J Comp Physiol A (1992) 170:509-520 Joumal of Sensory, Comparative.--,,

�9 Springer-Verlag 1992

Range fractionation in the locust metathoracic femoral chordotonal organ

Thomas Matheson

Department of Zoology, University of Canterbury, Christchurch 1, New Zealand 1

Accepted January 22, 1992

Summary. Insect femoral chordotonal organs are inter- nal proprioceptors which monitor the position and movements of the femur-tibia joint of the leg. The locust (Locusta migratoria) metathoracic femoral chordotonal organ is composed of approximately 100 neurones with a variety of response properties. In this study intracellu- lar recordings were used to examine the range fractiona- tion of phasic and tonic responses to tibial movements. Some neurones responded across the full range of leg angles, while others had restricted response ranges, and could therefore act as labeled lines. Neurones with max- imal firing at mid-angles are described for the first time in a locust femoral chordotonal organ. Responses are discussed in terms of underlying structural constraints on signal transduction.

Key words: Locust- Insect - Femoral chordotonal organ - Proprioceptor - Range fractionation - Hysteresis

Introduction

Many sensory systems must be able to respond to stimuli which vary in intensity over several orders of magnitude. In addition, they must be able to represent any relevant stimulus with a certain degree of precision. Because of the limited range of firing frequencies available to single neurones, these two requirements often conflict with each other. The decapod propodite-dactylopodite (PD) chor- dotonal organ is a multineuronal proprioceptor which contains 3 types of sensory neurones (Types 1-3). It monitors movements of the PD joint. Its Type 3 neurones which code for wide ranges of stimulus intensities are generally quite insensitive (Mill and Lowe 1972). On the other hand, sensitive receptors often respond only to a restricted subset of possible stimuli (e.g., some Decticus

Abbreviation: (mt) FCO (metathoracic) femoral chordotonal organ

1 Present address." Department of Zoology, University of Cam- bridge, Downing Street, Cambridge CB2 3EJ, England

primary auditory neurones (Kalmring et al. 1978)). Neurones which monitor appendage joint angles face a further complicating factor: they must unambiguously represent a parameter of movement such as velocity or acceleration over a range of leg angles. These neurones' firing frequencies may therefore be influenced by both velocity and absolute leg angle.

By increasing the number of receptor units (neurones) within a sense organ, it is possible to overcome these problems. At least 2 scenarios are possible: (1) each unit responds to the full range of stimuli, but is relatively insensitive. The sensitivity of the receptor organ as a whole is increased by subsequent convergence of afferent information onto interneurones (Aidley 1978, and see Fig. 5 in Hofmann et al. 1985); (2) individual units respond with great accuracy, but only within narrow ranges of stimulus intensity. Different neurones respond in discrete (or slightly overlapping) ranges, in such a way that the combined response of the organ codes for all stimuli (range fractionation) (Cohen 1963). If a neurone's response range is sufficiently narrow, then that neurone may function as a "labelled line". The movement units of the crustacean myochordotonal organ (MCO2) func- tion in this way. Their firing frequency codes for joint velocity, while the presence of firing indicates that the leg is within the response range (position information). Evoy and Cohen (1969) suggested that this type of neurone may be necessary in MCO2 to allow fine control of movement. Many vertebrate and invertebrate sensory systems exhibit range fractionation. These include taste and temperature receptors in the cat tongue (Cohen et al. 1955; Dodt and Zotterman 1952); arthropod limb pro- prioceptors (Burns 1974; Cohen 1963; Hofmann et al. 1985; Young 1970; Zill 1985a); insect auditory receptors (Kalmring et al. 1978; Rheinlaender 1975); insect mecha- noreceptors (Shimozawa and Kanou 1984a, b); insect vibration receptors (Kiihne 1982); equilibrium receptors in vertebrates (Lowenstein and Roberts 1950); and in- vertebrate statocysts (Cohen 1960).

The femoral chordotonal organs (FCOs) of insect legs are multineuronal sense organs which monitor move-

510 Th. Matheson: Chordotonal organ range fractionation

ments of the tibia relative to the femur. They have a role in shaping leg movements (Hofmann et al. 1985; Usher- wood et al. 1968; Zill 1985a, b). FCO afferent neurones make direct connections with motor neurones, spiking and non-spiking local interneurones, and interganglionic interneurones (Burrows 1987, 1988; Burrows et al. 1988; Laurent 1988; Siegler 1984; Wilson 1981). Several au- thors have mentioned range fractionation in insect joint chordotonal organs (Burns 1974; Hofmann et al. 1985; Matheson 1990a; Zill 1985a), but, in contrast to the locust auditory system (Kalmring et al. 1978; Rheinlaen- der 1975), there has not yet been a thorough study of this aspect of sensory function in joint chordotonal organs. Matheson (1990a) outlined the types of receptor neu- rones in the mtFCO, while a second report (Matheson 1992) described the relationship between response type and central projection pattern. Field (1991) has recently described a mechanical system underlying range frac- tionation in the mtFCO, but this has not yet been cor- related with detailed physiological studies. Hofmann et al. (1985) have discussed aspects of range fractionation in the stick insect FCO, giving examples of the main response types. This valuable information however, was not presented in a way that gave an appreciation of the overlap between response ranges; nor did their few exam- ples reveal much about the more subtle differences be- tween neurones (because the examples were either "typi- cal" of a class, or "extreme").

The main aim of the present paper is to illustrate and discuss the response ranges, variation between units, and overlap of responses in the locust mtFCO. I have recor- ded intracellularly from individual locust mtFCO neu- rones while moving the FCO apodeme to mimic leg movements in the range 0-120 ~ (encompassing the angles used by the locust during normal activities). I examine both tonic (position) and phasic (velocity) range frac- tionation. Because the acceleration of movements was not controlled I do not describe the range fractionation of putative acceleration receptors. The hysteresis of the tonic responses is also illustrated and discussed. This present paper complements and expands on information presented in 2 previous papers (Matheson 1990a, 1992) which describe (respectively) the relationships between these FCO neurone's responses and their peripheral and central morphology. In order to emphasise these rela- tionships, the previous papers avoided discussing the details of individual neurone's responses (although all the available information was used to ascribe each neu- rone to the response classes used in all three papers). This present paper addresses questions concerned with details of individual responses by presenting and discussing many specific examples. It should provide the basis for a greater understanding of the complex flow of informa- tion originating in this sense organ and impinging upon motor neurones and interneurones in the CNS.

Materials and methods

Adult locusts (Locusta migratoria) from our laboratory culture were used for all experiments. The results are based on 189 success-

ful recordings from 256 animals. These results are derived from experiments reported in Matheson (1990a, 1992). The responses of 110 FCO neurones were recorded from axons near the metathoracic ganglion so that the neurones' central projections could be stained by dye injection (Matheson 1992). The axons of a further 79 neu- rones were penetrated in mid-femur so that they could be physiolog- ically characterised before dye injection was used to reveal the positions of their somata in the FCO (Matheson 1990a).

The terminology, and the stimulating and recording techniques used have been fully described previously (Matheson 1990a, 1992), and are only briefly mentioned here for clarity.

