Ecology and Breeding Habits of the Savanna Hawk in the Llanos of Venezuela

Condor 84261-271 0 The Cooper Ornithological Society 1982

ECOLOGY AND BREEDING HABITS OF THE SAVANNA HAWK IN THE LLANOS OF VENEZUELA

WILLIAM J. MADER

ABSTRACT. -The ecology of the Savanna Hawk (Buteogallusmeridionalis) was studied in the palm-savannas of Venezuela in 1978-1980. Pairs produced only one egg per clutch, no more than one young per calendar year, and had eggsor young in nestsfrom February to December, coinciding largely with the wet season. In comparisonto North American Red-tailed Hawks (Buteojumaicensis), Savanna Hawks had significantly fewer young per pair per year, lower nest success(47.1% vs. 82.5%), mainly due to lower hatching success(64.1% vs. 84.4%) and a lower proportion of pairs laying eggs (73.3% vs. 88.0%). Survival of young Savanna Hawks from hatching to near fledging was 70.4%, and the adult survival rate, 7 1.1%; these values are approximately equal to those for Red-tailed Hawks. The brood size was artificially doubled from one to two at 10 Savanna Hawk nests, but the adults could find only enough food to feed and raise one chick in most cases.Food appeared to be both the ultimate and proximate factor controlling the timing of breeding. The Savanna Hawk has evolved interrelated habits that counter its low fecundity: (1) the capability of renesting after the initial eggs or young have perished; (2) a post-fledging dependency period of four to seven months to help insure survival of the single young; and (3) permanent or long- lived pair-bonds. The ability to renest is enhanced by hunting versatility. Long- lived pair bonds may improve intra-pair activities thereby enhancing the repro- duct&e successof experienced breeders.

Hawks and eaglesin the tropics, in contrast to ezuela which examined: (1) a test of the northern latitudes, may have evolved different hypotheses of Lack and Skutch, (2) what reproductive habits in response to greater breeding habits exist to help offset a low repro- diversities of prey and predators, both avian ductive potential inherent with a clutch size of and mammalian, and different climates. Little one, and (3) how the timing and duration of is known, however, about breeding habits and breeding were related to rainfall and food food relationshipsin neotropical falconiforms. resources. Tropical birds generally lay smaller clutches than their temperate counterparts (Lack 1947, STUDY AREA AND METHODS 1948, 1954). Of the several hypotheses that The Savanna Hawk ranges from Panama to have been proposed to explain this contrast, I central Argentina (Blake 1977). I studied a investigated two dealing with the evolution of population for two breeding seasonsand part tropical clutch sizes:Lack ’s (1947) hypothesis ofa third (15 Januarythrough 15 October 1978, of “maximum reproduction” and Skutch’s 25 April through 3 October 1979, and 26 April (1967) of “readjusted reproduction.” The for- through 3 May 1980) on a cattle ranch in Guar- mer theory, which is more widely accepted, ice State, Venezuela. Colleagues continued statesthat birds produce asmany eggsand thus checking nests after I left, from October to offspringas they can feed. The latter maintains December. The habitat was primarily palm that birds produce fewer young than they are savanna, although a few nests were found in capable of feeding becausethe adult survival tree savanna. The region (llanos) has one dry rate is high and the need to recruit breeders is season(generally December through April) and low. Additionally, Skutch believed that a one wet season (May through November), smaller clutch size may, in part, be a result of which have a pronounced effect on the land. heavy predation pressuresin the tropics, such In February for instance, the palm savannas that smaller broods would require fewer feed- were dry and dusty while in August they were ing visits and therefore less risk of nest detec- sometimescovered with over a meter of stand- tion by predators. The question of what selects ing water. for a smaller clutch size in a predatory bird I chose a study area of 9.4 km2 containing has not been addressed. Here, I present the 23 pairs of hawks for which I determined the results of a 16-month study of the Savanna number of eggsand young produced each year. Hawk (Buteogallus meridionalis, following Additional data, including clutch size and Mayr and Cottrell 1979) in the llanos of Ven- nestingsuccess, were gathered on nestslocated

