Diapause termination of Rhagoletis cerasi pupae is regulated by local adaptation and phenotypic plasticity: escape in time through bet-hedging strategies

Authors

  • C. A. Moraiti,

    1. Department of Agriculture, Crop Production and Rural Environment, Laboratory of Entomology and Agricultural Zoology, University of Thessaly, N. Ionia (Volos), Greece
    Search for more papers by this author
  • C. T. Nakas,

    1. Department of Agriculture, Crop Production and Rural Environment, Laboratory of Biometry, University of Thessaly, N. Ionia (Volos), Greece
    Search for more papers by this author
  • N. T. Papadopoulos

    Corresponding author
    1. Department of Agriculture, Crop Production and Rural Environment, Laboratory of Entomology and Agricultural Zoology, University of Thessaly, N. Ionia (Volos), Greece
    • Correspondence: Nikos T. Papadopoulos, Department of Agriculture, Laboratory of Entomology and Agricultural Zoology, Crop Production and Rural Environment, University of Thessaly, Fytokou St., 384 46 N. Ionia (Volos), Greece. Tel.:+30 24210 93285; fax: +30 24210 93285;

      e-mail: nikopap@uth.gr

    Search for more papers by this author

Abstract

Persistence and thriving of univoltine, herbivore insect species of the temperate zone rely on obligate diapause response that ensures winter survival and synchronization with host phenology. We used a stenophagous fruit fly (Rhagoletis cerasi) with obligate pupae diapause to determine genetic and environmental effects on diapause intensity of geographically isolated populations with habitat heterogeneity. Pupae from two Greek and one German populations with various gene flow rates were exposed at five constant chilling temperatures (0–12 °C) for different durations and then incubated at a high temperature until all adults have emerged. Pupae diapause intensity differs among Greek and German populations, suggesting an adaptive response to habitat heterogeneity (mostly differences in phenology patterns of local host cultivars). Moderately warm winter temperatures, such as 8 °C, promote diapause termination in all three populations. Insufficient chilling (short duration or warmer temperatures) regulates the expression of prolonged dormancy. Interestingly, extended chilling (longer than required for terminating diapause) ‘return’ pupae to another (facultative) cycle of dormancy enabling adults to emerge during the next appropriate ‘window of time’; a strategy first time reported for univoltine insects. Consequently, diapause duration of R. cerasi is determined both by i) the adaptive response to local climatic conditions (annual dormancy) and ii) the plastic responses to interannual climatic variability resulting in two types of long life cycles within populations, prolonged and facultative dormancy as response to insufficient chilling and extended exposure to chilling, respectively. Long life cycles are expressed as a part of dormancy bet-hedging strategies of R. cerasi populations.

Introduction

In seasonal environments, dormancy (diapause and/or quiescence) is a major element of the life history of insect herbivores, synchronizing their life cycles with specific stages of host plants such as fruit ripening and newly flashed leaves (Tauber et al., 1986). For temperate species of the northern hemisphere, diapause (for most individuals) ends in December or even earlier (Santiago-Alvarez et al., 2003; Koštál, 2006; Papanastasiou et al., 2011), followed by post-diapause quiescence that ensures survival, and allows all individuals to terminate diapause before temperature rises the coming spring (Tauber et al., 1986; Hodek, 1996; Koštál, 2006; Belozerov, 2009). However, some individuals may extend their diapause for one or more years (prolonged dormancy), based on environmental factors, such as temperature, humidity and body energetic reserves. Seasonally repeated temperature cycles enhance prolonged dormancy termination in temperate species (Neilson, 1962; Dean & Hartley, 1977; Higaki & Ando, 2000; Menu & Desouhant, 2002; Higaki, 2005). Individuals expressing prolonged dormancy are exposed to mortality risks and other fitness costs in adult life, such as reduced fecundity (Lyons, 1970; Moraiti et al., 2012), even though exceptions exist (Soula & Menu, 2005; Wei et al., 2010). Apparently, the overwintering period is a strong determinant of fitness, and insects have evolved various strategies to cope with location and time-dependent selective pressures, including local adaptation and phenotypic plasticity (Hereford, 2009; Scheiner & Holt, 2012). For instance, diapause duration usually increases with latitude (Rank & Rank, 1989; Hodkinson, 2005; Demont & Blanckenhorn, 2008), though exception exists (Zhao et al., 2010). Moreover, plastic responses to photoperiod and temperature account for diapause induction in some temperate insects (Wang et al., 2009; Fischer & Karl, 2010). Thus, herbivorous insects of the temperate zone highly vary in diapause traits among and within populations driven by the spatiotemporal environmental heterogeneity of their habitats.

