condor, 82: 170-175 @ The Cooper Ornithological Society 1980

SLEEPING BEHAVIOR OF PURPLE MARTINS

CHARLES R. BROWN

ABSTRACT.-1 studied the behavior of Purple Martins (Progne subis) at nightfall and the birds ’ sleeping arrangements, each night, from spring arrival until premigratory flocking began. Martins slept in martin houses until about 15 June, after which they commonly slept in trees. Birds that were firmly established on a territory slept in a room of that territory. Some pairs of birds slept together in the same room and others did not. This behavior was de- termined by the males. Pairs that occupied their territories for a week or longer often slept together. While building nests, pairs began to prefer certain rooms for sleeping. Dur- ing egg laying and incubation, all females slept in the nest. They slept with the young and probably brooded them at night until the nestlings were 13- 15 days old. Some females ceased sleeping with the nestlings after two weeks, and these females either slept in a tree or resumed sleeping with their mates. Many vagrant martins slept in martin houses, often on occupied territories. Martins frequently called from 03:OO until daybreak. I hypothesize that Purple Martins copulate in martin house rooms at night, probably in the early morning. Pairs ’ sleeping together facilitates mating. Nocturnal copulation benefits males by minimizing cuckoldry and neighbor interference. Females benefit by being less vulnerable to gang rapes. Pairs also may sleep together to conserve heat on cool spring nights.

Despite recent studies of Purple Martins 3 in 1969-1970,8 in 1971-1972, 19 in 1973,20 in 1974, (Progne subis), primarily by Finlay (1971, 25 in 1975, and 35 in 1976. A total of 123 pairs (246 1976), Niles (1972), Brown (1978a-e, 1979), individuals) was observed. Although some birds re- turned year after year, to simplify procedures I treated and E. J. Bitterbaum (pers. comm.), and ear- all birds each year as different individuals; thus the lier work by Allen and Nice (1952), John- reported total is slightly inflated. ston and Hardy (1962) and Johnston (1966), All individuals were recognized either by painted little attention has been paid to sleeping aluminum bands, colored plastic leg bands, behavior characteristics, or plumage features. Approximately behavior of martins after spring arrival and half of the birds each year were banded or color- before premigratory flocking. In fact, sleep- marked, and I could also recognize some individuals ing behavior has not been well studied in by their slight habits which became evident after care- any North American swallow. ful observation. These traits included establishing A consideration of sleeping patterns in ownership of certain perches on which no other birds were allowed, specific routes to and from the nesting Purple Martins is important for three rea- houses, unusually strong aggression toward other mar- sons. First, since martins are thought to cop- tins in the colony and toward other species that came ulate within the nesting cavity (Brown near, and individual patterns of the Claiming-Reclaim- 1978a), any sleeping arrangement that uses ing (Johnston and Hardy 1962) display which, for many males early in the season, were uniquely characteristic. this chamber could facilitate copulation. Plumage differences were useful in identifying first- Copulation within the cavity may, among year males because the purple mottling on the head, other functions, minimize male cuckoldry. breast, and belly is variable. Most females were also Second, sleeping patterns are reliable in- recognizable by the shading and patterning of gray on dicators of an individuals’ colony residency, the head, breast, and belly. My techniques for captur- ing adults had a low yield and I could not capture all and thus are useful in censuses of Purple birds for banding. Only those that were individually Martins. Third, additional knowledge of be- recognizable were included in the study. havior in the genus Progne may be helpful After the martins arrived and became established at in defining phylogenetic relationships the colony, I observed each pair daily and recorded sleeping behavior throughout the season. Toward the among member species. I present here an end of the season as the colony grew (especially in investigation of sleeping behavior in Purple 1975-1976), it became difficult to observe the behavior Martins. of each pair every evening, but all birds were observed at least every other evening. STUDY SITE AND METHODS During 1968-76, I noted sleeping behavior on 1,392 nights (24,324 bird-nights). By month, there were 60 I made all observations at a Purple Martin colony in nights in February, 262 in March, 253 in April, 242 in Sherman, Grayson Co., north Texas, each year from May, 270 in June, 265 in July, and 40 in August. I 1968 to 1976. The colony grew from 2 pairs in 1968, to distinguished five major stages during the nesting

