Department of Biology, Case Western Reserve University, Cleveland, Ohio 44106-7080
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Tryba, Andrew K. and Roy E. Ritzmann. Multi-Joint Coordination During Walking and Foothold Searching in the Blaberus Cockroach. I. Kinematics and Electromyograms. J. Neurophysiol. 83: 3323-3336, 2000. Cockroaches were induced to walk or search for a foothold while they were tethered above a glass plate made slick with microtome oil. We combined kinematic analysis of leg joint movements with electromyographic (EMG) recordings from leg extensor muscles during tethered walking and searching to characterize these behaviors. The tethered preparation provides technical advantages for multi-joint kinematic and neural analysis. However, the behavioral relevance of the tethered preparation is an important issue. To address this issue, we evaluated the effects of tethering the animals by comparing kinematic parameters of tethered walking with similar data collected previously from cockroaches walking freely on a treadmill at the same speeds. No significant differences between tethered and treadmill walking were found for most joint kinematic parameters. In contrast, comparison of tethered walking and searching showed that the two behaviors can be distinguished by analysis of kinematics and electrical data. We combined analysis of joint kinematics and electromyograms to examine the change in multi-joint coordination during walking and searching. During searching, middle leg joints extended during swing rather than stance (i.e., walking) and the coordination of movements and extensor motor neuron activity at the coxa-trochanteral and femur tibia joints differed significantly during walking and searching. We also found that the pattern of myographic activity in the middle leg during searching was similar to that in the front legs during walking.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The problem of coordinating several motor neuron
pools to produce a behavior is common to many motor systems. This
problem is complex as animals may use the same appendages to produce
multiple behaviors and each behavior may have its own degree of
variability. In walking systems, where joint oscillators exist (insect:
Buschges et al. 1995; mud-puppy: Cheng et al.
1998
), central or afferent influences orchestrate these
oscillators to achieve appropriate inter-joint coordination during
ongoing activity. Inter-joint coordination then must be reliably
modified to produce multiple behaviors in a context-dependent fashion.
Evidence for central and peripheral modulation of inter-joint
coordination exists (Angel et al. 1996
; Bassler
1993
; Brunn 1998
; Graham and Bassler
1981
; Grillner and Zangger 1979
; Hess and
Buschges 1999
; Robertson et al. 1985
).
However, few studies have begun to examine the underlying neural
principals involved in control of multiple leg joints (El Manira
et al. 1991
; Hess and Buschges 1999
; Wolf
1990
). Further, intracellular analysis of neural mechanisms
underlying control of inter-joint coordination in legged animals during
multiple behaviors has rarely been examined (Kitmann et al.
1995
). In approaching this problem, preparations that
allow detailed three-dimensional kinematic analysis concurrently with
intracellular and extracellular recordings would be highly
advantageous. However, such data sets are technically difficult to
obtain from behaving animals.
Tethered preparations provide a potential solution to this dilemma.
Several studies have examined neural strategies for coordinating behaviors using semi-intact tethered preparations that allow
simultaneous kinematic, intra- and extracellular analysis (e.g., crab:
Hienzel et al. 1993; mollusks: Hume and Getting
1982
; insects: Kitmann et al. 1995
). The tether
allows the animal to move relatively freely while the motor activity
and kinematics can be examined in a variety of active behaviors.
However, the tethered animal may not experience normal sensory inputs
or display its normal behavioral repertoire. These experimental
disadvantages raise the possibility that the tethered behaviors are not
the same as the freely moving behaviors of interest. To address this
concern and take full advantage of a tethered preparation, an
investigator can compare the behavior when the animal is freely moving
to that in the tethered situation (see Godden and Graham
1984
; Nye and Ritzmann 1992
).
We developed a tethered preparation that permits detailed kinematic
analysis of joint movements and intra- and extracellular analysis of
motor control while cockroaches perform a range of active behaviors. We
address the issue of behavioral relevance by comparing leg kinematics
and motor activity during tethered walking to that of freely moving
animals. For this comparison, we used a large data set from animals
walking freely on a treadmill that was obtained in a previous study
(Watson and Ritzmann 1999a).
Having established the behavioral relevance of the preparation, we then
examined a switch from walking to searching movements. When walking
insects suffer a loss of a reliable foothold, the behavior of their
legs switches from walking to "searching." The searching behavior
is used to find a foothold and thereby maintain stability
(Franklin and Pearson 1984). We take advantage of
insects' ability to switch between walking and searching to examine
multi-joint coordination during these behaviors. In this paper, we
compare coordination of the coxa-trochanter (CTr) and femur-tibia (FT) joints in the front (prothoracic) and middle (mesothoracic) legs during
walking and searching. The coordination of these principal joints
changes between walking and searching movements. In the DISCUSSION, we present hypotheses to account for the
differences in mesothoracic walking and searching movements. In the
companion paper (Tryba and Ritzmann 2000
), we use
intracellular analysis to begin to test these hypotheses.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Animals
Adult male death-head cockroaches (Blaberus discoidalis) were used in all experiments. Cockroaches were raised in our own colony descended from 250 adult animals generously provided by Dr. Larry L. Keeley of Texas A & M University. Cockroaches were housed in 20-l plastic buckets, half filled with aspen shavings, and were held at 27°C in a 12 h light:12 h dark circadian cycle. A commercial dry chicken starter and water were provided ad libitum. Only intact, undamaged cockroaches were used.
