1384 NOTES AND COMMENTS

Evolution, 39(6), 1985, pp. 1384-1388

THE USE OF PHYLOGENY IN BEHAVIOR AND ECOLOGY

F. STEPHEN DOBSON Museum ofZoology and Division ofBiological Sciences University ofMichigan, Ann Arbor, MI48109

Received July 13, 1984. Accepted July 12, 1985

The comparativeanalysis ofcharacteristicsofdif­ characteristics are examinedfor correlationsamong ferent species is extensively used in evaluating pos­ themselves, or for correlations among characteris­ sible selective contexts for behavioral and ecolog­ tics and environments. Ifa group that shares a char­ ical differences among species (e.g., Alexander et acteristic is also a monophyletic group, then there al., 1979; Blumer, 1979; Bradbury, 1981; Dobson, is an alternative to explanations ofnatural selection 1982; Greenwood, 1980;Ralls, 1976;Steams, 1977; in recent (modem) environments: the characteristic Wittenberger, 1976;WittenbergerandTilson, 1980). might have evolved in a past environment, and it Such analyses often explicitly or implicitly assume is retained because superior alternatives did not that when species resemble each other in some be­ arise in the taxa. In an interspecific comparison in havioral or morphological characteristic, the re­ which monophyletic and character trait groups co­ semblance is due to independent appearance ofthe occur, the a priori historical origin hypothesis is an characteristic under study in each species in the equally valid hypothesis compared to any expla­ analysis. This assumption can be (but rarely is)tested nation of superior function (i.e., superior adaptive­ by examining evidence for independent evolution ness) in recent environments. of characteristics, deduced from a well supported Shared morphological characteristics in a mono­ hypothesis of phylogeny. In this essay, I will explain phyletic group are commonly thought to be influ­ how non-systematists in general, and behaviorists enced strongly by a commonancestry. For example, and ecologists in particular, can benefit from con­ numerous characteristics of reproductive mor­ sidering historical influences on the evolution of phology ofmarsupial mammals are shared widely characteristics. I will also indicate problems that within the clade (the monophyletic group), and they may be encountered in extracting a well supported are probably descended from a single ancestor and phylogeny from the systematic literature. conserved over evolutionary time (Lillegraven, 1974, 1975).Whether marsupial reproductive mor­ Historical Analysis phology is strongly constrained against change (Lil­ The characteristics that contemporary speciesex­ legraven, 1974, 1975) or highly adaptive in certain hibit may have evolved in their recent environ­ environments (Low, 1978), or both, it would be ment, or they may be of relatively older evolution­ unsatisfactory to explain the distribution ofrepro­ ary origin. That is, a group of related species may ductive characteristics between marsupial and pla­ have derived characteristics independently or from cental mammals solely as the result of natural se­ a common ancestor. Ifa characteristic is obtained lection operating in recent environments. The through descent from an ancestor, then in com­ historical origin hypothesis is corroborated by phy­ parison with the ancestor the characteristic may be logenetic analysis in this case, and it should be con­ ofgreater, equal, or less utility to individuals in the sidered a valid alternative to adaptive explanations. descendent species in coping with their respective The hypothesis of shared similarity due to shared contemporary environments. Genes for the char­ history can also be applied to behavioral studies. acteristic should remain in the genomes until a more For example, Greenwood (1980) found that char­ beneficialalternative characteristic arises. In a group acteristics of juvenile dispersal, with several im­ of contemporary species that share a particular portant exceptions, differed between birds and characteristic, therefore, it might be difficult to as­ mammals. In about 90% of bird species, juvenile certain just how beneficial the characteristic is in males and females disperse equally, or juvenile fe­ the respective environments ofthe species. Itmight males exhibit somewhat greater dispersal than ju­ be even more difficult (or impossible) to judge venile males. In about 70% of mammal species, whether alternative characteristics that have never males are the predominant dispersing sex among evolved in the group ofspecies might be more ben­ juveniles. The hypothesis that the characteristics of eficialthan an existing characteristic. Thus, it might these taxa might be shared among species because be misleading to assume that a characteristic is an they share a common historical origin should be adaptation to contemporary environments, and a considered along with the adaptive hypotheses dis­ characteristic that descends from a common ances­ cussed by Greenwood (1980). tor to contemporary species cannot be assumed to The historical origin hypothesis might also prove be of "optimal" benefit to individuals. illuminatingin population ecology.One of the most In interspecific comparisons, sets of alternative intensively studied questions in population ecology NOTES AND COMMENTS 1385 is why many species of microtine rodents exhibit hypotheses that invoke adaptation to recent envi­ "cyclic" or erratic outbreaks in population sizeevery ronments. three to four years. Krebs and Myers (1974) argued A second method can be used to insure that pat­ that the widespread occurrence of fluctuating mi­ terns of characteristics among taxa are likely to