Global climate change is unequivocal, and as the United Nations Intergovernmental Panel (IPCC) notes down eleven out of twelve years within the period 1995–2006 were among the top 12 warmest years ever recorded since 1850. However, global warming is proceeding faster at northern than at southern latitudes in the Northern Hemisphere (Christensen et al., 2007). Insects respond to climate change through a) dispersal to more hospitable habitats, resulting in range shifts usually towards higher latitudes and/or altitudes, b) evolutionary adaptation or c) phenotypic and physiological plasticity (Hofmann & Togham, 2010; Angert et al., 2011; Jenouvrier & Visser, 2011). Evolutionary shifts to shorter day lengths for diapause induction have already been observed in nature (Gomi, 1997, 2007; Bradshaw & Holzapfel, 2001; Bradshaw et al., 2004; Yamanaka et al., 2008; Stoeckli et al., 2012). Multivoltine butterflies and moths in Central and North Europe have increased the number of generations per year as a plastic response to rising temperatures during spring and summer (Altermatt, 2010; Pöyry et al., 2011). Considering that changes are faster in autumn and winter compared with spring and summer (Christensen et al., 2007), univoltine insects of the temperate zone (with winter diapause) are likely to experience the most significant changes in their thermal overwintering environment (Bale & Hayward, 2010). Indeed, the duration of annual diapause for hibernating individuals has been found either to become shorter (Maeta, 1978) or to be extended (Gray et al., 2001; Bosch & Kemp, 2004; Vanhanen et al., 2007) by warm winter temperatures. However, the length of overwintering period remains relatively stable in the face of winter warming for univoltine species with temperature-independent diapause (Bosch & Blas, 1994; Fielding et al., 1999). It seems that temperate univoltine species respond differently to winter warming, and therefore, the impact of climate change on hibernating insects is challenging to predict (Robinet & Roques, 2010). However, the relative contribution of local adaptation and phenotypic plasticity to geographical variation in life-history traits is expected to determine the ‘vulnerability’ of each population to climatic changes (Hoffmann & Sgrò, 2011).

The European cherry fruit fly, Rhagoletis cerasi Linnaeus (Diptera: Tephritidae), is a univoltine, stenophagous species that infests fruit primarily of Prunus spp. (Rosaceae; P. cerasus, P. avium, P. mahaleb) and secondary of Lonicera spp. (Caprifoliaceae; L. xylosteum and L. tartarica) (White & Elson-Harris, 1992). It undergoes an obligatory autumnal-hibernal diapause at pupal stage that allows adults to emerge next spring when local host fruits are available (Boller & Prokopy, 1976). Climatic conditions interacting with host tree cultivars have a strong effect on blooming period, sweet cherry fruit set and the flight period of R. cerasi, as well (Garcia-Montiel et al., 2010). Geographical variation in diapause traits, such as diapause intensity and post-diapause development, accounts for differences in adult emergence patterns among R. cerasi populations at heterogeneous habitats (Baker & Miller, 1978; Kovanci & Kovanci, 2006; Papanastasiou et al., 2011). At population level, some individuals may fail to meet chilling requirements for diapause termination during a particular winter period (e.g. when winter temperatures are lower than 2 °C and/or pupae are exposed to winter temperatures for a period shorter than 3 months) and then follow an alternative strategy by prolonging their life cycle for one or more years (Vallo et al., 1976). Accordingly, the interannual variation in local temperatures (and possible that of other climatic factors) can also affect the emergence time of R. cerasi flies.

