Starve and survive: stress tolerance and life-history traits of a bdelloid rotifer

Summary

• 1
  Bdelloid rotifers occur in unstable habitats, such as moss, lichens and temporary pools, where conditions change unpredictably and dramatically. Their ability to survive in such habitats could be related to their physiological ability to enter metabolic stasis in response to severe environmental stress. We tested the response of a bdelloid species (Macrotrachela quadricornifera Milne, 1886) to prolonged starvation by evaluating survival and fecundity once food was resupplied on a regular basis.
• 2
  Animals were starved for different durations (20, 40, 60 days) and all survivors resumed reproduction. Life-history parameters of the 20-day starvation group subsequent to the resumption of normal feeding were studied.
• 3
  From resumption of normal feeding, mean longevity and age-specific traits of starved rotifers were the same as those of the fed rotifers. No reduction of lifetime fecundity was found in the bdelloids that experienced starvation, suggesting a time shift of the effective ‘age’ of the starved rotifers.
• 4
  The response of the rotifer to starvation is similar to that induced by drought. The presence of a previously undocumented form of dormancy is hypothesized.

Introduction

Bdelloid rotifers are aquatic animals that live in habitats such as temporary pools or terrestrial mosses or lichens, where conditions can change unpredictably and dramatically in a short period of time. Their capacity to cope with these changes, coupled with the obligatory parthenogenesis peculiar to all bdelloids, is a key feature determining their persistence. When the habitat dries out, bdelloids arrest ageing and enter a form of dormancy called anhydrobiosis (Càceres 1997; Ricci 2001). Upon rehydration, they resume activity and complete their life cycle apparently unaffected by having entered the anhydrobiotic state (Ricci, Vaghi & Manzini 1987; Ricci & Covino 2005). If temperature fluctuates greatly, bdelloids adjust life span in response to the temperature change, but lifetime fecundity remains constant (Barrows 1968; C. Ricci, unpublished data). The fact that bdelloids produce the same number of offspring under favourable conditions as they do under harsh conditions, temperature fluctuations or dormancy-inducing events suggests that survival is sustained without the use of resources that were committed to reproduction. In other words, trade-offs between fecundity and survival are not apparent.

Like other freshwater animals, bdelloids are subject to periodic shortages of food, a condition known to cause a trade-off between fecundity and survival in most invertebrates (e.g. Sibly & Calow 1986; Stearns 1989). In principle, resources from food are allocated to three processes: maintenance, somatic growth and reproduction. Under starvation, animals are expected to reallocate resources to maintenance that might have been committed to reproduction (Holliday 1989), with the consequence that fecundity is reduced to improve survival. However, species may differ in their responses when food is scarce. For example, under starvation, one group of monogonont species (e.g. Synchaeta pectinata) continued reproducing, but survived briefly. Another group (e.g. Keratella cochlearis) suspended reproduction and survived starvation for a relatively long time (Kirk 1997a). Facing starvation, some rotifers could even extend their life span; young Brachionus plicatilis suppressed reproduction and about half of them survived for more than 16 days without any food, while the fed controls reproduced normally and lived about 12 days (Yoshinaga, Hagiwara & Tsukamoto 2003a). The time that B. plicatilis survives starvation decreases with increasing age at starvation, suggesting that reproduction prior to starvation reduces resources that could be invested in maintenance (Yoshinaga et al. 2003b). Thus some monogonont rotifers seem flexible in allocating energy between maintenance and reproduction, but if they expend resources for reproduction prior to starvation, their life span will be shortened. Monogononts are relatively short-lived and tend to occur in habitats where environmental conditions change cyclically and hence predictably; they survive harsh winter periods as resting eggs produced by sexual reproduction (Gilbert 2003).

The occurrence of bdelloid rotifers in unstable habitats could be related to their physiological ability to regulate their expenditure of resources and suppress activity when conditions deteriorate. The present study analyses the response of a bdelloid species to food deprivation and addresses three major aspects: (1) tolerance to starvation of different periods, (2) recovery and resumption of reproduction when food is resupplied and (3) the effect of prolonged starvation on life histories when food is resupplied.

