The effects of temperature on peripheral neuronal function in eurythermal and stenothermal crustaceans

The ambient temperature range of the environment in which a poikilotherm lives governs almost completely the temperatures at which the nervous system must function. Temperate marine invertebrates experience large daily and seasonal fluctuations of temperature [maximum daily variation: 12°C,annual range: 2 to 23°C, rocky shore, NE England(Willows, 1987)]. In contrast,Antarctic marine invertebrates experience permanently low temperatures (annual range –1.9 to +1.7°C, at 15 m depth)(Barnes et al., in press). The evolutionary adaptations associated with life in such different thermal environments over the past 20 million years(Kennett, 1977) have been comprehensively studied for species of fish but, with few exceptions (e.g. Macdonald, 1981), have been neglected for marine invertebrates. We have compared the effects of temperature on the propagation of action potentials in the leg nerves of two temperate and two Antarctic crustaceans, and we have examined the responses of sensory neurones in one of the Antarctic species. Our choice of species permits comparisons of similarly sized peracarids from the two environments(Ligia oceanica, Paraceradocus gibber) and permits contrasts with a larger peracarid Glyptonotus antarcticus, and a similarly large decapod Carcinus maenas.

Propagation of action potentials in the peripheral nervous system represents a potential rate-limiting step in the relay of motor and sensory signals to and from the central nervous system, and can therefore constrain rates of behaviour. For example, the large Antarctic pycnogonid Colossendeis robusta has legs often in excess of 15 cm long, and neuronal conduction velocity is only 0.33 m s^–1(Macdonald, 1981). This leads to an exceptionally long conduction delay of 450 ms, which must contribute to the animal's extremely sluggish behaviour.

A positive relationship between temperature and neuronal conduction velocity (Helmholtz, 1850) has been characterised in sensory transduction of the tactile spine afferents in the legs of the cockroach P. americana(Chapman and Pankhurst, 1967),sensilla involved in the detection of orientation in the crab Carcinus maenas (Fraser, 1990) and a slit sensillum of the Central American hunting spider Cupiennius salei (Höger and French,1999). The effects of temperature on the conduction velocities of individual motor axons involved in jumping and kicking(Burrows, 1989) and flight(Xu and Robertson, 1994) have been assessed for two species of locust (Schistocerca gregaria and Locusta migratoria, respectively).

Only a few assessments of the effect of temperature on neuronal conduction velocity have been applied in the comparison of temperate and polar species,primarily in fish (Macdonald,1981; Harper et al.,1990; Moran and Melani,2001). If the polar species are physiologically compensated(Prosser, 1958), then the conduction velocities of different species measured at their respective environmental temperatures (e.g. –1.8°C for Antarctic, +5°C for Arctic, and +20°C for Mediterranean fish) should be the same. Perfect compensation is not achieved in either Antarctic fish [rates are 2–3 times slower than temperate fish measured at 16°C(Macdonald, 1981)] or Arctic fish [rates are 2–2.5 times slower than Mediterranean fish at 12°C(Melani and Moran, 1988)].

We show how temperature affects neuronal conduction velocity in four marine crustaceans: the amphipod Paraceradocus gibber and isopod Glyptonotus antarcticus from Antarctica; and the decapod Carcinus maenas and isopod Ligia oceanica from the British coast. For the first time in an Antarctic invertebrate we describe the responses of mechanosensory neurones, their conduction velocities and the thermal dependence of receptor responsiveness. The diameter of axons in the leg nerves of L. oceanica and G. antarcticus is related to conduction velocity.

Carcinus maenas Linnaeus (35–50 mm in length) were obtained from Wells-next-the-Sea (Norfolk, UK) using baited traps. They were housed in 40 cm×80 cm×16 cm aquaria containing stones and artificial seawater (in mmol l^–1: Cl^–, 541.6;Na⁺, 465.2; Mg²⁺, 54.3;SO₄^–, 28.1; Ca²⁺, 11.9; K⁺,10.5; CO₂^–, 4.4; B³⁺, 0.5;Br^–, 0.03; Kent Sea Salt, Kent Marine Inc., Acworth, USA),which was aerated and filtered. They were fed on dead fish and maintained at+4°C.

