allow for the possibility that additive genetic variances, which probe the expression of our phenotype has not been optimized (Orzack & Sober, genetic variation in a given environment and also genetic 1994). A strategy set can, however, always be expanded to correlations, which are due to pleiotropy and linkage include formerly `maladaptive' phenotypes, which in e€ect disequilibrium. A low level of additive genetic variation, or negates the possibility of maladaptation and equates natural a substantial genetic correlation, may indicate the possibility selection with adaptation (Rose et al., 1987). The main of maladaptation with regard to the traits involved (Price & usefulness of teleonomy and optimality approaches is ®rstly, 4Langen, 1992). Aspects of developmental mechanisms (a.k.a. that they can generate the trade-o€ curves that describe the `developmental constraints') can be depicted by matrices of relationships between complexly integrated traits and ®tness, genetic variances and covariances (Cheverud, 1984), although which no other approach, including quantitative genetics, can G is always environment-speci®c and may not capture do (Partridge & Sibly, 1991). Secondly, cycles of observation, essential elements of life-history or other tradeo€s (Clark, modelling and experiments can identify the causes and objects 1987; Houle, 1991; Partridge & Sibly, 1991). In our adaptive- of selection and their adaptive (environmental) signi®cance. peak depiction of maladaptive evolution, mutation moves Deviations from optimality predictions may then direct us populations downhill and a lack of variation strands them on towards maladaptation. The main objection of many research- a slope, or slows their climb. Drift and inbreeding normally ers to optimality is that this latter road is not taken; genetically drive populations downhill. Genetic correlations may accel- based causes of deviation from optimality are seldom consid- erate uphill movement or cause curved trajectories (Arnold, ered, as the space between prediction and observation can 1992), and gene ¯ow between multiple local adaptive peaks always be ®lled with ad hoc explanations. This is a valid pulls populations down towards some centre of gravity (see complaint but it does not vitiate the tremendous successes of also Fear & Price, 1998). Selection, however, de®nes the teleonomy and optimality methods. landscape on which populations move, and to the extent that Phylogenetic approaches to analysing adaptation provide environments and selective pressures change our landscape a vital long-term temporal dimension to data via analysis of becomes a sea where waves rise, sink and move like water in convergence between traits and between traits and aspects a bath. If populations are normally on or near a crest, of their environments, or also via analysis of evolutionary movements will tend to displace them downward ± Fisher's trajectories along speci®c lineages. Convergence analyses (1958) constant deterioration of the environment, due to both (e.g. independent contrasts, see Doughty, 1996) entail abiotic and biotic causes. The degree of maladaptation of macroevolutionary tests for functional relations and strong populations thus depends on rates of change in selective deviations from expected relationships, or invariance in surfaces, in relation to rates of genetic and phenotypic change traits in particular clades (Stearns, 1984), may be indicative 5(Kirkpatrick, 1996). But how can one empirically capture this of maladaptation. Inference of evolutionary trajectories may complex scene and put a maladaptationist research program be used to identify apparent temporal lags between the into practice? appearance of a selective pressure and evolutionary response

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629. 626 B. J. CRESPI

Table 1 Examples of apparent and putative maladaptations. `Apparent causes' include both ultimate and proximate (genetic) factors