12611 262 WILLIAM J. MADER outside the study area. Nest contents were and exchangesof prey from adult to immature checked periodically using a mirror attached birds. to a pole. I travelled by motorcycle, horse, and on foot. Fifty adult Savanna Hawks were cap- RESULTS tured with balchatri traps (Berger and Mueller 1959). Forty-nine were individually color- PRODUCTIVITY banded, including 13 pairs in which both Savanna Hawks laid eggs from February to members were banded. I could not tell the sex September (n = 116 clutches, Figs. 1 and 2). of adults becausethey differ little in size and Laying peaked in the first part of July 1978 color (mean wing chord = 39.7 cm, range but had no peak in 1979. The number of 31.5-42.3, n = 49 adults; mean weight = 844 clutches in 1979 was nearly equal for each g, range 720-958, y1= 50 adults). Later, I month from April to July, except for a low in determined the sex of several individuals by June. In 1978, hatching occurred from March their behavior. Territories of five neighboring to October with a peak in mid-August. In 1979 pairs(8 of 10 adultswere initially color-banded) nest failures caused two well-defined periods were determined during the dry and wet sea- of hatching, the first in mid-May and the sec- sonsby plotting 50 or more sightingsfor each ond in early September. In both years, I noted pair from February to October 1978. fledging from May to December. Individual Lack’s hypothesis of maximum reproduc- hawks generally did not nest at the same time tion was tested in 1978 by artificially doubling each year. In 1979, marked adults laid eggs, the normal brood size of one, at nine nests on the average, about 30 days from the date during the first week after hatching. The ages they laid in 1978 (Fig. 3). For both years the of the chicks did not differ by more than 1.5 entire breeding season (egg laying-fledging) days. Growth (weight) of two chicks in a spannedabout 250 days (260 days in 1978 and manipulated nest and a chick in a control nest 240 days in 1979) and coincided to a large was measured every five days until fledging. I extent with the wet season(Figs. 1 and 2). doubled the brood size at another nest in 1979 All clutches( yt = 116) had one egg.Nest suc- in order to observe chick interactions and to cess,from egglaying until a chick was at least determine if doubling affected the amount of 5 weeks old, both years combined, was 47.1% prey brought to the nest. In a blind placed level (n = 85 nests, includes 10 renests).The num- with the nest 85 m away, I watched the nest ber of young producedper nest attempt (started for about 44 h from 7 September to 2 October, with a single egg) was 0.47 (n = 85). Of 75 and for 6 h before the secondchick was added. initial attempts in two years, 36 (48.0%) were The average period of observation was 2 h:5 successful;only 4 out of 10 renestsproved suc- min. (range 1:20-2:33). Another blind was cessful (six other renests were noted but not erected at a nest with one chick in 1979 so that checked to determine productivity). For 16 I could collect information about prey (30 h renests, 10 (62.5%) of the initial nestsfailed at observation). Observationsat both blinds were the egg stageand 6 (37.5%) at the chick stage. often interrupted becauseof storms and flood- Of 24 territories thoroughly checked after an ing. I estimated weights of prey from known initial nest failed, 16 (66.7%) pairs renested. weights of comparably sized animals. All No Savanna Hawks attempted to raise two observations were aided by a 32x spotting young in one calendar year. scope. More nests were successful(P < 0.05) in Platforms were built at the top of nine palm 1978 than in 1979. Of 44 initial nestsin 1978, trees (typical nesting sites) containing neststo 25 (56.8%) were successful.Only three ofseven allow a close inspection of chicks and prey (43%) renests were successful. In 1979, 11 remains. Flooding reduced the number of vis- (35.5%) of 3 1 initial nestswere successful.Only its to some nests. Therefore, I considered a one of three renestswas successful.For both nest successfulif a chick was raised to an age years combined, the time of nest failure was of at least five weeks, even though chicks determined for 48 nests,including six renests. fledged(left the nest) at roughly 6.5-7.5 weeks. Thirty-two (66.7%) failed with eggs and 16 I used an incubation period of 39 days (range (33.3%) with young. Hatching success was 37-40, yt = 4) for back-dating some nestsfrom 64.1% (n = 78), while nestling successfrom known hatchingdates to estimatedlaying dates. hatching to at least five weeksof agewas 70.4% Six immature hawks of known fledgingdates (n = 54). were followed to determine the duration of the Nestsbegun during the first half of the breed- post-fledgingdependency period. Dependency ing seasonhad about the same successas nests of a young bird on its parentswas substantiated begun in the last half. Combining data from by continuous food beggingcalls by the former both years, 18 (62.1%) of 29 nests were suc- ECOLOGY OF THE SAVANNA HAWK 263

;F lo EGG LAYING 6 0: 2 zz l”l”I”I”I A S 0 N D I;? 10 0: N:28 ; zz

J FMAMJJ k?_ 0: 6 FLEDGING N:16 zz 2

IVl’ ”I”I1 ’ JF’ M A M J

FIGURE 2. Breedingcycle of the SavannaHawk in 1979 in the llanos of Venezuela showing peaksand spreadsfor major events in the cycle and their relation to annual precipitation. Renestingattempts are included. Although FIGURE 1. Breedingcycle of the SavannaHawk in 1978 temporary flooding started in the last third of May, long- in the llanos of Venezuela showing peaks and spreadsfor term flooding started in the middle of June and did not major events in the cycle and their relation to annual begin to subsideuntil the last third of October. precipitation. Renesting attempts are included. Flooding startedin the last third of May and did not begin to subside until the middle of October. October 1979. Thirty-five ofthe 45 adults were alive, giving a minimum yearly adult survival cessful for the months before June while 29 rate of 77.8% (assumingthat they all remained (65.9%) were successfulfor June and after. in the area and none lost their bands). I also In 1978, 21 (91.3%) of the 23 pairs in the determined how many had survived to study area laid eggsresulting in 27 eggs,and approximately two years after capture. Of 35 16 young survived to at least five weeks (1.17 surviving adults in 1979, I checked 28 for sur- eggsand 0.70 young per pair). This included vivorship in 1980 (26 April-3 May). Eighteen six attempts at renesting, two of which were or 64.3% of the 28 adults were still alive, so successful(0.26 renest attempts per pair). The the averageannual adult survival rate over the number of laying pairs and production of eggs two-year period was 7 1.1%. Since Savanna and young per pair were significantly lower Hawks are resident and have eggsor young 11 (P < O.Ol)in 1979 than 1978.Twelve(54.5%) months of the year, an approximate breeding of the 22 pairs laid eggsresulting in 14 eggs replacement rate can be determined by sub- and six young (0.64 eggsand 0.27 young per tracting the annual presumed survival rate of pair). In 1979, only three renestsoccurred, one 7 1.1% from lOO%, giving 28.9%. of which was successful(0.14 renest attempts Eleven of the 13 pairs with both mates per pair, significantly less, P < 0.05, than in banded were checked one year later and seven 1978). For both years combined, an average (64%) pairs were intact. Only one member of of 0.9 1 eggsand 0.5 3 young were produced per each of the other four pairs remained, although pair, including nonbreeders (non-layers). each had re-mated. Of the seven intact pairs left in 1979 only three (42.9%) remained intact ADULT SURVIVAL AND MOVEMENT in 1980. After two years then, 3 (27.3%) of 11 Forty-one adults were trapped and color- pairs were still together. I consider the adult banded from 27 January to 13 March 1978, annual survival and fidelity ratesto be accurate and 8 more from 5-10 October 1978, all of because:(1) color-marked birds were not found which were members of pairs. The first 4 1 were outside their home ranges indicating little or checked 3-18 April in 1979 and all were still no dispersal, (2) the location of territories in present. For 45 of the 49, I determined how the study area remained essentially the same many had survived to approximately one year from 1978 to 1979, except for one pair, and after capture. The territories of the January- (3) I saw no evidence that birds who had lost March hawks were checked in 1979 from 26 their bands were misidentified (based on 32 April to 7 May, and the October hawks on 1 adults that received two bands each). 264 WILLIAM J. MADER

IA 0 2, Nr15 a

I I I 111. 91. 71. 51. 31. 11. 0 ll- 31. 51. 120 100 80 60 40 20 20 40 60 101. 81. 61. 41. 21, 1. 1. 21. 41. 111 90 70 50 30 10 10 30 50

NO. DAYS EARLIER NO. DAYS LATER

THAN 1978 CLUTCH THAN 1978 CLUTCH

FIGURE 3. A comparison of laying dates in 1979 to those of 1978 for the same individually marked SavannaHawks with the differencesexpressed in days earlier or later than 1978. Of 20 adults followed (all from different pairs), five did not produce eggsin 1979.

