SEARCH

SEARCH BY CITATION

Keywords:

  • cold-stress resistance;
  • correlated responses;
  • Drosophila melanogaster;
  • gene-by-temperature interaction;
  • longevity;
  • sex-specificity

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Thermal environments can influence many fitness-related traits including life span. Here, we assess whether longevity in Drosophila melanogaster can experimentally evolve as a correlated response to cold-stress selection, and whether genotype-by-temperature and sex-by-temperature interactions are significant components of variation in life span. Three replicated S lines were cold-stress selected and compared with their respective unselected controls (Clines) in the 16th generation of thermal selection. Cold-stress resistance exhibited a substantial direct response to selection, and also showed a significant interaction between sex and type of line. Mean longevity exhibited a significant interaction between adult test temperature (14 and 25 °C) and line (with suggestive evidence for increased longevity of S lines when tested at 14 °C), but there was no evidence for increased longevity in S lines at normal temperatures (i.e. 25 °C). Another temperature-dependent effect was sex-specific, with males being the longer lived sex at 25 °C but the less long-lived sex at 14 °C. Additionally, we tested in an exploratory way the relationship between longevity and cold-stress resistance by also measuring resistance to a prefreezing temperature before and after one generation of longevity selection at 14 °C (selection intensity, i = 1.47 for S lines, and 1.42 for C lines). In this longevity selection, we found that cold-stress resistance increased by about 6% in S lines and 18% in C lines. However, taken together, the results indicate no simple relationship between longevity and cold-stress resistance, with genotype-by-sex interactions in both traits. Temperature dependent interaction in longevity is apparent between S and C lines, and sex-specific variation in mean longevity also depends on temperature.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Temperature stress is one of the most common in terrestrial environments. It induces phenotypic, sex-specific, and potentially, evolutionary responses not only on the thermotolerance level itself but also other traits, including life-history traits (e.g. Hoffmann & Parsons, 1991; Khazaeli et al., 1997; Loeschcke et al., 1997; Gibert et al., 2001). In addition, tolerance to thermal stress might be affected not only by thermal selection but also by selection on life-history traits (Luckinbill, 1998).

One theory on the evolution of longevity claims that stress tolerance is causally related to extended life span (e.g. Lithgow & Kirkwood, 1996; references therein; Luckinbill, 1998). Support for this includes the finding of stress-resistant mutations that extend the life span of the worm Caenorhabditis elegans (Kenyon et al., 1993; Lithgow et al., 1994, 1995; Larsen et al., 1995). Positive relationships between heat-shock tolerance and longevity have also been reported for Drosophila. For example, brief exposure to elevated but nonlethal levels of heat increased the life span of Drosophila melanogaster at normal temperatures, independently of the suppression of reproductive activity (Khazaeli et al., 1997). Increased production of a heat-shock protein (Hsp70) extended adult longevity of transgenic strains of the same species at normal temperatures (Tatar et al., 1997; but see Minois et al., 2001). However, genetic variation in resistance to multiple stresses may not necessarily be associated to longevity (Harshman et al., 1999).

The relationship between cold-stress resistance and longevity has been less explored than relationships between heat-stress resistance and longevity. Recently, however, Luckinbill (1998) found that artificial selection for increased longevity confers resistance to low-temperature stress in laboratory D. melanogaster populations. One central question with respect to evolutionary responses to cold stress, however, remains: does selection for resistance to cold-stress increase life span at normal temperatures?