Materials and methods

Macrotrachela quadricornifera Milne, 1886 is a bdelloid rotifer in the family Philodinidae. The strain used in this study originated from a single individual and has been maintained under controlled conditions for several years (see Ricci 1991). Mean fecundity (eggs per female per lifetime) and mean life span of the strain vary with temperature and food; at 24 °C with suitable food (fish pellets), it lays between 25 and 27 eggs per female and lives for between 30 and 35 days (Ricci 1991; Ricci & Fascio 1995). Prior to the reported experiments, M. quadricornifera was cultivated in 5-ml embryo dishes and fed with powdered fish food (Friskies®) suspended in deionized water at a concentration of 1 mg ml⁻¹ (Ricci 1995). Medium and food were changed daily.

All experiments were run at 24 °C and replicated three times. Each experiment consisted of pairs of homogeneous cohorts; one cohort (S = starved) was starved for a fixed time and then fed daily for the rest of its life; another cohort (F = fed) was fed daily for its entire life and represented the control. Cohorts used for life table experiments consisted of clonal groups of individuals hatched from eggs laid by 10-day-old rotifers. Within 12 h of laying, the eggs were singly transferred to 0·7-ml dishes. Unless stated otherwise, in all experiments each isolated rotifer was fed with 1 mg ml⁻¹ suspension of fish food.

Starvation was induced by replacing the usual culture medium with 0·15 µm filtered water. When starved, bdelloids became inert and contracted and could easily be lost when removing the medium. To avoid disturbance or loss of the experimental rotifers, the water medium was changed irregularly. Although bacteria may have been present in the water medium, they are unlikely to have contributed significantly to nutrition. We set the beginning of starvation as the day of food removal, but note that the animals had food in their gut at this time. After the desired period (see below), food was resupplied, and the numbers of live and dead rotifers were recorded. To better discriminate between live and dead animals, survival was registered 24 h after food addition, and only active rotifers were considered alive. These were fed daily for their remaining life and were studied following a life table experimental protocol; for each cohort mean longevity and mean fecundity were calculated and averaged among the three replicates of each experiment. Age-specific fecundity (m_x) and survival (l_x) rates reported for each experiment were calculated over the ‘virtual’ cohort obtained by summing the three replicates of each treatment. These life table parameters were relative to pre- and poststarvation time and were computed on the bdelloids that survived starvation, only. In other words, l_x curves of S cohorts were set at 1 at resupply of food.

experiment a: tolerance

Rotifers were starved for different times to determine their recovery capacity. Rotifers from a homogeneous group (same age, genotype, birth order) that had been fed daily until reproduction started were isolated in separate wells and randomly assigned to four cohorts of 10–15 bdelloids each. Three of these were starved for 20, 40 and 60 days. The fourth group was fed (i.e. the F cohort). Each of these was replicated three times. After the starvation time, survival was recorded on the day after re-feeding and again 1 week later. Whether the animals had resumed reproduction or not was noted.

experiments b and c: life tables

To investigate the effects of starvation on life-history traits two experiments were designed that differed in the quality of food given to the animals. The food used in experiment B was powdered fish food (1 mg ml⁻¹), known to be suitable for M. quadricornifera (Ricci 1995). In experiment C, the food was Escherichia coli (10⁸ cells ml⁻¹), known to be ‘poor’ for this rotifer species (Ricci 1991). In each experiment newly hatched rotifers were fed until they started to reproduce and were then randomly divided in two groups of 7–10 rotifers each. One group, F, was checked daily for reproduction and survival to compile the life table sheet. The other cohort (S) did not receive food for 20 days, and thereafter was fed daily.

