The Journal of Experimental Biology 7, 7-86 Published by The Company of Biologists doi:./jeb.888 7 Stik inset loomotion in a omplex environment: limbing over large gaps Bettina Blaesing* and Holk Cruse Faulty of Biology, University of Bielefeld, Postbox, D-5 Bielefeld, Germany *Author for orrespondene (e-mail: bettina.blaesing@uni-bielefeld.de) Aepted January In a omplex environment, animals are hallenged by various types of obstales. This requires the ontroller of their walking system to be highly flexible. In this study, stik insets were presented with large gaps to ross in order to observe how loomotion an be adapted to hallenging environmental situations. Different approahes were used to investigate the sequene of gaprossing behaviour. A detailed video analysis revealed that gap-rossing behaviour resembles modified walking behaviour with additional step types. The walking sequene is interrupted by an interval of exploration, in whih the inset probes the gap spae with its antennae and front legs. When reahing the gap, loss of ontat of an antenna with the ground does not eliit any observable reations. In ontrast, an initial front leg step into the gap that often follows antennal non-ontat evokes slowing Summary down of stane veloity. An ablation experiment showed that the far edge of the gap is deteted by tatile antennal stimulation rather than by vision. Initial ontat of an antenna or front leg with the far edge of the gap represents a point of no return, after whih gap rossing is always suessfully ompleted. Finally, flow hart diagrams of the gap-rossing sequene were onstruted based on an ethogram of single elements of behaviour. Comparing flow harts for two gap sizes revealed differenes in the frequeny and suession of these elements, espeially during the first part of the sequene. Key words: stik inset loomotion, hexapod walking, gap rossing, tatile orientation, exploration, ethogram, Aretaon asperrimus, Carausius morosus. Introdution A fundamental goal of ethologial studies is to understand how omplex behaviour is ontrolled. This goal an be ahieved by breaking down the ontinuous flow of ativity into single ations that an easily be identified and ounted. For investigating the ontrol of adaptive loomotion in a natural environment, analysing the basi elements of the behaviour and their linking is a useful approah. The behaviour performed by insets when limbing over obstales suh as large gaps is a hallenging paradigm for investigating the adaptation of loomotion. In spite of its high variability, this behaviour ontains a variety of reognisable elements, on the basis of whih a onise desription of the omplete behaviour is still possible. In several previous studies, insets have been observed when rossing gaps (e.g. Cruse, 976a, 979; Duerr, ; Watson et al., ). The gaps used in these studies, however, were not wider than the orresponding mean step amplitude. Steps observed under these onditions were mainly homogenous regarding their spatial and temporal parameters. In the urrent study, gap size was deliberately hosen to be larger two and three times the step length to hallenge the adaptive apabilities of the ontroller of the inset s loomotor behaviour. Comparably large gaps relative to body size and step length have only been used by Pik and Strauss () in order to evaluate the role of visual and tatile orientation in Drosophila loomotion. The omplex gap-rossing behaviour of the stik inset desribed here is far beyond the apabilities of any atual hexapod robot. As inset loomotion has proved to be a useful model for the onstrution of walking mahines that have to ope with rough terrain (e.g. Beer et al., 997; Ritzman et al., ; Cruse, ), the analysis of the underlying mehanisms may help in the onstrution of robots with more animal-like abilities. When faing a large gap, the inset annot just ontinue its normal walking pattern; it has to ensure that there is a ontinuation of the path ahead. In vertebrates suh as humans, obstale avoidane behaviour during walking is mainly guided by vision (Patla et al., 999). In insets, orientation of the antennae towards visual stimuli has been observed (Honegger, 98). Tatile exploration an beome ruial as an alternative, espeially in noturnal speies. Both the antennae and the front legs an be used as tatile probes, and slow-walking stik inset speies with long antennae seem to make use of both options. It has been shown that the stik inset Carausius morosus uses its front legs as feelers when walking on a horizontal plane (Cruse, 976b) but also probes the spae in front of its body with its antennae (Duerr, ). The role of inset antennae as tatile sensors is impressively
7 B. Blaesing and H. Cruse demonstrated in studies of objet-guided orientation (Okada and Toh, ) and wall-following behaviour (Camhi and Johnson, 999) in the okroah. Beause of their robustness, inset-like antennae have been used to failitate fast loomotion and ative probing in walking robots (Duerr and Krause, ; Cowan et al., ). Using the antennae atively for tatile exploration has also been observed in Crustaea suh as rayfish, whih move their antennae into the walking diretion before walking or turning (Zeil et al., 985) and even loalise objets aurately from the reeived tatile input (Sandeman and Varju, 988). In agonisti enounters, both rayfish (Bruski and Dunham, 99) and rikets (Hofmann and Shildberger, ) use their antennae for tatile ommuniation. As an alternative or in addition to the antennae, the front legs are used for tatile exploration by different speies. Cokroahes use their front legs to explore the surrounding environment by forward and sideways reahing movements (Watson et al., ; Full et al., 99). Speial funtions of the front legs ompared to the other leg pairs have been demonstrated for urve walking and turning behaviour, to whih the front legs of the okroah ontribute more than the middle and hind legs (Jindrih and Full, 999). In the stik inset, the front legs also play an important role in urve walking by initiating the turning movement (Duerr and Authmann, ). Even the front legs of bipedal walkers that have been speialized for other tasks suh as grasping like human arms, still play an important role in the stabilisation of walking (Marigold and Patla, ; Marigold et al., ). In this artile, we will investigate loomotive behaviour during trials with varying gap width and investigate how stik insets mainly examine their path: by vision or by touh reeived by the antennae or front legs. Subsequently, gap-rossing behaviour from trials with two different gap sizes will be studied in detail by defining basi elements of the sequene and analysing their distribution and frequeny. As the temporal struture of the gap-rossing sequene is partly predetermined by