sign in sign up
<<
Home , Ammoperdix, Chrysolophus, Colinus, Francolin, Francolinus, Galloperdix

The Auk 109(1):24-42, 1992

PHYLOGENETIC, TAXONOMIC AND BIOGEOGRAPHICAL IMPLICATIONS OF GENETIC, MORPHOLOGICAL, AND BEHAVIORAL VARIATION IN (: )

TIMOTHY M. CROWE,• ERIC H. HARLEY,2 MARIOLAB. JAKUTOWICZ,•'2 JORISKOMEN, 3'• AND ANNA A. CROWE1 •FitzPatrickInstitute, University of CapeTown, Rondebosch 7700, South ; 2Departmentof ChemicalPathology, University of CapeTown, Rondebosch7700, South Africa; and 3Departmentof ,State Museum of ,Box 1203, Windhoek, Namibia

ASSTRACT.--Westudied restriction-fragment length polymorphisms (RFLPs) in mitochon- dria! DNA for 13 speciesof African franco!ins(Francolinus spp.) and the JapaneseQuail (Coturnixc. japonica). Phylogenetic analyses of RFLPsfor these14 speciesand of morphological and behavioralcharacters for the 41 francolinspecies and other perdicinetaxa do not confirm the monophylyof Francolinusas currently recognized.Analyses of morpho-behaviora!char- acters suggestthat Francolinusconsists of at least four major assemblages:the five Asiatic ;two groupsof Africanquail-like species; and the Africanpartridge-like species. Within theseassemblages, analyses of RFLPsand/or morpho-behaviora!characters support the mono- phy!y of six of eight speciesgroups attributed to Francolinus.Assuming the monophy!yof currently recognizedsupraspecific groups of ga!!iformbirds, morphometricanalyses of gal- !iform skeletonscorrectly classified 90-99% of specimensto , subfami!yand tribe, as well as 95% of the specimensto . Genetic distancesderived from RFLP data imply thatAfrican francolinsdiverged from their sistertaxa at or beforethe mid-lateMiocene, and that a!! speciesstudied diverged from their sister-speciesduring the Plioceneor early .Received 29 June1990, accepted 13 July 1991.

THESUPRASPECIFIC phylogenetic relationships skeletalanatomy has led to a proliferation of amongmembers of the family Phasianidae(sen- small genera(mean phasianidspecies-to-genus su Morony et al. 1975) remain poorly resolved ratio = 3.3; range = 1-41; Morony et al. 1975) (Verheyen 1956,Cracraft 1981,Stock and Bunch of uncertainphylogenetic affinity (Olson 1985). 1982,Gutierrez et al. 1983,Sibley and Ahlquist The most striking exception to this "small- 1985,Helm-Bychowski and Wilson 1986,Crowe genus rule" in the Phasianidaeis the genus 1988,Randi et al. 1991).A primary causeof this Francolinus.The francolins form the largest ge- phylogenetic uncertainty is that, while the di- nus in the (Morony et al. 1975)and agnostic morphological and behavioral attri- one of the largestgenera in the classAves (Bock butes (hereafter termed morpho-behavioral and Farrand 1980). Hall (1963:109;Fig. 1A) rec- characters)of phasianidsupraspecific taxa tend ognized 41 speciesof francolins (36 African, 5 to be highly divergentqualitatively, differences Asian) and concluded that francolins form a in the skeletalanatomy of thesetaxa tend to be single, monophyletic genus, the members of subtle and continuous.For example, although which are distinguishablefrom other members Steadman(1980) was able to assignavian fossil of the Perdicini (sensuMorony et al. 1975) by to various speciesof turkeys (Meleagri- a long, hooked bill, a short tail of 14 , dinae) by a relatively simple morphometricap- an upright stanceand, in the majorityof species, proach,only a few of the more than 100 skeletal spurs at least in the males. charactershe employed qualitatively distin- Hall (1963:110)proposed the monophyly of guish turkeys from other phasianids.Because Francolinusand then partitioned all but four osteologicalcharacters are among the more im- species(F. lathami,F. pondicerianus,F. nahani, and portant diagnosticfeatures used to assignspe- F. gularis)of this assemblageinto eight mono- ciesto avian genera, this marked divergence in phyletic groups,seven of which have represen- external morphology and relatively conserved tatives in Africa (Fig. 1A). Hall (1963:170)fur- 24 January1992] EvolutionofFrancolins 25

QUAIL-FRANCOUNS -FRANCOLINS 11 26 C•ow• ETA[. [Auk, Vol. 109 ther suggested that the genus may have and Milstein and Wolff's (1987) - and par- originatedin Asiaduring the (ca. 25- tridge-francolins(Fig. 1A); (2) determine the 35 x 106y.b.p.), and that extant speciesof Af- degreeof geneticvariation within and between rican francolins diverged from their sister spe- certain(primarily southern)African francolins; cies perhaps as recently as the last 10,000 to (3) estimatethe evolutionarydivergence times 100,000 years. of the genusFrancolinus and its componenttaxa; Milstein and Wolff (1987) argued for parti- (4) comment on the taxonomic and biogeo- tioning Francolinusinto two major com- graphical implications of these results;and (5) prising quail-like and partridge-likebirds (Fig. assessthe correlation between speciesgroup- 1A). "Quail-francolins"(called patryseby Mil- ings suggestedby morphometric, morpho-be- stein and Wolff 1987) are generally small, havioral,and geneticinformation for galliforms ground-roostingbirds with striped and barred in general and francolins in particular. rufousdorsal resembling that of (Coturnixspp.); they have high-pitched, tonal METHODS AND MATERIALS calls. "Partridge-francolins"(called fisanteby Milstein and Wolff 1987) are generally larger, MtDNA datacollection and analysis.-- tissue was tree-roosting birds with dark dorsal plumage excisedfrom one specimenof the JapaneseQuail and vermiculatedwith white or buff; they give low- eachof the 13 African francolin species(Table 1) and frozen in liquid nitrogen. Tissue from additional er-pitched, raucouscalls (Milstein and Wolff specimensof F. africanusand F. levaillantiiwas col- 1987).Taxonomically, following Wolters(1975 ), lectedat sites400-700 km from the original collection Milstein and Wolff split the southern African localities(Table 1) to assessgeographical variation in speciesinto four genera (Fig. 1A). mtDNA structure. In a preliminary phylogeneticstudy of fran- MtDNA from each specimenwas extractedfrom colin morphology and behavior, Crowe and about 4 g of whole tissue (Brown 1980, as modified Crowe (1985) failed to corroborate the mono- by Densmoreet al. 1985). We used 14 restrictionen- phyly of Francolinus.However, they confirmed donucleases:EcoRI, ScaI, SacI, SacII, PvuII, StuI, HindIII, the monophyly of Hall's Spotted,Red-winged, NcoI, BamHI, BclI, PstI, EcoRV, Asp 718, and XbaI. Red-tailed,Bare-throated and Montane groups, Conditions for restriction-endonucleasedigestions were those suggestedby the suppliers(Amersham as well as that of Milstein and Wolff's (1987) International,Boehringer Mannhelm, New England partridge-francolins(Fig. 1A). Crowe and Crowe Biolabs).We end-labeledmtDNA fragmentswith 32p_ (1985) also concludedthat quail-francolinsare dCTP by incubation with the Klenow fragment of a paraphyletic assemblage.Based primarily on DNA polymeraseI at 25øCfor 10 min to exposead- ontogenetic information, Crowe and Crowe ditional nucleotidesfrom fragmentswith blunt ends (1985) hypothesized that the ancestor of the or 3' overhangs.The fragmentswere incubatedfor francolinswas a small, quail-like phasianid,be- an additional 10 min after addition of unlabeled DTTP, causeplumage and other integumentary fea- dATP and dGTP plus 32P-dCTP.The labeled frag- tures (e.g. bill and tarsus color) of immature mentswere separatedby electrophoresisthrough 1.2% francolinsare remarkablyquail-like (Crowe et horizontal agarosegels in 1 x TAE buffer that in- cluded0.05% pyrophosphate. Gels were visualizedby al. 1986). Taxonomically,on the strength of autoradiography.Lambda phage DNA digestedwith overall morphometric osteologicalsimilarities HindIII was used as a molecular-weightmarker. amongfrancolins, Crowe and Crowe (1985)pro- We assessedrestriction-fragment length polymor- visionally kept the francolinsin one genus,but phism (RFLP) in mtDNA from restriction-endonu- proposeda systemof subgenerasimilar to that cleasefragment patterns, assumingthat fragments of Wolters (1975). with the same electrophoreticmobility are homolo- In this paper, we discussstudies of: (1) the gousbetween haplotypes.The percentageoverall nu- mitochondrial DNA (mtDNA) of 13 speciesof cleotidedivergence (b) between haplotypeswas es- African francolins and of the JapaneseQuail timated in a pairwisemanner by the iterative method (Coturnixcoturnix japonica); (2) the morphology of Nei (1987). Due to the large number of francolin speciesex- and behavior of the 41 speciesof francolins and amined, it was not feasible to map restriction sites. a range of other perdicine species;and (3) the Therefore, we used RFLPs as characters. We realize morphometricsof a representativerange of gal- that thisinvolves a potentialloss of phylogeneticin- lifotins (including 25 Francolinusspecies). Our formation (Swoffordand Olsen 1990),but RFLPsare aimsare to: (1) reassessthe monophylyof Fran- alsolegitimate synapomorphies in phylogeneticanal- ,Hall's (1963) francolin speciesgroups, ysis(Zink and Avise 1990).Therefore, phylogeneti- January1992] Evolutionof Francolins 27

TAISLE1. Speciesand sourcesof specimensused in mtDNA analysis.