In the experiments described here we investigate longevity variation between lines of high and low resistance to cold stress in D. melanogaster at two adult test temperatures, 14 and 25 °C. Three replicated lines were cold-stress selected and compared with their respective control lines in the 16th generation of thermal selection. Provided that cold-stress resistance successfully diverged between selected and control lines, two main questions are addressed: (1) Does longevity increase by cold-stress selection, as expected if life span is both genetically and positively correlated with cold-stress resistance? (2) Is longevity a temperature-dependent trait whose differential expression before and after cold-stress selection can be modulated by interactions with temperature (involving genotype-by-temperature and sex–by-temperature interactions)? The last is particularly contentious with regard to the possible effects of temperature on longevity because genotype-by-environment and sex-by-environment interactions are thought to be mechanisms maintaining genetic variation in life span (e.g. Vieira et al., 2000). Additionally, we have tested in an exploratory way the relationship between longevity and cold-stress tolerance by also measuring cold-stress resistance before and after one generation of longevity selection at 14 °C. Although cold-stress resistance increased by this longevity selection, we found that longevity when tested at normal temperatures (e.g. 25 °C as opposed to 14 °C) does not increase by selecting for cold-stress resistance. For mean longevity we found significant interactions between adult test temperature and cold-stress selection as well as between sex and temperature.

Cold-selected lines and their controls

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Flies originated from 19 inseminated females collected inHarjavalta, Finland (22°61N) in August 1996. The progeny of these females were pooled into a large mass population from which two sets of thermal regimes were initiated for this study. Three control lines, denoting C1, C2 and C3, were kept at constant 25 °C at a 12-h light : 12-h dark cycle. There other lines were subjected to a cold-stress selection regime, denoted S1, S2 and S3. These flies were cold stressed at an age of 4 ± 1 days by exposing them to 0 °C (20 vials per line, 40 flies per vial, separately for each sex). Selection intensity depends on time of exposure (generally between 3 and 4 days in this study) at this temperature. To avoid any possible cross-generation effects (e.g. Sgrò & Hoffmann, 1998), selection was performed every other generation. Throughout this experiment, mean selection intensity (in phenotypic units of SD) was about 0.55 per generation of selection, as a value averaged across lines, sexes and generations. For example, survival after stress treatment was 70–80% in the final generation (G16) of selection, but was less than 55% in G1. In case of higher mortalities in some of the first seven selection generations, a few males from a back-up selection regime (3–5-day-old males exposed to 0 °C but only for 18 h at an age of 4 ± 1 days) were added as parents of the next generation to avoid very low effective population size Ne. The selection lines used here had thus been cold selected for 16 generations (i.e. 32 generations from the start of thermal selection) before the start of the longevity experiment.

Measurement and analyses of longevity

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

First instar larvae were collected from each Si and Ci line (10 bottles per line, 10 pairs per bottle), and placed at a density of 35 in 95 × 20-mm shell vials containing 6 mL of culture medium (hereafter referred to as standard vials). All cultures were kept at 25 ± 1 °C. The enclosing adults were simultaneously collected from all lines, sexed under light CO2 anaesthesia, and used for measurement of longevity. Separately for each replicate line, 10 vials each containing 10 males and 10 females at 1-day of age were set up simultaneously at each of two adult test temperatures, 14 and 25 °C. The flies were transferred to fresh vials every 2 days at 25 °C and 4 days at 14 °C, when vials were examined for deaths. The number of vials was gradually reduced as deaths occurred, with surviving adults being kept at a density as close to 20 per vial as possible. On the whole, longevity was scored for 1200 individuals at 25 °C and, as a result of escaped flies, 1108 individuals at 14 °C, with an interval of 2 days (25 °C) or 4 d (14 °C) for each successive scoring.

For analysis, longevity (days) was ln-transformed to remove dependence of variances on means. A multiway, partially hierarchical anova, with the fixed factors being adult test temperature, sex and stress selection treatment (i.e. S vs. C lines) was applied (e.g. Zwaan et al., 1995; Reeve et al., 2000). The random factor was replicates nested within stress-selection treatment. However, there are restrictions on randomization of all factorial treatment combinations within the experiment. That is, adult test temperature and sex are subplots because thesefactors were randomized only over a whole-plot (Snedecor & Cochran, 1989, p. 369), with two whole-plot treatments (S vs. C) and three replicates within each of them (F.James Rohlf, pers. comm.). In split-plot designs there are two error terms: a whole-plot error a, and a subplot error b. Stress is tested over error a (i.e. replicate lines with 4 d.f.). Adult test temperature, sex, and interactions are tested over error b (12 d.f.). Therefore, only the split-plot anova is presented, but the results obtained by applying the above mentioned, partially hierarchical, block-randomized anova were similar to the split-plot results.