physial parameters the front legs have to ross the gap before the middle legs a framework of fixed events is used here to subdivide the sequene and to determine the temporal and spatial measures of the resulting setions. Within the different predefined setions of the gap-rossing sequene, the frequeny and order of behavioural elements an vary, and single elements an be modified, depending on the atual requirements. This approah is studied with an ethologial method by using an ethogram, in whih basi elements of gaprossing behaviour are defined. An ethogram is a atalogue of all ations or units or elements of behaviour that are observed in the general or speial behavioural repertoire of a speies (Immelmann and Beer, 989). It onsists of ategories of behaviour that are objetive, disrete, do not overlap with eah other and allow for the behaviour to be desribed as ompletely and preisely as possible. Ethograms are used in desriptive behaviour studies to analyse sequenes of behaviour. Early examples an be found in the work of Tinbergen (95), more reent examples are studies of bird song (e.g. Bradley and Bradley, 98) or loomotor behaviour (Berridge, 99). In inset studies, ethograms have mainly been used to desribe soial (Hoelldobler and Wilson, 99) or agonisti behaviour (Hoffmann, 987; Hofmann and Shildberger, ). Burrows and Morris () show hoie trees based on an ethogram of different avoidane and esape behaviours in Sipyloidea sp. The ethogram of gap-rossing behaviour used in the urrent study onsists of different types of steps that have been lassified aording to their swing amplitude and the ontext in whih they our. It does not inlude all elements of the behavioural repertoire of the stik inset, only the ones that are neessary to desribe the walking and gap-rossing behaviour relevant for this study. In another artile, we ompare gap-rossing behaviour to undisturbed walking on the basis of single step parameters suh as the swing amplitude and extreme positions of single steps (Blaesing and Cruse, ). The studies of stik inset behaviour different from walking are rather limited; a brief review an be found in Burrows and Morris (). Whereas most studies of stik inset loomotion have used the speies C. morosus for investigation, Aretaon asperrimus was preferred here, as previously by Cruse and Frantsevih (997). This speies walks slowly but more steadily than C. morosus and limbs readily over obstales and gaps. During undisturbed walking, it sans the ground more intensely with its antennae than C. morosus (Duerr and Blaesing, ). A. asperrimus is better amouflaged when sitting on the ground or on stems of trees than on leaves and twigs (for a speies desription, see Bragg, ). The morphology of the speies and personal qualitative observations suggest that this speies hides lose to the ground or on bark during the day, from where it moves up to the leaves to forage at night. As A. asperrimus is not able to jump or fly like other inset speies that inhabit a omparable environment, it depends on its ability to walk and limb in the foliage. Aordingly, the speies shows a high motivation for exploration and rossing gaps and obstales. This behaviour makes A. asperrimus a suitable biologial model for adaptive walking in a omplex environment. By investigating its performane in the gap-rossing paradigm, we hope to ontribute to our understanding of the ontrol of adaptive loomotion in insets and its appliation for autonomous artifiial agents that are thought to perform loomotive tasks in a natural environment. Materials and methods Animals Stik insets of the speies Aretaon asperrimus Rethenbaher 96 were kept in mesh wire ages on bramble (Rubus frutiosus) and water ad libitum with an artifiial day:night yle of h: h. Body length was 5±. mm (mean ± S.D.) in males (N= animals) and 76±. mm in females (N=). Males and females were treated separately in the first experiment due to their different size and body geometry. For the seond experiment and the detailed analysis of the gap-rossing sequene, only male subjets were used beause of their higher agility. The average step amplitude of
Stik inset loomotion 75 EFL IFM ML IMH HL Fig.. Conseutive photographs (from top to bottom) of Aretaon asperrimus male limbing aross a gap of 5 mm width between two ardboard footbridges; eah piture illustrates one of the setions of the gap-rossing sequene. EFL, exploration and front leg gap-rossing steps; IFM, interval between front leg and middle leg gap-rossing steps; ML, middle leg gap-rossing steps; IMH, interval between middle leg and hind leg gaprossing steps; HL, hind leg gap-rossing steps. males is approximately 7 mm (front legs: 7.±.6 mm, n=, middle legs: 6.8±.7 mm, n=6, hind legs: 7.6±. mm, n=). Distanes between the oxae of adjaent leg pairs are.±. mm between the middle and front leg oxae and 7.±.7 mm between the hind and middle leg oxae (N=). Experimental set-up In all experiments, animals were plaed on a ardboard footbridge of 6 mm width, mm length and mm height, faing a seond footbridge of the same type. The gap between the two footbridges was variable; an example of a trial with 5 mm gap width is shown in Fig.. The animals were reorded from a distane of.5 m by an overhead video amera (Sony EVI-D). A simultaneous side view image of the walking animal was obtained via a mirror attahed to the footbridges at a 5 angle. Video reordings (5 frames s ) were arried out in daylight with additional artifiial lighting. Reordings of the gap-rossing sequene were analysed in slow motion and single frame modus, using ustomized software designed to read marked pixel oordinates as ASCII data. Data analysis and statistial tests were arried out using Origin (Miroal, Northampton, MA, USA) and SPSS software. Experimental proedure In the first experiment, the insets (5 males, females) were tested with gaps of,,, 5 and 6 mm width. Eah animal was tested in trials per gap size, presented in random order. A trial was ounted as suessful only if the animal had rossed the gap. In a ontrol experiment ( males, females), a paper-strip of orresponding width was used instead of a gap. Duration was measured as number of frames from the first antennal ontat with the ground behind the gap or paper-strip to the touhdown of the sixth leg behind the gap or paper-strip. Antennal exploration before disovering the seond footbridge was not taken into aount. In the seond experiment, six males were reversibly blindfolded with solvent-free blak ink and tested in the same task. Individual animals started either sighted or blindfolded. Additionally, six males with shortened antennae (between 5 and 8 mm) but intat vision were tested in the same set-up. In the following experiments, gap-rossing behaviour was analysed in more detail. Two gap sizes were