4121 The Journl of Experimentl iology 214, 4121-4132 2011. Published by The Compny of iologists Ltd doi:10.1242/jeb.036954 RESEARCH ARTICLE The role of vision in odor-plume trcking by wlking nd flying insects Mrk A. Willis*, Jennifer L. Avondet nd Elizbeth Zheng Deprtment of iology, Cse Western Reserve University, Clevelnd, OH 44106, USA *Author for correspondence (mw27@cse.edu) Accepted 6 September 2011 SUMMARY The wlking pths of mle cockroches, Periplnet mericn, trcking point-source plumes of femle pheromone often pper similr in structure to those observed from flying mle moths. Flying moths use visul-flow-field feedbck of their movements to control steering nd speed over the ground nd to detect the wind speed nd direction while trcking plumes of odors. Wlking insects re lso known to use flow field cues to steer their trjectories. Cn the upwind steering we observe in plume-trcking wlking mle cockroches be explined by visul-flow-field feedbck, s in flying moths? To nswer this question, we experimentlly occluded the compound eyes nd ocelli of virgin P. mericn mles, seprtely nd in combintion, nd chllenged them with different wind nd odor environments in our lbortory wind tunnel. They were observed responding to: (1) still ir nd no odor, (2) wind nd no odor, (3) wind-borne point-source pheromone plume nd (4) wide pheromone plume in wind. If wlking cockroches require visul cues to control their steering with respect to their environment, we would expect their trcks to be less directed nd more vrible if they cnnot see. Insted, we found few sttisticlly significnt differences mong behviors exhibited by intct control cockroches or those with their eyes occluded, under ny of our environmentl conditions. Working towrds our gol of comprehensive understnding of chemo-orienttion in insects, we then chllenged flying nd wlking mle moths to trck pheromone plumes with nd without visul feedbck. Neither wlking nor flying moths performed s well s wlking cockroches when there ws no visul informtion vilble. Key words: orienttion, pheromone, vision, cockroch, behvior. INTRODUCTION It is well known tht flying insects control their steering nd flight speed using visul-flow fields cused by their movements through the environment (Srinivsn et l., 1999; Wrrnt nd Dcke, 2011). When trcking wind-borne plumes of odor, these visul control systems lso provide informtion on wind speed nd direction (Kennedy, 1939; Kennedy nd Mrsh, 1974) nd re modulted by olfctory informtion (hndwt et l., 2010; Chow nd Frye, 2008; Duistermrs nd Frye, 2008; Olberg nd Willis, 1990; Willis nd Avondet, 2005). However, reltively fewer studies hve ddressed the role of visul cues in the control of mneuvering while wlking, nd even fewer hve ddressed the role of vision in odor trcking while wlking (ell nd Tobin, 1981; Götz nd Wenking, 1973; Tobin, 1981; Znker nd Collett, 1984). Flying mle moths trcking pheromone plumes re not in physicl contct with the ground nd re thought to use windinduced drifting of the visul-flow-field feedbck cused by their movement through the environment to determine the wind speed nd direction (Dvid, 1986; Kennedy nd Mrsh, 1974; Mrsh et l., 1978). Mechnosensors on their bodies (e.g. ntenne nd wind-sensitive hirs) re thought to be ble to provide only informtion bout the movement of the nimls with respect to the ir (Dvid, 1986; Mrsh et l., 1978). Recent studies suggest tht the ntenne might lso serve s Coriolis force detectors for rpid flight stbiliztion (Sne et l., 2007). Steering with respect to the wind using visul feedbck from the environment is known s optomotor nemotxis (Kennedy, 1939). For wlking plumetrcking insects, constnt contct with sttionry reference (i.e. the ground) mens tht wind-deflected mechnosensors cn provide continuous informtion on wind speed nd direction (ell nd Krmer, 1979). Reltive to flight, much less is known bout the role of visul informtion in the control of wlking in insects, especilly in those tht trck wind-borne odor. Most observtions nd experiments hve shown tht, unlike flying insects trcking odor, the trnsltory component of optic flow hs little or no effect on the speed of wlking insects (Götz nd Wenking, 1973; Znker nd Collett, 1984). However, these sme, nd other (Weber, 1990; Weber et l., 1981), studies demonstrted tht rottionl optic flow cn be used to compenste for steering errors (Götz nd Wenking, 1973; Struss et l., 2001; Struss et l., 1997; Znker nd Collett, 1984). Recent results show tht the sme wide-field motion-sensitive cells thought to support flight orienttion in other flying insects re involved in processing rottionl imge flow in Drosophil melnogster flies when they re wlking (Chippe et l., 2010) nd flying (Mimon et l., 2010). Optic flow cn lso be used to judge the distnce trveled in wlking nd flying socil insects, which must return to their nest fter forging (Roncher nd Wehner, 1994; Srinivsn et l., 2000). In wlking nts, for exmple, the optic flow below the niml is importnt (Roncher nd Wehner, 1994), wheres tht to the sides ppers to hve no impct on speed control or distnce estimtion (Roncher et l., 2000). y contrst, honey bees use optic flow in their lterl fields of view to judge how fr they fly from forging sites to the hive (Srinivsn et l., 2000). Recent studies hve shown tht cockroches cn use visul informtion to control mny spects of their behviors. Individuls of the Americn cockroch, Periplnet mericn, re less likely
4122 M. A. Willis, J. L. Avondet nd E. Zheng to collide with visible trget thn trnsprent trget (b et l., 2010), nd light levels detected by the visul system of the tropicl cockroch lberus discoidlis influence the decision of whether to climb under or over n obstcle (Hrley et l., 2009). In ddition, drkened hiding plces re locted using visul cues by P. mericn (Okd nd Toh, 1998). Finlly, the positions of fixed visul cues hve been shown to be lerned nd then used s nvigtionl lndmrks by P. mericn to locte visully undetectble resources (Mizunmi et l., 1998). We re unwre of ny published observtions or experiments specificlly ddressing the optomotor responses of P. mericn or ny other cockroch species. However, study of neurl regenertion in P. mericn used the response to rotting striped drum s n ssy of post-lesion recovery in the brin (Drescher, 1960). Likewise, in n ongoing study of brin ctivity in nother cockroch species,. discoidlis, neurl nd behviorl responses to moving striped ptterns show cler directionlly selective neurl responses nd ssocited turning behvior (Kthmn