Longevity was also analysed with a nonparametric (log-rank) test, which uses the observed and expected death from each sampling interval to calculate a χ2 statistics (Miller, 1981; Partridge et al., 1995). In this case, tests were made by pairing Si and Ci replicates according to the line numbers that were assigned arbitrarily to them at the start of the experiment, with all three P-values being then combined (Roper et al., 1993; Partridge et al., 1995; see Sokal & Rohlf, 1981, p. 779). When the differences were opposite in sign, the overall P-value reported was obtained by using the distribution of chi (Roper et al., 1993), and corrected for multiple comparisons (Rice, 1989).

All statistics, including sums of squares for split-plot anova from GLM procedures were computed using the STATISTICA package (StatSoft, 1999).

Samples examined for stress resistance

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Survival under cold stress was examined in both the F1 of a sample of young flies (i.e. representing the preselective sample, because at that stage one did not know if they were short- or long-lived individuals), and F2 from long-lived flies. One generation of truncation selection on longevity is the main difference between these samples. This was carried out simultaneously for all Si and Ci lines. Experimental individuals for each sample were reared under standardized conditions of larval density (35 larvae per vial at 25 ± 1 °C) using the same batch of culture medium. Two kinds of samples were obtained as follows:

  • 1
    F1 from young flies . To obtain this sample, 10 vials each containing five young flies (4–7 days old) of each sex were set up to collect 0–2-h-old larvae from each line to get offspring under standardized conditions.
  • 2
    F2 from 14 °C long-lived flies. To reduce any possible cross-generation effects of parental acclimation influencing cold-stress resistance in F1 ( Watson & Hoffmann, 1995 ), sample two was the F2 from long-lived flies at 14 °C. Following the protocol described above, aged flies were transferred to fresh vials every 4 days. The vacated vials corresponding to surviving adults of ages equal or later than 170 days were stored to get offspring. For each line, flies were pooled in a F1-generation (60–70 flies per replicate line) by using similar numbers of F1 flies (6–9) from each successive (4 days)-interval of sampling. These flies were transferred to new vials tocollect 0–2 h-old larvae, obtaining the F2 from long-lived flies under standardized conditions. Selection intensity (i) was estimated as in Norry & Loeschcke (2002 ) .

Measurement and analysis of cold-stress resistance

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Pilot studies were performed and a prefreezing temperature of 3.3 °C for 120 h was chosen for the assessment of mortalities. All experimental flies were aged between 2 and 3.5 days and manipulation was carried out without anaesthesia. Cold resistance was measured using shell vials with culture medium in a horizontal position. The number of survivors to stress was scored (at 25 °C) as the number of flies that could walk 24 h after the cold treatment.

The test was performed in a block for comparison of sample 1 with sample 2, with 10 replicate vials per Si and Ci line (excepting sample 2 of line C2, where the number of available flies restricted replication to seven vials) and 20 flies per vial, 10 males plus 10 females. For analysis, the number of survivors in each vial (our unit of replication) was expressed as proportion alive, to which the arcsine transformation was applied (Luckinbill, 1998). Stress-selection treatment (C vs. S) was tested with anova on F1 of young flies, using stress selection and sex as fixed factors and replicate as a random factor nested within stress-selection treatment.

To test for effects of longevity selection, the same restrictions on randomization of all factorial treatment combinations within the experiment apply also here as for the above described longevity analysis. Therefore, split-plot anova was performed. That is, stress-selection treatment is a whole-plot treatment, and longevity selection and sex are subplot treatments.