hosen: mm (N=7 animals, n=5 trials) and 5 mm (N=5, n=). Definition of step types In the desription of individual steps, the terms posterior extreme position (PEP) and anterior extreme position (AEP) desribe the lift-off and touhdown position of the leg in a body-fixed oordinate system, respetively. The position at whih the tarsus moves below footbridge level when swinging into the gap has been alled fitive AEP (faep) by Duerr (). The swing amplitude is defined as the length of the vetor that points from the PEP to the AEP of the same swing movement. All individual steps reorded from the trials were assigned to the following four ategories: () tentative steps, () gap-rossing steps, () normal walking steps and () short steps (Fig. ). Tentative steps onsist of swing movements into the gap followed by pulling the tarsus bak and plaing it onto the first footbridge. Gap-rossing steps are haraterised by a swing trajetory that onnets the first to the seond footbridge. Normal walking steps and short steps were defined aording to their swing amplitude and swing diretion in a body-fixed oordinate system. For lassifiation as normal walking steps, steps had to fulfil two onditions: () a minimum swing amplitude of 8.5 mm and () forward diretion of the swing movement. The latter riterion, forward diretion of the swing movement, was met if the AEP was loated rostral of the PEP within an angle of ±5 relative to the body long axis in a bodyfixed oordinate system. Steps of more than 8.5 mm amplitude that did not fulfil this riterion were so rare that they were not onsidered in the analysis. All steps with an amplitude of less than 8.5 mm were assigned to the group of short steps, regardless of their swing diretion. The threshold of 8.5 mm was hosen based on the distribution of amplitudes of all steps
76 B. Blaesing and H. Cruse observed during gap rossing and undisturbed walking (Fig. ). As an example of the typial distribution of the four defined step types, stepping patterns of an individual mm gaprossing trial and a sequene of undisturbed walking are displayed in Fig.. of the setion. Advane was measured as the distane between the position of the body entre of mass (between the hind leg oxae) in the first and in the last frame of eah setion in an external oordinate system. Veloity of body movement over ground was alulated by dividing body advane by duration. Setions of the gap rossing sequene To subdivide the temporal sequene of gap-rossing behaviour, six events that our in a fixed order were defined. These events are () the first non-ontat of an antenna with the gap (see below), () reahing the AEP of the gap-rossing step of the seond front leg, (,) the PEP of the first and the AEP of the seond middle leg gap-rossing step and (5,6) the PEP of the first and the AEP of the seond hind leg gap-rossing step. By using these six events as a framework, the sequene of gap-rossing behaviour was divided into the following five setions (Fig. ): EFL (exploration/front legs ross the gap) this setion inludes antennal and front leg exploration movements and front leg gap-rossing steps; it starts when the tip of an antenna moves below the line that onnets the two footbridges ( antennal non-ontat with the gap ) and ends when both front legs are plaed on the seond footbridge; IFM (front leg/middle leg interval) from the touhdown of the seond front leg gap-rossing step to the lift-off of the first middle leg gap-rossing step; ML (middle legs ross the gap) from the lift-off of the first middle leg gap-rossing step to the touhdown of the seond middle leg gap-rossing step; IMH (middle leg/hind leg interval) from the touhdown of the seond middle leg gap-rossing step to the lift-off of the first hind leg gap-rossing step; HL (hind legs ross the gap) from the lift-off of the first hind leg gap-rossing step to the touhdown of the seond hind leg gap-rossing step. For these setions, duration, advane of the body over ground and forward veloity of the body were measured and the distribution of the step types defined above was determined. Duration was alulated as time differene between the first and the last frame () Tentativestep PEP } AEP st footbridge faep } Searh () Normal walking step nd footbridge () Gap rossing step PEP () Short step Ethogram and bigram analysis Finally, an ethogram (Fig. 5) was used to analyse the sequene of gap-rossing behaviour on the basis of its single elements and their order. It inludes the four previously defined step types and three elements of antennal and front leg exploration. As gap-rossing steps and tentative steps have a more variable struture than walking steps and short steps, the former have been broken down into finer parts: the elements swing and searh our in both step types, but tentative steps are ompleted by plaing the leg bak on the first footbridge ( AEP fbr_ ), whereas gap-rossing steps are ompleted by plaing the leg on the seond footbridge ( AEP fbr_ ). For analysing the sequene of the single elements of gaprossing behaviour, the onept of bigrams has been adopted from omputer linguistis (e.g. Jurafsky and Martin, ). A bigram is a pair of two elements that diretly follow eah other in a naturally ourring sequene. This means that eah element (exept for the first and the last one of the sequene) is reorded twie, with its preeding and following neighbour (for example the sequene ABCD onsisting of elements A, B, C and D results in bigrams AB, BC and CD). The resulting pairs, the bigrams, are then treated as new basi units in the analysis. In the urrent study, all elements of behaviour have been listed in the order of their ourrene for every trial. From this list, all pairs of two elements that diretly follow eah other have been defined as bigrams and have been used as basi units in the following analysis. The frequeny of bigrams was ounted; mm trials and 5 mm trials were treated separately. In total, 9 bigrams were ounted in the mm trials and 97 in the 5 mm trials. 6 different types of bigrams (AB is a different type of bigram ompared to BA or BC) faep Searh } } st footbridge AEP nd footbridge ourred in the mm trials and 87 in the 5 mm trials. We tested the hypothesis that ertain bigrams our more often in the observed behaviour than in a random distribution (if the raw data ontain A, B and 7 C, the bigrams AB and BA would be expeted to our twie eah and AC and CA seven times eah if the elements A, B and C were randomly PEP Footbridge } AEP PEP } AEP 8.5 mm Footbridge Fig.. () Tentative step, () gap-rossing step, () normal walking step and () short step, shown shematially. PEP, posterior extreme position; AEP, anterior extreme position; faep, fitive anterior extreme position; swing, initial swing movement (green); searh, subsequent searhing movement (red).
