Asymmetry in thermal tolerance trade-offs between the B and Q sibling species of Bemisia tabaci (Hemiptera: Aleyrodidae)

Authors


Shai Morin, Hebrew University of Jerusalem, Department of Entomology, Herzel 3, 76100 Rehovot, Israel.
Tel.: +972 8 9489365; fax: +972 8 9466768; e-mail: morin@agri.huji.ac.il

Abstract

We investigated life history trade-offs related to thermal tolerance in two sibling species, commonly referred to as the B and Q biotypes, of Bemisia tabaci. We focused on basal resistance to short unpredicted heat stress, which reflects the organism investment, during both optimal and stressful conditions, in insuring its survival. At 27 °C, the relative reproductive performance of B was seven-fold higher than Q. After short stress of 42 °C, these differences increased to 23-fold. A turnover took place after short stress of 43 and 45 °C, in which Q adults performed better. As the expression of the analysed Hsp70 and other stress-related genes was found to be higher in the Q species, our data likely reflects two different strategies for optimal performance. B lowers soma protection for achieving maximum reproduction (‘direct inhibitory’ trade-off model), whereas Q invests significant resources in being always ‘ready’ for a challenge.

Introduction

High temperatures are one of the main abiotic selection pressures that ectotherms, such as insects, have to face and adapt to in their physical environment (Bale et al., 2002; Fasolo & Krebs, 2004). To minimize the energetic cost imposed by high-temperature conditions, insects can divide the available energy, differently, between life history and physiological or behavioural traits to enhance their survival during the stressful periods (Hercus et al., 2003). Moreover, the adaptive changes acquired during the continuous interaction between the ambient temperature and the organism genetic background (Sørensen et al., 1999) lead to optimization of the cost/benefit trade-offs between the physiological or behavioural responses and their price in fitness currencies (Loeschcke et al., 1994).

One leading strategy in studying insect’s thermal adaptation involves the comparison of geographical populations of the same species or closely related sibling species (Stratman & Markow, 1998; Garbuz et al., 2002; Bahrndorff et al., 2006). These populations often experience different selection pressures owing to different geographical origin or other habitat properties and can exhibit large differences in thermal stress resistance (Krebs & Loeschcke, 1995). In some cases, these differences have been associated with modified expression of stress-related genes (Sørensen et al., 2001, 2003), which can also remain when the same populations are subjected to favourable or optimal conditions (i.e. basal expression level of stress-related genes) (Mahadav et al., 2009). It was suggested that populations which maintain higher constitutive levels of stress-related genes may insure a better performance under stressful conditions (Shudo & Iwasa, 2002), but may also experience detrimental effects on life history traits when occurring in more favourable conditions (Feder et al., 1992; Krebs & Feder, 1997a). Surprisingly, this line of research has focused, to date, mainly on Drosophilidae species, emphasizing the need for more broadened research approach that utilizes additional insect model systems.

Among insect ectotherms, phloem feeders like whiteflies have a limited ability to buffer thermal changes. This is because of their large surface area-to-volume ratio that prevents utilization of evaporation as a cooling mechanism (Prange, 1996) and the limited mobility of their immature developmental stages (Grill, 1990). Alternatively, to avoid high temperature, whiteflies favour the abaxial side of the leaf (Van Lenteren & Noldus, 1990) and utilize physiological mechanisms such as the accumulations of sorbitol to stabilize the conformation of proteins (Wolfe et al., 1998; Salvucci et al., 2000). Furthermore, whiteflies can utilize heat shock proteins (HSPs) (encoded by Hsp genes) and other stress-related genes to overcome thermal stress (Salvucci et al., 2000; Mahadav et al., 2009; Yu & Wan, 2009). HSPs prevent aggregation and assist mis-folded proteins during stress conditions (Bukau & Horwich, 1998). Within the Hsps, Hsp70s are the most studied group. These genes were found to be temperature sensitive in many insect orders (Dahlgaard et al., 1998; Salvucci et al., 2000; Edgerly et al., 2005; McMillan et al., 2005; Karl et al., 2009).

The whitefly, Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae), has emerged in recent years as one of the most serious agricultural pests in the world. It has been recognized as a complex of 11 well-defined high-level groups containing at least 24 morphologically indistinguishable species (Dinsdale et al., 2010; Xu et al., 2010; De Barro et al., 2011). The most widespread ones are the B and the Q [to avoid confusion, and to ensure a capacity to link this study with previous literature, we have retained the old biotype terminology but indicate that these biotypes now equate to the following putative species (species in parentheses), B (Middle East – Asia Minor 1) and Q (Mediterranean)]. B is believed to originate in the desert environments of north-eastern Africa, the Middle East and the Arabian Peninsula (Frohlich et al., 1999; De Barro et al., 2000). Q most likely originated in sub-Saharan Africa and spread throughout North Africa and the Mediterranean Basin (Boykin et al., 2007).

