Preferred Nest Spacing of an Obligate Cavity-Nesting Bird, the Tree Swallow

Home , Mangrove swallow, Progne, Sand martin, Tree swallow

The Condor87:356-363 0 The CooperOrnithological Society 1985 PREFERRED NEST SPACING OF AN OBLIGATE CAVITY-NESTING BIRD, THE TREE SWALLOW

ALISON MULDAL H. LISLE GIBBS

AND RALEIGH J. ROBERTSON

ABSTRACT. - In order to determine the preferred dispersion of a population of breeding Tree Swallows (Tachycineta bicolor) in Ontario, we set up an abundant supply of nest boxes with a variety of distancesbetween them. The 72 boxes were arranged in 12 equidistantly spacedspirals. Within a spiral, the distance between boxes was much smaller than between spirals. Over five breeding seasons,we observed the order and positions in which swallows settled in the spirals. Pairs of swallows usually settled in empty spirals before settling in spirals occupied by conspecifics,but they did not avoid nesting in spirals occupied by Eastern Blue- birds (Siulia sialis). Swallows did not show any spacing preferences when their nearest neighbors were in different spirals, and were therefore more than 36 m away. Within spirals,however, swallowsnested as far as possiblefrom each other when their nests were less than 14 days apart. Swallow nests in the same spiral also tended to be spaced out temporally. We conclude that, over the range of distances within a spiral, Tree Swallows prefer to space their nests as far from conspecificsas possible. The observed spacingpattern probably arises from territorial behavior that is directed toward defense of a nest site from intruders.

Breeding birds dispersetheir nestsin a variety ia; Bent 1942) and White-backed Swallows of patterns ranging from the tightly clumped (Cheramoecaleucosternum), which excavate distribution of colonial speciesto the uniform their own burrows in sandbanks (Serventy and distribution of territorial species(Lack 1968). Whittell 1976). On the other hand, another The spatial distribution of resources such as burrower, the Banded Sand Martin (Riparia food, nest sites, and nest materials; the intra- cincta),nests solitarily (McLachlan and Liv- specificcompetition for theseresources and for ersidge 1965) and another mud-nest builder, mates; and predation pressureare some of the the Barn Swallow (Hit-undorustica), nests sol- factors determining the breeding dispersion of itarily or in loose colonies (Snapp 1976). individuals (Crook 1965, Lack 1968, Hoog- Those speciesthat must nest in existingholes land and Sherman 1976). and cavities have distributions that are deter- In this paper, we report on an experimental mined primarily by the availability of suitable approach to determine the preferred nesting nest sites(von Haartman 1957, Holroyd 1975). dispersionof a cavity-nesting species,the Tree The Tree Swallow, Purple Martin (Prognesub- Swallow (Tachycinetabicolor). Swallows (Hir- is ), and Northern Rough-winged Swallow undinidae) show varying degrees of gregari- (Stelgidopteryxserripennis) are examples of ousnessduring the breeding season. Disper- obligate cavity-nesting swallows. Cavity-nest- sion may depend upon a particular species’ ing species show considerable intraspecific nestingrequirements. Some speciesbuild their variation in their nesting behavior, and may own nests, while others use existing crevices nest either solitarily or colonially, depending in trees, rocks, and walls, or old burrows ex- on the distribution of nest sites (Bent 1942). cavated by other animals. Since spacingpatterns are so diverse in cavity- Those speciesthat build their own nestshave nesters,it is of interest to determine which, if some choice in the nature of their association any, nesting dispersion pattern is actually pre- with conspecifics, and intraspecific interac- ferred. Nest spacingpreferences of cavity-nest- tions are probably the most important deter- ing species can be determined by providing minants of their dispersion patterns (e.g., many nest sites with different distances be- Hoogland and Sherman 1976, Snapp 1976). tween them. Once the preferred spacing is For example, Cave Swallows (Hirundo @vu) known, one can begin to look for the behav- and Cliff Swallows(H. pyrrhonota),which build ioral mechanismsby which the spacingpattern globular nests of mud pellets, breed in dense is achieved, and for any effects of the spacing colonies, as do Bank Swallows (Riparia ripar- pattern on fitness.