Stik inset loomotion 77 distributed). The expeted probability of any bigram in a random distribution was alulated by ontingeny tables (see M. Moens and C. Brew, : Data-intensive Linguistis. http://www.ltg.ed.a.uk/ ~hrisbr/dilbook/) and ompared to the observed probability by χ -tests. Only bigrams that ourred more than twie in the data and signifiantly more often than expeted in a random distribution (χ >.8, P.) were inluded in the analysis. Results Variation of gap size In the first experiment, we studied how the suess rate and duration of gap-rossing behaviour depends on gap width. Walking aross paper-strips of orresponding width was used as ontrol. The perentage of suessful trials and the average duration of rossing gaps and walking over paperstrips is displayed in Table. Males and females suessfully rossed gaps of and mm in almost every trial ( 97%). The suess rate dereased from about mm gap width in the males and 5 mm gap width in the females. Males needed more time to ross gaps of the same width than females, with exeption of mm gaps. The time differene between rossing gaps and rossing paper-strips of orresponding width inreased approx. exponentially with gap size in the males. In the females, the time differene hardly inreased up to 5 mm gap width. In separate experiments, no effet of previous experiene was observed with respet to duration of the sequene in insets repeatedly rossing gaps of the same width (N=8 animals, n= trials per animal and gap width). Sensory orientation In a seond experiment we tested whih sensory mode is used for deteting the far edge of the gap before limbing aross it. Both the visual and the tatile sensory systems ould be used by the inset to gain information about a possible ontinuation of the path. The results of this experiment show that blindfolding has neither any signifiant effet on the number of suessfully ompleted trials (Table ) nor on the duration of gap rossing (data not shown). The gap-rossing sequene was abandoned in 85 out of 8 ases (7.7%) in the sighted animals (A+V+) and in 8 out of 8 ases (7.5%) in the blindfolded animals (A+V ), both groups with intat antennae. This onsisteny shows that vision is not neessary for deteting the far edge of the gap, whih suggests that antennal ontat with the seond footbridge provides suffiient information. Having touhed the seond footbridge with a front leg, the gap-rossing sequene was always suessfully ompleted regardless of the animal s visual situation. % of steps taken A Short steps B C D Undisturbed walking Normal steps 5 5 5 Short steps amplitude (mm) Gap rossing Normal steps 5 5 5 Fig.. Histograms of swing amplitudes of all steps (with exeption of gaprossing steps and tentative steps) observed during undisturbed walking (left) and gap rossing (right). (A) Front legs, (B) middle legs, (C) hind legs, (D) pooled data of all leg pairs. The broken lines mark the threshold of 8.5 mm that separates short steps from normal walking steps (as indiated). Sighted animals with shortened antennae (A V+) rossed the gap only if they ould still reah the seond footbridge with an antenna or a front leg. This means that in every suessful trial in this group the animal had touhed the seond footbridge with its shortened antenna (one individual with extremely short antennae regularly touhed the far edge of the mm gap with the strethed front leg, whih also resulted in gap-rossing behaviour). Beause of the restrited working spae of their antennae, animals of this group performed fewer suessful trials, espeially with larger gap sizes than animals with intat antennae. Gap rossing was abandoned in 5 out of 8 trials (.7%). In only two of
78 B. Blaesing and H. Cruse these ases, gap-rossing behaviour was terminated after antennal ontat with the seond footbridge had already ourred. This behaviour was not observed in any trial of the two groups with intat antennae. Detailed analysis of gap-rossing behaviour Analysis of mm and 5 mm gaps revealed that after stepping into the gap with one or both front legs, the inset dereases its stane veloity to almost zero (Fig. 6). Additional forward movement onsists of short stops alternating with bouts of slow advane while the antennae perform extensive exploration movements. After reahing the seond footbridge with the front legs, body veloity is gradually aelerated throughout the sequene. In Fig. 6, slowing down of body veloity is shown in relation to the first non-ontat of the front leg (Fig. 6A) and the first nonontat of the antenna (Fig. 6B). The relation of slowing down after stepping into the gap with the front leg is more obvious. In the observed trials, antennal non-ontat takes plae between and ms before stepping into the gap with the front leg. The gap-rossing sequene has been subdivided into five setions EFL, IFM, ML, IMH and HL (Fig. ; explanation in Materials and methods). Duration, advane of the body overground and veloity of body movement for these setions are displayed in Fig. 7. For the entire gap-rossing sequene, the animals A lant lfl lml lhl rant rfl rml rhl B lfl lml lhl rfl rml rhl Walking sh a te te Walking a g g 5 5 5 Time (s) 5 6 7 sh Time (s) sh sh te sh sh sh sh g sh sh sh g needed approximately 6 s in the mm trials (mean ± S.D.=5.9±. s) and six times longer in the 5 mm trials g shsh sh shsh Fig.. Examples of stepping patterns. (A) Gap rossing, (B) undisturbed walking. Grey bars, ground ontats of antennae; oloured bars, swing movements of legs. FL, front leg (red); ML, middle leg (green); HL, hind leg (blue); l, left; r, right; a, first ontat of the antennae with the seond footbridge; g, gap-rossing step; te, tentative step; sh, short step. Normal walking steps are not marked. In A, the defined setions of gap-rossing behaviour (see Fig. ) are separated by vertial lines. sh g sh Table. Mean duration and suess rate of stik insets limbing over gaps and walking aross paper-strips Length of rossing (mm) 5 6 Duration % Duration % Duration % Duration % Duration % (s) Suessful (s) Suessful (s) Suessful (s) Suessful (s) Suessful Gap Males.8±. 97 8.±.8 98 7.7±.5 8 56.7±56. 8 Females 6.6±. 6.8±. 97 9.±. 97.±. 7.9±.8 Paper-strip Males.±.8.±..5±. 5.±.5 5.6±.9 Females 5.±.9 6.±. 7.±.6 7.±. 8.7±.5 Values are means ± S.D. N=5 males, females for gap rossings; N= males, females for paper-strip rossings. n= trials per animal and gap size. Mean body length = 5±. mm (males), 76±. mm (females) (see Materials and methods).