Several characteristics make the B and Q sibling species a promising model system for studying the evolution of thermal tolerance and its associated trade-offs. First, their small body size suggests that even with behavioural avoidance of hot microclimates, these insects may risk great degree of heat stress (Stratman & Markow, 1998). Second, several lines of evidence suggest that the B and Q sibling species went through an allopatric (geographical) divergence process (Moya et al., 2001), in which they were likely to encounter different thermal selection forces. Third, the two species are completely reproductively isolated (Elbaz et al., 2010) but are closely related (Dinsdale et al., 2010), as they retain sufficient genetic similarity to permit accurate comparisons of phenotypic performance (Karunker et al., 2008). Fourth, the two species occupy both natural habitats and agricultural patches (Khasdan et al., 2005) in which the environmental changes are expected to be less gradual, providing an experimental system in which the combined effect of natural and men-made thermal selection forces can be studied. Fifth, analyses of several independent B and Q populations around the world have indicated that the two species differ in many life history traits (Pascual & Callejas, 2004; Iida et al., 2009; Crowder et al., 2010). Also, the two species were recently shown to differ in the expression profile of stress-related genes when subjected to both heat stress and favourable conditions (Mahadav et al., 2009).

The leading research hypothesis in this study was that the B and Q sibling species of B. tabaci have evolved, during allopatric divergence, different strategies to optimize their performance under heat stress. Our experimental system was specifically designed to focus on basal heat shock resistance (i.e. resistance of nonhardened insects), which reflects the organism’s investment in the ability to handle sudden and nongradual changes in temperature. We had three major objectives: first, to establish how fitness-related traits such as oviposition, survival and courtship behaviour change in adults of the two species under heat stress. Second, to investigate whether the basal and induced expression profiles of Hsp70 genes differ between the two species, as Hsp70s expression under these conditions can be used to reflect not only the severity of the stress experienced by the organism but also its evolutionary background (Sørensen et al., 2001). Third, as expression of Hsp70s was shown to improve several aspects of adult fitness (Tatar et al., 1997), but was also associated with declined fecundity and reduced developmental rate and larva-to-adult survival (Krebs & Feder, 1997b), we also asked whether larval costs can be induced through maternal experience of heat shock, indicating a direct trade-off between adult survival and reproduction (Silbermann & Tatar, 2000). Larval costs were estimated by measuring egg maturation, progeny development and progeny survival rates.

Methods

Origin and maintenance of colonies

Two B. tabaci colonies, collected from fields in Israel, were used in this study: Ashalim-03 and Arava-03. Based on their cytochrome oxidase I sequences (Frohlich et al., 1999), Ashalim-03 was designated as B whereas Arava-03 was designated as Q. Ashalim-03 was collected in the western Negev and Arava-03 was collected in the Arava valley in 2003. Since their collection, the colonies were maintained in separate rooms and reared under standard greenhouse conditions of 26 ± 2 °C, photoperiod of L : D 14 : 10 h on cotton (Gossypium hirsutum L. cv. Acala).

Oviposition and development to first instar after heat shock

Male and female pupae were situated in 3.5-cm Petri dishes, with 1-cm hole covered with 50-mesh net, containing cotton leaf discs. All discs were placed with their adaxial surface downwards onto a bed of 1% agar. The plates were checked the next day for the presence of one male and one female newly emerged adults. Petri dishes containing one couple were subjected to 1.5 h of 42 °C heat stress in an incubator. Temperature was measured in all experiments using HOBO data logger (Onset Computer Corporation, Bourne, MA, USA) with sensor in Petri dish. Control plates were subjected to 1.5 h of 27 °C. After treatment, Petri dishes were placed in an incubator at 27 °C (photoperiod of L : D 14 : 10 h) for the remainder of the experiment. The number of eggs oviposited by each female was monitored at 1, 24, 48, 72 and 96 h. At each time point, the position of every egg was marked on a paper image of the cotton leaf. This method allowed us to follow the development time to first instar and viability of each egg separately. After 4 days (from heat shock), adults were removed and egg development was monitored for additional 9 days. If a male or a female adult died during the experiment, the plate was excluded from the analysis. Each (species × heat treatment) combination was repeated 10 times. Eggs that did not hatch 9 days after oviposition were scored as dead.