Rhagoletis cerasi seems to have all these life-history traits that make it vulnerable to global warming. First, it is a stenophagous species-infesting fruit with limited seasonal availability, and specialists are considered as more vulnerable to climatic change (McKinney, 1997; Colles et al., 2009). In addition, blooming of sweet cherry tree is highly sensitive to winter and early spring temperatures. Many sweet cherry growing regions around the Mediterranean Basin are projected to lose most of ‘their safe winter chill’ for fruit production, whereas growing regions in North Europe, and particularly in Germany, will not be much affected (winter chill reductions are expected to be compensated for by chilling gains caused by less frequent frost) (Chung et al., 2011; Luedeling et al., 2011). Early-flowering cultivars are expected to suffer the most by climate warming than late-flowering ones (Miller-Rushing et al., 2007). Accordingly, irregular dormancy releases resulting in unpredictable blooming and abnormal floral development are expected for cherry trees under climate warming (Oukabli & Mahhou, 2007; Luedeling et al., 2009). Secondly, R. cerasi is a univoltine species that spends more than 3/4 of its life cycle as hibernating pupae within the surface soil. Given that pupae require an extended chilling period to break dormancy and give adults next spring (Boller & Prokopy, 1976), R. cerasi is thought to be particularly exposed to winter warming (Bale et al., 2002). Thirdly, diapausing individuals (in pupae stage) seem to be vulnerable to energetic drain under warm winters since they have no opportunity to feed. Even though winter warming cannot always result in energetic drain in insects (Mercader & Scriber, 2008), some univoltine species suffer increased mortality due to high winter temperatures (Williams et al., 2003; Bosch & Kemp, 2004; Koštál et al., 2011; Sorvari et al., 2011). Fourthly, adults of R. cerasi have low flight capacity and reside in fragmented and specialized habitats (Fletcher, 1989; Kneifl et al., 1997). Limited dispersal activities of R. cerasi adults challenge the colonization of new patches of optimal environmental conditions under a climatic change scenario (Poisot et al., 2011). Overall, R. cerasi seems to be an excellent model organism to address the effects of winter warming on temperate univoltine species.

Even though several studies have already dealt with different aspects of R. cerasi diapause (Boller & Bush, 1974; Vallo et al., 1976; Baker & Miller, 1978; Papanastasiou et al., 2011), the interface between genetic and environmental factors that determine diapause termination and the expression of prolonged dormancy remain poorly understood. Here, we studied the response of geographically isolated R. cerasi populations with habitat heterogeneity to a wide range of temperature regimes for terminating diapause. Our first hypothesis was that diapause intensity varies among R. cerasi populations from ecological different habitats. We expect that pupae from warm regions need a shorter exposure to cold for terminating diapause than those from cold ones. In an attempt to distinguish between genetic and environmental effects on diapause intensity, we selected populations with different gene flow rates (Augustinos et al., 2013). Specifically, we used pupae from two Greek populations (Kala Nera and Dafni) with moderate gene flow rates and one German population (Dossenheim) genetically different from the Greek ones. Given that winter temperatures up to 5 °C are optimal for diapause termination of a wide range of European R. cerasi populations (Vallo et al., 1976), pupae from the above three populations were mainly challenged to terminate diapause under warmer winter temperatures (8, 10 and 12 °C). We expect that R. cerasi pupae from the warm, coastal area of K. Nera will successfully terminate diapause and yield adults after being exposed to higher (than usual) winter temperatures. In contrast, pupae from the cold regions of Dafni and Dossenheim are expected to fail terminating diapause under warmer winters. In order to examine how chilling conditions (temperature and duration of chilling) determine the expression of prolonged dormancy in each population, overlaying pupae were also recorded (after adult emergence was completed) in each temperature treatment. We tested the third hypothesis that R. cerasi pupae cannot meet their chilling requirements for diapause termination under warmer winter temperatures and/or short cold exposure and thus express prolonged dormancy (remain dormant for more than 1 year), whereas prolonged exposure to chilling results in high pupae mortality that may affect existence of R. cerasi in areas experiencing climatic disturbances.