Twenty-three pairs averaged 0.41 km* of nests. The average weight of the prey (84% palm savanna per pair. This value includes were whole) when delivered to the nests was some areas not used by the adults but by one- 19 g. or two-year-olds in predominantly immature plumage. The home range included a hunting CLUTCH SIZE MANIPULATIONS area that was defended against other hawks The adults acceptedand fed both chicks at the and a soaringrange where the adults flew over 10 manipulatednests. Of the nine nestsin 1978, their own area and their neighbor’s as well. two failed from unknown causes,six produced The home ranges of five neighboring pairs in only one young, and one produced two young. palm savanna averaged 0.14 km2 (range = The latter chicks were lighter than average 0.08-0.16 km2), none of which overlapped. weight, fledged two weeks later than normal, Fledged hawks remained inside their parent’s and were severely infested with botfly larvae. home rangeuntil they becameindependent four Thus, only one of the nine nestsproduced two to seven months after fledging (n = 6 young). chicks;this is statistically significant(binomial test, P < 0.005). In 1979, the parents at the singlemanipulated nest raised both young. FOOD For those six neststhat lost one of two chicks, The remains of prey at Savanna Hawk nests insufficient food was probably the cause. Star- were chiefly semi-terrestrial fresh-water crabs vation of one chick occurred at two nests,one (Dilocarcinus dentutus, 57.3%) birds (24.5%) at 48 days of age and the other at 44. The and reptiles (10.7%) (Table 1). Sampling of remaining chicks disappeared between 5-22 remains was biased becauselarge or partly in- days of age. The weights of both chicks at one edible items such as crabs, lasted longer than manipulated nest were compared to those small or entirely consumed prey (eels). Direct obtained at a control nest containing only one observation of prey delivery therefore gave chick (Fig. 4). Chicks in the manipulated nest more accurate data on food habits. Prey grew significantlyslower (P < 0.05, full model observedbrought to two nestswere mostly eels multiple regressiontest, F = 3.49, df = 2, 20) (44.9%) frogs(16.3%), crabs(12.3%) and birds than the control and lost weight at 28-33 days (12.2%; y1= 69, Table 2). All of these are wet- of age, suggestinginsufficient food. Delayed seasonprey. In the dry season,eels and crabs fledging was apparent at other manipulated likely comprise little of the available food nestsin which only one chick survived (three becauseso little water is present.Probably more to seven days late). At three nests, I marked mammals and birds are taken in the dry sea- both chicks to determine if adults raised their son; indeed, I saw two adults carrying rodents own or the introduced chick. At one nest, the then. adults raised only the introduced chick, while I determined the rate of prey delivery at two at two others the adults raised both (one each nestsduring the nestling phase (Table 3) -one in 1978 and 1979). nest had one chick and the other two chicks Observations at the blind of the 1979 (manipulation). Although incomplete, the data manipulated nest disclosed no sibling aggres- suggestthat about the same amount of prey sion even though chick two (C2) developed and number of items were brought to both slower than the other (Cl) and weighed only ECOLOGY OF THE SAVANNA HAWK 265

TABLE 1. Prey remains found at Savanna Hawk nests TABLE 2. Prey items observedbrought to two Savanna in Guarico State, Venezuela. Hawk nestsand nearby areasin Guarico State, Venezuela.

NO. % NO. 96 Prey individuals individuals individ- individ- % Prey uals uals biomass Mammals Unidentified rodents 3 6.1 1.1 Unidentified rodents 4 2.5 Cowbird (Molothrus sp.) 1 2.0 2.7 Birds Purple Gallinule 1 2.0 11.4 Purple Gallinule (Porphyruh Unidentified birds 4 8.2 10.7 martinica) 5 3.1 Total birds 6 12.2 24.8 Ani (Crotophagasp.) 2 1.3 Unidentified lizards 3 6.1 14.9 Unidentified doves 15 9.4 Unidentified snakes 1 2.0 5.3 Unidentified flycatchers 1 0.6 Unidentified birds 16 10.1 Total reptiles 4 8.2 20.2 Total birds 39 24.5 Unidentified frogs 8 16.3 7.3 Eel (Symbranchus Reptiles marmoratus) 22 44.9 42.0 Rattlesnake (Crotalus sp.) 1 0.6 Crab (Dilocarcinus Iguana (Iguana iguana) 1.9 dentutus) 6 12.3 4.6 Unidentified lizards : 5.0 TOTAL 49 100.0 100.0 Unidentified snakes 1.0 Unidentified reptiles ; 1.9 Total reptiles 17 10.7 a total of 1,45 3 bites up to 32 days of age (when Amphibians I ceased watching from the blind). Cl con- Unidentified frogs 5 3.1 sumed most of the food (63.5% of the bites) Crustaceans until the chicks were 21 days old, at which Crab (Dilocarcinusdentatus) 91 57.3 time C2 consumed the most until 32 days Insects (57.5%). At 3 1 days, C2 weighed586 g or 86.7% Unidentified insects(Hemiptera) 3 1.9 of Cl’s weight (676 g). Both chicks fledged, TOTAL 159 100.0 although C2 was underweight. DISCUSSION 73.8% as much as Cl at 26 days of age. When CONTROL OF CLUTCH SIZE the female fed one or both chicks (only the Many studiessupport Lack’s theory that birds female fed chicks, sex was determined by cop- raise as many young as they can feed (see ulation) during a visit to the nest, I defined this reviews by Lack 1954, 1968, Klomp 1970). as a “feeding bout.” When she offered food to However, this explanation appearsinadequate a chick with her bill and it was accepted, I for some species (see von Haartman 1971). termed this a “bite.” I witnessed 27 bouts for Indeed, the converse,that clutch sizeis adjusted

TABLE 3. Comparative rates of prey delivery at two Savanna Hawk nests, one with one chick and the other with two chicks (manipulation).