Ligia oceanica Linnaeus (22–28 mm in length) were obtained from a tidal wall at Wells-next-the-Sea. They were housed individually in containers 8.5 cm diameter × 6.5 cm high lined with seawater-moistened tissue paper. They were fed on vegetable peelings and were maintained at+4°C.

Glyptonotus antarcticus Eights (60–110 mm in length) and Paraceradocus gibber Andres (33–51 mm in length) were taken from baited traps at Rothera (67°34′S, 68°08′W) and Palmer(64°46′S, 64°03′W) stations (Antarctica) and transported by ship to the UK where they were maintained in a recirculating natural seawater aquarium at 0 to +1°C with a 12 h:12 h light:dark regime. They were fed on dead fish and krill. Salinity was maintained at 33–36 psu(practical salinity units; the conductivity ratio relative to a standard KCl solution, approximately equivalent to 33-36‰.

Experiments were performed on isolated legs that were cut from the rest of the animal at the coxa with a sharp pair of dissecting scissors. Following removal of a leg, a slow flow of haemolymph from the wound on the animal ceased within 30 s.

A leg from C. maenas was prepared with the ventral surface facing upwards. The merus was dissected open and the carpus flexor muscle removed to reveal the leg nerve. For L. oceanica and G. antarcticus,legs were mounted with the ventral surface facing upwards. The basis and ischium were dissected open, and the merus extensor, ischium extensor and ischium flexor muscles removed to expose the leg nerve. For P. gibber, a leg was fixed with the posterior surface facing upwards. The basis was dissected open to reveal the merus extensor, ischium extensor and flexor muscles, which were removed to reveal the leg nerve.

Experiments on the Antarctic species, G. antarcticus and P. gibber, were conducted in a controlled-temperature room, maintained at 5°C. Experiments on the temperate species, C. maenas and L. oceanica, were conducted in a laboratory at 20°C.

Two pairs of 50 μm (for C. maenas and G. antarcticus)or 25 μm (for L. oceanica and P. gibber) silver bipolar hook electrodes were placed as far apart as possible around the leg nerve which was lifted clear of the saline, and the distance between the electrodes was measured. A 75:25% Vaseline: paraffin mixture was applied to the nerve around each pair of hook electrodes with a drawn out plastic syringe. The preparation dish contained sufficient crustacean saline(Yamagishi and Ebara, 1985) to cover the preparation. The preparation was either cooled then warmed, or warmed then cooled, at approximately 0.25°C min^–1 using a thermocirculator (Grant LTD 20G, Cambridge, UK) circulating a 50:50 mixture of ethylene glycol and water around a jacketed Perspex preparation dish (3 ml volume) lined with Sylgard (Dow Corning, Midland, USA).

A train of ten square pulses of 1 ms duration and 130% amplitude of the voltage threshold for eliciting action potentials, was delivered at 20 Hz once every 30 s (Master-8 stimulator, AMPI, Jerusalem, Israel) from the proximal pair of electrodes. Measurements of temperature were made using a thermocouple probe placed adjacent to the preparation.

Signals were amplified using a custom-built extracellular amplifier with a 50 Hz notch filter, digitised at 1 kHz (CED 1401 or Micro 1401, Cambridge Electronic Design, Cambridge, UK), and stored on a computer for later analysis.

Conduction velocity was calculated by dividing the distance between stimulating and recording electrodes by the latency of the first peak of the compound action potential (Fig. 1).

Neuronal conduction velocity was interpolated for temperatures of +2 and+10°C using a linear relationship fitted to data measured at temperatures between approximately –2 and +20°C. Values were calculated at 2°C because this temperature represents the overlap in thermal tolerances of the temperate and Antarctic species. Values were calculated at 10°C– the minimum experimental temperature used in many of the previous assessments – so that the neuronal conduction velocities of C. maenas, L. oceanica, G. antarcticus and P. gibber, could be compared directly with published data from other species. In the case of C. maenas, neuronal conduction velocity and its thermal dependence were calculated at temperatures between 0 and 15°C, where the relationship between temperature and neuronal conduction velocity was linear.