Materials and methods

Study populations

We selected two Greek and one German R. cerasi populations originated from i) western Macedonia (Dafni, Kozani), ii) Thessaly (Kala Nera, Magnesia) and iii) the state of Baden – Württemberg (Dossenheim, Karlsruhe), respectively. The above populations reside in habitats with different geographical characteristics (Table 1) and climatic (temperature and rainfall) profiles (Table 2), allowing us to perform comparisons between populations from cold and warm regions at different spatial scales. Continental areas of Dafni and Dossenheim consist of late-flowering cherry cultivars, whereas early-flowering cherry cultivars can be found in coastal area of K. Nera. Given that R. cerasi adults have to emerge in synchrony with the ripe or ripening fruit of local cultivars, diapause termination of pupae from K. Nera is expected to be earlier than those of pupae from the other two cooler areas. In addition, winter temperatures (December-February) are often below 5 °C in both Dafni and Dossenheim, while temperatures around 10 °C are common in K. Nera (Table 2). Projected climate changes in Europe indicate that winter warming will vary from 2.6 °C to 8.2 °C in North Europe and 1.7 °C to 4.6 °C in the Mediterranean Basin until the end of this century, implying that German regions will be more affected than Greek areas (Christensen et al., 2007). In addition, Greek continental areas are expected to experience greater warming events compared to coastal ones (Giorgi & Lionello, 2008; Giannakopoulos et al., 2009; Tolika et al., 2012). Thus, choosing R. cerasi populations from areas that are expected to be differently affected by winter warming will provide a better understanding of diapause response to winter warming.

Table 1. Geographical areas and habitat characteristics for the three Rhagoletis cerasi populations.
PopulationLatitudeLongitudeAltitude (m)Phenology of sweet cherry cultivars
Dossenheim (Karlsruhe)N 49°27′0″E 8°40′0″153Late-flowering
Dafni (Kozani)N 40°17′8″E 21°8′53″1.050Late-flowering
Kala Nera (Magnesia)N 39°18′54″E 23°4′10″20Early-flowering
Table 2. Climatic data from the three areas where Rhagoletis cerasi populations were obtained from.
MonthTemperature (mean, °C)aMean precipitation, mma
DossenheimDafnibK. NeraDossenheimDafniK. Nera
  1. a

    Reference period: 2007–2010.

  2. b

    Data from Dafni obtained from the nearest meteorological station at Argos Orestiko (≈ 20 Km away).

January2.44.58.7353640
February4.14.09.9642129
March7.87.012.9775623
April12.71115.9454217
May15.915.821.31004531
June18.718.925.9714025
July20.223.328.59707
August19.423.228.379428
September14.617.522.6494435
October10.215.518.0457766
November6.913.014.45814647
December1.63.310.38562100

To separate environmental from genetic factors contributing to the potential variation in diapause intensity among the three populations, their genetic structure should be considered. Using microsatellite markers, it has recently been revealed that the two Greek populations used in our study form one cluster with moderate levels of gene flow, whereas the German population is quite different from the previous cluster (Augustinos et al., 2011, 2013).

Experimental procedures

Rhagoletis cerasi pupae obtained from field-infested sweet cherries (P. avium L.) that were collected from abandoned fields in Greece and from the cherry orchard of the Julius Kühn–Institute in Dossenheim. Infested fruits from Greece were taken to the Laboratory of Entomology and Agricultural Zoology at the University of Thessaly in Volos and placed in plastic containers over a layer of dry sand (1 cm thick) allowing mature larvae to pupate under constant conditions (25 ± 1 °C, 65 ± 5% relative humidity, and a photoperiod of L14/D10). Once a week, pupae were collected by sieving the sand. Infested cherries from Dossenheim were treated in a similar manner at Julius Kühn–Institute before pupae being sent to Greece.

Newly formed pupae remained for two months at 25 ± 1 °C, and then, they were exposed to five constant temperatures (0, 5, 8, 10 and 12 ± 1 °C) for a period ranged from 1 to 9 months. A sample of 100 pupae from each treatment was monthly transferred back to 25 ± 1 °C, and emergence rates were recorded. When adult emergence ceased, dead pupae and pupae still in dormancy were recorded.

Statistical analysis

Multinomial logistic regression was used to assess the effects of temperature, population and chilling period on the proportions of pupae yielding adults, dead pupae and overlaying pupae. Excluding dead pupae from analysis, binary logistic regression was used to assess the effects of temperature, population and chilling period on i) the proportion of pupae yielding adults and ii) the proportion of overlaying pupae. Polynomial regression methodology was employed in order to model the U-shape pattern of the proportion of overlaying pupae based on chilling period. Minimum values of the estimated equations are reported, thus providing the expected chilling period where the minimum percentage of overlaying pupae is achieved. All statistical analyses were performed using SPSS 20.0 (SPSS Inc. Chicago, IL, USA).