Species Locality Coturnixc. japonica Domestic . F rancolinus leucoscepus Athi River District, . levaillantii Sable District, Transvaal, and Giants Castle Nature Reserve, Natal. levaillantoides Balfour District, Transvaal. natalensis Sable District, Transvaal. adspersus Omaruru District, Namibia. sephaena Nylstroom District, Transvaal. Uitenhage District, Cape Province. shelleyi Nylstroom District, Transvaal. africanus Ceres and Molteno Districts,Cape Province. swainsonii Nylstroom District, Transvaal. capensis Cape Town District, Cape Province. coqui Nylstroom District, Transvaal. hartlaubi Omaruru District, Namibia.

cally informative mtDNA RFLPs (i.e. excluding au- culated a Nelson strict-consensus tree. The RFLP-based tapomorphiesand universally sharedfragments) for geneticdistances also were analyzedphylogenetical- the JapaneseQuail and the 13 African francolin spe- ly following the approachof Fitch and Margoliash cieswere scoredas present or absentfor each , (1967)using the FITCH programin Felsenstein's(1987) and polarized by using the JapaneseQuail as an out- PHYLIP (version 3.1). In this analysis,the Japanese group. Quail was the outgroupand the global searchoption Morpho-behavioralcharacters and polaritydecisions.- (G) was invoked to ensure that the minimum-length We analyzed 34 morpho-behavioralcharacters (Ap- tree was found. pendix 1 and Table2) for the 41 francolinspecies and In the analysisof the combinedRFLP and morpho- 9 speciesof perdicines(Margaroperdix rnadagarensis, behavioral character data for the 13 African franco- Rhizotheralongirostris, Caloperdix oculea, lins, we measuredthe congruencebetween the two spadicea,Lerwa lerwa, graeca, Ptilopachus petro- charactersets with the Mickevich-Farris incongru- sus,Barnbusicola thoracica and Arborophilatorqueola) that ence metric (iMr; Mickevich and Farris 1981; Kluge possesssome or all of the diagnosticcharacters offered 1989), which is calculated as: for the genusFrancolinus by Hall (1963).Morpho-be- ESc- (ES, + ... + ES. havioral character information was obtained from Hall iM• = x 100, (1963), Sch6nwetter (1967), Crowe and Crowe (1985), ESc Crowe et al. (1986), Milstein and Wolff (1987), Johns- where the values of ES1to ES,are the number of extra gard (1988), G. E. S. Robbins (unpubl. data), P. le S. stepsfor the data setswhen analyzed separately,and Milstein (unpubl.data), and D. Marais(unpubl. data). ES• is the number of extra steps for the analysisof Unlessotherwise stated in Appendix 1, characterpo- the combined data set, with extra steps being the larity was determined using ontogeneticcriteria. Our length of the calculated tree minus the minimum assumptionwas that quail-like featuresare plesiom- length (i.e. with no homoplasy)of the tree for the orphic for nonquail perdicines. data set in question.The iMris the percentageof the Phylogeneticmethods.--We analyzed charactersets total incongruencethat results from a lack of con- using Wagner parsimony(Farris 1970) with the pro- gruence between the data sets. We chose this ap- gram Hennig86 (version1.5; Farris 1988). For analyses proach to assesscharacter-set congruence because of morpho-behavioralcharacters for the 41 Francolinus others(e.g. Simberloff 1987)do not provide measures speciesand the 9 other perdicine speciesmentioned of incongruencethat result from combining character above,we employed the "rnhennig*bb*" tree-search- sets. ing commands.Although this strategy is not guar- We assessed the robustness of trees for the RFLP, anteed to find the tree(s) of minimal length, it con- morpho-behavioralcharacter, and combineddata in structsseveral trees,adding terminal taxa in different three ways. First, the charactersupport of each node sequences,and applies branch-swappingto each of leading to at leasttwo taxain eachtree (or the Nelson the initial trees,saving as many equally-parsimonious strict-consensustree if more than one tree was found) treesas the availabletree spacecan hold (Farris 1988). was determinedby collapsingthat node into a poly- In all analysesof the 13 African Francolinusspecies, furcation using the "\ \x;" command in "Dos Equis- the shortest-possibletree(s) was found by implicit mode" within Hennig86. Second,we examined the enumeration(Farris 1988). In all analysesthat pro- same nodes for each tree and determined the number duced multiple equally-parsimonioustrees, we cal- of unique and unreversedsynapomorphies that sup- 28 CROWEEl'At.. [Auk,Vol. 109 January1992] Evolutionof Francolins 29

portedeach node. Third, for analysesof African fran- colinsand the JapaneseQuail only, we usedthe boot- strapping program BOOT (with 100 replicates) in PHYLIP. BOOT repeatedlyresamples characters ran- domly with replacement(Felsenstein 1985) and then performsa parsimonyanalysis on eachpseudorepli- cate data set, ultimately identifying the frequency with which the topologicalresults agree with that producedby a majority-ruleconsensus tree. In BOOT analyses,all multistate characterswere recoded in an additive binary fashionto preservepolarity infor- mation. We did not attempt to analyze the 42- and 51-taxondata setsusing this method,because of the prohibitively large number of taxa involved. Morphometricanalyses of extantfrancolins.--To assess the utility of the "nearest-neighbor"morphometric approachemployed by Crowe and Crowe (1985) in assigninggalliform skeletons to monophyleticgroups and to determinethe correlationbetween morpho- metric and phylogeneticrelationships in galliform birds, we made 73 measurements(Fig. 2; Appendix 2) on 146skeletons of galliforms,including 99 species (25francolins) and 67genera (18 perdicine; Appendix 3). We included skeletons from both males and fe- males. Unstandardized mensural data were log•0- transformedand analyzedusing BMDP2M, a cluster- analysis program (Dixon 1985), to determine the morphometric"nearest neighbor" of each skeleton. By morphometricnearest neighbor, we mean the specimenwith the shortestEuclidian distance to the skeletonunder study as indicatedby the distance matrix output by BMDP2M prior to dendrogramcon- struction.If this approachhas utility in identifying natural groups, membersof highly corroborated monophyletic groups should have members of the same groups as nearest neighbors (i.e. quails should havequails as nearest neighbors, turkeys should link with turkeys, with grouse,and francolinswith francolins).

RESULTS

MtDNA results.--A total of 211 distinct re- strictionfragments was produced by the 14 re- striction enzyme digestsof francolin and Jap- anese Quail mtDNAs. Fragment raw data are availablefrom EHH on request.Ninety-nine of thesefragments were phylogeneticallyinfor- mative (Table 3). The • values(Table 4) for two intraspecific comparisons(0.3 between two F. africanusand 0.1 between two F. levaillantii collected hun- dredsof kilometersapart) were an of mag- nitude lower than thosefor nearly all the in- terspecificcomparisons. Therefore, birds from only one locality for each species(Ceres for africanusand Sabiefor levaillantii;Table 1) were 30 CROWEET ^L. [Auk, Vol. 109

13 9 8 22 / 7 6 31-..., 30 87 11 2510

10

18..• 17

b c d

6 7.5 77 /91

• ! 69 68 82 9 95

45---,• /,"t,•\ •--44

7/• n •øø q • 5

52 64 63 50

52

51

•53 F•õ. 2. Bo•es measuted [• motphome•ic studies.Re•ete•ce pendix 2. used in phylogeneticanalyses. The &values be- francolin was 6.4 (between F. levaillantii and F. tween pairs of francolin speciesranged from natalensis).The mean &values between the Jap- 2.0 (between F. capensisand F. natalensis)to 14.9 aneseQuail and quail- and partridge-francolins (between F. africanusand F. hartlaubi).The low- were 9.5 and 8.9, respectively. est 6 between a quail-francolinand partridge- Phylogeneticanalyses.--The Hennig86 analy- January1992] EvolutionofFrancolins 31 32 CROWEET At,. [Auk,Vol. 109

T^I•I,E4. Matrixof estimatesof percentnucleotide divergence (•; lowerhalf matrix)and proportion of shared mtDNA restrictionfragments (upper half matrix;Nei 1987)for Francolinusspecies.

