BehavioralEcology Behav. Ecol. Sociobiol. 6, 219-227 (1980) and Sociobiology ?) by Springer-Verlag 1980

FlexibleHelper Structure as an EcologicalAdaptation in the Pied (Ceryle rudis rudis L.) Heinz-Ulrich Reyer

Max-Planck-Institut fur Verhaltensphysiologie, D-8131 Seewiesen, Federal Republic of Germany

Received April 5, 1979 / Accepted October 9, 1979

Summary.The pied kingfisherhas two types of hel- satisfactorycorrelations between social structureand pers at the nest: primary ones, helping their own environment- a prerequisitefor understandingthe parents,and secondaryones, helpingbirds other than adaptivesignificance and evolution of behavior- are parents. Primary helpers are always accepted by lacking. Most investigatorsof cooperative breeding breedingpairs, secondaryhelpers only in poor envi- did not pay enough attention to ecological parame- ronmental conditions where the time and energy ters. With a few exceptions(Brown and Balda, 1977; budget of parents is not sufficient for rearing the Vehrencamp,1978), habitat qualities have eitherbeen offspring alone. Under these conditions, helpers in- ignoredtotally, or the categoriesemployed in ecolog- crease the breeding success of pairs (Tables2 and ical investigationshave been relativelycrude (Gaston 3) by providingadditional food for the young (Ta- and Perrins,1975). We cannot expect to discoverany ble 4). Thus the flexible helper structurecan be seen satisfactorycorrelation between social structureand as an ecologicaladaptation. It is arguedthat - origi- environment,however, as long as we continueto clas- nating from a skewed sex ratio and breedingin co- sify the distributionof animalsand their food merely lonies- this adaptationevolved through the combined as either stable or unpredictable;habitats merely as effects of individualand kin selection. desert, savannah, bushland, forest, or marsh; and food no more preciselythan as grain, fruit, nectar, or (for summariessee Emlen, 1978;Fry, 1977; Grimes, 1976; Woolfenden,1976). Introduction The presentand a followingpaper will show that finermeasurement of ecologicalparameters can reveal Helpersat the nest, as definedby Skutch(1961), have relationshipsbetween helper structureand environ- been reported to date in over 150 species ment, even for different populations of the same (Grimes,1976; Rowley, 1976; Woolfenden,1976; Za- species.These relationships,which are not biased by havi, 1976).The few featuresshared by most coopera- species-specificdifferences, can be interpreted in tive breedersare easily outnumberedby the diver- terms of ecological adaptation. They also can help sities. Consequentlythe adaptivesignificance of such to assess the importanceof individualand kin selec- cooperative (communal) breeding is poorly under- tion in the evolution of cooperativebreeding. stood and the drivingforces in the evolution of help- ing are controversial.Some scientistsfavor individual selection, others favor kin selection (Brown, 1974, Materials and Methods 1978; M. Dyer and C.H. Fry, unpublishedwork; rudis rudis 1978, 1980; Fry, 1977; J.D. Ligon, unpub- Two breeding populations of the (Ceryle Emlen, L.) in Kenya (East Africa) were compared, one (20-30 pairs) at lished work; Ligon and Ligon, 1978; Ricklefs, 1975; the Winam Gulf of (34?18' E, 0O21'S), the other Woolfenden, 1975, 1978; Woolfenden and Fitzpa- (15-20 pairs) at the southern end of Lake Naivasha (36?42' E, trick, 1978). 021' S). Most of the in these two colonies had been mistnet- This controversyseems to arise partlyfrom exist- ted and were individually color-ringed or wing-tagged. Some casual observations in a colony of over 50 pairs at Lake Simbi (34?38' E, ing species-specificdifferences, and partly from the 0O21'S) were also recorded. These investigations are part of our lack of data regardingthe effects of helping on inclu- continuing studies, commenced in 1975, on the ecology and the sive fitness,both for breedersand helpers.Moreover, behavior of various East African kingfisher species (Alcedinidae).