Results

The impact of different chilling treatments on the proportions of pupae that give adults, die or continue in dormancy (overlaying pupae) for the three R. cerasi populations is given in Fig. 1. In total, 30–35% of pupae from each population gave adults, 15–25% died and the rest of them (40–55%) remained alive (overlaying pupae). Individuals from Dafni have 91.2% increased odds of adult emergence compared to individuals from Dossenheim, whereas individuals from K. Nera have only 21.5% decreased odds of adult emergence compared to individuals from Dossenheim (Table 3). Emergence rates reach high levels (> 60%) within the range of 5–8 °C. At 5 °C, the ratio of pupae that give adults to overlaying pupae was high for all three populations and, in particular, for Dossenheim. Pupae from K. Nera achieved high proportion of diapause termination up to 10 °C. At 12 °C, diapause termination was substantially restricted (< 60%), and it was dramatically delayed at 0 °C (≥7 months) in all three populations. The ratio of pupae that give adults to overlaying pupae remained low in all three populations after chilling at low (0 °C) and higher temperatures (≥10 °C), indicating insufficient chilling requirements to terminate diapause. Pupae from K. Nera terminated diapause almost 1 month earlier than those from Dafni and Dossenheim at temperatures ranging from 5 to 10 °C. The ratio of pupae that give adults to overlaying pupae gradually increased until the peak of emergence in all temperature regimes. At 5 °C, no overlaying pupae were recorded at the peak of adult emergence. Interestingly, in all three populations, extended exposure at this temperature resulted in occurrence (increasing proportions in response to the length of exposure) of overlaying pupae. Specifically, the ratio of pupae that give adults to overlaying pupae increased after chilling for a period ranged from 4 to 6 months for pupae from K. Nera and from 6 to 7 months for pupae from both Dafni and Dossenheim, and then decreased for any additional chilling unit up to 9 months. It is noted that overlaying pupae (both before and after peak of adult emergence) can give adults at high proportions after one or additional years, depending on the population and temperature (cold/warm) regimes (Moraiti et al., 2012; C. A. Moraiti and N. T. Papadopoulos, unpublished).

Table 3. Results of the binary logistic regression testing the effect of temperature, population and chilling period on adult emergence. The population from Dossenheim forms the baseline.
Source of variationBSEExp (B)Sig.
Temperature0.1920.0191.211≤ 0.001
Population   ≤ 0.001
Dafni0.6480.2161.912= 0.003
Kala Nera−0.2410.2010.785= 0.229
Chilling period−0.2930.0290.746≤ 0.001
Population * temperature   ≤ 0.001
Dafni * temperature−0.0500.0140.951≤ 0.001
Kala Nera * temperature−0.1070.0140.899≤ 0.001
Population * chilling period   ≤ 0.001
Dafni * chilling period−0.0570.0340.945= 0.091
Kala Nera * chilling period0.1500.0311.161P ≤ 0.001
Temperature * chilling period−0.0280.0030.973≤ 0.001
Figure 1.

Effect of different temperature treatments on the proportions of Rhagoletis cerasi pupae that give adults, die and remain alive (overlaying pupae) in populations obtained from (a) Dossenheim, (b) Dafni and (c) Kala Nera.

The ratio of dead pupae to adult yielding ones remained low regardless of the chilling temperatures for the two Greek populations, but it increased at 12 °C for the German population. Adult yielding pupae outnumbered dead ones when the chilling period ranged from 6 to 7 months for pupae from K. Nera and from three to 8 months for pupae from Dafni and Dossenheim. Dead pupae outnumber adult yielding ones when chilling lasted < 3 months. Overall, more pupae from Dafni (that failed to give adults) remained alive than die, compared to pupae from K. Nera. Overlaying pupae outnumbered dead pupae at lower (0 °C) and higher (≥8 °C) temperatures in all three populations. The ratio of dead pupae to overlaying pupae remained low for the first 4–6 months in cold for pupae from K. Nera, and both Dafni and Dossenheim respectively; however, it increased when the chilling period lasted up to 8 months (see Fig. S1).

