Directory UMM :Journals:Journal of Insect Physiology:Vol 47.Issue1.Jan2001:

(1)

www.elsevier.com/locate/jinsphys

Critical thermal limits, temperature tolerance and water balance of

a sub-Antarctic kelp fly, Paractora dreuxi (Diptera: Helcomyzidae)

C. Jaco Klok

_{, Steven L. Chown}

Department of Zoology and Entomology, University of Pretoria, Pretoria 0002, South Africa Received 11 April 2000

Abstract

Paractora dreuxi displays distinct ontogenetic differences in thermal tolerance and water balance. Larvae are moderately freeze

tolerant. Mean larval onset of chill coma was 25.1°C, and onset of heat stupor was 35.5°C. Larval supercooling point (SCP) was

23.3°C with 100% recovery, although mortality was high below 24°C. Starvation caused SCP depression in the larvae. Adults were significantly less tolerant, with critical thermal limits of 22.7 and 30.2°C, no survival below the SCP (29.6°C), and no change in SCP with starvation. Moderate freeze tolerance in the larvae supports the contention that this strategy is common in insects from southern, oceanic islands. Fly larvae survived desiccation in dry air for 30 h, and are thus less desiccation tolerant than most other sub-Antarctic insect larvae. Water loss rates of the adults were significantly lower than those of the larvae. Lipid metabolism did not contribute significantly to water replacement in larvae, which replaced lost body water by drinking fresh water, but not sea water. Kelp fly larvae had excellent haemolymph osmoregulatory abilities. Current climate change has led to increased temperatures and decreased rainfall on Marion Island. These changes are likely to have significant effects on P. dreuxi, and pro-nounced physiological regulation in larvae suggests that they will be most susceptible to such change. 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Ontogenetic differences; CTMin; CTMax; Freeze tolerance; Osmoregulation; Marion Island; Climate change

1. Introduction

The majority of studies concerning temperature and water relations of arthropods in the sub- and maritime Antarctic regions has concerned strictly terrestrial spec-ies. With the exception of a few studies of semi-aquatic flies, beetles and copepods (Chown and Van Drimmelen, 1992; Chown, 1993; Convey and Block, 1996; Daven-port and MacAlister, 1996; DavenDaven-port et al., 1997), little emphasis has been given to environmental tolerances of arthropods living in shoreline or semi-aquatic habitats (see Block, 1984; Klok and Chown 1997, 1998; Sinclair, 1999). Nonetheless, these habitats, and particularly the shore zones, are characterized by high arthropod diver-sity (Bellido, 1981; Trave´, 1981; Chown, 1990; Marshall et al., 1999), of which a variety of species are major

* Corresponding author. Tel.: +2712-420-3236; fax: + 2712-342-3136.

E-mail address: cjklok@zoology.up.ac.za (C.J. Klok).

contributors to ecosystem functioning (Tre´hen et al., 1985; Crafford and Scholtz, 1987; Ha¨nel and Chown, 1998). In consequence, there is little information avail-able on the way in which these ecologically important arthropods cope with their often variable environments, and how their physiological responses differ to those of arthropods from terrestrial environments in the region.

Such comparisons are of considerable interest from three perspectives. First, the ways in which terrestrial and aquatic species cope with their respective environ-ments are usually quite different (e.g. Lee and Denlinger, 1991; Hadley, 1994), yet many insect species make the transition from one environment to the other as they develop. Although marked ontogenetic differences in physiology have been documented in some insect spec-ies (Morrissey and Baust, 1976; Tre´hen and Vernon, 1986; Vernon, 1986; Vernon and Vannier 1986, 1996), the subtleties of such differences remain to be thor-oughly explored (see Spicer and Gaston, 1999 for discussion). Second, it has been mooted repeatedly that the cold hardiness strategy adopted by most mid-latitude

(2)

southern hemisphere insect species is one of freeze toler-ance (e.g. Klok and Chown, 1997; Van der Merwe et al., 1997). This is thought to be a consequence of the oceanic nature of the region, which leads to generally moist habitats and, as a result, a high risk of freezing due to external inoculation by ice crystals. Nonetheless, data with which to test this idea are limited, and it has not been well-explored in species where adult and larval habitats might differ (see Addo-Bediako et al., 2000; Convey and Block, 1996 for exceptions). Third, differ-ences in the physiological tolerances of terrestrial and aquatic species are not only likely to influence the range of habitats they can occupy, and hence, ultimately differ-ences in their geographic ranges (Chown and Gaston, 1999), but they may also influence the extent to which changes in the abiotic environment will differentially influence these species. The latter is of particular con-cern in Antarctic and sub-Antarctic environments where climates are changing rapidly (Smith et al., 1996; Berg-strom and Chown, 1999).

In this study we examine thermal tolerances and water balance in adults and larvae of Paractora dreuxi Se´guy (Diptera: Helcomyzidae), a brachypterous kelp fly that frequents kelp deposits in the littoral habitats on the coastlines of the Prince Edward and Crozet islands in the sub-Antarctic (Crafford, 1984; Tre´hen et al., 1985; Crafford and Scholtz, 1987). The associations of P.

dreuxi with kelp deposits differ markedly between the

larvae and the adults. The larvae are less mobile than adults and are mostly confined to kelp, or the substrate below the kelp, where they either burrow into the fronds, feed between the fronds, or, in the later instars, feed on the underlying detritus (Crafford, 1984; Crafford and Scholtz, 1987). Here they are subject to substantial fluc-tuations in temperature and water availability. In con-trast, the shorter-lived adults, though brachypterous, are highly mobile as a consequence of their search for ovi-position and feeding sites (Crafford, 1984; Tre´hen et al., 1985). However, they often take refuge, at least on Marion Island (Prince Edward group), between the boul-ders below the kelp-substrate interface either when pred-ators are present, or when weather conditions are unfavourable to them. Using flies collected at Marion Island, we test the hypothesis that stage-related differ-ences in association of P. dreuxi with stranded kelp will result in differences in their environmental tolerances. We also examine the idea that these differences will affect the likely impact of climate change (especially declining rainfall) on the stages.

2. Material and methods

2.1. Study site and animals

Sub-Antarctic Marion Island (46°549S 37°459E) is the larger of two islands forming the Prince Edward Islands

group, and lies 2100 km south east of Cape Town, and to the north of the Antarctic Polar Front. It has a highly oceanic climate with mean monthly temperatures rang-ing from a winter low of 2°C to a summer high of 7°C with little annual variation (Schulze, 1971; Blake, 1996). Precipitation is in excess of 2000 mm per yr and the island experiences a high degree of cloudiness (climate data from Schulze, 1971; Smith and Steenkamp, 1990). Sea surface temperatures vary between 2°C in winter to 8°C in summer with an annual mean of 5°C (De Villi-ers, 1976).

On boulder beaches, P. dreuxi larvae and adults occur from just above the high water mark to the lower limit of closed vascular vegetation, feeding predominantly on decaying kelp deposits, but also, facultatively, on seal and penguin carcasses and other carrion (Crafford, 1984; Tre´hen et al., 1985). Larval duration is approximately two months, of which instar III lasts for 40 – 50 days and is responsible for most kelp consumption. Mature instar III larvae pupariate below the beach surface to a depth of 50 cm. The pupal stage lasts for 30 – 60 days and the adults can live for 14 – 21 days (Crafford, 1984), feeding on the slime that covers decaying kelp fronds (personal observation). The largest biomass of third instar larvae is found in the supralittoral zone, 7 – 9 m from the shore-line. Both adults and larvae avoid the intertidal zone but do occur in low densities in the splash zone (the upper part of the eulittoral zone) situated 3 – 7 m from the shoreline (Crafford and Scholtz, 1987). Nonetheless, this species may come into contact with seawater, either as saltspray blown ashore by high winds, or during storms when large swells, which override the small (71 cm) tidal range, inundate the boulder beaches well into the supralittoral zone (De Villiers, 1976; Smith, 1977; Craf-ford and Scholtz, 1987).

Large wrack beds are known to have stable internal temperatures compared to ambient temperatures (Crafford and Scholtz, 1987; Chown, 1996). However, sunshine, varying wind conditions, rainfall and sporadic snowfalls do cause wide fluctuations in the microclimate of smaller kelp deposits, especially the smaller strings that are also utilized by P. dreuxi (Crafford, 1984). Rela-tive to the stable ambient temperatures, temperatures in kelp deposits on Marion Island can fluctuate widely, from below zero (mean of ₂2.3±0.5°C, range ₂1.3 to 23.0, n=10 observation days with single temperatures recorded during the early morning) in winter, to 29 – 37°C during sunny days in early spring (Crafford, 1984, C.J. Klok, unpublished data). Furthermore, kelp deposits often desiccate completely despite the high rainfall at Marion Island (Crafford and Scholtz, 1987), although they rehydrate after rain, the main source of fresh water on boulder beaches.

This study was undertaken from May to October 1998, a period when low temperatures are common. Blake (1996) has demonstrated that in lowland areas of

(3)

Marion Island, freezing of the soil surface takes place regularly (weekly, and often more regularly) at this time of year, and that temperatures decline to approximately 24°C on a regular basis. Adults and third instar larvae were collected with an aspirator from decomposing kelp deposited on Trypot Beach. This site was selected because kelp deposition is regular (albeit random), and because of its proximity to the laboratory. Live speci-mens were taken to the laboratory within two hours of collection. In the laboratory, the animals were kept in a temperature-controlled cabinet (10±0.5°C, with an 11L:13D photoperiod) in plastic containers and supplied with decomposing kelp. Animals were not held for longer than 24 h in the cabinets prior to experiments. In instances where animals were kept for longer periods as part of an experimental protocol the temperature and light regimes were as above.

