Cooperative Breeding in Lanius Shrikes

COOPERATIVE BREEDING IN LANIUS SHRIKES. I. HABITAT AND DEMOGRAPHY OF TWO SYMPATRIC SPECIES

STEVEZACK • AND J. DAVID LIGON Departmentof Biology,University of New Mexico,Albuquerque, New Mexico87131 USA

ABSTRACT.--Twosyrupattic speciesof Laniusshrikes were studied near Lake Naivasha, Kenya, in an effort to understandthe ecologicalfactors favoring the evolution and mainte- nanceof cooperativebreeding in one of them. The Gray-backedFiscal Shrike (L. excubitorius), a cooperativebreeder, occupied territories of significantlyhigher tree and shrub coverthan did the Common Fiscal Shrike (L. collaris),which is not a cooperativebreeder. Areas with greatervegetational cover held significantlymore insectsin the dry monthsof the year than did relatively open sites. Possiblyassociated with the shrikes' differencesin habitats and resourceswas their significantdifference in disappearancerates during the study. Turnover of individualsamong gray-backs was abouthalf that of commonfiscals. Similarly, territorial stability (asmeasured by the percentof territoriescontinually occupiedduring the 18-month study) was nearly twice as high in the gray-backs.Within the restrictedacacia woodland gray-backswere dominant to commonfiscals, and in four instanceswe observedgray-backs expel common fiscalsfrom the latter's territories. We suggest that cooperative breeding in gray-backsis related to occupancyof a temporally stable,but spatially restricted,habitat of high quality. This in turn may lead to relatively higher survivorshipin gray-backsand, as a result,their habitatbecomes relatively "saturated."In suchan ecologicaland demographic setting, options for juvenile dispersalare restricted,and one evolutionary solution is group living. Received5 October1984, accepted 11 April 1985.

COOPERATIVELYbreeding birds have received Stacey(1979) related differences in groupsize, considerableattention in recent years (see re- site tenacity, and reproductiverates between views by Brown 1978; Emlen 1978, 1984). Se- New Mexico and California populationsof the lander (1964) first suggestedhabitat saturation Acorn Woodpecker(Melanerpes formicivorus) to asa critical factorfavoring the evolution of such differences in the degree of habitat saturation socialsystems. Habitat saturation occurswhen and in demographic patterns in the two areas. all usable habitat in an area is filled with breed- Koenig and Pitelka (1981) refined the habitat ing pairs. An important consequenceof habitat saturationhypothesis (emphasizing "resource saturationis the inability of nonbreedingbirds localization" and "habitat localization") and to establisha territory in the local habitat. Re- postulatedthat a relative scarcityof marginal tention of young in the natal territory until a habitat is instrumental in the evolution of co- breeding vacancy occurs is one response to operative breeding. Atwood (1980) used this habitat saturation.Cooperative breeding often line of reasoningto explain why Santa Cruz is thought to be a result of such habitat restric- Island ScrubJays (Aphelocoma coerulescens insu- tions and retention of offspring in the natal laris), which occupyareas with extensive mar- territory. Only recently, however, has the in- ginal habitat, are not group-breeders,whereas fluence of habitat on the development or main- Florida ScrubJays (A. c. coerulescens),which are tenance of cooperativebreeding systemsbeen more habitat-restricted, do exhibit cooperative investigated in a quantitative manner (e.g. breeding (Woolfenden 1975, Woolfenden and Brown and Balda 1977, Gaston 1978a, Craig Fitzpatrick 1978). 1979, Dow 1980a, Trail 1980, Koenig 1981a, In this paper we comparetwo sympatriccon- Brown et al. 1983). generic shrikes, one species a cooperative breeder and the other a more typical (for the •Present address:Department of Biological Sci- genus) noncooperativebreeder. We then con- ences, Purdue University, West Lafayette, Indiana sider the possibleroles of habitat and resources 47907 USA. in the evolution and maintenance of these two

754 The Auk 102: 754-765. October 1985 October1985] ShrikeCooperative Breeding 755 socialsystems. The Gray-backedFiscal Shrike num band on one leg and a unique color combina- (Laniusexcubitorius) is a cooperativelybreeding tion of two color bands on the other. Nestlings of species,with socialunits ranging in size from both specieswere similarlybanded at 10 daysof age. 2 to 9 or more birds. Gray-backsappear to be Groups(gray-backs) and pairs(common fiscals) were monogamous,with only one breeding pair per censusedat irregularintervals, but generallyat least once a month. Both speciesare sexuallydimorphic group (Zach MS). The supernumerarybirds in in plumage,as femaleshave auburnflanks (Grimes a flock act as "helpers." The breeding biology 1979)that, althoughdifficult to observein the field, of gray-backsis very similarto that described are readily observedin the hand. for anothercooperatively breeding laniid, the Both shrike speciesnested with the onsetof the Yellow-billed Shrike (Corvinella corvina; Grimes majorrains. Most nests were inaccessible due to their 1980). In contrast,the sympatricCommon Fis- heightand frailty of the supportinglimb. The num- cal Shrike (L. collaris)occurs only as breeding ber of fledgedyoung was recorded from all success- pairs (Zack MS). First-year Common Fiscal ful nestsfor both species. Shrikes disperseupon maturity, usually coin- Habitat sampling.--Twelvegray-back and 6 common fiscal territories were measured using a modi- cident with molt into adultlike plumage. Gray- backs are restricted to Acacia woodlands in east fied point-quartermethod (Cottamand Curtis 1956) asfollows: Beginning 50 m eastof the southwestcor- and central Africa, whereas common fiscals oc- ner of a given territory, a north-southtransect was cur throughoutmost of the Afrotropicalregion established. Each transect consisted of five 40-m in- in a variety of habitats--most commonly in tervals,with eachinterval having one randomly de- openand disturbed(especially cultivated) areas terminedpoint. Fromeach point, the distance(in cm) (Hall and Moreau 1970, Britton 1980, Mack- to the nearesttree (>1.5 tn in height) to the west, worth-Praed and Grant 1960). and then againfrom the randompoint to the nearest individual tree to the east, was measured. Occasion-