et l., 2011). Our working hypothesis is tht P. mericn mles trcking odor plumes through their environment use visul cues (e.g. optic flow or lndscpe cues) to steer more precisely to the odor source. Rther thn removing visible cues from our experimentl environment, we physiclly occluded the two visul systems of our P. mericn mles. These included the lrge multifceted compound eyes tht detect motion nd ptterns, nd pir of ocelli (simple eyes) tht seem to be incpble of imge formtion nd re thought to function minly in detecting chnges in light intensity (Mizunmi, 1994). If visul cues re importnt to the control of steering mneuvers while trcking plumes of ttrctive odors, we predicted tht individuls with occluded eyes should show incresed errors in their steering, possibly resulting in them turning more often, steering less directly towrds the source or filing to locte the source. To determine the importnce of visul informtion to odor-plume trcking in wlking P. mericn mles, we compred the plumetrcking performnces of intct control cockroches (uncovered eyes) with those of individuls for which the eyes hd been covered by pinting: (1) the compound eyes only, (2) ocelli only nd (3) both sets of eyes. We first imed to determine whether vision ws importnt to wind orienttion by observing the steering response of the cockroches when exposed individully to: (1) still ir nd no odor, (2) wind nd no odor or (3) wind-borne odor plume. Two dditionl experiments were imed t determining the importnce of visul input on upwind steering during sustined plume trcking. In the first, we introduced intct control cockroches nd those with their eyes pinted into wind-borne plume of femle sex-pheromone nrrow enough for their ntenne to spn. In the second, we chllenged nother group of similrly mnipulted cockroches to trck pheromone plume issuing from source ~20 times the width of the nrrow plume. This second plume is t lest twice s wide s the spn of the ntenne of mle cockroch. The im of this tretment ws to mke it difficult for the cockroches to use the high-contrst olfctory lndmrk provided by the edge of the plume. We resoned tht, if vision is importnt to steering control in wlking plume-trcking cockroches, but they cn compenste for its loss by using the edges of the plume s olfctory cues for steering, then we expect the steering precision of eye-pinted mles to be worse when wlking fr from the edges of our wide plume. y contrst, if they cn perform odor-gted nemotxis without visul inputs (i.e. directionl cues provided solely by mechnosensory inputs), then their odor-ctivted orienttion to the wind direction should cuse them to orient nd wlk directly upwind in the plume. ecuse of the importnce of vision for detecting the speed nd direction of the wind during plume trcking, it hs been ssumed (but never tested) tht removing the visul input from flying moth will mke it unble to trck n odor plume. To gin more complete understnding of the importnce of visul informtion to odor trcking in moths, we compred the bility of intct controls nd moths, with their eyes pinted s in the cockroch experiments, to trck pheromone plume while wlking nd flying. For these studies, we used mles of the tobcco hornworm moth, Mnduc sext, from our lbortory colony. As in our cockroch experiments, the mle moths were either controls with intct visul systems or hd their eyes covered with pint to eliminte their possibility of detecting visul informtion while ttempting to trck n odor plume. MATERIALS AND METHODS Insects Cockroches We collected penultimte-instr mle P. mericn (L.) from our lbortory colony nd llowed them to molt to dults isolted from femles. They were housed in plstic continers with wter nd chicken feed d libitum until sexul mturtion (t lest 3 weeks old). The continers were held in n environmentl chmber on 12 h:12 h light:drk cycle, t 27 C nd 50% reltive humidity, until the nimls were used in experiments. Moths The mle moths, M. sext (L.), used in this study cme from the Willis lbortory colony. Lrve were rered on n rtificil diet (ell nd Jochim, 1976) d libitum nd mintined t ~27 C on 14 h:10 h light:drk cycle. Mture pupe were removed from the colony, seprted ccording to sex, nd the mles held in n environmentl chmber t ~27 C on 14 h:10 h light:drk cycle. Three-to-four-dy-old virgin mle moths were used in our experiments. Wind tunnel Cockroches Mle cockroches were individully relesed onto our 0.92 1.52 m flt luminum experimentl ren, which ws held 25.4 cm bove the floor of the 1 1 2.5 m working section of our lbortory wind tunnel. In ll experiments with n odor stimulus, the source ws held 1 cm bove the upwind end of the experimentl ren, with the wind speed set t 25 cm s 1. The pheromone plume ws removed from the room nd the building by mens of n exhust system positioned t the downwind end of the wind tunnel. In ll experiments using point-source plume, 0.1 ng of ( )-periplnone- (Kithr et l., 1987; Kuwhr nd Mori, 1990) ws pplied to 0.7 cm dimeter disk of filter pper (Whtmns No.1). Solutions of ( )-periplnone- were mde with n-hexne (Acros Orgnics, Geel, elgium). To generte turbulent plume, the plne of the filter pper disk odor source ws oriented perpendiculr to the ir flow in the wind tunnel. The wide plume tretment ws generted by incresing the source to strip of filter pper (14.3 0.7 cm). This increse in width incresed the surfce re of the source ~25 times, which we compensted for by incresing the dosge of pheromone solution proportiontely. This resulted in the point nd wide source both bering ~0.05 0.1 ng cm 2 of ( )-periplnone-. We did not control the visul environment during these experiments. The working section of the wind tunnel is trnsprent Plexigls. The white-pinted wlls nd ceiling of the room s well s the cmer mounts nd infrred light sources would hve been visible to nimls in the wind tunnel. Aside from the elevted luminum