Trait Description Apparent cause Reference Large fruits, poorly Large fruits are not dispersed, Large mammalian Janzen & Martin (1982) dispersed rot under parent dispersers went extinct Wood duck egg Wood ducks engage Novel environment Reeve & Sherman (1993) dumping frequency in extremely high levels (nest sites much of egg dumping when less hidden) using obvious nest boxes Fig wasps sex ratios Wasp species under weaker Weak selection Herre (1987) sex-ratio selection show larger deviations from optimality Spider behaviour Spiders in riparian zone exhibit Gene ¯ow from Riechert (1993) nonoptimal behaviour surrounding desert Lack of paternal care Duck species exhibit lack Lag in response to selection? Johnston et al. (1999) of male care where it is expected Tit clutch sizes Clutch sizes are maladapted Gene ¯ow Dhondt et al. (1990) to local habitat Transference Females exhibit aspects High positive genetic Muma & Weatherhead of male secondary correlation between sexes (1989) sexual traits Vestigial traits Nonfunctional traits, High genetic correlations? Fong et al. (1995) present in functional form in ancestors Lack of disease Selection for disease resistance High levels of sel®ng Parker (1991) resistance but no response prevent independent evolution of traits Gall form Some species exhibit gall Lag in response to selection? Crespi & Worobey (1998) form of recent ancestors, and mismatch between gall form and life history Thorax morphology Large thoraces in wingless High genetic correlation? Crespi (1988) males that developed in environment where winged males developed Toes in dogs Large breeds of dogs often Genetic correlation? Alberch & Gale (1985) have an extra toe

in traits (e.g. Crespi & Worobey, 1998; Johnston et al., approaches can help determine if, and how, G changes in 1999) (Table 1). Phylogenies can also be used to infer the short and long-term (Shaw et al., 1995). They can also evolutionary change in the aspects of genetic systems that help determine whether any observed changes are due to can produce maladaptation. For example, phylogenetic selection, drift, or both, if drift causes only proportional

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629. EVOLUTION OF MALADAPTATION 627 changes in G, and to what extent the short and long-term traits be selected for in females? Some aspects of allometry evolution of species is directed by the main axis of multitrait appear to be modi®able by short-term selection, but others do variation (Schluter, 1996). Phylogenies are thus primarily not (Wilkinson, 1993; Emlen, 1996). Analysis of developmen- useful in that they may direct us towards putative maladap- tal mechanisms should allow us to assess if these latter traits tations, which are then dissected using other means. are intrinsically resistant to change; alternatively, selection Data on genetics and phenotypic selection are necessary may not be acting upon them at all. for the analysis of maladaptation under our chosen de®nition Using our conception of maladaptation, we can generate a of the term. Whereas teleonomic and optimality approaches list of cases of putatively maladaptive traits (Table 1), that help to identify the most selectively relevant aspects of traits have been subjected to analysis by various of the four and environments (i.e. the causes and objects of selection) approaches described above. The list is short and a surprising and characterize the direction and extent of deviation from proportion of the studies have been published in top journals. predicted optima, quanti®cation of selection and anticipated Some studies have put two of the approaches together (e.g. or observed response to selection are required for an Alberch & Gale, 1985) but none have used three or four, explicitly adaptive-peak perspective on trait evolution. Path- perhaps because of the independence of researchers using analytic methods allow for the joint analysis of the causes genetic, developmental, optimality, or phylogenetic perspec- and objects of selection (Crespi, 1990), and multiple tives and tools (Lewontin, 1979). A combination of all the regression methods (Lande & Arnold, 1983) and methods approaches allows connection of microevolution with macro- for