STUDY AREA AND METHODS ally, this led to repeatedsampling of the sametree in relativelyopen areas. These measures formed the Studyarea.--The study site is locatedon Morendat basisof the point-treedata. From eachof thesetrees, Farm near Lake Naivasha, Kenya, in the Rift Valley the distanceto its nearestneighboring tree was mea- of eastAfrica (0•40'S, 36ø23'E). Ligon and Ligon (1978) sured for use in dispersionestimates. Dispersion es- have describedthis area.The vegetationis character- timatesare from the equation• p2/(• p2+ • n2),where ized by a virtually monotypicstand of the yellow- p is the point-to-plantdistance and n is the plant-to- barked acacia (Acaciaxanthophloea) with an under- nearest-neighbordistance (Hopkins and Skellam 1954, story of Naivashastar grass(Cynodon plectostachyus) Batcheller and Bell 1971). The diameter at breast and a few speciesof perennial shrubs (most com- height (DBH) was measuredfor each tree encoun- monly Achyranthesaspera, Hypoestes verticillaris, and tered. DBH measureswere taken only once and were Solanumincanurn). Thickets of Euphorbia,Aloe, or not usedagain if the treewas encountered more than Opuntiaoccur uncommonly in woodlandopenings. onetime. A l-m: grid alsowas established at the orig- To the eastand south of the study site are more open inal random point, and the percent coverageof pe- areaswith only scatteredacacias. To the west toward rennial shrub cover was determined visually. This Lake Naivasha is a denser woodland composed of shrub measurementwas repeatedin l-m: quadrats5 several tree species.To the north is a large plowed and 10 m west of the point and 5 and 10 m east of field under constantcultivation (see Fig. 1). The farm the point, making five such measurementsper ran- is grazed by domesticcattle, domesticsheep, six dom point. At each quadrat,tree cover was recorded speciesof native bovids,and hippopotamus(Hippo- (presentor absent),permitting a relativemeasure of potamusamphibius ). percent shade cover. This processwas repeated 15 Rainfall measurements were taken from records times per territory (3 parallel transectscontaining 5 kept by the farm managersof Morendatand adjacent random point lines spaced75 m apart), giving 30 Marula farms. Rainfall sometimes occurs in a bimod- point-to-treemeasures, 30 tree-to-treemeasures, and al patternbut is highly unpredictablein both timing 75 perennial shrub cover and tree cover measures. and quantity.The generalpattern is that of substan- Five territories (2 gray-backand 3 common'fiscal) tial rains in April-May with lighter rains during oth- were too small to accommodate three transect lines, er months. and for these only two transects(for a total of 20 Shrikes were studied from June 1979 to December random point measurements)were sampled. All 1980 and from August 1981 to early January 1982. measurements were made in the woodland; no mea- Birdswere trapped in mist nets or caught in spring- sureswere taken in the plowed field. Three common loaded traps baited with large insectsor small frogs, fiscalterritories (osf, gpf, and tf in Fig. 1) contained and then banded with a uniquely numbered alumi- no acaciasand were not measured.The plowed field 756 ZACKAND LIGON [Auk, Vol. 102

TABLE1. Resultsof the vegetationalsampling per territory for each specieson Morendat Farm, Naivasha, Kenya. Territory namesare the sameas in Fig. 1. n refers to the number of random points sampledper territory from which two point-to-treemeasures were made.Tree diameterat breastheight (DBH) measures (sample sizes in parentheses)do not include repeat samplings.

Tree Random point to tree(cm) Shrubcover (%) cover DBH(cm) (%) Disper- Territory n œ SD œ SD • (n) • SD sion Gray-backedFiscal Shrikes RM 30 1,451.5 739.9 9.24 16.34 34.7 (49) 34.4 42.7 0.524 NWW 30 1,450.4 889.9 10.52 19.88 49.3 (56) 42.8 42.8 0.480 RYU 30 1,864.1 1,007.1 10.40 17.36 28.0 (47) 39.6 15.9 0.624 BM 30 1,259.5 615.4 8.71 24.01 28.0 (42) 31.8 23.8 0.500 CW 30 949.2 582.9 3.88 9.78 45.3 (49) 24.1 26.1 0.550 KM 30 901.0 421.4 10.67 14.76 52.0 (57) 23.6 14.3 0.622 MN 30 1,346.6 653.2 4.27 13.87 50.7 (58) 35.0 17.7 0.552 NYM 30 1,348.9 793.8 9.76 19.71 41.3 (46) 35.3 28.4 0.529 PYM/NS 20 988.1 593.5 6.30 14.81 46.0 (33) 29.9 23.4 0.476 SK 30 836.7 444.5 16.00 27.44 56.0 (51) 25.6 23.5 0.608 MYN 30 641.1 420.6 22.67 28.36 64.0 (52) 16.0 16.9 0.508 HKK 20 1,919.9 850.6 3.30 8.84 10.0 (29) 17.3 17.9 0.586 Total 340 1,234.2 779.1 9.93 19.81 42.9 (569) 30.0 27.4 0.546 Common Fiscal Shrikes hf 30 1,987.8 1,253.3 5.40 17.80 29.3 (48) 49.0 50.3 0.656 vpf 30 1,419.4 817.7 5.96 12.20 45.3 (55) 33.2 38.5 0.560 cpsf 20 1,946.8 1,083.5 0.60 2.60 16.0 (31) 34.8 41.4 0.647 cpnf 20 1,482.9 725.0 12.70 21.50 50.0 (28) 34.5 24.4 0.575 wwhf 30 1,634.7 1,255.0 1.67 8.32 25.3 (50) 16.0 17.5 0.613 wgf 20 1,790.1 854.4 5.70 18.46 20.0 (34) 37.4 20.7 0.545 Total 150 1,704.4 1,049.4 5.14 15.00 31.4 (246) 33.7 36.3 0.599