Visul inputs to plume trcking 4123 ren used in the wlking studies, ll spects of the wind tunnel were the sme for cockroch nd moths trils. We video-recorded the wlking trcks of individul cockroches t 30 frmes s 1 with urle TC355AC blck-nd-white cmer positioned overhed. This provided n overhed view of the entire experimentl ren. Moths Wlking mle moths were relesed into the sme wind tunnel used for the cockroch experiments. When wlking moths were the subject of the experiment, their performnce ws observed on the sme ren s tht used by the wlking cockroches. When flying moths were the subjects, the wlking ren ws removed. The sme cmer nd wind tunnel used in the cockroch experiments were used to record both the wlking nd flying moth performnces. We used freshly mde extrct of virgin femle pheromone glnds s our ttrctnt odor source (Willis nd Arbs, 1991) nd pplied it to the sme type of filter pper disk used in the point-source plume cockroch experiments. Wind speed for the wlking moths ws held t 25 cm s 1, s in the cockroch studies. The flying moths trcked plumes in 100 cm s 1 wind, the stndrd wind speed used in our studies of M. sext plume-trcking flight (Rutkowski et l., 2009; Willis nd Arbs, 1991; Willis nd Arbs, 1998). Visuliztion of odor plumes To visulize the odor plume, we video-recorded titnium tetrchloride smoke plumes issuing from filter pper sources of the sme size nd shpe s used to distribute pheromone in our behvior experiments. This is stndrd procedure for visulizing odor plumes in insect pheromone studies (ker et l., 1984; Chrlton et l., 1993; Willis nd Avondet, 2005). We then determined the time-verged plume boundries by using our motion-nlysis system to digitize smoke pckets t the lterl mrgins of the plume. As in previous studies (Willis nd Avondet, 2005), overlying digitized cockroch trcks on these time-verged plumes enbled us to view the turns in the wlking pths of the cockroches with respect to the time-verged plume boundries. Experimentl design Cockroches We covered the eyes of cockroches by pinting: (1) compound eyes only (N 27 for wind nd odor orienttion experiment, N 20 for the point-source plume experiment nd N 9 for the wide plume experiment), (2) ocelli only (N 30 for the wind nd odor orienttion experiment, N 20 for the point-source plume experiment nd N 8 for the wide plume experiment) nd (3) both sets of eyes (N 27 for the wind nd odor orienttion experiment, N 22 for the point-source plume experiment nd N 9 for the wide plume experiment). Intct cockroches with unpinted eyes were used s controls (N 29 for the wind nd odor orienttion experiment, N 19 for the point-source plume experiment nd N 10 for the wide plume experiment). We pinted the eyes of our experimentl nimls using methods previously used for P. mericn behvior experiments (Ye et l., 2003). First, we removed the lyer of oil on the cuticle of the eyes nd ocelli of the cockroches with n cetone-dmpened (nd lmost dried) cotton pplictor. Then, to ensure tht the eyes nd/or ocelli were covered, we pplied lyer of red enmel pint (Shrpie pint pen) until the blck compound eyes (especilly long the edges) could no longer be seen. Finlly, when the red pint dried (~30 s), lyer of blck crylic pint ws pplied over the red to completely block the light. After ech cockroch ws pinted, it ws plced individully in lbeled luminum window-screen relese cge corresponding to the specific tretment. This method ws shown to eliminte lmost completely the ntennl orienttion response of P. mericn to visul trget (Ye et l., 2003). Furthermore, our own observtions show tht mles in tretment groups with pinted compound eyes show no obvious response to shining red-filtered flshlight in their eyes, wheres unpinted control mles turn to fce wy from the light. Mles in our tretment groups with pinted compound eyes lso showed no response when we pproched to recpture them fter their trils in the wind tunnel. Unpinted control mles lwys oriented nd wlked wy from our pproches. The cockroches in relese cges were then plced into the wind tunnel room, under our stndrd low-light conditions (~3 lx), until experiments begn ~2 h into their scotophse, the beginning of their period of pek pheromone response. To ensure tht the pint hd not been groomed off the eyes of the cockroches, we exmined them under dissecting microscope fter ll pinting ws completed nd before they were plced in the wind-tunnel room. We lso verified the presence of pint on the cockroches by checking their eyes gin under the dissecting microscope fter experiments ech dy. If ny of the eyes ws not completely covered fter the trils, we removed the responses of such individuls from our smple. This occurred in ~10% of the cockroches with both sets of eyes pinted, 10% of the individuls with their compound eyes only pinted nd 50% of the cockroches with only the ocelli pinted. In ech of the three experiments presented here, we used ~16 cockroches ech dy; four unpinted controls, four with pinted compound eyes, four with pinted ocelli, nd four with both pinted compound eyes nd ocelli. In the initil experiment imed t understnding the effect of the loss of vision on the bility to orient to wind, nd odor plus wind, we plced ech cockroch in the center of the ren to llow them freedom to wlk in ny direction. For the two experiments focused on plume trcking, ech cockroch ws plced into the odor plume, t the center of the downwind end of the experimentl ren fter the pheromone source ws positioned nd the wind nd exhust were turned on. Ech cockroch ws rndomly selected for plcement into the ren. To cclimte the cockroch to the experimentl environment, it ws held in its relese cge on the ren in the wind tunnel for 1 min. The behviors of the cockroch were video-recorded fter it ws relesed. A cockroch ws scored s successful when it touched the pheromone source. Trils in which pheromone ws not used were stopped when the cockroch wlked to one of the edges of the ren. When cockroch did not leve the relese point within 5 min of being given ccess to the ren, it ws scored s non-responding. We performed pilot study to verify tht olfctory inputs from the ntenne were mediting the orienttion behvior tht we were studying. The behviorl responses of intct control cockroches were compred with those tht hd both ntenne surgiclly removed t their bses. None of the mles (0%) tht hd their ntenne removed before introduction into wind-borne pheromone plume left the relese point within the 5 min observtion period, wheres ll of the intct controls (100%) immeditely trcked the pheromone plume to its source upon relese into the ren (N 10 in both cses). Thus, mle cockroches with their ntenne removed do not respond to femle pheromone or the mbient wind direction. A similr lck of behviorl responses ws observed when both ntenne were coted with pint. Moths The ocelli of M. sext re inside the hed, well below the surfce of the cuticle (Eton, 1971). For this reson, we were only ble to pint the compound eyes, using the methods detiled bove for