[I701 MARTIN SLEEPING BEHAVIOR 171 cycle: I-arrival and pair formation, II-nest building, A residents’ bond to its territory was III-egg laying and incubation, IV-feeding young, tested at sleeping time, much as Johnston and V-post-breeding nest defense. During these and Hardy (1962) suggested for pair-bonds. stages, there were 4,876,4,744,4,910, 7,786, and 2,008 bird-nights of observation, respectively. The stages If a prospective resident failed to appear overlapped among pairs, reflecting dissimilar spring and sleep on a territory, that martin was not arrival times and low synchronization of colony activ- established there (though it may have ap- ities. peared to be established earlier in the day). Beginning about 1 April each year, I checked each Of birds that became established as breed- nest in the colony every other day to determine laying dates, clutch size, incubation period, egg and young ers and for those whose exact day of arrival losses, and hatching and fledging dates. These data at the colony was known (N = 202), 80.7% were important in relating sleeping behavior to stages slept their first night at the colony on the of the nesting cycle. Aluminum martin houses were territory which they were to hold that year. lowered on telescoping poles and doorpanels were raised to note nest contents; wooden martin houses Most of the remaining birds slept as va- were reached with ladders and nest contents were grants on various territories in the colony checked with a small dental mirror inserted through until they found a territory of their own. the entrance hole (see Brown 1978b). Both males and females followed these pat- All-purple male martins (i.e. at least two years old) were termed “adults,” and one-year-old males with terns. mottled purple feathering on the head, breast, and bel- The sleeping behavior of a female indi- ly were termed “yearlings.” Yearling females were dis- cated the strength of a pair-bond. Some fe- tinguished from adult females primarily by banding males visited several unmated males in the records; I assumed that very pale-breasted and pale- colony during the mornings, yet they typi- bellied females with little purple on the mantle who arrived late in the season were yearlings. cally slept in the room with their “favored” Because the term “roost” for the Purple Martin often male. This bird was likely to be chosen by connotes premigratory flocking (which is well docu- the female as her mate. Sixty-five females mented and not considered here), I use the term that had a choice of two or more males slept “sleep” in this paper. It might be incorrect to assume that martins in their nesting compartments sleep with the male they later chose as their mate throughout the night, but I think “sleep” is preferable on 71% of the nights. Some females associ- to ambiguous and awkward terms such as “house- ated with two or more males for several roosting.” The birds were not active enough after dark days, sleeping with a different one each to speak of “nocturnal behavior” (as in owls). I was night. These females did not appear to favor concerned with the birds ’ behavior at sleeping time and with their physical arrangements for spending any one male, although they later left all each night, rather than with the physiological state of except one. Rarely, a female slept with one sleep. In this paper, whenever I state that a martin male for several weeks and then left him, “slept” in a room, I mean only that the bird spent the but once established, pairs of martins were entire night in that room and did not come out until daybreak. remarkably stable. Some males claimed only one room, and females shared it with them. Most males, RESULTS however, claimed more than one room I. ARRIVAL AND PAIR FORMATION (Brown 1979), and such pairs slept either Upon arrival at a colony, Purple Martins es- together or in separate rooms. Pairs that tablished territories consisting of one or were established for less than a week were more martin house compartments. The more likely to sleep apart than were pairs birds slept in these rooms, behavior that I that had been together for more than a found to be a useful criterion for indicating week. During the first week a pair was pres- whether a martin was established on a ter- ent at the study colony (N = 123), 61% slept ritory. Residents slept in a room on their together on four or more nights, while 39% territories 98.8% of the bird-nights during slept separately. After the first week of re- Stage I. sidency, the respective percentages of pairs Resident birds behaved differently from sleeping together and separately during nonresidents, who also slept in martin Stage I were 78.1% and 21.9%. Although houses. Residents, after foraging away from sleeping together was more prevalent than the colony during the afternoons, returned sleeping separately, I found no correlation to the colony about an hour before twilight. between a pairs’ sleeping arrangement