Phylogenetic Analysis of Avian Parental Care
Total Page:16
File Type:pdf, Size:1020Kb
The Auk 109(4):828-846, 1992 PHYLOGENETIC ANALYSIS OF AVIAN PARENTAL CARE MARY C. MCKITRICK Museumof Zoologyand Departmentof Biology,University of Michigan, Ann Arbor, Michigan48109, USA ABsTRACT.--Phylogeneticpatterns can be usedto generatedetailed historical hypotheses about the evolution of charactersystems, including parental-carebehavior. Phylogenetic analysisof 60 taxa usingparental-care data (15 characters)in combinationwith anatomical data (69 characters)shows that biparental care is primitive for birds, and that biparental incubationarose from an ancestralcondition in which neither parent incubated.Components of parentalcare may be decoupled,such that incubationand feedingof nestlingsis biparental, but postfledgingcare is uniparental;the patternof decouplingvaries across taxa. Behavioral plasticity,environmentally induced variation, and quantitativevariation in charactersthat are codedqualitatively are someof the problemsinherent in suchan analysis.Nevertheless, historicalanalysis offers a goodvehicle for obtainingrigorous hypotheses about the coevo- lution of avian charactersystems. Received 5 July 1991,accepted 29 October1991. THEUSE of phylogeneticpattern to elucidate phylogeny(one of the "topthree heinous 'crimes questionsabout other evolutionaryand ecolog- againstphylogenetics' "; Brooksand McLennan ical patternsor processeshas becomeincreas- 1991). Looking at numerousspecies from the ingly commonin the last 15 years.Several rig- same taxonomic family actually may be quite orous attemptsto incorporatebehavioral data valid, dependingon phylogeneticrelationships into phylogeneticanalyses or to comparesuch amongthose species. What mattersabove all is datawith existingphylogenies have beenmade the number of times the trait in questionarose, in the lastfive years.Clutton-Brock and Harvey rather than how many genera or families show (1977) recognized the pitfalls of comparing the trait; this can only be determined by a de- closely related speciesto study the origin of tailed phylogeny of the organismsand traits. ecologicalphenomena, and this has been fur- The trait could well have arisen numerous times ther emphasizedby Clutton-Brockand Harvey within the family, in which caseit is not only (1979), Ridley (1978, 1983), Harvey and Mace meaningful but necessaryto include members (1982), Felsenstein(1985), Dobson (1985), Pagel of the samefamily in the study. With such in- and Harvey (1988), H/Sglund(1989), and Bj/Srk- formation, one can ask appropriate questions lund (1990). Coddington (1988) outlined a about the adaptive significanceof that trait. It method for using cladistichypotheses of rela- is not profitable to begin the investigationby tionship to test hypothesesof adaptation,and determiningwhether one shouldcount genera, this kind of approachhas been fruitfully ap- families or orders in which the trait occurs. plied in a numberof recentstudies (Prum 1990, Rather, one should begin with a detailed phy- B. N. Danforth unpubl. data,R. J.Smith unpubl. logeny and determine whether the trait arose data). once or many times, and the pattern of origin A historicalapproach is a profitable one, as and loss. Adaptive explanationscan only be it attemptsto find the correcthierarchical level tested with convergent cases,not with homol- at which hypothesesof adaptationare mean- ogousones. In other words,two speciesshow- ingful. It is not enough simply to dismiss"ad- ing the trait can only serve as two data points aptations"as artifactsof phylogeneticpropin- in supportof an adaptiveexplanation of origin quity; rather, one mustdissect out the effectsof if they did not inherit the trait from a common phylogeny so that adaptive explanationsmay ancestor. be viewed in the appropriatecontext. Clutton- McLennan et al. (1988) lamented what they Brockand Harvey's (1979) concernabout bias- perceivedas a declinein the use of behavioral ing investigationsof ecologicalphenomena by data to constructphylogenetic hypothesesin examining closely related specieswas appro- the last three decades,citing work by Lorenz priate. However, they may have been some- (1941) in which this was done successfullyfor what tyrannizedby taxonomyin their view of ducks(Anatidae) and work by later authorsin the problem--that is, equatingtaxonomy with which phylogeny was ignored. A noncladistic 828 October1992] PhylogenyofAvian Parental Care 829 attempt to extend Lorenz's work was made by Cracraft's (1981) classification, which lacks Johnsgard(1961), for example.McLennan et al. characterdata and explicit hypothesesof phy- (1988) were remarkablysuccessful in generat- logeneticrelationships, is of limited utility for ing phylogenies for gasterosteidfishes using the purposeof understandingthe evolutionof data on