be crotine populations makes it likely that a single due primarily to historical origin. Alexander et al. causal mechanism is responsible. Perhaps a pro­ (1979) examined the degree of polygyny and degree pensity for cyclic fluctuations due to characteristics of sexual dimorphism in three distantly related taxa of individual microtine rodents (in particular en­ (pinnipeds, primates, and ungulates). Each taxon vironments) has been inherited by many microtine contained a wide range of variation in the charac­ speciesfrom their commonancestor. Ecologistshave teristics of polygyny and sexual dimorphism, and recently emphasized causal mechanisms that are the range of variation likely evolved independently "intrinsic" to fluctuatingmicrotine rodents, and they at least three times. While a consideration of his­ have assumed thatthe characteristics of individuals torical influences was not explicitly mentioned by that result in changes in population size are adap­ Alexander et aI. (1979), a primary influence of his­ tive in contemporary environments (e.g., Krebs, torical origin on the pattern of characteristics that 1978; Lidicker, 1978). An influence of historical they examined is unlikely. origin (an intrinsic factor) has not previously been Ridley (1983) suggested a third method for iden­ suggested. tifying independent evolution that assigns charac­ I will not present examples of the relevance of teristics to ancestors by successive "outgroup" the historical origin hypothesis to the study offunc­ comparisons. Such inference of behavioral or eco­ tional morphology, life history, or coevolution. logical characteristics of ancestors, however, is not Lauder (1981, 1982) presented methods for testing required for testing hypotheses of historical origin. the hypothesis of historical influence on morpho­ As the previous two methods indicate, all that is logicalform and argued that functional morphology required is evidence of the minimum number of should be viewed as an historical discipline. Steams independent evolutionary events (i.e., assignment (1983, 1984) has indicated a strong influence of of a hypothetical ancestral condition that minimiz­ phylogeny on the evolution oflife histories (but see es potential characterevolution). For example, Rid­ Vitt and Seigel, 1985), and the influence ofhistor­ ley (1983) states that his method cannot be used in ical origin was applied to the evolution of ecological cases like the study by Alexander et aI. (1979), de­ associationsby Mitter and Brooks(1983)and Brooks spite considerable evidence of independent evolu­ (1985). These studies indicate a growing interest in tion. the value ofconsidering the influence ofhistory in A fourth method that corrects comparative stud­ interspecific studies. Tests of historical influences ies for phylogeny was developed by Felsenstein need to be extended to studies of the evolution of (1985). This statistical methodology has consider­ behaviors and to additional subfields of ecology. able barriers to practical application (discussed by Such tests can be used as an important first step in Felsenstein), but it illustrates the difficulties in­ the design of research, in which groups of species volved in trying to separate environmental and his­ that are most likely to exhibit adaptations to recent torical influences on the characteristics of organ­ environments are chosen for testing evolutionary isms. Until methods for partitioning these influences hypotheses. are available, methods that indicate when the in­ The historical origin hypothesis can be tested by fluence of history is minimized or maximized will comparing the distribution of behavioral or eco­ likely be most helpful to behaviorists, ecologists, logical characteristics among species to a well sup­ and other functional biologists. ported phylogeny for the species. Ifthe distribution ofcharacteristicsamong species is inconsistent with Example the phylogeny, then several independent evolution­ The cladogram (a hypothetical phylogeny that ary events are required to explain the distribution shows speciation events, but not the time between of characteristics. Under these conditions, it is un­ speciation events) that behaviorists and ecologists likely that the distribution of characteristics has use to test the common historical origin hypothesis been strongly influenced by historical origin. will, in most cases, come from the literature. Un­ fortunately, the systematic classification of a group Four Methods of animals cannot always be used to produce a I used this method in a recent examination of the cladogram. For example, prairie dogs, genus Cyn­ occurrence of territoriality by adult males during omys, are thought to be a sister group to the ground the mating season in the sciurid rodents of North squirrel subgenus Spermophilus (Black, 1963; America (Dobson, 1984). I compared the occur­ Bryant, 1945;Hight etaI., 1974;Nadleretal., 1971). rence ofterritoriality to the best supported phylog­ The group formed by prairie dogs and this subgenus eny of sciurid species, and found that in the ground (a monophyletic group, which by definition in­ squirrels (genera Cynomys and Spermophilus), male cludes all related species) is more distantly related territoriality probably arose independently more to the other ground squirrel subgenera (see Fig. 1c than once. Thus, I could reject the historical origin and Table 1A).The genus Spermophilus, therefore, hypothesis as a major determinant of the occur­ is paraphyletic (i.e., not all the related species are rence of male territoriality and proceed to test included in the genus). 1386 NOTES AND COMMENTS

a. b. c.