and At that time they sat and preened on their reproductive success (Brown, unpubl. data). territories or on wires, or they foraged in the It appeared that males mostly determined immediate vicinity. Nonresidents behaved whether pairs slept together or separately. hesitantly; typically they arrived near twi- When martins returned to their respective light and darted into rooms with none of the rooms at twilight, a male often entered and “leisure” activities which residents en- remained briefly in several rooms of his ter- gaged in before sleeping. ritory. His female followed him into each 172 CHARLES R. BROWN room. If the pair were to share a room, the where the nest was or where the nest was male either remained in the room or came to be built. out and then re-entered the room. If they were to sleep separately, the male went into III. EGG LAYING AND INCUBATION another room. The female often followed After laying their first egg, female martins him into this room, but then he exited again. slept in the nest room each night. I never A pair frequently moved between rooms for recorded a female sleeping anywhere other five minutes or more until the female than the nest once egg laying began. At that ceased following, and they slept separately. time most males (61.6%), if accustomed to Occasionally in newly established pairs, the sleeping with their mate prior to egg laying, female flew away and did not sleep at the continued to sleep with her then and during colony after the male repeatedly refused to incubation. The fact that a pair spent the sleep with her. She was likely to return on night in the nest room during egg laying following days, and the procedure was often probably was responsible for slightly stag- repeated. Inclination to sleep with a female gered hatching of martin clutches, indicat- seemed to vary among males, but after a pair ing that some embryonic development oc- had been established for six or more days curred in the heated nest at night before the male invariably was willing to sleep incubation began (Allen and Nice 1952). with the female. Other males readily slept The remaining males (38.4%) regularly with a female the first day she arrived. Al- slept in other rooms of the territory during though most males apparently were willing incubation. to sleep with their mates after six days of residency, they sometimes still slept sepa- IV. FEEDING YOUNG rately, especially after egg laying began. In Stage IV established sleeping patterns Some pairs holding large territories (e.g. began to dissolve. All female martins slept eight rooms or more) were so preoccupied in the nest with their young until the nest- with territorial defense at sleeping time that lings were two weeks old, after which some they were forced to sleep separately as it females ceased sleeping with them. became dark while they were chasing away Prior to 15 June in Sherman, most females intruders, and the female could not find the (89.2%) with broods over two weeks old male. continued sleeping with the young. After 15 Females seemed to strongly desire to June, females (68.4%), especially yearlings, sleep with males during Stage I and they with broods over two weeks old slept else- took the initiative in this behavior. A female where. Regardless of date, however, all fe- on a large territory often sought out her males slept with the young until they were mate by peering into all rooms until she 13-15 days old, suggesting that young Pur- found him. Most other interactions between ple Martins may require night-brooding for members of a Dair seemed to be initiated bv at least two weeks. (Mv studies show that the male. Sleeping with a male was an ap- nestlings are brooded during the daytime parent means by which females indicated until they are 10 days old.) mate choice, yet some males initially re- Although time of year influenced when a fused to sleep with females. female ceased sleeping with the young, Behavior described for this and following brood size was important also. No quanti- Stages applied equally to adult and yearling tative data are available, but for broods of martins, although yearling females some- five to seven young, female parents seemed times showed less initiative than did adults less likely to sleep with the young after the in seeking to sleep with a male. two-week period than were female parents of smaller broods. In large broods, which II. NEST BUILDING probably raise the temperature inside the Sleeping behavior during this stage was nest slightly, the young may stay warmer similar to that in Stage I. Pairs were well than do young in small broods, thus large established by the time nest building be- broods require less night-brooding. gan, and pairs commonly slept together al- Date and brood size also influenced the though they sometimes slept apart. During sleeping behavior of males. In broods of one this stage pairs showed sleeping prefer- to four young, a male was likely to