Longevity in cold-selected lines vs. their controls, and its sex-specific variation

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Maximal life span is about three times higher at 14 °C than at 25 °C, and its value is similar both between lines and between sexes, particularly at 14 °C (Fig. 1).

image

Figure 1. Survival curves for males and females at 14 and 25 °C. Solid lines are cold stress selected lines (S), dashed lines are control lines(C), bold lines are females, and thin lines are males. Each curve is the average of three replicated lines. Survival curves for each replicate line are shown elsewhere ( Norry & Loeschcke, 2002 ).

Download figure to PowerPoint

However, a temperature-dependent sex effect is observed for mean longevity, with males being, in all lines, the longer lived sex at 25 °C but the less long-lived sex at 14 °C (Table 1; Fig. 1). The sex-by-temperature interaction is highly significant (Table 2). Identical conclusions are achieved with partial anova for each temperature (i.e. using sex and stress selection treatment as fixed factors, and replicates within treatment as a random factor), with females living longer at 14 °C (MS = 1.098, F1,4 = 9.24, P < 0.05), and males living longer at 25 °C (MS = 1.026, F1,4 = 7.70, P < 0.05; it should be noted that this is a conservative test for sex because of restrictions on randomization of all factors within the experiment).

Table 1.  Mean longevity (±SD, in ln-days) ofmales and females from cold stress selected(S i ) and control (C i ) lines at two temperatures. Bold typeface indicates marginal mean values. P -values were first combined from log-rank tests and finally corrected for multiple comparisons within each temperature (i.e. two comparisons).
LineA. 25 °C experimentB. 14 °C experiment
MalesFemalesMalesFemales
C13.918 (0.245)3.794 (0.343)4.987 (0.332)5.074 (0.197)
C23.945 (0.339)3.896 (0.315)5.041 (0.384)5.047 (0.392)
C33.937 (0.353)3.911 (0.360)5.022 (0.467)5.108 (0.330)
C3.933 (0.315)3.867 (0.343)5.006 (0.438)5.076 (0.317)
S13.933 (0.488)3.905 (0.416)5.079 (0.386)5.171 (0.244)
S23.862 (0.397)3.847 (0.320)5.083 (0.328)5.145 (0.315)
S33.844 (0.388)3.733 (0.332)4.985 (0.395)4.996 (0.471)
S3.880 (0.427)3.828 (0.364)5.050 (0.373)5.102 (0.366)
C, S3.906 (0.371)3.848 (0.353)5.027 (0.405)5.089 (0.341)
S < C, n.s.S < C, n.s.S > C, n.s.S > C, P < 0.05
Table 2.  Split-plot anova on ln(longevity) performed to test for effects of (1) cold-stress selection treatment (i.e. S vs. C lines), (2) replicates within selection treatment, (3) adult test temperature (i.e. 14 vs. 25 °C) and (4) sex, and their (1) × (3) and (3) × (4) putative interactions. Bold p -values are significant.
Source of variationd.f.MSFP -value
Whole-plot
(1) Stress selection   1  0.0174   0.020.9054
(2) Error a[replicates within (1)]   4  1.0866  
 Within2284  0.1354  
Subplot
(3) Adult test temperature   1802.84737636.46<0.0001
(4) Sex   1  0.0030   0.030.8699
Interactions
(1) × (3)   1  0.9576   9.110.0107
(1) × (4)   1  0.0001   0.010.9260
(3) × (4)   1  2.1246  20.210.0007
(1) × (3) × (4)   1  0.0334   0.310.5879
Error b  12  0.1049  

Although there is considerable variation among the replicates within one stress-selection treatment, the interaction between stress-selection treatment (i.e. S vs. C) and adult test temperature is significant (Table 2). Inspection of mean longevity as well as survival curves indicates that Si lines tend to live longer than Ci lines at 14 °C, with an opposite trend being observed at 25 °C (Table 1; Fig. 1). However, perhaps because of the low statistical power of the test, no effect of cold-stress selection was significant in a partial anova for each temperature (MS = 0.344, F1,4 = 0.60 at 14 °C; MS = 0.643, F1,4 = 1.12 at 25 °C; fixed factors for each anova are sex and stress-selection treatment, whereas replicates within treatment is a random factor). The log-rank test suggested a significant difference between S and C lines, but only in females at 14 °C (Table 1).