Adult survival after heat shock

Ten newly emerged adult couples were collected and placed in each Petri dish prepared as described previously. The viability of the adults was monitored prior to heat treatment. Then, Petri dishes were subjected to 1.5 h of 42, 43 and 45 °C heat stress in an incubator. Control plates were subjected to 1.5 h of 27 °C. After treatment, Petri dishes were placed in an incubator at 27 °C for the remainder of the experiment. First measure of survival was conducted 1 h after heat stress to allow recovery. Individuals not showing repetitive movement of more than one appendage in response to tapping on the plate lid were scored as dead. The location and sex of each dead individual was marked on a paper image of the cotton leaf. Mortality was scored for 8 days (day 1, 2, 3, 4, 6 and 8) in the 27 and 42 °C heat treatments, and for 4 days (day 1, 2, 3 and 4) in the 43 and 45 °C heat treatments. Each (species × heat treatment) combination was repeated 9–10 times. The proportion of dead male and female adults in each plate was calculated each day. In the 27, 43 and 45 °C heat treatments, we also scored for B and Q the number of courting adults. This was done once a day, at the same time, across all plates. The proportion of courting adults from the total number of survivors was calculated.

In the same plates, we also compared the proportion of a second category of nondead individuals that were attached to the leaf disc with their wings because of wax melting. Most of these individuals were not able to feed on the leaf discs and therefore were analysed separately. This phenomenon, which was designated as ‘loss of mobility’, was observed only in experiments where adults experienced heat shock conditions of 43 and 45 °C.

Progeny development and survival after heat shock

Ten newly emerged adult couples were collected and placed in a clip cage that was attached to the abaxial surface of a cotton leaf of 3-week-old plants. Two clip cages were attached to each cotton plant, one containing B couples and the other containing Q couples. This was mainly done to reduce plant-related variances in developmental studies. It is true that in this experimental setting, one cannot rule out the possibility of potential interactions between B and Q on the same individual plant, without providing direct evidence for the given species system under investigation. However, we believe that these interactions were unlikely to play, if at all, a meaningful effect in our experimental system for two main reasons. First, Crowder et al. (2010) used cotton plants (separate for each species) for comparing life history traits between B and Q populations and obtained nearly identical estimations for relative progeny developmental rate and survival. Second, the potential interactions could only be systemic as our setting involved different and separated leaves. Previous studies have indicated that systemic effects of long pre-infestation periods of B. tabaci B (20 days) are limited only to the proximal noninfested leaf (Inbar et al., 1999), or to nondirect effects through physiological disorders (Xue et al., 2010). As cotton plants have never been reported to suffer from systemic or physiological disorders associated with feeding of B. tabaci, and our infested leaves were well separated, it is unlikely that significant interactions took place in our experimental system.

Plants with attached clip cages were subjected to 1.5 h of 42 °C heat stress in an incubator. Control clip cages containing 10 coupes were subjected to 1.5 h of 27 °C. In order to exclude from our analysis the effect of heat stress on cotton plants, control plants were also subjected to the 1.5 h of 42 °C heat stress treatment, after which control clip cages were attached. After treatment, plants carrying clip cages were placed in an incubator at 27 °C for the remainder of the experiment. B. tabaci adults within clip cages were allowed an oviposition period of 72 h after which adults were removed by collection. The total number of F1 progeny, their developmental stages and viability status (live/dead) were recorded after 19 days. Developmental progression was estimated by calculating the proportion of late fourth instars (red-eye stage and above) on each plant. Mortality ratio was estimated as the proportion of dead progeny from the total oviposited on each plant. After developmental documentation, the oviposition areas of each leaf were cut and placed on a 1% agar. When most of the fourth instars emerged, adults were counted and sexed, and the proportion of emerging females was calculated. Each (species × heat treatment) combination was repeated 15 times.

Basal and induced Hsp70 expression in Bemisia tabaci

Fifty newly emerged adult couples (1–4 days old) were collected and placed in a clip cage that was attached to the abaxial surface of a cotton leaf of 3-week-old plants. Two clip cages were attached to each cotton plant, one containing B couples and the other containing Q couples. Then, plants with attached clip cages were subjected to 1.5 h of 42 °C heat stress in an incubator. Control clip cages containing 50 couples were subjected to 1.5 h of 27 °C. Again, in order to exclude from our analysis the effect of heat stress on cotton plants, control plants were also subjected to the 1.5 h of 42 °C heat stress treatment, after which control clip cages were attached. After treatment, plants carrying clip cages were placed in an incubator at 27 °C for the remainder of the experiment. RNA was extracted from B. tabaci adults within clip cages at 0, 2, 4, 8, 16 and 48 h after heat treatment. At least three replicates were produced for each combination of time point and species except for the 48-h treatment in which two replicates were conducted.

RNA extraction and quantitative RT-PCR analyses

Total RNA was extracted from 100 adults using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instructions. First-strand cDNA was synthesized by Superscript II (Invitrogen) using 1 μg total RNA and an Oligo-dT primer according to the manufacturer’s instructions.