experiment d: neonates

This experiment tested the capacity of the rotifers to make use of maternally derived resources. In total, 17 newborn bdelloids were maintained without food for the first 20 days of their life. Subsequently their medium and food were renewed daily. S cohorts and F cohorts were studied in parallel.

data analysis

The mean longevity and mean fecundity of starved and fed cohorts were compared by one-way anova. To assess the effect of different foods on the recovery obtained in B and C experiments, numbers of individuals surviving starvation in S and F cohorts were compared using a contingency table and a Pearson's chi-squared test. Owing to low observed numbers in some cells of the contingency table, the P-value for the chi-squared test was calculated by random sampling from the set of all contingency tables that can be generated maintaining the given marginals using R (R Development Core Team 2004). The effect of starvation on the rotifer life history was assessed by correlation tests between age-specific fecundities of each S cohort and its paired control (F) cohort. The data of age-specific fecundity (m_x) of S cohort were matched with those of F by either including or excluding the days spent under starvation. The age-specific fecundity data used for the analyses were relative to the poststarvation time only.

Results

At 24 °C and regularly fed with powdered fish food, the control cohorts of Macrotrachela quadricornifera lived for about 30–40 days and produced 20–30 eggs.

In the ‘tolerance’ experiment (A) no bdelloid reproduced before removal of food. The 60 day starvation period produced the highest mortality and less than 10% of the individuals recovered (Table 1). On average, more than 60% and more than 90% recovered after 40 days and 20 days without food, respectively. After removal of food, rotifers commonly laid no more than one egg per individual before arresting reproduction for the whole starvation time (Table 1). When fed again, recovered rotifers resumed reproduction within 3 to 5 days: the longer the starvation, the longer the time needed to produce eggs. Survival of the cohorts was checked again 1 week later. In all experiments less than 10% of the recovered rotifers died during the first week of feeding.

Table 1.  Egg production (± SD) and recovery (%) after starvation of the bdelloid Macrotrachela quadricornifera: results of the experiments A, B, C and D

Experiment	Replicates	No. of rotifers	No. eggs/rotifer before starvation	No. eggs/rotifer during starvation	Days required to reproduce after refeeding	Recovery (%)
			X̄ (± SD)	X̄ (± SD)	X̄ (± SD)
60 days starvation
A	I	13	0	0·6 (± 0·6)	4 (± 0·0)	7·7
	II	13	0	0·8 (± 0·7)	–	0
	III	14	0	0·7 (± 0·6)	4 (± 0·0)	14·3
40 days starvation
A	I	13	0	0·5 (± 0·5)	4·5 (± 1·8)	53·8
	II	13	0	0·9 (± 0·9)	3·7 (± 0·5)	61·5
	III	14	0	0·6 (± 0·6)	3·5 (± 0·7)	78·6
20 days starvation
A	I	8	0	1·0 (± 0·5)	3·1 (± 0·6)	100
	II	8	0	0·9 (± 0·6)	3·5 (± 0·8)	100
	III	9	0	1·0 (± 0·5)	3·3 (± 0·9)	88·9
B	I	10	1·4 (± 0·7)	3·9 (± 1·2)	2 (± 0·0)	80
	II	10	1·7 (± 0·7)	3·2 (± 0·8)	1·8 (± 0·4)	70
	III	9	1·4 (± 1·0)	2·3 (± 0·7)	2·4 (± 0·5)	55·5
C	I	7	0·3 (± 0·5)	2·1 (± 0·6)	2 (± 0·0)	100
	II	7	0·1 (± 0·3)	2·2 (± 0·6)	2 (± 0·0)	85·7
	III	7	0·6 (± 0·5)	2·0 (± 0·5)	2 (± 0·0)	100
D	I	6	–	0	7 (± 2·7)	100
	II	6	–	0	5·2 (± 1·1)	83·3
	III	5	–	0	5·2 (± 1·1)	100