2.2. Critical thermal limits

The methods of Roberts et al. (1991) and Klok and Chown (1997, 1998) were followed to determine critical thermal minima, identified as the onset of and recovery from chill coma (CTMin/o for onset and CTMin/r for recovery), and critical thermal maxima, identified as the onset of heat stupor (CTMax), of the larvae and adults. Eleven larvae (carefully wiped dry using tissue paper, see below), or adults, were placed individually into vials, with one larva, or adult, serving as an operative ther-mometer. A 40-gauge copper-constantan thermocouple was inserted into the body core of the larva or the thorax of the adult to measure body temperature (Tb) during the course of the experiment. The Tb of the operative thermometer was assumed to be representative of the Tb of the ten experimental animals. The vials were sub-merged in a Grant LTD 6 water bath (0.1°C accuracy) connected to a PZ1 programmable temperature control-ler. For CTMin/o the bath’s temperature was allowed to stabilize for 15 min at 2°C after which it was lowered at 0.5°C min21 _{until the onset of chill coma was} recorded in all animals. To identify the onset of chill coma in larvae, their anterior parts were pinched gently using a fine wooden wand. The tip of the latter was retained in the vials to minimize heat exchange. Chill coma was indicated by the larvae’s inability to retract from this pinching and by the adults’ inability to right themselves after being turned on their backs. The bath temperature was allowed to drop 0.5°C below the last CTMin/ovalue measured. The animals were held there for five minutes to allow all animals to equilibrate, and the temperature was then increased at the same rate. The temperature at which an individual regained complete motor function was noted as the CTMin/rfor that individ-ual. For CTMax, a similar procedure was adopted. Ani-mals were held for 15 min at 20°C and the temperature was subsequently increased at 0.5°C min21

until the

onset of uncontrolled muscular spasms in the last animal. The experiments were repeated to increase the sample sizes to n=20. Because larvae in the field are often covered by a thin layer of slime produced by the decay-ing kelp, the effects of this slime on larval CTMin was also determined by examining a second set of larvae that had not been wiped dry.

2.3. Supercooling points

Supercooling points (SCPs) of individual larvae (wet and dry) and adults were determined by placing the ani-mals tightly in pipette tips and inserting 40 gauge cop-per-constantan thermocouples between the animal and the pipette tip wall. The pipette tips containing the ani-mals were transferred individually to air filled vials and submerged in the water bath at 0°C. After a 15 min equi-libration period temperature was lowered at 0.1°C min21 (see Klok and Chown 1997, 1998; Kelty and Lee, 1999). The thermocouples were connected to a Campbell Scien-tific CR10 datalogger that logged temperatures every second and calculated a mean every 15 s. The tempera-ture logged just prior to the freezing exotherm was taken as the supercooling point.

The SCPs of freshly collected larvae and adults were determined. To investigate whether feeding state might influence SCPs, larvae and adults were kept at 10°C (11L:13D) for four and three days respectively, with no access to decaying kelp, but access to fresh water.

2.4. Lower and upper thermal tolerances

To determine lower lethal temperatures, the methods of Worland et al. (1992) and Klok and Chown (1997, 1998) were followed. Batches of ten larvae and ten adults were placed in glass vials submerged in the pro-grammable water bath. After a 15 min equilibration per-iod at 0°C, the specimens were cooled at 0.1°C min21 until the first temperature,21°C, was reached. After one hour at this temperature, ten specimens were removed from the bath and were given 24 h to recover at 10°C. Subsequently, the temperature was lowered by 1°C inter-vals at 0.1°C min21_until

26°C was reached in the case of the larvae, and₂10°C in the case of the adults, corre-sponding to their respective mean field fresh SCPs. The removal procedure was repeated at each temperature interval. After 24 h, the larvae and adults were assessed for survival. Those specimens with full motor function were considered survivors and were kept at 10°C for seven more days and then reassessed. Lower lethal tem-perature was considered that temtem-perature after which survival was consistently less than 100%.

To determine upper lethal temperatures, larvae and adults were warmed at 0.1°C min21 _{to 30 and 25}°_C, respectively. After one hour, ten specimens were removed and the temperature raised by 1°C. This

(4)

pro-cedure was repeated until temperatures were reached that corresponded to the CTMax values of the larvae and adults respectively. Survival assessment was the same as above.

2.5. Desiccation and starvation resistance

For desiccation resistance, Chown’s (1993) protocol was followed. Twenty field-collected larvae were given access only to fresh water. The animals were kept for 24 h in the controlled-temperature cabinet to allow them to clear their digestive tracts. The individuals were num-bered, weighed (to the nearest 0.1 mg using a Mettler AE163 electronic microbalance) and placed in des-iccation chambers containing silica gel which reduced the relative humidity to 5% (measured using a Novasina Thermohygrometer). These were kept in the controlled temperature cabinets. Larvae were weighed at 6-h inter-vals until 100% mortality. Maximum tolerable water loss, expressed as the percentage fresh mass lost (% FM) and as water loss (mg), time to maximum water loss (h), and rate of water loss expressed as %FM h21 _{and mg} h21_{were determined. These measures were used because} cuticular and respiratory transpiration could not be sep-arated, and they were calculated from the mass recorded in the time interval directly prior to death in each indi-vidual. Weight loss consisted of incidental and respir-atory water loss (Wharton, 1985) and of minute faecal droplets produced only as a reaction to the disturbance during weighing. The effects of body size on the water loss parameters was determined using least-squares lin-ear regression analyses (Sokal and Rohlf, 1995). Because of differences in the size of the larvae used in the experiment and the unreliability of ratios, mass loss (mg), and rates of mass loss (mg h21_{) were corrected} for initial mass, using an analysis of covariance (Packard and Boardman, 1988). This experiment was repeated with the adults, but because of their tendency to attempt to burrow through the mesh and into the silica gel, was terminated after 5.5 h. Larval and adult water loss rate comparisons were made over this period only.

2.6. Larval water balance and osmoregulation

Seventy field-collected larvae were given access only to fresh water and another seventy only to seawater. The larvae were kept for 24 h in the controlled-temperature cabinet to allow them to clear their digestive tracts. At the start of the experiment, ten individuals were chosen at random from both groups. They were weighed, a 10 µl haemolymph sample was extracted with a capillary tube from a small, dorsal cuticular incision, and the lar-vae were then dried to constant mass at 60°C. Haemo-lymph osmolality was determined using a Wescor 5120 B vapour pressure osmometer. Larvae were weighed dry, broken up, and lipids were subsequently extracted using

three rotations (24 h each) of a 2:1 methanol–chloroform solution. Individuals were dried to constant mass again, and lipid content was assumed to be equivalent to the difference between dry mass and lipid-free dry mass (Naidu and Hattingh, 1988).

The remaining individuals of the two groups were numbered, weighed and then transferred to desiccation chambers containing silica gel, which in turn were placed in the controlled-temperature cabinets. At 4, 8 and 12 h all individuals were weighed, and at each inter-val ten individuals, of each group, were removed and their haemolymph osmolality, dry mass, and lipid-free dry mass were determined. After the 12 h interval, the remaining individuals were given access to fresh or sea-water, according to their original treatments. At 1, 2 and 4 h intervals after the start of the rehydration period, all individuals were weighed. At each post-rehydration interval, ten individuals of both groups were removed and their haemolymph osmolality, dry mass and lipid-free dry mass determined. In addition, fresh mass, dry mass, water content, lipid content and haemolymph osmolalities were determined for 10 field-fresh larvae. The osmolalities of freshly collected seawater, fresh water and decaying kelp slime were also determined.

Mass loss (mg), lipid content (mg), and water content (mg) were also corrected for initial mass, using an analy-sis of covariance (Packard and Boardman, 1988) before the means for each desiccation interval were calculated. The same was done for the larvae which were allowed to rehydrate, but this analysis had to be undertaken sep-arately because mass gain, rather than mass loss, was the variable of interest (see Klok and Chown, 1997).

To determine the efficacy of lipid metabolism as a means of supplementing body water during desiccation, we followed the method outlined by Naidu and Hattingh (1988). Both mass loss and lipid content were corrected for larval mass using an analysis of covariance. The cor-rected values were then used in a least squares linear regression of lipid content on mass loss.

The efficacy of haemolymph osmoregulation was determined by comparing haemolymph osmolality values predicted by mean whole body water contents at each time interval, and actual haemolymph osmolality and body water contents. Predicted values were calcu-lated from p_e/pi=V_i/V_e (V is volume, p is haemolymph osmolality, i and e are initial and experimental values respectively; Hadley, 1994), and linear regression lines were fitted to both data sets. For reasons outlined above we used body water content corrected for initial mass as the abscissa. Data for larvae subjected to desiccation, but not given access to water, and those that had access to water, were treated separately.

2.7. Larval tolerance of submersion

Tolerance of total submersion of larvae in fresh water and in seawater was determined by placing batches of

(5)

ten larvae in sealed vials filled with water. Care was taken not to include any air bubbles in the vials. These vials were placed in the temperature-controlled cabinet and a batch of larvae from both fresh water and seawater were removed after 6, 12, 18, 24, 48, 72 and 96 h. After removal the larvae were placed on decomposing kelp and assessed for survival after 24 h, and reassessed after seven days.

Changes in water content and haemolymph osmolality of submersed larvae were also determined. Field-fresh larvae were weighed and submersed individually, as described above, in both seawater and fresh water. Five larvae from both sea- and fresh water were removed and weighed again after 2, 5, 10, 20 and 48 h and their hae-molymph osmolalities and dry mass were determined. After the 48 h interval the remaining seawater submersed larvae were weighed and then switched to fresh water. The fresh water submersed larvae were switched to sea-water. After the switch, five larvae from both treatments were removed at 0.5, 1, 2, 5, and 24 h, re-weighed, and their haemolymph osmolalities and dry masses determ-ined. We adopted this dual procedure to assess whether larvae showed different responses to the order of treat-ments (hyper-osmotic to hypo-osmotic or vice versa, see Parkinson and Ring, 1983; Chown and Van Drim-melen, 1992).