Quail-francolins Partridge-francolins afr levo she coq sep levi har nat ads cap afer swa leu cot Quail-francolins africanus .521 .508 .225 .448 .154 .097 .123 .105 .113 .211 .123 .206 .212 levaillantoides 3.8 .606 .189 .257 .265 .123 .167 .177 .135 .203 .176 .197 .203 shelleyi 4.0 2.9 .250 .400 .207 .109 .135 .145 .156 .232 .172 .197 .169 coqui 9.2 10.3 8.5 .382 .212 .127 .171 .182 .194 .234 .152 .174 .239 sephaena 4.7 8.3 5.4 5.7 .258 .169 .128 .184 .176 .219 .161 .185 .222 levaillantii 11.7 8.1 9.7 9.6 8.3 .316 .342 .310 .212 .282 .233 .317 .262 Partridge-francolins hartlaubi 14.9 13.3 14.1 13.0 11.1 6.9 .247 .235 .286 .206 .140 .200 .276 natalensis 13.3 11.2 12.6 11.0 13.0 6.4 8.5 .621 .707 .414 .316 .481 .208 adspersus 14.4 10.8 12.1 10.6 11.3 7.1 8.9 2.7 .597 .439 .338 .459 .250 capensis 13.8 12.6 11.6 10.1 10.8 9.6 7.6 2.0 3.0 .442 .333 .493 .269 afer 9.6 9.8 9.0 8.9 9.3 7.7 9.7 5.2 4.9 4.8 .479 .649 .222 swainsonii 13.3 10.8 11.0 11.8 11.4 8.9 12.4 6.9 6.5 6.6 4.3 .508 .164 leucoscepus 9.7 10.0 10.0 10.9 10.5 6.9 9.9 4.3 4.6 4.1 2.5 4.0 .281 Coturnixc. japonica 9.6 9.8 11.1 8.8 9.2 8.2 7.8 9.7 8.5 8.0 9.2 11.3 7.7

sisof morpho-behavioralcharacter data for the 3), all but 2 of the 99 species(98%) examined 41 Francolinusspecies and 9 other perdicinetaxa had morphometric nearestneighbors from the "overflowed" after producing 1,232 equally- same family; all phasianidshad phasianidsas parsimonioustrees. The Nelson strict-consen- their nearest neighbors. Within the Phasiani- sustree is shown in Figure lB. A similar analysis dae, skeletonsfrom 81 of the 86 (94%) species restricted to francolins overflowed after pro- examinedhad nearestneighbors from the same ducing 1,486 trees (consensustree in Fig. 1C). subfamily.Within the ,skeletons of A third morpho-behavioralanalysis restricted 56 of 62 (90%) specieshad nearest neighbors to the 13 African Francolinusspecies (from which from the samesubfamily. Within the Perdicini, RFLP data were obtained)produced four trees all but one of the 83 (99%) skeletons from 43 (consensustree in Fig. 3A). Analysis of the 99 species(98%) had a perdicine nearestneighbor. phylogenetically informative mtDNA frag- Among francolins, 61 of the 64 (95%) skele- ments for the JapaneseQuail and the 13 Fran- tons and 22 of the 25 (88%) francolin species colinusspecies produced three equally-parsi- studied had francolin nearest neighbors. The monious trees (consensustree in Fig. 3B). The mean nearest-neighbormorphometric distance analysisof the combined mtDNA and morpho- amongthe 25 francolin species(0.34, SD = 0.06, behavioral character data set for the 13 African range 0.26-0.52), was significantlylower (P < francolinsproduced one tree (Fig. 3C) with an 0.0001, Mann-Whitney U-test) than that for iMr of 3.5%. comparisonsof francolinsand the 10perdicines The FITCH analysis of the genetic-distance which had francolinsas their nearestneighbors results(Fig. 4) examined554 trees and differs (0.46, SD = 0.07, range 0.38-0.57). Among from the Wagner-parsimonyanalysis (Fig. 3B) monophyleticspecies groups of francolins(Figs. in threerespects. First, FITCH groupedF. swain- 1 and 3), 41 of 46 (89%) skeletonsfrom par- sonii with the other two members of Hall's tridge-francolins(including all five speciesof Bare-throatedGroup. Second, it linked the quail- Asiatic francolins)had partridge-francolinsas francolin F. levaillantii with F. hartlaubi,a par- their nearestneighbors, but only 11 of 18 (61%) tridge-francolin.Third, it placedF. sephaenawith quail-francolins had quail-francolins as their the three membersof Hall's Red-wingedGroup, nearest neighbors. In Hall's (1963) species and not with F. coqui. groups,the classificationsuccess rate for skel- Morphometricanalyses.--The raw galliform etons was 6 of 9 (66%) for Asiatic francolins, 10 mensural data are available from TMC. In the of 15 (66%) for skeletons from members of the BMDP2M nearest-neighboranalysis (Appendix Bare-throatedGroup, 4 of 16 (25%) skeletons January1992] EvolutionofFrancolins 33

QUAIL-FRANGOLINS PARTRIDGE-FRANCOLINS 1[ o •

I l 34 CROWEEr ^•,. [Auk,Vol. 109

3,7 Cotumix colins(fisante) would emergeas monophyletic groups. •.• • 2.1sephaena Morpho-behavioralcharacter data also sup- L• 1.s ' a/rio=us• RED-WINGED port the monophylyof at leastfour majormono- '•shelley,ß I GROUPMINUS phyletic groups among the francolins (Figs. lB • levaillantoid•_JLEVAILLANTII and 1C):(1) Hall's (1963)Spotted Group of fran- colins, plus the two other Asiatic francolins 3.7coqui (pondicerianus and gularis);(2) Hall's Red-tailed hartlaubi Group plus F. lathami;(3) Hall's Red-winged Group plus F. streptophorus;and (4) a large as- natalensis• VERMICULATEDGROUP 1• • capensisI (SOUTHERN) semblageof partridge-francolins,including at 1.7 HARTLAUBI least three monophyletic groups that corre- spond to Hall's Scaly, Bare-throatedand Mon- 3o afer BARE- tane groups.RFLP analysis (Figs. 3B and 4) sup- ports the monophyly of three of the five o.• swainsoniileucoscepuss• THROATEDGROUP southern African members of Hall's Vermicu- SUM OF SQUARES = 2.36 lated Group (adspersus,natalensis and capensis). MEAN % STANDARD DEVIATION = 11.20 We provisionallyaccept the monophylyof these NUMBER OF TREES EXAMINED = 554 threespecies, plus F. hildebrandti,which (in male Fig. 4. Tree (rooted on JapaneseQuail) based on adults)closely resembles and is distributedpar- Fitch-Margoliash algorithm applied to matrix of apatricallywith natalensis(Crowe et al. 1986). mtDNA • values in Table 4. Numbers above nodes Furthermore,although there asyet are no syn- are branch lengths. apomorphiesto unite the four remaining mem- bers of Hall's Vermiculated Group from north- ern and eastern Africa (bicalcaratus,clappertoni, from membersof the Vermiculated Group, and icterorhynchus,harwoodi), their strong overall 6 of 6 for skeletons from members of the Red- similarity and parapatricdistributions (Crowe winged Group. At the specieslevel, in 69 in- et al. 1986)suggest that they also form a mono- stances for which more than one individual of phyletic group within the partridge-francolin the samespecies were analyzed,57 (83%)of the assemblage. nearest neighbors were conspecificsand the Two francolins(F. sephaenaand F. nahani)fall nearest neighbors of the remaining 12 were outside the major francolin clades, grouping congeners. paraphyleticallywith an assemblageof primar- ily Indo-Malaysian perdicines(Fig. lB). RFLP DISCUSSION data (Fig. 3B) and measuresof genetic distance for the 13 speciesof African francolins•(Table Francolinsas a monophyleticgroup.--Phyloge- 4, Fig. 4) suggestthat F. sephaenahas affinities netic analysesof a range of morpho-behavioral with F. coquiand other quail-francolins (e.g. characters,including those traditionally used to F. africanus)of Hall's Red-wingedGroup. Fur- distinguish francolins from other perdicines thermore, if all 41 francolin speciesare ana- (Appendix 1, Fig. lB), do not confirmthe mono- lyzed with the RFLP data included for the 13 phyly of Francolinus.If Figure lB is modified African species,sephaena joins the Red-tailed/ minimally to supportthe monophylyof Fran- lathamiGroup. colinus(i.e. if nonfrancolins that cluster within The affinities of F. nahani,which is virtually the francolin assemblageare shifted out to form unknown biologically, remain obscure. We a basalpolytomy), the resulting tree is 19 steps agree with Hall (1963) that its small size, dark longer.Moreover, if the francolinsare analyzed colorationand spotted plumage are probably alone, two major cladesemerge (Fig. 1C), one not indicative of affinities with F. lathami, be- comprisedof quail-francolins(minus pondice- causeall three attributesare commonadapta- rianus)and the other of partridge-francolins. tionsto living in tropical-forestconditions. One Thus, should Francolinusprove to be a mono- possibilityis that nahaniis not a francolin and, phyleticassemblage, Milstein and Wolff's(1987) perhaps like several other Afrotropical forest quail-francolins (patryse)and partridge-fran- taxa(e.g. Afropavo, Tigriornis, Pseudocalyptomena, January1992] EvolutionofFrancolins 35