[3561 TREE SWALLOW NEST SPACING 357

16m 8m

FIGURE 1. The arrangement of nest boxes in a spiral cell. The interbox distancesare drawn to scale.

We sought to examine the preferred nest spacing of Tree Swallows when their choice was not constrained by the availability of nest sites.Tree Swallows were chosenbecause they commonly breed in Ontario and readily nest in nest boxes; thus, it was easy to manipulate the spacingof available nest sites for a sizable FIGURE 2. Map of the study area showing the three population of swallows. We asked the ques- hayfields with the grid of spiral cells. tions: (1) Do swallowstend to clump together, spaceout uniformly, or settle randomly in the available nest sites when they arrive at the The nest boxes were checked early in the beginning of the season?(2) Over what range afternoon at 1-3-day intervals from mid-April of distances do swallows exhibit any spacing to mid-July each year, to determine the order behavior? and, (3) Are the nesting attempts of in which they were occupied. Since it was usu- neighboringswallows synchronized or are they ally difficult to be certain of the date on which temporally spacedas far apart as possible?We pairs settled in boxes, we used the date when answered these questions by setting up a reg- the first eggwas laid as an index of the settling ular arrangement of nest boxes with a wide date. From our available data, date of the first range of interbox distances,and recording the egg was highly correlated with the date when order and positions in which swallows settled nest material first appeared in the box, which during the breeding season. we assumeto be close to the settling date (Y = 0.677, y1= 41, P < 0.01). First egg date is METHODS therefore a reasonable indicator of the order The study was done at the Queen’s University in which different pairs settled. Biological Station, Chaffey’s Locks, Ontario, The spiral arrangement of nest boxes in the during the summersof 1977 to 198 1, inclusive. three fields enabled us to look at Tree Swallow The study site consistedof three adjacent hay- settling patterns both within single spirals, fields of different sizes but of similar vegeta- where potential nearest-neighbor distances tion. Nest boxes mounted on 1.75-m high poles were relatively short (l-l 9 m), and also be- were erected in a grid of spirals, each spiral tween different spirals,where potential nearest consisting of six boxes placed 1, 2, 4, 8, and neighbors were relatively remote (36-92 m). 16 m from a central box (Fig. 1). This arrange- Within a spiral, there were thirteen different ment provided a variety of distancesbetween permutations of interbox distance. Depending boxes, both within a single spiral and between on which box was chosen by the first pair of spirals. The entrance holes on all the boxes swallowssettling in a spiral, a secondpair nest- were oriented in the same direction so this did ing in an occupied spiral had a choice of five not affect their desirability. of thirteen possible nest-box distances from One of the fields had eight spiralsin a regular the first pair. The settling behavior between grid, with the central boxesof each spiral spaced and within spirals was considered separately. 64 m from neighboring spirals.The two small- Non-parametric statisticswere used to ana- er fields had two spirals each, with the central lyze the data (Sokal and Rohlf 1981). For the boxes again 64 m apart (Fig. 2). The distance analysesof dispersion patterns, we calculated between boxes in neighboring spirals was in a coefficient of dispersion (CD = variance/ the range of 36-92 m, which greatly exceeded mean). The difference between the calculated the distance between boxes in a single spiral CD value and 1.0 indicates whether the birds (1-19 m). were distributed uniformly (CD < l), ran- 358 A. MULDAL, H. L. GIBBS AND R. J. ROBERTSON