Stik inset loomotion 79 Table. Numbers and perentages of suessfully ompleted gap rossing trials in males with intat antennae, sighted (A+V+) and blindfolded (A+V ), and animals with defetive antennae, sighted (A V+) Length of rossing (mm) 5 Number % Number % Number % Number % Intat antennae A+V+ 7 98 8 98 97 8 6 5 A+V 9 99 87 5 Defetive antennae A V+ 9 99 9 78 5 8 8 5 N=6 animals eah for intat and defetive antennae; n= trials per animal and gap size. Note that animals only rossed the gap in trials in whih they had reeived tatile input by the antennae (or in one ase of A V+ by the front legs, as desribed in the text). (7.±6. s). Setion EFL takes almost times longer in the 5 mm trials (6.±.5 s) than in the mm trials (.8±. s) whereas the rest of the sequene takes only about three times longer. Forward movement of the body overground mainly takes plae during EFL, ML and HL, the largest differene between mm and 5 mm trials ourring during EFL (5 mm: + mm) and HL (5 mm: +5 mm). In the 5 mm trials, the animals move more slowly than in the mm trials. Mean veloity in the mm trials (5.7±. mm s, measured from the beginning of EFL to the end of HL) is about 5% of the veloity of undisturbed walking (.±. mm s, N=), whereas in the 5 mm trials only % of normal walking veloity is reahed (.8±.5 mm s ). During EFL, veloity is five times higher in the mm trials than in the 5 mm trials, whereas it is only twie as high during the rest of the sequene. All of the observed steps have been assigned to four ategories, namely gaprossing steps, tentative steps, short steps and normal walking steps (Fig. ; see explanation in Materials and methods). The average frequeny of the four step types in eah setion is displayed in Fig. 8. There is no qualitative differene between mm trials and 5 mm trials. In the 5 mm trials, more short steps an be observed ompared to the Exploration Ant,FL Tentative step FL, ML,HL Gap rossing step FL, ML,HL Normal walking step FL, ML,HL Short step FL, ML,HL mm trials. For the short steps, this is partiularly obvious for the middle and hind legs. Tentative steps are only different from gap-rossing steps regarding the end of their searhing movement. They our most often in the front legs in setion EFL. In the front legs, the number of tentative steps approximately equals the number of the gap-rossing steps in both data sets, refleting that on average, every seond step into Contat fbr_ Contat fbr_ Searh AEP fbr_ Searh AEP fbr_ Norm Short 8.5 mm Antenna touhes the seond footbridge for the firsttime Front leg touhesseond footbridge for the firsttime Leg swings into the gap Leg passes faep after swinging into the gap Legis plaed on the first footbridge Leg swings into the gap Leg passes faep after swinging into the gap Legisplaed on the seond footbridge Step on plane surfae, amplitude >8.5 mm Step on plane surfae, amplitude <8.5 mm Fig. 5. Ethogram of gap-rossing behaviour; elements of behaviour used in the flow hart diagram (Fig. ) are printed red. Ant, antenna; FL, front leg; ML, middle leg; HL, hind leg; fbr_, first footbridge; fbr_, seond footbridge; norm, normal walking step; AEP, anterior extreme position; faep, fitive anterior extreme position; swing, initial swing movement; searh, subsequent searhing movement. the gap results in reahing the far edge. In the middle legs, tentative steps are far less frequent than in the front legs, and no tentative steps were observed in the hind legs. The number of normal walking steps dereases before and inreases after the legs have performed their gap-rossing steps. Below, short steps of the mm trials (N= short steps) are onsidered in more detail to gain information regarding
8 B. Blaesing and H. Cruse their funtion. Three types of short steps an easily be haraterised aording to the ontext in whih they our. The first type has been desribed as levator reflex by Dean and Wendler (98): if the inset hits an obstale with its tarsus or the end of the tibia, the leg is pulled up and plaed on the ground again. During gap rossing, the levator reflex has been observed in the front and middle legs in response to hitting the side edge of the footbridge (front legs: EFL 5 steps, IFM 7 steps, IMH 5 steps, HL steps; middle legs: EFL 6 steps). Another reflex an be observed if the tarsus hits the anterior leg during the swing movement or is plaed on the anterior tarsus. In this ase the posterior leg is pulled up and plaed slightly bakwards [treading on tarsus (TOT) reflex; Graham, 979; Shmitz and Hassfeld, 989]. During gap rossing, the TOT-reflex mainly ourred during EFL in the middle legs (5 ases) or while the animal was trying to plae the leg on the seond footbridge (middle leg: ases during ML and IMH, hind leg: ases during HL). A third group of short steps ourred diretly after the tarsus had reahed the seond footbridge and was linging to the edge with the unguis rather than standing in a stable position. In this situation, short steps were apparently used to plae the tarsus on the footbridge surfae. This situation ourred times in the front legs, Body veloity (mm s ) 5 5 5 A B mm gap 5 mm gap 5 Time (s) Fig. 6. Veloity profiles of forward movement of the body overground during a time window of 6 s at the beginning of the gap-rossing sequene. (A) t= (broken line) taken as initial nonontat of the front leg (tarsus rosses footbridge level while swinging into the gap for the first time); (B) same data as in A, but t= (broken line) taken as initial non-ontat of the antenna. Heavy lines, arithmeti mean; thin lines, single trials; red dots in the upper panel mark nonontats of the antenna (t= in B), illustrating their temporal relation to orresponding nonontats of the front leg. Left, mm trials; right, 5 mm trials; N=7 trials in eah gap width, bin width= ms. times in the middle legs and 8 times in the hind legs. The distribution and the relative swing diretion (AEP relative to the PEP) of the remaining 66 short steps is displayed in Fig. 9. The majority of these short steps (8 steps) were performed by the front legs throughout the entire sequene, espeially during IMH and HL. During setion EFL, more short steps were performed by the middle legs and hind legs than by the front legs. Most of these short steps were direted to the front and, in the middle legs, also to the side, presumably ontributing to slow forward movement and slight side shifting of the body long axis to support front leg searhing. During