was usedextensively by the commonfiscals and to a orthopteranswere eaten by the shrikes,whereas no lesserextent by the gray-backsfor feeding.However, shrike was observedto eat hymenopteransor lepi- no measurements could be made of insect abundance dopteran larvae with urticating hairs. there. Vegetational characteristicsof the grid sites were To assessprey availability, three pitfall trap grids measuredin the following way: Perennialshrub cov- (traps 10 m apart in a 5 x 5 configuration) were set er wasestimated using the 1-m2 grid describedabove, up in areasof different vegetationalstructure (low except that measurementswere made 5 m west of tree density/low shrubcover, high tree density/high each pitfall trap and shrub cover was recorded as shrub cover, and an area of intermediate value for either present or absent. Tree shade cover was re- these measures;see Table 2 for values)to assesspos- cordedat thesesame points. Trees were mappedrel- sible differencesin the prey baseof the insectivorous ative to the position of the pitfall traps, and DBH shrikes in different microhabitats.Pitfall traps were was measured for each tree. opened for 8-h daytime periods during alternate Data were analyzedusing SAS programs(SAS In- weeks throughout 1980.Cattle dip (a locally usedin- stitutes Inc., SAS Circle, P.O. Box 8000, Cary, North secticidesprayed on cattle) was put in each of the Carolina 27511). trapsto retain capturedinsects. The insectswere then stored in methanol and water until all "edible" in- RESULTS sectsequal to or greater than 1 cm in length could be measured to the nearest mm. The number of edi- Habitatand shrike distribution. --Significant dif- ble insectscollected in pitfalls less than 1 cm was ferences existed between the territories of the negligible.Edibility was assessedby direct observationsof foragingand feedingshrikes throughout the two shrike specieswith regard to tree density study. Both shrike speciesappeared to have great (expressedin terms of random point-tree mea- overlapin the kindsand sizesof prey captured(Zack sures,one-way ANOVA, df = 1 and 489, F = in prep.).Coleopterans, "naked" lepid,0pteran larvae 350.14,P < 0.0001)and percentperennial shrub (without urticating hairs), somehemipterans, and all cover (Kruskal-Wallis test, df = 1 and 1,222, F = October1985] ShrikeCooperative Breeding 757

O :Grey-back Territory(orig. flock size) O :Common FiscalTerritory STUDY500m SITE• •1 : Immigration from outside study • :Inter-territory movement

-n : Number of birds died/disappeared

nj : Juveniles fledged (GBF only)

Fig.1. Thespatial and temporal pattern of Gray-backed Fiscal (open areas) and Common Fiscal (hatched areas)shrike territories during the study. The hatched regions in theupper left represent the area studied (birdsof both species did occur on the eastern edge of these areas). Original flock sizes for gray-backs in June1979 are indicated; forsubsequent months, only changes arenoted. Immigrations fromoutside thestudy areaare noted by short, bold arrows and designate single individuals unless otherwise noted. Interterritory movements(birds of known origin) are indicated bylong, thin arrows. The flock or pair letter designations correspondto thosegiven in Table 1.

44.58,P < 0.0001).Gray-backed Fiscal Shrikes Wallis test, df = 1, 0.05 < P < 0.10) nor the occupiedterritories of significantlyhigher tree DBH of acacias(one-way ANOVA, df = 1 and density (lower point-to-plantdistances) and 814,0.05 < P < 0.10)differed significantly be- higher perennial shrub cover than did com- tween territoriesof the two species. mon fiscals(Table 1). Gray-backsalso occurred Gray-backedFiscal Shrikes were strongly as- in habitat of higher tree cover than did the sociatedwith yellow-barkedacacia woodlands common fiscals (Table 1, G-test with WillJain's throughouttheir range, contrastingsharply correction,G = 14.57, df--- 1, P < 0.001). Nei- with the broad distributionalpattern of the ther the patternof tree dispersion(Kruskal- CommonFiscal Shrike. The datapresented here 758 ZACKAND LIGON [Auk, Vol. 102

TABLE2. Resultsof insectsampling during 1980 for the three grids of different vegetation cover.

Inter- o 200- 7, High cover mediate Low cover (shrub = (shrub = (shrub = 52%, 28%, 0%, tree = tree = tree = Date 60ø70) 48%) 12%) 2 Jan 4 2 1 15 Jan 1 0 1