Animals were restrained ventral side up in plasticene. Move- ments of the mtFCO apodeme at controlled velocities (Matheson and Ditz 1991) mimicked tibial movements between 0 and 120 ~ A complete stimulus sequence moved the apodeme from 0 ~ to 120 ~ and back in 12 equal ramps (6 each way), thus revealing any hysteresis in the tonic response at any given angle. Ramps were separated by 1 s or 3 s "hold" periods. The FCO flexor strand was held static at 60 ~ Matheson (1990a, 1992) discusses possible biases introduced by holding the flexor strand motionless while moving the apodeme strand. In summary, these biases are likely to be small (if present) because the flexor strand has at most a minor and indirect effect on mtFCO neurones (Field 1991).

Intracellular recordings were made from FCO axons either close to the metathoracic ganglion (in nerve 5), or in mid-femur (n5B1). All recordings were made with cobalt filled microelectrodes. The neurones presented in this paper were all identified as FCO afferent neurones by examination of either their central projections or the position of their cell body in the FCO. Drops of locust saline (pH 6.8) were added if required to prevent drying out of the preparation. Responses to stepwise ramp-and-hold stimuli at a range of velocities were recorded on tape for later filming and analysis. The position and velocity (but not acceleration) of the apodeme were controlled. The order in which different velocities were tested was not strictly controlled. The first 2 velocities tested were always 400~ and 40~ The subsequent velocities used depended on the sensitivity of the neurone, and were applied in an unpredictable sequence. Putative acceleration receptors have been discussed by Matheson (1990a), and the term "acceleration receptor" is used in the same way here to refer to these units.

Tonic firing frequencies were averaged from 3-8 complete step- wise ramp-and-hold stimulations (each with ramps at a different velocity). Error bars indicate 4- 1 SE, and are often hidden by the symbols. In cases where hysteresis caused an ambiguity in determin- ing the leg angle at which maximal firing occurred, the two curves were averaged in order to yield the value used in the neurone's response abbreviation.

Results

The terminology used here to describe FCO neurones' response types follows that described in detail by Math- eson (1990a). I f a neurone had a tonic discharge while the femur was held at a constant angle, the neurone was deemed to be position sensitive (abbreviation: P). The angle at which the highest firing frequency occurred was added as a superscript (e.g., p2o). The abbreviation px was sometimes used to refer to a group of tonic neurones irrespective of their individual tonic maxima. Neurones which responded to leg movements in a velocity depen- dent way were given a "V" code. The direction of move- ment that excited the neurone was added as a superscript "plus" or "minus" sign (+ =tibial flexion, - = t i b i a l extension). I f the neurone matched the criteria for ac- celeration sensitivity (see Matheson 1990a for discus- sion), it was given a "?A" code. Again, the sign of accel-

Th. Matheson: Chordotonal organ range fractionation 511

eration that excited the neurone was added as a super- script "plus" or "minus". Acceleration sensitive neurones which responded to both signs were classed as ?A § - . Another type of acceleration sensitive neurone fired in response to both signs of acceleration, but only at the start of movements . These were classed as "?AS" neu- rones (for Acceleration at the Start). When a neurone responded to more than one parameter of a movement the appropr ia te abbreviations were combined.

Tonic responses

All tonically active neurones (including phasic-tonic re- sponse types) have been pooled in this section. The full response type of any particular illustrated neurone is given in the appropr ia te figure legend. Further examples are given in Matheson (1990b).

Neurones chosen for recording were selected to give examples of as many response types as possible (i.e., not all neurones encountered were recorded and cat- egorised). I t is therefore invalid to compare the number of neurones having firing maxima at different angles (although I present the numbers for completeness). Neurones with max ima at either 0 or 120 ~ were encoun- tered much more often than those with maximal firing at mid-angles.

Maximal tonic firing frequencies differed between neurones, and occurred at different leg angles (Fig. 1). There was no consistent relationship between maximal firing frequency and leg angle causing maximal firing.

M a x i m u m at f u l l f l e x i o n (0~ Twenty-eight neurones had tonic maxima at 0 ~ Four of these were relatively insen- sitive to position between 0 and 80 ~ but responded with a marked reduction in firing to more extended angles (Fig. 2A, and upper curve in Fig. 2D). These neurones all had relatively high firing frequencies at 0 ~ ( > 25 Hz). Eleven other neurones had a more linear response curve (Fig. 2B). I t is likely that there is a cont inuum of response type, rather than separation into 2 distinct types. For example, the neurone represented in the upper curve in Fig. 2B has the steepest par t o f its response between 80 and 120 ~ yet does not have a plateau between 0 and 80 ~

Thirteen other neurones with tonic max ima at 0 ~ had narrow response ranges restricted to angles between 0 and 40 ~ (Fig. 2C). Generally these neurones had maximal firing rates of less than 25 Hz, but one (upper curve in Fig. 2C) reached 40 Hz at 0 ~ These neurones very rarely fired tonically at angles greater than 40 ~ .

Hysteresis in the response curves was evident in vary- ing degrees for different neurones (Fig. 2A-D). Figure 2D compares the extent o f hysteresis in 2 neurones with max imum tonic firing at 0 ~ The lower curve represents one of the most extreme cases recorded, with a hysteresis in excess of 20 Hz at a leg angle of 40 ~ (approx. 80% of its max i m um firing rate). This neurone was also quite sensitive to velocity of flexion (P~ § response), but there was no evidence that the tonic firing rate following flexions was affected by the velocity of the preceding movement . Other, equally sensitive, pxV neurones did not exhibit such a large hysteresis.

7 0 . 0 '

60.0

~ " 5 0 . 0 v

4 0 . 0

30.0- ~ i /" "~"" "-'--" ~ 2 0 . 0 -

0 . 0 , ," . . . . . . . ;" . . . . . , . . . . . , : . . . . . . . . . . ; ' ~ " ' ~ - , 0 2 0 4 0 60 8,0 1 0 0 1 2 0

L e g a n g l e ( d e g r e e s )

Fig. 1. Tonic firing of 7 neurones at static femur-tibia angles be- tween 0 and 120 ~ Each line (A-G) represents the firing frequency (vertical axis) of a single neurone, recorded with the chordotonal organ apodeme set at various displacements (calibrated as femur- tibia angle: horizontal axis). Tonic firing was measured over a 3 s period immediately after each angle was set. Each value was cal- culated by averaging the firing frequencies obtained from 3-8 tests, each with ramps of a different velocity preceding the static (hold) phases. Each test started with the leg angle set to one extreme (0 ~ or 120~ and consisted of a series of 20 ~ ramp-and-hold steps progressing to the opposite extreme angle, then returning to the start angle. Static angles were therefore approached from both more extended, and more flexed positions. Error bars and symbols have been omitted from this summary figure, which serves simply to illustrate a variety of response profiles. A = P2~ + ?A +, B = p6o .9A9., C=psov + , D=P~ +, E= ps~ - , F=p6~ -, G= p12oV-?A -

M a x i m u m at 20 ~ Two of the 8 neurones with a tonic peak at 20 ~ had max imum firing frequencies of 60-80 Hz (e.g., upper curve in Fig. 3A), al though mos t max ima were less than 40 Hz. Two types of curves were apparent (contrast upper and lower curves in Fig. 3A). The firing of some neurones declined linearly for angles greater than 20 ~ (lower curve), while others (upper curve) have a broad plateau of high firing frequency between ap- proximately 20 ~ and 80 ~ , with relatively sudden decreases at either extreme of leg movement . Note that the hys- teresis of the neurones in Fig. 3A and B reverses between 20 ~ and 60 ~ .