4124 M. A. Willis, J. L. Avondet nd E. Zheng cockroches. It is known tht M. sext mles fly upwind to femle pheromone redily nd flight ppers to be their preferred mode of locomotion. It is difficult to mke them wlk once they hve tken flight. Therefore, to ssess their bility to orient to pheromone plume while wlking, we surgiclly removed pproximtely onehlf to two-thirds of their wings. These individuls were cold nesthetized by plcing them in cler plstic cup on crushed ice until they stopped moving (~15 20 min). Once nesthetized, both sets of wings were removed using surgicl scissors. Once the experimentl moths were prepred, either by hving their wings removed, or eyes pinted, or both, they were plced into individul screen cges nd plced in the wind-tunnel room, where they experienced lights off ccording to their norml light drk cycle. They were chllenged to trck pheromone plumes, either wlking or flying, two hours fter lights off during their pek responsiveness to pheromone (Sski nd Riddiford, 1984). These introductory studies of flying nd wlking plume trcking in moths serve the long-term gol of our lbortory to develop comprehensive understnding of chemo-orienttion in insects by mking explicit comprisons between wlking nd flying odor trckers. Dt nlysis Cockroches We used the computerized motion-nlysis system clled Motus (Vicon Pek, Englewood, CO, USA) to digitize nd mesure the wlking trjectories of the mle cockroches. Motus splits ech 1/30 s video frme into two video fields, yielding 1/60 s temporl resolution. For ll experiments, we digitized the wlking trjectories of the cockroch by mrking its position every fifth time-point (i.e. every 83 ms). At ech digitized cockroch position, we mrked the center of the hed nd the distl tip of the bdomen. A line drwn from the cockroch relese point in the center of the ren through the point where ech individul encountered the edge of our experimentl ren defined their vnishing direction. We then generted men vnishing direction for ech tretment group. These men directions (i.e. ngle ) were compred with circulr rndom distribution using Ryleigh s test (Cbrer et l., 1991). In ddition to determining whether this men vnishing direction is significntly different from rndom, this test provides the reltive length for men vector (r) distributed between 0 (i.e. no movement in the men direction) nd 1.0 (i.e. ll individuls vnished in the men direction). These ngles were mesured with respect to the wind direction (0 deg), which is prllel to the longitudinl xis of the wind tunnel. We lso compred the men responses (i.e. r nd ) of cockroches with those of visul-system mnipultions to our experimentl environments with Wtson Willims F-test using commercilly vilble circulr sttistics softwre clled Orin (Rockwre, Golden, CO, USA), to determine whether they were sttisticlly different. In ddition to the orienttion of the vnishing direction, we mesured eight different response vribles from the movement vectors defined by successive positions of the heds of the cockroches digitized from the video-recorded trcks: the net velocity (the speed of the cockroch, if it hd trveled in stright line, from the beginning to the end of the trck), ground-speed (the speed of the cockroch long its trck), trck ngle (the ngle of ech movement vector with respect to the wind direction), bodyxis ngle (the ngle of the body of the cockroch with respect to the wind direction), trck width (the distnce between the pex of ech turn perpendiculr to the wind direction), inter-turn durtion (the time between the pex of ech turn) nd the number of turns per second for ech trck. Potentil turns were identified s the locl extremes in lterl position (i.e. positions left or right of their immeditely djcent digitized points in the wlking trcks). We then clculted turn identifiction distnce, which ws the product of vrible turn threshold vlue we set nd the overll width of ech trck. The overll width ws the distnce, perpendiculr to the wind, between the leftmost nd rightmost position of the cockroch over its entire trck. If the distnce, perpendiculr to the wind, between the cndidte turn nd the next djcent turn in the opposite direction ws longer thn the turn identifiction distnce, it ws ccepted s turn. If shorter, it ws rejected. The turn threshold vlue ws set ccording to visul comprison between the computeridentified turns nd the cockroch trcks. A turn threshold vlue of 0.5 worked for ll of the trcks. All of the bove trck prmeters were clculted with custom-written script in MATLA (The MthWorks, Ntick, MA, USA) (Rutkowski et l., 2009). We lso mesured the number nd durtion of stops mde by ech individul during plume-trcking performnce. The cockroch ws sid to be stopped if t lest two successive digitized positions were the sme. For the experiment with the wide odor plume, we lso clculted linerity index (degree of strightness of the trck) nd ngulr velocity (degrees turned through per second). We clculted our linerity index by dividing the totl length of ech wlking trck into the stright-line distnce between its first nd lst digitized position. During study of the experimentl designs used in olfctory orienttion studies, including those in our lbortory, it ws shown tht our design met ll of the ssumptions of nlyses of vrince (Pill et l., 2005). This study lso demonstrted uniquely pproprite ANOVA procedure for this design (Pill et l., 2005), which we hve used here. First, we verged the mesurements for ech individul cockroch nd then clculted grnd men of ll of the individuls in tretment group. We then conducted n nlysis of vrince (ANOVA) to identify ny sttisticl differences mong the grnd mens. When the ANOVA reveled significnt differences, we performed post hoc Tukey s test to determine which tretments nd prmeters differed significntly (P<0.05). In the point-source plume experiment, there were smll number of individuls in ech tretment group (no pint, N 6; compound only, N 3; ocelli only, N 5; both sets, N 5) tht generted long looping trcks before they locked on to the plume nd trcked it to the source. In ll cses, we nlyzed only the sections of the plumetrcking performnce fter ech individul hd initited plumetrcking response tht led to the source. We conducted post hoc power nlyses with the commercilly vilble softwre SigmStt (Aspire Softwre Interntionl, Ashburn, VA, USA) on our ANOVAs t 0.80 level (the ccepted convention for this sort of nlysis) to determine the smple sizes necessry to minimize the likelihood of our nlyses producing flse negtives (i.e. type II errors). In ll cses, the smple sizes in the experiments we conducted were in the rnge tht would revel sttisticlly significnt differences if ny existed nd would minimize the probbility of rejecting rel differences s not significnt. Moths For the purpose of the experiments presented here, our focus ws primrily on whether mle moths would trck n odor plume while wlking or flying with their eyes pinted. We compred the proportion of intct control individuls with unpinted eyes (N 20) with those with pinted eyes (N 19) tht were ble to trck the plume to the source. The sme comprison ws performed on flying mles with unpinted eyes (N 23), compring them with those with pinted eyes (N 16). We then pplied Ryn s multiple comprison test for