For quantitative real-time PCR analysis of Hsp70 mRNA expression (GenBank accession no. ACH85197– details on gene selection are provided in supporting information and Fig. S1), 1 μg of total RNA was treated with DNase (Promega, Madison, WI, USA), and the RNA was re-extracted in phenol/chloroform followed by an ethanol precipitation. cDNA synthesis was conducted as described earlier, using the Oligo(dT) primer. Real-time PCR primers were 5′ TGATGATACCGTCGATGG 3′ and 5′ TCAGGGTGTAA TGGTCGGTA 3′ for actin, and 5′ AGCGGGCCAAACGAACTCTTT 3′ and 5′ AAGTTCTTCGAACC TGGCACGG 3′ for Hsp70. These primers result in PCR products of 110 and 113 bp, respectively. Real-time PCR was performed as described in the Supporting Information. Quantification of the transcript level was conducted according to the ΔCT method (Yuan et al., 2006) using actin as the reference gene.

Statistical analyses

All statistical analyses conducted in this paper used jmp statistical software version 7.0.1 (SAS Institute, Cary, NC, USA). Statistical significance was assumed at  0.05. Proportionate data were arcsine-square-root transformed to prevent the variance to be a function of the mean and to improve normality. Bartlett’s test was used to test for equality of variances. Comprehensive data of all statistical analyses are summarized in Table S1.

To test for significant differences in oviposition, egg developmental time, adult survival and courtship behaviour, we used a three-way repeated-measures anova model as follows:

image

In which day was the repeated (within-subjects) factor whereas species and heat treatment were the main (between-subjects) factors. When anova yielded a significant F value, means were separated by the Tukey–Kramer honestly significant difference (HSD) test. In addition, specific means in the (species × heat treatment × day) triple interaction were selected a priori and orthogonal comparisons were conducted.

Significant differences in Hsp70 gene expression were tested by hierarchical two-way anova model as follows

image

Specific means in the (species × time) interaction were selected a priori and orthogonal comparisons were conducted.

Significant differences in progeny viability, F1 progeny developmental time, proportion of emerging females and loss of mobility were tested by two-way anova model as follows:

image

When anova yielded a significant F value, means were separated by the Tukey–Kramer HSD test.

Relative fitness estimates

The fitness of B and Q was compared under control (27 °C) and heat shock (42 °C) conditions. We used estimates of fecundity (eggs/female/day), progeny survival, proportion of late fourth instars at day 19, adult survival and proportion of emerging females to compute for each species the reproductive performance per generation (fecundity × progeny survival × proportion of late fourth instars at day nineteen × adult survival × proportion of emerging females). The fitness of Q relative to the B was estimated as the reproductive performance of Q divided by the reproductive performance of B.

Results

Fecundity

The number of eggs oviposited by B females was significantly higher (= 0.019) than the number of eggs oviposited by Q females. The heat shock treatment also caused a significant reduction in oviposition (= 0.009). Both species were affected by the heat stress in a similar manner as the species × heat treatment interaction was not significant. Pairwise a priori comparisons within each species uncovered a significant reduction in oviposition (in both species) 2 days (B; t124 = 2.02, P = 0.045 and Q; t124 = 2.38, P = 0.019) and 4 days (B; t124 = 2.38, P = 0.018 and Q; t124 = 2.46, P = 0.015) after heat shock (Fig. 1).

Figure 1.

 The mean daily number of oviposited eggs per female (from B and Q) treated at 42 °C for 90 min or held as control at 27 °C. Measurements at time = 0 were taken 1 h after heat shock. Asterisk indicates significant difference between means in the species × heat treatment × day interaction, as detected by a priori and orthogonal comparisons ( 0.05). Error bars represent standard error of the means.

Development to first instar

Developmental time to first instar was significantly shorter in B (< 0.0001). On the other hand, heat treatment and species × heat treatment were not significant. Only in B, pairwise a priori comparisons indicated significant increases in the development time (from egg to first instar) in eggs oviposited two (t641 = 2.63, P = 0.008) and 4 days (t641 = −2.86, P = 0.004) after heat shock (Fig. 2).

Figure 2.

 The effect of maternal heat shock (42 °C for 90 min or held as control at 27 °C) on the daily mean egg developmental time of B and Q. Asterisk indicates significant difference between means in the species × heat treatment × day interaction, as detected by a priori and orthogonal comparisons ( 0.05). Error bars represent standard error of the means.