M a x i m u m at 40 ~ Thirteen neurones had tonic max ima at 40 ~ Two neurones had consistently high firing rates between 0 ~ and 60 ~ and a steady decline between 60 ~ and 120 ~ The remaining 11 p40 neurones had quite low firing rates at both 0 ~ and 120 ~ (Fig. 3B). Hysteresis was present to varying degrees, with some neurones having markedly different firing rates following ramps in either direction (upper curve in Fig. 3B), while others showed little difference (lower curve in Fig. 3B). Neurones of the former type often did not fire tonically during holds following one direction of movement (cf. examples of p6o units in Fig. 3C). The hysteresis of 5 neurone 's curves reversed between 60 and 80 ~ (Fig. 3B).

M a x i m u m at 60 ~ . Responses seen in other mid-posi t ion sensitive neurones were present in p6o n e u r o n e s but are not illustrated again (i.e., high firing frequency between 0~50 ~ with sharp decline at 80-100~ broad peak of max- imum firing between 40 ~ and 80 ~ with sharp declines at 0-20 ~ and 100-120~ variable degrees of hysteresis). None of the 9 p60 neurones had response curves in which

512

A4o.o~

ao.o!

20.0

t t0.0.

0.0 . . . . 0 20 40 60

C 40.0

o ~ a0.0

~20.0

10.0

0.01 m,

o ~ ~ ~ 50 Leg angle (degrees)

B

80 100 120

A A

Th. Matheson: Chordotonal organ range fractionation

Fig. 2A-D. Tonic neurones with maximal firing at 00. Tonic firing was measured over a 3 s period immediately after each angle was set. Each value (mean 4- S.E.) was calculated by averaging the firing frequencies obtained from 3-8 tests, each with ramps of a different velocity preceding the static (hold) phases. Each test started with the leg angle set to 0 ~ and consisted of a series of 20 ~ ramp-and-hold steps progressing to 120 ~ then returning to 0 ~ Static angles were therefore approached from both more extended, and more flexed

4 0 . 0 ~

3 0 . 0 ~

2 0 . 0 ~

0 20 40 60 80 100 120

D 40.0

30'0~1

20.0

10.0

0.0- 20 40 60 80 100 120

Leg angle (degrees)

positions (direction of preceding movement indicated by arrows for each neurone). A Three units with a sharp reduction in firing between 80 and 100 ~ Circles, squares=P~ § triangles= P~247 B Three units with a gradual decrease in firing at extended angles. Circles= P~247 squares, triangles= P~ C Four units with ranges restricted to near 0 ~ Circles, diamonds=P~ +, squares=P~ triangles= P~ D Comparison of hysteresis in 2 units. Circles= P~ ?A § squares= P~247

A 80.0,

~ 6 0 . 0 v

~40.0 ID ~ .

~ 20.0-

0.0_ ~ 0

B

20 40 60 80 100 120 C 30.0.

25.0.

~ 20.0.

~ 15.0

& 10.0

5.0

0.01

Fig. 3A-D. Tonic neurones with maximal firing at Values were obtained as described in the caption to Fig. 2. A Two units with tonic maxima at 20 ~ and elevated firing over a wide range of leg angles. Circles= p2ov § ?A § squares=p2o?A+-. B Two units with tonic maxima at 40 ~ Circles=p4o?A §

25.0-

20.0-

15.0"

10.0.

5.0-

O.Ou o ~o io io 8'o 1~o i~o

,o.01 /

2~ 43 6? 8-0 10~ 1~-0 " ~ ~ 470 60 00 100 ' - - 1~0 Leg angle (degrees) Leg angle (degrees)

mid angles, squares = p4~ ?A- . C Three units with narrow tonic maxima at 60 ~ Circles = p6~ ?A- , squares= P6~ triangles = P6~ - D Three units with tonic maxima at 80 ~ or 100 ~ Squares = p s ~ circles = Pl~176 diamonds = ps~ ?A-

Th. M a t h e s o n : C h o r d o t o n a l o r g a n range f r ac t i ona t ion 513

A 40.0. B 40.0

30.0- 30.0

,.~ 10.0. 10.0

0.01 . . . . . . . . 0 2:0 40 60 80 100 120

C ~o.o D 6

~, 50 .0(

~ t0.0 5.0 40.0-

30,0 -

.~ t.0 20.0- 0.5

IO.C

O. t 0.r o zo 4'o 6'o 5o lOO 12o 2o 0

Leg angle (degrees)

Fig. 4 A - D . N e u r o n e s w i th tonic m a x i m a at 120 ~ Va lues were obtained as described in the caption to Fig. 2. A Four neurones illustrating response ranges and hysteresis. Circles, squares, trian- g les=P12~ diamonds=px2~ B Selection of 16 units chosen to illustrate a variety of response curves (extension and

Les angle (degrees)

flexion sequences have been averaged, and error bars omitted for clarity). C The same data as B on a log scale to emphasise the different response ranges. D Two tonic neurones with 2 peaks of elevated firing. Circles = P2~ squares = p6~ ?A-

the hysteresis crossed over. Three p60 neurones had nar- row peaks of tonic firing (Fig. 3C).

Two neurones (with maximum firing rates of less than 10 Hz, not illustrated) had low firing frequencies at 0-60 ~ increasing to reach a peak at 60 ~ and had main- tained higher firing between 60 ~ and 120*. Neurones with tonic maxima at angles less than 60 ~ never had main- tained high firing at more extended angles.

M a x i m u m at 80 or 100 ~ Only 4 neurones had maximum firing at 80 ~ (e.g., upper and lower curves in Fig. 3D), while 2 reached their maximum at 100 ~ (e.g., middle curve in Fig. 3D).

M a x i m u m at 120 ~ . The maximum firing frequencies at- tained by p120 neurones ranged between 3 and 40 Hz (Fig. 4A-C). The range of leg angles eliciting a ton- ic response also varied markedly between neurones (Fig. 4A-C). In order to highlight this, all 16 of the p120 responses recorded are presented together in Fig. 4B (values from the extension and flexion sides of the ramp-and-hold stimulus have been averaged for each neurone). In Fig. 4C the same data are plotted on a log scale to emphasise the different response ranges. There is a cont inuum of response width, rather than a sepa- ration into distinct response groupings.

Other responses. Four neurones had tonic maxima at relatively flexed angles, and a second, smaller, increase in firing frequency at 120 ~ (Fig. 4D). In all cases this secondary increase was large when compared to the size of the Standard Error bars, and therefore appears to be real. The phasic component of some of these neurones

also showed a secondary peak near 120 ~ (lower curve in Fig. 4D is from the same neurone as that represented in Fig. 6D).

Seven phasic neurones fired erratically during hold periods at some leg angles. This firing was caused by movements of tracheae or muscles in the distal femur distorting the FCO apodeme so these neurones were not classified as tonic.

Phas ic (ve loc i t y ) responses

Range fractionation may occur in either (or both) of 2 aspects of velocity sensitivity. These are: (1) the range of leg angles over which neurones respond at a given veloc- ity; and (2) the range of velocities over which neurones respond at a given angle. Most of the figures in this section are "families" of curves representing a neurone's response to different velocity movements in different parts of the leg's arc of movement. They therefore in- clude information about both aspects of the response. Each line represents a single repeat of one complete ramp-and-hold stimulus so error bars cannot be cal- culated.