Kinematics
We marked the ventral surface of the segments making up the
body-coxa (BC) joint, the CTr joint, the anterior surface of the tibia
just distal to the FT joint, and the tibia-tarsus joint with Pilot
silver permanent marker ink to facilitate visualizing them against the
dark thorax of the cockroach (Fig. 1,
A and B). Note that
in previous papers (Watson and Ritzmann 1998a) the
distal joint of the coxa was referred to (incorrectly) as the
coxa-femur (CF) joint because the joint between the trochanter and
femur was assumed to be fused. More recent observations indicate that the trochanteral-femur joint makes small but important movements that
significantly influence placement of the tarsus (foot) (Watson et al. 1998
). Therefore to avoid potential confusion regarding which joint is studied, we will refer to the distal coxa joint by the
more correct terminology, the CTr joint. When comparing joint angles
published here with those measured during treadmill running published
earlier by Watson and Ritzmann (1998a)
, the reader
should compare our CTr data with their CF data.
|
Animals were tethered above an 11.5 cm × 7.5 cm × 3.0 mm glass plate made slick with microtome oil (Lipshaw Manufacturing, microtome oil No. 288). They were tethered with two (#2) insect pins each bent at a 90° angle and pushed through the animal's pronotum, lateral to each side of the animal's head (Fig. 1C). The other end of the pins were glued together and attached to a glass rod that was mounted on a micromanipulator. The sharp ends of the pins were placed through the apices of the dorsal "keystone"-shaped marking on the pronotum. The pronotum markings were used to direct pin placement so animals were tethered at the same relative points from animal to animal. Each hole was made slightly larger than the width of the pin to allow the animals to slide up or down on the tether to adjust their position relative to the substrate. After tethering the animals, a small droplet of cyanoacrylate glue was placed about 4 mm above the bend in the pins. This glue prevented the animal from sliding off the tether during an experiment. The ventral view of the animal was imaged via a 7.0 × 7.0 cm plane mirror mounted at a 45° angle to the glass plate and approximately 1.3 cm below it (Fig. 1C).
Ventral and lateral views of the running cockroach were videotaped at
either 125 frames/s with a Redlake digital high-speed video (HSV)
system or at 200 frames/s using a NAC 400 analog high-speed video
system. The ventral and lateral projections of the CTr, FT, and BC
joints of the leg of interest, as well as the anterior tip of the head
and the posterior tip of the abdomen, were digitized from each video
frame (see Watson and Ritzmann 1998a). The true BC, CTr,
and FT joint angles in three-dimensional space were calculated from the
ventral and lateral projected images (Marx et al. 1993
). An earlier kinematic study provided extensive three-dimensional kinematics for mesothoracic (T2) and metathoracic (T3) legs
(Watson and Ritzmann 1998a
) but not for prothoracic (T1)
legs. We wished to compare T1 walking versus T2 leg searching and
walking joint movements. Therefore T1 leg movements were monitored in a
limited number of freely moving animals using a treadmill that was
described previously (Watson and Ritzmann 1998a
).
Prothoracic (T1) treadmill running data were collected, with a slight
modification of the methods described extensively in Watson and
Ritzmann (1998a)
. To obtain an unobstructed lateral view of the
T1 coxa, about 2 mm of the lateral edge of the pronotum was trimmed
off. The lateral pronotum was trimmed bilaterally to avoid unevenly
reducing the weight of the animal.
Electromyograms
In experiments involving muscle activity, 50-µm bipolar
electromyographic (EMG) electrodes insulated to the tip were placed in
muscles of a right leg. Electrodes were placed in the main depressor
muscle either of mesothoracic leg muscle 135D or prothoracic muscle 85D
(Carbonell 1947). EMG electrodes were also inserted in
the tibial extensor muscle (mesothoracic leg muscle 142A; prothoracic muscle 92A) (based on the numbering system of Dresden and
Nijenhuis 1953) that extends the femur. The coxal depressor is
innervated by one slow excitatory motor neuron (Ds) and one fast
excitatory motor neuron (Df) as well as three inhibitors
(Pearson and Iles 1971
). EMG electrodes were also
inserted in the tibial extensor muscle (142a) (Dresden and
Nijenhuis 1953) that is innervated by one fast extensor
(FETi) and one slow extensor (SETi) motor neuron (Atwood et al.
1969
). For details of the recording sites and implantation
methods, see Watson and Ritzmann (1998a)
. Electrodes implanted by these methods do not interfere with leg kinematics (Watson and Ritzmann 1998a
). Animals having EMG wires
implanted in their T1 legs also had the serrations of their mandibles
trimmed off to prevent them from biting through the EMG wires with
their mouth parts. EMGs were amplified and the signal digitally
recorded on VHS tape using a VCR equipped with an A/D converter.