Adult survival

Heat shock of 42 °C had no negative effect on the survival (for 8 days) of males and females from both species as the species, heat treatment and their interaction were not significant. Pairwise a priori comparisons between control and 42 °C treated individuals revealed that the survival of females, only from Q, was significantly higher 6 days (0.97 ± 0.014 in heat treated vs. 0.91 ± 0.041 in controls, t216 = −2.57, = 0.011) and 8 days (0.97 ± 0.014 in heat treated vs. 0.88 ± 0.044 in controls, t216 = −4.80, < 0.001) after heat treatment. At a more stressful heat shock condition of 43 °C, the effect of species was not significant for females and males but the heat treatment (females, = 0.008; males, = 0.0005) and species × heat treatment (females, = 0.028; males, = 0.053) effects were. Tukey–Kramer HSD test indicated that the two species responded differentially, as only B females receiving the heat treatment showed reduced survival. Pairwise a priori comparisons between control and 43 °C treated individuals revealed that reduction in survival occurred in both sexes of B and males of Q at all 4 days (Fig. 3). In contrast, Q females experienced significant reduction in survival only on the fourth day after heat shock. At the 45 °C treatment, the asymmetry in B and Q responses to heat stress was more pronounced, and all three effects were significant: species (females, = 0.029; males, = 0.018), heat treatment (females, < 0.0001; males, < 0.0001) and their interaction (females, = 0.006; males, = 0.043). Tukey–Kramer HSD test indicated that individuals (from both species and sexes) that received the heat treatment showed reduced survival. The effect, however, was significantly stronger in B. Pairwise a priori comparisons between control and 45 °C treated individuals revealed that in both species and sexes, the reduction in survival occurred on all 4 days (Fig. 3).

Figure 3.

 Daily survival of B and Q adults treated at 42, 43 and 45 °C for 90 min or held as control at 27 °C. (a) B♀; (b) B♂; (c) Q♀; (d) Q♂. Asterisk indicates significant difference between means in the species × heat treatment × day interaction, as detected by a priori and orthogonal comparisons ( 0.05). Error bars represent standard error of the means.

Adult courtship and loss of mobility

At 43 °C, the proportion of adults that courted was not significantly different between species and heat treatment. The interaction, however, was highly significant (= 0.005). Tukey–Kramer HSD test indicated that in Q, there was a significant increase in the proportion of heat-treated adults that courted compared with control. In B, the proportion of courting adults was not different between control and heat-treated adults. Pairwise a priori comparisons within each species on days 0–4 uncovered a significant elevation in courting proportion 1 h after heat shock in both species (B; t119 = −2.73, = 0.0073; Q, t119 = −2.98, = 0.0035) and 4 days after heat shock only in Q (t119 = −2.03, = 0.044) (Fig. 4). At this temperature, we also observed a significant higher proportion of B males and females that lost their mobility because of wax melting in their wings, a phenomenon not observed in Q males and females (Fig. S2). At 45 °C, the proportion of adults that courted among the survivors was analysed only for Q adults as there were no male survivors in B. Analysis indicated a significant reduction in courtship behaviour among Q heat-treated adults compared with control (F1,18 = 6.85, = 0.017). At this temperature, higher proportion of individuals, from both species and sexes, showed significant reduction in mobility because of wax melting. The effect, however, was much stronger in B (Fig. S2).

Figure 4.

 The daily mean proportion of courting adults, from B (a) and Q (b), treated at 43 °C for 90 min or held as control at 27 °C. Measurements at time = 0 were taken 1 h after heat shock. Asterisk indicates significant difference between means in the species × heat treatment × day interaction, as detected by a priori and orthogonal comparisons ( 0.05). Error bars represent standard error of the means.

Progeny development and survival

Analysis of the mean percentage of progeny that had reached advanced stages of development (late fourth instars) by day nineteen revealed that B developed significantly faster than Q in both control and heat shock treatments (< 0.0001) (Fig. 5). The heat treatment effect was not significant. The significant interaction between the species and heat treatment effects (= 0.007) indicated that the two species differ in their response to the heat treatment as B increased whereas the Q decreased its developmental rate (Fig. 5). In addition, progeny mortality was significantly lower in B than in Q (= 0.0026) whereas heat treatment and species × heat treatment had no significant effects. However, Tukey–Kramer HSD test indicated that progeny mortality did not differ between the B and Q at 27 or 42 °C (Table 1). Analysis of the mean proportion of emerging females did not detect significant differences between the four species × heat treatment combinations (data not shown).

Figure 5.

 The effect of maternal heat shock (42 °C for 90 min or held as control at 27 °C) on the proportion of progeny of B and Q that had reached advanced stages of development (late fourth instars) by day nineteen. Different letters indicate significant differences between the means (Tukey–Kramer HSD test, P ≤ 0.05). Error bars represent standard error of the means.