Age of chick(s)in days 1-7 8-14 15-21 22-28 29-35 3643 Avg.

One chick Total period of observation (h) 6:05 - - 7:05 10:30 12120 - Total no. of adult visits 8 - - 7 6 No. visits per h of observation 1.32 - - 0.99 0.57 120.97 0.96- No. prey items brought to nest - 5 No. prey items per h of observation 0.71 i48 120.97 0.62- Prey biomassbrought to nest per h (g) 18 14’ 24 15 Two chicks Total period of observation (h) - 17:03 12:43 10:30 4:oo - - Total no. of adult visits - 18 18 8 2 - - No. visits per h of observation - 1.06 1.42 0.76 0.50 - 0.94 No. prey items brought to nest - 5 12 6 2 - No. prey items per h of observation - 0.29 0.95 0.57 0.50 - O.sS Prey biomass brought to nest per h (g) - 4 14 14 9 - 10

*The data are from two nests.The chick in ageclass 1-7 daysis the samechick usedm the manipulationfor the secondnest, but the data are prior to the addition of the secondchick. The data for the otherthree age classes are from anothernest with one chick. 266 WILLIAM J. MADER to compensate for mortality is held by some regions (Ricklefs 1969) although some recent (Wynne-Edwards 1962, Skutch 1976). Results evidence suggests otherwise (Oniki 1979). of my brood manipulations with Savanna Skutch (1967) countered Lack’s hypothesis of Hawks support Lack’s explanation. Only one food limitation by presenting evidence that a of nine manipulated broods that were exper- small clutch and later small brood, reducesthe imentally increased to two chicks in 1978 risk of nest predation by requiring fewer feed- actually raised both chicks, although the one ing visits and less chance of nest detection by nest manipulated in 1979 raised both chicks. a predator. Skutch suggestedthat this was why In two casesthe dead chick apparently died food items brought to the nest are large and from starvation. That both manipulated chicks young do not often give beggingcalls. developed more slowly than normal and/or I have no evidence to say that a clutch of fledged later, supports the contention that one in the Savanna Hawk resulted from fewer inadequate food resourcesfavored a clutch size feeding visits and a reduced chance of preda- of one. Additionally, a comparison of a brood tion. First, Savanna Hawks bring prey items of one chick to a manipulated brood of two of smaller average size to nests(47 prey items indicated no significant differences in the rate brought to nestsweighed an estimated average of prey items delivered to the nest nor the of 19 g at the time of arrival) than comparably amount of prey carried to the nest (Table 3). sized North American buteonines, which gen- In short, the adults with two chicks brought erally bring squirrel- and rabbit-sized items no more food to their nest than the adults with (50-2,300 g; e.g., Red-tailed Hawk, Buteo one chick. jamaicensis, Fitch et al. 1946, Craighead and The above conclusion must also be dis- Craighead 1956, Seidensticker and Reynolds cussedin terms of the “sibling aggression”the- 1971, Luttich et al. 1971, Gates 1972, Smith ory proposed for certain species of African and Murphy 197 3, McInvaile and Keith 1974; eagles. The Lesser Spotted Eagle (Aquila Swainson’s Hawk, Buteo swainsoni,Smith and pomarina) and Verreaux’s Eagle (A. ver- Murphy 1973, Porton 1977, Dunkle 1977; reauxii) typically lay two eggs;invariably when Ferruginous Hawk, Buteo regalis, Smith and both hatch, only one chick survives, often Murphy 1973, Woffinden and Murphy 1977; because the first and larger kills the smaller Harris’ Hawk, Parabuteo unicinctus, Mader (Meyburg 1974, Gargett 1977). Brown et al. 1975). The Black Hawk (Buteogallus anthra- (1977) reviewed this phenomenon and con- cinus) in Arizona differs from the foregoing cluded that competition between siblingsand becauseit catchesreptiles and fish comparable survival are not affectedby availability of food. in size to Savanna Hawk prey (Schnell 1979). These authors also rejected the idea that a sec- Because Savanna Hawks bring less food to ond egg servesas a reserve if the first does not nests,they probably require as many or more hatch, but other evidence supports this inter- feeding visits each day than temperate rela- pretation (Meyburg 1974, Stinson 1979). Thus, tives, but I know of no comparable data on in theseeagles, the survival of one chick appears North American buteonines. Secondly, nest- to result from sibling aggressionand not an lingsappeared to call for food as often as North immediate scarcity of food. However, it American buteoninesthat I have observed(e.g., appears likely that fratricide avoids competi- Red-tailed Hawk, Harris’ Hawk). tion for a food supply that would ultimately Since Savanna Hawks have a greater nest be inadequateto feed two fledgednestlings (see successthan non-raptorial open nestersin the discussion by Stinson 1979). During obser- wet tropics (47% vs. 34% respectively; see vations (44 h) at a manipulated nest of a Skutch 1976 for latter), it is possiblethat pre- Savanna Hawk, I saw no sibling aggression dation is not an important cause of nest fail- even though one chick was larger and devel- ures and that there is less need for larger prey oped faster than the other (both fledged). Also, items and lessbegging. However, circumstan- I saw no physical injuries on the two starved tial evidence suggeststhat most nest failures chicks mentioned above or any of the other may indeed have resulted from predation. chicksat manipulated nests.Sibling aggression Savanna Hawks often chased Crested Cara- in eagles, and probably other raptors, often caras (Polyborus plancus), Harris’ Hawks, results in obvious injuries to the lesser chick Crane Hawks (Geranospiza caerulescens),and (Meyburg 1974, Brown et al. 1977). My find- conspecificsaway from nests.Also, tree