In the ‘life table’ experiments (B and C) two different types of food were supplied and this produced differences in the rotifers’ fecundity and longevity, which were both lower in experiment C in which the food was E. coli (Table 2). Starvation was induced in 4-day-old rotifers in experiment B and in 3-day-old rotifers in experiment C because they matured at different ages. In the first reproductive day, rotifers in experiment C produced 0·3 eggs, while those in experiment B produced five times as many eggs (Table 1). All were starved for 20 days and thereafter regularly fed. Recovery after starvation was significantly lower in experiment B than in experiment C (χ² = 5·25, P = 0·033, Table 1). Within each experiment, the mean fecundities of starved and fed cohorts did not differ (Table 2). Longevity differed between starved and fed groups, but the difference disappeared if the days of starvation were subtracted from the bdelloids’ age (Table 2). Age-specific rates of survival (l_x) and of fecundity (m_x) of the starved and fed rotifers are presented in Fig. 1. Both the m_x and l_x curves of the S cohorts follow the curves of their respective F controls when the 20-day period of starvation is disregarded. The age-specific fecundity rates (m_x) of the starved cohorts of both experiments were compared with those of the respective fed controls by a correlation test either including or excluding the 20-day starvation. Not surprisingly, the correlation between S and F cohorts was stronger when the starvation period was excluded (Table 3).

Table 2.  Fecundity and life span (X̄± SD) of Macrotrachela quadricornifera following recovery after 20 days of starvation. The rotifers were starved at maturity in experiments B and C, in which food sources differed. The rotifers of experiment D were starved at birth. Each parameter represents the mean of the three replicate experiments. Mean life span of S cohorts reports two values, corresponding to the life span either including (+20 days) or not including (−20 days) the 20 days of starvation

Experiment	B			C			D
Cohort	Starved	df 43	Fed	Starved	df 35	Fed	Starved	df 34	Fed
NS = non-significant; ***P < 0·001 at one-way anova. df = degrees of freedom.
Mean fecundity (egg per female) (± SD)	29·0 (± 0·7)	NS	28·9 (± 1·9)	17·7 (± 4·0)	NS	18·1 (± 3. 5)	13·6 (± 5·4)	***	21·5 (± 3·0)
+20 days starvation	62·3 (± 4·2)			51·0 (± 4·9)			54·8 (± 8·4)
Mean life span (days) (± SD)		***	44·3 (± 4·0)		***	32·4 (± 1·5)		***	35·4 (± 6·6)
−20 days starvation	42·3 (± 4·2)	NS		31·0 (± 4·9)	NS		34·8 (± 8·4)	NS

Figure 1. Age-specific fecundity rates (m_x, left y-axis) (open symbols) and age-specific survival rates (l_x, right y-axis) (solid symbols) of rotifers recovered after 20 days starvation starting on the first day of reproduction, and then fed a routine diet: ▵ fed cohort; □ starved cohort. Upper panel, rich diet (experiment B), lower panel, poor diet (experiment C).

Table 3.  Correlation coefficients for comparison of age-specific fecundity rates between each starved and fed cohort of experiments B, C and D

Experiment	Correlation coefficient (no. of pairs)
	+20 days starvation	−20 days starvation
NS = non-significant; ***P < 0·001. Results are shown either including (+20 days) or not including (−20 days) the 20 days of starvation.
B	0·54 NS (12)	0·76*** (27)
C	0·49 NS (13)	0·66*** (27)
D	0·15 NS (9)	0·55*** (29)

Total cessation of egg production did not occur until the third day after removal of food; rotifers in experiments B and C produced about three and two eggs per female, respectively, during this initial phase of starvation (Table 1). These eggs, not represented in Fig. 1, are included in the calculation of mean fecundity.