3. Results

3.1. Critical thermal limits

The results of the CTMin/o, CTMin/r, and CTMax experi-ments are summarized in Table 1. At temperatures close to the CTMin/o values of the larvae, the outer layer of slime covering the wet larvae froze, forming a sheath around their bodies. The unfrozen inner layer of slime

Table 1

Summary statistics of CTMin/o,CTMin/r and CTMax(°C) of wet and dry field-fresh P. dreuxi larvae and field-fresh adults.A,Bdenote significantly different means (Tukey HSD, P,0.001). (CTMin/o—onset of chill coma; CTMin/r—recovery from chill coma; CTMax—onset of heat stupor)

CTMin/o Mean±SE Minimum Maximum n

Wet larvae 25.1±0.09A ₂_5.5 ₂_4.0 ₂₀

Dry larvae 24.5±0.13A ₂_5.9 ₂_3.3 ₂₀

Adults 22.7±0.19B ₂_3.7 ₂_0.9 ₂₀

ANOVA: F=30.98, df=2, 57, P,0.0001 CTMin/r

Wet larvae 21.3±0.32A ₂_2.4 ₂_0.3 ₆

Dry larvae 20.3±0.90A ₂_3.9 +_6.5 ₁₃

Adults 1.9±0.40B ₂_2.4 +_4.3 ₁₉

ANOVA: F=6.25, df=2, 35, P=0.0048

CTMax Critical range

Field-fresh larvae 35.5±0.11 34.7 36.6 20 36.8 (40.6)a

Adults 30.2±0.27 27.9 31.9 20 28.3 (32.9)a

a_{Critical temperature ranges are calculated from CTMax}₂_CTMin/r_{to be comparable with previous studies but for the values in parentheses the} CTMin/oof wet larvae and adults were used.

allowed the larvae to free themselves from these sheaths. The newly exposed slime layers subsequently froze again, consequently small frozen tunnels marked the movements of the larvae. In some cases, larvae froze at the ends of these tunnels, after the onset of chill coma. Among the wet larvae, 14 out of 20 specimens froze and none of these recovered when the temperature was increased to determine CTMin/r. The dry larvae were not exposed to this possible external inoculation. Only nine dry larvae froze during the CTMin/o decline in tempera-ture. In contrast with the wet larvae, four of the nine frozen dry larvae recovered along with the unfrozen indi-viduals. However, two of these four specimens appeared much weaker than those that did not freeze. They were not considered survivors. None of the adults froze during the course of the CTMin experiments, but one specimen failed to recover from chill coma. Both the CTMin/o and CTMin/rvalues of the larvae were significantly lower than those of the adults (Table 1). However, when the adults were removed from the analysis, the slime-covered lar-vae had significantly lower CTMin/o values than the dry larvae (F(1,38)=14.28, P_,0.0005), but there was no dif-ference in CTMin/rvalues between the wet and dry larvae that recovered (F(1,17)=0.5, P=0.47). CTMin/o values of larvae that froze, wet or dry, did not differ significantly from those larvae, in either group, that did not freeze (F(1, 18)wet=1.61, P=0.22; F(1,18)dry=0.006, P=0.94).

Paractora larvae had significantly higher CTMax values (see Table 1) than did adults (F(1,38)=277.73,

P_,0.0001). The critical temperature ranges of the larvae and the adults were calculated using the CTMin/rvalues to make the results comparable with those of other similar studies. Larvae have a critical temperature range 8.5°C wider than that of the adults.

(6)

3.2. Supercooling points and temperature tolerances

All of the field-fresh larvae survived freezing and recovered completely. The SCPs of the field-fresh wet and dry larvae (Table 2) did not differ significantly (F(1,28)=2.624, P=0.116). Of the starved group, nine specimens recovered from freezing, but only two of those survived after 24 h and seven days. None of the adults survived freezing. The field-fresh larvae had sig-nificantly higher SCPs than the field-fresh adults (F(2,47)=230.8, P_,0.0001). Starvation drastically depressed the SCPs of the larvae, but had no significant effect on the SCPs of the adults (Table 2). The SCPs of kelp slime (Table 2) were significantly lower than those of both wet and dry field-fresh larvae (F(2,37)=46.89,

P_,0.0001).

Percentage survival in the larvae [Fig. 1(A)] showed a sudden decline with temperature corresponding to the measured CTMin/o and SCP values. All surviving speci-mens remained viable and reacted normally seven days after the experiment. In contrast, adult survival declined gradually as temperature declined, although mortality was high at temperatures higher than their mean field-fresh SCPs. At 25°C the adults suffered further mor-talities in the seven days following the experiment. Below₂5°C the ‘surviving’ adults did not live for more than 24 h.

In the determination of upper lethal temperatures [Fig. 1(B)], larvae showed 100% survival at 31°C (24 h and 7 days after exposure), whereas the adults had already suffered 100% mortality at this temperature. Once again, larvae showed a sudden increase in mortality over a small temperature range, 32–34°C. Surviving larvae from all the high temperature intervals were still alive seven days after exposure. Immediate survival of high temperatures declined more gradually in the adults than in the larvae, but further mortalities occurred in the adults seven days after the experiment.

Table 2

Summary statistics for the supercooling points of P. dreuxi larvae, adults and kelp slime

SCP Mean±SE Minimum Maximum n

Wet field-fresh larvaeA ₂_3.3±_0.05 ₂_3.8 ₂_2.8 ₂₀

Dry field-fresh larvaeA ₂_3.5±_0.08 ₂_3.9 ₂_3.1 ₁₀

Starved larvaeB ₂_7.4±_0.44 ₂_10.6 ₂_4.0 ₂₀

Adults 29.6±0.51 215.0 26.7 20

Starved adults 29.1±0.39 212.0 24.7 19

Kelp slime 25.1±0.28 26.6 24.1 10

ANOVAs F ratio df P

Larval SCPs 62.98 2,47 ,0.0001a

Adult SCPs 0.51 1,37 ,0.48

Wet CTMin/ovs Wet SCP 299.82 1,38 ,0.0001

Dry CTMin/ovs Dry SCP 30.03 1,28 ,0.0001

a_{Different letters denote significant differences.}

3.3. Water balance and osmoregulation

The large range in the maximum tolerable water loss, rate of water loss and time to maximum water loss values (Table 3) is a consequence of the size range of the individual larvae used in this experiment. This is clearly demonstrated by the significant relationships between initial mass and both mass loss prior to death and rate of water loss (Table 3). Most of the larvae reacted with jerking movements when handled and some of them accompanied these movements by excreting a small droplet (_,1 mm in diameter) of liquid from the anus. This excretion continued well into the desiccation period until the weighing intervals preceding death, and thus the gravimetric water loss provided here may be over-estimates of incidental and respiratory loss. Nonetheless, comparisons of water loss rates for larvae (1.187±0.138 mg h21_{) and adults (0.708}±_{0.19 mg h}21_{), corrected by} ANCOVA for differences in initial body mass, showed that these rates differed significantly even over short time periods (ANCOVA F(1,37)=7.136, P=0.011).

Table 4 provides summary statistics for fresh mass, lipid content, water content and haemolymph osmolality of field-fresh larvae and the osmolalities of kelp slime, seawater and fresh water. At the start of the osmoregul-ation experiments (i.e. after 24 h acclimosmoregul-ation in either fresh water or seawater), body mass corrected water con-tents in the two groups were not significantly different (F(1,138)=1.104, P=0.3). When desiccated for 12 h the mass corrected body water contents of the seawater treated larvae were significantly less than those of the fresh water treated larvae (F(1,78)=19.27, P_,0.0001, Fig. 2). On the other hand haemolymph osmolalities after 12 h desiccation were not significantly different (F(1,12)=2.25, P=0.16). During the rehydration period, the mass corrected data showed that the fresh water acclimated larvae, when given access to fresh water, gained significantly more water than those acclimated in

(7)

Fig. 1. The lower lethal temperatures (A), and upper lethal temperatures (B), of P. dreuxi larvae (slime covered) and adults. Specimens were kept for one hour at each treatment temperature and were assessed for survival at 24 h and 7 days after the experiments. Larval survival for both lower and upper lethal temperatures did not decline after seven days.

seawater, when given access to seawater (F(1,58)=25.74,

P,0.0001). In addition, after rehydration, haemolymph osmolality of the fresh water acclimated larvae was sig-nificantly lower than that of the seawater treated larvae that had been given access to seawater (F(1,54)=70.66,

P_,0.0001).

Within the fresh water group, mass corrected water contents and haemolymph osmolalities, after 12 h des-iccation, differed significantly from initial water contents and osmolalities (F(1,63)=4.646, P_,0.0006 and

F(1,59)=9.332, P_,0.0001 respectively). After rehydration, neither the mass corrected water contents nor the haemo-lymph osmolalities differed from the initial values [Fig. 2(A)]. The seawater group’s water contents and haemo-lymph osmolalities also differed significantly from the initial values after desiccation for 12 h, but in contrast, remained different when they were given access to

sea-water for rehydration [F(1,63)=10.403, P,0.001 and

F(1,55)=9.013, P,0.0001, respectively, Fig. 2(B)]. Linear regressions of the observed and predicted values of haemolymph osmolality on mass corrected water content during the desiccation period, for both the fresh water and seawater treated larvae, were all highly significant. The slopes of observed haemolymph osmol-alities on mass corrected water contents in both the fresh and seawater experiments also differed significantly from the slopes derived using predicted haemolymph osmolalities (Fig. 3). The values of the predicted slopes in both treatments are approximately three times larger than what was observed in the experiments. These regressions indicate that larval haemolymph osmolalities did increase somewhat with increasing water loss. How-ever, these osmolalities were much lower than those pre-dicted from a decline in body water content. In contrast,

(8)

Table 3

Summary statistics for initial mass, maximum tolerable water loss, rate of water loss and time to maximum water loss, and the least squares linear regressions of maximum tolerable water loss, rate of water loss and time to maximum tolerable water loss on initial mass in P.dreuxi larvae used in the desiccation experiments (n=20)

Variable Mean±SE Minimum Maximum

Initial mass (mg) 41.9±2.6 19.2 61.9 Maximum tolerable 23.4±1.5 7.3 34.2 loss (mg)

Maximum tolerable 55.60±1.25 37.24 64.14 loss (%)

Corrected 23.4±0.04 20.0 27.1

maximum loss (mg)

