Avocetta N° 17: 157-162 (1994) Echolocation acuity of the Palawan Swiftlet (Aerodramus palawanensis) CHARLEST. COLLlNS(l)and RICHARDMURPHy(2) (I)Dept. Biological Sciences, California St. Univ., Long Beach CA 90840 USA (2) The Cousteau Society, 870 Creenbrier Circle, Suite 402, Chesapeake VA 23320 USA Abstract - Echolocation acuity trials were conducted on Palawan Swiftlets (Aerodramus palawanensis) under natural conditions in Palawan, Philippine lslands. Detection of 3.2 mm diameter obstacles was significantly less than for 6.3 mm and 10 mm obstacles. These results are consistent with previous laboratory trials conducted on other swiftlets. They confirm that although echolocation is used for orientation in cave nesting and roosting areas it is unlikely to be effective in detecting their typically small « 5 mm) food iterns. Introduction report here similar experiments conducted with the PaIawan Swiftiet (Aerodramus palawanensis) under The ability to use echoIocation or animaI sonar is a naturai conditions in a Iarge nesting and roosting widespread and well studied specialization among cave. bats of the suborder Microchiroptera (Griffin 1958, Vincent 1963, Simmons et al. 1975, Busnell and Fish 1980). Among birds, echoIocation is known onIy for MetHoDs the OiIbird (Steatornis caripensis) a unique cave nesting caprimulgid of northern South America These experiments were conducted in the under- (Griffin 1954, Konishi and Knudsen 1979) and most, ground river cave in St. Pau I River Subterranean if not all, of the widespread species of cave swiftlets Nationai Park, PaIawan, Philippine Islands from 14- (Aerodramus sp.)(Medway and Pye 1977) which 21 September 1990. The tests were conducted in a occur from the western Indian Ocean isIands of side channei off the main cave about 200 m from the Mauritius and Reunion eastward to the islands of the entrance. This site was in constant totai darkness. southwest Pacific ocean (SibIey and Monroe 1990). The test apparatus consisted of two bamboo poles The echolocation sounds utiIized for orientation by erected next to the rock walls of the channel and birds are brief bursts or audible clicks Iasting for a connected by a Iowerabie rope, ca. 6 mrn in diameter, few milliseconds with a repetition rate of 3-20 clicks stretched between pulleys. Suspended from this rope per second (Novick 1959, Griffin and Suthers 1970, were 12 test obstacJes consisting of 3.6 m Iengths of Medway and Pye 1977, Fullard et al. 1993). The nylon or manila rope lO mm, 6.3 mm or 3.2 mrn in frequency range of these clicks is from 1-16 kHz diameter. The test obstacJes (ropes) were centered 30 (Medway and Pye 1977, Fullard et al. 1993) and there cm apart which allowed about another 30 cm space on are no ultrasonic frequencies as typically found in either side between the outermost rope and the rock microchiropteran pulses (Cranbrook and Medway wall. Depending upon the tidai influenced water level 1965). The severai tests of the acuity of echoIocation in the cave the test obstacJes stopped near the water in Aerodramus swiftIets (Griffin and Suthers 1970, surface or were about 50 cm above it. Small metai Fenton 1975, Griffin and Thompson 1982, Smyth bells were attached to the bottom of each test obstacJe and Roberts 1983) have indicated that it is probabIy as a further aid in detecting contact by a flying bird. only utiIized for flight orientation within caves where Twenty four of twenty six separate triais with the they nest and roost and not for foraging. However, severai test obstacJes were conducted between 16:45 most of the tests of swiftiet echoIocation acuity have and 20:00 when Iarge numbers of swiftIets were retur- involved small numbers of individuals flying in ning to roosting sites in the cave; two additional trials laboratory conditions or a man-rnade mine shaft. We were conducted from 10:20-11 :30. Observations were Accepted 30 December, 1993 158 C. T. Collins and R. Murphy made of individuai swiftlets which either passed unpublished) and until more definitive information is through the apparatus without striking a test obstacle available, we follow Sibley and Monroe (1990) in (= a pass), struck a test obstacle (= a hit), or turned giving p alawanensis full species status. Our back before passing through but after having observations and prior field work by Coleman (1981) approached to within approximately 2 m of it (= a indicate that the swiftlets of the underground river are turn). Visual observations were made under two sets attributable to this species. This was further confirmed of conditions. The first was under very low light by comparison of museum specimens and a voucher conditions achieved by reflecting a small incandescent specimen (#7391) housed in the ornithological headlamp beam off the ceiling above and slightly collections of California State University, Long back from the test apparatus. This gave just enough Beach. This specimen was one of a total of 19 light for the