In the ‘neonate’ experiment (D), about 95% of the neonates survived starvation and recovered when fed (Table 1). These individuals never produced eggs during the period of starvation, and subsequent to the resupplying of food they required about 5 days to initiate reproduction, while the control F group became reproductive at the age of 4. The mean fecundity of the S cohorts was significantly lower than that of their F controls, while the life spans of the two groups were remarkably similar if the no-food period was excluded from the S cohort's age (Table 2). Disregarding the starvation time, the patterns of m_x and l_x of the S cohort were remarkably similar to those of the F control, and reproductive periods and lifetimes had similar durations (Fig. 2). Correlations between m_x values of S and F rotifers were highly significant (P < 0·001) when starvation time was disregarded (Table 3). The difference between fecundities of S and F cohorts was due to the early reproductive effort, which was much lower in the starved rotifers.

Figure 2. Age-specific fecundity rates (m_x, left y-axis) (open symbols) and age-specific survival rates (l_x, right y-axis) (solid symbols) of rotifers recovered after 20-day starvation starting at birth and then fed a routine diet: ▵ fed cohort; □ starved cohort.

Discussion

Macrotrachela quadricornifera, like other bdelloids, does not switch its resource utilization between growth and reproduction according to a ‘bang-bang’ strategy, but rather follows an ‘intermediate’ pattern by investing in both activities at the same time (Ricci 1995; Ricci & Fascio 1995). At the time of first reproduction, energy is required for oogenesis, body growth and physiological maintenance; thus any shortage of food at this time of life is likely to affect survival. On this basis, we decided to starve the rotifers when they were starting reproduction. Because bdelloid age at first reproduction appeared to differ with the food source, starvation initiation varied between the 3rd and 4th day of life in different experiments.

Not surprisingly the longer the starvation time, the lower the recovery rate of M. quadricornifera. A few rotifers survived starvation that lasted longer than their average lifetime, about 35–40 days, but more than 60% survived a starvation lasting a period equalling their life span (Table 1, experiment A). In the week following the end of starvation few bdelloids died, and thus most individuals recovered from the stress of prolonged starvation. When the regular food supply was restored, the recovered rotifers resumed reproduction and produced viable eggs.

To our knowledge, such remarkable tolerance of starvation is uncommon among freshwater invertebrates, like monogonont rotifers or crustaceans (Tessier et al. 1983; Kirk 1997a). The only exception is the monogonont rotifer Brachionus plicatilis, which, if starved when prereproductive, extends its life span by as long as 16 days (Yoshinaga et al. 2003a). Survival of starvation can be enhanced if stored resources are directed to maintenance and other energy expenditures are reduced or eliminated. Monogonont rotifers (Kirk 1997a) and Daphnia (Tessier et al. 1983) were found to survive longer periods of starvation if resources were not committed to reproduction. However, if they had already initiated reproduction, the stress of starvation rapidly led to death. The greater investment into reproduction prior to starvation could be the reason why fewer bdelloids recovered after 20-day starvation in experiment B than in experiments A or C (Table 1). In experiment B, the rotifers matured 1 day later, laid about 1·5 eggs per female on the first day of reproduction before experiencing starvation, continued reproduction in absence of food, producing approximately three eggs each, and only 70% rotifers survived starvation. By contrast, more than 90% of the rotifers survived in experiments A and C, when each bdelloid produced less than 0·3 egg before and two eggs during starvation. It seems that resources already ‘canalized’ to reproduction could not be recovered and used for maintenance with the result that survival decreased. This would be an interesting avenue for further study.

‘Life table’ experiments (B, C) were planned to test the hypothesis that a more nutritious diet enhances survival and subsequent recovery from starvation, but our results were inconclusive. The ‘poor’ diet (exp. C) produced significantly higher recovery after starvation than the ‘rich’ diet (exp. B), but the result may have been produced by the low early egg production of the rotifers fed the ‘poor’ food.