visualizing adaptive surfaces (Schluter & Nychka, 1994) evolution, in the context of a trait's genetic architecture and can be used most e€ectively once the objects and causes of functional signi®cance, or lack thereof. Whereas the analysis of selection are reasonably well understood. Quantitative-genetic adaptation bene®ts from integration of multiple disciplines, analysis of G then permits inference of population trajectories the study of maladaptation demands it. Such a research on adaptive surfaces and assessment of if, and how, aspects program is challenging, but without it we will never thor- of genetic architecture prevent, delay, or facilitate approach oughly understand the causes of phenotypic variation. to local peaks (e.g. BjoÈ rklund, 1996), and study of if, and how, peaks move over time. One of the most important outcomes of such analyses will be whether or not G itself can Conclusion be considered adaptative (Thornhill, 1990; Schluter, 1996); To the extent that evolution is de®ned in terms of changes has it been shaped by a history of correlational selection, or in gene frequencies and phenotypes, and to the extent does it re¯ect inexorable, intrinsic relations between traits? that `natural selection is not evolution' (Fisher, 1958; p. 1), Ultimately, we will need to connect G with the genes that maladaptation can evolve. However, I believe that the lead to adaptation (Clark, 1987; Orr & Coyne, 1992). Genes integrative research needed to analyse it has largely been of larger e€ect may be more likely to exhibit maladaptive prevented by the rhetoric of Gould & Lewontin (1979), pleiotropy and may be more likely to be ®xed by selection, at chaotic de®nitions of constraint (Antonovics & van Tiender- least during the early stages of approach to an optimum (Orr, en, 1991), denigrations of optimality (Rose et al., 1987), the 1998). contrast between a genetic focus on variation and an Although quantitative-genetic methods are tremendously ecological focus on optimal states, and the contention of powerful for microevolutionary inference, G only more or less some behavioural ecologists that they need seldom inspect indirectly re¯ects the developmental mechanisms that translate Pandora's black box of genetic mechanisms (e.g. Thornhill, between genotypes and phenotypes (Houle, 1991). Empirical 61990). Many authors have called for synthetic studies that analysis and modelling of developmental processes can identify combine phylogeny, teleonomy, development, and genetics in potential maladaptation by demonstrating that aspects of the analysis of adaptation and constraint, however de®ned ontogeny that are recalcitrant to modi®cation by selection lead 7(e.g. Stearns, 1984; Arnold, 1992; Pigliucci & Kaplan, 2000). to maladaptive phenotypes (e.g. Slatkin, 1987). Understanding I have attempted here to sketch out a research program based the developmental mechanisms of maladaptation is important on their hopes. An understanding of how and why both because it allows for exclusion of alternative explanations of adaptation and maladaptation evolve, and how they are observed apparent maladaptation, such as failure to identify linked to population dynamics, is increasingly important as the correct trait or selective context. For example, multiple rapidly transform the globe into an environment lineages of vertebrates have gained and lost digits in concert where maladaptation is expected to be more prevalent for all with the evolution of larger and smaller body sizes (Alberch & species, including ourselves. Gale, 1985; Alberch, 1985). Is this pattern due to intrinsic inability of developmental systems to produce certain numbers of digits from a certain amount of limb bud tissue (i.e., a lack Acknowledgements of appropriate variation), or might fewer digits be adaptive in some unknown way for smaller species? Does `transference' of I am grateful to Arne Mooers, Andrew Pomiankowski, Mike secondary sexual characteristics from males to females (Lande, Whitlock and two anonymous reviewers for helpful comments, 1987; Muma & Weatherhead, 1989) result from highly and to Andrew Pomiankowski for inviting me to write this conserved hormonal e€ects on development, or might such short review.