Visul inputs to plume trcking 4125 proportions (Ryn, 1960) to determine whether our response proportions were sttisticlly different. We lso video recorded the trcking performnces of the moths to visulize their movement trjectories, but complete trck nlysis is beyond the scope of the study presented here. A A C 0 0 270 * 180 RESULTS Cockroch orienttion to wind nd odor Whether intct or with their eyes pinted, P. mericn mles responded to the conditions in our experimentl ren in brodly similr mnner. Of the 126 cockroches tested, 113 (90%) left the relese cge nd either wlked to n edge of the experimentl ren or trcked the odor plume, depending on the experimentl tretment (Fig. 1). Of the 13 individuls tht did not move from the relese point, eight were in the no wind nd no odor tretment, three in the wind nd no odor nd two in the wind nd odor tretment. All individuls remining on the relese point hd their compound eyes pinted over. Hlf of these were from the group tht hd only their compound eyes pinted, nd the other hlf hd both sets of eyes pinted. Our nlysis of these results ws stepwise. First we pplied Ryleigh s test to determine whether the vnishing directions of the cockroches were sttisticlly different from rndom circulr distribution. Then we pplied the Wtson Willims F-test to determine whether there were ny sttisticl differences in the verge responses to our experimentl environments or visulsystem mnipultions. Consistent with previous results (Willis nd Avondet, 2005), the vnishing directions of the control cockroches with unpinted eyes were significntly upwind (i.e. 0 deg) when in wind nd odor ( 0.5 deg, r 1.0, P 0.05), nd downwind when exposed to wind only ( 166.7 deg, r 0.59, P 0.05). Control cockroches wlking in still ir with no pheromone present left the experimentl ren in so mny different directions tht there ws no sttisticl difference from rndom ( 50.7 deg, r 0.19, P 0.05; Fig. 1A). On verge, cockroches with experimentl visul-system mnipultions generted men vnishing directions similr to those of the controls, leving the ren in upwind directions when in wind nd odor (compound eyes pinted, 6.2 deg, r 0.91; ocelli pinted, 355.8 deg, r 0.99; both sets of eyes pinted, 2.2 deg, r 0.78; P 0.05), downwind in wind only (compound eyes pinted, 174.5 deg, r 0.71; ocelli pinted, 154.9 deg, r 0.72; both sets of eyes pinted, 143.3 deg, r 0.55, P 0.05) nd in mny different directions in the bsence of wind or odor (compound eyes pinted, 108.9 deg, r 0.58; ocelli pinted, 233.8 deg, r 0.15; both sets of eyes pinted, 173.0 deg, r 0.28; P 0.05; Fig. 1 D). The Wtson Willims F-test uses the men direction ( ) nd vector lengths (r) clculted bove to test for sttisticl differences between tretment groups (Fisher, 2000). The results of our nlysis confirm the visul impression of the circulr plots in Fig. 1. First, we compred the responses of ech of the visul-system mnipultions with the three experimentl environments we used. Individuls in ech of our four visul-system mnipultions groups generted men vnishing directions in the wind-plus-odor environment tht were significntly different (P 0.05) from those in the wind-plus-no-odor environment (Fig. 1). The men vnishing directions in the no-wind-plus-no-odor environment were less consistent. Even though their low r vlues indicted generl lck of orienttion in the men direction, if the men direction ws sufficiently distinct from the wind-plus-odor or wind-plus-no-odor tretments, the nlysis indicted it ws significntly different (Fig. 1A,). However, if the vlue for men direction ws close to C D 0 A 0.5 1.0 90 270 A * 0 180 C 90 270 0 A 90 270 90 No wind, no odor 180 180 Wind (when present) Wind, no odor Wind + odor Fig. 1. Wlking orienttion of Periplnet mericn mles in wind plus odor (red), wind nd no odor (green) nd no wind nd no odor (blue) for cockroches with (A) unpinted eyes (control), () only compound eyes pinted, (C) only ocelli pinted or (D) both sets of eyes pinted. Ech dot on the outer circumference of the circle plot represents the direction t which n individul cockroch encountered the edge of the ren. Note tht, for ech plot, dots representing severl of the wind-plus-odor trils re plotted on top of ech other ner ʻ0ʼ. The length of ech colored rrow indictes the men resultnt length (r) ssocited with the men direction of the vnishing ngle ( ) s determined by Rleighʼs test. Plese refer to the text for vlues for nd r for ech combintion of eye-pinting tretment nd environmentl mnipultion. The lengths of the men vectors (r) re n index of how closely the vnishing directions of the individuls in the popultion re clumped. This index rnges from 0 to 1.0, with 0 being equivlent to circulr rndom distribution, nd 1.0 indicting tht ll individuls encountered the edge of the ren in the sme direction. The rdil xis in the circulr plot in (A) hs been mrked to represent this rnge, nd ll circulr plots in this figure re consistent. Results of Wtson Willims F- test compring: (1) the responses to the three environmentl tretments (i.e. wind plus odor, wind plus no odor, etc.) by individuls with ech visulsystem mnipultion re depicted with uppercse letters nd (2) the effects of visul-system mnipultions (i.e. intct controls, pinted compound eyes, etc.) on the responses of individuls to ech environment re depicted in typogrphic symbols (i.e. sterisk, section symbol nd dgger). Men vectors with uppercse letters or typogrphic symbols in common re not significntly different from ech other. There were no sttisticlly significnt differences in responses of individuls with ny of the visul-system mnipultions to the wind-plus-odor or wind-plus-no-odor environments. For the plots of the individul wlking trcks on the righthnd side of the figure, the rrowheds indicte the orienttion of the cockroch t the end of ech trck in the lbortory wind tunnel. The bold blck circle represents the relese cge.