In the ‘neonate’ experiment, the use of stored resources during starvation was prevented by starving neonates that had never been fed, and whose only resource was the energy derived from the egg reserves. Similar experiments have been conducted with monogononts. Starved newborns of the monogonont B. calyciflorus were able to survive for a few days depending on the amount of lipid reserves in the animal's body (Gilbert 2004) and Synchaeta pectinata resisted starvation longer if hatched from larger eggs (Kirk 1997b). Similarly, cladoceran (Daphnia) newborns can survive starvation for about 8–10 days (Tessier et al. 1983). However the starvation times of the monogonont species were up to 3 days, as was the starvation time of Brachionus plicatilis (Yoshinaga et al. 2003a). By contrast, as many as 95% of bdelloid neonates survived starvation for 20 days and, when subsequently fed, matured and produced viable eggs. Starved neonates matured at a later age, but required the same number of days of feeding to become reproductive, and their life spans were similar to those of the F controls if the starvation period is disregarded. Only their early age-specific fecundity rates were lower than those of the fed controls.

In conclusion, all bdelloids that recovered from starvation at birth or at maturity reproduced soon after receiving food and had extended life spans. Animals that were starved when reproductive had an average fecundity as high as that of fed controls, and food quality did not affect their poststarvation life-history traits. Age-specific fecundity rate of starved and fed rotifers were quite similar: S and F bdelloids had similar reproductive periods as well as mean life spans, excluding starvation duration. Only continued reproduction into the early phase of the starvation period decreased the bdelloids’ rate of recovery, an equivalent but much weaker effect than found in other freshwater animals (Kirk 1997a; Kirk, Ellis & Taylor 1999).

How can bdelloids tolerate long periods of starvation and resume life unaffected once food is present again? If animals do not receive resources, the only way to survive is to reduce expenditure. This can only be accomplished by depression of metabolic rates, a common strategy of animals with resting stages in the face of environmental stresses (Storey 1988). When starved, a number of freshwater animals are capable of reducing their respiration rate by about one-half (see Kirkwood 2005). Under severe prolonged starvation the respiratory rate of M. quadricornifera is almost undetectable (Santo and Ramløv, personal communication).

There is an interesting parallel between the response of bdelloids to starvation and anhydrobiosis. Recovery of bdelloids from anhydrobiosis is dependent upon the duration of the harsh period, and occurs with no ‘cost’ imposed on the life-history characteristics of survivorship and fecundity rates (Ricci et al. 1987; Ricci 2001). The ‘behaviour’ of M. quadricornifera after long starvation times thus resembles its response after anhydrobiosis. However, while typical bdelloid habitats such as terrestrial soils, mosses and lichens are prone to extended desiccation, it is unlikely that food resources, such as bacteria, algae or organic particulate matter, would be absent from a non-desiccated habitat for a comparable period. Since long-lasting starvation can hardly be a natural condition to which bdelloids are adapted, but evokes a reaction that is similar to dormancy, we suggest that the cue related to starvation is perceived as a dormancy inducer, and that the same processes used to ‘shut down’ vital processes and enter anhydrobiosis are followed. Food quantity and quality are common environmental cues that evoke production of dormant resting eggs in monogonont rotifers (Pourriot & Snell 1983); food shortage induces the production of ‘diapausing’ amictic eggs of Synchaeta pectinata (Gilbert 1995). Similarly, the dauer larva of the nematode Caenorhabditis elegans forms in response to several factors, among which is a limited food supply (Riddle & Albert 1997). However, whether other animals capable of anhydrobiosis, such as tardigrades or nematodes, also show substantial abilities to survive starvation is totally unexplored. Therefore, whether the remarkable tolerance of starvation is related to the capacity of dormancy common to all anhydrobionts, or is peculiar to bdelloid rotifers remains to be seen. The ability to shut down normal metabolic functions for extended periods might also contribute to the long-term survival of these fascinating animals (Mark Welch & Meselson 2000).

Acknowledgements

Giulio Melone, Diego Fontaneto, Timothy Barraclough and Manuel Serra substantially improved this manuscript with valuable comments and suggestions. This paper benefits from the critical help of Charles E. King, who read, corrected and discussed an earlier draft. Charles E. King and Timothy Barraclough greatly helped us with the use of English. The study was supported by ASI grant to C.R.