sleep antes for certain rooms; these were the with the female and the young throughout rooms (or room) where the female had been the four-week nestling period, especially doing most of h er nest construction. Sleep- prior to I5 June. In broods of five to seven ing occurred 79.6% of the nights in rooms young, males frequently ceased sleeping in MARTIN SLEEPING BEHAVIOR 173 the nest even before the young were two on either side of the entrance, but they were weeks old. Generally, if a male slept in the active during the night and moved often. I nest with the female during incubation, he do not think that the light, which was continued to do so for a time after the young flashed from the ground, influenced their hatched. If a male slept separately during behavior. While capturing Purple Martins incubation, he continued to do so after the at night for banding in Gainesville, Florida, young hatched. E. J. Bitterbaum (pers. comm.) found males When parents did not sleep with their resting on top of females; the birds possibly young, they occupied another room of the slept in this position (and copulated?). territory (often a pair resumed sleeping to- Intruders began appearing at the colony gether at that time), especially prior to 15 about an hour before twilight, but they were June. After 15 June parental martins either driven from occupied territories by resi- continued using another room, or they slept dents. As twilight neared, the “homeless” in trees, a prelude to premigratory roosting. vagrants tried vigorously to find a place to Sleeping in trees was infrequent before 15 sleep. They would enter a room very quick- June but rapidly became common after that ly, in contrast to their slow, cautious behav- date. If the parents slept in a tree, they nor- ior during the day. Rapid entry sometimes mally left the colony at twilight and re- enabled them to escape detection by the turned at daybreak to feed the young. Mar- residents, and the vagrants were then able tins from one to six different colonies to sleep in that room. Residents tenaciously repeatedly slept at a communal tree or defended their territories until dark, and grove within one km of the colonies. If a vagrants found sleeping rooms only with pair claimed only one room-the nest-as great difficulty. Vagrants sometimes entered their territory, sometimes they quietly a room containing other vagrants, residents, entered rooms on another pairs’ territory to or House Sparrows (Passer domesticus), and sleep in during Stage IV. the intruders were quickly repulsed. Rarely a male and a female vagrant entered the V. POST-BREEDING NEST DEFENSE same room and slept there together. Va- Martins rarely slept in their houses during grants often flew from one martin house to this stage although pairs spent much time another trying to find a room. I did not find there during the day. Only 6.5% of the res- out where room-less martins slept, but they ident birds slept in the colony during this probably used trees. The vagrants sleeping stage. Martins mainly slept in trees during in rooms departed at daybreak. When the Stage V, and premigratory flocking and study colony was relatively large (1973-76), roosting began then. varying numbers of vagrants used the col- Certain generalities pertain to Stages I- ony each night throughout Stages I-III and IV when martins always slept in their during much of Stage IV. Often at least 15 houses. Only once did I observe a martin vagrants slept at the colony per night. sleep outside on the porch of a house. This All martins in my study seemed reluctant individual was an intruder who was backed to fly or move outside their rooms after dark. into the corner of a porch and was unable I never observed them change rooms during to move because the defending resident sat the night, and whenever martins flew after in the entrance of a room nearby and pecked dark owing to major disturbances (e.g. jar- the intruder vigorously whenever he tried ring or lowering a martin house), they ap- to move. Males in particular often perched peared nervous, disoriented, and uncoordi- in the entrance of a room or on a porch until nated in flight. However, Southerns’ (1959) it was quite dark, but in my study they in- homing experiments showed that Purple variably retired inside the room to sleep. Martins indeed are able to see and success- However, J. C. Finlay (pers. comm.) found fully orient at night. that martins in Alberta, Canada, occasion- Purple Martins characteristically call at ally left their rooms at night, to return at night (Sprunt in Bent 1942, Oberholser daybreak. 