Cold-stress resistance: the correlated response to longevity selection

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

The lines have substantially diverged in resistance to cold stress after 16 generations of selection on this trait, with S lines being, in average, about 25% more tolerant to our experimental conditions of cold stress than C lines. This value of 25% of between-line divergence was obtained by comparing the untransformed trait means for F1s of young flies, averaging it between results shown in Table 3 (i.e. 23% of divergence for untransformed proportion alive) and results from an independent replicate of the same test, where the estimated mean divergence was 27.42% (MS = 0.928; F1,4 = 32.59; P < 0.005; mean values and SD not shown). The effect of cold-stress selection was significant in anova(Table 4). There was also a significant sex-by-line interaction for this trait (Table 4). The effect of cold-stress selection was also significant in partial anovas for each sex (MS = 0.253, F1,4 = 9.18, P < 0.05, for males; MS = 0.441; F1,4 = 17.08, P < 0.05, for females).

Table 3.  Mean survival (±SD, in arcsine-transformed proportion alive) to a cold stress (120 h at 3.3 °C) for cold-selected (S i ) and control lines (C i ). Results for F1 of young flies and F2 from 14 °C long-lived flies are shown. Bold typeface indicates marginal mean values for lines
 F1 of young fliesF2 of long-lived flies
LineMalesFemalesMalesFemales
C10.522 (0.089)0.538 (0.072)0.606 (0.145)0.612 (0.112)
C20.631 (0.166)0.637 (0.124)0.675 (0.133)0.669 (0.076)
C30.495 (0.079)0.493 (0.072)0.748 (0.158)0.718 (0.109)
C0.549 (0.111)0.556 (0.089)0.676 (0.145)0.666 (0.099)
S10.728 (0.172)0.695 (0.126)0.722 (0.095)0.675 (0.069)
S20.713 (0.156)0.675 (0.115)0.820 (0.061)0.751 (0.068)
S30.728 (0.172)0.687 (0.129)0.793 (0.050)0.748 (0.071)
S0.723 (0.167)0.686 (0.123)0.778 (0.069)0.725 (0.069)
C, S0.636 (0.133)0.621 (0.106)0.727 (0.107)0.695 (0.084)
Table 4. anova s on arcsine-transformed proportion alive after a cold stress (120 h at 3.3 °C). In the block-randomized anova (A), the analysis is performed only on F1 of young flies to test for a direct response to cold-stress selection. In split-plot anova (B), the analysis is performed on data combined over F1 of young flies and F2 of long-lived flies, with (1) being stress selection treatment (i. e. S vs. C lines), (2) replicate lines within (1), (3) sex and (4) longevity selection treatment (i.e. F1 of young flies vs. F2 of long-lived flies). Bold p -values are significant.
Source of variationd.f.MSFP
(A) Analysis of the direct response to cold-stress selection
(1) Stress selection10.689012.840.0231
(2) Replicates within (1)40.05373.270.0143
(3) Sex10.007328.630.0058
Interactions
(1) × (3)10.014055.100.0017
(2) × (3)40.00020.010.9995
Within1080.0164  
(B) Analysis of longevity selection, and its interactions
Whole-plot
(1) Stress selection10.780419.690.0114
(2) Error a[replicates within (1)]40.0396  
  Within2100.0134  
Subplot
(3) Sex10.03241.390.2612
(4) Longevity selection10.398617.190.0014
Interactions
(1) × (3)10.02711.170.3007
(1) × (4)10.07303.170.1003
(3) × (4)10.00380.160.6962
(1) × (3) × (4)10.00010.040.8448
Error b120.0229  

