Activity and dormancy in relation to body water and cold tolerance in a winter-active springtail ( Collembola )

Ceratophysella sigillata (Collembola, Hypogastruridae) has a life cycle which may extend for >2 years in a temperate cli­ mate. It exists in two main morphs, a winter-active morph and a summer-dormant morph in central European forests. The winteractive morph often occurs in large aggregations, wandering on leaf litter and snow surfaces and climbing on tree trunks. The summer-dormant morph is found in the upper soil layers of the forest floor. The cryobiology of the two morphs, sampled from a population near Bern in Switzerland, was examined using Differential Scanning Calorimetry to elucidate the roles of body water and the cold tolerance of individual springtails. Mean (SD) live weights were 62 ± 16 and 17 ± 6 pg for winter and summer individuals, respectively. Winter-active springtails, which were two feeding instars older than summer-dormant individuals, were significantly heavier (by up to 4 times), but contained less water (48% of fresh weight [or 0.9 g g-1 dry weight]) compared with summer-dormant animals (70% of fresh weight [or 2.5 g g-1 dry weight]). Summer-dormant animals had a slightly greater supercooling capacity (mean (SD) -16 ± 6°C) compared with winter-active individuals (-12 ± 3°C), and they also contained significantly larger amounts of both total body water and osmotically inactive (unfrozen) water. In the summer morph, the unfrozen fraction was 26%, compared to 11% in the winter morph. The ratio of osmotically inactive to osmotically active (freezable) water was 1:1.7 (summer) and 1 : 3.3 (win­ ter); thus unfrozen water constituted 59% of the total body water during summer compared with only 30% in winter. Small, but sig­ nificant, levels of thermal hysteresis were detected in the winter-active morph (0.15°C) and in summer-dormant forms (0.05°C), which would not confer protection from freezing. However, the presence of antifreeze proteins may prevent ice crystal growth when feeding on algae with associated ice crystals during winter. It is hypothesised that in summer animals a small decrease in freezable water results in a large increase in haemolymph osmolality, thereby reducing the vapour pressure gradient between the springtail and the surrounding air. A similar decrease in freezable water in winter animals will not have such a large effect. The transfer of free water into the osmotically inactive state is a possible mechanism for increasing drought survival in the summer-dormant morph. The ecophysiological differences between the summer and winter forms of C. sigillata are discussed in relation to its population ecology and survival.


