Change in the signal-response sequence responsible for

Proc. Natl. Acad. Sci. USA Vol. 90, pp. 5813-5817, June 1993 Evolution Change in the signal-response sequence responsible for asymmetric isolation between Drosophila planitibia and Drosophila silvestris (evolution/sexual isolation/behavior/hawaii) ANNELI HOIKKALA* AND KENNETH KANESHIRO Hawaiian Evolutionary Biology Program, University of Hawaii at Manoa, Honolulu, HI 96822 Communicated by Hampton L. Carson, March 19, 1993 (received for review October 13, 1992) ABSTRACT Drosophilaplanitibia and Drosophila silvestris form a species pair that is an example of species diverged through a founder event. These species exhibit asymmetric sexual isolation, courtships between D. planitibia males and D. suvestris females being more successfl than courtships between D. silvestris males and D. planitibia females. When analyzing the signal-response courtship sequence in these species, we found that D. silvestris females responded to male circling by standing or preening while D. planitibia females required further signals from the male to stop walking. The main reason for the reduced mating success rate ofd. silvestris males with D. planitibia females was that the females responded to male circling by walking and the males did not proceed to the head-under-wings (HUW) position of a walking female. Another critical phase in these courtships was the HUW position in D. silvestris, where males proceeded almost immediately to wing and leg vibration. The courtships between D. planitibia male and D. silvestis female proceeded in a signal-response coordination until the male went to the HUW position, where he fanned his wings for too long a period before proceeding to wing and leg vibration. Thus, it seems that the asymmetric isolation between D. planitibia (ancestral species) and D. silvestis (derived species) is mainly due to a loss of transitions in the signal-response chain of D. silvestis. A change in the behavior of the males in the HUW position has caused further isolation between the species in both directions. It has been suggested that changes in the mating system have played a major role in the speciation of Hawaiian Drosophila flies (1). Kaneshiro (2) conducted mate preference experiments to examine sexual isolation among four of the most derived species (planitibia species complex) of the planitibia subgroup of the picture-winged group of Hawaiian Drosophila and found that females of the ancestral species discriminated against males of the derived species, whereas females of the derived species accepted the courtship of males of the ancestral species more readily. This kind of asymmetric sexual isolation was found in all except one of the species pairs in the planitibia species complex (2) and it has also been found to evolve as a result of bottlenecks that occur in laboratory strains (3-5). The reasons for the asymmetry are still largely unknown. Kaneshiro (1) first suggested that the asymmetric isolation between ancestral and derived species could be due to the loss of behavioral elements in the courtship of derived species. Ohta (6) and Kaneshiro (7, 8) suggested that following a founder event and during the initial stages of colonization when population size is small, there might be a shift in the distribution of female mating types toward less discriminating females because of the lower rate of encounter between choosy females and males that are able The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact. 5813 to satisfy their courtship requirements. Other factors such as changes in the relative emphases placed on various behavioral elements during courtship (9) and changes in the genetic covariance between a sexually selected trait and the female preference for that trait (10, 11) have been suggested to occur during bottlenecks. In the present paper we attempt to determine the reasons for the asymmetric isolation between Drosophila planitibia and Drosophila silvestris, a species pair used in the original work of Kaneshiro (1). We describe the signal-response sequences of these two species to trace the changes that may have taken place in the specific mate-recognition system (SMRS) during speciation. We also studied the interactions of males and females in successful and unsuccessful intraspecific courtships and in interspecific courtships to determine the critical phases in courtship. MATERIALS AND METHODS