The increased resistance to cold stress in S lines is also consistent in samples derived from long-lived individuals ( Tables 3 and 4). However, an important difference is found between samples before and after longevity selection. Specifically, survival to cold stress tends to be higher in post-selective samples (i.e. F2 from long-lived flies) than in F1 of young flies (Table 3). This effect of longevity selection is highly significant in our split-plot design (Table 4). There is a marginally significant interaction between stress-selection treatment and the longevity-selection subplot treatment (Table 4).

Longevity in cold-selected lines and their controls

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

Like most poikilotherms, D. melanogaster has a shorter life at higher temperatures (Table 1). Variation in life span is partially attributable to both genetic and environmental effects (Tower, 1996; Vieira et al., 2000). Longevity at two very different temperatures did not significantly differ between S and C lines when tested by relatively conservative anova models in this study (but see Table 1 for females at 14 °C), but the interaction between adult temperature and cold-stress selection was significant (Table 2). Therefore, we cannot conclude that cold-stress selection has no impact on longevity at lower temperatures, but an inspection of mean longevity indicated that this form of stress selection cannot increase adult life span at a moderate temperature (i.e. 25 °C; Table 1). Furthermore, although larger sample sizes are required to test for differences in age-specific mortality rate (see Driver, 2001; for caution against the use of the Gompertz function), S males at 25 °C tend to show lower survival than C males in the first 55 days of adult life (Fig. 1), an age equivalent to mean longevity at this temperature.

In a recent study on quantitative trait loci (QTL) affecting life span in D. melanogaster, Vieira et al. (2000) observed that the only significant genetic variation in their analysis pooled over several environments (starvation, low, control and high temperatures) and over sexes appeared in the gene-by-environment and gene-by-sex interaction terms. A trend of genotype-by-temperature interaction, with relatively extended longevity of cold-selected lines when tested at low as opposed to moderate temperatures, was suggested by some previous observations. In D. melanogaster, for instance, a consistent trend was observed in long-term (about 4 years) laboratory cultures that were continually maintained at two verydifferent thermal environments: low temperature (16.5 °C) and moderate temperature (25 °C) (Partridge et al., 1995). In this case, flies from each selection regime (i.e. culture temperature) showed higher longevity than flies from the other selection regime when they were tested at the same temperature at which their ancestors had evolved (Partridge et al., 1995). Whereas this trend may well be indicative of longevity adaptation to temperature, adaptation could also be the result of either trade-offs between the relative performance at different temperatures and/or a consequence of mutation pressure (Partridge et al., 1995). Although longevity at each of our two adult test temperatures was not significantly different between S and C lines, the interaction between adult temperature and cold selection was significant and in the same direction as in the Partridge et al.'s (1995) study. Both S and C populations used in the present study were continuously maintained at 25 °C (with cold-selected lines being obtained by exposure to a potentially lethal low temperature for no longer than 4 days of adult selection), i.e. thus discarding the possibility that mutation accumulation in Si lines reduces adaptation to (intermediate) temperatures that the organism no longer encounters. Overall, therefore, there is no evidence that cold-stress selection per se can increase life span at moderate temperatures.

