THE INFLUENCE OF MOONLIGHT ON THE BEHAVIOR OF GOATSUCKERS (CAPRIMULGIDAE) ALEXANDER M. MILLS • Departmentof Biology,Carleton University, Ottawa, Ontario KIS 5B6, Canada ABsTR•CT.--Whip-poor-wills(Caprimulgus vociferus) showed significantly higher levels of locomotory,vocal, and nestactivity during twilight and bright moonlightthan under moon- lessconditions. Field observationsand nestrecord card data indicated that Caprimulgusspecies usuallysynchronize their reproductivecycle with the lunar cycle.Hatching tends to occur during young waxing moons,presumably so that moonlight-enhancedforaging will be at a maximumwhile the dependentnestlings are an energyburden on the parents.Observations of foraging Whip-poor-wills suggestthey are primarily visually oriented, unlike the bats (Chiroptera). Received26 March 1985,accepted 17 November1985. MOSTnocturnal flying insectivoresare echo- 1983), as are correction formulae for coordinates. Be- locating bats (Fenton 1974, 1984; Simmons and causelunar light intensity increaseswith increasing Stein 1980), but a significant though unknown %MFI and moon height (Bowden 1973), I grouped proportion (15% by taxa) are non-echolocating nights into 5 categoriesbased on %MFI at midnight (0%, 1-25%, 26-50%, 51-75%, and 76-100%), and then birds, most of which are goatsuckers(Capri- subdivided them based on moon height (below the mulgidae) in the genus Caprimulgus. horizon, 1-100 min above the horizon, 101-200 min Most male caprimulgids are conspicuousby above,and 201-300 min above).Each 30 min during their songs,and many accountshave reported monitoring, I recordedcloud cover and wind strength increased singing in bright moonlight (e.g. basedon a predeterminedsubjective scale of 0-3. Wynne-Edwards 1930, Brauner 1952, Cooper Locomotoryactivity.--Temporal patterns in the lo- 1980, Bjorklundand Bjorklund 1983). This sug- comotory activity of Whip-poor-wills were moni- geststhat goatsuckersare moon-loving or lu- tored by radiotelemetry.The radio packagesworn by narphilic, unlike the insectivorousbats, which the birds as backpacksaveraged 4.9 g. The signals appear to be lunarphobic or lunar-indifferent (173.8-174.1 MHz) were detected with a Merlin 12 (Morrison 1978,Bell 1980,Anthony et al. 1981). (CustomElectronics, Urbana, Illinois) receiver oper- ated with a directional 5-element Yagi antenna. Ef- The purposeof this study was to seeif capri- fective range varied from 500 to 1,000 m, and all mulgids, especiallyWhip-poor-wills (Caprimul- transmitters performed for the entire field season. gus vociferus),adjust locomotory (and hence Eleven birds (7 males, 4 females) were outfitted feeding), vocal,and nest activitiesaccording to with radios. Two males disappearedand 1 female lunar conditions. was found dying from an eye injury a week after being tagged. The 8 remaining birds were each MATERIALS AND METHODS trackedfor 43-63 days.All 5 males held territories and all 3 females and 4 of 5 males fledged young Studysite.--I conductedfieldwork near Rideau Lake while tagged. Narrows (44ø43'N,76ø17'W), eastern Ontario, May- Individual pairs were monitored from blinds near July 1983 and 1984. This was a region of marginal their nests. Readings,taken every 4 min, were de- farmland,forest, rock outcrops, and many smalllakes, fined in terms of distance (near, medium, far), as as- where both Whip-poor-wills and Common Night- sessedby the intensity of the radio signal and the hawks (Chordeilesminor) were common. direction (1 o'clockthrough 12 o'clock).Any change Lunar,solar, and meteorologicalconditions.--Percent- from the preceding reading in distance or direction agesof moon faceilluminated (%MFI), moon heights qualified as a move. In all, 14,782telemetry readings (measured in time), and times of sunrise, sunset, and were taken. various solar positions below the horizon are given I distinguished periods of solar influence (twi- in the AstronomicalAlmanac (e.g. Vohden and Smith light) from periodsof lunar influence(night) by de- termining when moonlight (> 25%MFI) began to in- fluence activity after dusk and ceasedto influence activity before dawn. Presentaddress: Department of Zoology, Univer- Instancesof Whip-poor-will feeding sallies were sity of Guelph, Guelph, Ontario NIG 2W1, Canada. recordedvisually for distanceand direction. 37O The Auk 103: 370-378. April 1986 April 1986] Moonlightand Goatsucker Behavior 371 I could tell when Common Nighthawks were ac- ricanOrnithological Society for C. rufigenaand C. pec- tive from their distinctive flight calls. I divided the toralis;and from Sprunt (1940) and the NANRP for night into 10-min censusperiods beginning at sunset C. carolinensis.For the nighthawks (Chordeilesminor and ending at sunrise(with an adjustmentperiod and C. acutipennis),similar data (74 recordsfrom 29 midway of 0-9 min), and noted whether nighthawks yr) came from the NANRP, the ONRS, the MNRS, were heard in each period. the Prairie Nest Records Scheme, and the Quebec Vocalactivity.