The Long-Billed Lark Complex: a Species Mosaic in Southwestern Africa

The Auk 116(1):194-208, 1999

THE LONG-BILLED LARK COMPLEX: A SPECIES MOSAIC IN SOUTHWESTERN AFRICA

PETER G. RYAN •,3 AND PAULETTE BLOOMER •'2 •PercyFitzPatrick Institute of AfricanOrnithology, University of CapeTown, Rondebosch 7701, South Africa; and 2Departmentof Zoologyand Entomology, Pretoria University, Pretoria 0001, SouthAfrica

ABSTi•ACT.--TheLong-billed Lark (Certhilaudacurvirostris) is a geographicallyvariable species.Thirteen subspecies have been described, and at leastthree phenotypically well- definedgroups exist. We use concordantpatterns of genotypicand phenotypicvariation to recognizecryptic species within this complex.Genotype information was obtainedby se- quencingpart of themitochondrial DNA cytochrome-bgene. Three taxa are genetically highly distinct,differing from each other by 6 to 9% sequencedivergence. One of thesetaxa com- prisesthree phenotypically distinct taxa that differ from eachother by 2% sequencedivergence.These values far exceedthe variationwithin describedsubspecies (0.2 to 0.4%)and approachthe distanceto the presumedsister taxon, the Short-clawedLark (C. chuana;8 to 10%).Well-defined morphological and plumage differences exist between all taxaexcept benguelensisand northernpopulations of subcoronata.We proposethat the complexconsists of five species:the closelyrelated C. curvirostris,C. brevirostris,and C. semitorquata,and the more distantly related C. subcoronataand C. benguelensis.Comparison of male display songs providesadditional support for the recognitionof five species,but samplesizes are small. The positionof C. chuanarelative to the five speciesnoted aboveis not resolved.Species' rangesare allopatric,but more work is neededto elucidatethe distributionallimits of taxa in northernNamibia and Angola.These findings have implications for biodiversityconser- vation,and they suggestthat considerablediversity has been overlooked in wide-ranging, sedentarylark species.Received 19 December1997, accepted22 June1998.

THE LARKS(FAMILY ALAUDIDAE) are a pre- deserts,and mesicmontane grasslands (Dean dominantlyOld World family whoseaffinities et al. 1992, 1997). Being sedentary(Dean and within the passerinesare poorly understood Hockey1989; Dean et al. 1992,1997) apparently (Sibley and Monroe 1990). Most speciesare has promoted the evolutionof marked geo- found in Africa, where there are two main cen- graphicalvariation in plumageand morpholo- ters of diversityassociated with the northeast- gy. Theseregional differences have been inter- ern and southwestern arid zones. These re- preted as adaptationsto local conditions.For gionsare characterizedby highlevels of ende- example,upperpart coloration matches the lo- mism,with manyspecies having small ranges. cal substratum color (Macdonald 1952), and Recentresearch suggests that superficially sim- long bills and hind clawsoccur in birds living ilar allopatricpopulations, which have been on soft, sandy soils (Clancey1957). lumped together in polytypic species,may This geographicvariation has led to consid- have been isolatedfor considerableperiods of erable debate about the species'taxonomy. time (e.g. Ryan et al. 1998). Thirteen taxa have been described (Clancey The Long-billed Lark (Certhilaudacurviros- 1980,Dean et al. 1992),which havebeen split tris) is restricted to southwesternAfrica from into as many as four species(Roberts 1940), southernAngola throughwestern Namibia to and even placed in separategenera (Sharpe SouthAfrica. It formsa superspecieswith the 1904). However, the current consensusis to Short-clawedLark (C. chuana),which replaces lump all forms into a single polytypic species the Long-billedLark in arid acaciasavannas in (Clancey 1957, Dean et al. 1992,Maclean 1993). easternBotswana and adjacentSouth Africa This is despitea recentreview of the speciesin (Dean et al. 1992, 1997; Fig. 1). Long-billed South Africa that concludedthat the species Larksoccur in a wide rangeof habitats,includ- perhapsshould be viewed as a superspecies ing coastaldunes, open plains, rocky slopes, with at least three allospecies(Quickelberge 1967). E-mail: [email protected] Here, we use geneticsequence data to help

194 January1999] Long-billedLark Species Complex 195

25S

3O S

curvirostris

35 S bre viros tris

15 E 20 E 25 E 30 E

FIG.1. The distributionof speciesin the Long-billedlark complex,including the presumedsister taxon, Short-clawedLark (Certhilaudachuana), based on examinationof specimensand atlasdata from Dean et al. (1997).Open squaresdenote collection localities of birds whosemtDNA was sequenced. resolvethe evolutionaryrelationships among mM Tris pH 7.6, 100 mM NaC1, 1 mM NaEDTA pH taxa within the Long-billedLark complex.We 8.0, 0.5% SDS, i mg/mL proteinaseK) overnightat sequencedpart of the mitochondrialcyto- 37øC. Following RNAse treatment, total genomic chrome-bgene, which has provedvaluable in DNA was extractedtwice with phenol,once with chloroform:isoamylalcohol(24:1), and precipitated resolvingrelationships at the specieslevel in with 2M ammonium acetate and 100% ethanol. Pu- larks and otherbirds (e.g.Shields and Helm- rified mtDNA was extractedfrom three specimens Bychowski1988, Smith et al. 1991,Helbig et al. (Table1) followingthe methods of Essopet al. (1991). 1996, Ryan et al. 1998).We also reanalyze geo- A 500 base-pair(bp) fragmentof the mtDNA cyto- graphicvariation in plumageand morphology chrome-bgene was amplifiedusing primers L14990 in relation to the geneticallydefined taxa and (L14841;Kocher et al. 1989)and H15499 (CBINT; Av- make recommendationsabout the taxonomyof ise et al. 1994).Double-stranded amplifications were the complex. performed in 50-p,L volumes using 2.5mM MgC1v 2mM dNTPs,25 pmol of eachprimer, and 1.5 units METHODS of TaqDNA polymerase(Promega). Reaction mix- tures were subjectedto 35 cyclesof denaturationat Samplingand DNA analysis.--BetweenOctober 94øC(30 s), annealingat 48 to 50øC(1 min), and ex- 1994and July1997, we collected19 birdsfrom 17 lo- tensionat 72øC(1 min).The first cycle was preceded calities(Table 1, Fig. 1). Individuals were assignedto by an initial denaturationof 3 min at 94øC,and the taxabased on plumagecharacters and locality(see last cyclewas followed by a final extensionof 10min below).We dissectedout the heart, liver, and pec- at 72øC.PCR productswere purified through 2% toral muscletissues and preservedthem in a satu- low-meltingpoint agarose gels (MS8; Whitehead Sci- rated NaCI/20% DMSO solution (Amos and Hoelzel entific)and recoveredusing the QIAEX II gel extrac- 1991). tion kit (Qiagen).Both light andheavy strands were In the laboratory,tissue samples were digested (50 sequencedby dideoxy sequencing(Sanger et al. 196 RYANAND BLOOMER [Auk,Vol. 116

TABLE1. Collectionlocalities and subspecificdesignation of the 19 birds in the Long-billedLark complex whosecytochrome-b gene was sequenced(one per site,except for UniabRiver).