Fly Strains. Adult flies of D. planitibia and D. silvestris used in the present study were taken from laboratory stocks maintained by the Hawaiian Drosophila Research Stock Center at the University of Hawaii. D. planitibia flies were taken from the Y14 laboratory stock established from females collected from Waikamoi on the island of Maui in 1987, whereas D. silvestris flies were taken from the U26B9 laboratory stock established from a single fertilized female collected from Kahuku Ranch on the island of Hawaii in 1977. Earlier studies indicated that the basic courtship sequence of four different strains of D. silvestris was species-specific (A.H., unpublished observations). The flies were cultured using standard culture procedures (12) and used in the experiments at the age of 4-6 weeks. Each fly was used only once. Recording the Behavior of the Flies. We videotaped and analyzed 15 successful and 15 unsuccessful courtships of D. planitibia, 19 successful and 10 unsuccessful courtships ofd. silvestris, 17 courtships between D. planitibia female and D. silvestris male, and 15 courtships between D. silvestris female and D. planitibia male. To videotape the behavior ofthe flies, one male and one female were aspirated without anesthesia into a 5 x 3.5 x 1.5 cm balsawood mating chamber with a glass roof. The chamber was placed inside a soundproof box above a velocity-sensitive microphone (13) to record the sounds produced during abdomen purring and wing vibration of the males (14). The chamber was illuminated with a fiber optic illuminator (Dolan-Jenner, model 180) and the behavior of the flies was videotaped with a JVC videocamera and a Sony VO-2610 U-matic (1.91-cm tape) recorder. The court- Abbreviations: HUW, head-under-wings; SMRS, specific materecognition system. *Present address: Department ofgenetics, University of Oulu, 90570 Oulu, Finland.

5814 Evolution: Hoikkala and Kaneshiro Proc. Natl. Acad Sci. USA 90 (1993) ship of each pair was recorded until copulation occurred or until 20 min had elapsed from the first courtship element of the male or the female. The recordings were made at a temperature of 19 C ± PC. We reviewed the videotapes and recorded the courtship behavior of the flies using an MS-DOS-adapted version of a behavioral analysis program developed by Welbergen et al. (15). The behavior of D. planitibia and D. silvestris flies was classified into eight male and seven female elements (Table 1). Two of the elements, "no courtship" and "copulating," were scored simultaneously for both sexes. Changes in the behavior of the flies were recorded on a continuous time scale by pressing assigned keys on the computer keyboard each time the male or the female changed from one behavioral element to another. Thus, strings of behavioral displays and matching time points were saved on computer files. Data Analysis. Analysis ofbehavioral transitions was based on the assumption that the probability of a given act depends only on the identity of the act immediately preceding that act-i.e., a first-order Markov chain. Our analysis used the method described in Welbergen et al. (15, 16) and Liimatainen et al. (17). For each courtship, we first calculated the frequency of changes from each behavioral element to any other element displayed within and between individuals of each courting pair. These preceding-following behavior matrices within each experimental group (intraspecific and interspecific combinations) were added. The total transition matrix was divided into two within-individual (male or female) and two intersexual (male-female and female-male) matrices. Each matrix involved a different number of transitions from preceding to following behaviors; the smallest of these was 529 and the largest was 1242. Conditional probabilities of behavioral transitions as given in Table 2 and Fig. 1 were calculated as a proportion of a preceding behavior leading to a following behavior divided by the total frequency of the preceding behavior. For example, the probability of 33.1 for circling wing-waving transitions shown in the first -- row of Table 2 is obtained by dividing the number of transitions from male circling to female wing waving (86) by the total number of male circlings leading to a change in female behavior (260). The expected value for this transition is the number of transitions from male circling (260) x the number of transitions leading to female wing