INTRODUCTION
Mass aggregations of insects and other arthropods occur rarely in winter, but some Collembola form colo nies, often in the coldest part of winter (e.g.Uchida & Fujita, 1968), which behave as superorganisms, moving over the forest floor as entities in a highly synchronised manner.The springtail Ceratophysella sigillata (Collem bola, Hypogastruridae) is one such species in European forests (Fig. 1a).It emerges from beneath the snow cover (Fig. 1b), forming dark columns or fronts, which migrate across the snow surface and it also feeds on algae on tree bark in great numbers (Fig. 1c).Few invertebrates exhibit such winter activity followed by a dormant state in summer (e.g.Hagvar, 1995).Two polymorphisms are expressed by C. sigillata: an epitoky (reproductive dimor phism) and a seasonal dimorphism, which includes a summer dormancy and surface-active instars during winter (for details of phenology and polymorphisms see Zettel & Zettel, 1994a, b, 1996).Summer dormancy is triggered by temperature and under natural conditions it is obligatory.
The summer morph remains inactive in the soil from June until late October, the duration of dormancy being controlled by temperature and this is used to synchronise winter activity with season.In experiments using constant thermal conditions (winter), the summer morph and the relative instar were omitted from the developmental cycle (J.Zettel, unpublished).C. sigillata is the only temperate species of Collembola known to have its main growing season during winter, and also the only species with a summer dormancy.Although summer dormancy has been reported for related species under summer drought condi tions in the Mediterranean region (eg.Bedos & Cassagnau, 1988;Cassagnau & Lauga-Reyrel, 1992), growth is not restricted to the winter season in these species.C. sigillata is hygrophilic and requires humid conditions during winter for migration similar to "the snow flea" Isotoma hiemalis (Zettel et al., 1997).Normally it is patchily distributed in the forest floor habitat, and forms dense aggregations or colonies, which move directionally across Fig. 1. a -massed colonies of Ceratophysella sigillata moving across the snow surface of the study plot near Bern, Switzerland; b -individuals emerging from the snow before migrating; c -a monolayer of springtails feeding on algae on tree bark.Body length of mature springtails is1.5 mm. the substrate at speeds o f up to 1 m h"1.The species has w ell developedjum ping and climbing behaviour (Zettel et al., 2000).
These surface colonies are observed in two periods: December to early January and mid-February to mid" M arch (Zettel & Zettel, 1994a;Zettel, 1999), and may be band-or sickle-shaped, measuring up to 50 m or more in width (Fig. 1a).Even without snow cover, the dense colo nies, often comprising many millions o f individuals, are very conspicuous (1 m 2 o f densely covered substrate has been estimated to contain 2 x 10s individuals, whereas the largest aggregations comprise >109 individuals).In late winter, the colonies disperse by migrating over consider able distances.Two strategies are found that reduce the risks o f such a dispersal pattern.Firstly, aggregation in dense colonies guarantees that the sexes meet even after migration.Secondly, reproduction in two consecutive years may be a risk-splitting strategy when summer drought reduces survival o f juveniles during summer dor mancy.
Ceratophysella sigillata is dormant during summer and autumn and has its main growing period in winter.During surface activity, individuals feed intensively at tempera tures down to -2.5°C on coccal algae covering dead wood on the ground, tree trunks and branches (Suter et al., 1993).All available pieces o f dead wood that are cov ered with algae or fungi are coloured purple due to feeding springtails.They may cover tree trunks to a height o f several metres above the ground with a purple monolayer o f Collembola (Fig. lc).Feeding reduces their supercooling ability, but if cold-hardened algae are con sumed springtail cold hardiness is significantly improved (Zettel et al., 2002).However, the process o f sequestra tion o f antifreezes from winter algae is not fully under stood.
The aims o f the present study were to determine the body water status o f the winter-active and summer-dormant morphs, to characterise their cryobiological features and to assess the relation between their freezable and unfreezable water components.The results enable a closer definition o f the survival strategy o f this species to be made and they attempt to answer the question "Does a winter-active arthropod jeopardise its survival through its unusual life cycle?"

Field sampling and culturing
Samples of live C. sigillata were collected by hand from a field site (mixed beech-spruce forest) at 640 m a.s.l., 10 km north of Bern, Switzerland in March 1997, February 1998, February 2001(winter-active individuals), and in September 1997, September 1998, August 2001 (summer-dormant individuals).The samples were maintained at 5°C with a natural photoperiod for 1-2 days after collection and prior to experimentation in the laboratory, except for some samples when mailed from Bern to Cambridge, UK, which were in culture for 2-3 weeks.The animals were kept in glass jars with a moist basal layer of plaster of Paris and fed with coccal algae (Stichococcus sp., Klebsorminium sp., Pleurococcus sp.) scraped from tree trunks at the field site.

W ater content
Individual springtails were weighed live in small aluminium containers on a Sartorius M3P microbalance (accuracy ±1 pg) and after air drying at 60°C over 24 h to constant weight to determine their total body water content.