Table 1.   Life history traits of B and Q colonies treated at 42, 43 and 45 °C for 90 min or held as control at 27 °C.
VariableQ/BMean ± SE
QB
  1. *ND, not determined.

27 °C
 Eggs/female/day0.766.94 ± 0.989.16 ± 1.2
 Proportion of viable progeny at day 190.930.84 ± 0.030.91 ± 0.01
 Proportion of late fourth instars at day 190.210.10 ± 0.030.48 ± 0.04
 Survival of adult females0.960.94 ± 0.030.98 ± 0.01
 Survival of adult males0.980.95 ± 0.020.97 ± 0.01
 Proportion of courting pairs0.750.33 ± 0.030.45 ± 0.03
 Proportion of emerging females1.010.52 ± 0.070.51 ± 0.02
42 °C
 Eggs/female/day0.684.52 ± 0.546.68 ± 0.50
 Proportion of viable progeny at day 190.950.82 ± 0.020.87 ± 0.01
 Proportion of late fourth instars at day 190.070.04 ± 0.020.59 ± 0.02
 Survival of adult females0.980.97 ± 0.010.99 ± 0.01
 Survival of adult males1.010.96 ± 0.20.95 ± 0.03
 Proportion of emerging females1.030.57 ± 0.070.55 ± 0.02
43 °C
 Survival of adult females1.570.77 ± 0.10.49 ± 0.12
 Survival of adult males1.580.76 ± 0.060.48 ± 0.12
 Proportion of courting pairs1.200.47 ± 0.030.39 ± 0.04
45 °C
 Survival of adult females2.890.24 ± 0.10.083 ± 0.05
 Survival of adult malesND*0.22 ± 0.10.00 ± 0.00
 Proportion of courting pairsND*0.16 ± 0.040.00 ± 0.00

Induction of HSP70 expression after heat shock

Comparisons of the mean ΔCt levels revealed a significant difference in the expression level of Hsp70 between B and Q (= 0.0115), with higher expression levels in Q. Significant differences were also observed across the different time periods after the heat shock (< 0.0001). Pairwise a priori comparisons between the two species at each time point indicated significantly higher basal expression levels of the analysed Hsp70 gene in Q (t33 = 2.171, = 0.037) (Fig. 6). In both species, Hsp70 expression levels were up-regulated several 100-folds after the heat treatment. These induced expression levels did not differ between B and Q up to 8 h (Fig. 6). In contrast, expression levels of Hsp70 were significantly lower in B 16 h after heat shock (t33 = 1.97, = 0.05) (Fig. 6), suggesting a more ‘transitory’ overexpression event in this species. A similar trend was observed 48 h after the heat treatment.

Figure 6.

 Levels of Hsp70 (mRNA) (mean ΔCt ± SE) in B and Q at different times elapsed after heat shock (42 °C for 90 min or held as control at 27 °C). After the heat treatment, individuals were allowed to recover for different lengths of time (0, 2, 4, 8, 16 and 48 h) at control conditions (27 °C). Asterisk indicates significant difference between means in the species × time after heat treatment interaction, as detected by a priori and orthogonal comparisons ( 0.05). Error bars represent standard error of the means.

Relative fitness of B and Q at 27 and 42 °C

At a control temperature of 27 °C, the combined effect of fecundity, progeny survival, proportion of late fourth instars at day nineteen, adult survival and proportion of emerging females gave a reproductive performance per generation estimate of 1.94 for B and 0.27 for Q (Table 1), which yielded 0.14 (0.27/1.94) for the fitness of Q relative to B. After heat shock of 42 °C, the fitness advantage of B was more pronounced as the estimated reproductive performance per generation for B (1.77) and Q (0.078) yielded relative fitness (Q/B) of 0.044 (Table 1). Comparisons of the two species’ relative reproductive performance per generation under 27 and 42 °C (B42 °C/B27 °C and Q42 °C/Q27 °C) revealed only a small reduction in performance in B (0.91) in contrast to severe reduction in performance in Q (0.28) (Table 1). Combined fitness was not computed for the 43 and 45 °C heat treatments as only adult survival and courtship behaviour were estimated.

Discussion

The main goal of this research was to compare thermal tolerance and its associated life history trade-offs between the B and Q sibling species of B. tabaci. The experimental system was specifically designed to focus on basal heat shock resistance to short stressful conditions, which reflects the organism investment in insuring its performance under unpredictable environments at the price of lowering its performance at more stable and favourable conditions. Our major findings can be divided into three: (i) At favourable conditions of 27 °C, the relative fitness of B, estimated as the relative reproductive performance per generation, was about seven-fold higher than that of Q. (ii) After heat stress of 42 °C, the differences between the species, in the relative reproductive performance per generation, increases to 23-fold, mainly due to significant differences in the progeny developmental rate trends (the B increased whereas the Q decreased its developmental rate; Fig. 5). (iii) At more stressful heat shock conditions of 43 and 45 °C, a complete turnover took place in which adults of Q showed not only enhanced survival and movement ability compared with B but also an increase in their courtship behaviour frequency. This discussion will focus first on possible explanations for these findings and then address the implications of the differences found for the biology and evolutionary ecology of the B and Q sibling species of B. tabaci.