Rate (mg h21₎ _0.85±_0.07 _0.43 _1.52 Rate (% h21₎ _2.07±_0.14 _1.17 _3.58 Corrected rate (mg 0.85±0.05 0.53 1.36 h21₎

Time (h) 29.30±2.10 16.50 47.50

Slope±SE Intercept±SE r2 _F _P _df

Water loss (g) 0.5676±0.0369 20.00034±0.0016 0.929 236.32 ,0.0001 18 Rate (g h21₎ _0.0204±_0.0046 _0.00001±_0.0002 _0.517 _19.27 _,_0.0004 ₁₈

Time (h) 81.99±191.99 25.86±8.33 0.01 0.18 0.67 18

Table 4

Summary statistics for fresh mass, water and lipid contents (absolute and percentages) and haemolymph osmolality of field-fresh P. dreuxi larvae (n=10 throughout)

Variable Mean±SE Minimum Maximum

Fresh mass (mg) 39.2±3.8 28.0 60.7

Water content (mg) 33.0±3.4 23.1 50.1

Water content (%) 83.84±0.98 79.28 88.27

Lipid content (mg) 2.2±0.5 0.4 5.8

Lipid content (%) (of dry mass) 38.54±3.32 14.29 52.38

Osmolalities (mOsmol kg21₎

Field-fresh haemolymph 339.9±8.48 292 390

Kelp slime 691.2±12.01 608 736

Seawater 928.0±0.95 922 932

Fresh water 67.7±0.33 66 0

during the rehydration period of both the fresh and sea-water treated larvae, neither the observed changes in haemolymph osmolalities in relation to the increase in mass-corrected water contents, nor the predicted changes gave significant regressions.

No changes in the mass-corrected lipid contents occurred over the 4 h intervals during the 12 h des-iccation period in either the fresh water (F(3,36)=0.9,

P=0.45) or seawater experiments (F(3,35)=1.83, P=0.15) (Table 5). The regressions of lipid contents (corrected for mass) on corrected-mass loss in both fresh water and seawater experiments were likewise not significant (Fresh water; F=0.018, P=0.894: Seawater: F=5.76,

P=0.216). The mass loss observed was thus mostly the result of water loss.

Percentage survival of total submersion in fresh and seawater was high. The few mortalities that did occur did not exhibit any readily comprehensible pattern. Apart from a 20% mortality at 24 h in fresh water, and 10% mortality at 18 h and 20% at 48 and 72 h in seawater, all other larvae survived total submersion in fresh and seawater up to 96 h without any observable adverse effects, when assessed both 24 h and seven days after the conclusion of the experiments.

When larvae were switched from seawater (928 mOs-mol kg21_{) to fresh water (67.7 mOsmol kg}21_{) or vice} versa, after submersion for 48 h, this did not lead to isosmotic changes in the haemolymph or pronounced changes in water contents of the submersed larvae (Fig. 4). Even within each treatment larval body water

(9)

con-Fig. 2. Mass corrected water contents (j_{) and haemolymph osmolalities (}G) of P. dreuxi larvae during desiccation and subsequent rehydration

(means±SE). (A) Larvae were pre-treated in fresh water and after desiccation were given access to fresh water. The recovery during rehydration of lost water content and lower haemolymph osmolalities to levels not significantly different from the initial levels are clearly evident. (B) Larvae were pre-treated in seawater and after desiccation were given access to seawater. Neither lost water content nor haemolymph osmolalities recovered to initial levels.

tents and haemolymph osmolalities did not follow decreasing or increasing trends over time (Fig. 4).

4. Discussion

4.1. Thermal tolerances

The mean supercooling point determined for field-fresh larvae (23.3°C) was close to larval lower lethal temperatures (20% survival at 24°C, but no survival from 25°C onwards). These data contrast strongly with

those from the CTMin/o trials, where mean CTMin/o tem-peratures were significantly lower than the SCPs determ-ined for both the wet and dry larvae (Tables 1 and 2). These discrepancies in the data are most likely a conse-quence of differences in experimental protocol. We sus-pect that the primary way in which experimental proto-col had an influence was through the differences in cooling rate used in the two trials, because it is well known that higher cooling rates often result in lower SCPs (Franks, 1985). However, it should also be noted that during the SCP experiments larvae were completely immobilized, whereas during CTMin/odetermination they

(10)

Fig. 3. Haemolymph osmolalities as a function of body water con-tent, corrected for initial mass, in larvae subjected to desiccation. The hashed lines indicate the regressions fitted to the predicted values (I)

in individuals denied access to water and the solid lines indicate the regression fitted to the observed values (s) for individuals denied access to water. (A) Larvae treated in fresh water, (s) Y=500.5725.02X, r2=_17.84%, _P_,_0.01, _n=₃₆ _and ₍_I₎ Y=951.25216.87X, r2=_{38.65%, P}_,_{0.0001, n}=_{36. The slopes of the} lines differed significantly (t=6.25, P,0.01). (B) Larvae treated in sea-water, (s_{) Y}=_654.03₂_{10.0X, r}2=_{62.77%, P}_,_{0.0001, n}=_{36 and (}_I₎ Y=1378.9231.76X, r2=_{77.94%, P}_,_{0.0001, n}=_{36. The slopes of the} lines differed significantly (t=16.46, P,0.001).

Table 5

Summary statistics of changes in lipid contents during the 12 h desiccation periods of P. dreuxi larvae treated with fresh water and seawater

Fresh water Mean±SE Minimum Maximum n

Initial lipid content (mg) 2.1±0.5 0.4 5.9 10

Corrected initial lipid content (mg) 3.1±0.4 2.4 6.9 10

Dehydrated lipid content (mg) 3.6±0.3 2.7 4.9 10

Corrected dehydrated lipid content (mg) 3.2±0.2 2.2 4.6 10

Seawater

Initial lipid content (mg) 2.2±0.5 0.4 5.8 10

Corrected initial lipid content (mg) 3.2±0.3 1.8 5.3 10

Dehydrated lipid content (mg) 2.9±0.3 1.5 3.8 9

Corrected dehydrated lipid content (mg) 2.5±0.2 1.8 3.5 9

were able to move freely until the onset of chill coma. Thus a difference in movement ability might in some way have affected ice nucleation, because both wet and dry larvae showed the same responses. In addition the decaying kelp slime had a significantly lower SCP than either wet or dry field fresh larvae, but its SCP was vir-tually identical to the CTMin/o of slime covered larvae (F(1, 28)=0, P=1), and was significantly lower than the CTMin/o of dry larvae (F(1,28)=1,28, P,0.038). Thus for confined larvae, ice inoculation is clearly not initiated externally and does not influence their survival. For free moving larvae, freezing of the external slime layer does have a negative influence on their survival as is evi-denced by the lower survival of frozen slime covered larvae in the CTMin/o determinations. This suggests that external ice nucleation is not as well tolerated by the larvae as is inoculation via the gut contents. The latter is clearly important for controlled freezing survival because the SCPs of starved larvae are much reduced and they did not survive freezing (see also Strong-Gund-erson et al., 1992; Lee et al., 1993; Klok and Chown, 1998). The initiation of nucleation in the gut, which leads to survival of the larvae is not unique to P. dreuxi, but has also been described in arctiid and tineid caterpil-lars (Fields and McNeil, 1988; Klok and Chown, 1997), and a tenthridinid wasp (Shimada, 1989). However, the way in which ice nucleation in the gut acts to enhance freezing survival is not clear (Shimada, 1989), and the role of the gut contents (or their absence) in cold hardi-ness remains the subject of some contention (Baust and Rojas, 1985; Parish and Bale, 1990). Thus, while the possibility exists that movement affects the response of larvae to low temperatures, we have no adequate expla-nation for this phenomenon, nor do we have information at our disposal on whether or not larvae become immo-bile at low temperatures in the field. Clearly such behav-ioural information is critical for determining larval sur-vival of a low temperature event.

(11)

Fig. 4. Haemolymph osmolalities (j_{) and water contents (}G) (means±SE), of P. dreuxi larvae initially submersed in water of a given osmolality

(freshwater or seawater) and then switched to water with a different osmolality (seawater or fresh water). Both the haemolymph osmolalities and the water contents of the larvae in either the fresh water to seawater (A), or the seawater to fresh water (B) submersion experiments, did not show any isosmotic changes corresponding to the osmolalities of the submersion media.

If these contradictory results are assumed to be a consequence of differences in experimental design, and if the SCP and lower lethal temperature experiments are accepted as a reasonable indication of the cold hardiness strategy adopted by P. dreuxi larvae, then they must be considered moderately freeze tolerant insects (see Sin-clair’s 1999 cold hardiness classification scheme which is essentially an extension of the original one proposed by Bale (1993, 1996)). Even so, field-fresh P. dreuxi lar-vae had remarkably high SCPs compared to most other freeze tolerant insects. Of a total of 53 freeze tolerant insect species discussed by Sinclair (1999), only three other species had SCP values higher than P. dreuxi’s 23.3°C. Thus the relatively high SCP and proximity of the SCP and lower lethal temperatures in P. dreuxi lar-vae suggest that their moderate freeze tolerance is

none-theless sufficiently well-developed for their particular thermal conditions in kelp depositions. This level of freeze tolerance in P. dreuxi larvae is similar to the situ-ation in the South Georgian beetle, Hydromedion

spar-sutum (Block et al., 1998), and some New Zealand wetas

(Sinclair et al., 1999), and may reflect a general strategy amongst sub-Antarctic insects, which are exposed to low intensity freezing events on a regular basis (see Klok and Chown, 1997; Sinclair et al., 1999 for further discussion). In the case of P. dreuxi larvae, it is signifi-cant that habitat temperatures are unlikely to decline below approximately 23 to 24°C (see Materials and Methods and Blake, 1996). Nonetheless, it should be borne in mind that most other studies have used cooling rates substantially greater than those here, and conse-quently report much lower SCPs (see Franks, 1985; Kelty and Lee, 1999)

(12)

Compared to the larvae, P. dreuxi adults were more sensitive to low temperatures. Additionally they proved to be intolerant of freezing. The mean SCP of field-fresh adults (29.6°C) and the decline in adult survival during the LLT determinations, at temperatures well above the mean SCP, suggest that the adult response is typically that of a summer active insect with no cold hardiness capacity (Zachariassen, 1985). This response does not appear to be influenced by feeding state, presumably either because the gut is not wholly evacuated during starvation, if nucleation is initiated there (see Parish and Bale, 1990), or because, like many other insects, nucleation is initiated in the haemolymph or tissues, and gut contents have no influence over this process (Block, 1990).