swiftlets to be seen and their performance swiftlets, randomly captured in the underground river quantified, but not enough illumination to allow them cave, which were weighed, measured and examined to decrease their utilization of echolocation. before release, all of which seemed to belong to this EIsewhere in the cave when either bright lights were single species (Collins, unpublished). Two other non- used or when near the mouth of the cave and natural echolocating species, the Glossy Swiftlet, Collocalia light was available, we noticed a clear decreased esculenta and Pygmy Swiftlet, C. troglodytes were utilization of echolocation. Under these brighter seen near the mouth of the underground river cave but light conditions, but not during the acuity trials, were never seen to penetrate beyond the area of echolocation clicks were reduced in emission rate or naturallight (personal observation). were discontinued altogether (personal observation). Echolocation acuity and obstacle avoidance were The second method of observation entailed the use of judged by the percent passes and hits of birds flying a battery powered infrared light source to illuminate through the apparatus and also by means of a model the test apparatus and the observation of the swiftlets formulated by Smyth and Roberts (1983) of a by means of a night vision scope. Again, no theoretical missile the same width as the wing span of diminution of echolocation clickemission rate seemed the swiftlets traveling through the test apparatus. This to occur under these conditions. In both observation model determines that the probability of a bird conditions the birds' performance was thus thought striking a test object to be k(t+m)/D where k is the to be strictly on the the basis of their ability to detect number of test obstacles of diameter t stretched across the test obstacles by echolocation alone. a passage with a width D. The width of the missile or We recorded a total of 1554 interactions in the form wingspan of the swiftlet is m. The number of obstacle of 716 passes and hits and 838 turns in test and strikes (S) compared to the total number of flights ( ) control observation periods.As ali of the swiftlets should be approximately equal to k(t+m)/D if the bird observed were free-flying and unmarked the total is flying at random. Thus the ratio (R) of k(t+m)/D number of separate individuals observed could not be over SIN can be used to indicate the degree of determined; it is quite possible that observations were randomness of the swiftlet' s flight performance. An R made of the same individual's performance on more value equal or close to l indicates random flight than one day. Many thousands of swiftlets utilized the independent of the obstacles, and a value appreciably entire 8.2 km long underground river cave with some greater than l indicates avoidance behavior. An R value belo w l would indicate deliberate collisions nests located at least 4 km from the cave mouth with the obstacles or at least a collision rate greater (Coleman 1981). A vastly smaller component of the than predicted by random flight. As also noted by population including those individuals with nests or Smyth and Roberts (1983) "this method of analysis is roosts in the immediate vicinity of the test apparatus, independent of the distance between obstacles, or or which used the channel to reach other more remote between an obstacle and the side wall, providing such side chambers of the cave, were likely to have been distance is greater than m." In our calculations a wing observed during our experiments .• span (m) of 28.5 cms was used which is slightly less The taxonomy and identification of swiftlets, than the maximum (stretched) wing span (30.16 cm particularly in the field, is very confusing. Recent +0.73 SD; Collins unpublished) but more closely reviews of the Philippine Island swiftlets (Dickinson approximates the normal wing span of this swiftlet in 1989, Sibley and Monroe 1990) indicate the presence flight. in Palawan of a swiftlet which can be considered either as an endemic subspecies, palawanensis, of the widespread Mossy-nest Swiftlet Aerodramus Results vanikorensis (Dickinson 1989) or an endernic species A. palawanensis (Sibley and Monroe 1990). In light The duration of each observation peri od varied of several distinctive aspects of this swiftlet (Collins, according to the number of birds passing the test Echolocation acuity oj the Palawan Swiftlet 159 apparatus. Thus we have chosen to present data on passes, hits or turns as percentages of the total number of birds observed during each of the observation periods. Before analysis of variance was conducted, the data were arcsine transformed (Sokal and Rohlf 1981). These data were then tested for normality by Lilliefors test and homogeneity of group variance by Boxs small sample F approximation for cell counts less than lO; both were not significant (SYSTAT 1990) indicating parametric statistics are appropriate.