Synchronizing the electrical and kinematic data was facilitated by
recording an onset trigger pulse marking the start of HSV recording
onto both the electrical and video records. Additionally, a timing pulse for each video frame (125 Hz) was recorded along with EMG records. The start of the electrical record coinciding with the beginning of the digitized sequence was then located by finding the
correct time from the video onset trigger pulse.
Data analysis
Joint-angle records sampled at 125 Hz were smoothed using a moving boxcar averaging method of three data points each of 8 ms width. Smoothed joint-angle records were expanded by 40 times to give the same number of data points as the EMGs that were digitized at a sampling rate of 5 kHz. Relevant kinematic and EMG sequences were synchronized and merged into a common file using Data-Pac III software from Run Technologies. All data conditioning and subsequent analysis were also carried out with the Data-Pac III software.
Slow or fast motor neuron potentials were discernible in the depressor
coxa and extensor tibia records used for analysis. The time of onset
and termination of individual EMG spikes was determined by the time
when the voltage exceeded a common threshold. The threshold was set
independently for each file and was chosen to maximize the number of
EMG spikes detected. Deflections that clearly resulted from movement
artifact or cross-talk from other muscles were edited out by hand. Slow
depressor and extensor potentials were distinguished from cross-talk
due to other muscles by the broad muscle potential, the relatively
larger size and the presence of a neural spike preceding each muscle
potential. The burst onset and termination were defined as 0.5 ms
before the first spike and 0.5 ms after the last spike in a burst
(Watson and Ritzmann 1998a). The mean EMG frequency was
calculated as the mean of instantaneous frequencies within a burst. The
mean joint angular velocity was calculated as the regression slope of
the joint-angle amplitude for the time interval between EMG spikes. The
start of each joint movement was taken as 0.5 ms before the first
detectable joint movement in one direction, and the end was taken as
0.5 ms before the first detectable opposing movement. Extension onset
phase was calculated as the onset of extension (i.e., flexion to
extension transition) relative to the peak extension at the start (0)
and end (1) of each joint cycle. The tarsus touchdown and foot lift-off were determined directly from visual inspection of the video data.
Experimental protocol
Tethered cockroaches were videotaped while they generated leg
movements on an oiled glass plate (Fig. 1C). Tethered
animals running in a tripod gait could then be induced to switch to
searching leg movements by altering the distance between the body and
substrate. That change was accomplished by either raising the position
of the tether thus pitching the animals' anterior up or pitching the
substrate near the animals' anterior downward. The former method was
used to collect T1 tethered data and the latter method was used to
collect T2 data. T2 data were collected by the latter method as it
allowed for direct comparison with subsequent intracellularly recorded
data as described in the companion paper (Tryba and Ritzmann 2000). Tethered walking data were then compared with treadmill walking data collected at similar joint cycle rates.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The first step in establishing the behavioral relevance of our
tethered preparation was to determine how similar tethered walking
behavior is to free-ranging movements. Here we compare our tethered
walking data to that obtained during treadmill walking at similar joint
cycle rates. Comparison between pooled data from cockroaches walking on
a treadmill versus walking on the tether were made from data collected
at similar joint cycle rates because it is known that several variables
vary with walking speed (Delcomyn and Usherwood 1973;
Pearson 1972
). For example, motor neuron inter-burst interval, burst duration, and frequency vary with walking rate (Delcomyn and Usherwood 1973
; Pearson
1972
).
T2 tethered walking
FOOTFALL PATTERN.
Our study includes walking rates between about 3.8 and 7.5 Hz. At these
rates (previously referred to as slow running by Watson and
Ritzmann 1998a), cockroaches typically use a tripod gait during walking (Delcomyn 1971
). All walking video records used
for detailed analysis of single leg walking kinematics were visually
examined to determine the coordination between legs. In each case,
front and hind legs of one side were in-phase with the opposite side middle leg, which is consistent with a tripod gait. We further performed a detailed analysis of all six legs for 10 steps each in five
animals. In each case, they used a tripod gait and an example of that
analysis is shown in Fig. 1D.
COMPARISON OF T2 LEG CTr AND FT JOINT KINEMATICS.
One of our goals in this paper is to examine the changes in T2 leg
joint kinematics as the animal performs searching movements. Therefore
our analysis concentrated on movements of the T2 leg. Tethered walking
is qualitatively and quantitatively similar to treadmill walking in
terms of T2 CTr and FT extension duration, joint-angle excursions, and
maximum and minimum joint angles (Fig. 2A;
Table 1). The combined treadmill data
compared in Table 1 included 36 steps from nine animals (Watson
and Ritzmann 1998a), and the tethered data represent 17 steps
from three animals. In each comparison, mean values are used, and the
data are from animals that did not have EMG electrodes implanted in
their legs. Of the variables examined, significant differences existed
only for CTr and FT flexion duration and FT mean joint angular velocity
during extension (Table 1). When comparing many factors, there is a possibility that statistical tests will report false differences simply
by chance. At the 0.05 confidence level, one can expect that 0.7 of 14 of such comparisons to be statistically different by chance. Thus the
three differences we describe here appear to be characteristic of
tethered walking. The fact that flexion duration was different in both
the CTr and FT joints further supports the notion that the differences
we report here are not due to chance. The fact that 11 other parameters
were not significantly different in treadmill and tethered walking
suggests that tethered walking is similar (overall) to treadmill
walking.
|
|
|
T2 DS AND SETi EMGS.