setions IFM to HL, most short steps were performed by the front legs, mostly direted to the front and to both sides, and only few short steps ourred in the middle and hind legs. The temporal sequene of the different behavioural elements is illustrated in the flow hart diagram of gap-rossing behaviour (Fig. ). This flow hart is more omplex than a hoie tree, as the sequene of behavioural elements in gaprossing behaviour often ontains loops, and bifurations an our at any time. To make the temporal struture more perspiuous, the most frequent assoiations between single elements of behaviour (Fig. 5) are displayed in the framework of the three setions EFL, ML and HL. In general, more different transitions our in the mm trials (red and blak arrows in Fig. ) than in the 5 mm trials (green and blak arrows in Fig. ), in whih the body position of the animal and therefore the number of possible subsequent movements is more restrited. This differene is more obvious in the middle and hind legs than in the front legs. Only in the mm trials, transitions between AEP fbr_ and swing or searh, i.e. red arrows pointing upward, our in every leg pair. In the 5 mm trials, middle leg swinging is less often diretly followed by the ontralateral middle leg reahing the first or seond footbridge. After plaing a front leg on the first or (in the mm trials) on the seond footbridge, the ontralateral front leg often immediately begins a swing movement. In the 5 mm trials, front leg searhing is often aompanied by a short step of the hind leg. Antennal ontat with the seond footbridge ourred in different situations in the mm and 5 mm trials. In the mm trials, the antenna was likely to touh the seond footbridge while searhing with a front leg during a tentative step. Therefore antennal ontat with the
Stik inset loomotion 8 seond footbridge was ommonly followed by AEP fbr_. This was not the ase in the 5 mm trials, as the body entre of mass had to be pushed too far forward for plaing the front leg bak onto the first footbridge before the seond footbridge ould be reahed by an antenna. In the 5 mm trials, antennal ontat with the seond footbridge regularly ourred after finishing a middle leg tentative step and before the front legs reahed the seond footbridge. This observation reflets a tendeny of front leg gap-rossing steps and middle leg tentative steps to overlap when rossing extremely large gaps. Disussion The aim of this study was to desribe how loomotion is adapted to the hallenges of a omplex environment. The behaviour of stik insets limbing over large gaps has proved to be a useful paradigm to approah this question: it is based on a well-studied objet, stik inset walking behaviour on even ground, and the environment an easily be varied in a ontrolled way by hanging the size of the gap. The speies A. asperrimus is partiularly suited for this study beause it appears to be highly motivated to ross even large gaps. The results of the study show that during gap-rossing behaviour, steps vary more strongly than during undisturbed walking. In addition to normal walking steps and gap-rossing steps, tentative steps and short steps have been observed. The number of both step types was higher in trials with a larger gap and dereased from the front to the hind legs (Fig. 8). Tentative steps espeially of the front legs ourred almost exlusively during setion EFL, the first part of the sequene that inludes antennal and front leg exploration movements and front leg gap-rossing steps. Front leg tentative steps are losely related to gap-rossing steps. Both step types start in the same way with a swing movement into the gap. In a tentative step, the leg is then pulled bak at some point during the searhing movement and plaed on the first instead of the seond footbridge. This behaviour enables the animal to use a trial and error strategy to searh for a ontinuation of the path. In the flow hart (Fig. ), antennal ontat with the seond footbridge is followed by finishing a tentative step, whereas front leg ontat with the seond footbridge is followed by finishing a gaprossing step. Taking into aount that gap-rossing behaviour is always ompleted after touhing the seond footbridge with an antenna, there might be a working memory funtion involved in whih the information path ontinues is stored while the single leg is pulled bak and possibly a more appropriate position is adopted. The observation that tentative steps our rarely in the middle legs and never in the hind legs may also be based on this stored information. Short steps our in various situations during gap rossing, two of whih have previously been identified as reflex reations: the levator reflex (Dean and Wendler, 98) and the Duration (s) Body advane (mm) Body veloity (mm s ) 7 6 5 8 A 6 8 6 EFL ML HL IFM IMH B C mm gap 5 mm gap EFL ML IMH HL 5. IFM 7 6 5 TOT (treading-on-tarsus) reflex (Graham, 979; Shmitz and Hassfeld, 989). The majority of the remaining short steps are performed by the middle and hind legs during setion EFL or by the front legs throughout the entire sequene. This indiates that during EFL, the funtion of the short steps is to adjust the body position of the animal while examining the gap. The short steps performed by the middle and hind legs during setion EFL may also be a part of the exploration behaviour, they support slow forward movement and slight side shifting of the body long axis during antennal and front leg searhing. In the following setions, after the front legs have rossed the gap, short steps mainly our in the front legs. Short steps performed by the front legs after reahing the seond footbridge might represent tatile investigation of the ground to find an appropriate tarsus position. This is important beause 5 7.8 EFL IFM ML HL IMH Fig. 7. Box and whisker plots of duration (A), body advane (B) and body veloity (C) of setions of the gap-rossing sequenes (note the different sales for body veloity and duration). Red, IFM and IMH; blak, EFL, ML, HL (see Fig. for abbreviations); arithmeti means are added as squares. Left, mm trials (N=7 animals, n=5 trials); right, 5 mm trials (N=5 animals, n= trials); mean walking veloity (.±. mm s ) alulated from ten sequenes of undisturbed walking is added as a horizontal broken line in B. Negative duration values of interval setions indiate overlapping of gap-rossing steps of neighbouring leg pairs ( mm: 5 ases in IFM, 5 ases IMH; 5 mm: ases in IMH).