In two G. antarcticus, we recorded the activity of leg tactile sensory neurones using paired hook electrodes as described above, with both channels set to record. One electrode was placed in the ischium, the other in the basis. Individual moveable spines on the merus or carpus were adducted by a standard brief (approximately 500 ms) deflection, delivered by hand. The number of action potentials elicited by the stimulus, and the conduction velocity in the axon, were recorded at temperatures ranging from–1.8°C to +25°C.

Short (1–5 mm) lengths of leg nerve to be used for transmission electron microscopy (TEM) were taken from the segments used in the electrophysiological recordings. Tissue embedding for TEM followed a protocol modified from that used by Postel et al.(Postel et al., 2000). The section of nerve removed from the leg was immediately washed in 0.1 mol^–1 phosphate buffer (9 mmol l^–1NaH₂PO₄ and 44 mmol l^–1Na₂HPO₄ in distilled water) at pH 7.0 and 4°C,before being fixed in 6% glutaraldehyde in phosphate buffer for 14 h at 4°C. Fixed nerves were washed in phosphate buffer and then post-fixed in 1% osmium tetroxide for 1 h at 4°C, before being dehydrated in an alcohol series, and then treated twice for 15 min in propylene oxide. Samples were then embedded overnight in Araldite. Semi-thin (10 μm) and ultra-thin (70 nm) sections were cut with an ultramicrotome (Reichert OmU2, Vienna, Austria)using glass and diamond knives, respectively. For light microscopic investigation, the semi-thin sections were stained with Toluidine Blue (0.8%Toluidine Blue, 0.8% borax, 0.2% pyronin in distilled water) for 30 s at 70°C. Ultra-thin sections placed on 100 Mesh grids were contrasted with uranyl acetate for 40 min, and lead citrate for 10 min, then viewed in a transmission electron microscope (Phillips, EM300).

Electron micrographs were taken at a magnification of approximately×3300, and calibrated against a standard grid. Negatives were developed,then scanned at 1200 dpi, and the resulting images assembled to form a montage of each nerve (Canvas 7SE, Deneba).

Axon diameters were measured using the method described by Graham(Graham, 2003).

Electrical stimulation of the leg nerve of C. maenas, L. oceanica, G. antarcticus and P. gibber yielded compound action potentials recorded at a second pair of hook electrodes placed distally on the nerve. The latency between electrical stimulation and the first peak of the action potential decreased with increasing temperature(Fig. 1A). At any given temperature there was a distinct voltage threshold for the excitation of axons in the leg nerve. In G. antarcticus, for example, the threshold was 1.25 V (Fig. 1B). Increasing the stimulation voltage above this threshold value caused a slight change in the compound action potential that led to a small increase in the maximum conduction velocity (Fig. 1B). In all experiments we stimulated at 130% of the threshold for each animal to ensure that we recorded the maximum conduction velocity (e.g. Fig. 1B).

Changing the temperature did not affect the threshold for stimulation of axons in the leg nerve in any species. In G. antarcticus, for example, an increase in temperature from –1.8°C to +16°C had no significant effect on threshold (Fig. 1C), although the conduction velocity was faster at higher temperatures (Fig. 1B). The upper temperature at which electrical stimulation failed to elicit a compound action potential, `the upper thermal block', was similar in C. maenas,L. oceanica, G. antarcticus and P. gibber,occurring at +24.5, 19.5, 21.5 and 19.5°C, respectively(Fig. 2E).