Phodilusand Hirnantornis), it is a relictual form testestwo to three timesthe size of thoseof any mostclosely related to an Indo-Malaysiantaxon other francolin we have examined (T. M. Crowe (Olson 1973). However, we agree with Hall and J. Komen, unpubl. data). The genetic, mor- (1963:166-167) that a combination of features phometricand morpho-behavioraldata suggest (e.g.lack of sexualdimorphism, red tarsi,white- that, if it is a francolin, Hartlaub's Francolin is streaked belly, crimson-basedbill, and bare a product of an early divergence within the skin below and behind the eye) place nahani Francolinuslineage. with the partridge-francolinclade, possibly with The relatively isolatedgenetic position of the Hall's Scaly Group. Redwing Francolin (F. levaillantii)is much more Geneticvariation.--In two speciesfor which difficult to explain. Its unresolved or basal po- RFLPsof individuals from widely separatedlo- sition within the African francolins (Fig. 3B) calities were available, we found intraspecific and its apparentgenetic affinities to partridge- divergence values of an order of magnitude francolins(Fig. 4, Table4) wereunexpected giv- lower than the • values between even the least en that Hall (1963) named her Red-winged divergent pairs of francolins(• = 0.3 between Group (Fig. 1A) after the common name of this two F. africanusand • = 0.1 between two F. typical quail-francolin. Nevertheless,the Red- levaillantii).This level of mtDNA divergenceis Francolin's first five genetic nearest similar to that found within other bird species neighbors are partridge-francolins (Table 4). (Shields and Helm-Bychowski 1988) and mam- mals (Wilson et al. 1985). This genetic similarity between a quail-fran- At the interspecificlevel, 11 of the 13 fran- colinand the morphologically,behaviorally and colins form four clusters(coqui/sephaena; levail- ecologicallydistinct partridge-francolins can be lantoides/ shelleyi / africanus ; adspersus/ natalen- explainedby three hypotheses.First, the • value sis/ capensis; and leucoscepus/ afer / swainsonii)of betweenthe Redwing Francolinand its nearest geneticallysimilar species(• = 2-5.7; œ= 3.4). geneticneighbor, F. natalensis,is 6.4;F. natalensis The geneticdistances between members of dif- is a partridge-francolinand, perhapsas recently ferent clusterstend to be much larger (• = 8.9- as 3 x 106y.b.p., there was gene flow between 14.4;œ = 9.4), about the samemagnitude as dif- quail- and partridge-francolinssuch that levail- ferencesbetween francolins and the Japanese lantiimales hybridized successfully with female Quail. partridge-francolins,and "partridge" mtDNA The two remaining francolin species,hartlaubi introgressedinto the levaillantiilineage through and levaillantii,appear to be relatively distantly subsequentback-crossing of fertile fe- related to the other francolins studied (• = 6.4- maleswith levaillantiimales. Alternatively, le- 14.9).The remoteplacement of one of thesetwo vaillantiirepresents a stem speciesthat possesses "outsiders,"F. hartlaubi,is not surprising given an ancestralquail-like phenotypeand mtDNA that, in many other ways, it is the most diver- haplotype.Finally, levaillantiiis either a par- gent francolin included in this study. It is the tridge-francolin that convergently acquired most distinct francolin morphometrically (Ap- quail-francolinmorphology and behavior,or a pendix 3), with a Euclidian distance (0.52) to quail-francolinthat convergentlyacquired par- the nearestfrancolin much greater than the next tridge-francolinRFLPs. These hypothesescan largest nearest-neighbor distance (0.46) be- be testedthrough further study of mtDNA by tween pairs of francolins. From a morpho-be- restriction-sitemapping and sequencing,as well havioral perspective,it is a basal taxon in the as by study of the nucleargenome (e.g. using partridge-francolinassemblage (Figs. lB, 1C,and DNA-DNA hybridization, restriction-frag- 3A). Unlike other francolins, it is found in a ment/site analysis,or protein electrophoresis) highly specifichabitat (isolated,rocky outcrops and other charactersystems (e.g. the syrinx). surrounded by subdesertsteppe) and has an Our RFLP resultstend to favor the stem-spe- extremely complex,antiphonal advertisement cieshypothesis. Of the 11 RFLPsin F. levaillantii call (Komen 1987). Osteologically,its morpho- not shared with , three are shared with metric nearestneighbor is the quail-like (Frost both quail- and partridge-francolins,and four 1975) Partridge (Margaroperdix each with only quail-francolinsand and par- rnadagarensis;Appendix 3). Like Coturnixspp., tridge-francolins.Furthermore, levaillantii lacks malesof F. hartlaubihave extremelylarge, ovoid most of the RFLPsthat are synapomorphiesfor 36 CROWEET AL. [Auk, Vol. 109 the Red-wingedGroup (e.g. characters6, 25, trio, is much closerto leucoscepusthan to swain- and 95 in Table 3). sonii.Perhaps F. afer evolved even more re- Divergencetimes.--Applying the estimateof a cently in central Africa from a relictual popu- mean rate of divergence of 2 • units per 10• lation of proto-leucoscepus.Assuming that F. years (Brown et al. 1982, Shields and Wilson hartlaubi,one of the manybird speciesendemic 1987,Shields and Helm-Bychowski1988), which to southwesternAfrica (Crowe and Crowe 1982), is basedon fossil-calibratedstudies of primates evolvedin situ,its relatively remotegenetic dis- and a rangeof avian taxa(including galliforms), tancefrom other partridge-francolinssuggests the francolinsappear to havediverged from the that there was at least a third, much more an- quail lineage at least 3.8 x 106y.b.p. The most cient, bout of francolin speciationin arid south- recent speciation event was at least 106years ern Africa. Thus,the openingof the "arid cor- ago. These divergencetimes agree with those ridor" between east and southern Africa suggestedby Helm-Bychowski and Wilson (Winterbottom 1967) was the vicariance event (1986)for othergroups of phasianids.Our min- that led to the speciationof thesebare-throated imum estimatefor the age of the genusis sup- francolins (Crowe et al. 1986). It occurred much ported by the existenceof a well-differentiated more recently than the closure of the "corri- francolinin southernAfrica nearly 5 x 10• y.b.p. dor," which may have promoted speciation (Crowe 1992). Thus, our results are consistent within the mole-rats and of F. hartlaubi. Hall with Hall's hypothesesof an ancientorigin for (1963) has already commentedon the impor- Francolinus. However, if mid- fossil hu- tance of the corridor as the commonboundary meri from Arrisdrift, Namibia, prove to be from between the "black-and-white" and "vermic- a primitive francolin (Crowe 1992),this diver- ulated" subspeciesgroups within F. afer. gencemight not have been as long ago as the Amongthe morequail-like membersof Hall's Oligocene(Hall 1963,Sibley and Ahlquist1985). Red-winged Group, our results suggesta bio- At the interspecificlevel, our resultsdo not sup- geographicalscenario similar to one Hall (1963: port Hall's hypothesisthat there mayhave been 158-160)proposed. Assuming that F. levaillantii a majorbout of speciationin African francolins is a memberof the Red-wingedGroup, the an- during the late Pleistocene, or Sibley and cestral quail-like francolin could have been a Ahlquist's(1985) suggestion that F. capensisand widespreadtaxon, and present-dayextant taxa F. natalensisdiverged from one another9 x 106 that are more localized could have speciatedin y.b.p. situwhen isolatedduring periodsof long-term, Biogeography.--ShouldFrancolinus prove to be climatic fluctuations. During such periods, a monophyleticassemblage, our results(Fig. lB) membersof the Red-winged Group, most of do not supportHall's hypothesisthat the genus which seemto be generallyadapted to cooler evolved first in . A more likely scenariois climaticconditions (Crowe et al. 1986),may have that the genusevolved in Africa, with an early been isolated (and subsequently speciated) offshootof the partridge-francolinlineage sec- within relatively cool, disjunctly distributed ondarily dispersinginto and, subsequently,di- highlands.If Hall (1963) was correctin assum- verging within Asia. With regard to intra-Af- ing that F. levaillantoidesevolved in eastAfrica, rican francolin biogeography,a comparisonof its low fi value (2.9) to its sister-species(F. shel- fi values for francolins and African mole-rats leyi)suggests that it only recentlyspread to, and (Honeycuttet al. 1987)suggests that vicariance became secondarily isolated in, southwestern eventsthat promotedspeciation in thesegroups Africa. were not contemporaneous.For example,the fi .--Previousresearchers (e.g. Ogil- values between the endemic east African mole- vie-Grant 1896, Hall 1963, Crowe and Crowe rats (Heterocephalusglaber and Heliophobiusar- 1985) have lumped the francolinsinto one or, gentocinereus)and their endemic southern Af- at most, two genera becausethere are "linking rican sistertaxa range from 14 to 28 (i.e. 7-14 taxa"that form a "gradedseries" (Chapin 1926) x 106y.b.p.), whereas that between eastAfrican between otherwisedistinct subgroupings(i.e. endemicF. leucoscepusand the southernAfrican Hall's groupsin Fig. 1). For example,F. sephaena endemics(F. aferand swainsonii)are 2.5 and 4.0, is the key taxon that links the quail- and par- respectively.This implies a much more recent tridge-francolins from a morpho-behavioral divergence.Furthermore, F. afer,the geograph- perspective(Figs. lB and C), becauseit has a ically intermediatespecies in this bare-throated quail-francolinbill and plumage.However, it January1992] EvolutionofFrancolins 37 also has red tarsi and roostsin trees like many membersof the subgenusSquamatocolinus differ partridge-francolins.Nevertheless, in the light from otherPternistis species in having dark edg- of the additionalmorpho~behavioral and RFLP ing to their belly plumage (17), which gives characterdata, this francolinfalls decisively with them a scalyappearance (Hall 1963).Members the quail-francolinsand probablywith thoseof of Oreocolinus differ from those of other sub- Hall's Red-tailedGroup. Therefore, we feel that, generain Pternistisby the uniformbrown crown, independentof the questionof monophylyof back,primaries, and tails of the males(character the francolins,Francolinus should be partitioned 13; Hall 1963). into at least four genera: Morphometricsand phylogenetics.--Assuming the correctnessof the morpho-behavioralphy- GenusFrancolinus Stephens 1819 logenies(Figs. lB and 1C) and the highly con- [SubgenusFrancolinus Stephens 1819] gruent phylogeny basedon morpho-behavioral pictus,francolinus, pintadeanus and RFLP characters(Fig. 3C), our results(Ap- [SubgenusOrtygornis Reichenbach 1853] pendix 3) suggest that a morphometric ap- pondicerianus proachcan correctlyassign galliform osteolog- [Subgenusnov. Limnocolinus] ical material to independently derived gularis monophyletictaxa. This finding is in marked contrastto thosefrom other morphometricand GenusPeliperdix Bonaparte 1856 phylogeneticstudies (e.g. Zink and Avise 1990). [SubgenusPeliperdix Bonaparte 1856] Our empiricalresults could be usefulfor studies coqui,albogularis, schlegelii, lathami of fossilgalliforms for which there are few re- [SubgenusDendroperdix Roberts 1922] liable osteologicalsynapomorphies below the sephaena family level (Cracraft1981). For example,Crowe GenusScleroptila Blythe 1849 and Short (1992) and T. M. Crowe (unpubl. streptophorus,finschi, levaillantii, africanus, manuscript) have shown that both morpho- psilolaemus,shelleyi, levaillantoides metric and synapomorphiccharacters place a fossilhumerus from the Oligoceneof Nebraska GenusPternistis Waglet 1832 and Gallinuloideswyomingensis (a fossil galliform [SubgenusAcentrortyx Chapin 1926] nahani from the Eoceneof Wyoming) within the Gal- linuloididae,the sistergroup of the Phasiani- [SubgenusChapinortyx Roberts 1928] hartlaubi dae. This challengesTordoff and Macdonald's (1957) hypothesisthat Gallinuloideswas a cracid [SubgenusChaetopus Swainson 1837] and eliminatesthe needto hypothesizethat the bicalcaratus,icterorhynchus, clappertoni, har- woodi Cracidaeoriginated in the Nearctic and, sub- sequently,dispersed and diversifiedwithin the [Subgenusnov. Notocolinus] Neotropics (Vuilleumier 1965). adspersus,capensis, natalensis, hildebrandti At the genuslevel, we suggestthat the fran- [Subgenusnov. Squamatocolinus] colinsform a relatively homogeneousmorpho- squamatus,ahantensis, griseostriatus metric assemblagewithin the Perdicini (Ap- [SubgenusPternistis Waglet 1832] pendix 3). Indeed, the phylogeneticallymost leucoscepus,rufopictus, afer, swainsonii distinctfrancolins (e.g. hartlaubi,sephaena, coqui, [Subgenusnov. Oreocolinus] andlathami) also are the mostmorphometrically jacksoni,nobilis, camerunensis, swierstrai, cas- distinct francolins(Appendix 3). However, at taneicollis,erckelii, ochropectus the species-grouplevel (=subgenus),the near- The new subgenusLimnocolinus differs from est-neighbormorphometric approach had much the other subgenerain Francolinusin having: poorer successin correctly placing skeletons. long toes(Hall 1963);large-bodied males (mor- In our study, 93% of the skeletonsfrom par- pho-behavioralcharacter 1 in Appendix 1 and tridge-francolins(e.g. membersof Francolinus Table 2); streakedbelly plumage(character 15); and )had partridge-francolinsas their uniformbrownish-grey primaries (character 19). nearestneighbors, and all six skeletonsof Scle- Three of the membersof the subgenusNoto- roptilaspp. had congeners as nearest neighbors. colinus(mtDNA unavailable from F. hildebrandtO No more than two-thirds of the skeletons from are distinguishedby four synapomorphicRFLPs any of the other francolingenera or subgenera (characters 3, 23, 66, and 96 in Table 3). The had nearestneighbors from the samesupra- 38 CROW•ET At. [Auk,Vol. 109 specific taxon. Thus, the congruence between BROWN, W. M., E. M. PRAGER,A. WANG, AND A. C. morphometric and qualitative morpho-behav- WILSON. 1982. Mitochondrial DNA sequences ioral charactersappears to break down below of primates: Tempo and mode of . J. the level of the genus. Mol. Evol. 18:225-239. CHAPIN,J. P. 1926. A new genus Acentrortyxpro- ACKNOWLEDGMENTS posedfor Francolinusnahani Dubois. Auk 43:235. CRACRAI•T,J. 1981. Toward a phylogeneticclassifi- TMC acknowledgesfinancial support from the Uni- cation of the Recent birds of the world ( versityof CapeTown ResearchCommittee, the Frank Aves). Auk 98:681-714. M. Chapman Memorial Fund (American Museum of CROWE,T. M. 1988. Moleculesvs morphologyin Natural History),the ErnstMayr Fund (Harvard Uni- phylogenetics:A non-controversy.Trans. R. Soc. versity), the SchmidtFund (Field Museum of Natural S. Aft. 46:317-334. History), the Jessup-McHenry Fund (Philadelphia CROWE,T.M. 1992. Morphometric researchon Mio- Academy of Natural Sciences),the Deutsche For- ceneand Pliocenefossil phasianids from the west schungsgemeinschaft,and the Core Programme of coastof southernAfrica: Somepreliminary tax- the Foundation for ResearchDevelopment (F.R.D.). onomicand phylogeneticconclusions. Proc. 7th EHH acknowledgesfinancial support from the Na- Pan African Ornithol. Congr. In press. tional Programmesand Core Programmeof the F.R.D. CROWE, T. M., AND A. A. CROWE. 1982. Patterns of TMC thanks the staffsof the bird departmentsof the distribution, diversity and endemismin Afro- American Museum of Natural History, Museum of tropical birds. J. Zool. (Lond.) 198:417-442. Comparative Zoology (Harvard University), Phila- CROWE,T. M., AND A. A. CROWE.1985. The genus delphia Academyof Natural Sciences,Carnegie Mu- Francolinus as a model for avian evolution and seum of Natural History, U.S. National Museum of biogeographyin Africa: I. Relationshipsamong Natural History (SmithsonianInstitution), Field Mu- species.Pages 207-231 in Proceedingsof the In- seumof Natural History,Los Angeles County Natural ternational Symposium on African Vertebrates History Museum,British Museum (Natural History), (K.-L. Schuchmann, Ed.). Museum Alexander Koninklijk Museum voor Midden Afrika, Museum Koenig, Bonn. National d'Histoire Naturelie, Humboldt Museum CROWE,T. M., G. S. KEITH, AND L. H. BROWN. 1986. (University of Berlin), Durban Natural ScienceMu- Galliformes.Pages 1-75 in Birds of Africa, vol. 2 seum, State Museum of Namibia, South African Mu- (E. K. Urban, C. H. Fry, and G. S. Keith, Eds.). seum, and Transvaal Museum for accessto, and loan Academic Press, London. of, specimenmaterial. We also thank the Transvaal CROWE,T. M., ANDL. L. SHORT.1992. A new galli- Nature ConservationDivision and the ForestryBranch naceousbird from the Oligocene of Nebraska of the Department of Environment Affairs for in- with commentson the phylogeneticposition of valuable assistancewith the collection of specimen the Gallinuloididae. ScienceSeries, Los Angeles material. Our specialthanks go to Peter le S. Milstein County Natural History Museum. No. 36. for assistancein the field and for stimulatingdiscus- DENSMORE,L. D., J. W. WRIGHT, AND W. g. BROWN. sionsabout francolin biology, and to Steve Farris for 1985. Length variation and heteroplasmyare suggestionsconcerning methods of phylogenetic frequent in the mitochondrial DNA from par- analysis.Gary Robbins, Peter le S. Milstein, and Daan thenogeneticand bisexual lizards (genus Cne- Marais kindly providedunpublished information on midophorus). 110:689-707. the natural history of francolinsin general and on DIXON, W. J. (ED.) 1985. BMDP statisticalsoftware. their downy plumagein particular.For help in the Univ. California Press,Los Angeles. field we thank hunters and their pointerswho helped FARMS,J.S. 1970. Methods for computing Wagner us to find theseelusive phasianids.We thank George trees.Syst. Zool. 19:83-92. Barrowclough,Richard Brooke,Scott Edwards,Ned FARRIS,J. S. 1988. Referencemanual for Hennig86 Johnson,Carey Krajewski, Peter le S. Milstein, Robert version 1.5.Published by the author. Port Jeffer- Zink, and an anonymousreviewer for their construc- son Station, New York. tive comments. FELSENSTEIN,J. 1985. Confidencelimits on phylog- enies: An approachusing the bootstrap.Evolu- tion 39:783-791. LITERATURE CITED FELSENSTFaN,J. 1987. PHYLIP (phylogenyinference BOCK,W. J., AND J. FARRAND,JR. 1980. The number package),version 3.0. Univ. Washington,Seattle. of speciesand genera of Recentbirds. Am. Mus. FITCH, W. M., AND E. MARGOLIASH. 1967. Construc- Novit. No. 2703:1-29. tion of phylogenetictrees. Science 155:279-284. BROWN,W. M. 1980. Polymorphismof mitochon- FROST,P. G.H. 1975. The systematicposition of the drial DNA of humansas revealedby restriction MadagascarPartridge Margaroperdixmadagasca- endonucleaseanalysis. Proc. Natl. Acad.Sci. USA riensis(Scopoli). Bull. Br. Ornithol. Club 95:64- 77:3605-3609. 68. January1992] EvolutionofFrancolins 39

