Species Richness Covaries with Mating System in Birds
Total Page:16
File Type:pdf, Size:1020Kb
The Auk 113(3):544-551, 1996 SPECIES RICHNESS COVARIES WITH MATING SYSTEM IN BIRDS SHAIBAL MITRA, •'3 HANS LANDEL,TM AND STEPHENPRUETT-JONES 2 •Committeeon EvolutionaryBiology and 2Department of Ecologyand Evolution, Universityof Chicago,1101 East 57th Street,Chicago, Illinois 60637, USA ABSTRACT.--Manystudies have soughtto identify traitsthat influencethe relative number of speciesin related taxa.We examinedwhether speciesrichness was associatedwith social mating systemin birds. Taxa with promiscuousmating systemstended to be more species- rich than their nonpromiscuoussister taxa. This associationwas statistically significant when examinedwith teststhat take into accountthe magnitudesof paired contrasts.The results do not arisefrom covariationbetween mating system and bodysize. We discussthese findings in the contextof the hypothesisthat sexualselection promotes speciation. Received 16 February 1995, accepted25 April 1995. AMONGTHE MOST important questionsin evo- ation in coloration and ornamentation, whereas lutionary biology are thoseconcerned with fac- the females are relatively uniform in appear- tors affectingspeciation. Many traits have been ance. Such variation among males often pro- proposedas key innovationsor adaptivebreak- vides criteria for species'boundaries according throughs responsiblefor the diversificationof to the phylogenetic speciesconcept (Cracraft particular taxa, but empirical testsof such hy- 1983, McKitrick and Zink 1988). Moreover, these potheseshave proven methodotogicatlydiffi- traits are inferred to function in premating re- cult (see Raikow 1986, Stowinski and Guyer productiveisolation under the biologicalspe- 1993). Testsof associationin comparative stud- ciesconcept (Mayr 1942,1963). Clearly, any force ies must be based on samplesreflecting inde- tending to accelerateevolutionary changesin pendent derivationsof the trait of interestacross secondarysexual characterswould hasten the a series of taxa whose historical relationships rate at which allopatric populations achieved can be reconstructed(Felsenstein 1985, Harvey speciesrank under both the phylogenetic and and Paget 1991, Guyer and Slowinski 1993, biologicalspecies concepts. Slowinski and Guyer 1993). Traits for which Sexual selection is recognized as a process effectson relative speciesnumbers have been that can lead to the rapid modificationof a va- analyzed in this manner include phytophagy riety of characters,particularly those involved in insects (Mitter et at. 1988), breadth of ovi- in mating preferences (Fisher 1930; Kaneshiro position sites in insects(Zeh et at. 1989), and 1980, 1989;Lande 1981, 1982;Kirkpatrick 1982; viviparity in fishes(Lydeard 1993).All of these Kaneshiro and Giddings 1987; Lande and Kirk- are traits at the level of individual organisms. patrick 1988).In fact, the mutually reinforcing In our study, we examined speciesrichness in elaborationof mating preferencesand second- relation to a population-level trait, social mat- ary sexualcharacters during "run-away" sexual ing system. selectionimplies that this processcould be very It has long been recognizedthat closelyre- effective in producing the differentiation nec- latedspecies often differ moststrikingly in their essaryfor the developmentof major morpho- secondarysexual characteristics (Darwin 1871). logical differences,and even premating repro- For example,within groupsof birds as diverse ductive barriers, between populations (Kane- asgrouse, hummingbirds, and birdsof paradise, shiro 1980; Lande 1981, 1982; West-Eberhard the males of many speciesshow striking vari- 1983;Wu 1985; Lande and Kirkpatrick 1988). Although there is this theoreticaljustification • Presentaddress: Department of Natural Resources for predicting a relationshipbetween the action Science,University of RhodeIsland, Kingston, Rhode of sexualselection and the processof speciation, Island 02881, USA. the methodologicaldifficulties involved in test- 4 Presentaddress: Department of Biology,Carleton ing sucha prediction are not trivial. Precisedata College, Northfield, Minnesota 55057, USA. regardingsexual-selection gradients (Lande and 544 July 1996] SpeciesRichness and Mating Systems 545 Arnold 1983) are known for very few natural ical patterns and taxonomic artifacts. Barra- populations,and pastrates of speciationare dif- clough et al. (1995) reported a significant rela- ficult or impossible to reconstruct. Neverthe- tionshipbetween sexualdichromatism and spe- less,one can approachthe hypothesisindirectly cies richness in passefine birds. To the extent by choosingconservative indicators of the vari- that dichromatism reflects the intensity of sex- ables of interest. For example, although it is ual selection, this result also is