Cold-stress resistance: the apparent correlated response to longevity selection

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

As noted by Promislow & Bugbee (2000), with selection regimes like the present, the more successful individuals will be those that can survive to late ages and, having survived, are the most fecund. Although only one generation of longevity selection was performed, it should be noted that environmental variation in our experiment was reduced as much as possible (i) by rearing experimental individuals in standardized conditions, (ii) by transferring adults to fresh vials every 4 days at 14 °C and (iii) by holding a fly density within vials as constant as possible. The results indicate that F2 from long-lived flies at 14 °C are more tolerant to cold stress than F1 of young flies (Tables 3 and 4). This positive relationship between longevity and resistance to cold stress was also significant in an independent replicate of this experiment when longevity selection was performed at 25 °C (results not shown). The results imply therefore that longevity selection can increase resistance to cold stress in D. melanogaster. This is consistent with the recent study by Luckinbill (1998), where artificial selection for increased longevity conferred resistance to low-temperature stress in the same species. Whereas this finding suggests that the traits are genetically correlated, we cannot reject the hypothesis that longevity selection resulted also in a purging of accumulated mutations with side-effects on cold-stress resistance. This could be a case in our C lines because the correlated response to only one generation of longevity selection was relatively high (about 18%) when compared with the direct response over 16 generations of cold-stress selection (about 25%, as estimated in the way indicated above). If so, it should be noted that the cold-stress test was performed on a very early age of the adult life, so that any possible accumulations of late-acting, age-specific mutations would be an issue here (Promislow & Tatar, 1998). Previous studies suggest also that longevity and physiology of resistance to cold stress are not genetically independent. For example, Luckinbill (1998) observed that long-lived populations also had significantly higher in vitro levels of glycerol, a cryoprotectant metabolite produced from glycogen. Whether cold-stress selection also affects the in vivo levels of this cryoprotectant metabolite is, however, unknown.

In an independent experiment of single-generation longevity selection, using a relatively high selection intensity with replicated lines as in the present study (1.28 ≤ i ≤ 1.36 at 25 °C), we found that the heat-induced expression of Hsp70 is strongly down-regulated under a weak heat-shock in F2s of long-lived flies (i.e. after one generation of strong selection for extended life), whereas 25 generations of heat-stress selection increased mean longevity by about 13% in males at 25 °C (V. Loeschcke & F. Norry, unpublished results). Presumably, short-lived flies are very susceptible to stress in terms of Hsp70 induction, whereas long-lived individuals remain more homeotic in the presence of stress because they do not need emergency defences at the same degree than short-lived individuals. Future work in this direction might reveal some mechanism of longevity selection involving Hsp70 regulation as well as variation among stress traits in their possible correlated responses to longevity selection.

Sex-specific variation in longevity

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

A temperature-by-sex interaction was significant in the present study, with males living longer than females at 25 °C and females living longer than males at 14 °C. This sex-specific effect of temperature was substantial for mean life span in all of the Si and Ci populations (Table 1), but not for maximal life span (Fig. 1). Sexual dimorphism in life span has not been extensively studied in Drosophila, but a demographic analysis suggested that males have higher age-independent but lower age-dependent mortality rates relative to females at moderate temperatures (Zwaan, 1999; see also Zwaan et al., 1995). Perhaps, the sex-by-temperature interaction we detected could partially be related to the level of reproductive activity at each temperature (if mating and ¤or egg-laying activity is lower at 14 °C than at 25 °C, although offspring from females at 14 °C can be verified at very late ages as in the present study). However, life span at 24 °C was also higher in males than in females in both a single-sex environment and a mixed-sex environment in highly inbred lines of D. melanogaster (Khazaeli et al., 1997), suggesting that the between-sex difference in life span is at least partially independent of the suppression of reproductive activity. Sexual dimorphism in life span may be population-specific and, as noted by Nuzhdin et al. (1997), most information on this topic in Drosophila comes from experiments of longevity selection in both sexes, which increases the between-sex genetic correlation. A minimum of 17 QTLs were detected for longevity in D. melanogaster (Nuzhdin et al., 1997; Curtsinger et al., 1998), all of which were sex- and/or environment-specific, and more than 50% of these had sexually antagonistic or antagonistic pleiotropic effects in different environments (Vieira et al., 2000). Furthermore, in the cactophilic D. buzzatii, second chromosome inversions exhibit also sex-specific (2st, 2jz3) and/or sexually antagonistic (2j) effects on longevity in nature (Norry et al., 1995; Rodriguez et al., 1999). Gene-by-sex and environment interactions are thought to be one of the mechanisms maintaining genetic variation in life span (e.g. Vieira et al., 2000). The present results show that sex-by-temperature interaction is a component of variation in longevity.