It is important to remember that the phasic firing frequencies shown in my graphs are calculated from the number of spikes per ramp. Because all ramps were 20 ~ steps, higher velocity ramps were shorter in duration. This means that spike frequencies calculated for these shorter (faster) ramps will be derived from fewer spikes, and may exhibit greater variation. For example, a neu- rone could fire 10 spikes at 50 Hz in response to a 100~ movement (20 ~ in 0.2 s), but only 4 spikes at 200 Hz

514 Th. Matheson: Chordotonal organ range fractionation

A 250

~ 200,

tsol

tOO

"Z ~ 5O

0 - 2 0

B

2

20 '-40 4 0 - 8 0 60-80 8 0 - 1 0 0 100-120

200q

150

100

400 50 200

4O 0 20 0 - 2 0 2 0 - 4 0 4 0 - 6 0 8 0 - 8 0 8 0 - 1 0 0 1 0 0 - 1 2 0

C too D 300. " "

~ 60 200 1 ~ 4 0 0

looo ts0 1 40 400

"~ 500

. . . . 200 0 20 0 - 2 0 2 0 - 4 0 4 0 - 6 0 6 0 - 8 0 8 0 - 1 0 0 100-120 0 - 2 0 20~40 40~60 80~80 80~100 100~120

Leg m o v e m e n t ( d e g r e e s )

Fig. 5A-D. Extension sensitive neurones. All phasic neurones were tested with both flexion and extension movements. Extension sen- sitive neurones did not fire during flexion movements (or only fired occasional, erratic spikes), so for clarity, only extension movements are plotted in these figures. A Five extension sensitive neurones with maximal responses close to 0 ~ Each line represents the response of one neurone to 20 ~ ramps at a velocity of 400~ Similar response

Leg m o v e m e n t (degrees)

profiles (but with lower firing frequencies at each angle) occur at other tested velocities (not shown). 1 = V -, 2 = P4~ 3 = P12~ -, 4=p4~ , 5=V-. B One P4~ ?A - neurone stimulated by different velocities of tibial movement (numbers to the right of each line (~ C One V- unit with maximal firing at mid angles. Stimulus velocities are indicated to the right (~ D One p6~ unit sti- mulated at velocities of 20, 40 and 400~

during a 1000~ movement (20 ~ in 0.02 s). A single spike occurring during a 1000~ (0.02 s) ramp equates to an instantaneous firing frequency of 50 Hz.

I have pooled all the neurones possessing a velocity sensitive componen t irrespective of any additional re- sponsiveness to position or acceleration. I first examine the response ranges of extension sensitive and then flexion sensitive neurones, before going on to describe the relationship between tibial velocity and firing frequency.

E x t e n s i o n s e n s i t i v e neurones . Neurones which responded to the velocity of leg extension (V-) must be relaxation sensitive (tibial ex t ens ion=apodeme ligament relaxa- tion). The flexor strand was held at a fixed angle and could not have contributed to this response.

The majori ty of extension sensitive neurones (47/74) responded mos t strongly to extensions close to 0 ~ (e.g., the 0-20 ~ step of a ramp-and-hold stimulus: Fig. 5A, B). Higher velocities caused increased firing at all angles (Fig. 5B). At the highest velocity tested (1000~ the frequency of action potentials during the 0-20 ~ ramp varied between neurones within the range 100-350 Hz.

Six neurones were very sensitive and responded to movements at 10~ with firing frequencies of 20-30 Hz (slower movements were not tested). These neurones generally responded over a wide range of leg angles. The range of leg angles over which any one neurone respon- ded sometimes depended on the velocity of movement : in all such cases a higher velocity caused the neurone to respond over a wider range of angles. This phenomenon

occurred in the responses of virtually all phasic neurones which had restricted ranges at lower leg velocities (e.g., Figs. 5C, 6A, B).

Neurones with V - maxima at 0 ~ included pure veloc- ity units (V-), and neurones with mixed position, veloc- ity, and acceleration responses (e.g, P~ p s~ -). Tonic maxima (for the phasic-tonic units in this group) occurred anywhere in the range of leg angles (0-120~

Twenty-four extension sensitive neurones were most responsive to movements at mid-angles, and had reduced firing at both extremes of leg movement (Fig. 5C, D). The maximal firing frequencies of these neurones ranged f rom 50 to 300 Hz (for a 1000~ movement) , with most (20) firing at or below 150 Hz. The neurone represented in Fig. 5D fired at 300 Hz during a 400~ ramp, but was not tested at any higher velocities. At low velocities of movement (e.g., 20~ lowest curves in Fig. 5C) most of these neurones (20) did not fire at all at extreme leg angles. At higher velocities their firing frequencies in- creased, and the range of leg angles at which extensions elicited spikes widened (other curves in Fig. 5C). One neurone did not fire at 0 ~ even during movements at 1000~ (Fig. 5C). Four other neurones fired in response to movements at all leg angles (Fig. 5D). Neurones with mid-angle phasic maxima included both pure velocity units (V-) and units with mixed position, velocity, and acceleration responses (e.g., P t2~ p6~ ?A-) . All the phasic-tonic neurones in this group had their tonic firing maximum between 60 and 120 ~ .

Three extension-velocity sensitive neurones had max-

Th. Matheson: Chordotonal organ range fractionation 515

/ ~ -0 1000 IO0- A 140 " B ,20 / , 0

t oo ~ ~ 7,o; / J ' .4oo 6~ _

0-20 20-40 40-60 60-80 80-100 100-_120 0-20 20-40

;~176

~ 4O

40-60 60-80 60-100 100'-120

C 3 o o ~ . ~ D zso T

zso~ ~ ~ zoo~L_ ~

i . '~176 40100 5 0 ~

or = . , _ : o 0-2o 20-40 4o-80 6o-8o re-lee 100-120 o-2o

Leg movement (degrees)

Fig. 6A-D. Extension sensitive neurones. Each graph shows the responses of a single neurone to extension ramps of different veloci- ties (1 trial at each velocity). These neurones did not fire during flexion ramps (not shown). A, B Effect of different stimulus veloci- ties on firing of 2 extension sensitive neurones with maximal phasic responses near 120 ~ . The velocity of movement is indicated in

= .,0oo

~ 200

20

20-40 40-60 60--80 80-100 100--120 Leg moveltnent (degrees)

degrees per second on the right side of each graph. Both examples are P~2~ units. C, D Two extension sensitive neurones exhibiting increased phasic firing at both extremes of leg movement for all velocities tested (numbers to right (~ Both examples are Pa~ units

250 1

120-100 100-80 80-60 60-40 40-20 20--0 Leg movement (degrees)

Fig. 7. Seven flexion-velocity sensitive neurones with maximal phas- ic responses near 0 e. Each curve represents a different neurone. These lines show the firing frequencies elicited by flexion move- ments at 400~ Similar relative differences in firing are also seen if a different velocity (e.g., 1000~ is chosen (all neurones were tested at several velocities). These neurones did not fire during extension movements. 1,3-7= V § 2 = P~

imal phasic firing frequencies a t 120 ~ (Fig. 6A, B). F o r 1000~ m o v e m e n t s , these rates were 50 H z (V- ) , 100 H z (PX2~ : Fig. 6B) and 150 H z (P12~ : Fig. 6A), respec- tively.