0340-5443/80/0006/02 19/$01.80 220 H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation

Results Hand-reared yearling pied bred success- fully in our aviaries, their first clutch containing five General Remarks on the Biology eggs from which four young hatched. These figures of the Pied Kingfisher do not differ from those which we found in older breeders. Further information on the biology of the Within their East African range, pied kingfishers in- pied kingfisher is given by Douthwaite (1970, 1973, habit many lakes and rivers and are particularly fre- 1976, 1978), Sugg (1974), and Migongo (1978). quent in the marginal regions of the big freshwater lakes. They live almost exclusively on fish. To catch their prey, they fly over the water searching, some- Comparisonof Two Colonies times hovering above the surface, and plunge swiftly when they see a fish. They swallow small fish while Sex Ratio and Population Structure in the air, and continue searching. Bigger fish are carried to perches on the shore and battered before In both study areas there was a significant surplus being swallowed. Direct dives from such perches were of &d. At Lake Victoria, SS outnumbered Y? by rarely observed in the areas investigated. 1.8:1 (107,3S, 61 ?Y; P<0.01, x2 test), at Lake Nai- Outside the breeding season, pied kingfishers can vashaby 1.7:1 (33 SS, 19 YY;P<0.05). These figures be seen singly, in pairs, or in small groups along are based on counting both ringed and unringed birds. the whole shore area. In the evening, they congregate For the same L. Victoria colony, Sugg (1974) obtained at communal roosting sites. During this time of the a figure of 2: 1, both in 1971 and 1972. Similar sex year, considerable local movements may be under- ratios ranging from 1.5 to 2.5 have been reported taken; these can carry individuals over hundreds over by Douthwaite (1973, 1978) for breeding and non- kilometers (Backhurst, 1974). In the breeding season, breeding populations in Uganda and Zambia. Only the birds usually concentrate at rivers, canals, road one of his counts, made at a roost, resulted in an embankments, and other places having sandy or clay even sex ratio. All these data taken together suggest banks, not too far from the lake. Here 20, 50, 100 that in populations of adult pied kingfishers, there or more birds excavate nesting holes. In this way, are 1.8 as many SS as Yy (1074 S, 610 yy; breeding colonies are formed. The average distances P<0.001). Among the young, however, an even sex between adjacent nesting holes were 5.2 m at L. Victo- ratio seems to exist. Out of 40 nestlings and fledglings ria and 1.6 m at L. Naivasha. Both are well above which were sexed either by laparatomy of by hand- the possible minimum (0.5 m) found in the pied king- rearing them to maturity, 21 were 3S, 19 ii (1. 1:1). fisher and in other colonial breeders such as bee One reason for the later-occurring skew toward eaters. Therefore - and because suitable, unused nest- a surplus of S3 probably lies in a higher mortality ing places were in the vicinity of both colonies - rate of breeding Yy. Incubating and brooding at night the carrying capacity of the habitat was probably fall to the Y exclusively, and even by day the Y per- not exhausted in respect to nesting sites. It was not forms the greater part of these activities. Conse- exhausted in respect to food either. All members of quently it is more endangered if nesting holes cave the colony hunted in the same area over the lake in or are flooded, as well as by nest predators such and did not defend individual fishing grounds. Only as the monitor lizard ( Varanusniloticus), cobras (Naja in the vicinity of the nesting holes did SS and ? spec.), and the ichneumon (Herpestes spec.) (Douth- behave territorially. waite, 1970, 1978). All four breeders which disap- The sexes can be distinguished by plumage differ- peared during our study were Yy, and though this ences: ?? have only one broad, interrupted chest- toll on breeding Yy was low in our colonies (6%), band; SS have in addition a second narrower, unin- it may be much higher in other colonies where preda- terrupted band. This is more pronounced and even tion rates of more than 40% have been found (Douth- in older SS than in yearling &S. The SS and ?W waite, 1978). take turns incubating the eggs from which the blind, A second reason for the skeward sex ratio may naked young emerge after 18 days. From the first be a higher mortality rate of juvenile YY. In contrast day, they are fed with fish brought from the lake to yearling SS, yearling Yy do not apparently return mainly by the X, and possible helpers; later, when to their natal colonies (Table 1). A stronger dispersal brooding declines, the ? joins increasingly in feeding of Yy has also been observed in other bird species. the young. Nestlings are fully fledged after about It has been suggested that on their routes into distant, 26 days and can fish for themselves roughly 14 days unknown areas, Yy might be endangered more than later, but stay with their parents for several months. &S remaining in familiar areas (e.g., Zahavi, 1974; Sexual maturity can be reached within the first year. Woolfenden and Fitzpatrick, 1978). This phenome- H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation 221