Multinomial logistic regression reveals that i) population, temperature, chilling period and their interactions have a significant effect on the ratio of pupae that give adults to overlaying pupae, ii) population, chilling period and the interactions between population and chilling period, between population and temperature, and between temperature and chilling period have a significant effect on the ratio of dead pupae to overlaying pupae, and iii) population, temperature, chilling period and the interactions between population and chilling period, and between temperature and chilling period have a significant effect on the ratio of dead pupae to adult (see Table S1, Fig. S1). Focused on emergence rates and the proportion of overlaying pupae, binary logistic regression analyses reveal that i) population, temperature, chilling period and their interactions have a significant effect on adult emergence (Table 3), and ii) population, temperature, chilling period and the interaction between population and chilling period have a significant effect on the proportion of overlaying pupae (Table 4).

Table 4. Results of the binary logistic regression testing the effect of temperature, population and chilling period on the numbers of overlaying pupae. The population from Dossenheim forms the baseline.
Source of variationBSEExp (B)Sig.
Temperature0.0670.0061.070≤ 0.001
Population   ≤ 0.001
Dafni0.7270.1442.068≤ 0.001
Kala Nera−0.3000.1360.741= 0.027
Chilling period−0.3810.0210.683≤ 0.001
Population * chilling period   = 0.004
Dafni * chilling period−0.0990.0300.906≤ 0.001
Kala Nera * chilling period−0.0280.0300.972= 0.351

The proportion of overlaying pupae as a function of chilling time at 5 °C follows a U-shape pattern for all three populations with maximum values at short and long chilling times (Fig. 2). The second-order polynomials with chilling period (months at 5 °C) as the predictor and the proportion of overlaying pupae as the dependent variable were significant for all three populations (Fig. 2, P < 0.01) with very high r-squared values (R2 > 0.83). We calculated the minimum of these polynomials in order to estimate the chilling period with lowest proportions of overlaying pupae. Minimum proportions of overlaying pupae are estimated at 3.97, 6.02 and 6.20 months at 5 °C for K. Nera, Dafni and Dossenheim, respectively.

Figure 2.

Effect of chilling period on the proportion of overlaying pupae for three Rhagoletis cerasi populations obtained from (a) Dossenheim (= 4.3x2−53.3+ 170.7; R2 = 0.83, < 0.01), (b) Dafni (= 5.3x2−63.8x + 198.2; R2 = 0.89, < 0.01) and (c) Kala Nera (= 1.5x2−11.9x + 29.6; R2 = 0.87, < 0.01).

Discussion

Testing a wide range of low temperatures, our results reveal geographical variation in diapause intensity of R. cerasi pupae. However, the difference in diapause intensity among populations is eliminated at both low (0 °C) and high (12 °C) winter temperatures. Long life cycles have been found to be a plastic response to both winter temperature and the duration of chilling period within optimal chilling temperatures. Interestingly, the chilling period affects the proportions of overlaying pupae in a dual mode: i) insufficient chilling for terminating annual dormancy (after the initial exposure to warm temperatures) and ii) extended chilling (lasting longer than required for terminating annual dormancy) that makes pupae refractory to warm cues that terminate diapause and thus ‘return’ to a dormant state by extending their life cycle. Contrary to negative effects of high winter temperatures, moderately warm temperatures, such as 8 °C, promote diapause termination in all three populations. Therefore, it seems that there is a complex interface between genetic and environmental cues that regulates diapause termination and long life cycles in R. cerasi.