nests, ingssuggest, therefore, that the survival of only particularly those with vines to the ground, one SavannaHawk chick at a manipulated nest were probably preyed upon as easily by arbo- was not a result of sibling aggressionbut inad- real mammals and large snakesas other bird equate food resources. nests; snakes are thought to be the principal The rate of predation on bird nests appears predators of tropical bird nests(Skutch 1976). to be much higher in tropical than in temperate John Robinson (pers.comm.) observeda group ECOLOGY OF THE SAVANNA HAWK 267 of Cebus monkeys (Cebus nigrivittates) sur- 900 _ round and molest a nearly fledged Roadside mean weight of 50 adults Hawk (Buteo magnirostris) and Tropical 800 _ Screech-Owl(Ohs choliba)near my study area. 700 _ Such monkeys are known nest robbers (Rob- inson, pers. comm.). Tayras (a weasel, Eira v) 600 _ barbara), Cebus monkeys, ocelots (Felis par- I Q 500 _ dalis), jaguarondis (F. yagouaroundz), opos- sums (Didelphis sp.), raccoons (Procyon can- crivorus), and other possible predators were - seen in or near Savanna Hawk territories. It thus seemsreasonable that predation contributed to nest failures. In summary, I suggestthat a clutch size of one evolved in the Savanna Hawk primarily becauseof a limited food supply. This supports -7 Lack’s idea of the control of clutch size, but 3 8 13 18 23 28 33 38 43 48 53 does not exclude other factors that may have AGE IN DAYS contributed indirectly, suchas predator avoid- FIGURE 4. Weight increasesof three young Savanna ance, or the effectsof competition on food sup- Hawks measured once every five days. The three devel- ply and clutch size (see Cody 1966 for a dis- opment curves are from two nests. The first representsa cussion). control nest that contained a normal brood of one chick (l), while the secondnest contained its original chick (3) TIMING OF BREEDING plus an introduced one (2). Only the introduced chick survived at the manipulated nest. Most Savanna Hawks (72% of 113 pairs) laid eggs during roughly the first half of the wet season,May to August (see Figs. 1 and 2). I and a proximate factor controlling the timing have no data on prey abundance (surveys of breeding. proved impractical because of flooding), but Two other factors that may influence the eels, crabs, and frogs seemed far more numer- timing of breeding are: (1) the availability and ous during the wet seasonthan the dry. Avian exploitation of diverse prey types, and (2) time densities and frequency of breeding were also of initial nest failures. The length of a hawk’s greatestduring the wet season(Thomas 1979). breeding seasondepends not only on how long The remaining hawks laid either before May prey is available (see review by Newton 1979) (24%) or near the end of the wet season(4%, but also on how prey is exploited (Mader 1978). September through October). Laying thus For example, neighboringRed-tailed and Har- occurredover many months, especiallyin 1979. ris’ hawks in Arizona take the same types of Since Savanna Hawks have non-overlapping prey (birds, mammals, reptiles) but the latter territories, suchunsynchronized breeding can- are more versatile in what they catch and their not be the result of pairs trying to reduce intra- breeding seasonis twice as long (includes sec- specific competition with their neighbors. ond broods, Mader 1978). Savanna Hawks are Instead, pairs may become adequately nour- equally versatile, catching fish, crabs, snakes, ished to breed at different times, possibly iguanas,and birds (see Tables 1 and 2). Those because prey availability differs within terri- pairs that nest before flooding or in locally tories and some individuals hunt more effi- unflooded savanna certainly take mostly ter- ciently than others. Fogden (1972) found that restrial or arboreal prey. Exploitation of diverse various tropical passerineslaid eggswhen their types of prey probably provides a greater food protein reservesand physiology were ready to base and helps insure adequate food supplies bear the added stressof breeding, and not in during renesting attempts, thereby contribut- response to rainfall. The absence of a clear ing to a longer breeding season. correlation between the onset of the rains (or Laying dates are also influenced by failures flooding) and laying in the Savanna Hawks of first nest attempts. When pairs of Savanna (Figs. 1 and 2) suggeststhat rainfall was not a Hawks failed with eggsor young at the begin- proximate cue to breeding. In short, for those ning of the rains they often renested a month hawks that breed in the wet season,it may not or two later. The same may be true of other be the rains that trigger breeding but the neotropical falconiforms. A pair of White- increase in food supply that follows. Those tailed Hawks that nested and failed in the dry hawks that breed before the rains may do so season,renested in the wet, even though this becausethey are already physiologically pre- speciesnormally breeds in only the dry season pared. Food may therefore be both an ultimate in the llanos of Venezuela (Mader 198 1). New- 268 WILLIAM J. MADER

TABLE 4. Breeding parameters in two North American buteoninesand the South American Savanna Hawk.

Red-tailed Hawk Black Hawk Savanna Hawk

Average incubation period in days 35 39 39

Hardy 1939 Schnell 1979 Mader 1978 Average clutch size (n) 2.79 (476) 1.67 (12) 1.00 (108) Henny and Wight 1972 Schnell, pers. comm. (476) Average hatching success% (n) 84.4 (379) - 64.1 (78) Fitch et al. 1946 (38) Craighead and Craighead 1956 (33) Seidenstickerand Reynolds 1971 (22) Luttich et al. 1971 (172) Smith and Murphy 1979 (114) Average nestling successfrom hatching 73.0 (152) - 70.4 (54) to fledging % (n)