E 2 Feb 5 7 2 22 Feb 4 3 1 13 Mar 17 34 10 26 Mar 5 13 4 11 Apr 99 92 100 24 Apr 137 156 100 8 May 112 128 110 23 May 11 17 37 6 June 5 11 21 20 June 9 23 7 4 July 15 11 10 18 July 5 5 8 29 July 6 7 13 18 Aug 6 8 8 31 Aug 6 8 2 13 Sept 5 2 4 29 Sept 7 2 3 15 Oct 5 0 2 31 Oct 15 2 3 Fig. 2. Patternof rainfall and "edible" insect(see 14 Nov 8 5 3 text) abundanceduring 1980 on Morendat Farm. For each month, the mean value of insectssampled (two samplingperiods, three grids) and the range of the grids are shown. tober 1979,aggression between the two shrike speciesbecame conspicuous. Such aggression consideronly commonfiscal territories occur- was directly observedin 4 of the 5 woodland ring in or at the edgeof acaciawoodland, and territoriesof the commonfiscals. The gray-backs thusare a very conservativecontrast of the dif- always won such encounters,apparently due ferences in the habitat occupied by the two to their larger size and numerical advantage. species.Common fiscals commonly were ob- We notedpersistent aggression by the gray-back servedthroughout the Rift Valley along road- flock BM (7 members) toward the common fis- sidesand in many kinds of open country. cal pair hf on 5 September(Fig. 1), in the form Behavioralinteractions of Gray-backedand Com- of displacementfrom perch sites and active monFiscal shrikes.--Although the data showed chases.Final expulsionof the commonfiscals a significantdifference in habitat characteris- took place on 5 October,when we observed tics of territories of the two shrikes, the two severallong chasesthat culminatedin a single shrikescan be highly syntopicon the study gray-backchasing the male Common Fiscal site. This was seen most clearly in June 1979 at Shrikeup and over the tall canopyand out of the beginningof the study.Five pairsof Com- sight.This pair of commonfiscals was not seen mon Fiscal Shrikes held territories within the again,and from that time on membersof the acaciawoodland away from the plowed field gray-backBM flock foragedregularly in what (Fig. 1). This degreeof overlapwith the larger previouslyhad been hf territory. The common Gray-backedFiscal Shrikes had not been ob- fiscalpair wgf also disappearedsoon after we servedduring less systematicobservations in first noted aggressionon 23 June between it the previous4 yr. Bothshrike species held ex- and the 9-membergray-back SK flock (Fig. 1). clusiveterritories against both conspecificsand At that time, pair wgf had fledged two off- each other. As the study area becameincreas- springin its territory.The breedingfemale of ingly drier during the monthsof August-Oc- the wgf pair and one offspringdisappeared in October1985] ShrikeCooperative Breeding 759

TABLE3. Statisticalcomparisons between the three pitfall trap grids and between seasons.All pairwise comparisonsare Wilcoxon's signed rank testsfor two groups (Sokal and Rohlf 1981; Ts refers to the test statistic;n refers to the number of pairwise comparisonsper category).Comparisons within seasons(wet season:April-August; dry season:January-March and September-November)comparing all three grids and the combinedcomparison for all grids for 1980are Friedmanrank tests(Brownlee 1965),with the Chi- squareapproximations listed.

Wet season Dry season Combined Grid comparison (amount of cover) Ts n P Ts n P Ts n P High-intermediate 7.0 8 NS 35.5 12 NS 79.5 20 NS High-low 20.0 9 NS 4.0 12 <0.01 71.5 21 NS Intermediate-low 22.0 9 NS 24.5 12 NS 85.5 21 NS All grids X2 = 1.06 X2 = 4.40 X2 = 0.06 P> 0.5 0. I

0.5

early August,probably taken by a predator,as SD = 0.009, n = 1,288; F = 1.48, P < 0.0005), but the upper mandible of the juvenile was found the absolutenumber of prey differed tremen- in the territory about a week later. The wid- dously between the samplesand so this result owed male and the remainingjuvenile disap- probably is not biologically meaningful. The peared in late September.By December 1980, two other differencesfound in average prey only one commonfiscal territory not adjacent size were observed when the low-cover grid to the plowedfield, wwhf, waspresent on the was comparedwith the intermediate-covergrid study area. No new common fiscal territories in the dry season (low cover: mean = 1.188, were established wholly within the acacia SD = 0.311, n = 34; intermediate cover: mean = woodland after the above-described events of 1.130, SD = 0.508, n = 69; F = 2.67, P < 0.005) late 1979. and when the high-cover grid was compared Prey abundanceand vegetationalcover.--There with the intermediate-covergrid during the wet was a generally concordantpattern between months (high cover:mean = 1.124,SD = 0.351, rainfall and insectabundance (Fig. 2; raw data n = 411; intermediate cover: mean = 1.118, SD = in Table 2). Data from the three pitfall grids 0.302, n = 461; F = 1.35, P < 0.005). Again, these were lumped together to show the seasonal differencesdo not appearto be biologicallyim- pattern (insectcaptures from the three grids portant to the shrikes. did not differ significantly overall and thus Comparativedemography.--Individual survi- were combined)(Table 3). There were signifi- vorship of adult gray-backswas 67% (41/61 cant differences(Kruskal-Wallis test, P < 0.01), birds surviving or staying on the study site) however, between the high- and low-cover from August 1979 to August 1980and 64% (34/ areasin edible insectavailability during the six 53 surviving) from August 1980 to August 1981 driest monthsof 1980 (Januarythrough March (note,however, that mortality couldnot be dis- and September through November). Other tinguishedfrom long-distanceemigrations). In pairwise comparisonsbetween pitfall areasand contrast, survival of adult Common Fiscal season(wet months, dry months) showed no Shrikes in the first period was 39% (7/18 sur- significantdifferences (Table 3). Threesignifi- viving). Adult common fiscalswere defined as cant differencesin averageprey size (of prey only those birds with bold black-and-white greaterthan or equalto 1 cm in size)were found plumage. Thus, dispersingjuveniles were not in 10 tests[with Bonferroni(1936) adjustment]. considered. Too few common fiscals were The testswere of all pairwisecomparisons of banded in 1981 to evaluate survival in relation the differentgrids in the two seasons,plus a to gray-backs.A Mantel-Cox Survival Analysis comparisonof overall prey size between the (Mantel 1966)of the 18-monthperiod from June wet and dry months.The averagesize of prey 1979 to December 1980 by speciesshowed a sampledin the dry monthswas significantly significantdifference in ratesof disappearance largerthan the averagesize of prey sampledin (P < 0.001; Fig. 3). No relationship was found the wet months(dry months:mean = 1.146cm, for either speciesas to when individuals dis- SD = 0.029, n = 183; wet months: mean = 1.137, appeared (wet seasonvs. dry seasonin 1980; 760 ZACKAND LIGON [Auk,Vol. 102