Taxon Locality Coordinates falcirostris Port Nolloth 29ø15'S,16ø50'E falcirostrisa GroenrivierMouth 30ø50'S,17ø35'E curvirostris Paternoster 32ø48'S, 17ø55'E brevirostris Roodekleigat(20 km W of Robertson) 33ø50'S,19ø40'E brevirostris Bredasdorp(15 km NE) 34ø25'S,20ø10'E semitorquata Cradock(25 km NW) 32ø08'S,25ø24'E semitorquata Stutterheim(20 km NW) 32ø23'S,27ø42'E semitorquata NottinghamRoad (15 km W) 29ø22'S,29ø54'E transvaalensis Wakkerstroom(10 km SE) 27ø22'S,30ø12'E gilli Toorberg(35 km E of Murraysburg) 32ø03'S,24ø08'E gilli Middlestevlei(40 km N of Laingsburg) 32ø55'S,20ø57'E subcoronata• Bloubosfontein(40 km NW of Prieska) 29ø24'S,22ø38'E subcoronata Dikpens (100 km NW of Brandvlei) 30ø10'S,19ø32'E bradshawia Pofadder 29ø10'S, 19ø22'E bradshawi Nieu-Tsaus (40 km S of Aus) 27ø00'S,16ø12'E kaokensisb Uniab River (100 km W of Kamajab) 19ø54'S,13ø59'E chuana Pietersburg 29ø20'S,29ø25'E Sequencesfrom purifiedmtDNA (seeMethods). btI --3.

1977)with the Sequenaseversion 2 enzymesystem have been depositedin GenBankunder accession (United StatesBiochemicals). The two PCR primers numbers AF033249-58. and two internal primers (H15 298 [Kocheret al. Morphometricsand plumage.--PGRmeasured 414 1989]and L15245[Palumbi et al. 1991])were usedfor Long-billedLarks and Short-clawedLarks; 400 were generating sequences.Sequences were aligned by from skins housed in the Durban, East London, Na- eye, and no insertionsor deletionswere found. mibian, South African, and Transvaal museums, and A matrix of pairwise nucleotidesequence diver- 14 were birds measured in the field. Measurement gencesfor neighbor-joininganalysis (Saitou and Nei protocolsfollow Ryan et al. (1998).Tail lengthand 1987) was calculated in MEGA version 1.01 (Kumar wing length(flattened chord) were measured to the et al. 1993)using the Kimura two-parametercorrec- nearest1 mm. Tarsuslength, bill length(distal edge tion (Kimura 1980). This model accountsfor the of naresto bill tip), andbill depth(depth at proximal higher rate of transitional substitutionsin animal edgeof nares)were measuredto the nearest0.1 mm. mtDNA and usesa 2:1 TV:TI ratio;no gammacor- Bill lengthfrom the nareswas preferred to the stan- rection was applied to correct for possibleamong- dard measureof bill length(from the baseto the tip) site rate variation. Maximum-parsimonyanalysis becausemolt and damageto museumskins make it was performedusing PAUP 3.1.1 (Swofford 1993). difficult to consistentlymeasure standard bill length Phylogeneticsignal was assessedby analyzingtree- (seeGrant et al. 1985). Coefficientsof variation with- lengthdistributions of 1,000random trees (Hillis and in taxainvariably were smaller using bill lengthfrom Huelsenbeck1992). In additionto treatingall phy- the nares. This measureaverages 68% of standard logenetically informative charactersas unordered bill lengthin Long-billedLarks, with the proportion and unweighted,different weighting schemes were increasingwith bill length (range66% for semitor- used to account for the transition bias in mitochon- quataand 71% for curvirostris-falcirostris). drial DNA and to reducehomoplasy. MacClade 3.0 Sexof freshlycollected specimens was determined (Maddison and Maddison 1992) was used to calcu- by inspectionof the gonads.The accuracyof sexes late the transition:transversion(TI:TV) ratio by reportedon museumlabels was testedby estimating countingthe averagenumber of transitionand trans- discriminant functions for each main taxon. Given versionevents on 1,000random trees. Successive ap- the markedsexual dimorphism, there were very few proximation characterweighting based on mean skinswhose sex could not readily be assigned.Two characterconsistency indices across equally parsi- skins(0.5%) were found to be incorrectlysexed (P < monioustrees (Farris 1969) was also used. Reliability 0.001)and werereassigned accordingly. of topologieswas assessedusing 1,000 bootstrap Plumagevariation in the Long-billedLark com- replicates(Felsenstein 1985). The Karoo Lark ("Cer- plex has two main components:(1) ground colorof thilauda"albescens; Ryan et al. 1998) and the Gray- the upperpartsand underparts;and (2) the extent, backedFinch-Lark (Eremopterix verticalis) were in- size,and intensityof dark brownfeather centers or cluded for outgroup comparisons.Sequence data streaking.Upperpart colorationwas noted but not January1999] Long-billedLark Species Complex 197

TABLE2. Pairwiseestimates of nucleotidesequence divergence between Long-billed Larks (sensu lato). Per- centagesequence divergence (Kimura two-parameterdistance) is abovediagonal, and numberof transitions/ tranversionsis below diagonal.