SPERMOPHILUS SPERMOPHILUS

CIl CIl :;, CIl .-:::: :;, .2 CIl -s: .-:::: :E .2 €Sa ts ~ CIl E 0 !!l E 0 .2 t: E :::: Q) t: Q) '- E - Q) '- .3 Q. Q) z: CIl Q. v 5} 8. 'iJ o CIl 0 o CIl 0 ~2::: m-... 0 0'- U 0 £ U 0 Q bbbbbaa b b b

FIG.1. a) A hypothetical phylogeny ofground squirrels, derived from the current classification system (see Table lA). b) The hypothetical phylogeny derived from the monophyletic classification system in Table lB. c) The phylogenetic relationships of the ground squirrels, modified from Bryant (1945). The genus Cynomys and the subgenera of the genus Spermophilus are shown. The placement of Cynomys is consistent with several systematic studies (Bryant, 1945; Black, 1963; Nadler et al., 1971; Hight et al., 1974). In a), b), and c), homologous behavioral characteristics a (the run-jump) and b (absence of the run-jump) are assigned to taxa (see text). Genera are completely capitalized and subgenera include lower case lettering.

Taxonomies that contain paraphyletic groups basis for comparative study of the run-jump be­ cannot be used to test hypotheses that invoke his­ havior (Table 1A), the results would be inconsistent torical origin of behavioral or ecological character­ with respect to the best supported phylogeny (Fig. istics. For example, I have observed a peculiarity lc). Using the phylogeny constructed from the cur­ in the running gait of black-tailed prairie dogs (c. rent system of classification (Fig. la), more than ludovicianus) and Richardson's ground squirrels (S. one independent evolutionary origin of the run­ richardsonii, subgenus Spermophilus). When fleeing jump or its homolog (the absence ofthe run-jump) from an observer(and, presumably, from terrestrial is required. A monophyletic system of classification predators) individuals of these species often push (Table IB) can be used to test the common histor­ the anterior part of the body off the ground with ical origin hypothesis, and in this case only one the forelegs, while continuing to run with the hind independent evolutionary event is required (Fig. legs (in the prairie dog this behavior is associated Ib), Note that some classifications do not allow with vocalization; King, 1955). I have not observed testing of the common historical origin hypothesis this behavior (the "run-jump") in observations of (e.g., Table lC is monophyletic, but it cannot be other ground squirrels, nor have I found mention used to provide a relevant test in my example). In of it in the ethological literature. Prairie dogs may such cases, a more complete hypothetical phylog­ be most closely related to Richardson's ground eny must be examined (e.g., Fig. lc). squirrels (Nadler et al., 1971), and they might share An adaptive scenario to explain the evolutionary the run-jump behavior because it was present in a origin of the run-jump behavior might be that it common ancestor. improves visual detection of terrestrial predators The presence ofthe run-jump in the genera and in prairie habitats, perhaps due to visual obstruc­ subgenera ofprairie dogs and ground squirrels fol­ tion of tall grass. Because the historical origin hy­ lows the pattern of character state a in Figure 1, pothesis is supported, however, an investigation of and the absence of the run-jump (according to the the habitat ofthe ancestor would be required to test current literature) follows the distribution of char­ the scenario. Ifit were observed that the behavior acter state b. Ifthe current systematic classification occurs irrespective of the visual obstruction of the of ground squirrels and prairie dogs is used as a habitat, the scenario might be falsely rejected be- NOTES AND COMMENTS 1387