4.2. Desiccation resistance and osmoregulation

Survival time and water loss rates calculated for the larvae are slight under-, and overestimates, respectively. This is because some excretion took place during hand-ling. Nonetheless, they are unlikely to be greatly in error due to the small amounts of liquid excreted. Bearing this in mind, it appears that P. dreuxi larvae are not parti-cularly desiccation resistant compared to many other insect species on Marion Island (Chown, 1993; Klok and Chown 1997, 1998). Larval survival time (ca 30 h) was even shorter than that found for desiccation intolerant

Pringleophaga marioni caterpillars (Lepidoptera: Tineidae) (ca 59 h) (Klok and Chown, 1997). In the case of the adults, mass corrected water loss rates over a short period (5.5 h) were significantly lower than those of the larvae. Although neither the adults’ maximum tolerable water loss nor their survival time was estimated, this lower water loss rate suggests that P. dreuxi adults are considerably more desiccation resistant than larvae. Given the adults mobile and exposed lifestyle, this dif-ference is not entirely unexpected. Nonetheless, it remains clear that neither the adults nor the larvae of

P. dreuxi are particularly desiccation resistant, especially

when their mean mass corrected water loss rates are compared with P. marioni (0.66 mg h21

) and E.

hal-ticella (0.49 mg h21_{) larvae over similar short periods} (see Klok and Chown 1997, 1998). An ANCOVA showed that water loss rates of both P. dreuxi larvae and adults do not differ significantly from those of the desiccation prone P. marioni larvae, but that they are significantly higher than those of the desiccation resist-ant larvae of E. halticella (F(3,95)=6.78, P_,0.001).

Despite the fact that P. dreuxi is not truly aquatic, it does have a remarkable osmoregulatory capability. The largely terrestrial P. dreuxi larvae were able to osmoreg-ulate in both hypo- and hyper-osmotic media. In addition, it appears that the larvae are able to osmoregul-ate, at least to some extent, during desiccation. In this respect they are similar to a variety of aquatic Diptera

and other insect species living in habitats with high salt concentrations. In general, these species can maintain stable haemolymph osmolalities in media with widely varying osmolalities (0 to 1500 mOsmol kg21_{) (Nemenz,} 1960; Shaw and Stobbard, 1963; Barnby, 1987; Herbst and Bradley, 1988; Herbst et al., 1988). However, P.

dreuxi larvae differ substantially from several aquatic

dipterans (mostly Chironimoidae, Culicidae and Ephydridae) that live in brackish water and survive as osmoconformers when the osmolality of the medium rises above that of their haemolymph (Shaw and Stob-bard, 1963; Garrett and Bradley 1984, 1987; Herbst and Bradley, 1988).

Gainey (1984) compared the efficiency of osmoregul-ation in various insects using the slope of the regression of haemolymph osmolality on total dissolved solids. The slope of this regression for strong osmoregulating Dip-tera species, that are thought to be amongst the insects with the most pronounced osmoregulatory ability (Gainey, 1984; Barnby, 1987; Herbst et al., 1988; Schwantes and Schwantes, 1990), ranged from 0.02 to 0.64. If the media in which P. dreuxi larvae were kept are regarded as a solution of NaCl ions, then the slope of this regression for P. dreuxi is 0.27. This places P.

dreuxi amongst these strong osmoregulating Diptera.

Such a pronounced osmoregulatory ability of the non-aquatic P. dreuxi larvae should not only enable them to survive brief periods of inundation with seawater, but should also ensure their survival of the high osmolalities associated with sporadic desiccation of kelp, if they are trapped inside the kelp fronds.

Although P. dreuxi larvae can maintain haemolymph osmolality at benign levels when exposed to seawater they do not utilize seawater to replenish lost body water during non-lethal periods of desiccation, nor do they appear to supplement their free water by metabolism of lipids. In the latter case they are like many non-desert insects (e.g. Klok and Chown, 1997; Le Lagadec et al., 1998), but remarkably different to Anatalanta aptera, a sub-Antarctic sphaerocerid fly that is remarkably des-iccation tolerant, and which is thought to metabolize lip-ids to survive dry conditions (Tre´hen and Vernon, 1986; Vernon, 1986; Vernon and Vannier, 1986).

4.3. Physiological tolerances, ontogenetic variation and climate change

Our study of P. dreuxi has provided limited, additional support for the contention that in the moist higher lati-tude environments of the southern hemipshere, freezing tolerance is likely to predominate as a cold hardiness strategy in insects (Klok and Chown, 1997; Addo-Bedi-ako et al., 2000). Nonetheless, to date, Diptera have appeared to constitute an exception to this trend, with most species being freezing intolerant (e.g. A. aptera) (Vernon and Vannier, 1996). Although freezing

(13)

toler-ance is absent in P. dreuxi adults, and not particularly well-developed in the larvae, its presence in this species provides an exception to the general tendency towards freezing intolerance amongst sub-Antarctic and Antarc-tic Diptera, and supports the broader latitudinal pattern. It also suggests that the relationship between habitat moisture content and cold hardiness strategy is an important one. Antarctic and sub-Antarctic flies living in dry environments tend to be freezing intolerant, whereas those from moist environments are freezing tolerant (Convey and Block, 1996; Vernon and Vannier 1986, 1996).

We also found pronounced ontogenetic differences in physiological tolerances in P. dreuxi. Adult flies are not cold hardy, and generally have a reduced thermal toler-ance range compared to the larvae. In addition, they are somewhat more desiccation resistant, although not to the extent that they differ from the larvae when compared with other insect species on Marion Island. Such onto-genetic differences in physiological tolerance are not uncommon in Diptera. For example, they have been found in the partially freezing tolerant Tipula excisa, from high altitudes in Norway (Todd and Block, 1995), and in the goldenrod gall fly, Eurosta solidaginis from North America. Paractora dreuxi differs from these species in that it has overlapping generations that are present throughout the year (Crafford, 1984; Crafford et al., 1986), in a habitat that shows only limited tempera-ture variation compared to high latitudes in the northern hemisphere (see Gaston and Chown, 1999). However, in its case, although the adults and larvae share the same habitat, they behave quite differently. The larvae are much less mobile than the adults, and this means that they are unlikely to be as capable as the adults of pro-nounced behavioural responses to altered environmental conditions (Tre´hen et al., 1985; Crafford and Scholtz, 1987). Thus larvae are able to tolerate a broader range of conditions than the adults, at least as far as tempera-ture is concerned. Such differences in physiological tol-erances associated with ontogenetic differences in behaviour are not uncommon in insects (e.g. Klok and Chown, 1999), and may constitute a source of variation in physiological ability that deserves further, more criti-cal study (see Spicer and Gaston, 1999).

Climate change, and particularly the pronounced dry-ing and warmdry-ing that has taken place at Marion Island, and is predicted to continue (Bergstrom and Chown, 1999), will undoubtedly have a significant effect on P.

dreuxi. Over the past few decades, mean annual rainfall

on Marion Island has declined by 600 mm (Bergstrom and Chown, 1999), leading to general drying of terres-trial habitats (Chown and Smith, 1993). Over a similar period, mean annual temperature has increased by approximately 1°C, and this increase is apparently set to continue in step with global climate change (Tett et al., 1999). We expect that differences in the behaviour of

adults and larvae, coupled with ontogenetic differences in their physiological tolerances, may result in differen-tial susceptibility of these stages to climate change. Elev-ated mean annual temperatures, especially if accompanied by increased sunshine durations (see Smith and Steenkamp, 1990) and decreased rainfall, are likely to mean high microclimate temperatures (Chown and Crafford, 1992; Blake, 1996), and pronounced kelp des-iccation. The desiccation intolerant P. dreuxi may well be affected substantially by this general drying out of habitats. The less mobile larvae are likely to be most affected, particularly because they require fresh water to replenish that lost due to desiccation. Nonetheless, given the nature of our initial physiological work, these predic-tions must remain largely speculative. Thus we suggest that further, more careful investigations of the relation-ships between climate change, environmental suitability, and physiological tolerances of indigenous species in the sub-Antarctic should be undertaken (see also Bergstrom and Chown, 1999).

Acknowledgements

K. Storey, J. Storey and W. Block are thanked for their helpful insights on critical minimum temperatures, mobility and sub-zero freezing. M. McGeogh, J. Bar-endse, R. Mercer and A. Addo-Beddiako commented on an earlier version of the manuscript. C. Jacobs provided partial assistance in the laboratory. B. Sinclair and an anonymous referee provided useful comments on an earlier version of the ms. This research was supported by the South African Department of Environmental Affairs and Tourism (SADEA&T) and the University of Pretoria. Logistic support at Marion Island is provided by the SADEA&T.

References

Addo-Bediako, A., Chown, S.L., Gaston, K.J., 2000. Cold hardiness, climatic variability and latitude. Proceedings of the Royal Society of London B 267, 739–745.

Bale, J.S., 1993. Classes of insect cold hardiness. Functional Ecology 7, 751–753.

Bale, J.S., 1996. Insect cold hardiness: a matter of life and death. Euro-pean Journal of Entomology 93, 369–382.

Barnby, M.A., 1987. Osmotic and ionic regulation of two brine fly species (Diptera: Ephydridae) from a saline hot spring. Physiologi-cal Zoology 60, 327–338.

Baust, J.G., Rojas, R.R., 1985. Review—insect cold hardiness, facts and fancy. Journal of Insect Physiology 31, 755–759.

Bellido, A., 1981. Les biocenoses du littoral rocheux aux Iles Kerg-uelen. CNFRA 51, 81–92.