TABLE l(a). The number of times Tree Swallows chose nest boxes before the Tree Swallows arrived between empty spirals and spirals occupied by bluebirds. in the spring (1977, 1978-one pair; 1979, For each spiral category, the lefthand column gives the 1980-three pairs; 1981-four pairs). Blue- number chosen (0) and the righthand column gives the expected frequency (E) based on the availability of each birds aggressivelycompete with Tree Swallows categoryat each stageof the settling sequence(see text for for nest sites when these are in short supply details). l(b). The number of times swallows chose be- (Kuerzi 194 1). Although boxeswere abundant, tween empty spirals and spirals occupiedby conspecifics. (on average, only 3% were already occupied by bluebirds when the swallowsarrived in the (a) spring), it was important to ascertain first Empty Bluebird Year 0 E 0 E whether swallowsavoided nesting in the same spiralsas bluebirds, or whether they “ignored” 1971 5 5.35 0.65 bluebirds. 1978 11 8.90 :, 2.10 1979 4 5.62 3 1.38 As settling progressed,the number of empty 1980 9 7.69 2 3.31 spirals and spirals containing swallows 1981 6 6.75 4 3.25 changed.Each time a new pair (or pairs) settled Total 35 34.31 10 10.69 in the spirals, we computed a posteriorithe (b) number of spirals available that were either Empty StWllOW empty or occupied by Tree Swallows or blue- Year 0 E 0 E birds, and the number of empty and occupied 1977 10 5.44 2 6.56 spirals that were actually chosen by swallows. 1978 10 5.45 1 5.55 The expected number of times that empty or 1979 7 4.14 3.86 occupied spirals would be chosen randomly 1980 4.00 :, 4.00 1981 : 3.66 1 3.34 was calculated by summing the probability of Total 41 22.69 5 23.3 1 each category of spiral (i.e., empty, occupied by swallows, or by bluebirds) that was chosen at each stage in the settling sequence.Devia- domly (CD = l), or in a clumped fashion tions of the swallows’ observed choices from (CD > 1). Data were tested against a positive the expected values were then compared to a binomial (uniform) distribution using a Chi- standard normal distribution using the Z sta- squared test for goodness-of-fit (Elliot 1973). tistic (Sokal and Rohlf 198 1). Since the probability of swallows choosing spirals in each RESULTS category changed continually as more swal- SETTLING PATTERN BETWEEN SPIRALS lows settled, Z was calculated using the for- mula: The occupancyrate of all of the nest boxes was relatively low each year (1977-35%; 1978- observed-expected choice 24%; 1979-24%; 1980-22%; 1981-29%), sug- Z= gesting that nest sites were plentiful for this product of probabilities population of swallows. c for each spiral category, To test whether Tree Swallows preferred particular spirals in the grid, we used a Krus- kal-Wallis ANOVA test on the annual order which takes into account the changing vari- of settling in the twelve spirals.Since swallows ance. did not tend to choose spiralsin any particular When swallowshad a choice between empty order each year (H = 5.58, P = 0.899), we spirals and spirals occupied by bluebirds, 10 concluded that all spirals in the grid were of 45 (22%) settled in spiralsoccupied by blue- equally attractive during the settling sequence. birds (Table la). This result is not significantly If the swallowspreferred to clump their nests, different from a random settling pattern when we would have expected to find two or more compared with empty spirals (Z = 0.25, P = pairs settling in the same spiral, even when 0.40). Although there were more than three empty spirals were still available. If the birds times as many spirals available that were oc- spacedthemselves perfectly uniformly, each of cupied by other swallows than by bluebirds, the twelve spirals should have been occupied only five of 46 (11%) swallows chose spirals by a singlepair of swallowsbefore any of them occupied by conspecifics(Table lb). This dif- was occupied by a second pair. By examining fered markedly from a random settling pattern the sequenceof settlingin the spirals,we tested (Z = 6.20, P K 0.001). Therefore, Tree Swal- the hypothesis that swallows chose randomly lows spaced themselves with respect to con- among the available boxes. specificsuntil all the spirals were occupied by At our study site, severai pairs of Eastern a single pair of swallows, but they settled ran- Bluebirds (Sialia sialis)became establishedin domly in spirals occupied by bluebirds. TREE SWALLOW NEST SPACING 359

TABLE 2. Annual frequency distributions of the maximum number of nest boxes occupied concurrently per spiral.