1974). The birds at my colony typically be- I installed one martin house with reflect- gan calling about 03:OO and continued until ing interiors so that I could observe the mar- sunrise. Apparently the vocalizations were tins sleeping inside the room in silhouette given while the birds were inside the mar- whenever a light was flashed onto the tin houses, although limited evidence sug- house. This showed that a pair, when sleep- gested that martins occasionally left their ing together, frequently changed position houses shortly before daybreak and called. during the night. Typically, the birds slept Nocturnal calls consisted of typical martin 174 CHARLES R. BROWN contentment notes and songs, including the Rough-winged Swallows (Stelgidopteryx guttural trill used by males when courting ruficollis) copulate within the nest cavity females. (possibly at night?), and Petersen (1955) made a similar suggestion for Bank Swal- DISCUSSION lows. Concealed copulation could be a com- Since Purple Martins have rarely been seen mon habit in the Hirundinidae, but addi- copulating during the day, I (1978a) sug- tional studies are needed. gested that copulation probably occurs in- Since male Purple Martins choose the side the rooms. Martins do not move be- sleeping arrangement, the seeming indiffer- tween rooms at night, so frequent sleeping ence of some males as to whether they sleep together likely facilitates copulation by with a female, especially early in the sea- bringing the sexes together. son, is puzzling. If martins copulate at night, From the standpoint of a male, copulating males would be expected to encourage fe- at night could be a habit that ensures that males to sleep with them. Yet, although his sperm and not that of a competitor are some pairs (21.9%) slept separately more transferred to his mate (and see Hoogland often than not (primarily because of the and Sherman 1976 on Bank Swallows, Ri- males’ choice), all pairs were found to sleep pa& riparia). Attempts at forced copula- together at least once during Stages I and tion (“rape”) are frequent in Purple Martins II, the periods when fertilization was most (Brown 1978a) and also are known in Gray- likely. Theoretically, a pair needs to sleep breasted Martins (Progne chalybea; Bitter- together on only one night to achieve fertil- baum, pers. comm.). Conspicuous diurnal ization. Also, perhaps the males’ option of copulations elicit massive interference and sleeping with a female or not could indicate attempted rapes by nearby males (Allen and a degree of male choice of mates. All pre- Nice 1952; Brown, unpubl. data). It is there- vious interpretations of pair formation in fore almost impossible for martins to mate Purple Martins have implied female choice during the daytime near a colony; if they do only (Allen and Nice 1952, Gaunt 1959, so at all, they must be away from the colony. Johnston and Hardy 1962). It seems to me more likely that they copu- Bitterbaum (pers. comm.) found diurnal late in rooms at night, when interference copulations more frequent in Gray-breasted from other males is nil. Copulation probably Martins of Trinidad than in the Purple Mar- takes place in conjunction with the noctur- tins he studied in Florida. He rarely heard nal calls. Similar calls and songs accom- Gray-breasted Martins calling at night. If panied 10 observed diurnal copulations they copulate during the day, they might (Brown, unpubl. data). not do so at night which could explain their From the standpoint of a female, by cop- lack of nocturnal vocalizations. Although ulating at night, she is not subjected to fre- rape occurs in Gray-breasted Martins, there quent and intense rapes. These are poten- is less interference from neighbors when tially dangerous to her because she abruptly mating than in Purple Martins (Bitterbaum, plunges into tree branches and onto martin pers. comm.), so perhaps nocturnal copula- houses in efforts to escape (Brown 1978a), tion has not been selected for in Gray- and attacking males are aggressive toward breasted Martins. her. Owing to the contagiousness of such On three occasions in Sherman, I saw behavior (Brown 1978a), a receptive female more than two martins share a room for would probably be raped if she flew about sleeping, and T. Dellinger (pers. comm.) the colony in daylight soliciting either her noted the same on three occasions at a col- mate or other males. Even though rapes oc- ony in Duncanville, Dallas Co., Texas. cur somewhat regularly in Purple Martin More than two birds (sometimes up to 12) colonies (Brown 1978a), more might be at- sleep together only during periods of se- tempted if martins often copulated diurnal- verely cold weather when food is scarce ly. The apparent eagerness of females to (Brown 1976). Martins abandon usual terri- sleep with males is consistent with the hy- tory defense during cold weather, and many pothesis of nocturnal copulation. huddle and sleep together, evidently to con- Emlen (1954) similarly reported that Cliff serve heat. In the early part of the nesting Swallows (Petrochelidon pyrrhonota) cop- cycle in Sherman (i.e. February and March), ulate within the nest, and, because of inter- nightly temperatures frequently range from ference from neighbors, copulations were -1” to 10°C. Even when starvation is not rarely successful unless the nest was sub- imminent, single pairs may sleep together stantial enough to screen the pair from other apparently to keep warm on cool spring Cliff Swallows. Lunk (1962) suggested that nights. MARTIN SLEEPING BEHAVIOR 175