Heat shock of 42 °C had no negative effect on the survival of adult males and females from both species. In contrast, the maternal heat shock had a negative effect on fecundity and progeny developmental rates in both B and Q, although the latter occurred at different stages in each species. Only the progeny of B females experienced a reduction in development rate in the egg to first instar developmental stage. However, this reduction in developmental rate was only ‘transitory’ as the proportion of the individuals that reached the late fourth instar stage at day 19 did not differ between the heat-treated and control treatments. In contrast, Q progeny experienced a more delayed negative effect that reduced the proportion of the individuals that reached the late fourth instar stage at day 19. One of the several possible mechanistic explanations for such negative maternal effects is the expression of HSPs (Krebs & Feder, 1998). In our system, the expression of the Hsp70 gene increased several 100-folds in both species, immediately after the heat stress (Fig. 6). Overall, however, the expression of the Hsp70 gene was significantly higher in Q. In addition, significant lower expression levels of the gene were observed in B, 16 h after the heat stress, suggesting a more ‘transitory’ overexpression event in this species.

Expression of Hsp70s was associated previously with declined fecundity and reduced larva-to-adult survival (Krebs & Feder, 1997b). Hsp70 overexpression may produce reproductive costs in either of two general ways. First, the expression of Hsp70 during oogenesis may disrupt specific cellular processes of development, such as protein trafficking, cell polarization or signalling (Silbermann & Tatar, 2000). Second, the allocation of energy or nutrients to somatic HSP70 production may reduce the pool available for egg production and maturation. Here, we bring data suggesting that overproduction of Hsp70 in B. tabaci may impose an energetic demand upon reproduction. This is mainly due to the reduction in oviposition observed in both species and the slow egg maturation rate and delayed larvae development observed in B and Q, respectively, which may help in preserving limited resources. Lengthening the time over which energetic costs are paid was suggested before as a general mechanism to reduce heat shock effects on fitness (Krebs & Feder, 1998). The effect may be, however, temporal (as seen in B) and directly correlated to the magnitude of the energetic demand.

If both B and Q are capable of inducing Hsp70 expression (although not to a similar duration) for achieving thermotolerance, why only Q increased its survival 6 and 8 days after heat treatment of 42 °C and was superior to B at surviving acute high-temperature exposures of 43 and 45 °C? Taking into account that animals face continuous evolutionary decisions, whether to allocate limited energy resources into current reproductive output or to functions that protect the soma (Speakman, 2008), we can use a generalized version of the ‘direct inhibitory’ trade-off model proposed by Silbermann & Tatar (2000) to provide a conceptual explanation to the observed differences between the B and Q sibling species. In its generalized form, the model argues that the evolution of HSPs (and other heat stress–related protein) regulation is likely to show trade-offs if lowering the expression levels of stress-related genes minimizes reproductive costs, but at the ‘price’ of reduced ability to protect the soma from the cumulative effects of intrinsic and extrinsic challenges. The model also predicts that as many cellular and systemic physiologies are necessary to support reproduction, there could also be ‘direct negative’ trade-offs. For example, enhanced mitochondrial metabolism aiming to provide sufficient energy for reproduction will also produce enhanced levels of reactive oxygen species (ROS). This potential for ROS-induced damage can be reduced by antioxidants and repair mechanisms activity. However, if ROS production exceeds this protection due to enhanced metabolic rate or lower regulatory state of the repair mechanisms, the net result can be severe oxidative damage, loss of cellular homeostasis and increased mortality rate (Selman et al., 2008).

Recently, Mahadav et al. (2009) provided first evidence supporting the existence of differences in regulation of stress-related genes between the B and Q sibling species, as the research identified several genes belonging to this class (for example Hsp40 and Hsp90) which were significantly induced following heat stress in Q relative to B. These findings together with the observed overall higher expression, in Q, of the Hsp70 gene analysed in this study (Fig. 6) suggest the possibility that under stress, Q is investing more than B in processes beneficial to somatic maintenance, in the ‘price’ of reduced reproduction (Fig. 1), whereas B is relatively reducing the expression of such defence mechanisms in order to lower their negative effect on reproduction. Following this line, we believe that the acute high-temperature exposures of 43 and 45 °C produce cumulative effects of intrinsic and extrinsic somatic stress that B can no longer cope with because the expression level of many defence proteins in this species is too low. In addition, we cannot rule out the possibility that minor differences in surface lipid composition (both in wax particles and in lipids of the cuticular surface) may also contribute to the observed differences in resistance to acute high temperatures in B and Q (Buckner et al., 1994).