Six neurones had unusua l responses : they fired mos t in response to extensions close to 0 ~ bu t also had a secondary peak o f sensitivity a t 100-120 ~ (Fig. 6C, D). These neurones had max ima l firing frequencies (for 1000~ tibial ex tens ions ) rang ing f r o m 150 to 300 Hz. All o f these neurones were o f the response type P W - ? A - , where x is in the range 40-120 ~ N o t all P W - ? A - neu- rones had this secondary peak. In some cases the tonic

c o m p o n e n t o f a neu rone ' s response paral le led the phasic c o m p o n e n t (cf. Fig. 4D (lower curve) and Fig. 6D: bo th responses f r o m a single neurone) .

Flexion sensitive neurones. Neurones which r e sponded to s tretch o f the a p o d e m e l igament (tibial flexion) (n = 67) r e sponded to higher velocities wi th higher firing frequen- cies. As for extension-veloci ty sensitive neurones , at higher velocities these neurones of ten fired in response to m o v e m e n t s in a greater range o f leg angles.

F o r t y - f o u r f lexion-velocity sensitive neurones respon- ded mos t s t rongly dur ing flexions close to 0 ~ (e.g., Fig. 7). Twen ty - fou r neurones o f this type fired m o r e spikes at each step successively closer to 0 ~ and hence had re- sponse curves which did no t f lat ten out. Twen t y o ther neurones did reach a p la teau over the last 2-3 r amps . O f the fo rmer type, 21 were pure ly phasic, and 3 were phasic- tonic (P~ Eight o f the lat ter (pla teau) type were pure ly phasic, while 12 had a P~247 response. This difference is statistically significant (Z 2 : P < 0.001, 1 df). The only PI~176247 neurone recorded fell into this la t ter group. M a x i m a l firing frequencies (20-0 ~ step, 1000~ ranged f r o m 100 to 350 Hz, wi th m o s t (28) in the range 100-200 Hz. Flexion-veloci ty sensitive neurones wi th m a x i m a at 0 ~ included some with the na r rowes t phas ic response ranges observed (e.g., curve 7 in Fig. 7). In ex t reme cases, firing could only be elicited in the last step (20-0 ~ ) at high velocities.

Sixteen flexion sensitive neurones had very wide re- sponse ranges, with cons tan t firing rates across the full range o f leg angles a t a given veloci ty (Fig. 8A, B). These neurones were general ly very sensitive: the m e a n firing

516 Th. Matheson: Chordotonal organ range fractionation

A 200 q

t50

~ tO0 I E

B 8oo~ r ~ 400 ,

t50

~ ~ : ~ 40 100

I 0- @ ~ 6.7 50

-- r ~ 400 ~ k ~ < l ' - ~ :. -- 200

~ 4 0

20

120-100 100-80 80-60 60-40 40-20 20-0 120-100 100-80 80-'60 60"40 40'-20 20-0

C zs0t / - -*- * * ~ . ~ . D 2oo,

2 0 0 4 / ~ ' 1000 ~ 150

o ~ r r 400 150

100 1 0 0 - -

"~, - - - - 5 0 ~., 50 ~ -- 40

0 120-100 100-00 80-60 60-40 40-20 20-0

Leg movement (degrees)

Fig. 8A-D. Flexion sensitive neurones. Each graph shows the re- sponses of a single neurone tested at several velocities (numbers to the right of each graph (~ The purely phasic neurones did not fire during extension movements. For phasic-tonic units, tonic firing was completely or largely inhibited during extension movements

: 400

-- 200 4O

120-100 100-80 80'-60 60-40 40-20 20'-0 Leg movement (degrees)

(not shown). A, B Units which responded with uniform firing across the full range of leg angles (A = V +, B = P~247 C One p2~ + unit with maximal firing at mid-angles. D One P~247 unit with maximal firing at 120 ~

rate during movements at 40~ was 37 Hz. The neurone in Fig. 8A fired at 50 Hz during movements at 6.7~ Slower movements were not tested methodically, but this neurone at least, responded to movements as slow as 0.50/s. This corresponds to a full tibial extension taking 4 rain. Phasic and phasic-tonic neurones were included in this group. Tonic max ima were in the range 0-80 ~ .

Only 5 neurones had max imum sensitivities to flex- ions at mid-angles. These responses were rather weak (Fig. 8C). The m ax i m um firing frequencies (1000*/s ramp) ranged between 150-250 Hz. Two neurones were not tested at 1000~ their maximal firing frequencies during 400~ ramps were 200 and 250 Hz, respectively. All these mid-posit ion neurones were phasic-tonic: the tonic maxima ranged f rom 0-40 ~ .

Three flexion sensitive neurones fired most strongly during ramps near 120 ~ In 2 cases the responses were rather weak, while the other neurone had a more marked peak, at least at the higher velocities (Fig. 8D). In all cases, the maximal firing frequency during a 1000~ ramp was 200 Hz. All these neurones had pxv* ?A § responses (where x = 0 - 2 0 ) . Other neurones with this classification did not have maximal phasic responses at 120 ~ .

Effect of tibial velocity on firin9 frequency durin9 ramps

All recorded neurones had higher firing frequencies at 1000~ than at the next slowest movement tested (usually 400~ but occasionally 500 or 600~ (i.e., firing fre- quency was never seen to completely saturate). Some neurones did however have different lower thresholds to

velocity. The most sensitive neurone responded to move- ments at least as slow as 0.5~ Another neurone respon- ded to movements at 1000~ but not those at 400~ or slower. Neurones which only responded to higher veloci- ties were usually those with narrow response ranges near either 0 ~ or 120". I did not a t tempt to determine velocity thresholds for the majori ty of neurones, but it is interest- ing to examine the above-threshold response to velocity. Twenty-three neurones chosen f rom 6 categories of velocity sensitive units (including wide and narrow range V § and V - types, and V - mid position types) were examined in detail (4 units from each type except wide range V § (3)). Firing rates were plotted against ramp velocity for the ramp which elicited the greatest firing (i.e., 20-0 ~ ramp for V § units which fired most as the leg approached 0~ For the wide range units, ramps in 3 ranges of leg angle were analysed: 0-20 ~ 60-80 ~ and 100-120 ~ .

When plotted on linear axes the majori ty of neurones exhibited some degree of saturation at higher velocities. There was some variation both between and within the chosen classes with respect to absolute firing frequency and the shape of the response curve, but this did not appear to be systematically related to any other features of response type. Regressions were calculated for in- dividual neurones f rom 3 classes: Fig. 9A: V +wide (20-0 ~ ramp), Fig. 9B: V -wid~ (0-20 ~ ramp), and Fig. 9C: V -mid (60-80 ~ ramp). Solid and open symbols and dashed lines simply distinguish between neurones within each part of the figure: they have no other significance. On average, r 2 values were higher for data plotted on log-log axes than on either linear or log-linear axes, indicating that the general form of the regression is:

Th. Ma theson : Chordo tona l organ range f ract ionat ion

A

100, v

o

.~ 10.

.h r~

l iD"

. IO

2 /

3o"

10 100 1000

Flexion velocity (*s- 1)

B

~vN 100

o

lO

r,,.

C

~" 100.

lO.