My observations suggest two important Van Tyne Memorial Library was helpful in providing points for Purple Martin researchers. First, references. My martin studies have been supported those who use the banding technique of largely by my parents. Klimkiewicz and Jung (1976) (involving lowering the martin house at night to band LITERATURE CITED sleeping birds) are likely to capture a num- ALLEN, R. W., AND M. M. NICE. 1952. A study of the ber of vagrants sleeping in the houses. breeding biology of the Purple Martin (Progne These nonresidents should be taken into subis). Am. Midl. Nat. 47:606-665. account when calculating survival rates and BENT, A. C. 1942. Life histories of North American site tenacity of banded birds. Nonresidents flycatchers, larks, swallows, and their allies. U.S. Natl. Mus. Bull. 179. usually depart immediately after banding BROWN, C. R. 1976. Minimum temperature for feed- and rarely reappear at the colony of capture. ing by Purple Martins. Wilson Bull. 88:672-673. I know of no reliable way, short of observ- BROWN, C. R. 1978a. Sexual chase in Purple Martins. ing the birds entering the rooms, to deter- Auk 95:588-590. mine if a captured bird is a resident or a BROWN, C. R. 197813. Clutch size and reproductive success of adult and subadult Purple Martins. vagrant, although birds that are caught as Southwest. Nat. 23:597-604. pairs using the same room are likely to be BROWN, C. R. 1978~. Post-fledging behavior of Purple residents who are sleeping together. Martins. Wilson Bull. 90:376-385. Second, proprietors of Purple Martin BROWN, C. R. 1978d. Juvenile Purple Martins: field identification and post-fledging nest defense. Bull. houses who wish to estimate populations in Tex. Ornithol. Sot. 11:25-27. their colonies should use sleeping behavior BROWN, C. R. 1978e. Double-broodedness in Purple to determine residency status of birds. Day- Martins in Texas. Wilson Bull. 90:239-247, 657. time counts of birds on wires and even of BROW, C. R. 1979. Territoriality in the Purple Mar- birds perched on martin houses are not ad- tin. Wilson Bull. 91:583-591.. EMLEN. I. T.. Tn. 1954. Territorv. nest building. and equate to determine population size. Dur- pair formation in the Cliff Swallow. Auk 71: I6-35. ing the early, often cool part of the nesting FINLAY, J, C. 1971. Breeding biology of Purple Mar- cycle, the birds spend much time feeding tins at the northern limit of their range. Wilson away from the colony during the day. Un- Bull. 83:255-269. FINLAY, J. C. 1976. Some effects of weather on Purple less one observes the martins at sleeping Martin activitv. Auk 93:231-244. time when all residents appear, one is likely GAUNT, A. S. 1959. Behavior in the Purple Martin. to underestimate population size. At other Kans. Ornithol. Sot. Bull. 10:14-16. times of the year, site-seeking vagrants, ju- HOOGLAND, J. L., AND P. W. SHERMAN. 1976. Advan- veniles, and post-breeding vagrants visit the tages and disadvantages of Bank Swallow (Riparia ripariu) coloniality. Ecol. Monogr. 46:33-58. colonies. At such times one may overesti- JOHNSTON, R. F. 1966. The adaptive basis of geo- mate population size unless observations graphic variation in color of the Purple Martin. are made at sleeping time when residents Condor 68:219-228. can be recognized by their behavior. JOHNSTON, R. F., AND J. W. HARDY. 1962. Behavior of the Purple Martin. Wilson Bull. 74:243-262. Detailed observations of sleeping pat- KLIMKIEWICZ, M. K., AND P. D. JUNG. 1976. A new terns in Purple Martins, such as those pre- banding technique for nesting adult Purple Mar- sented here, are valuable for an ethological tins. North Am. Bird Bander 2:3-6. comparison of this species with other mem- LUNK, W. A. 1962. The Rough-winged Swallow Stel- bers of the genus Progne and with other gidopteryx ruficollis (Vieillot), a study based on its breeding biology in Michigan. Publ. Nuttall Or- hirundinids. Bitterbaums’ (pers. comm.) ob- nithol. Club 4:1-155. servations on the Gray-breasted Martin in- NILES, D. M. 1972. Molt cycles of Purple Martins dicate differences in sleeping patterns be- (Progne suhis). Condor 74:61-71. tween these congeners. Knowledge of OBERHOLSEH, H. C. 1974. The bird life of Texas. Vol. 2. Univ. Texas Press, Austin. sleeping behavior may prove useful in un- PETERSEN, A. J. 1955. The breeding cycle in the Bank derstanding the ecology and relationships Swallow. Wilson Bull. 67:235-286. of swallows and other birds. SOUTHERN, W. E. 1959. Homing of Purple Martins. Wilson Bull. 71:254-261. ACKNOWLEDGMENTS 2601 Turtle Creek, Sherman, Texus 75090. Accepted Erik J. Bitterbaum discussed martins with me and crit- ~__ icized the manuscript. Janet Hinshaw of the Josselyn for publication 13 November 1979.