The similarities in T2 CTr and FT joint extension kinematics during
treadmill and tethered walking suggested that the neural control of
these joints was largely unaffected by the tether. We tested this
hypothesis by recording EMGs from the extensor muscles of the CTr and
FT joints during tethered walking (Fig. 2B) and compared the
results for Ds and SETi to those collected by Watson and
Ritzmann (1998a) during treadmill walking. The leg joint
movements in the tethered animals did not appear to be altered following implantation of EMG electrodes (see also Watson and Ritzmann 1998a
).
|
|
RELATIONSHIP OF T2 DS AND SETi ACTIVITY TO JOINT KINEMATICS. While the distribution of Ds or SETi potentials within a burst remained largely unchanged whether the animal was walking on a treadmill or tethered, it is possible that the relationship of the potentials relative to the joint cycle is different when the animal walks on the two substrates. Two factors that are clearly different and could be important here are the differences in friction encountered by the tarsi (foot) and the lack of inertia experienced by the tethered animals. For example, the onset of joint movement could occur earlier (or later) and/or the joint velocity could be significantly altered during tethered walking. In that case, the distribution of EMG activity relative to joint cycle may be significantly different during tethered versus treadmill walking.
To look at the relationship of muscle activity to the joint cycle, we re-plotted the EMG data, now relative to the joint kinematics. Cumulative frequency histograms of Ds or SETi motor neuron potentials within the CTr or FT joint cycle (4.4 Hz) are shown in Fig. 5, A and B. The representative data shown are compiled from 13 CTr and FT joint cycles from one tethered animal. These data were also consistent with data compiled for six joint cycles taken from another tethered animal. In contrast to the histogram plotted in Fig. 4A, the histograms in Fig. 5, A and B, start and end at peak joint extensions (i.e., extension-flexion transition). As was the case for T2 treadmill walking data (Watson and Ritzmann 1998a
|
|
Tethered searching
SWITCHING FROM TRIPOD WALKING TO SEARCHING. Having demonstrated the behavioral relevance of tethered walking, we can now examine a context dependent change in behavior; i.e., the switch from walking to searching. In all cases examined, the tethered position allowed the cockroach to make tarsal contact with the substrate for all three legs during the stance phase. The distance of the anterior end of the animal relative to the substrate could be increased by either raising the glass rod attached to a manipulator, which resulted in pitching the tethered animals' anterior upward or by dropping the glass plate on which the animal walked. In either case, this change faithfully resulted in a change in leg movements from walking to searching. That is, the animal ceased running in a tripod gait and the T1 and T2 legs engaged in searching. The T3 legs stopped cycling and remained extended. The mean CTr joint angle measured when T3 legs stopped cycling was 103.31% of the maximum CTr joint angle measured at the end of stance phase during walking (n = 3 animals; 19 trials). These data are indicative of the extended position of the T3 legs while the animal performed searching movements with its T1 and T2 legs. Lowering the tether or raising the substrate usually resulted in all six legs once again engaging in a tripod gait. This switch was very reproducible. Mesothoracic (T2) leg data from five animals revealed that in 100% of the trials, including 10 consecutive trials from each animal, the animals switched from walking to searching on lowering the substrate. In 86% of the trials, the animals reestablished a tripod gait following the return of the substrate to the horizontal position.
For both cats (Pearson et al. 1992T2 LEG CTr AND FT JOINT KINEMATICS DURING SEARCHING.
Unlike walking, searching may involve simultaneous extension of
segmental pairs of legs (searching: Franklin and Pearson
1984; walking: Pearson 1976
; Pearson and
Fourtner 1975
). As was the case for locust searching
(Franklin and Pearson 1984
), we occasionally observed
nearly synchronous extension of pairs of either T1 or T2 legs during
the aerial phase of a searching episode (n = 5 animals,
15 observations). We did not quantify the frequency of occurrence that
both legs of a thoracic segment extended during the aerial phase of
searching as it was not necessary for the conclusions drawn based on
these data. In tethered animals, this frequency may increase due to the
fact that the requirements for the legs to support the animal above the
substrate are reduced compared with freely behaving animals.