Geographical variation in diapause intensity

Diapause intensity differs among the Greek and German populations within the optimal winter temperature range (5–10 °C). Specifically, the fact that pupae from K. Nera terminated diapause almost 1 month earlier than those from both Dafni and Dossenheim is in line with the phenology patterns of sweet cherry cultivars (early vs. late-flowering cultivar) in each area. It seems that R. cerasi individuals ‘count time’ during dormancy (under ambient overwintering conditions) for adjusting diapause progress to fruiting seasonality of local host cultivars. Time below threshold temperatures is comparable to the short day length of insects responding to photoperiod, which is considered as a more reliable predictor of seasonal change than temperature (Bradshaw & Holzapfel, 2007). Adults from K. Nera are known to emerge earlier than those from Dafni in local field conditions, and reciprocal transplant experiments (between these two populations in field) revealed a genetic basis to geographical variation of diapause intensity (Papanastasiou et al., 2011), implying that the above populations are not ecologically exchangeable, despite the presence of gene flow (Rader et al., 2005). Thus, the existence of local adaptation in the timing of diapause termination of R. cerasi pupae, even under conditions of ongoing gene flow, is in agreement with recent related studies regarding to the yellow dung fly and Drosophila montana (Demont et al., 2008; Tyukmaeva et al., 2011). Nevertheless, conditions experienced mostly by egg and larvae in the field before being brought back to the laboratory for collecting pupae remain uncontrolled (the initial rearing environment cannot be controlled in R. cerasi s because of lack of efficient rearing methodology; Köppler et al., 2009) and might add some variability in diapause responses. However, in our study, the infested fruits had been collected and transferred to the laboratory, where temperature was controlled; thereby, the experimental noise and possible plastic responses was substantially reduced during the last larval instar and early pupae stage (directly related to obligate diapause). Nonetheless, lack of geographical variation in diapause intensity following chilling at nonoptimal temperatures (both lower and higher) suggests phenotypic robustness under temperature stress (Stewart et al., 2012; Wagner, 2012).

Long life cycles as a bet-hedging strategy

The proportion of overlaying pupae gradually decreases until the peak of adult emergence in all treatments, suggesting that more individuals meet their chilling requirements for diapause termination in response to increasing length of chilling period, which are expected to become refractory to cues that would terminate diapause when warmed (Vallo et al., 1976). There were no overlaying pupae at peak of adult emergence at 4 °C, demonstrating that once a pupa has met its chilling requirement, it becomes susceptible to terminating diapause if it receives the proper warming cues. Thus, prolonged dormancy is not obligate for R. cerasi. Contrary to earlier predictions (Vallo et al., 1976), gradual decrease in emergence rates after peak of adult emergence was not related to increased mortality but to a gradual increase in the proportions of overlaying pupae in all three populations. It seems that if R. cerasi pupae ‘fail’ to receive the environmental cues for morphogenesis and development during the adaptive period of diapause completion and adult emergence in each area, they will become refractory to terminating diapause and ‘return’ to dormancy by expressing long life cycles (lasting more than a year). Repeated diapause has already been recorded in dragonflies (Corbet, 1956; Norling, 1971), lepidopterans (West et al., 1972) and antlions (Furunishi & Masaki, 1982), resulting usually from the instability of environmental factors, primarily food and temperature. In most cases, the facultative, ‘repeated’ diapause takes place in a subsequent developmental stage, for example early and later larval instars (Saulich, 2010). In contrast, R. cerasi pupae, having already met chilling requirements for diapause termination, respond plastically and become refractory, whenever warm temperatures (acting as environmental cues for pupal development) are absent throughout a specific ‘time window’ that ensures the timely adult occurrence in field. On the other hand, pupae that did not meet chilling requirements for diapause termination are also becoming refractory to warm temperatures and expressing prolonged dormancy. Plastic responses to environmental cues that cause phenotypes beyond the optimum result from selection for bet-hedging strategies (Scheiner & Holt, 2012).

Bet-hedging is expected to evolve under conditions of unpredictable environmental variance (Simons, 2011). In this context, it seems that the most important determinant of selection for diapause bet-hedging in R. cerasi is the multi-annual climatic variability, driven mainly by an unseasonable timing of temperature increase and winter warming. Whenever R. cerasi pupae hibernate within favourable winter temperature range, any temperature increase before the optimum chilling time for diapause termination can prolong the life cycle of some individuals, allowing some of them to emerge after a second chilling stimulus, depending on population and temperature regimes (Moraiti et al., 2012; C. A. Moraiti & N. T. Papadopoulos, unpublished data). Given that plants with an obligate chilling requirement need a minimum number of cold units before budburst, even after a sufficient long heat exposure (Harrington et al., 2010), we expect that the bet-hedgers (emerge at subsequent years) act as a ‘reservoir’ ensuring R. cerasi existence. Similarly, extended cold exposure (or differently delayed temperature increase) makes some individuals within populations refractory to warm temperatures that would allow diapause termination and ‘return’ to a dormant state in order to avoid emerging ‘out of season’ since the fruit availability of their hosts is limited in a narrow ‘time window’ each year. Individuals that ‘return’ to dormancy after an extended cold exposure are capable of emerging after a second chilling stimulus (depending on population and temperature treatments) (C. A. Moraiti & N. T. Papadopoulos, unpublished data) and therefore act as an ‘insurance’ group. When crucial resources, such as the existence of egg-laying sites, are challenged by climate change, R. cerasi distribute the risk of untimely attaining adulthood among an array of phenotypes based on difference in timing of dormancy termination, indicating diversified bet-hedging mechanisms (Simons, 2011).