Thermal analysis
Thermal analysis techniques are widely used in studies of cold resistance and water dynamics of invertebrates (Block, 1994(Block, , 2002(Block, , 2003)).A Differential Scanning Calorimeter (DSC820, Mettler-Toledo) was utilised in this study to examine the thermal changes in individual Collembola subjected to a cooling and warming programme.The DSC system incorporated a heat flux module, a RP100 Intracooler (to -60°C) and an Epson TAS811 workstation.The system was calibrated using indium (melting point 156.6°C,enthalpy 28.71 Jg-1) as an upper temperature and enthalpy standard and dodecane (melting point -9.65°C) as a lower temperature standard.The melting point of HPLC grade water was used as a calibration check.
Individual collembolans were weighed in sealed 40 pl alu minium pans and subjected to the following DSC programme: Cooled from +5 to -30°C at 1°C min-1, held isothermally for 1 min at -30°C and re-warmed to +5°C at the same rate.The resultant thermograms were evaluated using STARe software (version 6.0) to provide data on temperatures of freeze onset (supercooling point, SCP), melt onset, melting point (peak tem perature of endotherm) and enthalpies of both the freeze and melt transitions.Calculations of the quantity of frozen (osmotically active: OA) water in individuals were made using the melt endotherm and the enthalpy value for water (334.5 Jg-1) cor rected for temperature.OA values were expressed as the propor tion of total body water which was frozen in particular samples.The amount of osmotically inactive (OI, unfrozen water) was calculated by subtraction of the OA values from the total water content of each insect.The terms "frozen" and "unfrozen" water are used in respect of experimental results in the present study to avoid confusion with "freezable" and "unfreezable" water, which may be different depending upon time, cooling rate and temperature (see Franks, 1986).

Thermal hysteresis
Using antennal punctures, haemolymph samples (1.2-1.7 nl) were drawn up into glass capillaries (under immersion oil) from freshly-collected individual springtails of the two morphs (23 February 2001 and22 August 2001, respectively) and their osmolality determined using a Clifton nanolitre osmometer (Zettel, 1984).Measurements were made on haemolymph sam ples from individual springtails; no pooling of samples was done.The samples were quickly frozen in the osmometer and then warmed at a constant rate of 0.12°C min-1 until the last minute ice crystal could be observed at x 60 magnification and the temperature was recorded.Then the samples were cooled at the same rate and the temperature at which ice crystal growth began was noted.The difference between the two temperatures was the freezing point depression (thermal hysteresis).Because the haemolymph droplets were very small, the procedure was repeated on each sample and the values were used only when repeated melting point and freezing point temperatures were obtained.Two controls (0 and 1000 mOsm) were used with each haemolymph sample.The amount of thermal hysteresis was cal culated from the osmolal melting point depression of 1.86°C Osm-1 (Duman, 1977).

Statistical analyses
Data were analysed using the non-parametric Mann-Whitney U test (software SYSTAT version 7.0).

Water status
Individuals o f the winter-active morph o f C. sigillata sampled from the field were significantly (P < 0.001) heavier (up to 4x) and contained a greater weight o f water (up to 2.5x) than those o f the summer-dormant morph (Table 1), because they were two (feeding) instars older.However, their relative total water contents, both on fresh and dry weight bases, were reversed; the summer morph having a significantly (P < 0.001) higher value (2.5 g g-1 dry weight) than winter individuals (0.9 g g-1 dry weight).Individual live weights ranged from 5 to 95 pg with summer-dormant animals being between 5 (juveniles) and 20 pg (adults) and winter-active animals being in the The developmental stages within the summer-dormant m orph differed in mean weight and water content (Table 2).The females were heavier than the males and con tained a greater w eight o f water (all by more than 1.5*).Juveniles were smaller than either adult stage, but con tained a similar weight o f water as the males.The relative water content ranged from 68 to 71% o f fresh w eight (2.2 to 2.7 g g-1 dry weight) over the three stages, but the dif ferences were not significant.