Bergstrom, D., Chown, S.L., 1999. Life at the front: history, ecology and change on southern ocean islands. Trends in Ecology and Evol-ution 14, 472–477.

(1)

in individuals denied access to water and the solid lines indicate the regression fitted to the observed values (s) for individuals denied access to water. (A) Larvae treated in fresh water, (s) Y=500.5725.02X, r2=_17.84%, _P_,_0.01, _n=₃₆ _and ₍_I₎ Y=951.25216.87X, r2=_{38.65%, P}_,_{0.0001, n}=_{36. The slopes of the} lines differed significantly (t=6.25, P,0.01). (B) Larvae treated in sea-water, (s_{) Y}=_{654.03210.0X, r}2=_{62.77%, P,0.0001, n}=_{36 and (}_I₎ Y=1378.9231.76X, r2=_{77.94%, P,0.0001, n}=_{36. The slopes of the} lines differed significantly (t=16.46, P,0.001).

Table 5

Summary statistics of changes in lipid contents during the 12 h desiccation periods of P. dreuxi larvae treated with fresh water and seawater

Fresh water Mean±SE Minimum Maximum n

Initial lipid content (mg) 2.1±0.5 0.4 5.9 10

Corrected initial lipid content (mg) 3.1±0.4 2.4 6.9 10

Dehydrated lipid content (mg) 3.6±0.3 2.7 4.9 10

Corrected dehydrated lipid content (mg) 3.2±0.2 2.2 4.6 10

Seawater

Initial lipid content (mg) 2.2±0.5 0.4 5.8 10

Corrected initial lipid content (mg) 3.2±0.3 1.8 5.3 10

Dehydrated lipid content (mg) 2.9±0.3 1.5 3.8 9

Corrected dehydrated lipid content (mg) 2.5±0.2 1.8 3.5 9

were able to move freely until the onset of chill coma.

Thus a difference in movement ability might in some

way have affected ice nucleation, because both wet and

dry larvae showed the same responses. In addition the

decaying kelp slime had a significantly lower SCP than

either wet or dry field fresh larvae, but its SCP was

vir-tually identical to the CT

Min/o

of slime covered larvae

(F

(1, 28)

=0, P=1), and was significantly lower than the

CT

Min/o

of dry larvae (F

(1,28)

=1,28, P

,

0.038). Thus for

confined larvae, ice inoculation is clearly not initiated

externally and does not influence their survival. For free

moving larvae, freezing of the external slime layer does

have a negative influence on their survival as is

evi-denced by the lower survival of frozen slime covered

larvae in the CT

Min/o

determinations. This suggests that

external ice nucleation is not as well tolerated by the

larvae as is inoculation via the gut contents. The latter

is clearly important for controlled freezing survival

because the SCPs of starved larvae are much reduced

and they did not survive freezing (see also

Strong-Gund-erson et al., 1992; Lee et al., 1993; Klok and Chown,

1998). The initiation of nucleation in the gut, which

leads to survival of the larvae is not unique to P. dreuxi,

but has also been described in arctiid and tineid

caterpil-lars (Fields and McNeil, 1988; Klok and Chown, 1997),

and a tenthridinid wasp (Shimada, 1989). However, the

way in which ice nucleation in the gut acts to enhance

freezing survival is not clear (Shimada, 1989), and the

role of the gut contents (or their absence) in cold

hardi-ness remains the subject of some contention (Baust and

Rojas, 1985; Parish and Bale, 1990). Thus, while the

possibility exists that movement affects the response of

larvae to low temperatures, we have no adequate

expla-nation for this phenomenon, nor do we have information

at our disposal on whether or not larvae become

immo-bile at low temperatures in the field. Clearly such

behav-ioural information is critical for determining larval

sur-vival of a low temperature event.

(2)

Fig. 4. Haemolymph osmolalities (j_{) and water contents (}G) (means±SE), of P. dreuxi larvae initially submersed in water of a given osmolality

If these contradictory results are assumed to be a

consequence of differences in experimental design, and

if the SCP and lower lethal temperature experiments are

accepted as a reasonable indication of the cold hardiness

strategy adopted by P. dreuxi larvae, then they must be

considered moderately freeze tolerant insects (see

Sin-clair’s 1999 cold hardiness classification scheme which

is essentially an extension of the original one proposed

by Bale (1993, 1996)). Even so, field-fresh P. dreuxi

lar-vae had remarkably high SCPs compared to most other

freeze tolerant insects. Of a total of 53 freeze tolerant

insect species discussed by Sinclair (1999), only three

other species had SCP values higher than P. dreuxi’s

2 3.3°C. Thus the relatively high SCP and proximity of

the SCP and lower lethal temperatures in P. dreuxi

lar-vae suggest that their moderate freeze tolerance is

none-theless sufficiently well-developed for their particular

thermal conditions in kelp depositions. This level of

freeze tolerance in P. dreuxi larvae is similar to the

situ-ation in the South Georgian beetle, Hydromedion

spar-sutum (Block et al., 1998), and some New Zealand wetas

(Sinclair et al., 1999), and may reflect a general strategy

amongst sub-Antarctic insects, which are exposed to low

intensity freezing events on a regular basis (see Klok

and Chown, 1997; Sinclair et al., 1999 for further

discussion). In the case of P. dreuxi larvae, it is

signifi-cant that habitat temperatures are unlikely to decline

below approximately

2 3 to

2 4°C (see Materials and

Methods and Blake, 1996). Nonetheless, it should be

borne in mind that most other studies have used cooling

rates substantially greater than those here, and

conse-quently report much lower SCPs (see Franks, 1985;

Kelty and Lee, 1999)

(3)

Compared to the larvae, P. dreuxi adults were more

sensitive to low temperatures. Additionally they proved

to be intolerant of freezing. The mean SCP of field-fresh

adults (

2 9.6°C) and the decline in adult survival during

the LLT determinations, at temperatures well above the

mean SCP, suggest that the adult response is typically

that of a summer active insect with no cold hardiness

capacity (Zachariassen, 1985). This response does not

appear to be influenced by feeding state, presumably

either because the gut is not wholly evacuated during

starvation, if nucleation is initiated there (see Parish and

Bale, 1990), or because, like many other insects,

nucleation is initiated in the haemolymph or tissues, and

gut contents have no influence over this process

(Block, 1990).

4.2. Desiccation resistance and osmoregulation

Survival time and water loss rates calculated for the

larvae are slight under-, and overestimates, respectively.

This is because some excretion took place during

hand-ling. Nonetheless, they are unlikely to be greatly in error

due to the small amounts of liquid excreted. Bearing this

in mind, it appears that P. dreuxi larvae are not

parti-cularly desiccation resistant compared to many other

insect species on Marion Island (Chown, 1993; Klok and

Chown 1997, 1998). Larval survival time (ca 30 h) was

even shorter than that found for desiccation intolerant

Pringleophaga

marioni

caterpillars

(Lepidoptera:

Tineidae) (ca 59 h) (Klok and Chown, 1997). In the case

of the adults, mass corrected water loss rates over a short

period (5.5 h) were significantly lower than those of the

larvae. Although neither the adults’ maximum tolerable

water loss nor their survival time was estimated, this

lower water loss rate suggests that P. dreuxi adults are

considerably more desiccation resistant than larvae.

Given the adults mobile and exposed lifestyle, this

dif-ference is not entirely unexpected. Nonetheless, it

remains clear that neither the adults nor the larvae of

P. dreuxi are particularly desiccation resistant, especially

when their mean mass corrected water loss rates are

compared with P. marioni (0.66 mg h

) and E.

hal-ticella (0.49 mg h

_{) larvae over similar short periods}

(see Klok and Chown 1997, 1998). An ANCOVA

showed that water loss rates of both P. dreuxi larvae

and adults do not differ significantly from those of the

desiccation prone P. marioni larvae, but that they are

significantly higher than those of the desiccation

resist-ant larvae of E. halticella (F

(3,95)

=6.78, P

,

0.001).

Despite the fact that P. dreuxi is not truly aquatic, it

does have a remarkable osmoregulatory capability. The

largely terrestrial P. dreuxi larvae were able to

osmoreg-ulate in both hypo- and hyper-osmotic media. In

addition, it appears that the larvae are able to

osmoregul-ate, at least to some extent, during desiccation. In this

respect they are similar to a variety of aquatic Diptera

and other insect species living in habitats with high salt

concentrations. In general, these species can maintain

stable haemolymph osmolalities in media with widely

varying osmolalities (0 to 1500 mOsmol kg

_{) (Nemenz,}

1960; Shaw and Stobbard, 1963; Barnby, 1987; Herbst

and Bradley, 1988; Herbst et al., 1988). However, P.

dreuxi larvae differ substantially from several aquatic

dipterans

(mostly

Chironimoidae,

Culicidae

and

Ephydridae) that live in brackish water and survive as

osmoconformers when the osmolality of the medium

rises above that of their haemolymph (Shaw and

Stob-bard, 1963; Garrett and Bradley 1984, 1987; Herbst and

Bradley, 1988).

Gainey (1984) compared the efficiency of

osmoregul-ation in various insects using the slope of the regression

of haemolymph osmolality on total dissolved solids. The

slope of this regression for strong osmoregulating

Dip-tera species, that are thought to be amongst the insects

with the most pronounced osmoregulatory ability

(Gainey, 1984; Barnby, 1987; Herbst et al., 1988;

Schwantes and Schwantes, 1990), ranged from 0.02 to

0.64. If the media in which P. dreuxi larvae were kept

are regarded as a solution of NaCl ions, then the slope

of this regression for P. dreuxi is 0.27. This places P.

dreuxi amongst these strong osmoregulating Diptera.

Such a pronounced osmoregulatory ability of the

non-aquatic P. dreuxi larvae should not only enable them to

survive brief periods of inundation with seawater, but

should also ensure their survival of the high osmolalities

associated with sporadic desiccation of kelp, if they are

trapped inside the kelp fronds.