Conduction velocity increased with temperature in all four species(Fig. 2). Action potentials were elicited by electrical stimulation over maximum ranges of –2.5 to+24.5°C in C. maenas (Fig. 2A), –3.5 to +21.5°C in G. antarcticus(Fig. 2B), –1.5 to+19.5°C in L. oceanica (Fig. 2C), and –2.5 to +19.5°C in P. gibber(Fig. 2D). Neuronal conduction velocity increased linearly with temperature in L. oceanica, G. antarcticus and P. gibber(Fig 2, Fig 3A; Table 1). In three preparations of C. maenas, neuronal conduction velocity increased exponentially on warming. In the other three preparations, neuronal conduction increased sigmoidally with temperature (Fig. 2A,E). The difference between these two groups of recordings from C. maenas only became apparent at temperatures above 15°C (groups A and B in Fig. 2E).

The leg nerve contains the axons of both sensory and motor neurones. In G. antarcticus only we recorded the conduction velocity of individual sensory neurones activated by touching specific articulated spines on the leg. The conduction velocities and temperature dependence of conduction velocity were similar to the values obtained by electrical stimulation of the whole nerve (Fig 2F, Fig 3B; Table 1). Touching a single spine could elicit action potentials of more than one size, indicating that the spines are multiply innervated. The conduction velocities of the smaller amplitude action potentials were slower than those of larger action potentials, but their temperature dependence was similar(Fig. 3).

There was a greater effect of temperature on neuronal conduction velocity in C. maenas (0.162 m s^–1 deg.^–1, N=6) than L. oceanica (0.067 m s^–1deg.^–1, N=7), G. antarcticus(0.058 m s^–1 deg.^–1, N=7) or P. gibber (0.039 m s^–1 deg.^–1, N=6) (Kruskal–Wallis, χ²=13.36, P<0.01) (Fig. 3A). There was no significant difference between the thermal dependence of neuronal conduction velocity in the three peracarid crustaceans, L. oceanica, G. antarcticus and P. gibber (Kruskal–Wallis,χ ²=2.86, P=NS).

Neuronal conduction velocity at 2°C was significantly greater in C. maenas than in the other species (Fig. 3B; Table 2;Kruskal–Wallis, χ²=19.55, P<0.01). It was significantly lower in L. oceanica than in the other peracarid species G. antarcticus and P. gibber (Kruskal–Wallis,performed on 2°C data, χ²=12.78, P<0.01).

In G. antarcticus, moving individual articulated spines on the leg in a repeatable way led to bursts of action potentials recorded in the leg nerve (Fig. 4A). The number of action potentials elicited by each touch depended on the temperature. At low temperatures relatively few action potentials were elicited, but as temperatures rose to 5–10°C the number increased to a maximum before declining again to fail at 20–22°C(Fig. 4B).

The number and diameter of axons within the main nerve at the recording site in the basis of the leg were assessed for L. oceanica and G. antarcticus (N=2 for each species). In both species, there were a large number of small diameter (<1 μm) axons and fewer large (>5μm) diameter axons (e.g. G. antarcticus; Fig. 5A,B). The median axon diameters from nerves of both specimens of L. oceanica (0.43 μm and 0.52 μm; Fig. 6) were significantly different (Mann–Whitney U test, Z=–3.89, P<0.001), and the distribution of axon diameters was different (Kolmogorov–Smirnov, Z=6.03, P<0.001). Similarly, the median axon diameters from nerves of both specimens of G. antarcticus (0.89 μm and 0.28 μm; Fig. 6B) were significantly different (Mann–Whitney U test, Z=–61.09, P<0.001), and the distribution of axon diameters was different(Kolmogorov–Smirnov, Z=31.54, P<0.001).

The number of axons was also different between animals of the same species. For L. oceanica, the two values were 929 and 1570 axons. In G. antarcticus, the values were 3638 and 10313 axons.

When the data from pairs of animals were pooled, the median axon diameters of L. oceanica and G. antarcticus were significantly different (Mann–Whitney U test, Z=–15.87, P<0.001), and the distribution of axon diameters was different(Kolmogorov-Smirnov, Z=10.77, P<0.001)(Fig. 6).

The number and frequency distribution of large (>5 μm) diameter axons was similar for the two species; there were 20 axons between 10 and 30 μm in two specimens of L. oceanica, and 12 in two specimens of G. antarcticus (Fig. 6A,Binsets).