Frequencyof concurrentlyoccupied boxes per spiral Year 0 1 2 3 4 5 6 MWI CD x” P

1977 2 8 0 0 0 2.00 0.18 2.00 co.005 1978 10 2 0 0 0 1.67 0.13 1.oo co.005 1979 1.33 0.32 3.50 co.025 1980 0” ; : : 0” : 0” 1.25 0.16 1.80 co.005 1981 0 6 4 2 0 0 0 1.67 0.36 4.00 co.05 * Relative to a positive binomial distributmn

The coefficient of dispersion for concur- of times close and far spirals would be ran- rently nesting swallows was significantly less domly chosen was calculated by the method than 1.O in each year, which also indicates that described earlier. The birds’ observed choices the final distribution of concurrently nesting were then compared with these expected val- pairs among spirals was uniform (Table 2). ues. To see whether the swallows preferred to The resultssuggest a random settlingpattern spacethemselves at relatively great distances, (Table 3). Tree Swallows neither preferred nor we tested the hypothesis that, during the set- avoided settling in a spiral that was close to tling sequencein the grid, birds would choose another occupied spiral (Z = 0.75, P = 0.23) the empty spiral closest to a spiral occupied over distancesin the range of 36-92 m. by a conspecific at a random-level frequency. Hereafter, only spiralsoccupied by Tree Swal- SETTLING PATTERN WITHIN SPIRALS lows are referred to as “occupied.” Since the To seewhether swallowstended to spacetheir central boxes of each spiral in a particular field nestsaway from relatively close neighbors, we were evenly spacedin a grid at 64-m intervals, looked at which boxes were chosenby the sec- and distancesbetween boxes within spiralswere ond pairs to settle in occupied spirals. We list- much smaller than distances between spirals ed the nest-box distancesavailable to the sec- (1-19 m vs. 36-92 m, respectively), a spiral ond pair, given the position in which the first was defined as “close” to an occupied site if it pair settled, and then compared this with the was an immediate neighbor to an occupied distance actually chosen by the second pair. spiral, and “far away” if it was two or more The results in Table 4 show that the second spirals from the nearest occupied site. Spirals pairs of swallowsto nest in spiralssignificantly in different fields were consideredfar from one more often chose the box farthest from the another. From the first egg dates, we deter- original occupants than any other box in the mined a posteriori how many spirals close to spiral (x2 = 22.10, P < 0.001). In the three and far from occupied spirals were available caseswhere the closestbox was chosen, it was each time a new pair, or pairs, settled, and then at distancesof 14, 14, and 16 m, respectively. compared this with the number of close and Tree Swallows chose proportionately more of far sitesactually chosenby Tree Swallows.This the available boxes at greater distances from procedilre was done for each stepin the settling the resident pair (Fig. 3). Within spirals,there- sequencewhere swallowshad a choice between fore, Tree Swallowstended to spacetheir nests “close” and “far” sites. The expected number away from their nearest neighbors.

TABLE 4. The number of times second pairs of Tree TABLE 3. Frequency with which Tree Swallows chose Swallows that settled in the spirals chose the box closest spirals “close” to, and “far” from, occupied spirals. Left- to, or farthest from, the box occupiedby the resident pair, hand columnsgive number of closeand far spiralsactually or any other box in the spiral. For each category of box, chosen, righthand columns give expected choices based the lefthand columns give the observed (0) and the right- on the availability of close and far spirals at each step in hand columns give the expected (E) frequencies. the settling sequence(see text for details). Number of times box chosen ClOX Far Closest Farthest Other Year 0 E 0 E Year 0 E 0 E 0 E

1977 2 3.37 4 2.63 1977 1 1.8 5 1.8 5.4 1978 3 3.31 3 2.69 1978 0 1.0 2 1.0 3.0 1979 4 4.72 : 2.28 1979 1 k?: 1 0.4 0 1.2 1980 4 3.58 3.42 1980 0 2 0.6 1 1.8 1981 4 3.90 2 2.10 1981 1 l:o 4 1.0 0 3.0 Total 17 18.88 15 13.12 Total 3 4.8 14 4.8 7 14.4 360 A. MULDAL, H. L. GIBBS AND R. J. ROBERTSON

2L ? . . . . : ...... :: . -1016 . . . . .