4126 M. A. Willis, J. L. Avondet nd E. Zheng Tble 1. Men ± s.d. of trck prmeters mesured from Periplnet mericn mles wlking in different combintions of wind nd odor in lbortory wind tunnel with different eye pinting tretments Net velocity Groundspeed Trck ngle* ody-yw ngle* Trck width Interturn Number Stop Environment Tretment + N (cm s 1 ) (cm s 1 ) (deg) (deg) (cm) durtion (s) of stops durtion (s) No wind, no odor Not pinted 8 4.60±2.68 8.07±3.48 63.22±29.49 b,c 62.83±39.68,b,c 4.42±3.87 2.52±2.82,b 22.75±32.80,b 0.28±0.15 Compound eyes 8 3.10±3.23 10.06±11.14 74.60±19.31 b,c 86.68±35.84,b 13.80±17.75 b 7.06±8.39 b 35.75±38.29,b 0.25±0.11 Ocelli 10 4.17±5.31 8.28±6.07 88.22±32.42,b 108.56±38.66 1.96±1.94 1.78±1.65 22.70±29.92,b 0.23±0.09 Ocelli nd compound eyes 9 5.05±7.86 11.13±10.40 86.03±33.83,b 82.05±46.63,b 8.08±9.87,b 4.67±5.68,b 46.22±77.51 0.28±0.18 Wind, no odor Not pinted 11 12.26±20.39 15.89±25.15 102.57±35.83 92.47±68.35 2.15±1.92 1.17±0.83 9.36±6.38,b 0.29±0.19 Compound eyes 9 6.90±4.06 8.62±3.48 117.75±33.77 89.52±70.00,b 2.38±1.90 1.49±1.69 9.78±9.91,b 0.21±0.09 Ocelli 11 6.21±3.47 8.36±3.14 99.41±36.54 85.46±61.76,b 2.44±1.29 1.47±0.73 10.81±12.73,b 0.38±0.20 Ocelli nd compound eyes 9 2.98±2.32 12.57±21.05 89.17±33.34,b 97.47±49.90 6.95±7.48,b 5.11±5.24,b 27.55±27.47,b 0.30±0.09 Wind, odor Not pinted 10 16.04±6.64 18.09±5.38 34.42±37.75 c 26.58±31.38 c 1.93±1.07 0.53±0.28 0.5±0.71 b 0.12±0.08 Compound eyes 10 9.08±7.09 10.97±5.59 37.42±19.58 c 34.29±26.37 b,c 4.04±4.11 1.65±1.72 5.20±4.66,b 0.32±0.22 Ocelli 9 10.19±5.90 15.23±4.13 42.35±25.46 c 48.41±45.12 b,c 4.45±3.81 1.06±1.18 4.55±6.52,b 0.26±0.22 Ocelli nd compound eyes 9 6.11±2.84 9.67±3.63 59.56±30.64 b,c 65.87±52.53,b,c 3.43±1.91 1.25±1.08 10.00±12.68,b 0.25±0.09 Vlues in ech column with different letters re significntly different ccording to two-wy ANOVA (P 0.05) nd post hoc Tukeyʼs test. + Eyes pinted before experimenttion. *All ngles depicted re bsolute vlues of the distributions mesured. those for one of the other tretments, they were not significntly different (P 0.05) (Fig. 1C,D). As none of the men vnishing directions generted in response to the no-wind-plus-no-odor environment were significntly different from circulr rndom distribution ccording to the Ryleigh s test, we put little weight in the sttisticl significnce or lck thereof in the no wind-no-odor environment. There were no sttisticlly significnt differences (P 0.05) mong the men vnishing directions of the four visul-system mnipultions in the wind-plus-odor nd wind-plus-no-odor environments (Fig. 1). There were sttisticlly significnt differences (P 0.05) in the men vnishing directions generted by the four visul-system mnipultions in the no-wind-plus-no-odor environment (Fig. 1). ut, s with the environment-specific comprisons bove, we put little weight in these results becuse the men vnishing directions generted by ech of the four visulsystem mnipultions in the no-wind-plus-no-odor environment were not significntly different (P 0.05) from circulr rndom distribution ccording to Ryleigh s test. Anlysis of the wlking trcks of these individuls reveled brod similrities mong the cockroches with the four visul-system mnipultions s they responded to the three experimentl environments. Intct nd eye-pinted cockroches generted, on verge, similr net velocities, similr wlking speeds long their trcks nd similr men stopping durtions when they stopped, regrdless of the wind nd odor environment (P>0.05; Tble 1). Our nlysis did revel sttisticlly significnt differences in the men trck ngles, yw ngles, trck widths nd numbers of stops executed by the cockroches in this experiment, but there were few obvious trends of incresing or decresing men vlues cross the eye-pinting tretments or environmentl mnipultions (Tble 1). We did observe consistent trends of incresing trck nd yw ngles s the visul systems were occluded in cockroches in the windplus-odor environment (Tble 1), suggesting some effect of removing visul input on steering behvior, but these differences were not sttisticlly significnt. However, this response is noteworthy becuse steering more off of the wind direction with loss of visul inputs mtches our predictions for how steering might chnge if visul informtion were importnt during odor trcking. When wind ws introduced to the ren, most mles oriented with respect to it nd left the ren in the downwind direction, whether their eyes were pinted or not (Fig. 1). In ll cses, except cockroches with both sets of eyes pinted, individuls orienting to odor plumes in wind steered their trcks more directly upwind, s evidenced by their trck nd body-yw ngles (Fig. 1; Tble 1). Cockroches responding to wind-borne pheromone plume lso hd significntly higher rtes of turning (turns s 1 ), regrdless of whether their eyes were pinted, thn those orienting in the bsence of ttrctive odors (Tble 1). Few sttisticlly significnt differences were reveled in our nlysis of this experiment, indicting tht the men responses of our experimentl popultions were similr. Importntly, it should be noted tht, in most cses, the stndrd devitions re lso similr cross the experiment (Tble 1). Thus, the vrition of the cockroches bout their verge performnces does not increse when the eyes re pinted, indicting tht vribility in the behvior does not increse significntly with loss of visul inputs. The trcks generted by cockroches with experimentlly mnipulted visul systems responding to no-wind-plus-no-odor nd wind-plus-no-odor environments hve somewht different ppernce thn those of intct controls or ny of the cockroches trcking odor plumes (Fig. 1D). It is probble tht the interction
Visul inputs to plume trcking 4127 A C 100% to source Wind Strt 10 cm D 100% to source Fig. 2. Exmples of typicl (center) nd extreme (upper nd lower) P. mericn mles trcking plume from point-source of pheromone upwind (right to left) in wind tunnel with: (A) unpinted eyes (control), () pinted compound eyes, (C) pinted ocelli nd (D) pinted ocelli nd compound eyes. To illustrte the vribility in the trcking responses, we hve plotted three trcks from ech experimentl group. In ech pnel, the three trcks plotted represent n exmple of trck with few turns (top), trck with n intermedite number of turns (middle) nd trck with mny turns. These trcks were sorted from ech experimentl group nd selected by counting the turns in ech trck by eye. The percentge of cockroches successfully trcking the pheromone to the source is stted in the lower-left corner of ech pnel. The time-verged plume boundries of titnium tetrchloride smoke visuliztion, s viewed from bove, re illustrted in drk gry. The ple-gry envelope round the blck line representing the wlking trck is n pproximtion of the re swept by the ntenne on either side of the body. This re ws generted by determining the verge distnce n ntenn projects to the side of plume-trcking cockroch (Willis nd Avondet, 2005) nd dding it to, or subtrcting it from, ech cockroch position in the trck. In this figure, wind blows from left to right t 25 cm s 1. 