GUTIf•RREZ,R. J., R. M. ZINK, AND S. Y. YANG. 1983. SHIELDS,G. F., AND A. C. WILSON. 1987. Calibration Genic variation, systematic,and biogeographic of mitochondrial DNA evolution in geese.J. Mol. relationshipsof some galliform birds. Auk I00: Evol. 24:212-217. 33-47. SIBLEY,C. G., AND J. E. AHLQUIST. 1985. The rela- HALL,B. P. 1963. The francolins,a study in speci- tionships of some groups of African birds, based ation. Bull. Br. Mus. (Nat. Hist.) 10:8-204. on comparisonsof the genetic material, DNA. HELM-BYCHOWSKI, K. M., AND A. C. WILSON. 1986. Pages115-161 in Proceedingsof the International Ratesof nuclearDNA evolution in -like Symposium on African Vertebrates (K.-L. birds:Evidence from restrictionmaps. Proc. Natl. Schuchmann,Ed.). Museum Alexander Koenig, Acad. Sci. USA. 83:688-692. Bonn. HONEYCUTT, R. L., S. V. EDWARDS,K. NELSON, AND E. SIMBERLOFF,D. 1987. Calculating probabilities that NEvO. 1987. Mitochondrial DNA variation and cladogramsmatch. A method of biogeographic the phylogeny of African mole rats (Rodentia: inference.Syst. Zool. 36:175-195. Bathyergidae).Syst. Zool. 36:280-292. STEADMAN,D.W. 1980. A review of the osteology JOHNSGARD,P.A. 1988. The quails,, and and paleontology of turkeys (Aves: Meleagridi- francolins of the world. Oxford Univ. Press, Ox- nae). Contrib. Sci. Nat. Hist. Mus. Los Angeles ford. County 330:131-207. KLUGE, A. G. 1989. A concern for evidence and a STOCK, A.D., AND T. D. BUNCH. 1982. The evolu- phylogenetichypothesis of relationshipsamong tionary implications of chromosomebanding Epicrates(Boidae, Serpentes). Syst. Zool. 38:7-25. pattern homologiesin the bird order Galliformes. KOMEN,J. 1987. Preliminary observationsof the so- Cytogenet.Cell Genet. 34:136-148. cial pattern, behaviour and vocalization of Hart- SWOFFORD,D. L., ANDG. J. OrSEN. 1990. Pages411- laub's Francolin.S. Afr. J. Wildl. Res. Suppl. 1: 501 in Molecular systematics(D. M. Hillis and C. 82-86. Moritz, Eds.). Sinauer Associates, Sunderland, MICKEVlCH,M. F., AND J. S. FARRIS. 1981. The im- Massachusetts. plicationsof congruencein Menidia.Syst. Zool. TORDOFF,H. B., AND J. R. MACDONALD. 1957. A new 30:351-370. bird (family )from the early Oligocene MILSTEIN, P. LE S., AND S. W. WOLFF. 1987. The over- of South Dakota. Auk 74:174-184. simplificationof our "francolins."S. Afr. J.Wildl. VERHEYEN, R. 1956. Contribution fi l'anatomie et fi Res. Suppl. 1:58-65. la syst6matiquede Galliformes.Inst. R. Sci. Nat. MORONY, J. J., W. J. BOCK, AND J. FARRAND. 1975. Belg. 32:1-24. Reference list of the birds of the world. American VUILLEUMIER,F. 1965. Relationshipswithin the Cra- Museum of Natural History, New York. cidae. Bull. Mus. Comp. Zool. 134(I):I-27. NEI, M. 1987. Molecularevolutionary genetics. Co- WILSON, A. C., R. L. CANN, S. M. CARR, M. GEORGE, lumbia Univ. Press, New York. U. B. GYLLENSTEIN,K. M. HELM-BYCHOWSKI,R. G. OGILWE-GRANT,W.R. 1896. A hand-bookto the - HIGUCHI, S. R. PALUMBI, E. M. PRAGER,R. D. SAGE, birds. Edward Lloyd, Ltd., London. AND M. STONEKING. 1985. Mitochondrial DNA OLSON, S. L. 1973. A classification of the Rallidae. and two perspectiveson evolutionary genetics. Wilson Bull. 85:381-416. Biol. J. Linnaean Soc. 26:375-400. OLSON,S. L. 1985. The fossil record of birds. Pages WINTERBOTTOM,J. M. 1967. Climatologicalimplica- 80-238 in Avian biology, vol. 8 (D. S. Farher,J. tions of avifaunal resemblances between south- R. King, and K. C. Parkes,Eds.). Academic Press, western Africa and Somaliland. Pages 77-79 in New York. Paleoecologyof Africa, vol. 2 (E. M. Van Zin- RANDI, E., G. Fusco, R. LORENZINI, AND T. M. CROWE. deren Bakker, Ed.). A. A. Balkema, Amsterdam. 1991. Phylogenetic relationships and rates of WOLTERS,H.E. 1975. Die Vogelarten der Erde. Paul allozyme evolution in the Phasianoidea.Bio- Parey, Berlin. chem. Syst.Ecol. 19:213-221. ZINK, R. M., AND J. C. AvISE. 1990. Patterns of mi- SCHONWETTER,M. 1967. Handbuchder Oologie,Lie- tochondrialDNA and allozyme evolution in the ferung 8. Berlin, Akademie-Verlag. avian genusAmmodramus. Syst. Zool. 39:148-161. SHIELDS, G. F., AND K. HELM-BYCHOWSKI. 1988. Mi- tochondrial DNA of birds. Curr. Ornithol. 5:273- 295. 40 CROWEET AI•. [Auk,Vol. 109