Table 1. Recaptures and resightings of juvenile and adult pied king- Table2. Number of pairs with and without helpers in two pied fishers, ringed in the same colony at L. Victoria the previous year. kingfishercolonies. The differenceis significant(P<0.005) when Data from 1976 to 1979 pooled. Juvenile birds were nestlings when in both colonies only the categoriespairs withouthelpers and pairs ringed, adults were ringed as > yearlings. The data ca. 32 O'S and withhelpers are considered (x2 test). No differenceis madebetween ca. 31 i? are based on the assumption of a sex ratio approximately primaryand secondaryhelpers equal to 1 (see. p.220) No. of breeding L. Victoria L. Naivasha No. marked No. recaptured % pairswith + resighted No helpers 14 13 Juveniles SS ca. 32 10 31 1 Helper 19 5 yy ca. 31 0 0 2 Helpers 7 0 Total 63 10 16 3 Helpers 1 0 >3 Helpers 0 0 Adults Ss 72 25 35 WV 48 12 25 Total 120 37 31 Total 41 18 non may preventinbreeding depressions (Greenwood 3) Despite poor reproductivesuccess, the colonies et al., 1978). on the Entebbe peninsula (Uganda) have increased Recaptureand resightingdata of ringed birds at in the last years, which suggeststhat many breeders L. Victoria showed that only 25% of adult Y? and are immigrants from more successful colonies 35% of adult SS returnin the following year to the (Douthwaite,1978). same breedingcolony (Table 1). At L. Naivasha,the figurewas 35% for each sex Helper Structure (six out of 17). The resultsin the two colonies, how- ever, cannot be comparedwith each other since the Althoughhelpers at the nest were found in both co- Naivashafigure might be biasedtwofold: (a) the sam- lonies, their frequencieswere different,both per pair ple size is small, and (b) thoughthe birdscongregated and per colony. At L. Naivasha, no pair had more in the colony the year afterringing them, no breeding than one helper and only 28% of all pairs were activity was possible since the risen water level had assisted when feeding nestlings. At L. Victoria, as swamped the nesting banks. Nevertheless,the rate many as three helpers per pair did occur and 66% of return seems to be low for both colonies, and of all pairs were assisted (Table2). In both colonies, seems to be fairly consistentfrom year to year. This and in others, all helperswere male (n=60). follows for L. Victoriafrom comparingour own data, However, the numerical difference between the collected from 1977 to 1979, with those of Sugg two colonies did not exist at the beginning of the (1974), who worked in the same colony in 1971 and breedingseason. Duringnest-digging, courtship, egg- 1972. He obtained return rates of 20% for ?y and laying, and incubating,67% of all L. Victoria pairs 30%for 3%.With his and our data pooled, the differ- had no helperseither (n=52). This is similar to the ence betweenthe sexesjust reachessignificance (P= L. Naivashafigure. 0.05; X2 test). This difference seems to be at least partly a consequenceof the different mortalitiesin Primary Helpers. The other pairs which had helpers breedingWV and &S (p. 220). The low returnrates in duringthese initial periods had alreadyarrived with both sexes may result from a generallyhigh rate of them in the colony. Such helpers which accompany mortality, also found in other kingfisher species pairs from the very beginningof the breedingseason (McClure,1974). It also may result from movements are termed'primary helpers' in this paper. between the various colonies. We checked 123pied During the excavating,courtship, and egg-laying kingfishersin three neighboringcolonies which were periods,primary helpers feed the d of the pair, sup- 2, 4, and 9.5 km away from the L. Victoria study port it in feedingthe V (both by feedingthe V directly area. Though not a single ringed bird was found, and via the d), and help the pair ward off rivals thereare indicationsfor dispersalof adult pied kingfi- from the nesting site. After the eggs are laid, helpers shers: participatein driving off nest predators, and later 1) Outsidethe breedingseason, considerablelocal in feedingthe young. Tunneling,copulating, incubat- movementsare frequent. ing of eggs, and broodingof young were not observed 2) In 1978 and 1979, unknown pairs appeared in primaryhelpers. in the L. Victoriacolony, and known pairs left with- To date, 17 primaryhelpers have been recorded out breeding. at L. Victoria, and five at L. Naivasha. For ten of 222 H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation the 17 L. Victoria birds, the descent is known from lier. This is demonstratedby the following examples ringing. They all were yearling sons of at least one from L. Victoria. of the breederswhich they now accompanied.Three 1) A year-oldS attachedhimself to a neighboring furtherunringed &S were identified as yearlingsby pair feeding young, after his father and his father's their fainter chest patterning; two of them helped new V had left the colony, abandoning an un- pairs each with one mate having bred successfully completednesting hole. the year before. The third S helped at a nesting site 2) Three pairs discontinuedtunneling, two after where young had fledged in the previous year. As coming up against obstacles, one after losing part those breedingpairs which returnto the same colony of theirnest in a landslip.Each S acted as a secondary also return to precisely the same nesting site, this helper at a neighbor'snesting hole. The ?? stayed last bird, too, probablywas helpinghis own parents. with their mates, but did not join in helping. In the remainingfour cases, therewere no indications 3) A year-old &flew back and forth through the either for or against a relationship between pairs colony with his father, after his mother had fallen and their helpers. victim to a nest predator.Both SS were seen with Only observationsfrom 1977 are availablefor the fish in their beaks outside various holes; at one of colony at L. Naivasha, as the water level had risen these, the son was eventuallytolerated by the pair, in 1978 and 1979 and swamped the nesting banks. while his fatherwas chased away. Thereforenothing is known about the relatednessbe- 4) SeveralSS which had obviouslyarrived in the tween the five primaryhelpers and the breeders.But colony singly,flew throughit repeatedlycarrying fish. according to their faint chest bands, all five were They tried to feed the young at various holes. At yearling d&. least five of these SS functionedlater as helpers at From all the evidence, we may assume