1.0 Individual turnover of common fiscals also is readily seen in Fig. 1. The number of fledged z • • Grey-backFiscal young is not indicated for this species,as no offspringbecame established on the study site .8 after fledging. An indication of the high level -) of individual and territorial turnover charac- .6 • •O•o z teristic of this speciesis shown by the cpwf o CommonFiscal .5 •o P • .001 pair. On 10 March 1980 the female breeder died •o (directly observed),and she was quickly re- o .,• placed by a bird from outside the study area. o By August 1981, the territory of the pair had .• •O•o shifted or been displacedfrom the edge of the .2 woodland to entirely within the plowed field, and a new pair of common fiscalsoccupied the •ONTHS former territory of the cpwf pair. Movement Fi•. •. Su•ivors•i D cu•es for Common from one territory to another by a breederwas S•ri•es (• = 21) nnd Grny-bnc•ed•iscn• S•r•es recordedonce in the gray-backsand four times 6]) for t•e Deriod June ]979 to December in the common fiscals. Productionof young.--A final comparisonbe- tween speciesrelates to the relative production gray-backs:x 2 = 0.39, P > 0.50; common fiscals: of young. Brown (1974) predicted that cooper- X2 = 0.82, P > 0.50). ative specieswould have lower reproductive Interspecific differences in number of terri- rates than their noncooperative relatives. Our tories lost or vacated during the period from resultsdo not supportthis prediction (Table 4). June 1979 to December 1981 were significant No significant differences were detected when (X2 = 5.76, P < 0.025). Of the gray-back terri- comparing young per pair (here comparing tories recorded in June 1979, 59% (10 of 17) only 2-member gray-back groups with the were occupiedcontinuously (although not nec- commonfiscals), young per group (comparing essarilyby the same individuals) through De- production of young per territory, regardless cember 1981. In contrast,only 24% (4 of 17) of of group size), or young produced per adult the territories of common fiscals were contin- (even though only one pair of gray-backscon- uously occupied(see Fig. 1). tributes directly genetically to the production A few examples for each specieswill illus- of young). trate the dynamicsof demographicflux during The major reason for these similarities in the study. The gray-back territory NYM (Fig. production of young appearsto be the high 1) held 5 birds from June 1979 until sometime incidence of nest predation for both species. between December 1980 and August 1981, Eighty-five percent of the nests initiated for when 1 flock member disappeared.No success- both species failed (Zack MS). In the acacia ful breeding occurredin this flock during the woodland site these two Lanius shrikes were study. In contrast,the gray-backflock SK (Fig. among the very few bird speciesthat nestedin 1) had 9 membersfrom June through Decem- open cups.(Turdoides babblers and the drongo ber 1979. By December 1980, however, consid- Dicrurus adsimilis have nests similar to the erable flux had occurred. Four adults had died shrikes.) Most major groups (weavers, star- or disappeared,and 4 juvenileshad been reared lings, coraciiforms,sunbirds, old world war- and recruited into the flock. By August 1981, 6 biers) nested either in highly cryptic or en- birds had disappearedfrom this group (includ- closednests. Shrike nestsappeared to be among ing the female breeder, which was replacedby the most conspicuous and vulnerable of all a female helper from the adjacentKM flock); 4 speciesnesting on the study site. Given the of these6 had emigratedto nearby flocks.Only similarity in predation rates,it appearsthat the 1 was known to have gained breeding statusin differences in social structure did not affect a its new territory, a female hatched in 1980 that given nest's vulnerability to predation. How- attempted unsuccessfullyto breed in the FBH ever, when considering only those nests that territory. successfullyfledged young, gray-backsfledged October1985] ShrikeCooperative Breeding 761

T^I•LE4. Comparisonof Common Fiscal and Gray-backedFiscal shrikes' production of young. Data are combined from 1980 and 1981.

Cornrnon Gray-backed Comparison Fiscal Shrike Fiscal Shrike Test P Young/pair 1.15 0.78a X2 = 0.75 >0.10 Young/group 1.15 1.39b X2= 1.70 >0.10 Young/adult 0.58 0.35c X2 = 2.83 0.05 < P < 0.10 Young fledged/ 1.88 2.65 Wilcoxon two- <0.05 successfulnest sampletest, z = -1.98 Pairs only. All groups. Includes helpers. more young on average than did common ils- 1970), and western populations of the Scrub cals(gray-backs: n = 17 nests,mode = 3, mean = Jay (e.g. Brown 1974, Atwood 1980). The iso- 2.64, SD = 0.78; common fiscals: n = 8, mode = lated Florida populations of the Scrub Jay ex- 1, mean = 1.87, SD = 0.83; Table 1). This result hibit cooperativebreeding in about 50% of the opposesBrown's (1974) prediction. The differ- groups (Woolfenden 1975, Woolfenden and ence could be due to the presence of helpers Fitzpatrick 1978). Although Brown (1974) out- in the gray-backs,the occupanyof higher-qual- lined several testable predictions for the jays, ity habitat, or a combination of these and other no comparative measures of habitat and de- factors. mography have been published. Our compar-