APPENDIX1. Descriptionsof 34 morpho-behaviora! APPENDIX 1. Continued. charactersfrom 41 Francolinusspecies, several non- francolin perdicines,and a hypotheticalquail-like vertisement call 1: (0) absent; (1) 8-9 low-pitched, ancestor.Suggested plesiomorphic character states cooingnotes, ascending and then descendingin pitch; listedfirst. Characters for which polaritydecisions (2) sameas state 1, but overall pitch much higher; (3) basedon ontogeneticcriteria marked with asterisk same as state 2, but speed of delivery doubled or (*). Otherwise,polarity based on outgroupanalysis. trebled. 27. Specializedtonal advertisementcall 2: (0) absent;(1) quail-like "wheet whit-it"; (2) high-pitched 1'. Mean male mass(g): (0) 95-110; (1) 120-370;(2) "whee-hee-hee-hee,whee-pee-eu" or "whee-hee-hee- > 400.2*. Sexualsize dimorphism(based on wing and hee, whee-hee";(3) high-pitched "kee-bee-tillee."28. tail measurements):(0) male -<10% larger; (1) male Responseto playbackof advertisementcall: (0) none >10% larger. 3*. complement: (0) absent; (1) 1 or poor; (1) strong, at least seasonally.29. Known to spur; (2) 2 spurs,upper shorter; (3) 2 spursof equal in trees:(0) no; (1) yes. 30. Number of tail feath- length;(4) 2 spurs,upper longer.4. Spur position:(0) ers:(0) = <14; (1) = 14.31. Tail length/wing length: absent or closer to distal end of tarsus; (1) closer to (0) -<0.66;(1) >0.66. 32. Stance:(0) squat;(1) upright. proximal end of tarsus. 5*. Cere: (0) feathered; (1) 33. Incubationperiod: (0) -<21days; (1) >21 days.34. moderatelycartilaginous; (2) strongly cartilaginous. Shell thickness: (0) -<0.5 mm; (1) >0.5 mm. 6. Culmen length/log•0 wing: (0) -<11.7; (1) >11.7. 7*. Bill color: (0) yellow or black; (1) with at least some orange or red; (2) all red. 8*. Leg color: (0) yellow; (1) orange or red; (2) brown or black. 9*. Na- ked skin aroundeye: (0) absent;(1) present.10'. Throat: (0) feathered; (1) unfeathered. 11. Throat color: (0) undifferentiatedfrom surroundingskin; (1) yellow; APPENDIX2. Description of mensural charactersof (2) red. 12'. Dorsal plumage 1: (0) quail-like or any skeleton. Character number and abbreviation fol- of statesin characters13 and 14; (1) quail-like with lowed by indication of referencepoints in Figure buff-grey vermiculations;(2) vermiculated with bar- 2 (in parentheses)and characterdescription. ring and/or streaking,or streaked.13. Dorsal plum- age 2: (0) any of statesin characters12 and 14; (1) 1. PMAXL1 (b. 1-2): premaxillalength, mediallyfrom uniform brownishor rufousgrey in males.14. Dorsal posterioredge to anteriortip of premaxilla.2. PMAXL2 plumage 3: (0) any of statesin characters12 and 13; (b. 1-3): premaxilla length from narial opening, lat- (1) upper backspotted. 15'. Belly plumage1: (0) uni- eromedially from anterior edge of nares to anterior form buff; (1) barred;(2) streaked.16. Belly plumage tip of premaxilla. 3. PMAXD (a. 4-5): upper mandible 2: (0) states 0-2 under character 15; (1) black and depth, medially at anterior edge of nares.4. PMW1 streakedwhite with broadblack shaft streak. 17. Belly (c. 6-7): premaxillawidth at anterior margin of nares. plumage 3: (0) states0-2 under character15 and states 5. PMW2 (c. 8-9): premaxillawidth midway between 0-1 under character16; (1) featherswith very narrow anterior margin of naresand tip of premaxilla.6. FW1 dark edgesgiving a scalyappearance. 18. Blackand (b. 10-11): width of frontals at distal end of nasal white "necklace" plumage on side of head and neck: processof premaxilla. 7. FW2 (b. 12-13): width of (0) none;(1) restrictedto narrow strip behind eyeand frontals at midorbit. 8. FW3 (b. 14-15): width of fron- bordering throat; (2) extensive strip running from tals above palatine pterygoid junction. 9. POW (a. 16 behind eye to below throat. 19'. Color of primaries: to samepoint on other sideof skull): postorbitalskull (2) uniform brownishgrey; (3) grey with lighter ver- width at baseof postfrontalorbital processes.10. BTPW miculations;4 = grey, streakedwith white or buff; (d. 17-18):basitemporal plate width. 11. BTPL(d. 19- (0) rufous-chestnut;(1) grey with somerufous-chest- 20): basitemporal plate length from anterior margin nut. 20*. Color of under-tail coverts: (0) rufous; (1) of occipital condyle to anterior edge of projection rufous with black and white barring; (2) barred and/ abovebasisphenoid. 12. OD (a. 21-22): occipitaldepth or barred and vermiculated with black and white; (3) from anterior margin of occipital condyle to dorsal barred and streaked, or streaked with black and white; edge of supraoccipital.13. MNPW (c. 23-24): mid- (4) black. 21. Tail shape:(0) flat; (1) vaulted. 22. Hallux narial nasal processwidth. 14. NAP (b. 3-25): nares claw: (0) well developed;(1) absentor rudimentary. anterior-posteriorlength. 15. LNBW (a. 26-27): width 23. Chick eye stripe:(0) single stripe; (1) faint second of lateral nasal bar at midpoint between top of nares stripe; (2) well-developed second stripe. 24. General and jugal-nasaljunction. 16. SKAP (b. 2-28): skull form of advertisementcall 1: (0) quail-like "wheet anterior-posterior length from supraoccipitalto pos- whit-it"; (1) grating "kee-raack."25. General form of terior tip of premaxilla nasal process.17. MPW (a. 29 advertisement call 2: (0) states 0-2 under characters to same point on other side of skull): mid-parietal 24; (1) whistling, tonal call. 26. Specializedtonal ad- skull width, measureddorsally at junctionof quadrate January1992] EvolutionofFrancolins 41