Nonetheless, these data suggest that left and right legs of the same
segment are not as tightly coupled (in anti-phase) during searching as
they are during walking.
|
T2 DS AND SETi EMGs. Consistent with the advance of the FT joint movement relative to CTr movement, during the extension phase of searching versus walking, SETi activity onset precedes Ds during searching (Fig. 3C). The mean onset time for Ds during searching occurs at 15.76% (n = 4 animals, 20 cycles) of the SETi cycle, that is, after SETi activity onset (Fig. 3A). We also found systematic changes in instantaneous motor neuron frequency in searching relative to walking at similar joint cycle rates (CTr joint cycle period, P = 0.081; FT joint cycle period, P = 0.186). The mean and peak instantaneous SETi frequency was always higher during searching versus walking (Fig. 8, A and B). In contrast, the mean and peak Ds frequency was lower during searching than during walking (Fig. 8, A-D; P < 0.0001 for both mean and peak SETi and Ds values). The data shown (Fig. 8, A-D) include five SETi cycles and four Ds cycles from one animal during searching and walking. The same relation was also found in another animal for seven searching and five walking SETi cycles and six walking and six searching Ds cycles (data not shown). In contrast to tethered walking the mean depressor or extensor motor neuron firing rate was not directly proportional to the mean CTr or FT joint extension velocity (data not shown).
|
Comparison between T2 searching and T1 walking
Although the joint kinematics of the T2 legs go through
considerable changes in switching from walking to searching, T2
searching has some features in common with the walking movements of the front legs. In insects, each segmental pair of legs plays a unique role
during walking (Full et al. 1991). In cockroaches, hind
(T3) and middle (T2) legs provide propulsive forces while front (T1) legs provide braking forces, investigate footholds, and contribute to
changes in body orientation. If the movements of T1 legs during walking
are similar to T2 searching movements, it is possible that during
searching the T2 legs move away from their typical walking behavior and
closer to the walking movements of the T1 legs. Indeed, in walking
crickets, the FT joint of the T1 leg extends prior to the CTr joint
during swing phase, whereas the T2 leg CTr extends prior to the FT
joint during stance (Laurent and Richard 1986
). The T1
legs of crickets during walking thus show similarity to the T2 legs of
the cockroaches during searching.
Unfortunately, the treadmill data that we used to compare to our tethered data did not include the front legs. Thus we could not compare T2 searching data to cockroach T1 walking data without making our own observations of T1 legs during treadmill walking (Fig. 9, A and B). A complete analysis of walking and searching in the T1 legs was beyond the scope of this study. However, we were able to obtain sufficient kinematic data to confirm the similarity to cricket front leg data and to make a qualitative comparison to our T2 searching data. Recording EMGs from T1 legs during free walking presented serious technical challenges. Because of the wide range of motion of the front legs, coupled with the proximity to the head, the animal can readily tangle wires with the other legs or bite through them, thereby breaking the wires. Thus we were only able to obtain clear EMG data for two animals (of more than 50 trials). Again, these data were sufficient to make some qualitative comparisons.
|
T1 WALKING KINEMATICS.
As in the cricket, the cockroach protracts the T1 leg during swing
phase. Reaching forward during swing involves a reduction of the
body-coxa (BC) joint angle (points A-B in Fig. 9A;
n = 12 steps, n = 2 animals). To
accomplish foot set-down, the FT joint is extended at mid-swing phase
followed by extension of the CTr joint that depresses the leg onto the
substrate (point B in Fig. 9A). These actions are in
contrast to those of the T2 and T3 legs during walking where both FT
and CTr joints extend during stance with CTr joint extension preceding
that of the FT joint (Watson and Ritzmann 1998a).
Subsequent to foot set-down (point B Fig. 9A), the CTr joint
flexes and then extends again, while the FT joint either flexes
throughout stance (Fig. 9A) or is held at a relatively
constant angle after a brief flexion (Fig. 9B). During the
stance phase, the coxa is rotated posteriorly at the body-coxa (BC)
joint (Fig. 9A). The second extension of the CTr joint
during stance coupled with the rotation of the BC joint thrusts the T1
leg away from the anterior of the body (Fig. 9A).
T1 TREADMILL WALKING EMGS AND KINEMATICS. The two animals from which we successfully recorded EMGs during treadmill walking used strategy 1 kinematics (Fig. 9B; n = 10 steps). The use of strategy 1 kinematics by these animals was not a consequence of wiring the animals for EMGs as strategy 1 kinematics were also observed during tethered walking from an animal not wired for EMGs (data not shown). The motor activity was consistent with the unique kinematics of T1 legs and is in contrast to T2 and T3 legs during walking. That is, the SETi burst onset occurred about 40 ms prior to that of Ds (Fig. 9B). Also, the SETi and Ds activity began prior to foot touch down during the swing phase (Fig. 9B). During protraction of the T1 leg, where the animal brought the leg forward, SETi bursts at an initial high frequency and was occasionally coupled with recruitment of the fast extensor tibia motor neuron (FETi; Fig. 9B). After the FT joint extended, SETi activity declined and Ds activity began. The early high-frequency SETi was followed by much lower frequency secondary activity after foot set-down and resulted in SETi having a distinct biphasic pattern of activity (Fig. 9B).