/ ,.-O

20- ~" / ~ / /

/ /

3 / / 4 . . . . . - I ~ - Z ' . . I ~ . . . . . . . . . 10 100 i000

2 ~ ~ ' ~ " ~ // 0

A3 O / 4

10 100 1000

Extension velocity (*s- 1)

Fig. 9A-C. Effect o f tibial velocity on firing frequency o f phasic neurones. Each graph plots the responses o f several neurones to different velocity movements over a given arc. See text for a de- scription o f the arcs chosen. A Three wide range flexion sensitive units (20-0 ~ ramp). 1 = V § :log Y= 0.3371ogX+ 1.44, r 2 = 0.987; 2 = p o v § 1.44, r2=0 .952; 3 = P ~ § 0.761ogX+ 0.25, r 2 = 0.988. B Four wide range extension sensitive units (0-20 ~ ramp). 1 = p12o V - :log Y= 0.501ogX+ 0.639, r 2 = 0.995; 2=p12oV-:logY=O.911ogX-0.406, r2=0 .997; 3 = V - : l o g Y = 1.551ogX-2.44, r2=0 .962; 4 = V - :logY= l.451ogX- 2.52, r z= 0.892. C Four mid-posi t ion extension sensitive neurones (60-80 ~ ramp). 1 = P 4 ~ : log Y= 0.491ogX+ 0.99, r 2 = 0.979; 2 = P~ 2~ : log Y= 0.251ogX+ 1.27, r 2 = 0.820; 3 = p s ~ - ?A - : log Y= 0.541ogX+0.71, r2=0 .971; 4=V-?A-:logY=O.921ogX-0.31, r 2 = 0.949

log(firing frequency) = a- log(ramp velocity)+ b

where a is the slope of the regression line, and b is the Y intercept. This equation can alternatively be expressed a s :

(firing frequency) = l0 b. (ramp velocity)"

517

The individual regression equations, regression coef- ficients, and details of each neurone's response type are given in the figure caption.

Two neurones in Fig. 9B appear to have steeper re- sponse curves than other neurones. It is not possible to statistically test the significance of this difference because each line is derived from only 1 complete ramp-and-hold stimulus (i.e., the errors cannot be calculated). Both these neurones were V- units, while the other 2 in Fig. 9B were P12~ units.

Discussion

Range fractionation

Range fractionation clearly does occur in the locust mtFCO, and there are some units which could act as labelled lines to signal leg position. Narrow-range units are present almost exclusively at the extremes of leg movement, but a few mid-range units also have restricted ranges (see Fig. 3C). The response ranges of most neu- rones (both phasic and tonic types) tend to be wide. Half (40) of the tonic neurones fired at all leg angles (0-120~ while three quarters (109) of the velocity sensitive neu- rones fired at all leg angles when stimulated at the fastest stimulus velocities. This contrasts to the crab myochor- dotonal organ, for example, where only 1 out of 48 flexion sensitive neurones responded over the full range of leg angles (Cohen 1963). Despite the absence of nar- row range "mid-position labeled line" units, the locust CNS should be able to extract accurate positional in- formation about all leg angles by evaluating the across- fibre pattern of firing supplied by the many neurones which fire differentially at different leg angles. Neurones with sensitivities restricted to near 0 ~ could act as labelled lines to indicate when the tibia is fully flexed. This in- formation should be important when the animal is preparing for a jump. In addition, Zill and Jepson-Innes (1990) have shown that locust leg reflexes differ when the leg is held at different initial angles. In particular, the usual resistance reflex to imposed movements of the tibia was replaced by a different reflex (termed the flexor reflex) when the leg was initially nearly fully flexed. In this mode of reflex action, tibial flexor motor neurones were excited by movements of the tibia in either direc- tion. It would be of considerable interest to investigate the connections of sensory afferents which have narrow response ranges near 0 ~ onto interneurones which trigger jumping, and the different classes of tibial motor neu- rones.

No neurones had narrow response ranges restricted to near 120 ~ however 4 phasic-tonic (P12~ neurones only fired tonically at 100-120 ~ (although they fired pha- sically over most of the range 0-120~ Some other posi- tion sensitive units have uniformly high firing frequencies between 0 and 100 ~ and a sudden decline in tonic firing at more extended angles. These neurones also can therefore precisely signal extreme extension.

It is interesting to note that some (8) phasic-tonic neurones had secondary peaks of firing at 120" (Figs. 4D,

518

6C). In 2 cases both the phasic and tonic components showed this trend, but in the rest only one component did so (2 tonic only; 4 phasic only). These neurones appear to be the first examples of U-shaped firing curves in a femoral chordotonal organ. Zill (1985a) stated that "phasic units were active in ranges of either extension or flexion but not both".

The lack of neurones with narrow-range extension sensitivity may reflect methodological constraints rather than a physiological difference: it is possible that such units do exist, but are activated by stretch of the flexor strand. In my experiments this ligament was held at a fixed angle of 60 ~ In addition, the metathoracic leg of Locusta can extend as far as 150 ~ so it is possible that there are units which begin to fire beyond 120 ~ to signal full extension. Zill (1985a) illustrated 1 such unit which only responded between 150 ~ and 170 ~ in Schistocerca. The relationship between tibial angle and FCO apodeme movement is approximately linear between 0 ~ and 120 ~ but is distorted at more extended angles (see curve for Sehistocerea in Field and Burrows (1982)). In particular, extensions between 0 ~ and 120 ~ push the apodeme to- wards the FCO, while further extensions (beyond 120 ~ tend to pull the apodeme. Thus, both a flexion from 120 ~ to 100 ~ and an extension from 120 ~ to 140 ~ exert a pull of similar magnitude on the apodeme. Some complex mechanism must exist if FCO neurones are to make a distinction between these movements. Field (1991) has shown that the main ligament to the apodeme is multi- stranded, and that different strands are under differential tension. It is possible that this mechanism is responsible for the observation that groups of neurones (connected to different bundles of strands in the ligament) respond differently to the same movement of the apodeme. Clear- ly, further research is required in this area.

The firing frequencies of many tonic neurones exceed 40 Hz, yet the whole-nerve recordings of locust mtFCO tonic activity at different leg angles presented in Zill (1985a) and Usherwood et al. (1968) have surprisingly low values (maximum < 160 Hz). Theophilidis (1986) has higher values for Decticus (minimum approximately 230 Hz, maximum 550 Hz). These low rates suggest that only the largest tonic spikes are being detected in ex- tracellular records. This introduces the possibility of con- siderable bias in the shape of the extracellularly recorded response curve because neurones with different sized axons are likely to have different responses. It is interest- ing therefore to note that the shapes of the response curves presented by Theophilidis are different to those presented by Zill (1985a) and Usherwood et al. (1968). Considerable caution should be used when interpreting such extracellular recordings. In particular, these ex- tracellular recordings certainly hide the responses of neurones which have mid-position tonic maxima. Al- though the extracellular recordings fail to represent small spikes, these neurones must still make significant contri- butions to information processing in the CNS.

Th. Matheson: Chordotonal organ range fractionation

Hysteresis in tonic responses

Hysteresis in the responses of mtFCO neurones has been discussed by Matheson (1990a) and Zill and Jepson- Innes (1988). Mill and Lowe (1972) presented a theoreti- cal model to explain how the CNS could extract accurate positional information from an otherwise ambiguous position response, but this does not seem to have been investigated at the neuronal level. Examples in the present paper show that hysteresis varies greatly between neurones: even between those with otherwise similar responses. Zill (1985a) and Zill and Jepson-Innes (1988) discussed hysteresis in terms of movements away from, and back to the central leg angle because all the tonic units in those studies had maximal firing at the extremes of leg movement. It is therefore interesting to examine the hysteresis of neurones with mid-position maxima reported in the present paper.