waving (154) divided by the total number of transitions in the male -> female matrix (609). The actual and expected values of the matrix cells were tested for significance using x2. To identify the categories responsible for a significant x2 value, we estimated an adjusted residual (standardized residual divided by the square root of its variance) for each cell (18). Adjusted residuals >1.96 (standard normal deviate) indicate that the given transition had occurred significantly more often than expected (P < 0.001). Two criteria were applied in these tests: at most, 20% of the matrix cells were allowed to contain expected values of <5, and the total matrix contained at least 5 x n (number of preceding behaviors) x m (number of following behaviors). Significant transitions that occurred at least five times in the courtships of a group and had a conditional probability of at least 15% were used to set up the kinematic graphs (Fig. 1). These figures also comprise information on the relative frequencies of male and female behavioral elements (boxes), calculated by dividing the number of behavioral elements in question by the total number of behaviors displayed by the relevant sex. RESULTS Courtship Behavior of D. planitibia. D. planitibia males began courtship by waving their wings or, less frequently, by standing alone and purring their abdomen (transitions from no courtship to "wing waving" were 72% and from no courtship to "purring alone" were 26%, Fig. 1). The female responded to both of these male behaviors, or initiated the courtship herself, by waving her wings. The male reacted to the female wing waving by waving his wings or by walking in front of the female and facing her. This was a very strong signal for the female and almost invariably (after >90% of the male facing bouts) she "slashed" at the male with her forelegs. The male then either began to wave his wings or circled around the female. The female responded to male wing waving by waving her wings and to male circling by walking. The male proceeded to a HUW position of the female while she was "standing," "preening," or "walking." As females were more likely to engage in walking (26% of the time) than in standing or preening (8% and 5% of the time, respectively), 70% of the male HUW bouts were preceded by female walking. D. planitibia males spent an average of about 9.1 sec in the HUW position, before proceeding to wing and leg vibration. This period included male wing fanning that lasted for 2-10 sec. The female responded to male HUW by standing (in 66% of the bouts) or preening (18% of the bouts). The male then supinated his wings in the forward position and vibrated them Table 1. Male and female behavioral elements in D. planitibia and D. silvestris used for the transition analysis d behavioral elements Purring alone: The male bobs his abdomen producing purring sounds while standing at a distance of at least 1.5 cm from the female. Wing waving: While oriented toward the female, the male extends both wings and waves them up and down. Facing: The male positions himself in front of the female facing her. He may keep his wings extended. Circling: The male circles around the female or follows the female circling behind her in a small arc. He may stop while circling and keep his wings extended. Head-under-wings (HUW): The male stands behind the female with his head under the wings of the female. At this position he extends his wings in lateral position and fans them in a slow speed. Vibration: While still in the HUW position the male supinates his wings in the forward position and vibrates them rapidly producing sound bursts. The male also vibrates the tibia of his folded forelegs against the dorsal surface of the abdomen of the female. 9 behavioral elements Wing waving: The female extends both wings and waves them up and down. Slashing: The female slashes at the male with her forelegs. Standing: The female is motionless. Preening: The female preens her body parts while standing. Walking: Any locomotion by the female toward or away from the courting male (includes decamping, which comprised <1% of female behavior bouts). Behavioral elements occurring coincidentally in both sexes No courtship: The male is not within close vicinity of the female (distance is >1.5 cm) and he is not purring his abdomen. Accordingly, the female is not being courted. Copulation: The male mounts the female and the flies copulate.