Cryobiology
The main cryobiological characteristics o f the two morphs o f C. sigillata are given in Table 3.The signifi cant difference in m ean live weights o f the experimental samples was o f the same order o f magnitude as for field animals (see also Table 1), and the mean temperatures for the onset o f freezing (SCP) were significantly different (P < 0.001) for the two morphs, the SCP for the summerdormant animals being almost 4°C lower than in winteractive animals.The mean temperature for the onset o f melting did not differ between the morphs.More water froze in summer than in winter individuals and this was reflected in the quantities o f unfrozen water, 14% more water remaining unfrozen in the summer m orph than in In terms o f cold hardiness, all the individuals tested were killed by freezing at their supercooling point tem peratures.The supercooling ability o f the summer form and the amount o f freezable water were greater than in the winter morph, and a larger proportion o f the total body water remained unfrozen in the summer morph in these experiments.As a consequence, almost twice the quantity o f osmotically inactive (bound) water was pre sent in summer compared with winter individuals.

Thermal hysteresis
Thermal hysteresis existed in both the winter-active and the summer-dormant morph, being 3 times greater in winter (P < 0.001) (Table 4).Haemolymph melting point was not significantly different in the two morphs, being -0.41°C for winter and -0.38°C for summer animals.However, m ean haemolymph (or hysteretic) freezing point was significantly (P < 0.001) lower in winter-active animals (-0.57°C) than in summer-dormant individuals (-0.44°C).

Water relations
The ratio o f unfrozen : frozen water o f 1 : 3.3 in the winter-active morph o f C. sigillata is comparable to values found by W orland (1996) and W orland et al., (1998) for the Arctic springtail Onychiurus arcticus and for other, larger, cold-adapted arthropods (Block et al., 1998;W orland et al., 2000).In contrast, the ratio for summer morphs o f C. sigillata (1 : 1.7) is low compared with several other cold-hardy invertebrates (Table 5), which means that summer-dormant individuals contain a larger proportion o f unfrozen water compared to other species.For overwintering freeze-sensitive species, it is conceivable that a smaller quantity o f osmotically active water may reduce the potential for lethal freezing.How ever, cold-hardiness is not a problem during aestivation, so an alternative explanation has to be sought for the low level o f water frozen in summer individuals in this popu lation o f C. sigillata.A possible, highly significant, role Table 5.Comparison of the ratio of osmotically inactive (unfrozen) to osmotically active (frozen) body water in selected cold hardy invertebrates.o f bound (osmotically inactive) water could be to main tain a basal level o f body water in an unfrozen state, which is protected from transpiration losses, and which m ay be mobilised during periods o f extreme drought when free water is in short supply and protect the indi vidual from lethal dehydration.It was found that in experiments w ith O. antarcticus individuals could lose almost all their osmotically active water during desicca tion, whereas only a very small amount o f the osmotically inactive water was lost (W orland et al., 1998).A similar situation has been reported for cocoons o f some species o f earthworms, which led to the proposal o f a protective dehydration mechanism for cold survival (Holmstrup & Westh, 1994).

Population ecology
In the study population o f C. sigillata, water is likely to be a limiting factor only during summer.Over ten years o f population monitoring, similar distribution patterns were observed: in late winter (M arch-April), the colonies were distributed throughout the research plot, but in the following December, they reappeared only in the litter layer o f the moist part o f the plot, which was dominated by spruce; in the beech-dominated, drier areas only a few individuals were found at that time with large colonies being absent (Willisch, 2000).The correlation between soil humidity and colony distribution in December was highly significant (P < 0.001).It is suggested that most individuals aestivating in the beech areas die and are replaced by immigration from the adjacent sprucedominated areas in the following February.Dormant springtails are not immobile, but they are unable to penetrate into the soil deeper than the upper limit o f the humus-horizon o f the podsolic soil profile, which is dense and highly compacted, so that the avail able pore spaces are too small for C. sigillata to pass through.W hen active on the surface under favourable humidity conditions in summer the animals are unable to sense the water content o f the underlying soil.Migrating individuals move down into the soil either when weather conditions become unfavourable or when an internal clock (or endogenous rhythm) forces the animals to retreat into the soil in preparation for summer dormancy, independent o f environmental quality.Thus, during periods o f surface activity, colonies cannot select optimal places for aestivation.Therefore insufficient soil moisture during summer drought may cause an ecological bottle neck for dormant individuals.In this context, a physio logical mechanism which increases the amount o f bound (osmotically inactive) water and thus sustains sufficient body water for survival will be crucial in dry periods in summer.Reduction o f cuticular transpiration, in combina tion with an increase in osmotically inactive (OI) water, may be part o f the survival strategy o f the summerdormant morph o f C. sigillata.Alternatively, with a high level o f OI water in summer morphs (59% o f total water), even small decreases in the OA water fraction will pro duce a substantial rise in haemolymph osmolality and with that a decrease in the vapour pressure gradient between the animal and the surrounding air.For example, a loss o f 20% o f the total body water content would approximately double the concentration o f the haemo lymph osmolytes in the OA water compartment (as Collembola do not osmoregulate).In winter individuals, the same loss o f body water would only lead to a 1.4-fold increase o f the original osmolytes.