Each G. antarcticus possessed a small nerve bundle, as part of the main leg nerve, containing fewer than 10 large (>5 μm) axons, surrounded by a wrapping, typically 0.5–3 μm in thickness(Fig. 7). The sheath surrounding the entire nerve bundle was loosely bound; the membrane was convoluted, and in places, gaps were present between adjacent layers. Similarly wrapped axons were not observed in the leg nerve tissue of L. oceanica (N=2).

The conduction velocity of axons in the leg nerve differed between species in the order C. maenas > G. antarcticus > P. gibber > L. oceanica (Fig. 2E). For the temperate species, the difference between conduction velocities at+2°C and +23°C(i.e. spanning their normal environmental range) was 3-fold inC. maenas, and 3.9-fold in L. oceanica. For both of the Antarctic species, there was only a 1.2-fold difference between conduction velocities at the extremes of their environmental range (–1.8°C vs +1.5°C; see Table 2).

The lower neuronal conduction velocity of L. oceanica thanC. maenas coincides with a large difference in leg size. Leg lengths are an order of magnitude greater in C. maenas than in L. oceanica(approximately 10 cm for a 6 cm long C. maenas, approximately 1 cm for a 2.3 cm long L. oceanica). At 10°C the conduction velocity of axons in L. oceanica was 1.24 m s^–1 so it takes 7 ms for an action potential to travel 1 cm along the leg. If the conduction velocity were exactly the same in C. maenas, it would take an action potential 70 ms to travel down its leg. The more rapid conduction velocity in C. maenas(3.2 m s^–1 at 10°C), means that it takes only 31 ms to travel the length of the leg.

Neuronal conduction velocities in similar sized peracarid crustaceans from Antarctic (P. gibber) and temperate (L. oceanica)environments were very similar when measured at the appropriate species-specific environmental temperatures. At the minimum Antarctic environmental temperature of –1.8°C, conduction velocity in P. gibber was 0.71 m s^–1. Conduction velocity in its temperate relative L. oceanica was 24% slower than this, at 0.54 m s^–1, when measured at the higher environmental minimum temperature for that species of +2°C(Table 2). Even at 10°C,approximately the mid point of the temperature range for L. oceanica,and well above the maximum environmental temperature for P. gibber,the Antarctic species maintained a faster conduction velocity than its temperate relative. Although partial compensation(Prosser, 1958) has been observed in both Antarctic fish(Macdonald, 1981) and Arctic fish (Melani and Moran, 1998),our data provide evidence for perfect compensation of neuronal conduction velocity in temperate and Antarctic peracarid crustaceans. Conduction velocity in L. oceanica only became faster than that in P. gibber at temperatures above 10°C (Table 2). This means that, when compared at their respective upper environmental temperature limits of +1.5°C and +23°C, conduction velocity in the Antarctic species (0.85 m s^–1) was 60% slower than in the temperate one (2.11 m s^–1).

Neuronal conduction velocity was much more rapid in our recordings from C. maenas than in a previous study(Fig. 8)(Fraser, 1990), perhaps because previous data were from a C. maenas sensory nerve. The neuronal conduction velocity of G. antarcticus in our study was slightly slower than reported in a previous study(Fig. 8)(Macdonald, 1981), although the thermal dependence of neuronal conduction velocity of G. antarcticus was similar [0.058 m s^–1deg.^–1, this study; 0.052 m s^–1deg.^–1 (Macdonald,1981)].

Neuronal conduction velocity had a higher temperature dependence in temperate C. maenas than in the Antarctic species G. antarcticus and P. gibber(Fig. 3A). In contrast, the other temperate species L. oceanica, did not, so eurythermality alone cannot explain the different thermal dependencies of the different species. L. oceanica has a similar thermal dependence to that of Antarctic species tested in the present and previous studies(Fig. 8), but the thermal dependence of conduction velocity in C. maenas is more similar to those of the primarily tropical insects studied to date(Fig. 8). It may be beneficial toC. maenas, and perhaps tropical insects, to have a higher thermal dependence of neuronal conduction velocity, so that behavioural rates can be modified strongly to utilise resources that fluctuate with temperature.