Taxon 1 2 3 4 5 6 7 8 9 10 11 1 falcirostris -- 0.2 2.1 2.3 7.3 7.5 8.4 8.5 9.4 16.9 15.3 2 curvirostris 1/0 -- 2.3 2.6 7.5 7.7 8.7 8.7 9.7 17.2 15.6 3 brevirostris 10/0 11/0 -- 2.3 7.3 7.5 8.7 8.7 9.9 16.7 15.8 4 semitorquata 11/0 12/0 11/0 -- 6.6 6.8 7.7 8.2 8.7 15.8 14.0 5 subcoronataAa 27/6 27/6 27/6 24/6 -- 0.2 6.1 6.6 8.2 15.1 14.8 6 subcoronataBa 28/6 28/6 28/6 25/6 1/0 -- 6.4 6.8 8.5 15.3 15.1 7 kaokensisA b 30/8 30/8 30/8 27/8 24/4 25/4 -- 0.4 9.4 17.0 14.8 8 kaokensisB b 30/8 30/8 303/8 29/8 26/4 27/4 2/0 -- 9.9 17.3 15.4 9 chuana 33/9 34/9 34/9 30/9 30/7 31/7 33/9 35/9 -- 18.0 15.6 10 albescens 49/23 50/23 50/23 45/23 44/21 45/21 51/21 53/20 52/24 -- 15.0 11 Eremopterix 43/23 44/23 45/23 38/23 45/19 46/19 43/21 45/21 45/22 41/24 -- • A = five birdscollected south of the OrangeRiver, including gilli, subcoronata,and bradshawi; B = onebird (bradshawl?)collected near Aus, southern Namibia. bOf threebirds collected near the UniabRiver, two weretype A and onetype B. quantified.Streaking was measured in threeparts of this latitudewas used to separatekaokensis from da- the body.On the upperparts,streak width was es- marensis.Benguelensis was restrictedto specimens timatedas a percentageof total featherwidth on the from Angola. exposedportion of backfeathers. On the flanksand Variationin morphology(size and plumagepat- belly,the extentof streakingwas scored from 0 to 3 tern) was analyzed at the level of the 12 described where 0 = absent and 3 = extensive. For ease of in- taxa, as well as for taxa identifiedfrom the phylo- terpretation,group averagescores are presentedas geneticanalysis. Univariate differences in morphol- percentages.On thebreast, the extentof streakingis ogy weretested using ANOVA and t-tests.Principal correlatedwith the size of the streaks.The length componentsanalysis (PCA) was used to summarize and width of breast streaks were measured to the phenotypicvariation in sizeand the extentof plum- nearest0.5 mm, andthese values were multiplied to age streaking.Discriminant function analysiswas givean indexof breaststreaking. used to identify "incorrectlyclassified" individuals. Individuals were allocated to taxa on the basis of Data werestandardized (relative to the largestmea- collectionlocality, following published ranges (Clan- sure = 1, smallestmeasure = 0) for multivariate an- cey 1980).We recognized12 putativetaxa in three alyses. well-definedgroups. Grayish, heavily streaked birds Vocalizations.--Long-billedLarks utter an array of occuron the westerncoastal plain (curvirostrisand calls but have a relativelysimple song associated falcirostris;Clancey 1957) and Agulhas Plain (brevi- with dramaticaerial displays. Males fly low overthe rostris;Roberts 1941). Rufous, mostly unstreaked groundbefore rising almostvertically 10 to 15 m in birdswith buffyunderparts occur in grasslandseast the air,the lastpart of theascent being driven solely of 25øE,with algidarestricted to thecoastal ranges in by the bird's momentum,because the wings are the southwest,semitorquata throughout most of the closed.The bird stallsat the apexof the displayand range,and transvaalensisin the north (Quickelberge plummetsdown, only openingits wingsjust before 1967).Like Quickelberge(1967), we considereddav- it reachesthe ground.As the bird descendsit utters iesi synonymouswith semitorquata.West of 25øE, a far-carrying,descending whistle, typically ren- more heavily streaked birds with rufous-brown dered "peeeoooo"(Dean et al. 1992, Maclean 1993). backsand whitish underparts occur in Karooshrub- This displayapparently occurs in all C. curvirostris lands (gilli and subcoronata;Roberts 1936). Farther taxa and is sharedwith the Short-clawedLark (Her- north, these birds become less streaked, redder, and remanset al. 1994).In Long-billedLarks, the same smaller,with bradshawiin the OrangeRiver basin songalso is uttered from prominentperches in the and southern Namibia, damarensisin central Namib- territory;flight songsdo not differ from thosegiven ia, kaokensisin northernNamibia, and benguelensisin by perchedbirds. southern Angola (Sharpe 1904, Clancey 1980). Lark vocalizationswere recordedby PGRusing a Boundaries between the northern taxa are not well TECT supercardiodshotgun microphone (model definedowing to limited and patchycollecting. We UEM-83) and a Sonycassette recorder (model TCM- arbitrarily separatedbradshawi from damarensisat 17); additionalrecordings were suppliedby Callan 25øS,which corresponds with a largegap in collec- Cohen,John Graham, Penn Lloyd, the FitzPatrick tions.Atlas data (Dean et al. 1997)indicate a gap in SoundLibrary, and commerciallyavailable bird re- the distributionof Long-billedLarks at 21ø30'S,and cordingsby Guy Gibbon.Only male display songs 198 RYANAND BLOOMER [Auk, Vol. 116

Eremopterix ve r tic a Ils

"Certhilauda " •/bescens

chuana

curvirostris

1O0 bre virostris --'•80 lOO 1 oo 1 oo

16 994•4semitorquafa Icirostris ta subcoronata A

83 1•--lOOI benguelensissubcoronata BA-- 1•--- l 56 lO L I benguelensisB -- B FIG.2. Phylogenyof larksin the Long-billedLark complexbased on 483 bp of the cytochrome-bgene, usingthe KarooLark (Certhilaudaalbescens) and Gray-backedFinch-Lark (Eremopterix verticalis) as outgroups (Table2). (A) Maximum-parsimonytree (branch and bound search) with bootstrapvalues above the branches and the numberof unambiguouschanges occurring along a branchbelow the branches. (B) Neighbor-joining tree basedon Kimuratwo-parameter distances. Numbers indicate the frequency(%) with whicheach clade was identifiedin 1,000bootstrap replicates. Branches with lessthan 50% bootstrapsupport are left unresolved. were compared. Sonogramswere produced using genceand the TI: TV ratiofor eachcomparison SoundEdit(MacRecorder, Farallon Computing Inc.) are given in Table2. on a PowerMacintosh. Because of small samplesiz- Eight haplotypeswere found among the es, no attempt was made to quantify variation in Long-billedLarks (Table 2). Identicalsequences songstructure within and amongtaxa. were found amongfalcirostris(n = 2), brevirostris (n = 2), semitorquata-transvaalensis(n = 4), RESULTS subcoronata-gilli-bradshawi(excluding the sole Genotypicvariation.--We sequenced 483 bp of Namibianspecimen, n = 5), and two of the the cytochrome-bgene from 18 birds in the three kaokensis.Three taxa were genetically Long-billedLark complex(Appendix). Includ- highly distinct(6 to 9% sequencedivergence), similar to the distance to the Short-clawed Lark ing Short-clawedLark andthe two outgroups, therewere 129variable sites. As expected,most (Table2): (1) a northernNamibian group (ka- variable sites were at third-codonpositions okensis),(2) a Karoo-southernNamibia group (105 sites,82%), with first-position(21 sites, (gilli, subcoronataand bradshawi),and (3) a 16%) and second-positionsites (3 sites, 2%) group comprisingthe coastaland grassland more conserved. Most were silent substitu- taxa.Within thislast group, three taxa differed tions,with only20 variableamino acid sites re- from eachother by at least2% sequencediver- corded.Including the outgroups,the ratio of gence:birds from the west coast(curvirostris transitions(TI) to transversions(TV)was 2.9:1, andfalcirostris),the Agulhas Plain (brevirostris), whereasthe ingroup(Long-billed Larks and and the easterngrassland (semitorquata and Short-clawedLark) showeda biasof 4.3:1.Pair- transvaalensis).These values contrast with the wise estimatesof nucleotidesequence diver- variation within describedsubspecies (0.2 to January1999] Long-billedLark Species Complex 199