Most tonic neurones (77/91) had a velocity sensitive component (the remaining 14 responded to position and acceleration). At any given leg angle the tonic firing of most flexion-velocity sensitive neurones was higher if the set angle had been approached by a flexion than if the set angle had been approached by an extension (e.g., lower curve in Fig. 2D). The converse was generally true for extension-velocity sensitive neurones (i.e., tonic firing was higher following extensions). This could suggest that the hysteresis is simply a long-lasting aftereffect of the phasic response. However, 4 flexion sensitive phasic- tonic units had higher tonic firing following extension movements which transiently inhibited the tonic dis- charge than following flexion movements which caused a phasic burst of spikes (e.g., middle curve of Fig. 3D). Position-and-acceleration sensitive neurones exhibited similar hysteresis to PW units, yet their phasic response was limited to 1-2 spikes per ramp. Eight neurones with mid-angle tonic maxima had response curves which clearly crossed over (standard error bars did not overlap: e.g., Fig. 3A, B). In these cases the hysteresis meant that the tonic firing frequency at any set angle was greater if that angle had been approached from a position further from the crossover angle. Hofmann et al. (1985) recorded phasic-tonic neurones in the stick insect which responded tonically as the leg was set to progressively more exten- ded angles, but responded phasically to flexions. This information, and the differences in hysteresis mentioned above suggest that in many phasic-tonic neurones the tonic and phasic responses are largely uncoupled from each other. Mill and Lowe (1972) have also noted this phenomenon in the PD proprioceptor of decapods. Be- cause the mechanism of sensory transduction in chor- dotonal scolopidia is still unclear it is difficult to spec- ulate about the possible basis for such apparently con- tradictory responses. Some insight may be gained by recording from, and staining individual receptors before fixing and sectioning the FCO for TEM examination. In this way it should be possible to clarify structural dif- ferences in the dendritic endings, scolopidia, and attach- ment cells of the various response types. As noted earlier, Field (1991) has begun this task by examining the fine structure of the FCO's main ligament. Further informa-

Th. Matheson: Chord,tonal organ range fractionation 519

tion about the mechanics of tension development in the various strands described by Field (1991), and the degree of interaction between strands is now required. The strikingly non-linear responses of many tonic neurones to linear movements of the apodeme (especially the units with secondary maxima at 120 ~ (Fig. 4D)) highlight the need to clarify the interactions between mechanical and electrophysiological factors controlling FCO neurone's responses. Zill (1985a) made a brief study of the effect of injected current on the responses of various locust mtFCO neurones, but did not thoroughly characterise neurones before testing their response to current. In addition, he gave very little information about the details of each elicited response. A more thorough study of the relationships between injected current and firing fre- quency in characterised neurones would complement a morphological study, and would help explain the basis for range fractionation and hysteresis in these neurones.

Effect o f velocity on firin9 f requency

Although velocity thresholds were not systematically ex- amined in the present study it is possible to make some generalisations about the responses of velocity-sensitive neurones. The range of angles and velocities tested in this work are similar to those used by the animal during walking: Burns (1973) reported that the metathoracic leg moved in the arc 40-90* at stepping frequencies of up to 9 Hz during straight walking. Assuming a sinusoidal protraction/retraction cycle, it is possible to determine that the maximal angular velocity during such move- ments will be approximately 1400~ This is likely to be an upper limit because velocity is known to be controlled during both the stance and swing phase (Dean and Cruse 1986). For example, in the stick insect at least, a graph of leg position versus time tends to resemble a rounded saw-tooth pattern with constant velocity during both phases of the step. It is clear though, that the locust leg can be moved at higher velocities during kicking for example. At these higher velocities, proprioceptive feed- back is likely to have a reduced influence because reflex latency will significantly interfere. In my work, velocities between 20 and 1000~ were regularly used to test phasic responses of neurones in the range 0-120" of leg angle. Some neurones were tested at lower velocities (0.5-20~

The majority of velocity-sensitive neurones (115/141) responded to all velocities in the range 40-1000*/s, al- though different neurones could have different firing fre- quencies at a given velocity (Fig. 7). Some neurones were markedly more sensitive to low velocities than most (e.g., Fig. 8A), indicating that there is at least some range fractionation of velocity sensitivity. No neurones sat- urated completely at higher velocities, and no neurones had maximal firing at a submaximal velocity (i.e., higher velocities of apodeme movement always led to higher firing frequencies). Note, however, that the 2 upper velo- cities tested were usually 400 and 1000*/s: some neurones could have reached their peak firing frequencies at an intermediate value (e.g., 800 Hz). Zill (1985a) illustrated the responses of 2 phasic neurones to increasing frequen-

cies of sinusoidal movement (from 0-9 Hz) over a 12 ~ arc. Both these neurones were almost saturated at 5 Hz (corresponding to m a x i m a l velocities of 188~ How- ever, in his text Zill stated that "at frequencies greater than 50-75 Hz these [same] units show range saturation" (my italics). If this is accurate, then the correct value for response saturation velocity would be 1880~ corre- sponding better with my information. Zill (1985a) also stated that below the saturation level, firing frequency increased linearly with increased frequency of stimula- tion. This does not agree with my results, nor with those of Hofmann et al. (1985): in these cases a log-log rela- tionship was chosen. Mill and Lowe (1972) illustrated velocity sensitive neurones from the PD organ of Cancer, one of which had a linear increase in firing with increased velocity, and another which responded linearly to the log of velocity. Although Mill and Lowe gave a correlation coefficient (r = 0.966) for their log-linear curve, they did not give corresponding values for the same data plotted on other coordinate systems, so it is not possible to confirm that this is the best possible fit. Hofmann et al. (1985) did not fit regressions to their data. In my study, average r 2 values (i.e., mean values from the regressions for many neurones) were always higher for data plotted on log-log scales than for the same data plotted on log-linear or linear axes. However, individual neurones sometimes had higher r z values when they were plotted on linear or log-linear axes. Regression coefficients for all 3 possibilities were always high (worst fit: r 2= 0.77). It is possible that different neurones have different stim- ulus-response relationships. Because I did not consistent- ly test velocities at both extremes of sensitivity I am unable to compare the upper and lower thresholds of different neurones. This restricts any interpretation of the variation of shape or slope of the regression lines because each neurone may have been tested in a different part of its sensitivity range.

Cohen (1963) illustrated another type of response for a flexion-sensitive neurone from the my,chord, tonal organ of Cancer: the slope of its response/velocity curve increased at higher velocities (the opposite of saturation). It is not clear, however, if these velocities (0-16~ re- present the full physiological range used by the animal, so this observation cannot be clearly compared to other studies.