APPENDIX 2. Continued. APPENDIX3. Scientific names of speciesin morpho- metric analysesand their morphometric"nearest and parietals.18. PPW (d. 30-31): posteriorpremaxilla neighbors."In casesin which nearestneighbor was width at origin of zygomatic maxillary process.19. a conspecific,the nearestspecies is alsolisted. Spe- OPQL (a. 32-33): length of orbital processof quadrate. cies not correctly assigned by nearest-neighbor method signified with an asterisk.Family or sub- 20. PRAL (g. 34-35): pre-acetabularpelvis length. 21. family name followed by species, with nearest POAL (g. 36-37): postacetabularpelvis length. 22. TPL neighbor(s)and Euclidian distance(s)in parenthe- (e. 34-38): total pelvic length. 23. PPL (g. 35-39): pec- ses. tineal processlength. 24. PRW (e. 40-41): pelvis rear width. 25. PFW1 (e. 42-43): pelvis frontal width 1.26. Megapodiidae:Megapodius freycinet* ( melea- PFW2 (e. 44-45): pelvis frontal width 2. 27. IF (g. 46- grides,0.82); Alecturalathami (M. ,0.66); Macro- 47): ischiadic foramen anteroposterior diameter. 28. cephalonmaleo (A. lathami,0.66). Cracidae: Ortalisve- PD (g. 40-48): pelvis depth. 29. RBW (f. 49-50): renal tula* (Polyplectron bicalcaratum, 0.69); Penelope bar width. 30. STL (r. 51-52): sternumlength. 31. SD1 purpurascens(C. globulosa,0.51); Aburriapipile (C. uni- (r. 51-53): sternum depth at anterior end. 32. SD2 (r. color,0.50); Champaetes unicolor (A. pipile,0.50); Notho- 54-55): sternum depth at half-way point. 33. SND (s. crax urumutum(C. globulosa,0.83); Mitu mitu (P. pauxi, 52-56): depth of internal sternal notch. 34. SW (s. 57- 0.42); Pauxipauxi (M. mitu,0.42); Crax rubra(C. alberti, 58): sternum anterior width. 35. ALPL (r. 59-60): an- 0.55); C. alberti(M. mitu,0.51); C. globulosa(P. purpuras- terior lateral sternal processlength. 36. ALPW (r. 61- cens,0.51). Phasianidae,: Meleagrisgal- 62): anterior lateral sternal processwidth. 37. PLPW lopavo(M. gallopavo,0.43; M. ocellata,1.02); M. gallo- (r. 63-64): outer posterior lateral processwidth. 38. (M. gallopavo,0.43; M. ocellata,0.89); M. ocellata* CL1 (h. 65-66): coracoldlength 1.39. CL2 (h. 65-67): (P. cristatus,0.69). Tetraoninae: Lagopusmutus (B. um- coracoldlength 2. 40. CW (h. 68-69): coracoldwidth bellus,0.57); Tetraotetrix (T. cupido,0.59); T. urogallus half-way down shaft. 41. CSW (h. 65-70): coracold (C. urophasianus,0.89); Bonasa umbellus (L. mutus,0.57); width from head to distal edge of scapularfacet. 42. Centrocercusurophasianus (T. tetrix,0.73); CDTW (h. 67-71): total width of coracoid at sternal cupido(T. cupido,0.36; T. tetrix,0.59); T. cupido( T. cupido, end. 43. CDW1 (h. 66-71): coracold sternal end width 0.36; T. tetrix,0.59). Odontophorinae: leu- 1.44. CDW2 (h. 66-67): coracold sternal end width 2. cophrys*(Melanoperdix nigra, 0.57); Oreortyxpictus (C. 45. CDW3 (i. 72-73): coracoid sternal end width 3. 46. virginianus,0.48); squamata (L. californicus, 0.49); CDW4 (i. 74-75): coracold sternal end width 4. 47. Lophortyxcalifornicus (C. virginianus, 0.36); Philortyx fas- CPWIII (j. 76-77): width of metacarpalIII at midpoint. ciatus(L. californicus,0.43); Colinus virginianus (L. cali- 48. CPWIV (j. 78-79): width of metacarpalIV at mid- fornicus,0.36); Odontophorus guttatus* (Melanoperdix ni- point. 49. CPWT (j. 76-79): total width of carpometa- gra,0.49); Dactylortyx thoracicus (C. montezumae,0.48); carpus. 50. CPL (j. 80-81): carpometacarpuslength. Cyrtonyxmontezumae (C. virginianus,0.46); Rhynchortyx 51. FL (k. 82-83): furcula length. 52. FW (k. 84-85): cinctus(C. montezumae,0.49). Phasianinae, Perdicini: furcula width half-way along its length. 53. ML1 (a. Lerwalerwa (A. graeca,0.51); Ammoperdixheyi (F. coqui, 86-87): mandible length 1.54. ML2 (a. 86-88): man- 0.42); Alectorisgraeca (F. swainsonii,0.39); Francolinus dible length 2. 55. MD (a. 89-90): mandible depth. francolinus(F. francolinus,0.24; F. harwoodi,0.33); F. 56. TMTL (1.91-92): tarsometatarsuslength. 57. TMDW francolinus(F. francolinus,0.33; F. harwoodi,0.36); F. (1.93-94): tarsometatarsusdistal width. 58. TMHW (1. francolinus(F. francolinus,0.24; F. pintadeanus,0.35); F. 95-96): tarsometatarsuswidth half-way along shaft. pictus(F. coqui,0.37); F. pintadeanus(F. pondicerianus, 59. TMHD (m. 97-98): tarsometatarsushypotarsal 0.36); F. pintadeanus(F. pintadeanus,0.26; F. francolinus, depth. 60. SCL (n. 99-100): scapula total length. 61. 0.35); F. pintadeanus(F. pintadeanus,0.26; F. francolinus, SCFW (n. 101-102): scapularfacet width. 62. SCW (n. 0.35); F. afer (F. swainsonii,0.26); F. afer (F. afer,0.29; 103-104): scapular width two-thirds along its total F. bicalcaratus,0.30); F. afer (F. afer,0.25; F. swainsonii, length from coracoid.63. TBTL (q. 105-106):tibiotar- 0.30);F. afer(F. afer,0.25; F. icterorhynchus,0.28); F. sus length. 64. TBW (q. 107-108).'tibiotarsus width swainsonii(F. swainsonii,0.26; F. bicalcaratus,0.27); F. half-way down shaft. 65. TBDW (q. 109-110): tibio- swainsonii(F. swainsonii,0.23; F. leucoscepus,0.26); F. tarsusdistal width. 66. TBPW (q. 111-112):tibiotarsus swainsonii(F. afer,0.26); F. swainsonii(F. swainsonii, 0.23; proximal width. 67. FEMW (o. 113-114): width F. leucoscepus,0.31); F. swainsonii(F. swainsonii,0.23; F. half-way down shaft. 68. FEML (o. 115-116): femur leucoscepus,0.33); F. swainsonii(F. swainsonii,0.26; F. length. 69. HMTL (p. 117-118): humerus length. 70. adspersus,0.34); F. leucoscepus(F. leucoscepus,0.25; F. HMPW (p. 119-120): humerus proximal width. 71. swainsonii,0.30); F. leucoscepus(F. leucoscepus,0.25; F. HMDW (p. 121-122): humerus distal width. 72. swainsonii,0.26); F. leucoscepus(F. leucoscepus,0.25; F. HMHW (p. 123-124): humerus width half-way down adspersus,0.31); F. leucoscepus(F. leucoscepus,0.25; F. shaft. 73. HMFD (p. 125-126): humerus pneumatic afer,0.30); F. leucoscepus(F. leucoscepus,0.29; F. afer, foramen diameter. 42 CROWEET ^I•. [Auk,Vol. 109