The tropical arthropods studied to date are relatively inactive at low temperatures (e.g. Uvarov,1977). Our study, therefore, broadens considerably the diversity of animals for which the temperature dependence of neuronal function has been studied. Moreover, our data extend by more than 10°C the lowest temperatures that have been analysed in most previous studies, and include for the first time temperatures significantly below 0°C. The linear temperature dependence of velocity reported in a range of species(Fig. 8) is shown to continue to –1.8°C in all four species of crustacean we studied. The high thermal dependence of conduction velocity in tropical species(Fig. 8) presumably contributes to the high thermal dependence of behavioural functions, such as wing-beat frequency in flying locusts (Xu and Robertson, 1994).

Of previous studies, only that of the squid Loligo vulgaris(Chapman, 1967) demonstrated a non-linear temperature dependence of conduction velocity(Fig. 8). Here, the rate of conduction dropped markedly at temperatures below 0°C. Our data show that this low-temperature non-linearity is not a general feature across species. Our data for C. maenas, in contrast, reveal a positive non-linearity at high temperatures (>10°C) in some animals and a weak plateau in others (Fig. 2E). It is unusual for a physiological process to have a linear temperature dependence, and yet with very few exceptions (e.g. C. maenas, this study), neuronal conduction does (Fig. 2)(Macdonald, 1981), so it is not clear whether this linearity results from the interaction of many non-linear relationships, or is due to one single underlying factor that has a positive linear relationship with temperature.

All four species studied have similar temperatures of upper thermal block(Fig. 2). Although temperate species survive temperatures exceeding 10°C, which cause high mortality in Antarctic invertebrates (Wells,1979; Peck and Conway,2000), there is no corresponding difference in their upper limit for action potential propagation. This suggests that the failure to conduct action potentials is not what determines mortality at 10°C in the Antarctic species. Moreover, it suggests that there is little difference in the thermal stability of the membrane ion channels involved in action potential propagation between temperate and Antarctic species.

For G. antarcticus and C. robusta Macdonald(Macdonald, 1981) reported the upper thermal block in leg nerves at 31°C and at 28°C, i.e. around 10°C higher than the values reported in the present study. This difference may result from different rates of imposed temperature change or from a differing thermal dose between the two studies.

Tropical insects and a temperate squid have higher temperatures of upper thermal block, propagating action potentials at temperatures greater than 30°C (Fig. 8)(Chapman, 1967; Chapman and Pankhurst, 1967; Burrows, 1989; Xu and Robertson, 1994; Höger and French, 1999),which is clearly an essential adaptation to carrying out behaviour at these high temperatures, to which the marine crustaceans are never naturally exposed.

The lowest temperature at which action potentials can be conducted is poorly documented because most studies have not tested temperatures below 10°C (e.g. Burrows, 1989; Höger and French, 1999),and in other cases, the minimum experimental temperature was not given (e.g. Fraser, 1990). The squid L. vulgaris has a lower thermal block of –0.5°C(Chapman, 1967). Our data demonstrate that in both stenothermal Antarctic species and eurythermal temperate species of crustacean, the lower limit for action potential conduction is close to the freezing point of seawater, –1.8°C. Terrestrial species of insects that continue to behave at even lower temperatures [e.g. the Himalayan midge Diamesa Meigen sp., active at–16°C (Kohshima,1984)] must conduct action potentials at these extreme temperatures, but this has not been demonstrated explicitly.

For the Antarctic stenotherm G. antarcticus, we demonstrate that some sensory receptors produce more action potentials per stimulus at temperatures exceeding the normal environmental range (i.e. 5–10°C),than within it. A consequence of this must be that the dynamic range of the output firing of the receptors is not optimal under normal conditions. Perhaps the energy costs of generating higher frequencies of activity are prohibitive at low temperatures. Alternatively it may be that the output effects (e.g. post-synaptic effects in down-stream neurones) saturate at even low frequencies of firing when temperature is low, meaning that higher rates would simply waste energy to no effect.