8 B. Blaesing and H. Cruse mm gap 5 mm gap HL ML FL A Front legs EFL 8 IFM 6 7 Average frequeny per trial B Middle legs C Hind legs Fig. 8. Average frequeny of different step types in the setions EFL to HL (see Fig. for abbreviations). (A) Front legs, (B) middle legs, (C) hind legs, left: mm trials (N=7 animals, n=5 trials), right: 5 mm trials (N=5 animals, n= trials). Blak olumns, normal walking steps; red olumns, short steps; green olumns, tentative steps; blue olumns, gap-rossing steps. the front legs have to support the main part of body weight during setions ML and HL. Pearson and Franklin (98) have mentioned an inrease of loal stepping in lousts when walking on a slippery surfae. Stepping has also been observed in the okroah after passive defletion of a leg (Zill, 99). The desribed stepping strategies are not just used by invertebrates: ats show orretive response movements when losing ground (Gorassini et al., 99) and perform a stumbling orretive reation (Forssberg, 979) that is similar to the levator reflex. In humans, different obstale avoidane reations result in lengthening or shortening a step if enough time is available to adapt the step length (Patla et al., 99). If the obstale is pereived within the same step yle, ML IMH HL 5 lengthening the step is the preferred strategy (Patla et al., 999). Comparing mm to 5 mm gap-rossing trials revealed that most differenes our during setion EFL. When limbing aross a 5 mm gap, animals need ten times longer for the exploration phase than when limbing aross a mm gap, whereas they only need three times longer for the rest of the sequene (Fig. 7). Also, during EFL there are more different elements of behaviour and their order is more variable than during later parts of the sequene (Fig. ). Comparison of the flow harts reveals further differenes that seem to be aused by body geometry: the variety of body postures adopted by the insets while moving aross the gap is more restrained in the 5 mm trials than in the mm trials. Starting a gap-rossing step after finishing the intrasegmental leg one ourred only in the mm trials, whereas a tendeny for front leg gap-rossing steps and middle leg tentative steps to overlap was only observed in the 5 mm trials (Fig. ). Antennal movements are very distint in A. asperrimus and more learly direted towards the ground than in C. morosus, o b f i 5 7 8 7 9 7 7 Fig. 9. diretion and temporal distribution of short steps. diretion (AEP relative to PEP) is defined in four groups (f, forward; b, bakward; o, outward; i, inward; oordinate system is shown in the inset figure below). Temporal distribution is defined aording to the five setions EFL to HL (see Fig. for abbreviations); short steps aused by the levator or TOT-reflex and by linging to the seond footbridge have not been taken into aount.
Stik inset loomotion 8 EFL FL swing A FL searh Ant ontat fbr_ Searh FL AEP fbr_ AEP fbr_ HL short Ant ontat fbr_ FL ontat fbr_ FL AEP fbr_ FL AEP fbr_ B Searh AEP fbr_ ML ML swing ML searh C ML AEP fbr_ Ant ontat fbr_ ML AEP fbr_ AEP fbr_ ML AEP fbr_ HL HL swing χ < HL searh < χ <5 HL AEP fbr_ HL AEP fbr_ 5< χ < < χ< < χ Fig.. Flow hart diagrams of gap-rossing behaviour. Arrows onnet any two elements of behaviour that were performed in diret transition (bigrams), pointing from the first to the seond element. The size of arrow (see key) represents the χ -values of the orresponding bigrams (explained in the text; see Materials and methods). Arrows marked by onnet behavioural elements performed by limbs of ontralateral body segments (e.g. left antenna and right front leg). Only bigrams that ourred more than twie and with higher probability than expeted in a random distribution (P.) have been inluded in the flow hart. Typial ombinations that represent steps desribed in the text are shematially displayed on the right side: (A) tentative step, (B) gap-rossing step, (C) gap-rossing step without searhing movement. Blak, elements and transitions that our in both the mm and 5 mm trials; red, elements and transitions that our only in the mm trials; green, elements and transitions that our only in the 5 mm trials; FL, front legs; ML, middle legs; HL, hind legs; Ant, antennal. For further abbreviations see Fig. (setions of gap rossing) and Fig. 5 (ethogram; elements of behaviour used in the flow hart diagram are printed red).