Cryobiology
Individuals o f C. sigillata had significantly (P < 0.001) lower supercooling points (SCP) in summer; their alimen tary canals being completely empty as feeding does not occur after moulting into the dormant stage.In winter, the guts o f active individuals are 60-90% full, and when starved, food may be retained in the gut for a considerable time: after 22 h o f starvation, Suter et al. (1993) found >40% o f the gut remained full o f food material.The m od erately low SCP o f C. sigillata in winter is sufficient to allow activity on the litter or snow surface down to -2.5°C without freezing (Zettel et al., 2002).Below this thermal threshold the springtails retreat into the protection o f the litter layer.
Invertebrates in freezing terrestrial conditions are sub jected to lower water vapour pressures than in an unfrozen environment.Supercooled terrestrial inverte brates will lose water to the surrounding atmosphere under such conditions (Somme, 1999).Soil inhabiting freeze-sensitive invertebrates probably dehydrate and equilibrate their body fluid melting points to ambient tem peratures in their microhabitats (Holmstrup et al., 2002) and do not rely on supercooling for survival overwinter.In the absence of any accumulated polyhydric alcohols and sugars in C. sigillata, which would lower the water vapour pressure of the haemolymph in the winter morph, water loss will not be reduced by this mechanism.The difference in body water content of the summer and winter morphs is not due to transpiration losses immedi ately prior to the experiment.Although, it has been pro posed that respiratory transpiration may be of considerable importance to the survival of a wide range of xeric species (Addo-Bediako et al., 2001), this is not the case for Collembola which do not possess spiracles or tra chea (except the Sminthuridae).
Lethal freezing is avoided in some earthworm cocoons by a protective dehydration strategy (Holmstrup & Westh, 1994), and a similar mechanism has been reported in the Arctic collembolan O. arcticus  (Worland, 1996;Holmstrup & Somme, 1998;Worland et al., 1998).C. sigillata is similar in one respect to O. arcticus as osmotically active water is lost during desiccation, but in con trast, the osmotically inactive water component of C. sigillata is increased between winter and summer morphs.