Both the joint kinematics and EMG activity of the T1 leg are consistent with data reported from front legs of other insects (Burns and Usherwood 1979T1 SEARCHING KINEMATICS. We also briefly looked at joint kinematics during T1 searching movements (Fig. 9C; n = 2 animals, n = 10 cycles). There were some differences in the coordination of T1 CTr and FT joint kinematics during searching versus walking. Searching involves larger joint excursions than walking (Fig. 9C). This finding is not unexpected as we induced searching movements by raising the tether, pitching the animals' anterior end away from the substrate. In contrast to T1 walking, searching involved CTr extension primarily during leg protraction (Fig. 9C). Thus T1 searching kinematics (Fig. 9C) resemble those of T2 searching (Fig. 7A) as both involve extension of the FT joint prior to the CTr joint and both joints extend during leg protraction.
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
A variety of experiments using deafferented cockroaches suggested
that coordination of muscles acting on a joint can be centrally patterned (Pearson 1972; Pearson and Iles
1970
). However, it is unclear whether the fictive coordination
produced by these restrained preparations is the same as that
underlying actual walking, struggling or rocking behaviors
(Reingold and Camhi 1977
). Further, it is known that
feedback from leg sensory structures is important in coordinating leg
joint movements in a normal way during walking (cat: Duysens and
Pearson 1980
; Pearson et al. 1992
; stick insect: Bassler 1983
; locust: Burns 1979
;
cockroach: Delcomyn and Usherwood 1973
;
Krauthamer and Fourtner 1978
; Spirito and
Mushrush 1979
) and searching for a foothold (stick insect:
Karg et al. 1991
).
It is clear from these studies that understanding the neural basis of behaviors such as walking and searching requires clearly defining the behavior examined and studying the behavior under circumstances where essentially normal movements and sensory feedback are produced. To define the behaviors examined, we characterized the relationship between motor neuron activity and kinematics of some of the joints involved. To ensure that the behaviors studied involved essentially normal movements, we compared the behavior of our tethered preparation to similar data sets from freely walking animals on a treadmill. We found that many of the joint kinematic and motor parameters measured during tethered walking are similar to those during treadmill walking. Our subsequent analysis focused on those aspects of the tethered movements that are similar to treadmill walking. Where differences in the kinematics and neural control were observed, we will speculate on the possible role of sensory feedback in causing these differences.
Our data further showed that T2 CTr and FT joint kinematics during
searching were markedly different from those observed for walking.
Preliminary observations suggest that there are also some differences
of CTr and FT joint coordination in the T1 legs during walking versus
searching. While the T1 and T2 searching joint kinematics were similar,
the T1 walking joint movements also shared some similarities with T2
searching during initial leg protraction. Our data suggest several
possibilities for neural control of two leg joints during two different
behaviors. For example, compared with walking, there is a delay in
onset of T2 CTr extension during searching. This delay may be the
result of direct inhibition of Ds, reduced excitation, delayed
excitation or, more likely, a combination of these mechanisms. We begin
to address these hypotheses in the companion paper that follows
(Tryba and Ritzmann 2000).
Comparison of tethered and treadmill T2 walking
While many of the T2 joint kinematics and EMG parameters for tethered and treadmill walking were similar, there were some differences. For example, although we found a linear relationship between mean EMG activity and mean joint velocity, the slope of the line was lower for tethered data than had been previously shown for treadmill walking.
There are several possible explanations for this difference. First, it is possible that the insertion of EMG wires into the legs somehow influenced the rate of leg extension during tethered walking. This was not the case in treadmill walking but could be when the animal is tethered. While we did not test this hypothesis directly, FT joint angular velocity during extension was actually lower in tethered animals that are not wired for EMGs than in animals during treadmill walking (Table 1). These data suggest that electrode implantation alone cannot account for the decline in joint velocity during tethered walking. Second, the tethered animals are not experiencing the normal whole body inertia of a freely moving animal, and that could account for some of the decrease in joint velocity relative to EMG frequency. Third, the decline in mean joint velocity may also result from the relative degree of slippage or drag that the foot experiences during tethered walking. For example, it may be that with application of different amounts of oils to the glass plate, or using oils of different viscosity, the tethered joint velocity during walking would more closely resemble that during treadmill walking. One would expect a lower joint velocity if the resistance to foot movement on the glass plate is higher than during walking on a treadmill.
There is some evidence to suggest that the tethered animals experience
an increase in resistance to foot movement during stance, and this
results in a decline in extension velocity at a given muscle activation
rate. Compared with freely walking cockroaches, the relationship
between Ds frequency and step period is lower for animals that have an
increase in resistance to extension when they drag a weight
(Pearson 1972). Along these lines, animals that were
climbing over obstacles had a lower slope in this relationship than
animals walking on a treadmill (Watson et al.