Bill length 18 (mm) 15

12 brevirostris•

70 75 80 85 90 Tail length (mm)

breviro 21 B ($tri• •uthern Streaking • • J forms

semitorquata / / ' / • • northern -1 ( • •rms enguelensis -2 -1 0 1 2

Size PC 1

F•G. 3. Bi•]ots o• male Long-billed Lat• mot- •o]ogy. (A) Un•atiate comparisono• tail •etsus bi]] length. (B) Multivariatecomparison o• size (BC • accounts•ot 5•% o• •atiance in Q•e mea- surementso• morphology)•etsus •lumage streaking (•CI accounts•ot 8•% o• •atiance in t•e t•tee streaking indices).•oiygons encompass >95% o• ua]s measured (n = 289).

0.4% for kaokensisand bradshawi)or between taxa within groups(0.2% for curvirostris-falcirostris). The sequencedifferences are alsoreflected at the amino acid level, with three to six amino aciddifferences among the threedistinct Long- billed Lark taxa and four to seven differences between the Long-billed and Short-clawed larks.There was only one amino acid difference within the curvirostris-brevirostris-semitorquata group, with brevirostrisdistinct from curvirostris and semitorquata.There were no aminoacid differenceswithin or betweensubspecies. The unweightedcladistic analysis of the 71 phylogeneticallyinformative characters yielded two equally parsimonioustrees (length = 200 RYANAND BLOOMER [Auk, Vol. 116

TABLE4. Indicesof plumagestreaking (œ _+ SD) amongtaxa in the Long-billedLark complex(sexes com- bined).Back and belly streakingare expressedas percentages;breast streaking is a measureof streaksize (seeMethods). Within columns, values with the samesuperscript are not significantlydifferent.

Taxa(n) Backstreaking Breaststreaking Bellystreaking curvirostris(67) 48.2 _+7.4 ^ 21.0 -+ 4.8^ 71.3 +_13.9 ^ brevirostris(15) 44.3 _+4.9 ^ 17.0 -+ 4.7• 69.1 _+17.0 ^ semitorquata(102) 14.3 _+7.6 B 8.3 +--4.1 c 7.8 _+8.2 c subcoronata(199) 16.2 _+10.1B 13.3 -+ 6.1B 22.6 ___24.2 B benguelensis(25) 10.7 _+5.1B 8.5 +--2.2 c 2.3 _+4.1 c chuana(6) a 49.2 _+10.2 6.2 -+ 1.2 0.0 +--0.0 Overall F (5 and 408 df)b 190.4 56.2 126.7 Excludedfrom analysesbecause of small samplesizes. P <: 0.001 in all cases.

124 steps;CI = 0.710;RI = 0.752;gl =-0.886). This bias presumablyreflects the greaterde- Similartopologies were obtained when placing tectabilityof males,which are morevocal and greateremphasis on transversionsrelative to tend to sit in exposedsites more than do fe- transitions(TV:TI weight = 3:1 or 5:1), or males. Plumage charactersdo not differ be- weightingcharacters according to their mean tweenthe sexes,but malesare much larger than consistencyindices. The latter analysispro- females(Table 3), so morphologicalcompari- duced a single most-parsimonioustree after sonswere restrictedto same-sexgroups. Males one round of successiveweighting (Fig. 2A; averagedat least10% larger than females in all length= 89 steps;CI = 0.795;RI = 0.828;gl = characters,with size dimorphismgreatest in - 1.146).The neighbor-joining analysis yielded bill length,which was more than 25% larger in an identicaltopology (Fig. 2B), and in both malesin all taxaexcept for curvirostris(Table 3). trees the five Long-billedLark specieswere Significantmorphological differences existed supportedby high bootstrapvalues (Fig. 2). amongthe taxa identifiedby geneticanalysis The positionof the Short-clawedLark remains (Table3). The west coastcurvirostris was the unresolved.Most analysesyielded a trichoto- largesttaxon in all charactersexcept tail length, my between chuana, curvirostris-brevirostris-whereas the eastern grassland semitorquata gen- semitorquata,and subcoronata-benguelensis.With erallywas the smallest. The Agulhas Plain pop- a 5:1 TV:TI weighting,chuana fell outsidethe ulation (brevirostris)has a shorttail and a short, Long-billedLark group (bootstrapvalue only slender bill, and the western interior forms 51%), but basedon the amino acid sequence, (subcoronataand benguelensis)have the longest bothparsimony and neighbor-joining analyses tails and relatively long wings. The same placedchuana with the curvirostris-brevirostris-trends occurred among females. However, semitorquatagroup (80% and 78% bootstrap overlapexisted between individuals from dif- values,respectively). ferent taxa, especiallyamong benguelensis, sub- Our resultssuggest that the Long-billed Lark coronata,and semitorquata(Fig. 3A). complex comprises five species:curvirostris, Streakingindices on the back, breast, and brevirostris,semitorquata, subcoronata and ben- belly were stronglypositively correlated, with guelensis(Fig. 1). We use benguelensisfor the significant differencesamong taxa (Table 4). northernNamibian species because, to be con- Combiningstreaking indices with morphology servative,we presumeit is thesame as kaokensis increased the separationbetween taxa, but con- and has precedence(Sharpe 1904). The status siderableoverlap remained, especially between of damarensisis unresolved(but seebelow). Fal- benguelensisand the northernpopulations of cirostrisis consideredto be a regionalvariant, subcoronata(Fig. 3B). Discriminant functions becauseits degreeof sequencedivergence from correctlyidentified only 78% of male birds on curvirostris(0.2%) is typicalof intraspecificvar- the basisof morphologyand plumage,with iation (seeHelbig et al. 1996,Ryan and Bloomer most misclassificationsbeing betweenbengue~ 1997, Ryan et al. 1998). lensisand subcoronata.The major factor con- Phenotypicvariation.--There was a strong foundingcorrect classification was the signifi- male bias among collectionsof Long-billed cant variationwithin geneticallydefined taxa Larks (71%of specimensexamined were male). in termsof morphologyand, especially, plum- January 1999] Long-billedLark SpeciesComplex 201

curvirostris

brevirostris

subcoronata

benguelensis 0.5 semitorquata 0 chuana -0.5

-1

bengue•ns•

kaokens•

chuana damarensis

falcirostris

curvirostris )lgida

brevirostr• 1.5

0.50 • -0.5'• -1

FIG.4. Variationin plumagestreaking (both sexes) and size(males only) amongdescribed subspecies in the Long-billedLark complex,showing clinal variationwithin subcoronataand semitorquata.Approximate rangesof subspeciesare depictedon the map.Principal components are derivedfrom the samemeasure- mentsas thosein Figure3. 202 RYANAND BLOOMER [Auk,Vol. 116