Fitch et al. 1946 (26) Seidenstickerand Reynolds 1971 (15) Smith and Murphy 1979 (84) Craighead and Craighead 1956 (27) Nestling period in days 41-47 43-50 45-50 Fitch et al. 1946 Schnell 1979 Johnson 1973 Wiley 1975 Average no. young/nest attempt (n)b 1.47 (281) 1.0 (58) 0.47 (85) Orians and Kuhlman 1956 Schnell, pers. comm. (87) Gates 1972 (3 1) Johnson 1975 (121) Mader 1978 (42) Average nest success% (n) 82.5 (309) 81.0 (58) 47.1 (85) Orians and Kuhlman 1956 Schnell, pers. comm. (87) Gates 1972 (31) Johnson 1975 (78) Wiley 1975 (33) Mader 1978 (42) Smith and Murphy 1979 (38) Average no. pairs laying eggs% (n) 88.0 (382) - 73.3 (45) Craigheadand Craighead 1956 (12) Orians and Kuhlman 1956 (60) Luttich et al. 1971 (66) McInvaile and Keith 1974 (107) Johnson 1975 (137) Approximate length of breeding season 114 100 250 in days (laying to fledging) ECOLOGY OF THE SAVANNA HAWK 269

TABLE 4. Continued.

Red-tadedHawk Black Hawk SavannaHawk

Smith and Murphy 1973 Schnell, pers. comm. - (83 days) Wiley 1975 (141 days) Mader 1978 (118 days)

r Fledgingwas not consistentlydefined in the referencesfor Red-tailed Hawks and the criteria for fledging probablyvaried. SavannaHawk chickswere followedto at least five weeks,and wereassumed fledged (left the nestand perchednearby in sametree or flew) if they survived to this age.Of 28 chicks followedfrom live weeksto fledging,27 (96%) fledged(does not includechicks from manipulatednests). b Data for nestattempts and nestsuccess include only thosenests that beganwith at leastone egg. ton (1979:107) commented that it was not of lation that lay eggs (73.3% vs. 88.0%, much importance when large tropical raptors P < 0.05); the survival of young from hatch- (kites to large eagles)with prolonged breeding ing to near fledging is about the same (Table seasonsnested, sincethe breeding cycle was so 4). Savanna Hawk nest successwas also sig- long. I suggestthat this is probably true only nificantly lessthan that for Black Hawks (47.1% for those speciesthat (a) do not renest after a vs. 8 l.O%, P < 0.01). Johnson (1975) found failure within the same year, and (b) do not an average of 1.36 fledglings/pair/year have to lay primarily in one seasonin regions (n = 137) in Montana, and Craighead and that have one wet and one dry season. For Craighead ( 19 5 6) 0.9 1 fledglings/pair/year those raptors that do renest and lay in prin- (n = 22 pairs) in Michigan. Savanna Hawks cipally one seasonsuch as the Savanna Hawk, raised far fewer young/pair/year (avg. = 0.53 a pair increasesfitness by breeding as early as in 9.4 km2 study area from 1978-1979) than possible in a given seasonin order to renest did Red-tailed. While the annual survival rate during favorable conditions should the first of adult Red-tailed Hawks is estimatedbetween clutch be lost. 76.2-80.0% (Henny and Wight 1972) the Savanna Hawk has a rate of about 7 1.1%. Pre- REPRODUCTIVE RATES AND BREEDING PAT- vious studies of tropical birds indicate that TERNS they usually have a greater adult survival rate Tropical vs. temperate hawks. The reproduc- than their temperate counterparts (Fogden tive rates (number of fledglings per adult per 1972, Snow and Lill 1974). year) of land birds are generally lower in trop- Tropical hawks and eaglesoften have longer ical than temperate species(Lack 1968, Skutch incubation and nestling periods than temper- 1976). To find out if this is also true in hawks, ate latitude ones of comparable size and taxo- I compared the reproductive rate of the nomic group (Brown 1976, Newton 1979) even Savanna Hawk with those of two related North though it probably increases the chances of American species,the Red-tailed Hawk and reproductive failure by storms, predation, or the Black Hawk. All are buteonines of com- death of an adult. Indeed the Savanna Hawk parable size (Red-tailed Hawk 1,126 g, Craig- has a longer incubation period than the Red- head and Craighead 1956; Black Hawk 945 g, tailed and a slightly longer nestling period. Schnell in Newton 1979; Savanna Hawk 844 However! the Savanna Hawk has incubation g); they occupy similar trophic levels, and live and nestling periods roughly equal to those of in largely open habitat. The Red-tailed Hawk temperate populations of Black Hawks (Table is migratory in at least parts of its range but 4). Black Hawks are primarily neotropical and does not breed in South America. In contrast, their long incubation and nestling periods in the Black Hawk has a primarily neotropical Arizona may simply reflect their generally distribution but does breed as far north as the southern distribution. southwesternUnited Stateswhere it is migra- Savanna Hawks have evolved at least the tory (Blake 1977); the data below are from following interrelated habits that help counter Arizona (Schnell 1979, pers. comm.). low fecundity: (1) the ability to renestafter eggs SavannaHawks averagefewer eggsper clutch or young have perished in an initial nesting (1 .O) than Red-tailed Hawks (2.79) and Black attempt, (2) a long fledglingdependency period Hawks (1.67) and far fewer young per nest of four to seven months to help insure survival attempt (0.47 young) than Red-tailed (1.47) of the single young, and (3) pair-bonds that and Black (1.0). Savanna Hawks have sig- remain intact from one breeding seasonto the nificantly lower averages than Red-tailed next or until a member dies. Hawks for nest success (47.1% vs. 82.5%, Renesting.Although Savanna Hawks renest P < 0.0 l), hatching success(64.1% vs. 84.4%, after a nest failure with young, this behavior P < 0.01) and percentageof pairs in a popu- is rare among North America hawks. Of the 210 WILLIAM J. MADER