8 B. Blaesing and H. Cruse with one up- and down-stroke and several ground ontats of eah antenna per step yle (Duerr and Blaesing, ; Duerr, ). Therefore, the first tatile pereption of the gap ould our when one or both antennae miss ground ontat after reahing the end of the first footbridge. It has been observed, however, that after this first antennal non-ontat, the animal steps into the gap, whih results in a non-ontat of the front leg. Only the latter, the front leg non-ontat, is followed by a derease in veloity of forward movement (Fig. 6). Slowing down seems to our in response to stepping into the gap, but not in response to lowering the antenna into the gap. This suggests that antennal non-ontat does not influene the subsequent behaviour of the animal whereas non-ontat of the front leg does. The inident of non-ontat, however, an only provide information in a situation in whih the expetation of ontat has been thwarted. In the ase of the antenna, therefore, expetation of ground ontat seems to be absent, whereas ground ontat seems to be expeted at a ertain height by the front leg, and thwarting of this expetation evokes a hange of behaviour. The shift of the extreme positions of onseutive steps provides independent support for this interpretation, as shown by Blaesing and Cruse (). How ould the information neessary for ground expetation be reeived? It ould be provided by the joint angles of the femur tibia joint and the oxa trohanter joint of the neighbouring legs. Passing a position with orresponding joint angles without sensing ground ontat would evoke slowing down of stane movement. Alternatively, the joint angles of the same leg ould be remembered throughout the swing movement and used as a referene. In ats trained to walk on a flat surfae, Gorassini et al. (99) showed that extensor musle exitation was similar to normal undisturbed stepping even when unexpetedly stepping into a hole. This is interpreted in suh a way that an expetation of the ground substrate exists in the at as well. It has been argued that this expetation is not based on leg position but on the lak of load on the supporting musles (Hiebert et al., 99). Unlike antennal non-ontat during walking, physial ontat of the antenna with the seond footbridge during exploration has a lear impat on the subsequent behaviour. The results of the seond experiment (Table ) suggest that when a stik inset is reahing the end of a walking path, information about a ontinuation of the path is not gained by vision, even though A. asperrimus shows orientation behaviour towards ertain visual stimuli (Frantsevih and Frantsevih, 996). The observation that animals only ontinue gaprossing behaviour if they reeive tatile input from touhing the far edge of the gap with an antenna or front leg independent of the state of their visual system indiates that vision alone does not provide suffiient information. The tatile stimulus of touhing the far edge of the gap with an antenna provides the animal with suffiient information about the existene of a far edge and, inherently, about gap size to limb aross the gap suessfully. After the far edge has been touhed by an antenna, loomotion towards this stimulus is always ontinued, and the gap-rossing sequene is not abandoned any more. This indiates that if an animal an reah a footbridge with its flagellum while probing its environment, it an also reah it with its front legs if it adopts a position in whih its body entre of mass is pushed forward even past the edge of the supporting footbridge. The male speimens of A. asperrimus used in this study were able to limb aross gaps of up to 5 mm (Table ), whih is equivalent to their body length. When trying to ross larger gaps, they ould hardly reah the far edge with their antennae and therefore failed to reeive information about the far side of the gap. It ould be argued that the additional distane that an be reahed by the antenna but barely by the front leg (about 5 mm in A. asperrimus) allows the animal to plan ahead and exploit speifi inonvenient body postures to reah far ground. In okroahes, antiipatory rearing of the thorax after deteting an obstale with the antenna has been observed (Watson et al., ; Tryba and Ritzman, ). Animals that do not have intat antennae might lak the hane to plan ahead as they have to rely on front leg exploration. To approah this issue, animals with artifiially elongated flagella (see Camhi and Johnson, 999), short but intat front legs or a shifted entre of mass (weights attahed to the thorax or abdomen), ould be tested in the same task. Immediate slowing down of forward movement ours in response to stepping into the gap, i.e. non-ontat of a front leg (Fig. 6), and further forward movement onsists of single bouts while the gap is investigated by explorative movements of the antennae and front legs. Derease of veloity helps to maintain stati stability, as has been shown by Cymbalyuk et al. (998) who, in a simulation study, investigated body stability when starting to walk from different leg onfigurations. Lousts have been reported to stop before stepping aross a mm dith (front leg step amplitude: mm), adjust their body position by hind leg flexing and middle leg stepping and perform extensive searhing movements with the front legs (Pearson and Franklin, 98). No stop has been reported for C. morosus when walking aross a mm dith (Cruse, 979) or a mm gap (Duerr, ). After swinging into the gap, the legs perform various searhing movements before touhing the seond footbridge. In lousts, searhing movements of the front legs after stepping into a gap have been desribed qualitatively as yles of elevation and depression (Pearson and Franklin, 98). An example of searhing movements in C. morosus and their simulation has been given by Duerr (). At the end of the searhing movement, the tarsus is often pulled upwards after hitting the seond footbridge from below, a reation that has been desribed as levator reflex (Dean and Wendler, 98) and has also been found in the same ontext in lousts when stepping aross a dith (Pearson and Franklin, 98). The results desribed above suggest that most adaptations of walking behaviour to the gap-rossing situation our during the interval of antennal and front leg exploration of the gap. Lowering an antenna into the gap does not influene the subsequent behaviour whereas stepping into the gap with a front leg initiates slowing down of walking veloity. Tatile ontat of the antenna or a front leg with the seond footbridge