TABLE 1. Alternate systems ofclassification ofthe logeny (see Eldredge and Cracraft, 1980; Wiley, ground squirrels. The order of listed genera and 1981). Naturally, the strongest tests of historical subgenera has no special meaning. The current sys­ origin hypotheses will be in groups in which sys­ tem (A) is paraphyletic, and it cannot be used to tematists have little disagreement concerning hy­ pothetical phylogeny. reconstruct a phylogeny that is consistent with Fig­ ure Ie. The monophyletic systems (B and C) can ACKNOWLEDGMENTS be used to reconstruct phylogenies that are consis­ lowe A. G. Kluge and P. Myers special thanks tent with Figure Ic, Monophyletic system II (C) is for patient discussions ofthe ideas presentedabove. most consistent with both the phylogeny and the I thank R. D. Alexander, L. S. Blumer, D. R. Brooks, current generic and subgeneric designations. G. K. Creighton, P. W. Ewald, W. L. Fink, D. J. Hafuer, S. Hatfield, L. R. Heaney, J. L. Hoogland, A. Current system T. H. Hubbell, J. A. King, A. G. Kluge, G. V. Lau­ der, B. L. Lundrigan, J. O. Murie, P. Myers, M. Genus Cynomys Robertson, and R. S. Voss for reading and com­ Genus Spermophilus menting on drafts ofthe manuscript. A discussion Subgenus Spermophilus with G. R. Michener was very helpful. J. D. Kjel­ Subgenus Ictidomys gaard drew the figure. Kananaskis Centre, Univer­ Subgenus Poliocitellus sity of Calgary, provided facilities for preparation Subgenus Xerospermophilus of the manuscript during my field season. Subgenus Otospermophilus Subgenus Callospermophilus LITERATURE CITED B. Monophyletic System I ALEXANDER, R. D., J. L. HOOGLAND, R. D. HOWARD, Genus Cynomys K. M. NOONAN, AND P. W. SHERMAN. 1979. Subgenus Cynomys Sexual dimorphisms and breeding systems in Subgenus Spermophilus pinnipeds, ungulates, primates, and humans, pp. Genus Ictidomys 402-435. In N. A. Chagnon and W. G. Irons Genus Poliocitellus (eds.), Evolutionary Biology and Human Social Genus Xerospermophilus Behavior. Duxbury Press, North Scituate, MA. Genus Otospermophilus BLACK, C. C. 1963. A review ofthe North Amer­ Genus Callospermophilus ican Tertiary Sciuridae. Bull. Mus. CompoZool. C. Monophyletic System II 130:109-248. BLUMER, L. S. 1979. Male parental care in the Genus Cynomys bony fishes. Quart. Rev. BioI. 54:149-161. Subgenus Cynomys BRADBURY, J. W. 1981. The evolution ofleks, pp. Subgenus Spermophilus 138-169. In R. D. Alexander and D. W. Tinkle Subgenus Ictidomys (eds.), Natural Selection and Social Behavior. Subgenus Poliocitellus Chiron, N.Y. Subgenus Xerospermophilus BROOKS, D. R. 1985. Historical ecology: A new Subgenus Otospermophilus Subgenus Callospermophilus approach to studying the evolution ofecological associations. Ann. Missouri Bot. Gard. In press. BRYANT, M. D. 1945. Phylogeny ofnearctic Sci­ uridae. Amer. Midi. Natur. 33:257-390. cause the behavioris not evolutionarilymaintained DOBSON, F. S. 1982. Competition for mates and as the scenario predicts. In this case, the possible predominant juvenile male dispersal in mam­ historical inheritance of the characteristic renders mals. Anim. Behav. 30:1183-1192. the evolutionary scenario oforigination untestable ---. 1984. Environmental influences on sci­ in recent environments. urid mating systems, pp. 229-249. In J. O. Mu­ rie and G. R. Michener (eds.), The Biology of Conclusion Ground Dwelling Squirrels. Univ. Nebraska Because some systematists produce taxonomies Press, Lincoln. that do not reflect phylogeny (i.e., paraphyletic tax­ ELDREDGE, N., AND J. CRACRAFT. 1980. Phylo­ onomies), the behaviorist or ecologist cannot sim­ genetic Patterns and the Evolutionary Process. ply refer to a taxonomy to test historical origin Columbia Univ. Press, N.Y. hypotheses, without study ofthe method that pro­ FELSENSTEIN, J. 1985. Phylogenies and the com­ duced the taxonomy. The safest procedure is to parative method. Amer. Natur. 125:1-15. extract the best supported phylogeny from the sys­ GREENWOOD, P. J. 1980. Mating systems, philo­ tematic literature. Taxonomies designed by Hen­ patry and dispersal in birds and mammals. Anim. nigian cladists, however, should be very useful to Behav.28:1140-1162. behaviorists and ecologists. Cladists are primarily HIGHT, M. E., M. GOODMAN, AND W. PRYCHODKO. concerned with accurate inference of patterns of 1974. Immunological studies ofthe Sciuridae. shared history, and their taxonomies can be ex­ Syst. Zooi. 23:12-25. pected to reflect the most accurate estimates ofphy- KING, J. A. 1955. Social behavior, social orga- 1388 NOTES AND COMMENTS