that pri- one hole or another. mary helpersare the sons of least one pair member. To date, 19 secondaryhelpers have been recorded The two colonies do not seem to differ with respect at L. Victoria.The status for 7 of them is unknown. to this helper type. The relativefrequencies of pairs The other 12 were certainly not the yearling sons with primaryhelpers were similarfor both colonies: of the pairs which they joined, and the rings gave 33% for L. Victoria and 28% for L. Naivasha. And no evidence for any other close genetic relatedness with one exceptionin which a pair had two primary betweenthem and the breeders.Whether this means helpers at L. Victoria,the numberper pair was one that theyhelped unrelated birds will be discussedlater in both areas. (p. 226). Prolonged chasing away of potential secondary SecondaryHelpers. The differencebetween the two helpers, which was the rule at L. Naivasha, was colonies arose 3-4 days after the young had hatched. observedat L. Victoriain only two cases. Both cases Thena secondhelper type appeared,apparently being involvedbreeding pairs which had only two nestlings attracted by the nestlings' intensive begging cries. from the very beginning. Normally, 4-5 young These 'secondary helpers' flew through the colony, hatchedin each colony. fish in beak, landed in the vicinityof various nesting These two pairs had not acceptedsecondary hel- holes, and waited.Later they approachedthe entrance pers until their young were 12 days old, which was repeatedlyand also triedto feed the adult V belonging unusual at L. Victoria. Then two young of similar to the respectivehole. In both colonies, they were age were added to each clutch with the result that initially repelled, particularlyby the d of the pair, both pairshad one secondaryhelper each withinthree But if persistentat L. Victoria,a helperwas accepted days.This suggeststhat the parents'decision to accept after2-4 days and was allowed to remainat the nest, secondaryhelpers depends on the number of young wherehe participatedin feedingthe young and ward- they have to rear and can rear. The results further ing off nest predators. suggest that pairs at L. Victoria need more helpers At L. Naivasha, however, expulsion continued than theirL. Naivashacounterparts to achievea simi- throughoutthe nesting period.None of the nine sur- lar breedingsuccess. plus d& whichshowed feeding tendencies as accepted. Thus, in contrast to L. Victoria, secondaryhelpers were not observed at L. Naivasha, This difference BreedingSuccess and Feeding Rates accountsfor the differentfrequencies of assisted and unassistedpairs in the two colonies (Table2). This last hypothesisis strengthenedby a comparison Secondaryhelpers can be recruitedfrom single of the breedingsuccess in the two colonies (Table3). surplus d& as well as from &S of those pairs and Averageclutch size and hatchingrates were the same groupswhose own breedingattempts have failed ear- in the two colonies, and also within the colonies for H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation 223

Table 3. Clutch size, number of hatched, and number of fledged young for pairs with and without helpers in two breeding colonies. 25- x = mean, SD = standard deviation, n = sample size. Values per bird are given in brackets, values per nest without brackets. For signifi- cance of differences, see the text 20 L. Victoria L. Naivasha

x SD n x SD n 15-

Clutch size 4.9 0.6 22 5.0 0.6 8 c-a Young hatched 4.6 0.5 14 4.5 0.7 2 10 Young fledged No helpers 1.8 0.6 14 3.7 0.9 9 (0.9) (1.9) 5- 1 Helper 3.6 0.5 12 4.3 0.5 4 (1.2) (1.4) 2 Helpers 4.7 1.0 6 - - 0 (1.2) A B C D E 61 55 45 41 40 Fig. 1. Number of fish eaten by five nestlings of different weight. The letters below the bars indicate the individuals, and the numbers show the respective weight at the beginning of the experiment. For further details, see the text pairs with and without helpers. Nevertheless, the 27 and 93 g, and 12 comparableyoung from nests breedingsuccess of individualpairs, expressedin the with one helper weighed between 53 and 88 g. The number of their fledging young, varied clearly. At average weights of the heaviest birds in these two L. Victoria, only 39% of hatched birds survivedto groups did not differ significantly(xc-=74 and 76 g), fledging when cared for exclusively by parents, as but the average weight of the smallest young was opposed to 78% when fed additionallyby one helper, lower in nests without helpers (- =36 g, n = 7) than and 100% when two helpers assisted. The improve- in nests with one helper (XH=57g, n=6; P<0.01, ment with each additionalhelper is significantin all U-test, two-tailed). The smallest nestlings usually cases (P < 0.001, P < 0.01, and P < 0.025; Mann-Whit- died, probably because their stronger siblings ney U-test, one-tailed). Even on a per-adult-bird snatched most of the food. This was confirmed by basis, groupswith one and two helpersare more suc- the following experiment. cessful than pairs without helpers (P < 0.01 and At the age of about ten days, five nestlings of P<0.05). But on this basis, there is no difference different weights were placed in an artificial nest- betweenthe success of pairs with one and those with chamber.The chamberhad only one circularopening two helpers. of 5 cm in diameter. The birds were individually At L. Naivasha, pairs without helpers succeeded color-markedat the tips of their bills. On three con- in raising as many as 80% of their hatched young, secutivedays, they were given less fish than they need which is double the numberraised by their counter- at this age and the fish were only fed to that nestling parts at L. Victoria (P<0.001) and almost the same which put out his head through the opening. The number as in pairs with one helper at L. Victoria. experimentclearly showed that heavierbirds pushed Probablydue to the small sample, no significantim- the