consistent with now clear that sexualselection can occurin spe- the hypothesislinking sexualselection and spe- cies exhibiting a wide variety of social mating ciation. In a study of the relationship between systems(Birkhead and Moller 1992), sexualse- lek mating systemsand sexual dimorphism in lection generally will be strongerin speciesex- birds, H6glund (1989) presented results sug- hibiting promiscuousmating systemsthan in gesting that lek-breeding, sexually dimorphic specieswith nonpromiscuousmating systems taxa were relatively species-rich.Nevertheless, (Arnold and Duvall 1994).Promiscuity thus may he did not statethis result explicitly nor discuss be an indicator of relatively intense sexualse- it in the contextof a hypothesislinking species lection. richnessto mating systems. Similarly, for the purposes of the test de- Ryan (1986) reported that three successive scribedhere, the number of extant species(spe- transitions in the neuroanatomy of anuran am- cies richness)that we observe in a given taxon phibians coincided with increasesin species is a conservativeestimator of past rates of spe- richness.The changesin neuroanatomywere ciation in that taxon. Becausespecies richness associatedwith increasesin the range of fre- is actuallythe net differencebetween speciation quenciesover which females were sensitive to and extinction events over a given period of the mating calls of males,and Ryan attributed time, one must be careful not to mistake the the increasedspecies richness to acceleratedrates effect of reduced extinction rates with that of of speciationvia divergencein mating calls. His increased speciation rates, and vice versa. We results are consistentwith the hypothesis that must examine the likely effectsof sexualselec- differentiation of traits associated with mate tion on both processes.There is some reasonto choicepromotes speciation, but the data do not believe that sexual selection increases the like- permit a general statisticaltest. lihood of extinction through its ability to re- duce the mean fitness of a population (Fisher METHODS 1930, McLain 1993). Conversely, there is no in- dication that sexual selection reduces extinction To identify phylogeneticcontrasts in the intensity rates. Therefore, a positive relationship be- of sexualselection, we comparedtaxa with promis- tween the intensity of sexualselection and spe- cuous mating systemswith taxa exhibiting other cies richnessis most reasonablyinterpreted as (mostly monogamous)mating systems.For our pur- a reflection of increased rates of speciation. poses,we defined promiscuityas a mating systemin The prediction that avian taxa with mating which there is no lastingsocial pair bond and males systemsindicative of intense sexual selection do not provide parental care or resourcescritical to femalereproduction. We alsoincluded species having have greater numbers of speciesthan do their non-resource-basedpolygyny with no male parental sister taxa with other mating systemshas not care(e.g. someAnatidae). First, we identified all cases been tested rigorously. Nevertheless, several of avian mating systemsknown to meet these criteria empiricalstudies have yielded resultsrelevant (see Appendix). Next, we sought phylogeniesthat to hypotheseslinking sexualselection and spe- identifiedthe sistertaxa of monophyletictaxa having ciation. For example, Pierotti and Annett (1993) promiscuousmating systems.We rejected casesin reported that avian families exhibiting non- which phylogenetic relationships could not be re- monogamousmating systemsand sexual di- constructedwithin groupsthat were polymorphic with morphism were more species-richthan families respectto mating system.For instance,many mem- exhibiting monogamousmating systemsand bers of the Phasianidaeare promiscuous,but their phylogeneticrelationships are ambiguous(Johnsgard sexualmonomorphism. This result is consistent 1986, 1988).Similarly, one speciesin the genusAn- with the prediction described above, but the dropadus(family Pycnonotidae)is promiscuous,but method employedto compareaverage numbers its relationshipswith its congenersare unknown. of speciesper family acrossgroups of families, In each casewhere the sistertaxon was uniformly treating each family as independent, was not nonpromiscuous,we comparedthe numbers of spe- intended to distinguishbetween purely biolog- cies in the two sister taxa. We examined whether the 546 MITRA,LANDEL, AND PRIJETT-JOIqES [Auk, Vol. 113 speciesrichness of the promiscuoustaxon exceeded Our data set comprised 14 comparisons,as that of its sistertaxon more frequently than expected detailed in Table 1. One of these, comparison by chance,using three statisticaltests: a sign test,a 14, involved a group (Quiscalusgrackles) for Wilcoxonsigned-ranks test, and a bootstraprandom- which two availablephylogenies differed