5 -A semitorquata

1 'semitorquata transvaalensis $tutterheim NottinghamRd Wakkerstroom Johannesburg 0

4 tB ,subcoronata

1 •g i l l t subcoronate Beaufort West Woodhouse 0

,, falcirostris .curvirostris Port Nolloth Olifants River Vredenburg Paternoster

-D brevirostris benguelensischuana --

Protein Elim Brandberg Pietersburg

1 2 3 4 5 Duration (s) January1999] Long-billedLark Species Complex 203 age(Fig. 4). Thisvariation appears to be clinal: differencescoinciding with speciesboundaries body sizeand streakingdecrease from southto definedby mtDNA increaseconfidence in the northin bothsemitorquata and subcoronata (Fig. validity of the five species.Further work is 4). This is especiallymarked in the amountof needed,especially in Namibia,but it is prudent streakingwithin the subcoronatagroup, which to recognize five speciesfrom a conservation is bimodal(Fig. 3B), suggesting a steppedcline. perspective.The cytochrome-bdivergence val- The abrupt changein theseclinal trendsbe- uesbetween the species(2 to 9%) are similarto tweendamarensis and kaokensis (Fig. 4) supports those reported for other bird species(Shields our contention,based in part on atlasdata, that and Helm-Bychowski1988, Helbig et al. 1996), birds from central Namibia (damarensis)are includinglarks (Ryan et al. 1998). part of the subcoronatagroup rather than the At least in the south,the Long-billedLark northernbenguelensis-kaokensis group. speciesare allopatric or narrowly parapatric, Only 25 male songswere availablefrom 19 with cleargaps (or at leastareas of low report- localities.There is little structureto the display ed occurrence)between species' ranges (Dean song,with thatof Short-clawedLark being sim- et al. 1997;Fig. 1). The only area where two ilar to mostLong-billed Lark taxa (Fig. 5A-E). speciesare known to co-occur is where the Most data are available for the eastern semitor- Richtersveldabuts the coastalplain near the quata,where variation among four individuals OrangeRiver mouth,where curvirostris(falci- within a singlelocality is equivalentto that rostris)and subcoronatahave been collectedto- among six widely spacedlocalities. Song dif- gether with no evidence of interbreeding ferencesappear to be consistentbetween some (Quickelberge1967). Interestingly,this is the taxa. The mostdistinctive feature is the pres- samearea where the newly describedBarlow's enceor absenceof an initial notepreceding the Lark (C. barlowi)replaces the KarooLark (Ryan descendingwhistle. This initial note is devel- et al. 1998).Further work is requiredin Namib- oped most fully in brevirostrisand, to a lesser ia to elucidatethe boundarybetween subcoron- extent, curvirostrisfrom the southernpart of ata and benguelensis,which is problematicbe- the range(nominate curvirostris from nearVre- causethe two are morphologicallythe most denburg;Fig. 5C, D). However,a reducedini- similarspecies pair. However,the cleargap in tial note also occurs in subcoronata from south reported range at 21ø30'S(Dean et al. 1997), of the OrangeRiver that is absentfrom semi- coupledwith the steppedshift in phenotypes torquata(Fig. 5B). Two subcoronatarecorded in at this latitude,suggest this asthe boundary. If southern Namibia lacked the initial note, but this is the case,Roberts' (1936, 1940)damarensis thisdifference did not coincidewith thechange merely formsthe northernlimit of subcoronata, in plumagefrom subcoronatato bradshawLThe henceour use of benguelensisfor the northern only speciesthat appearsto exhibitsignificant species. variationin its displaysong is the west coast Our recommendationthat five speciesbe rec- curvirostris(Fig. 5C). ognized concurswith Roberts' (1936, 1940) treatment of the Long-billed Lark complex, DISCUSSION with the addition of brevirostris(which Roberts describedin 1941).Roberts (1940) used the En- The large mtDNA sequencedifferences glish names:Cape Long-billedLark (curviros- amongtaxa suggestthat the Long-billedLark tris), Eastern Long-billed Lark (semitorquata), complex comprisesfive species. Given the Karoo Long-billedLark (subcoronata),and Da- small samplesizes, the possibilityof intraspe- mara Long-billedLark (damarensis).Given the cificpolymorphism in mtDNA cannotbe ruled confusionregarding the statusof damarensisas out,but the consistentmorphological and vocal a possiblesubspecies of subcoronata,we suggest