12 buteonines that breed in North America, SUMMARY only the Harris’ Hawk is known to renest after Savanna Hawks laid only one egg per clutch the loss of young (Mader 1978); this species in the llanos of Venezuela becausethey usually also ranges into the neotropics. However, cannot feed more than one chick. However, in renestingof buteonines after the loss of eggsis contrast to all but one North American not uncommon. Little is known about renest- buteonine, Savanna Hawks renested after the ing in tropical hawks after the loss of young. lossof a chick. Although most nestingoccurred Melierax goshawksin equatorial East Africa in the wet season,rainfall by itself was prob- breed twice a year (Smeenk and Smeenk- ably not the cue that started Savanna Hawks Enserink 1975), sopresumably they renestafter breeding; rather it appeared to be the increase the loss of young. Thus it is not clear if renest- in food that followed. Home rangeswere very ing of this kind is a universal adaptation of small, averaging0.14 km2, but pairs neverthe- tropical hawks, but it clearly increases the lesstook a great diversity of prey-fresh water reproductive successof Savanna Hawks. eels,frogs, crabs,birds, reptiles and mammals. Fledglingdependency. The long post-fledg- Fifteen marked adults laid eggs,on the aver- ing dependency period of the Savanna Hawk age, about 30 days from the date they laid in is about two to three times longer than that of the previous year. Nest successwas about 47%, the Red-tailed Hawk (l-2.5 mo., Johnson regardlessof whether pairs bred in the first half 1973), Swainson’s Hawk (1.0 mo., Fitzner in of the wet seasonor the last. Little is known Newton 1979), or Black Hawk (1.5-2.0 mo., about the breeding ecology of other neotrop- Schnell 1979). Hawks in Venezuela of com- ical falconiforms. parable size to the Savanna Hawk show the same phenomenon. I found apparent depen- ACKNOWLEDGMENTS dency periods of ten months in the White- It is a pleasureto thank Tomas Blohm, owner of the ranch tailed Hawk and about seven months in the I lived and worked on, without whose generoushelp this Great Black Hawk (Buteogallusurubitinga, project would not have been possible.Sr. Carlos Colonel Mader 198 1). The onset of migration contrib- also kindly let me work on his ranch. I gratefully acknowl- utes to a shortened period of fledgling depen- edge supporting grants from the Frank M. Chapman Memorial Fund of the American Museum of Natural His- dence in migratory North American hawks. tory, the National Geographic Society, Brigham Young Nevertheless, even those that are resident in University, and Tomas Blohm. Specialthanks go to Peter parts of their range have shorter dependency August, Miguel Carballo, Susy Mader, Jose I& Perez, periods than that of the Savanna Hawk. Red- John Robinson and Robert Brooks for their helo in the field. Kerry Rabenold also assisted,and allowed me to use shoulderedHawks (Buteolineatus) in southern his observation tower at hawk nests. Betsy Thomas pro- California have a fledgling dependency period vided me with wei&htsof eels and frogs. John Eisenberg, of 2.0-2.5 months (Snyder and Wiley 1976) R. Rudran and Carolyn Crockett of the-SmithsonianInsti- and Harris’ Hawks in southern Arizona gen- tution provided field equipment. The late Paul Schwartz, erally two to three months (Mader 1975). Taken and Dean Amadon, Steve Hilty, William H. Phelps, Jr., and EdgardoMondolfi helped me get startedin Venezuela. together, these facts suggestthat a long depen- I appreciatethe valuable criticis& made on this project dency may provide added parental care and and paper by Clayton M. White. Other helpful comments increasedchances of survival for small, infre- were made on portions or drafts of this paper by Joseph quent broods in the tropics. R. Murphy, Hal L. Black, Clyde L. Pritchett, Dean Ama-

don. Kern,_ Rabenold. and James Wilev.- MvI wife Susy, PairJidelity.I tried to test the idea that pro- who was ever inspiring and supportive, brought humor longed pair-bonds increase reproductive suc- and perspectiveto many long days on the savannas. cessbut could not find any information on pair fidelity for any North American buteonines. In areas where the individuals must migrate, LITERATURE CITED pair bonds probably last only through the BERGER,D. D., AND H. C. MUELLER.1959. The bal-chatri: breeding season.However, since some migra- a trap for birds of prey. Bird Banding 30: 18-26. BLAKE,E. 1977. Manual of neotropicalbirds. Vol. 1. Univ. tory hawkshave a high degreeof nest-sitefidel- of Chicago Press, Chicago. ity, the same individuals may pair at the same BROWN,L. H. 1976. Eaglesofthe World. Universe Books, nest-site year after year, creating a situation New York. similar to permanent pair-bonds. Pair fidelity BROWN,L. H., V. GARGETT,AND P. STEYN.1977. Breeding in seabirds improves intra-pair activities such successin someAfrican eaglesrelated to theoriesabout sibling aggressionand its effects. Ostrich 48:65-7 1. ascourtship and synchronyof mating and often CODY, M. L. 1966. A general theory of clutch size. Evo- increases successful reproduction (Coulson lution 20: 174-l 84. 1966, Mills 1973, Davis 1976). Long-term pair- COULSON,J. C. 1966. The influence of the pair-bond and bonds may help offset a low reproductive age on the breeding biology of the Kittiwake Gull potential in the Savanna Hawk, principally by &a tridactyla.J. Anim. l%ol. 35:269-219. CRAIGHEAD,J. J., AND F. C. CRAIGHEAD,JR. 1956. Hawks, insuring fertilization and egglaying as soon as owls and wildlife. Stackpole,Harrisburg, PA. the adults are capable. DAVIS,J. W. F. 1976. Breeding successand experiencein ECOLOGY OF THE SAVANNA HAWK 271