lighter ones aside and receivedmost of the food provement of breeding success was recorded at L. (Fig. 1). Naivashaif therewas one helper.On a per-birdbasis, At L. Naivasha, the offspring from nests with assistedpairs were even less successfulthan unassisted helpersand withouthelpers weighed the sameat equi- ones (P < 0.05). Probablybecause of the same small valent stages of growth. These results suggest that sample,no differencewas found between the success at L. Victoriahelpers are necessaryto preventa food of pairs with one helper at L. Naivasha and those deficit in the young, whereas at L. Naivasha they at L. Victoria(P<0.2). are not. Indeed,at L. Victoria,from about the fourth Regularcounts and weight measurementsof nes- to the tenth day after hatching, pairs with helpers tlings have shown that at L. Victoriamany offspring provide more fish per offspringthan those without of pairswithout helpers starve. Fourteen 9- to 12-day- (P < 0.02; U-test, two-tailed;Table 4), and pairswith- old youngfrom nests without helpers weighed between out helpers do not meet the energy requirementsof 224 H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation

Table 4. Average number of fish/offspring x h fetched by pairs with less time (5.9 min, n= 26; P<0.0002, U-test, two- and without helpers to nests with 3-6 young which were 4-12 days tailed). old. All observations were made for a 5-h period from 0730 to 3) At L. Victoria, the distance from the colony 1230 h. Each mean x~is based on observation of at least 41 fish and 15 h at three nests, at most 142 and 55 h at 11 nests. SD = stan- to its fishinggrounds is about 500-700 m, about dou- dard deviation, n = sample size. For significance of differences, ble the L. Naivasha distance. see the text 4) Contraryto L. Naivasha,the L. Victoriacolony is in an area thickly populated by humans. Breeding pairs Throughoutthe day, people come down to the river Without helpers With helpers for various activities. Adult birds, arriving with a fish, are often disturbed and have to wait several x SD n x SD n minutesuntil they can feed it to the nestlings. Theseenvironmental differences mean that L. Vic- L. Victoria 0.57 0.13 9 1.03 0.32 11 toria kingfishersmust expendconsiderably more L. Naivasha 0.84 0.15 4 0.86 0.27 3 pied time and energyon fishingand on raisingyoung than their counterpartsat L. Naivasha (H.-U. Reyer, in preparation),but thereare probablyscarcely any dif- theiryoung (H.-U. Reyer,in preparation).At L. Nai- ferencesamong birds within a colony, as members vasha, however, pairs with helpers fetch approxi- all hunt in the same area. mately the same number of fish as those without (Table4) and both groups meet the energy require- ments of their young at all ages. Discussion

Ecological Conditions. These differences in feeding Cooperativebreeding in the pied kingfisheris charac- ratesand reproductivesuccesses can be explainedwith terizedby a flexiblehelper structure, which is adapted the differentecological conditions which exist in the to the particularenvironmental conditions in that two colonies. potential helpers,available because of a general SS 1) At L. Victoria, 56% of the fish fed to the 4- surplus,are recruitedmore in unfavorable(L. Victo- to 12-day-old young were Engraulicypris argenteus ria) than in favorable conditions (L. Naivasha). (Cyprinidae) (n 128). The remaining 44% con- Therebythe breedingsuccess improved especially un- sisted of various Haplochromisspecies (Cichlidae). der poor conditions, making the total number of At L. Naivasha, Tilapia leucosticta, T. zillii (Cich- fledglings/pairsimilar for both the colonies investi- lidae), and Micropterus salmonides (Centrarchidae) gated (L. Victoria, 3.2; L. Naivasha, 3.9). With one were fed to the nestlings.For comparablelength, all exception,in neithercolony were more helpersfound Cichlid fishes and Micropterusare very similar in per pairthan were necessaryfor the maximalpossible their caloric values, but because of its slender body breedingsuccess. The numberof helpersis apparently an Engraulicyprisof the same lengthyields less energy determinedby the breedingpairs which, in unfavor- than the other species (H.-U. Reyer, in preparation). able conditions,accept secondary, probably unrelated To compensate for that, the adults at L. Victoria helpers as well as primary,related helpers, but bar have to feed more fish to their young than do the the secondaryhelpers in favorableconditions. Even adults at L. Naivasha. individual pairs switched their behavior from one 2) The mean wind speed at L. Victoria in the strategyto the otherwhen conditions were experimen- observation months May-Septemberis consistently tally changed(p. 222). Accessibility and quality of higher than at L. Naivasha (Chipeta, 1976). At L. food, the distancebetween the colony and its fishing Victoria, the wind sweeps from across the lake by grounds, as well as the frequencyof disturbanceby day, rougheningthe surface and forming breakers people proved to be the decisive environmentalfac- which disturbthe sand and muddy the waters along tors in the study areas. For the birds, these factors the shore. The pied kingfishershave to fly far out result in more favorableor less favorabletime and to fish and only 24% of their dives are successful energy budgets,which make the breedingsuccess of (n= 107). Catchinga fish takes an average 13.0 min pairswithout helpers, pairs with one helper,and pairs of flying and hovering over the lake (n= 22). At L. with two helpers roughly predictable(H-U. Reyer, Naivasha, the wind comes from the land, and there in preparation). is a beach zone of calm, relatively clear water Occasional observationsat Lake Simbi confirm protectedby a papyrusbelt. Here the pied kingfishers that with decreasing quality of the biotope, the are more successful (79% of 52 dives) than at L. numberof helpersper pair increases.In