. . . .

i, B 1'6 it $2 Time interval (days)

I I I I 1 FIGURE 4. The relationship of the time interval be- 0 20 tween two pairs of swallows nesting in the same spiral to DistancSefrom original :ir (m) the distance between the two nests. Data from five years for 27 spirals occupied concurrently by two pairs. Spear- FIGURE 3. The proportion of available boxes chosen man’s rank correlation r = 0.375, P -c 0.05. by secondpairs of swallows nesting in spirals, in relation to the distance of the chosen boxes from the original occupants of the spirals. Spearman’s rank correlation r = 0.637. P < 0.02. nestsat closedistances. In studiesat other sites, Tree Swallows nested much closer than we observed in our study. They nested in boxes 2- TEMPORAL SPACING 4 m apart in Massachusetts(Whittle 1926), The first swallows to nest in the spirals each 15.2 m apart in Connecticut (Kuerzi 1941) year were highly synchronized with one and 10 m apart in New York (Sheppard 1977). another, with 68% of the birds (? 1 SD from Harris (1979), in New Brunswick, noted four the mean) laying their first eggswithin three pairs nesting in a single spiral similar to ours days of the mean for the population (data from and also documented one successfulnesting in five years combined). The secondpairs to nest a pair of boxes only 1 m apart. The temporal in occupied spirals, however, laid their first separationof the nests,however, was not men- eggs,on average, 13 days later than the first tioned in any of these studies; it is therefore pairs. Nests within spirals therefore tended to possiblethat although the nestswere close to- be temporally spaced out from one another. gether in distance, they may have been widely Within a spiral, the closer together the nests separated in time. were in time, the farther the secondpair ’s nest The very high occupancy rates documented was from the first (Fig. 4). From inspection of in some of these earlier studies suggestthat the figure, it seemedthat behavior might switch population pressuremay have resulted in clos- as the interval between first and second nests er spacingof nests.The occupancyrate for our increased, from choosing only boxes farthest population was relatively low (on average, only from the nearest neighbor, to choosing boxes 45% of the boxes 4 m or more apart were randomly. Dividing the data in half, we found occupied). With high occupancy rates, it is that swallows that started nests less than 14 likely that the availability of nest sites deter- days after the first pair in the spiral were sig- mines the observed spacing pattern. We can nificantly more likely to choose the box far- be fairly confident, however, that nest siteswere thest from the first pair than any other box in not limiting in our study site, and that we ob- the spiral (x2 = 22.72, P < O.OOl), while the served the preferred spacing pattern of Tree choice of nests initiated more than 14 days Swallows in the spirals. after the first pair had settled was not signifi- The spacingpattern we found may have aris- cantly different from random (x2 = 3.56, P > en either from avoidance of occupied boxes 0.1). by arriving swallows, or from aggressionby resident swallows,preventing others from set- DISCUSSION tling close by. It is difficult to distinguish be- Although Tree Swallows have been reported tween these two possibilities from our data. to nest colonially (Whittle 1926, Sheppard The spacing pattern at the beginning of the 1977), those in this Ontario population did not breeding season,when arriving swallowstend- choose to nest close together when there were ed to settle in empty spirals while these were abundant suitable nest sites. Our results dem- available, may have arisen as a result of avoid- onstrate both spatial and temporal spacingof ance behavior by the arriving swallows. It is TREE SWALLOW NEST SPACING 361 likely, however, that the spacingpattern later breeding cycle has also been reported for Pur- in the season, when second pairs settled in ple Martins, which usually defend more than spiralsas far as possiblefrom the resident pair, one room of a martin house at the start of the resulted from the aggressivebehavior of the breeding season, but defend fewer rooms as resident pair. the breeding seasonprogresses (Brown 1979). That Tree Swallows are aggressivetowards We can offer several alternative hypotheses conspecificshas been reported by several au- for the function of territoriality in Tree Swal- thors (Kuerzi 194 1; Harris 1979; Leffelaar and lows: (1) Tree Swallows might prevent other Robertson 1985). In an experiment using individuals from nestingclose to them because