the Arctic Skua, Stercorariusparasiticus(L). J. Anim. ONIKI, Y. 1979. Is nesting successof birds low in the Ecol. 45531-535. tropics?Biotropica 11(1):60-69. DUNKLE, S. W. 1977. Swainson’s Hawks on the Laramie ORIANS,G., AND F. KUHLMAN. 1956. Red-tailed Hawk Plains, Wyoming. Auk 94:65-7 1. and Homed Owl populations in Wisconsin. Condor FITCH, H. S., F. SWENSON, AND D. F. TILLOTSON.1946. 58:371-385. Behavior and food habits of the Red-tailed Hawk. PORTON,I. 1977. Some aspectsof the breeding behavior Condor 48:205-237. of Swainson’s Hawk. M.S. thesis, Univ. of Arizona, FOGDEN,M. P. L. 1972. The seasonalityand population Tucson. dynamics of equatorial forest birds in Sarawak. Ibis RICKLEFS,R. E. 1969. An analysisof nesting mortality in 114:307-342. birds. Smithson. Contrib. Zool. No. 9. GARGETT,V. 1977. Sibling aggressionin the Black Eagle SCHNELL,J. H. 1979. Black Hawk (Buteogallusanthraci- in the Matopos, Rhodesia. Ostrich 49:57-63. nus). In Habitat management series for unique or GATES,J. M. 1972. Red-tailed Hawk populationsand ecol- endangeredspecies. U.S. Dep. Inter. B.L.M. Rep. 18. ogy in east-central Wisconsin. Wilson Bull. SEIDENSTICKER,J. C., AND H. V. REYNOLDS.197 1. The 84:421433. nesting, reproductive performance and chlorinated HARDY,R. 1939. Nesting habits of the Red-tailed Hawk. hydrocarbon residues in the Red-tailed Hawk and Condor 4 1:79-80. Great Homed Owl in south-centralMontana. Wilson HENNY,C. J., AND H. M. WIGHT. 1972. Populationecology Bull. 83:408-418. and environmental pollution: Red-tailed and Coo- SKUTCH,A. 1967. Adaptive limitation of the reproductive per’s hawks, p. 229-250. In Population Ecology of rate of birds. Ibis 109:579-599. Migratory Birds:A symposium.U.S. Dep. Inter. Wildl. SKUTCH,A. 1976. Parent birds and their young. Univ. Res. Rep. No. 2. Texas Press,Austin. JOHNSON,S. J. 1973. Post-fledging activity of the Red- SMEENK,C., AND N. SMEENK-ENSERINK.1975. Observa- tailed Hawk. Raptor Res. 7:4348. tions on the Pale Chanting Goshawk Melieraxpoliop- JOHNSON,S. J. 1975. Productivity of the Red-tailed Hawk term, with comparative notes on the Gabar Goshawk in southwesternMontana. Auk 921732-736. Micronisusgabhr. Ardea 63:93-l 15. KLOMP,H. 1970. The determinationof clutch-sizein birds. SMITH.D. G.. AND J. R. MURPHY. 1973. Breedine ecoloev Ardea 58: 1-124. ofraptori in the easternGreat BasinofUtah.%righaG LACK,D. 1947. The significanceof clutch-size,parts 1 and Young Univ. Sci. Bull., Biol. Ser. 18 (3). 2. Ibis 89:302-352. SMITH,D. G., AND J. R. MURPHY.1979. Breeding responses LACK,D. 1948. The significanceof clutch-size,part 3. Ibis of raptors to jackrabbit density in the eastern Great 90:2545. Basin Desert of Utah. Raptor Res. 13:1-14. LACK,D. 1954. The natural regulationof animal numbers. SNOW,D. W., AND A. LILL. 1974. Longevity records for Clarendon Press, Oxford- some neotropical land birds. Condor 761262-267. LACK, D. 1968. Ecological adaptations for breeding in SNYDER,N. F. R., AND J. W. WILEY. 1976. Sexual size birds. Methuen, London. dimorphism in hawks and owls of North America. LUTTICH,S. N., L. B. KEITH,AND J. D. STEPHENSON.197 1. Omithol. Monogr. 20:1-96. Population dynamics of the Red-tailed Hawk (Buteo STINSON,C. H. 1979. On the selective advantage of frat- jamaicensis)at Rochester, Alberta. Auk 88:75-87. ricide in rautors. Evolution 33:1219-1225. MADER,W. J. 1975. Biology of the Harris’ Hawk in south- THOMAS,B. T. 1979. The birds of a ranch in the Vene- em Arizona. Living Bird 14:59-85. zuelan llanos, p. 213-232. In J. F. Eisenberg [ed.], MADER,W. J. 1978. A comparative nestingstudy of Red- Vertebrate ecologyin the northern neotropics.Smith- tailed Hawks and Harris’ Hawks in southernArizona. sonian Inst. Press., Washington, DC. Auk 95~327-337. VON HAARTMAN, L. 1971. Population dynamics, p. MADER,W. J. 1981. Notes on nestingraptors in the llanos 391459. In D. S. Famer and J. R. King [eds.],Avian of Venezuela. Condor 83:48-5 1. _ biology. Vol. 2. Academic Press, New York. MAYR. E.. AND G. W. COTTRELL.1979. Birds ofthe World. WILEY,J. W. 1975. The nesting and reproductive success Vbl. ‘1. Harvard Univ. Press, Cambridge, MA. of Red-tailed Hawks and Red-shouldered Hawks in MCINVAILE,W. B., JR., AND L. B. KEITH. 1974. Predator- OrangeCounty,Califomia, 1973. Condor77:133-139. prey relations and breeding biology of the Great WOFFINDEN,N. D., AND J. R. MURPHY. 1977. Population Homed Owl and Red-tailed Hawk in central Alberta. dynamics of the Fermginous Hawk during a prey Can. Field-Nat. 88: l-20. decline. Great Basin Nat. 37:41l-425. MEYBURG,B.-U. 1974. Sibling aggressionand mortality WYNNE-EDWARDS,V. C. 1962. Animal dispersionin rela- among nestling eagles.Ibis 116:224-228. tion to social behavior. Oliver and Boyd, Edinburgh. MILLS,J. A. 1973. The-influence of age and pair-bond on the breeding bioloav of the Red-billed Gull Larus 275 E., 8180 South, Midvale, Utah 84047. Received 15 novaeholla>diae -&opulinus. J. Anim. Ecol. January 1981. Final acceptance9 November 198 1. 42:147-162. NEWTON,I. 1979. Population ecology of raptors. Buteo Books, Vermillion, SD.