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629. 628 B. J. CRESPI

References JOHNSTON, K. P., MCKINNEY, F.AND SORENSEN , M. D. 1999. Phylogenetic constraint on male parental care in the dabbling ducks. Proc. R. ALBERCH, P. 1985. Developmental constraints: why St. Bernards often Soc. Lond. B., 266, 759±763. have an extra digit and poodles never do. Am. Nat., 126, 430±433. KIRKPATRICK, M. 1996. Genes and adaptation: a pocket guide to the ALBERCH, P.AND GALE , E. A. 1985. A developmental analysis of an theory. In: Rose, M. R. and Lauder, G. V., eds. Adaptation, pp. evolutionary trend: digital reduction in amphibians. Evolution, 39, 125±146. Academic Press, New York. 8±23. LANDE, R. 1987. Genetic correlation between the sexes in the evolution ANTONOVICS, J.AND VAN TIENDEREN , P. H. 1991. Ontoecogenophylo- of dimorphism and mating preference. In: Bradbury, J. W. and constraints? The chaos of constraint terminology. Trends Ecol. Andersson, M. B., (eds) Sexual Selection: Testing the Alternatives, Evol., 6, 166±168. pp. 83±94. John Wiley, New York. ARNOLD, S. J. 1992. Constraints on phenotypic evolution. Am. Nat., LANDE, R.AND ARNOLD , S. J. 1983. The measurement of selection on 140, S85±S107. correlated characters. Evolution, 37, 1210±1226. BARTON, N. H. 1990. Pleiotropic models of quantitative variation. LEROI, A. M., ROSE, M. R.AND LAUDER , G. V. 1994. What does the Genetics, 124, 773±782. comparative method reveal about adaptation? Am. Nat., 143, BAUM, D. A.AND LARSON , A. 1991. Adaptation reviewed: a phylogenetic 381±402. methodology for studying character macroevolution. Syst. Zool., LEWONTIN, R. C. 1979. Fitness, survival, and optimality. In: Analysis of 40, 1±18. Ecological Systems, pp. 3±21. Ohio State University Press, Colum- BJOBJORKLUNDÈ RKLUND, M. 1996. The importance of evolutionary constraints in bus, Ohio. ecological time scales. Evol. Ecol., 10, 423±431. LOESCHCKE, V. 1987. Genetic Constraints on Adaptive Evolution. CHEVERUD, L. M. 1984. Quantitative genetics and developmental Springer-Verlag, Berlin. constraints on evolution by selection. J. Theor. Biol., 110, 155±171. MAYNARD SMITH, J. 1976. What determines the rate of evolution? Am. CLARK, A. G. 1987. Senescence and the genetic-correlation hang-up. Nat., 110, 331±338. Am. Nat., 129, 932±940. MAYNARD SMITH, L., BURIAN, R., KAUFFMAN, S., ALBERCH, P., CAMPBELL, CRESPI, B. J. 1988. Adaptation, compromise, and constraint: the L., GOODWIN, B., LANDE, R., RAUP, D.AND WOLPERT , L. 1985. development, morphometrics, and behavioral basis of a ®ghter-¯ier Developmental constraints and evolution. Quart. Rev. Biol., 60, polymorphism in male Hoplothrips karnyi (Insecta: Thysanoptera). 265±287. Behav. Ecol. Sociobiol., 23, 93±104. MUMA, K. E.AND WEATHERHEAD , P. J. 1989. Male traits expressed in CRESPI, B. J. 1990. The measurement of selection on phenotypic females: direct or indirect sexual selection? Behav. Ecol. Sociobiol., interaction systems. Am. Nat., 135, 32±47. 25, 23±31. CRESPI, B. J.AND WOROBEY, WOROBEY,M.M. 1998. Comparative analysis of gall ORR, H. A. 1998. The of adaptation: the distribution morphology in Australian gall thrips: the evolution of extended of factors ®xed during adaptive evolution. Evolution, 52, 935±949. phenotypes. Evolution, 52, 1686±1696. ORR, H. A.AND COYNE , J. A. 1992. The genetics of adaptation: DARWIN, C. 1859. On the Origin of Species by Means of Natural a reassessment. Am. Nat., 140, 725±742. Selection. John Murray, London. ORZACK, S.H. AND SOBER , E. 1994. Optimality models and the test of DHONDT, A. A., ADRIAENSEN, F., MATTHYSEN, E.AND KEMPENAERS , B. adaptationism. Am. Nat., 143, 361±380. 