APPENDIX 3. Continued. APPENDIX 3. Continued.

0.30);F. jacksoni (F. capensis,0.44); F. squamatus(F. squa- dix madagarensis(F. pintadeanus,0.42); Melanoperdixni- matus,0.30; F. afer, 0.31); F. squamatus(F. natalensis, gra (Rollulusroulroul, 0.33); Coturnixc. coturnix(C. c. 0.27); F. squamatus(F. bicalcaratus,0.25); F. squamatus africana,0.32); C. c. africana(C. c. coturnix,0.32); C. c. (F. squamatus,0.29; F. icterorhynchus,0.29); F. bicalcara- japonica(C. c.coturnix, 0.41); C. delegorguei(C. c.coturnix, tus (F. bicalcaratus,0.32; F. swainsonii,0.33); F. bicalcar- 0.43);Synoicus ypsilophorus (C. c. africana,0.45); Excal- atus (F. swainsonii,0.27); F. bicalcaratus(F. squamatus, factoriachinensis (P. argoondah,0.65); argoon- 0.25); F. icterorhynchus(F. levaillantii,0.28); F. clapper- dah (C. delegorguei,0.50); brunneopectus (F. toni (F. afer, 0.32);F. clappertoni(F. harwoodi,0.29); F. lathami,0.43); Tropicoperdixcharltonii (A. brunneopectus, natalensis(F. natalensis,0.28; F. harwoodi, 0.30); F. na- 0.46);Rollulus roulroul (M. nigra,0.33); Ptilopachus pe- talensis(F. natalensis,0.29; F. squamatus,0.35); F. natalen- trosus*(P. petrosus,0.26; Lophortyx californicus, 0.48); P. sis(F. squamatus,0.27); F. adspersus(F. capensis,0.26); petrosus(P. petrosus,0.26; A. heyi, 0.48); Bambusicola F. hartlaubi*(Margaroperdix madagarensis, 0.44; F. pon- thoracica(F. sephaena,0.38); Galloperdix spadicea (F. se- dicerianus,0.52); F. harwoodi(F. squamatus,0.29); F. ca- phaena,0.43). Phasianini:lthaginis cruentus* (Franco- pensis(F. capensis,0.25; F. swainsonii,0.35); F. capensis linus squamatus,0.56); Tragopantemminckii (P. macro- (F. capensis,0.31; F. adspersus,0.37); F. capensis(F. ca- lopha,0.41); Pucrasiamacrolopha (T. temminckii,0.41); pensis,0.25; F. swainsonii,0.40); F. capensis(F. adspersus, Lophophorusimpeyanus (C. mantchuricum,0.68); Gallus 0.26); F. capensis(F. capensis,0.26; F. swainsonii,0.36); gallus(G. varius,0.50); G. varius(L. swinhoii,0.42); Lo- F. sephaena(F. sephaena,0.33; F. coqui,0.37); F. sephaena phura leucomelanos*(Francolinus adspersus, 0.46); L. (F. sephaena,0.33; Rollulusroulroul, 0.37); F. sephaena swinhoii(G. varius,0.42); L. ignita(A. ,0.54); Cros- (F. sephaena,0.33; F. lathami,0.46); F. sephaena(F. se- soptilonmantchuricum (L. impeyanus,0.68); Catreuswal- phaena,0.33; F. coqui,0.37); F. africanus(F. levaillantii, lichii(P. colchicus,0.41); Syrmaticushumiae* (Francolinus 0.30); F. levaillantoides(F. levaillantoides,0.28; F. levail- capensis,0.44); Phasianuscolchicus (C. wallichii,0.41); lantii, 0.35); F. levaillantoides(F. levaillantoides,0.28; F. Chrysolophuspictus (L. leucomelanos,0.52); Polyplectron levaillantii,0.31); F. shelleyi(F. levaillantoides,0.31); F. bicalcaratum• (Francolinus gularis, 0.57); Rheinartiaocel- levaillantii(F. levaillantii,0.25; F. icterorhynchus,0.28); lata (A. argus,0.62); Argusianusargus (L. ignita,0.54); F. levaillantii(F. levaillantii,0.25; F. africanus,0.32); F. Pavo cristatus*(Meleagris ocellata, 0.69). Numidinae: coqui(F. coqui,0.22; F. sephaena,0.37); F. coqui(F. coqui, Agelastesmeleagrides (A. niger,0.35); A. meleagrides(A. 0.26; F. pintadeanus,0.40); F. coqui(F. coqui,0.22; F. meleagrides,0.46; A. niger,0.50); A. niger(A. meleagrides, pictus,0.39); F. coqui(F. coqui,0.32; F. pictus,0.38); F. 0.35);Numida meleagris (N. meleagris,0.31; G. pucherani, coqui(F. coqui,0.34; F. pictus,0.39); F. lathami(F. lathami, 0.51); N. meleagris(N. meleagris,0.38; A. vulturinum, 0.31; Arborophilabrunneopectus, 0.43); F. lathami(F. la- 0.53); N. meleagris(N. meleagris,0.32; A. vulturinum, thami,0.29; F. sephaena,0.45); F. lathami(F. lathami,0.29; 0.49); N. meleagris(N. meleagris,0.31; A. vulturinum, F. sephaena,0.42); F. pondicerianus(F. pintadeanus,0.36); 0.53); plumifera (G. pucherani,0.45); G. pucherani F. gularis(F. francolinus,0.39); perdix (F. coqui, (G. plumifera,0.45); Acryllium vulturinum (N. meleagris, 0.55); longirostris (F. afer,0.43); Margaroper- 0.49).