reproductivebehavior, and argued that character systems(e.g. behavior). Without a behavioral data are extremely useful for con- phylogeny,van Rhijn was limited to making structing phylogenies. hypothesesbased on statistical probabilities The evolution of parental care was an early rather than historical evidence. focus of the phylogenetic approach. Ridley In this study, I use characterdata from the (1978) reviewed the circumstancesof paternal hindlimb musculature (from McKitrick 1991) in care in a broad array of taxa to consider the combinationwith published behavioral data to effectsof fertilization mode, territoriality, fe- test van Rhijn's hypothesisthat male-only pa- male choice and other factors on the evolution rental careis primitive in birds,and to generate, of paternal care.He recognizedthe possibility in effect,phylogenies of avian parental-carebe- of phylogeneticbias in formulatinghypotheses havior. In doing so, I also was able to test some about the evolution of the phenomenon. Git- of the hypothesesof Silver et al. (1985) about tleman (1981) used phylogeny to test predic- the causalnature of associationsamong certain tions about probabilities of transition among aspectsof parental care. In considering the typesof parentalcare in bony fishes(i.e. among question of what aspectsof parental care are states of uniparental [male], uniparental [fe- primitive for birds,however, it shouldbe borne male], biparental, and no parental care). He in mind that there is nothing magical about foundthat the mostcommon evolutionary tran- statementsregarding the primitive or derived sition was from no parental care to uniparental nature of these character states. A character state care by males, but pointed out possiblestatis- that is derived for birds is primitive within birds. tical problems with the conclusion that this A statemay be derivedfor birds,but showearly transition had the highest likelihood. transitions to a different state, such that the ma- Van Rhijn (1984, 1985, 1990) proposed that jority of birds show a different state from the pure male parental care is the primitive con- one that aroseat the baseof the avian lineage. dition in birds. His hypothesisis basedupon These terms are relative, and the hierarchical the premise that "monogamouspaternal care level to which they refer shouldalways be con- systemscan easily evolve towards all recent sidered. mating systemsin birds" (van Rhijn 1984:103), Phylogeniesare hypothesesabout character whereasbiparental caresystems cannot. This is evolution and, as such, they allow the refine- in contrastwith the suggestionby Emlen and ment of ecologicalquestions such as, why does Oring (1977:220)that "complete male parental speciesX exhibit biparental care?With a phy- care is most likely to develop in groups with logeny, one can go beyondthe proximatean- ... a phylogenetic history of shared incuba- swer to this question,namely, "Becauseits an- tion." If van Rhijn's hypothesisis correct,then cestors exhibited biparental care," and seek complete male care would probably have ultimate answersby examiningthe ecological evolved from a condition where incubation was parametersthat may have contributed to the lacking,as in crocodiliansand possiblynonavi- origin of biparentalcare in the variouslineages an dinosaurs;van Rhijn, however, limited his in which it arose.My study is not an attempt discussion to birds. to poseand answer suchquestions, but rather Van Rhijn (1990) attempted to support his to begin to constructa historical framework view primarily by using Cracraft's(1981) clas- within which suchquestions may be explored. sificationof birdsand Strauch's(1978) phylog- The hindlimb musculature of birds has been eny of the order Charadriiformes (shorebirds usedeffectively to generatecladistic hypotheses and allies). Testing van Rhijn's intriguing hy- for restrictedlineages (e.g. Old World subos- pothesis was somewhat hampered at the time cines;Raikow 1987), and has recently been an- by the unavailabilityof cladisticanalyses of avi- alyzed for a large and diversegroup of birds an character data, with which the behavioral (McKitrick 1991). The latter data set was chosen data could be either comparedor combined to to complementthe behavioraldata summarized generatea detailed hypothesisabout the evo- here becauseit is the most taxonomicallycom- lution of parental care in birds. Furthermore, prehensiveset of characterdata available for 830 MiddyC. MCKIT•CK [Auk,Vol. 109 birds. No assumptionsare made here about the number of trees and thereby reduce the number of independenceof the hindlimbmusculature and assumptionsthat were necessaryfor formulatinghy- parental-carebehavior. potheses about character evolution, I examined each of thesegroups of "same"taxa and arbitrarily elim- METHODS inated all but one of the taxa in each group.