stuffed models of Tree Swallows, Robertson there is some fitness-related disadvantage to and Gibbs (1982) showed that resident Tree nesting at high densities; for example, in- Swallows were aggressiveat distancesof up to creased competition for food or other neces- 8 m from their nest box, and that aggression sary resources,higher risk of cuckoldry, pre- was centered around the box. They concluded dation, or parasitism (e.g., Davies 1978). (2) from the model tests that one function of in- Tree Swallowsmight defend territories, includ- traspecific aggressionwas the territorial de- ing boxes additional to that required for nest- fense of a nest site from usurpation by con- ing, either in order to (a) enable males to be specifics, and that Tree Swallows defended polygamous (Brown 1979), or (b) prevent oth- symmetrical territories of 6-8 m radius around er birds in the population from breeding, thus their nest sites. These results suggest that increasing the resident’s relative genetic con- aggressionis an important behavioral mech- tribution to future generations (Harris 1979). anism that determines nest spacing patterns, (3) Tree Swallows might defend territories as although avoidance may be operating as well. the most efficient means of defending a nest When Tree Swallows first settled in the grid, site from intruders (Hinde 1956). each spiral of six boxes was defended by a Tree Swallows do not defend feeding terri- single pair, and an average of 13 days elapsed tories, sincethey feed almost entirely on highly before a second pair settled into any of the mobile, aerial insects,a resourcewhich would spirals.We can offer two possibleexplanations be difficult to defend (Kuerzi 194 1, Sheppard for this temporal spacing.(1) It may be a func- 1977). Thus, we would not expect that nest tion of the pattern of arrival of birds into the spacing would be a factor influencing fitness area. Second-year brown female Tree Swal- through competition for food. Although little lows are known to arrive in their breedingareas predation existed in our population, there may 2-3 weeks later than blue-green adults (Kuerzi have been a density-related effect on fitness of 1941). At our study site, however, only four blowfly larval (Protocalliphoru sp.) parasitism or five second-year females settled in the grid of nestlings,which was sometimes severe. each year; therefore, a late wave of settling Brown (1979) suggestedthat male Purple second-yearfemales can only partially account Martins defended more than one room of a for the observed 13-day temporal spacingbe- martin house early in the breeding seasonin tween the first and second pairs nesting in the order to have the potential to mate polygy- spirals. (2) Resident swallows might actively nously. Five percent of males in his population prevent other pairs from settling in the spiral were polygynous. Quinney (1983) found that for approximately 13 days, but then might re- 5% of male Tree Swallows at a site with rela- lax their defense as they become involved in tively abundant food were polygynous. Polyg- incubation and the feeding of nestlings,so that yny in our population was rare, however, and the defended area becomes smaller with time. sincein most reported casespolygyny involved Behavioral observations of breeding swallows two females nesting in the same box (Quinney indicate that aggressiondecreases as the breed- 1983; this study), it is unlikely that males in ing seasonprogresses. Gibbs (1980) found that our population defended extra boxes to obtain Tree Swallows were most aggressivetowards additional mates. conspecificswhen nest siteswere being chosen, Harris (1979) suggestedthat Tree Swallows before nest-building began. A pair’s aggressive defended territories containing extra boxes in responsedropped off during the nest-building order to prevent other individuals in the pop- period, and declined further as the breeding ulation from breeding. This is the “superter- cycle progressed.Stutchbury (1984) also found ritory” hypothesis of Verner (1977). Robert- that aggressive responses of breeding Tree son and Gibbs (1982) showed in their study Swallows towards models of conspecificsde- site, however, that territorial defense was not creasedas the nestingcycle progressed.Aggres- oriented towards additional nest boxes and, sion between Eastern Bluebirds similarly de- furthermore, relative territory sizewas the same clines during the breeding season (Gowaty for solitary-nesting birds as for those that had 198 1). A decreasein territory size during the the potential to defend extra boxes. 362 A. MULDAL. H. L. GIBBS AND R. J. ROBERTSON