Although P. dreuxi larvae can maintain haemolymph

osmolality at benign levels when exposed to seawater

they do not utilize seawater to replenish lost body water

during non-lethal periods of desiccation, nor do they

appear to supplement their free water by metabolism of

lipids. In the latter case they are like many non-desert

insects (e.g. Klok and Chown, 1997; Le Lagadec et al.,

1998), but remarkably different to Anatalanta aptera, a

sub-Antarctic sphaerocerid fly that is remarkably

des-iccation tolerant, and which is thought to metabolize

lip-ids to survive dry conditions (Tre´hen and Vernon, 1986;

Vernon, 1986; Vernon and Vannier, 1986).

4.3. Physiological tolerances, ontogenetic variation

and climate change

Our study of P. dreuxi has provided limited, additional

support for the contention that in the moist higher

lati-tude environments of the southern hemipshere, freezing

tolerance is likely to predominate as a cold hardiness

strategy in insects (Klok and Chown, 1997;

Addo-Bedi-ako et al., 2000). Nonetheless, to date, Diptera have

appeared to constitute an exception to this trend, with

most species being freezing intolerant (e.g. A. aptera)

(Vernon and Vannier, 1996). Although freezing

(4)

toler-ance is absent in P. dreuxi adults, and not particularly

well-developed in the larvae, its presence in this species

provides an exception to the general tendency towards

freezing intolerance amongst sub-Antarctic and

Antarc-tic Diptera, and supports the broader latitudinal pattern.

It also suggests that the relationship between habitat

moisture content and cold hardiness strategy is an

important one. Antarctic and sub-Antarctic flies living in

dry environments tend to be freezing intolerant, whereas

those from moist environments are freezing tolerant

(Convey and Block, 1996; Vernon and Vannier 1986,

1996).

We also found pronounced ontogenetic differences in

physiological tolerances in P. dreuxi. Adult flies are not

cold hardy, and generally have a reduced thermal

toler-ance range compared to the larvae. In addition, they are

somewhat more desiccation resistant, although not to the

extent that they differ from the larvae when compared

with other insect species on Marion Island. Such

onto-genetic differences in physiological tolerance are not

uncommon in Diptera. For example, they have been

found in the partially freezing tolerant Tipula excisa,

from high altitudes in Norway (Todd and Block, 1995),

and in the goldenrod gall fly, Eurosta solidaginis from

North America. Paractora dreuxi differs from these

species in that it has overlapping generations that are

present throughout the year (Crafford, 1984; Crafford et

al., 1986), in a habitat that shows only limited

tempera-ture variation compared to high latitudes in the northern

hemisphere (see Gaston and Chown, 1999). However, in

its case, although the adults and larvae share the same

habitat, they behave quite differently. The larvae are

much less mobile than the adults, and this means that

they are unlikely to be as capable as the adults of

pro-nounced behavioural responses to altered environmental

conditions (Tre´hen et al., 1985; Crafford and Scholtz,

1987). Thus larvae are able to tolerate a broader range

of conditions than the adults, at least as far as

tempera-ture is concerned. Such differences in physiological

tol-erances associated with ontogenetic differences in

behaviour are not uncommon in insects (e.g. Klok and

Chown, 1999), and may constitute a source of variation

in physiological ability that deserves further, more

criti-cal study (see Spicer and Gaston, 1999).

Climate change, and particularly the pronounced

dry-ing and warmdry-ing that has taken place at Marion Island,

and is predicted to continue (Bergstrom and Chown,

1999), will undoubtedly have a significant effect on P.

dreuxi. Over the past few decades, mean annual rainfall

on Marion Island has declined by 600 mm (Bergstrom

and Chown, 1999), leading to general drying of

terres-trial habitats (Chown and Smith, 1993). Over a similar

period, mean annual temperature has increased by

approximately 1°C, and this increase is apparently set to

continue in step with global climate change (Tett et al.,

1999). We expect that differences in the behaviour of

adults and larvae, coupled with ontogenetic differences

in their physiological tolerances, may result in

differen-tial susceptibility of these stages to climate change.

Elev-ated

mean

annual

temperatures,

especially

if

accompanied by increased sunshine durations (see Smith

and Steenkamp, 1990) and decreased rainfall, are likely

to mean high microclimate temperatures (Chown and

Crafford, 1992; Blake, 1996), and pronounced kelp

des-iccation. The desiccation intolerant P. dreuxi may well

be affected substantially by this general drying out of

habitats. The less mobile larvae are likely to be most

affected, particularly because they require fresh water to

replenish that lost due to desiccation. Nonetheless, given

the nature of our initial physiological work, these

predic-tions must remain largely speculative. Thus we suggest

that further, more careful investigations of the

relation-ships between climate change, environmental suitability,

and physiological tolerances of indigenous species in the

sub-Antarctic should be undertaken (see also Bergstrom

and Chown, 1999).

Acknowledgements

K. Storey, J. Storey and W. Block are thanked for

their helpful insights on critical minimum temperatures,

mobility and sub-zero freezing. M. McGeogh, J.

Bar-endse, R. Mercer and A. Addo-Beddiako commented on

an earlier version of the manuscript. C. Jacobs provided

partial assistance in the laboratory. B. Sinclair and an

anonymous referee provided useful comments on an

earlier version of the ms. This research was supported

by the South African Department of Environmental

Affairs and Tourism (SADEA&T) and the University of

Pretoria. Logistic support at Marion Island is provided

by the SADEA&T.

References

Addo-Bediako, A., Chown, S.L., Gaston, K.J., 2000. Cold hardiness, climatic variability and latitude. Proceedings of the Royal Society of London B 267, 739–745.

Bale, J.S., 1993. Classes of insect cold hardiness. Functional Ecology 7, 751–753.

Bale, J.S., 1996. Insect cold hardiness: a matter of life and death. Euro-pean Journal of Entomology 93, 369–382.

Barnby, M.A., 1987. Osmotic and ionic regulation of two brine fly species (Diptera: Ephydridae) from a saline hot spring. Physiologi-cal Zoology 60, 327–338.

Baust, J.G., Rojas, R.R., 1985. Review—insect cold hardiness, facts and fancy. Journal of Insect Physiology 31, 755–759.

Bellido, A., 1981. Les biocenoses du littoral rocheux aux Iles Kerg-uelen. CNFRA 51, 81–92.

Bergstrom, D., Chown, S.L., 1999. Life at the front: history, ecology and change on southern ocean islands. Trends in Ecology and Evol-ution 14, 472–477.

(5)

Marion Island. M.Sc. Thesis, University of the Orange Free State, 109 pp.

Block, W., 1984. Terrestrial microbiology, invertebrates and ecosys-tems. In: Laws, R.M. (Ed.), Antarctic Ecology, vol. 1. Academic Press, London, pp. 163–236.

Block, W., 1990. Cold tolerance of insects and other arthropods. Philo-sophical Transactions of the Royal Society of London B 326, 613–633.

Block, W., Worland, M.R., Bale, J., 1998. Respiratory responses to chilling and freezing in two sub-Antarctic insects. Cryobiology 37, 163–166.

Chown, S.L., 1990. Possible effects of Quarternary climate change on the composition of insect communities of the South Indian Ocean Province Islands. South African Journal of Science 86, 386–391. Chown, S.L., 1993. Desiccation resistance in six sub-Antarctic weevils

(Coleoptera: Curculionidae): humidity as an abiotic factor influ-encing assemblage structure. Functional Ecology 7, 318–325. Chown, S.L., 1996. Kelp degradation by Paractora trichosterna

(Thomson) (Diptera: Helcomyzidae) at sub-Antarctic South Geor-gia. Polar Biology 16, 171–178.

Chown, S.L., Crafford, J.E., 1992. Microhabitat temperatures at Marion Island (46°549S 37°459E). South African Journal of Antarc-tic Research 22, 51–58.

Chown, S.L., Gaston, K.J., 1999. Exploring links between physiology and ecology at macro-scales: the role of respiratory metabolism in insects. Biological Reviews 74, 87–120.

Chown, S.L., Smith, V.R., 1993. Climate change and the short-term impact of feral house mice at the sub-Antarctic Prince Edwards Islands. Oecologia 96, 508–516.

Chown, S.L., Van Drimmelen, M., 1992. Water balance and osmoreg-ulation in weevil larvae (Coleoptera: Curculionidae: Brachycerinae) from three different habitats on sub-Antarctic Marion Island. Polar Biology 12, 527–532.

Convey, P., Block, W., 1996. Antarctic Diptera: ecology, physiology and distribution. European Journal of Entomology 93, 1–13. Crafford, J.E., 1984. Lifecycle and kelp consumption of Paractora

dreuxi mirabilis (Diptera: Helcomyzidae): a primary decomposer of

stranded kelp on Marion Island. South African Journal of Antarctic Research 14, 18–22.

Crafford, J.E., Scholtz, C.H., Chown, S.L., 1986. The insects of sub-Antarctic Marion and Prince Edward Islands; with a bibliography of entomology of the Kerguelen Biogeographical Province. South African Journal of Antarctic Research 16, 42–84.

Crafford, J.E., Scholtz, C.H., 1987. Phenology of stranded kelp degra-dation by the kelp fly Paractora dreuxi mirabilis (Helcomyzidae) at Marion Island. Polar Biology 7, 289–294.

Davenport, J., MacAlister, H., 1996. Environmental conditions and physiological tolerances of intertidal fauna in relation to shore zonation at Husvik, South Georgia. Journal of the Marine Biologi-cal Association UK 76, 985–1002.

Davenport, J., Barnett, P.R.O., McAllen, R.J., 1997. Environmental tolerances of three species of the harpacticoid copepod genus

Tigri-opus. Journal of the Marine Biological Association UK 77, 3–16.

De Villiers, A.F., 1976. Littoral ecology of Marion and Prince Edward Islands (Southern Ocean). South African Journal of Antarctic Research 1 (Suppl), 1–40.

Fields, P.G., McNeil, J.N., 1988. The cold-hardiness of Ctenucha

vir-ginica (Lepidoptera: Arctiidae) larvae, a freezing-tolerant species.

Journal of Insect Physiology 34, 269–277.