Evolution: Hoikkala and Kaneshiro D. planitibia Proc. Natl. Acad. Sci. USA 90 (1993) 5815., ow 9 1 - - - - - - - - - G 8060 ~~~~400 o- <60Y.< < 20 h 6 200% - <40%/ - 1001/. We< 5% - <10%0 10N - <15%/ 9 15% - <200oO = 200/s <25% - 2.25% FIG. 1. Kinetographs of successful courtships between D. planitibia males and females and between D. silvestris males and females. Within-male and within-female transitions (open arrows) and between-individual transitions (black arrows) should be studied separately (see text). Boxes refer to the relative frequencies of male and female behavioral elements. Follow the description of the sequence of within-individual and between-individual transitions in Results. rapidly producing a long sound burst. At the same time, he vibrated the tibial bristles of his folded forelegs against the dorsal surface of the female's abdomen. The female reacted to 85% of the male wing vibration bouts by walking away. If the female remained stationary, the male mounted her and copulation occurred. Females of the planitibia subgroup species extend their ovipositor when they are ready for copulation (19). However, this acceptance signal could not be observed in our videotape recordings because the flies had been filmed from above. Unsuccessful courtships of D. planitibia proceeded about the same way as the successful ones, but only two-thirds of these courtships proceeded to male wing vibration. Courtship Behavior of D. silvestris. The courtship of D. silvestris proceeded much the same way as that of D. planitibia (Fig. 1). The male began the courtship by waving his wings or by purring his abdomen. Then he went to the front of the female in the facing position. The female slashed the male and the male began to circle around the female. In contrast to the courtship of D. planitibia, the male behaviors facing and circling were tightly connected: the betweenindividual transitions from male facing to female slashing and from female slashing to male circling had conditional probabilities close to 100% (99% and 98%, respectively). Male circling was followed by female standing, preening, or wing waving. The male proceeded to HUW when the female was standing or preening. D. silvestris males spent on average only about 1.2 sec in the HUW position before proceeding to wing and leg vibration. The wing fanning bout preceding wing vibration was extremely short (<1 sec). Wing vibration followed >90%o of the HUW bouts and the female did not change her behavior until the male had produced the song. Female standing led to copulation and female walking to a break in the courtship. Courtships of unsuccessful males resembled those of the successful ones. There were, however, a few transitions that were significant in one, but not in both, of these groups (Table 2). In unsuccessful courtships, male circling led to female standing and preening and, instead of wing waving, to walking. Also, transition from female walking to male purring alone was significant although it occurred in <10% of the walking bouts. The most obvious difference in male-female transitions of D. planitibia and D. silvestris was that D. silvestris females reacted to male circling significantly more often by standing or preening than did D. planitibia females [G(l) = 115.34, P < 0.001]. Female-male transitions show that D. planitibia males went under the wings of a walking female more often than D. silvestris males [G(ll) = 34.71, P < 0.001]. In D. planitibia, male HUW led to female standing in 84% of HUW bouts, whereas in D. silvestris, the females did not change their behavior when the male went to HUW. Behavior of the Flies in Interspecific Courtships. In interspecific courtships, the male and the female responded to the behavior of the opposite sex almost the same way as they did in intraspecific courtships (Table 2). Because ofthe difference in the signal-response transitions between D. planitibia andd. silvestris, none of the interspecific courtships led to copulation within the 20-min recording period. In courtships between D. silvestris female and D. planitibia male, the male responded to