Thermal hysteresis
The role that thermal hysteresis proteins play in Col lembola is not fully understood, but they have been detected in several species of drought-resistant arthro pods (Duman et al., 1991).It was postulated that thermal hysteresis is important for water retention in Tenebrio molitor larvae (Patterson & Duman, 1978).Further, Graham et al. (2000) showed that desiccation had the same effect as cold exposure in increasing the level of thermal hysteresis in larvae of T. molitor.THPs were also found in summer-acclimatised oribatid mites (Phauloppia sp.), which were very drought resistant, but had lost up to 90% of their total body water (Sjursen & Somme, 2000).Small, but significant, levels of thermal hysteresis were detected in the winter-active morph (0.15°C) and in summer-dormant form (0.05°C) of C. sigillata, which would not confer protection from freezing.However, the presence of antifreeze proteins may reduce or prevent ice crystal growth when feeding on frozen algae with associ ated ice crystals during winter.
Thermal hysteresis (TH) has not been observed in edaphic species of Collembola which have no winter activity on the snow surface.In winter, the TH of C. sigil lata haemolymph is 10* smaller than in subalpine winteractive species such as I. hiemalis, I. propinqua and A. bidenticulata, which exhibit similar cold avoidance behaviour, retreating into the snow pack when the surface temperature reaches c. -3°C.In summer, the TH of C. sigillata haemolymph (0.05°C) is higher than, for exam ple, than that of I. hiemalis (0.012°C).TH values for B. index (5.2°C)and B. westwoodi (5.4°C) are much greater than in Collembola, which have been studied (Husby & Zachariassen, 1980).Summer TH levels in C. sigillata are comparable to those of Entomobrya nivalis (0.06°C; Zet tel, 1984), which remains on spruce tree bark and endures significant periods of drought.Although thermal hys teresis appears to be important for water retention in larval Tenebrio molitor under dry conditions, where up to 6 °C have been recorded (Patterson & Duman, 1978), it is unlikely that such a small amount (0.15°C) of thermal hysteresis will have a significant influence on water con servation in C. sigillata.

Survival
These results have implications for the population ecology of C. sigillata.Winter activity at low environ mental temperatures appears to present less of a problem for the survival of this population of C. sigillata than summer drought.With a well-developed supercooling ability (SCP -12.5°C) and a small, but significant, level of thermal hysteresis, the species is able to feed and migrate considerable distances, only retreating into deeper layers of the forest floor when surface tempera tures decline below -2.5°C.Feeding at low temperatures does not compromise individual survival: winter-adapted algae significantly increase the cold hardiness of C. sigil lata, when ingested, compared with non-adapted algae or starved control animals (Zettel et al., 2002).Ingestion of ice crystals with food during winter will pose a consider able risk of freezing via inoculation and thermal hystereis proteins may protect individuals by reducing or pre venting ice crystal growth.Recent results (Zettel et al., 2002) suggest that algal antifreezes are sequestrated by springtails during winter, thereby improving their cold tolerance.
The regular extinction of colonies in this springtail population in the beech-dominated part of the study site is brought about by summer aridity in the forest floor habi tat, which causes an ecological bottleneck.In addition, there may be a stress synergy between drought and common environmental contaminants (e.g.agricultural pollutants), which could significantly alter its drought tol erance as has been found for the springtail Folsomia Can dida (Hojer et al., 2001).The results o f the present study suggest that the ecophysiological mechanism, which operates to retain sufficient body water to survive drought, is based upon the transfer o f part o f the freezable (osmotically active) water to the bound (osmotically inac tive) water compartment in C. sigillata.
It is concluded that the survival o f such a winter-active collembolan is not jeopardised by its unusual life cycle as it possesses w ell developed physiological mechanisms to overcome the problems presented by winter activity in freezing conditions and by being dormant during aestiva tion in summer drought.

Table 1 .
Mean (SD) individual body weight and water con tent of winter-active and summer-dormant morphs of the springtail Ceratophysella sigillata from field samples collected inMarch 1997 and September 1998, respectively.

Table 2 .
Mean (SD) individual body weight and water con tent of life stages of the summer-dormant morph of the springtail Ceratophysella sigillata from a field sample collected in September 1998.

Table 3 .
Cryobiological parameters of winter-active and summer-dormant morphs of the springtail Ceratophysella sigil lata from thermal analyses using Differential Scanning Calorimeter techniques.Water contents are as given in Table1.OA: osmotically active; free water; OI: osmotically inactive; bound water.Values are mean (SD) in each case.

Table 4 .
Mean (SD)values of haemolymph osmolality for melting point (MP), freezing point (FP) and thermal hysteresis (TH) in winter-active and summer-dormant morphs of the springtail Ceratophysella sigillata.Temperature equivalents are also given.