100% to source 100% to source (18% did not leve relese cge nd were excluded) of wlking speeds somewht slower thn those in intct control cockroches, or those in the wind-plus-odor tretment, together with similr turning rtes, generted trcks tht looked different but yielded mesured movement prmeters tht were not sttisticlly different. Point-source plume All but four of the 81 individuls chllenged to trck the pointsource plume were successful in reching the source (i.e. 95% were successful). The four mles tht did not trck the plume hd both sets of eyes pinted. Visul inspection of the rnge of trcking performnces exhibited by our experimentl popultion revels no obvious consistent differences cused by occluding either of the visul systems (Fig. 2). Our trck nlysis showed tht, on verge, intct mles wlked long their upwind trck the fstest, wheres both tretments tht included pinting the ocelli wlked significntly slower (P>0.05; Tble 2). The men wlking speed of mles with only their compound eyes pinted ws not significntly different from tht of the other tretments. Other thn the ground-speed, our nlysis reveled no sttisticlly significnt differences ssocited with occluding the eyes. However, there re observble trends. Specificlly, individuls in tretment groups with their ocelli pinted (including those with both sets of eyes pinted) tended to generte slower net velocities nd steer greter trck ngles, on verge, thn those with only compound eyes pinted nd non-pinted controls. Those with both sets of eyes pinted lso stopped more nd for longer durtions thn the other three tretments. Wide plume Thirty-three of the 36 cockroches tht initited wlking in response to the wide plume locted the source. The few tht did not locte the source behved in wy tht suggested tht they did not detect the pheromone or responded to it only trnsiently. They left the time-verged plume re, did not initite the serch behvior typiclly performed when contct with pheromone is lost (Willis et l., 2008) nd wlked until they left the ren (Fig. 3D). As in the previous experiment, visul inspection of the trcks showed brod similrity in the movement trcks cross the experiment, with similr rnge of performnces cross ll the tretments nd the intct controls (Fig. 3). Most individuls ppered to exhibit odor-gted nemotxis, with their performnce Tble 2. Men (±s.d.) trck prmeters mesured from P. mericn mles trcking pheromone upwind with pinted eyes Net velocity Groundspeed Trck ngle* ody xis* Trck width Inter-turn Number of Stop Tretment + N (cm s 1 ) (cm s 1 ) (deg) (deg) (cm) durtion (s) stops durtion (s) Unpinted 19 16.67±8.95 23.89±4.75 20.89±9.04 16.09±8.74 3.18±1.57 0.55±0.20 0.39±0.98 0.04±0.12 Compound 20 16.82±6.28 20.69±4.45,b 21.24±6.92 16.18±11.24 2.67±1.48 0.54±0.40 0.86±2.10 0.05±0.10 pinted Ocelli pinted 20 13.72±7.87 20.10±3.95 b 25.72±10.75 20.07±11.91 2.92±1.10 0.54±0.16 0.65±1.04 0.06±0.08 Compound nd 18 11.68±7.49 19.39±5.87 b 25.50±12.61 17.21±12.21 3.70±3.91 0.64±0.42 1.39±2.03 0.09±0.15 ocelli pinted Vlues in the sme column with the sme letters do not differ significntly ccording to split-plot ANOVA (P 0.05) nd post hoc Tukeyʼs test. + Eyes pinted before experimenttion. *All ngles depicted re bsolute vlues of the distributions mesured.
4128 M. A. Willis, J. L. Avondet nd E. Zheng A Strt Fig. 3. Exmples of typicl (center) nd extreme (upper nd lower) P. mericn mles trcking wide (14.3 cm wide source) pheromone plume upwind (right to left) in wind tunnel with (A) unpinted eyes (control), () pinted compound eyes, (C) pinted ocelli nd (D) pinted ocelli nd compound eyes. Detils for this figure re the sme s those of Fig. 2. 10 cm 100% to source 100% to source C D Wind 100% to source 77% to source (10% did not leve relese cge) chrcterized by fcing into the wind nd wlking upwind embedded in the plume. A few individuls did wnder more widely nd locked on to the lterl boundry of the plume when they encountered it, trcking the high-contrst pheromone clen-ir edge to the source. Our nlysis reveled no sttisticlly significnt differences cross ll of the tretments (Tble 3). Like the pointsource plume experiment, there ws trend for the tretments with occluded ocelli to be the most different from the intct controls. Moth orienttion to wind nd odor Flying plume trcking Intct moths with unpinted eyes initited wing fnning, took flight, mde upwind progress in the plume nd eventully locted the source in mnner similr to tht reported previously (Willis nd Arbs, 1991; Willis et l., 1995). The mles tht trcked the plume generted upwind zigzgging flight trcks chrcteristic of pheromone-modulted plume trcking in mny species of moths (Fig. 4A). However, the individuls with pinted compound eyes demonstrted very different behvior (Fig. 4). Of the 23 intct moths we tested, 61% mde upwind progress in the plume, with 43% locting the source (Fig. 5). The response of the moths with pinted eyes ws very different. They often initited wing fnning upon introduction to the pheromone plume, like norml intct moth, but showed much more vrible durtion of response thn the mles with unoccluded eyes. The men durtion of wing fnning ws longer in eye-pinted mles (55.8±68.1 s) thn in intct controls (18.7±15.3 s), but not significntly so (P>0.05). When moths with their eyes pinted did tke flight, they immeditely lost ltitude, drifted downwind nd lnded on the floor of the wind tunnel, rpidly cesing wing fnning. None of these mles took flight gin. Of the 16 moths tested, none with pinted compound eyes trcked n odor plume to the source in flight (Fig. 5). Wlking plume trcking All moths with surgiclly shortened wings nd intct visul systems initited wing fnning, nd mny subsequently left the relese cge nd wlked upwind nd locted the source. The wlking plumetrcking mles ppered to move in less-coordinted mnner thn the wlking cockroches nd generted trcks with fewer counterturns (see Fig. 2A, Fig. 4). Of the 20 moths with unpinted eyes whose wings we surgiclly removed, 80% mde upwind progress, with 40% locting the source (Fig. 5). The moths with wings removed nd pinted eyes ll responded like the intct control moths by moving their ntenne forwrd nd inititing wing fnning behvior upon introduction to the plume, but the proportion mking the trnsition to the next sequentil behvior decresed t ech trnsition until only 5% locted the source (Fig. 5). Thus, lthough wlking moths with pinted eyes mde more upwind progress thn flyers, there ws no sttisticlly significnt difference in their bility to locte the source (Fig. 5). DISCUSSION In most cses, P. mericn mles with ll or prt of their visul inputs removed continued to orient to wind nd wind plus pheromone in mnner similr to tht of intct control mles nd our erlier observtions (Willis nd Avondet, 2005). The plume-trcking behvior of mles with their eyes pinted nd plced in wind-borne point-source plumes of pheromone ws lso, for the most prt, sttisticlly similr to tht of intct controls nd our erlier studies (Willis nd Avondet, 2005). When we ttempted to mke the tsk more difficult by moving the high-contrst edges of the plume further thn ntennl width prt, P. mericn mles with their eyes pinted continued to perform s well s intct controls (Fig. 3; Tble 3). In ech of these experiments, smll number of individuls behved ccording to our predictions nd showed less-precise steering (Figs 2, 3). A few of these mles even demonstrted lck of pheromonemodulted behviors. Such behvior is rrely, if ever, observed in intct mle cockroches, where 100% of the popultion typiclly trcks the plume to the source (Willis nd Avondet, 2005; Willis et l., 2008). Intct wlking nd flying M. sext mles trcked pheromone plumes to their source with lmost equl success, but removing visul input in either of these cses renders these nimls reltively incpble of performing plume-trcking behvior (Figs 4, 5).