FIG.5. Sonogramsof maledisplay songs in theLong-billed Lark complex. Each song represents a separate individual.(A) Variationwithin semitorquata(note similarity between subspecies). (B) Variationwithin subcoronata(subspecies in SouthAfrica are similar,but Namibianbradshawl lack the initial note). (C) Variation within curvirostris(note marked difference between subspecies). (D) brevirostris,benguelensis and C. chuana. 204 RYANAND BLOOMER [Auk, Vol. 116 that BenguelaLong-billed Lark is moreappro- bers of Certhilauda(Dean et al. 1992), in bill and priatefor benguelensis,and that AgulhasLong- body proportions,display flight, and nest billedLark be usedfor brevirostris,given that it structure(Herremans et al. 1994).The sequence is virtuallyrestricted to the AgulhasPlain. data confirmthat the Karoo and Long-billed Intraspecificvariation.--Distinct regional var- lark complexesare not closelyrelated, being iants occurwithin at leastthree of the Long- unableto resolvethe trichotomybetween these billed Lark species.The widespreadspecies two taxaand a finch-lark(Eremopterix). Our un- subcoronataand semitorquataexhibit marked publishedsequence data suggestthat the Ka- clinal variationin plumagestreaking and, to a roo Lark complexis closelyallied to the Sabota lesserextent, size. It is intriguing that these Lark (Mirafrasabota) and is ratherdistant from clinesrun in the samedirection in bothspecies. the Long-billedLark complex(Ryan et al. The decreasein streakingfrom southto north 1998). Accordingly, we support Herreman et amongsubcoronata parallels that found in other al.'s (1994) recommendationthat the Karoo larksin the region(e.g. Macdonald 1953, Law- Lark complex be dropped from Certhilauda, son 1961, Ryan et al. 1998) and corresponds whichis now restrictedto theLong-billed Lark with a decreasein shrubcover. Streaking may complex,and possiblythe SomaliLark (C. so- thusbe adaptive,providing camouflage in ar- malica),which currently is placed in Mirafra easof dappledshade in the south,and being (Deanet al. 1992).The KarooLark complexcan replacedwith uniformplumage in more arid be includedin Mirafrafor the time being, al- openhabitats farther north. This explanation is thoughthis genus is in needof revision. lessplausible for variationwithin semitorquata, Range,habitat use, and conservation.--Together which occurs in primarily grassy habitats with the Short-clawedLark, the five species throughoutits range(although the southwest- within the Long-billedLark complexoccupy ern areascontain more shrubs).The variation mostof southwesternAfrica (Fig. 1). Basedon within semitorquataappears to be gradual,with atlas data (Dean et al. 1997), the crude (maxi- little need to recognizewell-defined subspe- mum)ranges of thefive speciesvary from ap- cies. However, variation within subcoronatais proximately 15,000km 2 for brevirostristo some bimodal(Fig. 3B), and we advocatethe use of 400,000km 2 for subcoronata.Habitats occupied subspecificnames, if only as convenientlabels, differ among species,and atlas data suggest for subcoronatafor the southern,large, heavily that differencesin abundanceexist among spe- streaked birds, and bradshawifor the northern, cies.Most populations are patchily distributed, smaller,plain birds. however,and occurat fairly low densitiesrel- Samplesizes available for thisstudy were too ative to other resident larks in southern Africa. small to assess the status of forms within ben- The northernform of curvirostris(falcirostris) guelensisand the otherNamibian taxa. This is a is common from south of the Olifants River to priority for further study.There were alsotoo extreme southern Namibia, where it is restrict- few specimensof curvirostris(sensu stricto) to ed to dunesand adjacentstrandveld vegetation fully assessits statusrelative tofalcirostris. At- of the Namaqualandcoastal plain. The south- las data (Dean et al. 1997) indicate a gap be- ern form (nominate curvirostris)occurs more tweenthe ranges of the two formssouth of the sparinglyon coastaldunes from St.Helena Bay OlifantsRiver, suggesting that they are rather to Langebaan, and inland on Malmesbury discrete.The southernspecies, brevirostris, has shalessouth and eastto Tygerbergand Gouda. the smallestrange, and no regionalvariation The latter area has largely been modifiedfor was detectedamong the small seriesof speci- agriculture,but the speciesoccurs in arid mens available. However, further work is re- wheatfields,especially on the ColumbinePen- quiredto elucidatethe relationships among bre- insula.It doesnot appearthreatened, provided virostris,curvirostris, and falcirostris, given that current land-use practicespersist. Dean et al. curvirostrisappears to be intermediatebetween (1997)suggest that its rangeincreased as a re- thetwo taxain morphology,and perhaps closer suit of agriculture. to brevirostrisin termsof songand habitat use. The AgulhasLong-billed Lark (C. breviros- Relationshipswithin the Alaudidae.--The Long- tris)is restrictedto the AgulhasPlain from Bot billed Lark complexdiffers markedly from the River to George,with an extensionup the Karoo Lark complex,the otherputative mem- Breede River valley. It occurs primarily in January1999] Long-billedLark Species Complex 205 stony wheatfieldsand pasture lands on the iousSouth African museums for allowingfree access Agulhas Plain but is also found in succulent to skin collections,and ColeenMoloney for assisting Karoo vegetationin the BreedeRiver valley. with capturing morphologicaldata. We are also Both brevirostris and curvirostris avoid moun- gratefulto regionalconservation agencies for grant- tain fynbos, with the Cape fold mountains ing permits to collect birds. We received financial and logisticalsupport from the Foundationfor Re- forming a naturalbarrier betweenthe two spe- searchDevelopment, the StateMuseum of Namibia, cies.Despite its limited range, brevirostrisalso and the Universityof Cape Town'sResearch Com- appears to be secureprovided current condi- mittee. tions persist.However, further study to assess habitat requirementsfor breedingis warrant- LITERATURE CITED ed, giventhe species'patchy distribution within its range. Very little, if any, of the species' AMOS,B., ANDA. R. HOELZEL.1991. Long-term pres- rangeis conservedwithin protectedareas. ervationof whaleskin for DNA analysis.Report The eastern semitorquataoccurs sparsely of the InternationalWhaling CommissionSpe- throughoutits large range, preferring rocky cial Issue 13:99-104. slopes and ridges in grasslands(Dean et al. AViSE, J. C., W. S. NELSON, AND C. S. SIBLEY.1994. 1997).Afforestation poses a threat to the spe- DNA sequencesupport for a closephylogenetic ciesin the eastof the range (the easternescarp- relationship between some storks and New ment and KwaZulu-Natal midlands),but it ap- World vultures. Proceedingsof the National Academyof SciencesUSA 91:5173-5177. pearsto be securein the easternKaroo and ad- CLANCEY,P. g. 1957.On the rangeand statusof Cer- jacent semiaridgrasslands. The Karoo Long- thilaudafalcirostris Reichenow, 1916: Port Nol- billed Lark (subcoronata)is widespread and loth, N.W. Cape. Bulletin of the British Orni- abundant,occurring in semiariddwarf shrub- thologists'Club 77:133-137. lands of the Nama and succulentKaroo, as well CLANCEY,P. g. 1980. S.A.O.S. checklist of southern as southernNamibia. It prefersrocky or stony African birds. SouthernAfrican Ornithological areas but also occurs in other habitats. It is not Society,Pretoria, South Africa. currentlythreatened, but verylittle of therang- DEAN, W. R. J., D. G. ALLAN, M. HERREMANS,P. G. esof eithersemitorquata or subcoronata fall with- RYAN,AND C. COHEN.1997. Family Alaudidae: in formal conservationareas (Siegfried 1989). Larks.Pages 4-47, 687 in The atlasof southern African birds. Vol. 2, Passerines(J. A. Harrison, Atlas data suggestbenguelensis is fairly com- D. G. Allan, L. G. Underhill, M. Herremans, A. J. mon in the fore-Namib (Dean et al. 1997);its Tree, V. Parker, and C. J. Brown, Eds.). BirdLife statusin southernAngola is little known but it SouthAfrica, Johannesburg. is apparently "not uncommon" (W. R. J. Dean DEAN, W. R. J., C. H. FRY, S. KEITH, AND P. LACK. pers. comm.). It occursprimarily in arid and 1992. Family Alaudidae: Larks. Pages13-124 in semiarid grasslandand shrubland on rocky The birds of Africa, vol. 4 (S. Keith, E. K. Urban, hills. It probablyfaces few threats,and a sub- and C. H. Fry, Eds.). AcademicPress, London. stantial proportion of its Namibian range is DEAN, W. R. J., AND P. A. R. HOCKEY. 1989. An eco- protectedwithin the SkeletonCoast Park. logical perspectiveof lark (Alaudidae) distri- In summary,although probably none of the bution and diversityin the southwest-aridzone of Africa. Ostrich 60:27-34. newly recognizedspecies is threatened,popu- DESJARDINS,P., AND R. MORALS.1990. Sequenceand lations of several speciesdo not receive ade- geneorganization of the chickenmitochondrial quate protectionin formal conservationareas. genome.A novel gene order in higher verte- Consequently,this type of study is important brates. Journal of Molecular Biology 212:599- to identify cryptic specieswithin polytypic 634. speciescomplexes. EssoP, M. E, P. H. LLOYD, H. J. VAN HENSBERGEN, AND E. H. HARLEY.1991. Estimationof genetic ACKNOWLEDGMENTS distancebetween Bontebokand Blesbokusing mitochondrial DNA. South African Journal of Coleen Moloney,Joris Komen, and David Winter Science 87:271-283. providedassistance in the field,and Joris Komen also GRANT, P. R., I. ABBOTT,D. SCHLUTER,R. L. CURRY, collected material from northern Namibia. Lark re- AND L. K. ABBOTT. 1985. Variation in the size and cordingswere suppliedby CallanCohen, John Gra- shapeof Darwin'sFinches. Biological Journal of ham,Penn Lloyd, and the FitzPatrickSound Library, the LinneanSociety 25:1-39. Transvaal Museum. PGR thanks the curators at var- FARRIS,J. S. 1969. A successiveapproximation ap- 206 RYANAND BLOOMER [Auk, Vol. 116