Non-linearities in the stimulus-response behaviour of mechanoreceptive neurones may be introduced at a vari- ety of steps in the pathway from mechanical input to firing frequency (Mann and Chapman 1975). These in- clude viscoelastic properties of the surrounding tissue and/or cuticle, transients in the conductances of strain- sensitive channels in the transducer membrane, and ac- commodation of the axonal spike generating zone. Rydqvist et al. (1990) have shown that the viscoelastic properties of the crayfish slowly adapting stretch recep- tor muscle play an important role in shaping the respon- ses of the receptor neurones. Because the locust mtFCO is linked to the tibia by a connective tissue strand, it is likely that viscoelastic properties of the strand are very important in determining velocity sensitivity at least. It would be of great interest to determine the viscoelastic

520 Th. Matheson: Chordotonal organ range fractionation

propert ies of this s t rand, and compare them to the propert ies of the comparab le s t rands which l ink the pro thorac ic and mesothorac ic F C O s to their respective t ib ia : in these legs, the s t rands are much longer than in the h ind leg. Direct m e a s u r e m e n t of the dynamic forces t ransmi t ted by the apodeme and l igament to the neu- rona l scolopidia m a y provide useful i n fo rma t ion abou t the c o m p o n e n t mechan i sms under ly ing range fract iona- t ion and hysteresis in this complex sense organ.

Acknowledgements. I thank Dr Larry Field for his support, enthu- siastic encouragement, and constructive comments throughout this study, and Franz Ditz for designing and building the ramp genera- tor.

References

Aidley DJ (1978) The physiology of excitable cells (2nd Ed.). Cam- bridge University Press, Cambridge London New York, 530 pp

Burns MD (1973) The control of walking in Orthoptera. I. Leg movements in normal walking. J Exp Biol 58:45-58

Burns MD (1974) Structure and physiology of the locust femoral chordotonal organ. J Insect Physiol 20:1319-1339

Burrows M (1987) Parallel processing of proprioceptive signals by spiking local interneurons and motor neurons in the locust. J Neurosci 7(4): 1064-1080

Burrows M (1988) Responses of spiking local interneurones in the locust to proprioceptive signals from the femoral chordotonal organ. J Comp Physiol A 164:207-217

Burrows M, Laurent GJ, Field LH (1988) Proprioceptive inputs to nonspiking local interneurons contribute to local reflexes of a locust hindleg. J Neurosci 8:3085-3093

Cohen MJ (1960) The response patterns of single receptors in the crustacean statocyst. Proc R Soc Lond B152:30~9

Cohen MJ (1963) The crustacean myochordotonal organ as a pro- prioceptive system. Comp Biochem Physiol 8:223-243

Cohen M J, Hagiwara S, Zotterman Y (1955) The response spec- trum of taste fibres in the cat: a single fibre analysis. Acta Physiol Scand 33:316-332

Dean J, Cruse H (1986) Evidence for the control of velocity as well as position in leg protraction and retraction by the stick insect. Exp Brain Res Ser 15:263-274

Dodt E, Zotterman Y (1952) Mode of action of warm receptors. Acta Physiol Scand 26:345-357

Evoy WH, Cohen MJ (1969) Sensory and motor interaction in the locomotor reflexes of crabs. J Exp Biol 51:151-169

Field LH (1991) Mechanism for range fractionation in chordotonal organs of Locusta mioratoria (L.) and Valanoa sp. (Orthop- tera :Acrididae). Int J Insect Morphol Embryol 20: 25-39

Field LH, Burrows M (1982) Reflex effects of the femoral chor- dotonal organ upon leg motor neurones of the locust. J Exp Biol 101:265-285

Hofmann T, Koch UT, B/issler U (1985) Physiology of the femoral chordotonal organ in the stick insect Cuniculina impiora. J Exp Biol 114: 207-223

Kalmring K, Lewis B, Eichendorf A (1978) The physiological characteristics of the primary sensory neurons of the complex tibial organ of Decticus verrucivorus L. (Orthoptera, Tet- tigonioidae). J Comp Physiol 127:109-121

Kiihne R (1982) Neurophysiology of the vibration sense in locusts and bushcrickets: response characteristics of single receptor units. J Insect Physiol 28:155-163

Laurent G (1988) Local circuits underlying excitation and inhibi- tion of intersegrnental interneurones in the locust. J Comp Physiol A 162:145-157

Lowenstein O, Roberts TDM (1950) The equilibrium function of the otolith organs of the thomback ray (Raja r J Physiol 110:392-415

Mann DW, Chapman KM (1975) Component mechanisms of sen- sitivity and adaption in an insect mechanoreceptor. Brain Res 97: 331-336

Matheson T (1990a) Responses and locations of neurones in the locust metathoracic femoral chordotonal organ. J Comp Physiol A166:915-927

Matheson T (1990b) An experimental and descriptive analysis of the responses and peripheral and central organisation of the locust metathoracic femoral chordotonal organ. Ph.D thesis, University of Canterbury, New Zealand

Matheson T (1992) Morphology of the central projections of phys- iologically charaeterised neurones from the locust metathoracic femoral chordotonal organ. J Comp Physiol A170:101-120

Matheson T, Ditz F (1991) A simple computer-controlled analogue ramp generator for producing multiple ramp and hold stimuli. J Neurosci Methods 39:45-52

Mill PJ, Lowe DA (1972) An analysis of the types of sensory unit present in the PD proprioeeptor of decapod crustaceans. J Exp Biol 56: 509-525

Rheinlaender J (1975) Transmission of acoustic information at three neuronal levels in the auditory system of Decticus verru- r (Tettigoniidae, Orthoptera). J Comp Physiol 97:1-53

Rydqvist B, Swerup C, L/innergren J (1990) Viscoelastic properties of the slowly adapting stretch receptor muscle of the crayfish. Acta Physiol Stand 139: 519-527

Shimozawa T, Kanou M (1984a) The aerodynamics and sensory physiology of range fractionation in the cercal filiform sensilla of the cricket Gryllus bimaculatus. J Comp Physiol A155:495-505

Shimozawa T, Kanou M (1984b) Varieties of filiform hairs: range fractionation by sensory afferents and cereal interneurons of a cricket. J Comp Physiol A155:485~93

Siegler MVS (1984) Local intemeurones and local interactions in arthropods. J Exp Biol 112:253-281

Theophilidis G (1986) The femoral chordotonal organs of Decticus albifrons (Orthoptera: Tettigoniidae) - II. Function. Comp Biochem Physiol 84A: 537-543

Usherwood PNR, Runion HI, Campbell JI (1968) Structure and physiology of a chordotonal organ in the locust leg. J Exp Biol 116:435-461

Wilson JA (1981) Unique, identifiable local non spiking inter- neurones in the locust mesothoracic ganglion. J Neurobiol 12 : 353-366

Young D (1970) The structure and function of a connective chor- dotonal organ in the cockroach leg. Phil Trans R Soc Lond B 256: 401--426

Young D (1975) Chordotonal organs associated with sound producing apparatus of cicadas (Insecta: Homoptera). Z Mor- phol Tiere 81 : 111-135

Zill SN (1985a) Plasticity and proprioception in insects. I. Re- sponses and cellular properties of individual receptors of the locust metathoracic femoral chordotonal organ. J Exp Biol 116:435-461

Zill SN (1985b) Plasticity and proprioception in insects. II. Modes of reflex action of the locust metathoracie femoral chordotonal organ. J Exp Biol 116:463-480

Zill SN, Jepson-Innes K (1988) Evolutionary adaptation of a reflex system: sensory hysteresis counters muscle "catch" tension. J Comp Physiol A 164: 43-48

Zill SN, Jepson-Innes K (1990) Functions of a proprioceptive sense organ in freely moving insects: characteristics of reflexes elicited by stimulation of the locust metathoracic femoral chordotonal organ. Brain Res 523 : 211-218