1990. Nonadaptive clutch sizes in tits. Nature, 348, 723±725. PARKER, M. A. 1991. Nonadaptive evolution of disease resistance in an DOUGHTY, P. 1996. Statistical analysis of natural experiments in annual legume. Evolution, 45, 1209±1217. evolutionary biology: comments on recent criticisms of the use of PARTRIDGE, L.AND SIBLY , R. 1991. Constraints in the evolution of comparative methods to study adaptation. Am. Nat., 148, 943±956. life histories. Phil. Trans. R. Soc. Lond. B., 332, 3±13. EMLEN, D. J. 1996. Arti®cial selection on horn length-body size PIGLIUCCI, M.AND KAPLAN , J. 2000. The rise and fall of Dr. Pangloss: allometry in the horned beetle Onthophagus acuminatus (Coleoptera: adaptationism and the Spandrels paper 20 years later. Trends Ecol. Scarabaeidae). Evolution, 50, 1219±1230. Evol., 15, 66±70. FEAR, K. K.AND PRICE , T. 1998. The adaptive surface in ecology. PRICE, T.AND LANGEN , T. 1992. Evolution of correlated characters. OIKOS, 82, 440±448. Trends Ecol. Evol., 7, 307±310. FISHER, R. A. 1958. The Genetical Theory of Natural Selection. Dover, REEVE, H. K.AND SHERMAN , P. W. 1993. Adaptation and the goals of New York. evolutionary research. Quart. Rev. Biol., 68, 1±32. FONG, D. W., KANE, T. C.AND CULVER , D. C. 1995. Vestigialization and RIECHERT, S. E. 1993. The evolution of behavioral phenotypes: lessons the loss of non-functional characters. Ann. Rev. Ecol. Syst., 26, 249± learned from divergent spider populations. Adv. Study Behavior, 22, 268. 103±134. GILLESPIE, J. H. 1991. The Causes of . Oxford ROSE, M. R.AND LAUDER , G. V. 1996. Adaptation. Academic Press, New University Press. New York. York. GOULD, S. J.AND LEWONTIN , R. C. 1979. The spandrels of San Marco and ROSE, M. R., SERVICE, P. M.AND HUTCHINSON , E. W. 1987. Three the Panglossian paradigm: a critique of the adaptationist pro- approaches to trade-o€s in life-history evolution. In: Loeschcke, V., gramme. Proc. R. Soc. Lond. B, 205, 581±598. (ed) Genetic Constraints on Adaptive Evolution, pp. 91±105. Spring- HARSHMAN, L. G.AND HOFFMANN , A. A. 2000. Laboratory selection er-Verlag, Berlin. experiments using Drosophila: what do they really tell us? Trends SCHLUTER, D. 1996. Adaptive radiation along genetic lines of least Ecol. Evol., 15, 32±36. resistance. Evolution, 50, 1766±1774. HERRE, E. A. 1987. Optimality, plasticity and selective regime in ®g wasp SCHLUTER, D.AND NYCHKA , D. 1994. Exploring ®tness surfaces. Am. sex ratios. Nature, 329, 627±629. Nat., 143, 597±616. HOULE, D. 1991. Genetic covariance of ®tness correlates: what genetic SHAW, F. H., SHAW, R. G., WILKINSON, G. S.AND TURELLI , M. 1995. correlations are made of and why it matters. Evolution, 45, Changes in genetic variances and covariances: G Whiz! Evolution, 630±648. 49, 1260±1267. JANZEN, D. H.AND MARTIN , P. S. 1982. Neotropical anachronisms: what SLATKIN, M. 1985. Gene ¯ow in natural populations. Ann. Rev. Ecol. the gomphotheres ate. Science, 215, 19±27. Syst., 16, 393±430.

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629. EVOLUTION OF MALADAPTATION 629

SLATKIN, M. 1987. Quantitative genetics of heterochrony. Evolution, 41, THORNHILL, R. 1990. The study of adaptation. In: Beko€, M. and 799±811. Jamieson, D., (eds). Interpretation and Explanation in the Study of STEARNS, S. C. 1984. The tension between adaptation and constraint in Behavior, 2, pp. 31±62. Westview Press, Boulder. the evolution of reproductive patterns. In: Engels, W., Clark, W. H. WILKINSON, G. S. 1993. Arti®cial sexual selection alters allometry in the Jr., Fischer, A., Olive, P. J. W., and Went, D. F. (eds) Advances in stalk-eyed ¯y Cyrtodiopsis dalmanni (Diptera: Diopsidae). Genet. Invertebrate Reproduction, 3, pp. 387±398. Elsevier Science Pub- Res., 62, 213±222. lishers, New York.

Ó The Genetical Society of Great Britain, Heredity, 84, 623±629.