We suggestinstead that the most likely ul- and Philip Stoddard, Edward Burtt and members of the timate reasonsfor Tree Swallow territoriality “Bird Lab” for helpful comments on the manuscript.Field facilities and accommodation were provided by Queen’s are the defense of the nest site against com- University Biological Station. This study was funded by petitors, which may be achieved most effec- a grant to RJR from the Natural Sciencesand Engineering tively by defending a certain minimum space ResearchCouncil of Canada. around the nest site (Hinde 1956), and the avoidance of cuckoldry by males. The reported LITERATURE CITED decline in aggressionduring the breeding sea- BENT, A. C. 1942. Life histories of North American fly- son (Gibbs 1980, Stutchbury 1984) supports catchers,larks, swallows, and their allies. U.S. Natl. Mus. Bull. 179. the hypothesisthat aggressionbetween females BROWN,C. R. 1979. Territoriality in the Purple Martin. is directed towards protection of the nest site Wilson Bull. 91:583-591. from intruders, rather than toward defense of CROOK,J. H. 1965. The adaptive significanceof avian food or againstpredators (Gowaty 198 1). Since social organization. Symp. Zool. Sot. Lond. 14:18 l- male Tree Swallows do not guard their mates 218. DAVIES,N. B. 1978. Ecological questions about terri- during their fertile periods (Leffelaar and Rob- torial behaviour, pp. 3 17-350. In J. R. Krebs and N. ertson 1984), aggressionbetween males near B. Davies [eds.], Behavioural ecology:an evolution- the nest box may serve to reduce the risk of ary approach. Blackwell Scientific Publications, Ox- cuckoldry. The decline in aggressionof males ford. during the breeding seasonis consistent with DYRCZ, A. 1984. Breeding biology of the Mangrove Swallow (Tachycinefaalbilinea) and the Gray-breast- protection of paternity and also with defense ed Martin (Progne chalybeu)at Barro Colorado Is- of the nest site against conspecific intruders. land. Panama. Ibis 126:59-66. Both members of the pair also vigorously de- ELLIOT,J. M. 1973. Some methods for the statistical fend their nest site from other cavity-nesting analysis of samples of benthic invertebrates. Fresh- water Biol. Assoc. Sci. Publ. 25. species, such as bluebirds, Purple Martins, GIBBS,H. L. 1980. Spacing behavior of nesting Tree House Sparrows (Passer domesticus), and Swallows(Zridoprocne bicolor): the dispersionpattern House Wrens (Troglodytes aedon; Kuerzi and its determinants. B.Sc. thesis, Queen’s Univer- 1941). sity, Kingston, Ontario. GOWATY, P. A. 1981. Aggression of breeding Eastern Defense of an area around the nest site has Bluebirds(Siulia siulis)toward their matesand models been reported for other cavity-nesting species. of intra- and interspecificintruders. Anim. Behav. 29: Pied Flycatchers (Ficedula hypoleuca) defend 1013-1027. an area of radius O-10 m around the nest site, HARRIS,R. W. 1979. Aggression, superterritories and and the only function of this defense is to pro- reproductive successin Tree Swallows. Can. J. Zool. 57:2072-2078. tect ownership of the nest site (von Haartman HINDE, R. A. 1956. The biological significanceof terri- 1956). Brown-chested Martins (Phaeoprogne toriality in birds. Ibis 98:340-369. taperu) in Venezuela fight at the entrances to HOLROYD,G. L. 1975. Nest site availability as a factor nest holes when more than one nest is present limiting population size of swallows.Can. Field-Nat. in the vicinity (Turner 1984). They are terri- 89~60-64. HOOGLAND,J. L., AND P. W. SHERMAN. 1976. Advan- torial, nesting singly or in small groupsin dead tages and disadvantagesof Bank Swallow (Ripariu trees and in holes in bridges. In Panama, Man- riparia) coloniality. Ecol. Monogr. 46:33-58. grove Swallows (Tachycineta albilinea) and KUERZI, R. G. 1941. Life history studies of the Tree Gray-breasted Martins (Progne chalybea) dis- Swallow. Proc. Linn. Sot. N. Y. 52-53: l-52. LACK, D. 1968. Ecological adaptation for breeding in perse in nest boxes, with only single pairs oc- birds. Methuen and Co., London. cupying groups of boxes (Dyrcz 1984). The LEFFELAAR,D., AND R. J. ROBERTSON.1984. Do male distance between neighboring pairs is always Tree Swallows guard their mates? Behav. Ecol. So- several hundred meters. Mangrove Swallow ciobiol. 16:73-79. LEFFELAAR,D., ANDR. J. ROBERTSON.1985. Nest usur- territories, unlike those of Tree Swallows, are pation and female competition for breeding oppor- also used as foraging areas, there being only a tunities in the Tree Swallow. Wilson Bull. 97:221- small area of overlap of feeding activity be- 224. tween neighboring territories. MCLACHLAN,G. R., AND R. LIVERSIDGE.1965. Roberts Our study demonstratesthat Tree Swallows, birds of South Africa. Trustees of the South African Bird Book Fund. when given the choice, space their nests both QUINNEY,T. E. 1983. Tree Swallows cross a polygyny spatially and temporally. This seemsto be re- threshold. Auk 100:750-754. lated to defense of nest sites, which are often ROBERTSON,R. J., AND H. L. GIBBS. 1982. Superteni- limiting for obligate cavity-nesters (von Haart- toriality in Tree Swallows: a reexamination. Condor man 1957, Holroyd 1975). 84:313-316. SERVENTY,D. L., AND H. M. WHITTELL. 1976. Birds of Western Australia. Univ. of Western Australia Press. ACKNOWLEDGMENTS Perth. SHEPPARD,C. D. 1977. Breeding in the Tree Swallow We are grateful for the field assistanceof ShelaghHurley (Zridovrocnebicolor) and its imolications for the evo- Mathers in 1980 and Lesley Cameron in 1981. We thank lution of coloniality. Ph.D. diss., Cornell Univ., Ith- Malcolm Griffin and Pete Blancher for help with statistics, aca, NY. TREE SWALLOW NEST SPACING 363