this colony, Victoria(P < 0.001; Z2 test), and catchinga fish takes the pied kingfishersare exposed to similarconditions H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation 225 as at L. Victoria, but their expenditureof time and (1978, 1980) came to the same conclusion recently energy must be greater still, as the fishing grounds for the white-fronted bee eater (M. bullockoides) are at least 1.5-2.0 km away. The breeders in this which, along with other bee eaters (Fry, 1972, 1977; colony seem to have adapted to the conditions by M. Dyer and C.H. Fry, unpublishedwork), shows recruitingthree or more helpers. many similaritieswith the pied kingfisher. This adaptabilityto the environmentis, of course, As already suggestedby Brown (1974), it seems limited by the sex ratio. At L. Victoria, 34% of the morelikely that the crucialfactors spurring the devel- pairs probablyhad no helpersbecause all the avail- opmentof cooperativebreeding in the pied kingfisher able {g were 'booked up.' If severalpairs have three are the skewed sex ratio and breeding in colonies. or more helpers,as at L. Simbi and in other colonies A ratio of 1.7 or 1.8 SS per V means that only 55%- (Douthwaite,1978), this must eitherbe at the expense 59% of all SS have a chance of mating. The others of other pairs, or else the surplus of && must be either can wait until next breeding season, or help greater still. The latter was the case in a colony in rear the young produced by others. Selection will Uganda (2.3: 1; Douthwaite, 1973). Investigationsto decide for the latter course if the birds' inclusivefit- clarifywhether a relationshipexists betweenthe incle- ness (Hamilton, 1964)is increasedthereby. Opportu- mencyof an environmentand the surplusof SS have nitiesfor suchhelp are numerousin breedingcolonies. been started. The increase in inclusive fitness through helping Fromthe resultsmentioned above, it becomesevi- is obvious for the primary helpers: by improving dent that the numberof helpersper pair in the pied the survivalrate of related birds, a primaryhelper kingfisher is an adaptation to and not a consequence at L. Victoria raises his inclusive fitness by about of differencesin the qualityof the environment.Some 0.9 or 0.45 genetic equivalents when he rears full authors have indeed thought it a consequence,since or half-siblings,respectively. For L. Naivasha, the in some specieswith all-purposeterritories, a positive valuesare 0.3 and 0.15, if the breedingsuccess figures correlationexists between territorysize and quality given in Table 3 are confirmed. These calculations and helpernumber (Brown and Balda, 1977; Gaston are based on the formulae given by Brown (1978) and Perrins, 1975; J.D. Ligon, unpublishedwork; and Emlen(1978). The calculationsassume one helper Parry, 1973; Woolfenden, 1975; Zahavi, 1976). A ne- son per pair in both colonies and the number of gativecorrelation as found in the pied kingfisheralso fledged young given in Table 3. Becauseof this and seems to exist in the colonial white-frontedbee eater other simplifications(see Brown, 1978; Emlen, 1978; (Merops bullockoides) (Emlen, 1980). Thus the differ- J.D. Ligon, unpublishedwork; Woolfenden, 1975), ence perhapsreflects a differencebetween territorial the values should be regarded not as precise, but and colonialspecies. There are numerousother differ- only as rough estimatesin comparisonwith the year- ences betweenpied kingfishersand most cooperative lings' chances of rearing their own young. These breeders.In addition to the defense of all-purpose changesare probablyalmost zero. The high SS sur- territories,most speciesshow many other characteris- plus will affect yearlingSS in particular.They proba- tics of K selection: delayed breeding(or even matu- bly cannot successfullycompete with older, more ex- rity),low fecundity,high survivalrate, sendentariness, perienced SS, and consequently will have little or diminisheddispersal, and almost no migration.Most no chance to win a mate and reproduce.Since this of these features are considered to be adaptations &&surplus seems to be a feature of the species as to the strong competition in stable environments a whole, dispersingto other colonies would not im- whose carryingcapacity in respectto food, territory, prove the reproductiveprospects. Therefore the gen- or nesting sites is reached or nearly reached. The etic equivalentsmentioned should be the highest at- resultis a slow populationturnover and a wide gener- tainable for yearlingprimary helpers in the two co- ation overlap(Brown, 1974, 1975; Emlen, 1978; Fry, lonies. Since this improvementof inclusivefitness re- 1977; Pianka, 1978; Woolfenden, 1976). For pied sults from raising relatives, kin selection (Maynard kingfishers,hardly any of these features apply (see Smith, 1964)must be held responsiblefor the evolu- p. 220, 221), and they show more characteristicsof r tion of this helpertype (see also MaynardSmith and selection than all other known cooperativebreeders Ridpath, 1972; Brown, 1974). among birds. For secondaryhelpers, the benefit of helping is Because of these numerousdifferences, it cannot more difficultto assess.The ringinggave no evidence be expectedthat the evolutionof cooperativebreeding for any close genetic relatednessbetween secondary in the pied kingfishercan be explainedwith the same helpers and breeders.It could be argued, however, 'ecological saturation' model which has been pro- that this relatednesswas not detected because the posedfor K-selectedspecies with all-purpose territories birds had hatched before the study had commenced (Brown, 1974; Fry, 1977; Selander, 1964). Emlen four yearsago. But then, how could one explainthat 226 H-U. Reyer: Flexible Helper Structure as an Ecological Adaptation the secondaryhelpers 'applied' to various pairs, re- confirmedby the observationthat it is mainly the maining where they were tolerated