5816 Evolution: Hoikkala and Kaneshiro Table 2. Between-individual transitions, where the signal-response sequences of D. planitibia and D. silvestris differ from each other Q D. planitibia Y D. silvestris d planitibia s d planitibia u 6 silvestris d silvestris s d silvestris u d planitibia - Y transitions Circling Wing waving 33.1 Standing 15.0 28.1 24.0 37.4 Preening 14.4 14.6 26.2 21.6 -~ Walking 60.2 52.9 72.7 28.8 HUW Standing 66.3 33.8 Preening 17.5 33.8 32.6 Walking 94.1 59.6 - d transitions Slashing Wing waving 47.5 55.1 45.6 - Circling 44.5 41.8 82.7 98.0 91.5 41.2 Standing HUW 27.8 54.3 81.8 88.9 84.2 88.1 -~ Vibration 64.8 17.4 Preening No courtship 30.4 Circling 42.0 25.9 HUW 30.4 26.0 28.6 68.5 45.2 26.9 - Vibration 10.2 Walking No courtship 28.8 40.3 58.5 35.7 32.9 30.8 Purring alone 8.6 9.8 Circling 32.5 30.0 50.8 -~ HUW 21.7 12.8 Significant (P < 0.001) conditional probabilities (percent of preceding behaviors leading to following behaviors) of male-female and female-male transitions have been presented for courtships involving D. planitibia female with a successful (s) or unsuccessful (u) conspecific male or with D. silvestris male and for courtships involving D. silvestris female with a successful (s) or unsuccessful (u) conspecific male for D. planitibia male. The most critical transitions leading to female walking and to a break in interspecific courtships are printed in bold type. female slashing in his species-specific way by circling around the female (after 41.2% of the slashing bouts) or by waving his wings (after 45.6% of the slashing bouts). The female responded to male circling by standing or preening. When the female stopped walking, the male proceeded to the HUW position. In contrast to D. planitibia courtships, there was no transition from female walking to male HUW position, mainly because the female was stationary at this courtship phase. While in the HUW position, the male fanned his wings slowly for several seconds. D. silvestris female did not wait for the male to proceed to wing vibration but walked away in 59.6% of male HUW bouts, which broke off the courtship. If the female remained stationary when the male fanned his wings, the male changed over to wing and leg vibration. In the courtships between D. planitibia females and D. silvestris males, the male, again, behaved in a way typical to his own species (within-individual transitions between facing and circling and between HUW and vibration were 84.0%o and 62.9%o, respectively). Also the female behaved in a way typical of her own species and responded to male circling by walking. As in intraspecific courtships, D. silvestris male responded to female walking rarely by going under her wings to the HUW position (in 4.6% of the female walking bouts). Thus, female walking led to a break in the courtship sequence. If the female stopped walking, which she did on occasion, the male immediately went into the HUW position and very quickly changed over to wing and leg vibration. At this stage, the female walked away, probably because the male proceeded so quickly to wing vibration without a proper wing fanning bout. As Table 2 shows, courtships between D. planitibia female and D. silvestris male had two critical phases. Male circling led to female walking in 72.7% of the bouts and male HUW led to the same female behavior in 94.1% of the bouts. DISCUSSION The most consistent difference between the two species was the lack of between-individual transitions from male circling to female walking, from female walking to male HUW, and Proc. Natl. Acad. Sci. USA 90 (1993) from male HUW to female standing or preening in D. silvestris. Lambert (20) pointed out that the loss of male-female or female-male elements (between-individual transitions) in different parts of the SMRS would have different effects on the outcome of crosses between individuals from ancestral and derived species. Lambert argued that the loss of behavioral elements in the middle of the SMRS may lead to asymmetric isolation only if the females of the derived species respond to a male behavior in a way similar to the way females of the ancestral species would react to the male behavior next in sequence. This is exactly what we found in D. planitibia and D. silvestris. D. silvestris females responded to male circling by standing or preening while D. planitibia females required further signals from the male to stop walking. The major reason for the inability of D. silvestris males to mate with D. planitibia females was that the females responded to male circling by walking and not by standing or preening, and the males did not go under the wings of a walking female. D. planitibia males, on the other hand, had no difficulty in going under the wings of a D. silvestris female, even though she already had stopped walking. Sexual isolation between D. planitibia and D. silvestris is not complete in either direction. In Kaneshiro's (1) malechoice experiments, Stalker's isolation index (21) for D. planitibia