1998
). In that case, the forward inertia of climbing
animals would presumably be less and extension resistance would
increase as the animal climbed against gravity. Additionally, during
walking in locust, there is a decrease in swing duration when extension
resistance is increased (Newland and Emptage 1996
). Thus
if there is an increase in resistance to extension during tethered
cockroach walking, one would a expect a decline in flexion duration at
a given step period for both the CTr and FT joints, and this was shown
to be the case (Table 1).
We also noted that the mean extension velocity during tethered walking (at a particular rate) shown in Table 1 was higher than would be expected to be achieved by increasing slow motor neuron activity alone (i.e., extrapolation of regression lines in Fig. 6, A and B). Although we did not record EMGs during tethered walking from animals represented in Table 1, we suspect these animals achieved higher extension velocities and faster walking rates (than animals in Fig. 6, A and B) via the recruitment of fast motor neurons.
Tethered T2 leg and freely behaving searching
The tethered preparation provided several technical advantages to
studying searching. While freely behaving cockroaches produce leg
searching movements following loss of a foothold as they walk over
uneven terrain (Franklin and Pearson 1984), it would be
difficult to define leg searching kinematics and motor coordination
under these circumstances. True joint-angle calculations in three
dimensions require a lateral and ventral view. Obtaining an appropriate
ventral view would not be tractable under conditions involving terrain complex enough to reliably evoke searching leg movements (see Franklin and Pearson 1984
). It would also be problematic
in freely moving animals to collect enough data at similar joint cycle
rates to make comparisons between data sets. Ideally, the true joint angles should be measured so there can be consistency among different investigators studying the same behaviors.
There is qualitative data from locusts to support the notion that the
tethered searching described here shares several similarities with
searching movements of freely behaving insects (Franklin and
Pearson 1984). These similarities include the following: the searching leg undergoes rapid elevation and depression, there is marked
extension at distal joints during the aerial phase, the searching
pattern continues for several cycles, the behavior is terminated when
the animal's legs stopped cycling, and the behavior ceases if the leg
engaged in searching encounters an object and finds a foothold.
Tethered searching and T2 motor neuron activity
Our data are also consistent with and provide further support for
the hypothesis proposed by Delcomyn (1987) that
searching and walking are distinct behaviors. Delcomyn
(1987)
examined searching Ds activity of cockroaches tethered
above a glass plate. He found that for T2/T3 legs there was a decrease
in Ds frequency and inter-burst interval during searching compared with
walking. We also found that there was a lower frequency of Ds activity
and a higher SETi frequency during searching. These findings may in
part account for the smaller CTr and larger FT joint excursions during
searching versus walking. Further, we extended the initial observations of Delcomyn (1987)
because we included the neural
control and kinematics of the T2 CTr and FT joints and we
quantitatively compared many aspects of searching joint kinematics to
those observed for walking.
We can now also define the searching behavior by extension of the FT
and CTr joints during leg protraction. In all cases of searching
examined, the onset of FT extension preceded that of the CTr joint.
Further, the onset of the extension cycle of searching included a
characteristic high-frequency SETi activity that was accompanied by
little or no Ds activity (Fig. 7B). Thus searching included
characteristic patterns of motor activity and the consequent joint
kinematics. As both the T2 CTr and FT joint kinematics and motor
activity of searching are markedly different from walking, either of
these parameters can be used to identify ongoing searching (Figs. 2, 3,
and 7). In contrast to our data, Delcomyn (1987) actually recorded two different patterns of Ds motor activity, and it
is not clear whether both result from variability in searching or are
different behaviors. Whether one pattern or the other was observed
depended on whether some or none of the legs were supported by
substrate (Delcomyn 1987
). There is evidence that
different behaviors may be expressed depending on degree of substrate
contact. For example, it is known that the flight motor activity is
evoked following substrate removal and loss of tarsal contact
(Kramer and Markl 1978
). To control for this
possibility, searching in this study was always examined under
conditions where the tarsi of all legs could contact the substrate.
Furthermore, characterization of multiple joints ensured that we were
examining the same behavior in each preparation.
T1 walking and searching
The CTr and FT joint movements vary among the three sets of legs.
T1 legs showed distinct differences from T2 legs during walking while
T1 leg data included fewer differences between searching and walking
than T2 legs did. For example, the onset of FT extension that is
followed by CTr extension during initial protraction of the leg is not
markedly different from T1 walking. Thus there may be similar neural
architecture underlying the coordination of the principal leg extensors
during onset of T1 searching and walking (Fig. 9, A-C).
During T2 searching (that bears kinematic resemblance to T1 searching)
and T1 walking, the onset of the extension cycle included
high-frequency SETi activity at a time when there was cessation of Ds
activity (Figs. 7B and 9B). There does appear to
be some consistency in the searching behavior of T1 and T2 legs. Both
T1 and T2 searching involves extension of the FT joint prior to CTr
extension and continued extension throughout protraction (Figs.
9C and 7B). This observation suggests there may
be similar Ds and SETi motor neuron activity during these behaviors.