DISCUSSION ative study of shrikes agrees with Brown's (1974) predictions of increasedsurvival, dimin- Fry (1972) commented that cooperative ished dispersal,and occupationof a stablehab- breeding systemsare a "mixed bag ecological- itat for the cooperativelybreeding relative to ly," and he felt that no commonselective agent the noncooperativelybreeding species.The di- could accountfor them. Dow (1980b) even sug- minished dispersal of the Gray-backed Fiscal gestedthat the observeddiversity of coopera- Shrikescan be readily seenin Fig. 1. All move- tive breeders indicates that this phenomenon ments were by nonbreedersdispersing to at- may be more likely to arise as a consequence tain breeding status. Most movements were of special social rather than strictly ecological only a few territories in distance. In contrast, factors.Reyer (1980) and Emlen (1982) also ob- no young Common Fiscal Shrikes were seen served that attempts to find ecological corre- again on the study site following their pre- lates have been largely unsuccessful.We com- sumed dispersal. The occupation of a stable pared habitat affinities, resource bases, habitat by the cooperatively breeding gray- demography,and territory stability of conge- backs is inferred from the data on insect abun- neric sympatric speciesthat differ in their so- dance in the dry season and the relatively cial systems (cooperative vs. noncooperative higher survivorship. Although our results are breeding) in an attempt to find meaningful in accordwith these predictions, it should be ecological correlates of cooperative breeding emphasizedthat our environmental measures (see Brown 1974). Our results strongly impli- were collected over a period of time much cate habitat stability and saturation in the evo- shorter than the lifetime of the birds. Thus, lution of cooperativebreeding, as has been much of our discussion of the evolution of the suggested often (e.g. Selander 1964; Brown social systemsof these speciesmust be viewed 1974, 1978; Emlen 1978; Woolfenden and Fitz- as tentative. patrick 1978). Habitat quality.--On Morendat Farm, Gray- The genus Aphelocomaof North America is backed Fiscal Shrikes occur only in relatively amenable to a similar analysis (Brown 1974), dense yellow-barked acacia woodland. High e.g. the Gray-breastedJay (A. ultramarina),a co- perennial shrub cover is associatedwith high operatively breeding speciesoccurring in the tree density (Table 1). Areas of high cover have southwestern U.S. and Mexico (Brown 1963, greater availability of insect prey throughout 762 ZACKAND LIGON [Auk, Vol. 102

Genus Lanius a more meaningful comparison is made between similar cooperatively and noncoopera- HabitatGeneralist Habitat Spe. c•i.alist tively breeding speciesin the same environment, as reported here. Habitat saturation.--One of the most important factors influencing the demographic environment is the degree of habitat saturation, becausethis may in effect control the dispersal optionsavailable to membersof the population (Brown 1974, Emlen 1982). If the annual production of young in such a habitat is, on the average,higher than the annual mortality rate of breeders,then the bestoption for nonbreed- ing individuals would be to stay in the natal territory and wait for openings (Ricklefs 1975; Pa•r Only Cooperative Breeding Brown 1978; Emlen 1978, 1982; Gaston 1978b). Fig. 4. Hypotheticalscenario for the evolutionof If parentsallow the young to stayin the breed- the socialsystem differences observed in Common ing territory (Gaston 1978b, Ligon 1981), non- Fiscaland Gray-backedFiscal shrikes. breeders may attain breeding statusby inher- iting the territory or part of it (e.g. Woolfenden and Fitzpatrick 1978).Alternatively, nonbreed- the annual cycle,but significantlyso only dur- ers may gain allies from within the natal ter- ing the dry months,when insectsare least com- ritory to competeas a group againstother non- mon (Table 3, Fig. 4). On Morendat Farm, the breeders for territorial vacancies(Ligon and habitat occupiedby gray-backscontrasts with Ligon 1978, 1983;Koenig 1981b). the more open areasused by common fiscals. These complex social strategiesappear to be The latter hassignificantly fewer availableprey the result of ecologicalrestrictions placed on in the dry monthsof the year than do the areas those populations occupying saturated envi- in which gray-backsoccur. The presumed ronments. This saturation is the result of hab- higher-quality,but spatiallyrestricted, habitat itat specificityand of a very low ratio of mar- occupiedby the gray-backsmay be the critical ginal to optimal habitat (Koenig and Pitelka factormaintaining cooperative breeding in this 1981). Habitat saturationis necessarilya rela- species.Strong seasonalityin rainfall and as- tive term. It is not clear from previous works sociatedinsect availability also have been re- whether "saturated" habitat refers to all suit- portedin studiesof othercooperative breeders able habitat, all suitable breeding habitat, or (e.g. Gaston1978a, Vehrencamp 1978, Emlen only thoseareas considered to be of high qual- 1982). ity. Most territories of Gray-backed Fiscal Becausethe vegetation characteristicof gray- Shrikes were occupied continuously through- back territoriesholds significantlymore insects out the study.These continuously occupied ter- during the dry months, Gray-backedFiscal ritoriestended to have high shrubcover. Those Shrikesmay benefit from their associationwith territories that were occupied intermittently high-coverareas prim, arily during the dry contained less cover and were inferior to months. However, since there was no increase higher-cover territories in terms of survivor- in the disappearancerate for either shrike ship and fecundity of the occupants(Zack and speciesduring the dry months, there was ap- Ligon 1985). parently no direct connectionbetween the re- In contrast,space for immature Common Fis- duction of the prey baseand the disappearance cal Shrikes attempting to enter the breeding of shrikes. population appearedto be frequently available A relationship between high survivorship as a result of the high turnover of territorial and cooperativebreeding has been suggested adults.Upon independence,common fiscals left frequently (e.g. Selander 1964; Brown 1974, the parental territory and were not seen again 1978; Emlen 1978; Gaston 1978b). However, as on the study site. On several occasions,we ob- survivorshipin tropical birds is typically high served immatures of unknown origin on the comparedto temperatespecies (e.g. Fry 1980), study area. October1985] ShrikeCooperative Breeding 763