Our measure of conduction velocity was based on latency of the fastest axons contributing to the compound action potential(Fig. 1). Conduction velocities were faster for G. antarcticus than L. oceanica(Fig. 2E), suggesting the presence of larger diameter axons in the former. Our transmission electron microscope data demonstrate, however, that the number of large diameter axons is the same in both species and that the median axon diameter is smaller in G. antarcticus than L. oceanica(Fig. 6).

The median diameters of axons from two individuals of G. antarcticus (0.89 and 0.28 μm) were greater than the average of 0.1μm reported by Macdonald (Macdonald,1981). A reason for this difference is not apparent, but Macdonald may have sectioned sensory nerves, which generally have smaller diameter axons than motor nerves (e.g. Xu and Robertson,1994) or may have used particularly small or young animals. Axons of 0.1 μm diameter and smaller were resolved in the present study, but they represented only a small fraction of the axons within the leg nerves.

The conduction velocity of an axon is proportional to its length constant(Hodgkin, 1954), and both can be increased through an increase in axon diameter or membrane resistance. As the difference in absolute conduction velocity between G. antarcticusand L. oceanica is not explained by axon diameter, this suggests that the layers of wrapping around some large diameter axons in G. antarcticus increase membrane resistance and therefore conduction velocity. Some calanoid copepods belonging to the superfamilies Megacalanoidea and Clausocalanoidea have axons and nerve bundles surrounded by myelin sheaths(Davis et al., 1999). Myelination enhances conduction velocity by increasing the membrane resistance and therefore the length constant of the axon(Eckert and Randall, 1983). The wrappings are responsible for the extremely rapid escape responses of the copepods Undinula vulgaris and Neocalanus gracilis(Lenz et al., 2000; Weatherby et al., 2000), which have latencies of around one quarter of those found in closely related species that do not have myelin-like wrappings. These more rapid reaction times apparently permit U. vulgaris and N. gracilis to colonise open water, whereas species without myelin are restricted to deep water. Despite the apparent advantages, myelin-like wrappings are surprisingly rare in invertebrates. The wrappings in G. antarcticus appear less dense than those of the copepods, but if they act to enhance conduction as we hypothesise, then this could provide a mechanism for cold adaptation in this large animal.

There was considerable variation in both the number and diameter of axons present in individuals of the same species. The largest difference was in the smallest axons (those less than 1 μm in diameter) with a fourfold difference between two individuals of G. antarcticus. There was no difference in the number of axons greater than 1 μm in diameter. The small sample size precludes a definitive explanation for this observation, but one possibility is that the animals were of different ages. As an arthropod ages and grows, the number of sensory sensilla increases(Jander and Jander, 1994; Sandeman and Sandeman, 1996; Brézot et al., 1997; Steullet et al., 2000). This proliferation would produce a difference in the number of primarily small diameter axons between two age groups. Such differences would not have resulted in intra-specific variation in neuronal conduction velocity, because our measure of velocity is based on the fastest action potentials,representing activity of the largest diameter axons.

We have demonstrated that neuronal conduction in the leg nerves of two species of Antarctic crustacean continues to temperatures at least 10°C higher than those that cause the animals to die. The upper thermal block for neuronal conduction is the same in temperate and Antarctic species, despite the temperate species surviving at higher temperatures. The temperature dependence of neuronal conduction velocity is the same in three peracarid species – the Antarctic G. antarcticus and P. gibber,and the temperateL. oceanica – but is greater in the temperate decapod C. maenas. There is no evidence to support an overriding effect of habitat (i.e. temperate vs Antarctic) on this aspect of neuronal function.

This work was supported by a BBSRC Studentship to J.S.Y. We thank the staff of Rothera and Palmer Stations for collection of animals, Lucy Conway for assistance with animal transport and maintenance, and Tony Walne at the Sir Alister Hardy Foundation for Ocean Science (SAHFOS) for the North Sea temperature data.