nization, and population dynamics in a black­ J. Futuyma and M. Slatkin (eds.), Coevolution. tailed prairiedog town in the Black Hills ofSouth Sinauer, N.Y. Dakota. Contrib. Lab. Vert. BioI. 67:1-123. NADLER, C. F., R. S. HOFFMANN, AND J. J. KREBS, C. J. 1978. A review of the Chitty hy­ PIZZIMENTI. 1971. Chromosomes and serum pothesis ofpopulation regulation. Can. J. ZooI. proteinsofprairie dogs and a model ofCynomys 56:2463-2480. evolution. J. Mammal. 52:545-555. KREBs, C. J., AND J. H. MYERS. 1974. Population RALLS, K. 1976. Mammals in which females are cycles in small mammals. Adv. EcoI. Res. 8: larger than males. Quart. Rev. BioI. 51:245­ 267-399. 276. LAUDER, G. V. 1981. Form and function: Struc­ RIDLEY, M. 1983. The Explanation of Organic tural analysis in evolutionary morphology. Pa­ Diversity. Oxford Univ, Press, N.Y. leobioI. 7:430442. STEARNS, S. C. 1977. The evolution oflife history --. 1982. Historical biology and the problem traits. Ann. Rev. EcoI. Syst. 8:145-171. ofdesign. J. Theoret. BioI. 97:57-67. ---. 1983. The influence ofsize and phylogeny LIDICKER, W. Z., JR. 1978. Regulation ofnumbers on patterns of covariation among life-history in small mammal populations-historical re­ traits in the mammals. Oikos 41:173-187. flections and a synthesis, pp. 122-141. In D. P. --. 1984. The effects of size and phylogeny Snyder (ed.), Populations of Small Mammals on patterns ofcovariation in the life history traits Under Natural Conditions. Pymatuning Labo­ oflizards and snakes. Amer. Natur. 123:56-72. ratory of Ecology, Special Publication No.5. VITI, L. J., AND R. A. SEIGEL. 1985. Life history Univ, Pittsburgh, Pittsburgh, PA. traits oflizards and snakes. Amer. Natur. 125: LILLEGRAVEN, J. A. 1974. Biogeographical con­ 480-484. siderations ofthe marsupial-placental dichoto­ WILEY, E. O. 1981. Phylogenetics. Wiley, N.Y. my. Ann. Rev. EcoI. Syst. 5:263-283. WITIENBERGER, J. F. 1976. The ecological factors ---. 1975. Biological considerations of the selecting for polygyny in altricial birds. Amer. marsupial-placental dichotomy. Evolution 29: Natur. 110:779-799. 707-722. WITTENBERGER, J. F., AND R. L. TILSON. 1980. Low, B. S. 1978. Environmental uncertainty and The evolution of monogamy: Hypotheses and the parental strategies of marsupials and pla­ evidence. Ann. Rev. EcoI. Syst. 11:197-232. centals. Amer. Natur. 112:197-213. MITIER, C., AND D. R. BROOKS. 1983. Phyloge­ Corresponding Editor: A. E. Dunham netic aspects of coevolution, pp. 65-98. In D.

Evolution, 39(6), 1985, pp. 1388-1392

FOUNDER-FLUSH SPECIATION: AN UPDATE OF EXPERIMENTAL RESULTS WITH DROSOPHILA

DIANE M. B. DODD' AND JEFFREY R. POWELL Department ofBiology, Yale University, P.O. Box 6666, New Haven, CT 06511

Received September 28, 1984. Accepted June 7, 1985

Mayr (1954) emphasized the possible causal role somal studies indicate that many species can be of small populations, his founder effects, in initi­ traced to ancestral species, and the derived species' ating speciation. Carson and colleagues rekindled chromosomal constitution is best explained by a interest in the role offounder events in speciation founder event. Carson has developed a rather spe­ by their studies of Hawaiian Drosophila (Carson, cificand comprehensive view ofhow founder events 1970; Carson et al., 1970; Carson andYoon, 1982). can promote or cause speciation (see Carson, 1982, Basically, Carson hypothesized a causal relation­ and references therein). One aspect of this model ship between two remarkable observations: 1) De­ is the possible importance of chromosomal rear­ spite its relatively young age, the Hawaiian archi­ rangements (e.g., inversions) in controlling rates of pelago contains an extraordinarily large number of recombination. In structurally heterozygous pop­ endemic species of Drosophila; and 2) Chromo- ulations, highly coadapted polygenic complexes build up; they become a closed system in the sense that recombination is rare; and, when recombina­ , Present address: Department ofBiological Sci­ tion occurs, the resultant scrambled chromosomes ences, University of North Carolina, Wilmington, are maladapted. In a founder event chromosomal NC28403. homozygosity may suddenly be achieved, and the