Since activity of Ds and SETi and the resultant CTr and FT joint
movements are clearly different for T1 versus T2 and T3 legs during
walking (T1: Fig. 9B; T2: Fig. 2B; T3:
Watson and Ritzmann 1998a), it appears that walking
involves different neural mechanisms coordinating the FT and CTr joints
in the T1 legs than in T2 (or T3) legs. The unique coordination of the
CTr and FT joints for each of the legs during walking may reflect the
roles these legs play during walking. For example, the fact that the T1
legs extend the principal leg joints (primarily the FT joint) during
the protraction phase of walking allows the animal to use the front
legs to investigate a larger area ahead of it for footholds than if T2
or T3 walking kinematics were used.
Coordination of CTr and FT joints during walking and searching
Several reports have suggested that movement at multiple joints in
a single leg is generated by the activity of separate joint oscillators
(Bassler 1993; Grillner 1981
;
Nothof and Bassler 1990
). The appropriate coordination
of the joint oscillators to produce walking or searching movements is
thought to be orchestrated by sensory feedback (Bassler
1993
; Bassler et al. 1991
; Braunig and Hustert 1983
; Hess and Buschges 1999
;
Pearson et al. 1976
; Zill et al.
1981
). Our data suggest that distinct mechanisms couple the
movements of the CTr and FT joints during the onset of the extension
cycle when the animal searches versus when it walks. The fact that
there is more variability in Ds onset times during searching suggests
that the CTr and FT joint coordination during walking is more tightly
coupled (Fig. 3A), and this result would not necessarily be
expected if the same mechanisms were responsible.
Different mechanisms are also necessary to account for the coordination
of multiple legs during searching and tripod walking. For example,
walking does not involve simultaneous extension of any segmental pair
of legs (Pearson 1976). However, the in-phase extension
of segmental leg pairs can be observed for T1-T2 legs during tethered
searching (see also Franklin and Pearson 1984
). The 0.5 phase relationship between legs on opposite sides of the same segment
during walking has suggested the possibility that reciprocal inhibition
of the CTr joint oscillators coordinates the phase relationship between
contralateral leg pairs (Pearson 1976
; Pearson
and Fourtner 1975
). This coordination does not appear to be the
case during searching.
Role of afferent information in the behavior of legs during searching and walking
Leg loading and joint angle information may play a role in
modulation of CTr and FT joint coordination (see also, Bassler 1993; Bassler et al. 1991
). Several lines of
evidence suggest that some of the differences in T1-T3 Ds and SETi
coordination and joint kinematics may be due in part to differences in
loading. For example, in both insects and cats, transition from stance to swing phase requires unloading of extensors and leg extension (Cruse 1976
; Cruse and Saxler 1980
;
Pearson 1976
; Whelan et al. 1995
). In our
study, induction of tethered T1 and T2 leg searching resulted in the T3
legs extending and remaining in stance phase until the tilted substrate
was rolled. Presumably rolling the pitched plate unloaded T3 legs
allowed them to engaged in searching. As is the case for walking, a
reduction in load supported by a leg probably permits the transition
from stance to swing during searching. We suspect that unloading the
legs alone allows leg protraction to occur but did not induce the
searching motor pattern. Instead, unloading the legs at a particular
joint angle may permit expression of the searching motor pattern. The
fact that induction of tethered searching occurs over a relatively
small range of glass plate angles, regardless of walking rate or rate
of pitching the glass plate, suggests that there may be specific joint
angle requirements met under these circumstances for induction of searching.
Conclusions
Our data show that tethered walking T1 and T2 leg CTr and FT joint coordination closely resembles that during treadmill walking. This conclusion is based both on joint kinematic data and principal leg extensor motor neuron activity. Additionally, the kinematics of walking were distinct from searching. Accordingly, we found a very different CTr and FT extensor EMG pattern during T2 searching versus walking.
In the companion paper, we take advantage of our tethered preparation
to investigate differences in control of the T2 CTr and FT joints
during walking and searching. In addition to the properties examined
here, we are also able to record intracellularly from the thoracic
ganglia during either behavior. We use this technique to examine the
hypothesis that inhibition patterns extensor motor neuron activity
during walking as was suggested for stick insects (Buschges
1998; Godden and Graham 1984
) and Ds during searching (this study).
![]() |
ACKNOWLEDGMENTS |
---|
The authors thank Drs. Sasha N. Zill and Joanne Westin for helpful comments on the manuscript and Dr. James T. Watson for technical support/advice, use of his T2 leg treadmill walking data for comparison, and very helpful comments. Thanks also to A. Pollack for providing tripod stance duration data for Fig. 1.
This work was supported by Office of Naval Research Grant N0014-99-1-0378 to R. E. Ritzmann.
![]() |
FOOTNOTES |
---|
Present address and address for reprint requests: A. K. Tryba,
Dept. of Organismal Biology and Anatomy, The University of Chicago,
Anatomy Bldg., 1027 E. 57th St., Chicago, IL 60637.
E-mail:
Tech10S{at}techsan.org
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 29 November 1999; accepted in final form 13 March 2000.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|