Summary of correlated selection responses

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

In summary, truncation selection on longevity can increase resistance to cold stress (Luckinbill, 1998; present results). However, an important conclusion in the present study is that longevity at 25 °C does not increase by cold-stress selection. Sixteen generations of selection for increased resistance to cold stress (0.35 < i ≤ 0.65, per generation of cold selection, with a net response of 25% as measured over 16 generations of selection to a prefreezing temperature) were unsuccessful in producing an increase in longevity at 25 °C (i.e. S lines tended even to survive relatively poorly at 25 °C, as opposed to 14 °C, when compared with C lines). However, only one generation of truncation selection on longevity (selection intensity, i = 1.47 for S lines at 14 °C, 1.42 for C lines at 14 °C, as estimated in Norry & Loeschcke (2002) improved cold-stress resistance by about 6% in S lines, and 18% in C lines (values obtained by comparing untransformed trait means before and after longevity selection). In addition, and consistent with the present study, Luckinbill (1998) showed that longevity selection also increases resistance to both prefreezing and freezing temperatures in adult D. melanogaster. If asymmetrical correlated responses between traits of stress resistance and longevity are apparent as in a number of previous studies, these could arise from physiological complexity (discussed in Shiotsugu et al., 1997). Correlated responses to selection depend not only on the genetic covariance between traits but also on other factors such as population size, number of loci and the details of pleiotropy (Gromko, 1995; references therein; see also van Tienderen & De Jong, 1994). In addition, stress might alter the latter (Krebs & Loeschcke, 1999). QTL analysis in D. melanogaster revealed that longevity QTL not only exhibit sex- and environment-specific genetic effects but also reversal in the signs of effects across sexes and environments (Vieira et al., 2000). Alternatively, asymmetrical correlated selection responses could be the result of at least one Mendelian factor with large effect on cold-stress resistance but small effect on longevity. Clearly, identifying the genes responsible for joint variation in these two traits is a necessary step to address these alternative hypotheses.

For mean longevity, we found a very consistent sex-by-temperature interaction, as well as a significant term of interaction between adult test temperature and cold-stress selection in adults. These two observations are consistent with the notion, strongly emphasized by Vieira et al. (2000), that genotype-by-temperature interaction is one of the general properties of this life-history trait.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References

We thank Trine Gammelgaard and Doth Andersen for excellent technical assistance, Stuart Barker, Miriam Hercus and Jesper Dahlgaard for helpful comments on the manuscript, and to F. James Rohlf for clarifying us the split-plot design in this study. We thank the reviewers for their thought-provoking comments. The Institute for Advanced Study and the Center for Environmental Stress and Adaptation Research at La Trobe University are greatly acknowledged for their hospitality to VL. This research was made possible by grants from the Danish Natural Sciences Research Council (No 9700177 and 9701412) to V.L., which partially supportedF.M.N.′s stay in Denmark. F.M.N.′s stay was alsosupported by a postdoctoral fellowship from ConsejoNacional de Investigación Científica y Técnica (Argentina).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Cold-selected lines and their controls
  6. Measurement and analyses of longevity
  7. Samples examined for stress resistance
  8. Measurement and analysis of cold-stress resistance
  9. Results
  10. Longevity in cold-selected lines vs. their controls, and its sex-specific variation
  11. Cold-stress resistance: the correlated response to longevity selection
  12. Discussion
  13. Longevity in cold-selected lines and their controls
  14. Cold-stress resistance: the apparent correlated response to longevity selection
  15. Sex-specific variation in longevity
  16. Summary of correlated selection responses
  17. Acknowledgements
  18. References