--The 5 radio-tracked males contrib- Nest RecordsScheme. These data were pooled by ge- uted equally to songactivity. For each 10-min census nus. period, I noted whether each monitoredmale sang I defined hatching datesin terms of days after the 1-10 "Whip-poor-will" units, 11-100 units, 101-1,000 most recent full moon, and divided the lunar month units, or not at all. These were respectivelyassigned into 10 3-day periods, such that the new and full scoresof 1, 2, 3, and 0. Song counts were made in moon dateswere in the middle of 3-day periods. Pe- 2,865 10-min periods.I usedmovement results to dis- riods, then, are days 2-4, 5-7 ..... 26-28, 29-1. Be- tinguish twilight 10-min periodsfrom night ones. causethe last period is only 83% the length of the I also measuredthe levels of aggressiveresponse others(because the lunar month is actually29.5 days), by male Whip-poor-wills to taped song. Five loca- expectedvalues in the statisticaltests were adjusted tions along a prescribedroute away from the telem- accordingly. etry site were used for solicitation trials. Care was Statisticalanal¾$is.--For each solar-definedand lu- taken to choose quiet times (wind category 0 or 1) nar-defined category,I generatedscores for move- and to spread the 15 test times (3 dusk, 3 dawn, 3 ment (percentageof radio readingsthat were moves), darkness,6 moonlight) over the seasonto prevent singing (percentageof theoretical maximum score), habituation to the tape. and nest departuresand nest feedings (numbers per Eachtrial (n = 75) involved 4 repetitionsof 15 s of 1,000 observationminutes). Birds temporarily out of song played at full volume on a portable cassette range as determined by telemetry and birds on nests player followed by 15 s of silence. A positive vocal were not included in calculating movement and responserequired that the local male sing, and a pos- singing scores.A 2 x 2 contingencytable for solar itive visitationresponse required that the male either and lunar conditionsdistinguished twilight periods fly over the tape or land within about 10 m. A bird from night periods. could respondafter the 1st, 2nd, 3rd, or 4th 15-spre- Night scoreswere subjectedto multiple regression sentation (or never), scoring 4, 3, 2, or 1 (or 0), re- and correlation analysisusing dummy variables for spectively. the various categoriesand weighted according to Becausesong tempo (the number of "Whip-poor- sample size. Factorswere %MFI, moon height, and will" units per minute of unbrokensong) might in- cloud (0-1 clear, 2-3 cloudy) for movement; %MFI, dicate an individual's level of arousal, I compared moon height, and wind (0-1 calm, 2-3 windy) for temposunder various light conditions. song;and %MFI and moon height for nest activity. Reproductiveactivity.--I monitored 5 successfulnests Differencesamong twilight, bright moonlight,and of 4 radio-taggedpairs from blinds about 6 m away. darknessscores were testedusing a Bonferroni Chi- During incubation, I recorded the number of times square(Miller 1966). Vocal and visitation responses the incubating bird left the eggs (nest departures), to taped song were analyzed for variance (ANOVA), either to feed or to allow a mate to assume incubation and between-categoryscores were testedby the Stu- duties.After hatching,I recordedthe number of times dent-Newman-Keuls procedure (SNK) (Sokal and adults fed the nestlings (nest feedings). Nests were Rohlf 1981).Song tempo measureswere subjectedto monitored for 14,690 min during incubation and for a Bonferroni t-test (Miller 1966), and a Chi-square 22,431 min during the nestling period. testwas usedto test the data for synchronybetween To assessthe possibility of a synchronized rela- the lunar month and the reproductivecycle. tionship between the lunar month and goatsucker reproductivecycles, I collected79 Caprimulgusnest records(from 38 different years)for which hatching RESULTS dateswere known or calculable.Whip-poor-will rec- ords camefrom Whedon (1906), Bailey (1912), Nau- Locomotoryactivity.--On moonless nights man (1925), Mousley (1937), Tyler (1940), Raynor Whip-poor-will activity appearedto end (dusk) (1941), Fowle and Fowle (1954), Kilham (1957), the and begin (dawn) when the sun was about 13ø Ontario Nest RecordsScheme (ONRS), the Marltimes Nest Records Scheme (MNRS), the North American below the horizon. A contingencytable for Nest RecordsProgram (NANRP), and this field study. moonless and moonlit (>25% MFI) conditions Other TemperateZone Caprimulgusdata came from verified this pattern for the periods when the Hewerr (1883), Gurney (1883), Soppitt (1883), Corbin sun was