Visul inputs to plume trcking 4129 A Tble 3. Men ± s.d. (coefficient of vrition) of trck prmeters mesured from P. mericn mles wlking in wide plume in lbortory wind tunnel Net velocity Groundspeed Trck ngle* ody-xis Trck width Inter-turn Number of Stop Linerity Angulr Tretment + N (cm s 1 ) (cm s 1 ) (deg) ngle* (deg) (cm) durtion (s) stops durtion (s) index velocity (deg s 1 ) Not pinted 10 17.23±6.04 20.77±4.62 31.82±13.64 29.65±22.77 3.67±2.59 0.61±0.41 2.80±3.58 0.23±0.14 0.77±0.16 293.20±95.80 (35.1) (22.2) (42.9) (76.8) (70.6) (67.2) (127.8) (60.9) (20.7) (32.7) Compound 9 15.96±5.71 18.61±5.22 29.71±12.14 19.93±8.23 4.56±2.14 0.85±0.60 3.77±3.0 0.21±0.11 0.79±0.16 297.17±137.68 eyes pinted (35.8) (28.1) (40.9) (41.3) (46.9) (70.6) (79.6) (52.4) (20.3) (46.3) Ocelli pinted 8 12.26±5.84 15.61±5.35 31.60±17.06 27.72±23.52 5.60±7.51 0.83±0.65 4.38±2.44 0.41±0.27 0.76±0.21 303.93±100.89 (47.6) (34.3) (54.0) (84.8) (134.1) (78.3) (55.7) (65.9) (27.6) (33.2) Ocelli nd 9 11.51±4.89 18.51±6.68 39.12±23.41 35.19±23.41 4.82±5.75 1.27±1.91 3.22±2.59 0.20±0.15 0.71±0.19 290.01±200.58 (42.5) (36.1) (59.8) (66.5) (119.3) (150.4) (80.4) (75.0) (26.8) (69.2) compound eyes pinted There re no significnt differences in ny of the vribles mong tretments ccording to repeted-mesures ANOVA (P 0.05) nd post hoc Tukeyʼs test. + Eyes pinted before experimenttion. *All ngles depicted re bsolute vlues of the distributions mesured. Source C Wind Strt 10 cm Fig. 4. Exmples of mle Mnduc sext moths trcking pheromone plume from point-source while wlking or flying, with nd without visul input. (A) Flight pth of mle trcking plume of femle pheromone with n intct visul system; () flight trck of mle plced in femle pheromone plume with n occluded visul system; (C) wlking pth of mle with its wings surgiclly removed trcking plume of femle pheromone with n intct visul system; nd (D) wlking trck of mle relesed into plume of femle pheromone with n occluded visul system. Wind speed ws 100 cm s 1 for flight nd 25 cm s 1 for wlking experiments. Moth positions were digitized ech 1/30 s while flying, nd ech 12 ms while wlking. efore these experiments, we thought tht wlking cockroches might steer their pths while trcking n odor plume more directly upwind using visul informtion bsed on known exmples of similr behvior ssocited with orienttion to distnt sources of sound (öhm et l., 1991; Weber, 1990; Weber et l., 1981). In these previous studies, the ddition of fixed visul cues enbled crickets wlking on servosphere locomotion compenstor to steer nd wlk more directly towrds fixed source of their uditory mting cll (Weber, 1990; öhm et l., 1991). Even when the visul cues were present, but not in the sme position s the sound source, crickets were ble to orient their wlking more directly towrds the speker (öhm et l., 1991). In mnner similr to these crickets, we thought tht our cockroches might steer their plume-trcking pths more directly upwind becuse of visul references to fixed visible cues in their environments. However, under the conditions of our experiments, lck of visul informtion seems to hve little effect on steering behvior during upwind orienttion to n odor plume in wlking P. mericn mles. It is possible tht the directionl informtion vilble from the wind, probbly detected by the ntenne (ell nd Krmer, 1979), together with their innte response to femle pheromone, could provide sufficient informtion to direct steering upwind nd mintin contct with the odor plume in the bsence of visul informtion. If this is the cse, then it is possible tht we simply hve not chllenged our cockroches with tsk difficult enough to require visul reference for success. Our next series of experiments will be designed to remove directionl wind cues while the cockroches re trcking plume. It is possible tht sudden loss of the primry directionl cues used to locte the source, together with the inbility to use visul inputs could cuse n increse in the vribility of their steering responses nd result in longer trcking times or n inbility to locte the source. In this cse, the cockroches would be forced to trck the plume using odor informtion lone. However, if the cockroches with occluded visul systems continued to perform s well s unpinted controls, it would suggest tht the D