For juvenile gray-backs,the ecologicaland survivorship. The larger and more numerous demographicenvironment posesa much more gray-backscan dominate common fiscalsand restricteddispersal option. Gray-backsmay re- hence can exclude the latter from acacia wood- main indefinitely in the natal territory. Several land. individuals fledged in June 1979 were still in Given the low turnover of gray-backs,juve- their natal territory in January 1982. The few niles can be retained indefinitely in their par- interflock movementsof gray-backswere as- ent's territory, as the option to disperse and sociatedwith specificopenings for breeding search for a territorial vacancy is unlikely to positionsor were related to the disruptionof a succeed.A proximate concern of parents that flockdue to mortality of one or more key mem- allow juveniles to stay in the natal territory is bers (Zack and Ligon 1985). the ability of the territory to provide resources Evolutionof Lanius socialstructure.--The evo- for all members of the developing flock. Pre- lutionary backgroundsleading to the habitat sumably, intraflock competition for food re- specificityof Gray-backedFiscal Shrikes and the sourcesis lesscritical to the breeding pair and habitatgenerality of CommonFiscal Shrikes are the supernumerariesthan interflock competi- not known. Part of the differences in habitat tion for space.This presumption follows from affinitiesmay lie in the spatialdistribution and the intensity of inter flock territorial interac- quality of the preferred habitat. A basicbehav- tions and the absenceof data suggestingthat ioral difference between common fiscals and flock members are expelled unless there is a gray-backsis the dispersalof young upon ma- replacementof a breeder. As flocksdevelop in turity in the formerand the retentionof young such a setting, selectionshould favor the evo- in the natal territory in the latter. The cost- lution of group cohesivenessand group coop- benefit relationshipsof dispersalmay be fun- eration (e.g. defending the territory). damentally different in the two species.The Once group-living has developed, individu- expectedbenefits of dispersingupon maturity als are subject to social as well as ecological to common fiscal young are much higher limitations. Sociallimitations are those placed (breeding status)than for young gray-backsin on individuals (e.g. breeding opportunity) by the more saturated and more limited habitat of the local social environment rather than by the latter. ecologicalfactors per se. Social limitations can The Common Fiscal Shrike, which exhibits be a function of an individual's dominance rank the "pair only" pattern typical of laniids, hasa within a flock, which is often related to its age wide rangeof habitatsavailable. Edge and open and sex. Social limitation also may be related habitatsare utilized throughout its range, and to an individual's ability to form alliances to there appearsto be little in the way of inherent compete successfullywith other groups for habitat restriction. With this apparently wide breeding positions (e.g. Koenig 1981b, Ligon availability of habitat and the high (relative to 1983). gray-backs) rate of disappearance,available In such a setting, ecologicaland social limi- habitat does not become saturated. Territorial tations may covary, as ecological limitation can "instability" is simply a function of the disap- determine much of the form of social structure pearancerate of its inhabitants.Adults readily in a population (e.g. the amount of available move between territories upon loss of a mate habitat,available nesting structures). The social (Fig. 1), and immaturesdisperse upon indepen- environment, however, can change independence and are unlikely to interact with their dently of the ecologicalenvironment, as in the parents again. death of one or more key flock members.This The important ecologicaland demographic dynamic interplay between ecologicaland de- differencesbetween the two speciesare ar- mographiclimitations presumably has affected ranged into two hypotheticalscenarios that lead the diversity of cooperativebreeding systems to differencesin their socialstructure (Fig. 4). observed(see also Ligon 1983). The pathway of gray-backsis determined by Our results suggest that habitat restriction their affinity for acaciawoodland. Although re- and habitat saturation are critical factors in the stricted, this habitat appearsto be one of rela- maintenance of cooperativebreeding in Lanius tively high quality comparedto the habitatused excubitorius.Ecological restrictions appear to by most common fiscals.One presumedeffect form the underlying basis of cooperative of higher-quality habitat is higher individual breeding in many bird species(Brown 1974; 764 ZACKAND LIGON [Auk, Vol. 102