Franks, F., 1985. Biophysics and Biochemistry at Low Temperatures. Cambridge University Press, Cambridge.

Gainey, L.F., 1984. Osmoregulation in the larvae of Odontomyia cincta (Diptera: Stratiomyidae). Physiological Zoology 57, 663–672. Garrett, M., Bradley, T.J., 1984. The pattern of osmotic regulation in

larvae of the mosquito Culiseta inornata. Journal of Experimental Biology 113, 133–141.

Garrett, M., Bradley, T.J., 1987. Accumulation of proline, serine and

trehalose in the hemolymph of osmoconforming brackish-water mosquitos. Journal of Experimental Biology 129, 231–238. Gaston, K.J., Chown, S.L., 1999. Geographic range size and

speci-ation. In: Magurran, A.E., May, R.M. (Eds.), Evolution of Biologi-cal Diversity. Oxford University Press, Oxford, pp. 236–259. Hadley, N.F., 1994. Water Relations of Terrestrial Arthropods.

Aca-demic Press, New York.

Ha¨nel, C., Chown, S.L., 1998. The impact of a small, alien macro-invertebrate on a sub-Antarctic terrestrial ecosystem: Limnophyes

minimus Meigen (Diptera Chironomidae) at Marion Island. Polar

Biology 20, 99–106.

Herbst, D.B., Bradley, T.J., 1988. Osmoregulation in dolichopodid lar-vae (Hydrophorus plumbeus) from a saline lake. Journal of Insect Physiology 34, 369–372.

Herbst, D.B., Conte, F.P., Brookes, V.J., 1988. Osmoregulation in an alkaline salt lake insect, Ephydra (Hydropyrus) hians Say (Diptera: Ephydridae) in relation to water chemistry. Journal of Insect Physi-ology 34, 903–909.

Kelty, J.D., Lee, R.E., 1999. Induction of rapid cold hardening by cooling at ecologically relevant rates in Drosophila melanogaster. Journal of Insect Physiology 45, 719–726.

Klok, C.J., Chown, S.L., 1997. Critical thermal limits, temperature tol-erance and water balance of a sub-Antarctic caterpillar,

Pringleo-phaga marioni (Lepidoptera: Tineidae). Journal of Insect

Physi-ology 43, 685–694.

Klok, C.J., Chown, S.L., 1998. Interactions between desiccation resist-ance, host-plant contact and the thermal biology of a leaf-dwelling sub-Antarctic caterpillar, Embryonopsis halticella (Lepidoptera: Yponomeutidae). Journal of Insect Physiology 44, 615–628. Klok, C.J., Chown, S.L., 1999. Assessing the benefits of aggregation:

thermal biology and water relations of anomalous emperor moth caterpillars. Functional Ecology 13, 417–427.

Lee, R.E., Denlinger, D.L., 1991. Insects at Low Temperature. Chap-man and Hall, New York.

Lee, R.E., Lee, M.R., Strong-Gunderson, J.M., 1993. Insect cold-hardi-ness and ice nucleating active microorganisms including their potential use for biological control. Journal of Insect Physiology 39, 1–12.

Le Lagadec, M.D., Chown, S.L., Scholtz, C.H., 1998. Desiccation resistance and water balance in southern African keratin beetles (Coleoptera, Trogidae): the influence of body size, habitat and phy-logeny. Journal of Comparative Physiology B 168, 112–122. Marshall, D.J., Gremmen, N.J.M., Coetzee, L., O’Connor, B.M., Pugh,

P.J.A., Theron, P.D., Ueckermann, E.A., 1999. New records of

Acari from the sub-Antarctic Prince Edward Islands. Polar Biology

21, 84–89.

Morrissey, R.E., Baust, J.G., 1976. The ontogeny of cold tolerance in the gall fly Eurosta solidagensis(!). Journal of Insect Physiology 22, 431–437.

Naidu, S.G., Hattingh, J., 1988. Water balance and osmoregulation in

Physadesmia globosa, a diurnal tenebrionid beetle from the Namib

desert. Journal of Insect Physiology 34, 911–917.

Nemenz, H., 1960. On the osmotic regulation of the larvae of Ephydra

cinerea. Journal of Insect Physiology 4, 38–44.

Packard, G.C., Boardman, T.J., 1988. The misuse of ratios, indices and percentages in ecophysiological research. Physiological Zoology 61, 1–9.

Parish, W.E.G., Bale, J.S., 1990. The effect of feeding and gut contents on supercooling in larvae of Pieris brassicae. Cryo-Letters 11, 67–74.

Parkinson, A., Ring, R.A., 1983. Osmoregulation and respiration in a marine chironomid larva, Paraclunio alaskensis Coquillet (Diptera Chironomidae). Canadian Journal of Zoology 61, 1937–1943. Roberts, C.S., Seely, M.K., Ward, D., Mitchell, D., Campbell, J.D.,

1991. Body temperatures of Namib desert tenebrionid beetles: their relationship in laboratory and field. Physiological Entomology 16, 463–475.

(6)

Schulze, B.R., 1971. The climate of Marion Island. In: Van Zinderen Bakker, E.M., Winterbottom, J.M., Dyer, R.A. (Eds.), Report on the South African Biological and Geological Expedition, 1965– 1966. A.A. Balkema, Cape Town, pp. 16–31.

Schwantes, U., Schwantes, P.A., 1990. Influences of different external ion-concentrations on osmolality, internal potassium and sodium levels and uric acid concentration in Musca domestica larvae. Com-parative Biochemistry and Physiology A 96, 273–279.

Shaw, J., Stobbard, R.H., 1963. Osmotic and ionic regulation in insects. Advances in Insect Physiology 1, 315–399.

Shimada, K., 1989. Ice-nucleating activity in the alimentary canal of the freezing-tolerant prepupae of Trichiocampus populi

(Hymenoptera: Tenthredinidae). Journal of Insect Physiology 35, 113–120.

Sinclair, B.J., 1999. Insect cold tolerance: how may kinds of frozen? European Journal of Entomology 96, 157–164.

Sinclair, B.J., Worland, M.R., Wharton, D.A., 1999. Ice nucleation and freezing tolerance in New Zealand alpine and lowland weta,

Hemideina spp. (Orthoptera; Stenopelmatidae). Physiological

Ento-mology 24, 56–63.

Smith, R.C., Stammerjohn, S.E., Baker, K.S., 1996. Surface air tem-perature variations in the western Antarctic Peninsula region. Ant-arctic Research Series 70, 105–122.

Smith, V.R., 1977. A qualitative description of energy flow and nutri-ent cycling in the Marion Island terrestrial ecosystem. Polar Record 18, 361–370.

Smith, V.R., Steenkamp, M., 1990. Climate change and its ecological implications at a sub-Antarctic island. Oecologia 85, 14–24. Sokal, R.R., Rohlf, F.J., 1995. Biometry. The Principles and Practice

of Statistics in Biological Research, 3rd ed. W.H. Freeman, San Francisco.

Spicer, J., Gaston, K.J., 1999. Physiological Diversity and its Ecologi-cal Implications. Blackwell Science, Oxford.

Strong-Gunderson, J.M., Lee, R.E., Lee, M.R., 1992. Topical appli-cation of ice-nucleating-active bacteria decreases insect cold hardi-ness. Applied and Environmental Microbiology 58, 2711–2716. Tett, S.F., Stott, P.A., Allen, M.R., Ingram, W.J., Mitchell, J.F.B.,

1999. Causes of twentieth-century temperature change near the Earth’s surface. Nature 399, 569–572.

Todd, C.M., Block, W., 1995. A comparison of the cold hardiness attributes in larvae of four species of Diptera. Cryo-Letters 16, 137–146.

Trave´, J., 1981. Les bioce´noses halophiles d’Acariens de l’archipel de Kerguelen. CNFRA 48, 149–158.

Tre´hen, P., Bouche, M., Vernon, P., Frenot, Y., 1985. Organization and dynamics of Oligochaeta and Diptera on Possession Island. In: Siegfried, W.R., Condy, P.R., Laws, R.M. (Eds.), Antarctic Nutri-ent Cycles and Food Webs. Springer, Berlin, pp. 606–613. Tre´hen, P., Vernon, P., 1986. Biologie des populations et approche

experimentale de la resistance au froid et des capacites de jeune chez un insecte aptere subantarctique: Anatalanta aptera Eaton (Diptera, Sphaeroceridae). In: Colloque National du C.N.R.S., Actes du Colloque ‘Biologie des Populations’. Lyon, 4–6 Septem-bre 1986. I.A.S.B.E.–Universite´ Claude Bernard, Lyon 1. 1987, 448–455.

Van der Merwe, M., Chown, S.L., Smith, V.R., 1997. Thermal toler-ance limits in six weevil species (Coleoptera Curculionidae) from sub-Antarctic Marion Island. Polar Biology 18, 331–336. Vernon, P., 1986. Evolution des reserves lipidiques en fonction de

l’etat physiologique des adultes dans une population experimentale d’un Diptere subantarctique: Anatalanta aptera Eaton (Sphaeroceridae). Bulletin des Societe´ de Ecophysiologique 11, 95–116.

Vernon, P., Vannier, G., 1986. Etude experimentale de la tolerance au froid chez les adultes d’un Diptere subantarctique: Anatalanta

aptera Eaton (Sphaeroceridae). Colloque sur les Ecosystemes

Ter-restres Subantarctiques 58, 151–167.

Vernon, P., Vannier, G., 1996. Developmental patterns of supercooling capacity in a sub-Antarctic wingless fly. Experientia 52, 155–158. Wharton, G.W., 1985. Water balance of insects. In: Kerkut, G.A., Gil-bert, L. (Eds.), Comparative Insect Biochemistry, Physiology and Pharmacology, vol. 4. Pergamon Press, Oxford, pp. 565–601. Worland, R., Block, W., Rothery, P., 1992. Survival of sub-zero

tem-peratures by two South Georgian beetles (Coleoptera Perimylopidae). Polar Biology 11, 607–613.

Zachariassen, K.E., 1985. Physiology of cold tolerance in insects. Physiological Reviews 65, 799–832.