male with D. silvestris and D. planitibia females was 0.25 and for D. silvestris male with D. silvestris and D. planitibia female was 0.73. In the present study, none of the interspecific courtships led to copulation during the 20-min observation period, whereas about 25% of D. planitibia and 35% of D. silvestris intraspecific courtships were successful within that time. Both kinds of interspecific courtships frequently broke off, while the male was in the HUW position. D. silvestris males proceeded to wing and leg vibration too quickly for D. planitibia females and D. planitibia males fanned their wings for too long a period for D. silvestris females. Wing fanning (also referred to as semaphoring; ref. 19) is an ancient feature in the planitibia subgroup species and it seems to be an essential part ofthe courtship in many species including D. planitibia. The males of some intermediate spe-

Evolution: Hoikkala and Kaneshiro cies of this subgroup fan their wings for several minutes and even produce an audible "pulse song" during the wing fanning period (A.H., K.K., and R. Hoy, unpublished data). In the derived species, the wing fanning bouts last, however, only for a few seconds at most and are inaudible. The D. silvestris strain, used in the present study, represents the two-row (tibial bristle rows) population of D. silvestris (22). On the basis of molecular and morphological data this population is regarded to be the oldest D. silvestris population closest to D. planitibia (23). The signal-response sequences ofthree "three-row" D. silvestris stocks analyzed in another study are similar to the sequences of the stock presented here (A.H., unpublished data). This indicates that the responses of the flies to the signals emitted by the opposite sex do not change readily during laboratory culture. To conclude, the main reason for asymmetric sexual isolation between D. planitibia and D. silvestris was found to be the loss of between-individual behavioral transitions. A possible explanation for isolation in both directions is a change in the behavior of the male during the HUW position. Thus, it is demonstrated that the loss of courtship elements (between-individual transitions) in the derived species can lead to asymmetric sexual isolation as Kaneshiro (1) has suggested. Occasionally, the species may also add new signals in their courtships (24), which must affect the whole signalresponse chains of the species. How this kind of change affects the sexual isolation between the ancestral and derived species remains to be studied. We thank Dr. Philip Welbergen for inspiring discussions and for the behavior analysis software. We also thank Kelvin Kanegawa for rearing flies for the present study and Kevin Kaneshiro for preparing the figures. This work has been supported by grants from Asla/ Fulbright commission and from The Academy of Finland to A.H. Proc. Natl. Acad. Sci. USA 90 (1993) 5817 1. Kaneshiro, K. Y. (1976) Evolution 30, 740-745. 2. Kaneshiro, K. Y. (1989) in Genetics, Speciation, and the Founder Principle, eds. Giddings, L. V., Kaneshiro, K. Y. & Anderson, W. W. (Oxford Univ. Press, Oxford). 3. Powell, J. R. (1978) Evolution 32, 465-474. 4. Arita, L. H. & Kaneshiro, K. Y. (1979) Proc. Hawaii. Entomol. Soc. 13, 31-34. 5. Ahearn, J. N. (1980) Experientia 36, 63-64. 6. Ohta, A. (1978) Evolution 32, 485-492. 7. Kaneshiro, K. Y. (1980) Evolution 34, 437-444. 8. Kaneshiro, K. Y. (1983) Annu. Rev. Entomol. 28, 161-178. 9. Spieth, H. T. (1951) Behavior 3, 105-145. 10. Lande, R. (1981) Proc. Natl. Acad. Sci. USA 78, 3721-3725. 11. Heisler, I. L. (1984) Evolution 38, 1283-1295. 12. Boake, C. R. B. (1989) Ethology 80, 318-329. 13. Bennet-Clark, H. C. (1984) J. Exp. Biol. 108, 459-463. 14. Hoy, R. R., Hoikkala, A. & Kaneshiro, K. Y. (1988) Science 240, 217-219. 15. Welbergen, P., Spruijt, B. M. & van Dijken, F. R. (1992) J. Insect Behav. 5, 229-244. 16. Welbergen, P., van Dijken, F. R. & Scharloo, W. (1987) Behavior 101, 253-274. 17. Liimatainen, J., Hoikkala, A., Aspi, J. & Welbergen, P. (1992) Anim. Behav. 43, 35-48. 18. Everitt, B. S. (1977) The Analysis of Contingency Tables (Chapman & Hall, London), p. 127. 19. Spieth, H. T. (1978) Evolution 32, 435-451. 20. Lambert, D. M. (1984) J. Theor. Biol. 109, 147-156. 21. Stalker, H. D. (1942) Genetics 27, 238-257. 22. Carson, H. L. & Bryant, P. J. (1979) Proc. Natl. Acad. Sci. USA 76, 1929-1932. 23. DeSalle, R. & Giddings, L. V. (1986) Proc. Natl. Acad. Sci. USA 83, 6902-6906. 24. Hoikkala, A., Hoy, R. & Kaneshiro, K. (1989) Anim. Behav. 37, 927-934.