Emlen 1982, 1984). The great diversity in avian measuresin phytosociologicalsampling. Ecolo- cooperative breeding systemsreflects adapta- gy 37: 451-460. tions both to ecological restrictions and to CRAIGßJ. L. 1979. Habitat variation in the social or- widely differing demographicenvironments. ganization of a common gallinule, the Pukeko, Porphyrioporphyrio melanotis. Behav. Ecol. Socio- biol. 5: 331-358. ACKNOWLEDGMENTS Dow, D. D. 1980a. Communally breeding Austra- lian birds with an analysisof distributional and We thank T. L. GeorgeßP. B. StaceyßL. W. Oring, environmental factors. Emu 80: 121-140. W. D. Koenig, C. H. Fry, J. A. Wiens, and an anony- ß 1980b. Systemsand strategiesof communal mous reviewer for constructive criticisms of earlier breeding in Australian birds. Proc. 17th Intern. draftsof this paper. S. H. Ligon and D. Schmittably Ornithol. Congr.: 875-881. assistedus in the field. R. E. Leakey and G. R. Cun- EMLEN,S. T. 1978. The evolution of cooperative ningham-vanSomeren and their staffat the National breeding in birds. Pp. 245-281 in Behavioural Museumsof Kenyamade our researchpossible. E. K. ecology(J. R. Krebs and N. B. Davies, Eds.). Ox- Ruchiami, of the Office of the President of Kenyaß ford, Blackwell Sci. Publ. kindly granted researchpermits. Dr. H.-U. Reyer ß 1982. The evolution of helping. I. An eco- kindly loanedus color bandsand snap-trapsearly in logical constraintsmodel. Amer. Natur. 119: 29- the study.The R. Terry family,the W. Hillyar familyß 39. and LakamiaBulinda provided friendship and logis- 1984. Cooperative breeding in birds and ticalsupport. This project was supported by NSF Grant mammals. Pp. 305-339 in Behavioural ecology, DEB-7905866to J. D. and S. H. Ligon. We thank all 2nd ed. (J. R. Krebs and N. B. Davies, Eds.). Ox- these individuals and institutions. fordß Blackwell Sci. Publ. FRY,C. H. 1972. The social organization of bee-eat- LITERATURE CITED ers (Meropidae) and co-operativebreeding in hot- climate birds. Ibis 114: 1-14. ATWOOD,J.L. 1980. Social interactions in the Santa 1980. Survival and longevity among tropi- Cruz Island ScrubJay. Condor 82: 440-448. cal land birds. Proc. 4th Pan-African Ornithol. BATCHILLER,C. L., & D. J. BILLß 1971. Experiments Congr. 4: 333-343. in estimating density from joint point- and near- GASTON,A. J. 1978a. Demography of the Jungle est-neighbourdistance samples. Proc. New Zea- BabblerßTurdoides striatus. J. Anim. Ecol. 47: 845- land EcoL Soc. 17: 111-117. 870. BONFERRONI,C. E. 1936. Publ. 1st Sup. Sci. Icon. 1978b. Factors affecting the evolution of Commun. Firenze 8: 1-62. group territorial behavior and co-operative BRITTONßP. L. (Ed.). 1980. Birds of EastAfrica, their breeding in birds. Amer. Natur. 112: 1091-1100. habitatßstatus and distribution. Nairobi, East Af- GRIMESßL. G. 1979. Sexual dimorphism in the Yel- rican Natural History Soc. low-billed Shrike Corvinella corvina and in the BROWNßJ. L. 1963. Socialorganization and behavior other African shrikes(subfamily Laniinae). Bull. of the Mexican Jay. Condor 65: 126-153. Brit. Ornithol. Club 99: 33-36. 1970. Cooperative breeding and altruistic 1980. Observationsof group behaviour and behavior in the Mexican Jay, Aphelocomaultra- breeding biology of the Yellow-billed Shrike marina. Anim. Behav. 18: 366-378. Corvinella corvina. Ibis 122: 166-192. 1974. Alternate routes to sociality in jays HALLßB. P., & R. E. MOREAU. 1970. An atlas of spe- with a theory for the evolution of altruism and ciation in African passerinebirds. London, Brit. communalbreeding. Amer. Zool. 14: 63-80. Mus. (Nat. Hist.). 1978. Avian communal breeding systems. HOPKINS,B., •r J. G. SKILLAM. 1954. A new method Ann. Rev. Ecol. Syst. 9: 123-155. for determining the type of distribution of plant --, & R. P. BALDA. 1977. The relationship of individuals. Ann. Bot. London N.S. 18: 213-227. habitat quality to group size in Hall's Babbler KOENIG,W. D. 1981a. Reproductive success,group (Pomatostomushalli). Condor 79: 312-320. sizeßand the evolution of cooperative breeding --, D. D. Dow, E. R. BROWN, 8r S. D. BROWN. in the Acorn Woodpecker.Amer. Natur. 117:421- 1983. Socio-ecologyof the Grey-crownedBab- 443. bler: populationstructure, unit size and vegeta- 1981b. Space competition in the Acorn tion correlates. Behav. Ecol. Sociobiol. 13: 115- Woodpecker: power struggles in a cooperative 124. breeder. Anim. Behav. 29: 396-409. BROWNLEE,K.A. 1965. Statisticaltheory and meth- ß& F. A. PITELKA.1981. Ecologicalfactors and odology,2nd ed. New York,John Wiley andSons. kin selection in the evolution of cooperative COTTAM,G., & J. T. CURTIS.1956. The use of distance breeding in birds. Pp. 261-280 in Natural selec- October1985] ShrikeCooperative Breeding 765

tion and social behavior (D. Tinkle and R. Alex- RICKLEFS,R. E. 1975ß The evolution of co-operative ander, Eds.). New York, Chiron Press. breeding in birds. Ibis 117: 531-534. LIGON,J. D. 1981. Demographicpatterns and com- SELANDER,R. K. 1964. Speciation in wrens of the munal breeding in the Green Woodhoopoe, genus Campylorhynchus.Univ. California Publ. Phoeniculuspurpureus. Pp. 231-243 in Natural se- Zool. 74: 1-305. lection and social behavior (D. Tinkle and R. SOK^L,R. R., & F. J. ROHLF. 1981. Biometry, 2nd ed. Alexander, Eds.). New York, Chiron Press. San Francisco, W. H. Freeman. ß 1983. Cooperationand reciprocityin avian STACEY,19. B. 1979. Habitat saturation and commu- social systems.Amer. Natur. 121: 366-384. nal breeding in the Acorn Woodpecker.Anim. , & S. H. LIGON. 1978. The communal social Behav. 27: 1153-1166. systemof the GreenWoodhoopoe in Kenya.Liv- TRAIL,P. W. 1980. Ecologicalcorrelates of socialor- ing Bird 17: 159-197. ganization in a communally breeding bird, the , & --. 1983. Reciprocityin the Green Acorn Woodpecker Melanerpescalifornicus. Be- Woodhoopoe(Phoeniculus purpureus). Anim. Be- hav. Ecol. Sociobiol. 7: 83-92. hav. 31: 480-489. VEHRENCAMP,S. L. 1978. The adaptive significance MACKWORTH-PRAED,C. W., &: C. H. B. GRANT. 1960. of communalnesting in Groove-billedAnis (Cro- Birds of eastern and northeastern Africa, vol. 2. tophagasulcirostris). Behav. Ecol. Sociobiol.4: 1- New York, Longman, Inc. 33. MANTEL, N. 1966. Evaluation of survival data and WOOLFENDEN,G.E. 1975. Florida ScrubJay helpers two new rank order statisticsarising in its con- at the nest. Auk 92: 1-15. sideration.Cancer Chemotherapy Repts. 50: 163- , 8r J. W. FITZPATRICK.1978. The inheritance 170. of territory in group-breedingbirds. BioScience REYER,H.-U. 1980. Flexible helper structureas an 28: 104-108. ecological adaptation in the Pied Kingfisher ZACK,S., 8rJ. D. LIGON. 1985. Cooperativebreeding (Ceryle rudis rudis L.). Behar. Ecol. Sociobiol. 6: in Laniusshrikes. II. Maintenance of group-liv- 219-227. ing in a nonsaturated habitat. Auk 102: 766-773.