The higher investment of Q in processes beneficial to somatic maintenance can also explain the huge difference in the relative reproductive performance of B and Q under favourable (27 °C) and nonacute stress conditions (42 °C). As mentioned earlier, comparison of female fecundity, mortality of immature instars and developmental rate in several independent B and Q populations around the world indicated higher reproductive potential in B populations compared with Q populations on many plant hosts, mainly due to shorter developmental time in B (Pascual & Callejas, 2004; Iida et al., 2009; Crowder et al., 2010). On some host plants, however, no differences (cucumber, kidney bean – cv. Satsukimidori) or a shorter developmental time (sweet paper) was observed in Q populations (Muñiz & Nombela, 2001; Iida et al., 2009), suggesting that developmental speed in B and Q could also be affected by the plant host. It is interesting to note that Iida et al. (2009) also found significant reduction in B nymph survival (compared with Q) on all tested kidney bean cultivars and a similar nonsignificant trend on cucumber, which raises the possibility that maturation and survival rates of B on some host plants are negatively affected by the species relative lower investment in defence mechanisms. This might reduce the species ability to handle not only acute high temperatures but also toxic host plant chemicals, a matter which requires further investigation into future studies. In complement, several independent lines of evidence indicate higher basal expression levels of stress-related genes in Q. First, the basal expression levels of the Hsp70 gene analysed in this study were significantly higher in Q than in B (Fig. 6). Second, Mahadav et al. (2009) found that at 25 °C, several of the aforementioned stress-related genes as well as several genes encoding microfilament and cytoskeleton proteins show significant higher levels of basal expression in Q than in B. Third, our recent data indicate that many metabolic genes from the families cytochrome P450 monooxygenases (P450s), glutathione S-transferases and the carboxylesterases, capable of detoxifying xenobiotics, also show significant higher basal levels of expression in Q, relative to B (E. Halun, unpublished). Taken together, the ecological and molecular data accumulated so far support the conclusion that lower investment of B in somatic maintenance can be beneficial to the fitness of this species at favourable and nonacute stress conditions but harmful when the stress becomes acute. On the other hand, the high investment of Q in somatic defences reduces its reproductive performance at favourable and nonacute stress conditions, but gives it advantage when facing severe environmental constraints. It also, most likely, explains why only Q females showed increased survival 6 and 8 days after heat treatment of 42 °C, as elevated tolerance to oxidative, heat and irradiation stresses was previously associated with increased lifespan (Lithgow et al., 1995; Tatar et al., 1997).

We can only raise hypotheses about the selection forces that shaped the different basal thermal stress tolerance in the B and Q sibling species of B. tabaci. Both hypotheses are stemming from the documented differences in their geographical origins (Boykin et al., 2007). The first hypothesis suggests that Q experienced in the past, natural and agricultural habitats that were more heterogenic or less predictable, which could have led this species to become more generally resistant. For example, comparison of heat shock resistance in different populations of Drosophila buzzatii found that populations coming from mountain areas, which are considered stressful and variable, were more resistant to heat stress without hardening and also express HSP70 to higher levels after hardening than populations from lowland areas, which are hotter but more stable (Sørensen et al., 2001). In hydra, species restricted to habitats of stable temperature and pH showed only small increase in Hsp70 transcript level, at maximum induction temperature, whereas species living in a wide range of freshwater environments under variable temperatures displayed significant increase (∼200-fold) in Hsp70 activity after heat shock (Bosch et al., 1988; Brennecke et al., 1998). The alternative hypothetic mechanism suggests that the two sibling species were selected under environments that were thermally different. It was previously suggested that long-term exposure to elevated temperatures results in selection for reduced HSPs expression as a consequence of fitness costs associated with expression (Sørensen et al., 1999, 2001; Garbuz et al., 2002). According to this, we would expect the B species, which induces HSPs to lower levels, to have experienced in the past hotter habitats. The reduction in expression of thermal stress–related genes together with behavioural thermoregulation and utilization of host microclimate may have allowed this species an optimal trade-off between thermal resistance and performance (Krebs & Loeschcke, 1994; Stratman & Markow, 1998).

Acknowledgments

We thank Ofer Ovadia for helpful comments on the experimental design and data interpretation. This work was supported by the Israel Science Foundation grant 971/04.

Ancillary