first? Moreover, male mates whichattack candidate helper SS (p.222). the yearlypopulation turnover is about 65% for SJ The secondaryhelpers should be acceptedonly when and 75% for ?Y. Thus the probability of finding a this disadvantageof losing inclusive fitness is more close relativeafter 3-4 years is very low. Yet, at least than compensatedfor by the helpers' contribution in those cases in which secondary helpers joined to reproductivesuccess. Since the effect of helpers neighboringpairs, relatednesscannot be excluded, on reproductivesuccess differs markedlybetween L. particularlywhen we considersite attachment(p.222). Victoriaand L. Naivasha(Table 3), this competition It also could be argued that, due to inbreeding, hypothesiscan explainthe differenttreatments of sec- the average relatednessbetween all membersof the ondaryhelpers in the two colonies. It also can explain colony is fairlyhigh. Again, the high populationturn- why usuallythere were no more helpersper pair than over plus the probably existing immigration from werenecessary for the maximalpossible breeding suc- othercolonies (p. 221) makesthis unlikely.In an effec- cess, and why breedersdo not accept secondaryhel- tive populationsize of 20-30 &{, an annualimmigra- pers if they have only a few young to rear (p.222). tion rate of 65% (making good mortalityand immi- These cases illustratethat there is a breeder-second- grationlosses) would result in an averagerelatedness ary helper conflict which can shift with ecological of less than 0.01 (calculated according to Brown, conditions(Emlen, 1978). 1974). One explanationof the greater tolerance shown From all the evidence,we may assume that most by pairs toward primary,related helpers in both co- secondaryhelpers probably are not closely related loniescould be that relatedhelpers represent less com- to the breedersand their young. Thereforeit seems petition for 9?. In many species, familiarity unlikely that kin selection was the driving force in with other membersof the family and/or stable hier- the evolution of this helper type. Various authors archies either are assumed or have been proved to have suggestedand summarizedthe possibilitiesby be mechanismspreventing such possibly incestuous which helping can improve the helpers' individual matings(Bischof, 1972; Ligon and Ligon, 1978; May- fitness, and some regard individual selection as the nardSmith, 1964;Parry, 1973; Woolfenden and Fitz- mainforce behind the evolutionof cooperativebreed- patrick, 1978; Zahavi, 1976). ing (Brosset, 1978; Brown, 1974, 1978; Emlen, 1978, In this case, the taking over of Yy, one of the 1980; J.D. Ligon, unpublishedwork; Ligon and Li- possible benefitsfrom helpingfor secondaryhelpers, gon, 1]978;Ricklefs, 1975; Selander, 1964; Woolfen- would not apply for primaryhelpers. This does not den and Fitzpatrick,1978; Zahavi, 1974, 1976). mean, however, primaryhelpers gain no individual Accordingto our present,scarce data, secondary advantagesfrom cooperativebreeding. The gain of helpersmay increasetheir own reproductivesuccess experienceand the recruitingof helpers mentioned in two ways: above may hold for them as well. On the other hand, 1) Pied kingfishers which return to the same the findingthat primaryrelated helpers have individ- colony the next year usually are the earliest to feed ual benefitsis not sufficientfor assumingthat individ- young, tend to have more secondary helpers than ual selectionhas been the drivingforce in the evolu- birds breedinglate (12 vs 7), and consequentlyhave tion of cooperativebreeding. When, for socioecolog- more young. Thus a secondary helper may gain ical reasons,birds are deprivedof the chanceto breed, knowledgeof a colony in one year, which improves and whenthey also profitfrom their helping activities, his futurereproductive success via early breeding,and their altruism is certainly in question (J.D. Ligon, recruitingof secondaryhelpers. unpublishedwork), but this is no argumentagainst 2) In 1979, one S bred with a V which he had kin selection. As long as the conditions relatedness 'served' as a secondaryhelper in 1978. This also had and increasedbreeding success are fulfilled, helping yieldedhim a primaryhelper since the two were ac- behaviorcannot help but being kin selected. companiedby one of the Y's yearling sons. The Y's The describedhelper structure in the pied kingfi- previousmate was still alive and mated with a new sher is an exampleshowing that both individualand Y. Thus, secondary helpers may improve their future kinship componentsof natural selection must have reproductivesuccess by taking over Y? and recruiting been at work in the evolutionof cooperativebreeding primary helpers from the young they have helped in this species.Perhaps kin selectionwas more impor- to rear. tant in evolvingprimary helpers, and individualselec- The last observationsuggests that the secondary tion in evolving secondaryhelpers. In order to esti- helpers are at the same time potential competitors mate the relative importance of these two com- for the scarce Yy. Thus they may decrease the breed- ponents, it is not sufficientto know the advantages ing SS's inclusive fitness. This hypothesis is of breedersand helpersin terms of inclusivefitness; H.