SNAPP,B. D. 1976. Colonial breeding in the Barn Swal- VON HAARTMAN,L. 1956. Territory in the Pied Flycatch- low (Hirundo rustica) and its adaptive significance. er, Muscicapa hypolecua.Ibis 98:460-475. Condor 78:471-480. VON HAARTMAN,L. 1957. Adaptation in hole-nesting SOKAL,R. R., AND F. J. ROHLF. 1981. Biometry. The birds. Evolution 11~339-347. principles and practice of statistics in biological re- WHITTLE,C. L. 1926. Notes on the nestinghabits of Tree search. W. H. Freeman and Co.. San Francisco. Swallows. Auk 43:247-248. STUTCHBURY,B. J. 1984. The adaptive significanceof delayed plumagematuration in female Tree Swallows (Tachycineta bicolor). B.Sc. thesis, Queen’s Univ., Kingston, Ontario. TURNER,A. K. 1984. Nestingand feedinghabits ofBrown- Biology Department, QueenS University,Kingston, On- chested Martins in relation to weather conditions. tario K7L 3N6. Addressof secondauthor: Devartment of Condor 86:30-35. Ecology and Evolutionary Biology, Universi& of Michi- VERNER,J. 1977. On the adaptive significanceof terri- gun, Ann Arbor, Michigan 48109-1048. Received 19 July toriality. Am. Nat. 111:769-775. 1984. Final acceptance23 February 1985.