proachto characterweighting. Systematic Zo- simplefools guide to PCR, version2.0. Depart- ology 18:374-385. ment of Zoologyand KewaloMarine Laboratory, FELSENSTEIN,J. 1985. Confidencelimits of phyloge- Universityof Hawaii, Honolulu. nies:An approachusing the bootstrap.Evolu- QUICKELBERGE,C. D. 1967.The racial taxonomyof tion 39:783-791. south-eastAfrican populationsof the Long- HELBIG, A. J., J. MARTENS, I. SEIBOLD,E HENNING, B. billed Lark. Annals of the Cape ProvincialMu- SCHOTTLER,AND M. WINK.1996. Phylogeny and suem (Natural History) 6:39-46. specieslimits in the PalaearcticChiffchaff Phyl- ROBERTS,A. 1936. Ornithologicalnotes. Annals of loscopuscollybita complex: Mitochondrial genetic the Transvaal Museum 18:255-269. differentiation and bioacoustic evidence. Ibis ROBERTS,A. 1940. The birds of South Africa. H. E and 138:650-666. G. Witherby,London. HERREMANS,M., N. D. HUNTER, AND D. ALLAN. 1994. ROBERTS,A. 1941.Notes on somebirds of the Cape The displayof the Short-clawedLark Certhilauda Province. Ostrich 11:112-135. chuanaand commentson the genusCerthilauda. RYAN,P. G., ANDP. BLOOMER. 1997. Geographic var- Bulletin of the BritishOrnithologist' Club 114: iation in Red Lark Certhilaudaburra plumage, 27-31. morphology,song and mitochondrialhaplo- HILLIS,D. M., AND J.P. HUELSENBECK.1992. Signal, types.Ostrich 68:31-35. noiseand reliability in molecularphylogenetic RYAN, P. G., I. HOOD, P. BLOOMER,J. KOMEN, AND t. analysis.Journal of Heredity 83:189-195. M. CROWE.1998. Barlot's Lark: A new species KIMURA,M. 1980. A simple method of estimating in the KarooLark complexof southwestAfrica. Ibis 140:605-619. evolutionaryrates of basesubstitutions through comparativestudies of nucleotide sequences. SAITOU,N., ANDM. NEI. 1987.The neighbor-joining Journal of Molecular Evolution 16:111-120. method:A new methodfor reconstructingphy- KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. ED- logenetictrees. Molecular Biology and Evolution 4:406-425. WARDS, S. PAABO, E X. VILLABLANCA, AND A. C. SANGER, F., S. NICKLEN, AND A. R COULSON. 1977. WILSON.1989. Dynamics of mitochondrialDNA DNA sequencingwith chain-terminatinginhib- evolution in animals: Amplification and se- itors. Proceedingsof the NationalAcademy of quencingwith conservedprimers. Proceedings Sciences USA 74:5463-5467. of the National Academyof SciencesUSA 86: SHARPE,R. B. 1904. On a collection of birds from the 6196-6200. districtof Deelfonteinin Cape Colony.Ibis 8(4): KUMAR, S., K. TAMURA, AND M. NEI. 1993. Molecular 1-29, 313-367. evolutionarygenetics analysis (MEGA), version SHIELDS,G. E, AND K. M. HELM-BYCHOWSKI.1988. 1.01. Institute of Molecular and Evolutionary Mitochondrial DNA of birds. Current Ornithol- Genetics,Pennsylvania State University, Penn- ogy 5:273-295. sylvania. SIBLEY,C. G., AND B. L. MONROE, JR. 1990. Distri- LAWSON,W. J. 1961. The races of the Karoo Lark Cer- bution and taxonomyof birds of the world. Yale thilaudaalbescens (Lafresnaye). Ostrich 32:64-74. University Press,New Haven, Connecticut. MACDONALD,J. D. 1952.Notes on the Long-billLark SIEGFRIED,W. R. 1989. Preservationof speciesin Certhilauda curvirostris. Ibis 94:122-127. southernAfrican nature reserves. Pages 186-201 MACDONALD,J.D. 1953.Taxonomy of the Karrooand in Biotic diversityin southernAfrica: Concepts Red-back larks of western South Africa. Bulletin and conservation(B. J. Huntley, Ed.). Oxford of the BritishMuseum (Natural History) Zool- University Press,Cape Town. ogy 1:321-350. SMITH, E. E G., P. ARCTANDER,J. FJELDSA,AND O. G. MACLEAN,G. L. 1993. Roberts/birds of southernAf- AMIR. 1991.A new speciesof shrike(Laniidae: rica. Trustees of the John Voelcker Bird Book Laniarius)from Sorealia,verified by DNA se- Fund, Cape Town. quencedata from the only known individual. MADDISON, W. P., AND D. R. MADDISON. 1992. Ibis 133:227-235. MacClade, version 3. Sinauer Associates, Sun- SWOFFORD,D. L. 1993.PAUP: Phylogenetic analysis derland, Massachusetts. using parsimony,version 3.1.1. Illinois Natural PALUMBI, S., A. MARTIN, S. ROMANO, W. O. MCMIL- HistorySurvey, Champaign. LAN, L. ST•CE, AND G. GRABOWSKI.1991. The AssociateEditor: A. J. Baker January1999] Long-billedLark Species Complex 207