-U. Reyer: Flexible Helper Structure as an Ecological Adaptation 227 the relative frequenciesof both helper types must Emlen, S.T.: The evolution of co-operative breeding in birds. In: also be known. In the two colonies investigated,the Behavioural ecology. Krebs, J.R., Davies, N.B. (eds.), total numbers of primary, related helpers (22) and pp. 245-281. Oxford, London, Edinburgh: Blackwell 1978 Emlen, S.T.: Altruism, kinship, and reciprocity in the white- secondary,probably unrelated helpers (19) were simi- fronted bee-eater. In: Natural selection and social behaviour. lar. Accordingto preliminaryinvestigations in other Alexander, R.D., Tinkle, D. (eds.). New York: Chiron Press colonies, the L. Victoria system with both types of (in press) 1980 helpersseems to be morerepresentative for the species Fry, C.H.: The social organization of bee-eaters (Meropidae) and as a co-operative breeding in hot-climate' birds. Ibis 114, 1-14 whole than the L. Naivasha systemwith primary (1972) helpersonly. Fry, C.H.: The evolutionary significance of co-operative breeding By comparing the inclusive fitness of primary, in birds. In: Evolutionary ecology. Stonehouse, B., Perrins, probablymore kin-selectedhelpers with that of sec- C.M. (eds.), pp. 127-136. London: Macmillan 1977 ondary,probably more individuallyselected helpers, Gaston, A.J., Perrins, C.: The relationship of habitat to group size in the genus Turdoides. Emu 74 [Suppl.], 309 (1975) I hope it will be possible one day to express the Greenwood, P.J., Harvey, P.H., Perrins, C.M.: Inbreeding and contributionsof individualand kin selectionin terms dispersal in the great tit. Nature 271, 52-54 (1978) of genetic equivalents. Grimes, L.G.: The occurrence of co-operative breeding in African birds. Ostrich 47, 1-15 (1976) Hamilton, W.D.: The genetical theory of social behaviour. I., II. Acknowledgements. I thank all those Kenyans who made this pro- J. Theor. Biol. 7, 1-52 (1964) ject possible, especially E.K. Ruchiami (Office of the President), Ligon, J.D., Ligon, S.H.: Communal breeding in green wood- J. Mburugu (Divisional Game and J. Warden, Nakuru), Hopcraft hoopoes as a case for reciprocity. Nature 276, 296-298 (1978) (Nakuru Wildlife Trust), who also arranged for our quarters. I McClure, H.E.: Migration and survival of the birds of Asia. Bang- also thank the managers of the 'Oserian Farm, Ltd.' and J.O. kok 1974 Ndolo, and numerous Luo families who put up with the 'mad Maynard Smith, J.: Group selection and kin selection. Nature white researcher' on their terrains at Lakes Naivasha and Victoria. 201, 1145-1147 (1964) J.L. Brown, J. Lamprecht, D. Limberger, C. Rechten, M. Ta- Maynard Smith, J., Ridpath, M.G.: Wife sharing in the Tasmanian borsky, F. Trillmich, S. and W. Wickler Vehrencamp, kindly helped native hen, Tribonyx mortierii: A case of kin selection? Am. with constructive criticism of an earlier version of the manuscript, Nat. 106, 447-452 (1972) and P. Rechten helped with the English translation. Migongo, E.W.K.: Environmental factors affecting the distribution thanks to assistant D. for My particular go my Schmidl, his of malachite and pied kingfishers in Lake Nakuru National enthusiastic and also to W. for proficient cooperation; Wickler Park. M.S. thesis, University of Nairobi (1978) his manifold and unbureaucratic to the support; Max-Planck-Ge- Parry, V.: The auxiliary social system and its effect on territory sellschaft for the and not to wife financing work; last, least, my and breeding in kookaburras. Emu 73, 81-100 (1973) H. for her labor and in our work and Reyer patience supporting Pianka, E.R.: Evolutionary ecology. New York: Harper and Row welfare. 1978 Ricklefs, R.E.: The evolution of co-operative breeding in birds. References Ibis 117, 531-534 (1975) Rowley, I.: Co-operative breeding in Australian birds. Proc. XVI Backhurst, G.C.: East African ringing report 1972-1973. J. East Int. Ornithol. Congr. (Canberra, Australia). Frith, H.J., Calaby, Afr. Nat. Hist. Soc. Nat. Mus. 146, 1-9 (1974) J.H. (eds.), pp. 657-666. Canberra: Australian Academy of Bischof, N.: Inzuchtbarrieren in Saugetiersozietaten. Homo 23, Science 1976 330-351 (1972) Selander, R.K.: Speciation in wrens of the genus Campylorhynchus. Brosset, A.: Social organization and nest-building in the forest Univ. Calif. Berkeley Publ. Zool. 74, 1-224 (1964) weaver birds of the genus Malimbus (Ploceinae). Ibis 120, 27-37 Skutch, A.F.: Helpers among birds. Condor 63, 198-226 (1961) (1978) Sugg, M.S.: Mensural and moult data from a breeding colony Brown, J.L.: Alternate routes to sociality in jays - with a theory of pied kingfishers. Ostrich 45, 227-234 (1974) for the evolution of altruism and communal breeding. Am. Vehrencamp, S.L.: The adaptive significance of communal nesting Zool. 14, 63-80 (1974) in groove-billed anis Crotophaga sulcirostris. Behav. Ecol. So- Brown, J.L.: The evolution of behaviour. New York: Norton 1975 ciobiol. 4, 1-33 (1978) Brown, J.L.: Avian communal breeding systems. Annu. Rev. Ecol. Woolfenden, G.E.: Florida scrub jay helpers at the nest. Auk Syst. 9, 123-155 (1978) 92, 1-15 (1975) Brown, J.L., Balda, R.P.: The relationship of habitat quality to Woolfenden, G.E.: Co-operative breeding in American birds. Proc. group size in Hall's babbler (Pomatostomus halli). Condor 79, XVI Int. Ornithol. Congr. (Canberra, Australia). Frith, H.J., 312-320 (1977) Calaby, J.H. (eds.), pp. 674-684. Canberra: Australian Academy Chipeta, G.B.: A study of wind power availability in Kenya. M.S. of Science 1976 thesis, University of Nairobi (1976) Woolfenden, G.E.: Growth and survival of young Florida scrub Douthwaite, R.J.: Some aspects of the biology of the lesser pied jays. Wilson Bull. 90, 1-18 (1978) kingfisher Ceryle rudis Linn. Ph.D. thesis, Makerere College, Woolfenden, G:E., Fitzpatrick, J.W.: The inheritance of territory University of East Africa (1970) in group-breeding birds. BioScience 28, 104-108 (1978) Douthwaite, R.J.: Pied kingfisher Ceryle rudis populations. Ostrich Zahavi, A.: Communal nesting by the Arabian babbler. Ibis 116, 44, 89-94 (1973) 84-87 (1974) Douthwaite, R.J.: Fishing techniques and foods of the pied kingfi- Zahavi, A.: Co-operative nesting in Eurasian birds. Proc. XVI sher on Lake Victoria in Uganda. Ostrich 47, 153-160 (1976) Int. Ornithol. Congr. (Canberra, Australia). Frith, H.J., Calaby, Douthwaite, R.J.: Breeding biology of the pied kingfisher on Lake J.H. (eds.), pp. 685-693. Canberra: Australian Academy of Victoria. J. East Afr. Nat. Hist. Nat. Mus. 31, 1-12 (1978) Science 1976