APPENDIX.Nucleotide sequences of 483 basepairs of the mitochondrialcytochrome-b gene of the Long- billed Lark complexaligned with that of the KarooLark ("Certhilauda"albescens) and Gray-backedFinch- Lark (Eremopterixverticalis). Dots indicateidentity to thefalcirostrissequence, and questionmarks indicate undeterminedbases. These sequences correspond to positionsL15 010 to 15 492 in the publishedchicken sequence(Desjardins and Morais 199).

[601 falcirostris ATCTGCCTGA TCATGCAAAT CATCACAGGC CTCCTACTAG CCATGCACTA CACTCCGGAC curvirostris brevirostris semitorquata ...... A ...... A ...... subcoronata A ...... A ...... A- -T- ß T- -A- .A. subcoronata B ...... A ...... A..T- - T..A- -A- kaokensis A ...... G-A ...... G ...... A- .T. ß T- -A- -A. kaokensis B ...... G-A ...... G ...... A- .T. ß T..A. -A- chuana ...... A ...... CA ..... T- -A- -A. albescens ...... A .... CA ...... T ...... CA ...... AG-A. Eremopterix ß .T ..... A ..... T ...... T ...... A ..... C ..... A ...... AG ..... [1201 falcirostris ACCTCCCTAG CCTTCGCCTC TGTCGCTCAC ATCTGCCGAG ATGTTCAATT CGGCTGACTA curvirostris brevirostris semitorquata subcoronata A ..... T ...... T ...... C ...... C..C ...... T ...... subcoronata B ß -T- -T ...... T ...... C ...... C- -C ...... T ...... kaokensis A ..... T ...... C ...... C. -C ...... T ...... kaokensis B ..... T ...... C ...... T ..... C ...... T ..... ? chuana ..... TA ...... T. -T- ß C..T- -C ...... C ...... albescens ...... C ..... A ..... T ...... C. .C ..... T ...... C E remopterix ...... T ...... C ..... T ...... C..C ...... A ...... [1801 falcirostris ATCCGAAACC TCCACGCAAA CGGAGCATCC CTCTTCTTCA TCTGCATCTA CCTCCACATT curvirostris breviros tris ...... C semitorquata subcoronataA .... "C ...... C ..... A ...... TT ...... C subcoronata B ..... C ...... C ..... A ...... T ...... C kaokensis A ..... T ...... C ...... T. ß .A ...... C kaokensis B ..... T ...... C ...... T- - -A ...... C chuana albescens ..... C- -T. -A ...... C- - - T ...... T ...... C Eremopterix ...... A- .T ...... T ..... A ...... G ...... T ...... [240] falcirostris GGCCGAGGAC TATACTACGG CTCATACCTA AACAAAGAGA CCTGAAACGT AGGAATCATC curvirostris brevirostris ...... A ...... semitorquata ...... G ...... A ...... subcoronata A ...... A ...... T ...... G ..... subcoronata B kaokensis A ...... A ...... GG.T. ß - kaokensis B ...... A ...... GG.T. - - chuana ß .T ...... T ...... A ...... T- - ß albescens ...... T -C ...... G ...... AG ...... A- C- - -G- .C. - Eremopterix ...... T .T ...... A ...... C. - -G ..... [3oo] falcirostris CTCCTCCTGC TACTCATGGC CACTGCTTTC GTAGGGTACG TTCTCCCCTG AGGACAAATA curvirostris brevirostris ...... A ...... semitorquata ...... A ...... C ...... A ...... subcoronata A ...... A ...... A ..... C..T ...... 208 RYAN AND BLOOMER [Auk, VoL 116

APPENDIX. Continued. subcoronata B kaokensisA ß -T ..... AT ...... A- ß A ...... A. kaokensis B ß -T ..... AT ...... A. ß A ...... A. chuana ...... C ...... A- albescens ..... T- -AG CC ..... A- - A ..... C ...... A- ---C-.T--A ...... G Eremopterix -C ..... A- - A- -C- -A- -T ..... A. -- -C ...... C ...... [360] falcirostris TCATTCTGAG GGGCTACAGT AATCACAAAC CTACTCTCAG CCA?CCCATA TGTTGGTCAG curvirostris brevirostris semitorquata subcoronataA subcoronata B kaokensis A ...... C ...... T- .C ...... kaokensis B chuana ß -C ...... T ...... T--C ..... C..C... albescens ...... A ...... T ..... GT ...... A ...... CA .... A..A Eremopterix ...... C ...... T.T ...... C- - CA .... C. ß ß [4201 falcirostris ACCCTGGTAG AGTGGGCGTG GGGGGGGTTC TCAGTAGATA ACCCTACTCT CTCCCGATTC curvirostris ...... T brevirostris semitorquata subcoronata A ..... A ...... A .... G- - - subcoronata B kaokensis A ...... A ...... C ...... A ...... kaokensis B ...... A ...... C ...... A ...... chuana ..... A ..... A. -A ...... A ...... C ...... A ...... albescens ..... A ..... A- -A ...... T - -G ..... C. -T- -C ...... A .... C. ß ß Eremopterix ..... A ..... A..A- .A- - A- .A- -A ..... C ..... C ..... C- -C- - -A-T ...... [483] falcirostris TTCGCCCTCC ATTTCCTACT CCCATTCCTT ATCGCAGGCC TCACACTAGT TCACCTCACC CTC curvirostris brevirostris ...... ? ...... semitorquata subcoronataA ...... G ...... T ...... T subcoronata B ...... T ...... C ..... T kaokensis A ...... C ...... T. -C ...... T kaokensis B ...... C ...... T..C ...... T chuana ß-C ...... T ...... T--C ..... C--C ..... T albescens ...... A ...... T ..... GT ...... A ...... CA .... A..A ..? Eremopterix ...... C ...... T-T ...... C.. CA .... C ......