Diapause response to winter warming

The phenology of host plant and herbivore, which are closely tied, can be regulated by similar temperature mechanisms (Forrest & Thomson, 2011). Temperature ranging from 3 to 5 °C satisfies chilling requirements of plants with obligate dormancy and also promotes diapause termination of R. cerasi pupae in Central Europe (Vallo et al., 1976). Focused, for first time, on chilling temperatures warmer than 6 °C, we demonstrated that R. cerasi pupae respond successfully to temperatures up to 10 °C for annual dormancy termination. Even pupae from the cold regions of Dafni and Dossenheim can successfully terminate diapause at 8 °C, a winter temperature quite high for the prevailing climatic conditions in the above areas (Table 2). Higher temperatures (8–10 °C) are more efficient for diapause termination for pupae obtained from the warmer, coastal region of K. Nera than those from the two cooler regions, suggesting an adaptive response to warm habitats. Additional studies revealed that individuals from Agia (Thessaly, Greece), a warm area close to K. Nera, are able to terminate diapause under moderately high chilling temperatures (8–10 °C), supporting that ‘warmer’ chilling temperatures up to 10 °C do not restrict diapause termination of R. cerasi pupae obtained from relatively warm areas (see Fig. S2). However, temperatures above 10 °C, which are less effective in satisfying chilling in most plants (Raulier & Bernier, 2000), hinder diapause termination for R. cerasi pupae regardless of the area they were obtained from. This could be considered as a mechanism to avoid a potential dearth of host fruit for adults either because of light crops or delaying ripening; both symptoms of prolonged dormancy caused by insufficient chilling of cherry trees (Oukabli & Mahhou, 2007; Luedeling et al., 2009; Chung et al., 2011). Interestingly, the dramatically low numbers of pupae yielding adults at 12 °C were not related to an increase in winter mortality, as it is expected for univoltine species due to energetic drain (see introduction), but they were inversely related to high proportions of overlaying pupae. We therefore expect that R. cerasi would respond to winter warming by shifting the timing of life-cycle events, instead of going extinct. On the other hand, winter temperatures as low as 0 °C are less effective than slightly warmer temperatures in satisfying chilling requirements in most plants (Harrington et al., 2010). Therefore, diapause response of R. cerasi pupae is an alignment for avoiding a potential mismatch with the phenology of host fruits for low temperature regimes as well. Our data support that temperature regimes ranging from 3 to 10 °C are optimal for diapause termination of R. cerasi pupae from at least three populations; nonetheless, populations respond differently between higher and lower temperatures, which is in accordance with plant response to extreme winter temperatures.

Conclusions

Overall, it seems that R. cerasi has evolved a mixed life-history strategy for coping with spatiotemporal environmental variation, based on a combination of local adaptation to diapause traits and diversified bet-hedging strategies. In this context, R. cerasi populations respond adaptively, even in the face of gene flow, to predicted spatial environmental variation of overwinter habitats, whereas diapause bet-hedging strategies are necessary for encountering unpredictable interannual (temporal) variation in climatic factors by ‘transferring’ to the next year(s) all those individuals that failed to emerge or did not ‘choose’ to give adults within the first year of diapause. Thus, life-cycle variability among and within R. cerasi populations is an adaptive response to the genetic and environmental cues (Leimar, 2009; Shea, 2012). In addition, we found that moderately high temperatures are stimulatory for diapause termination, but there was no linear response to low temperatures since freezing temperatures inhibit diapause termination. Studies on R. cerasi populations from a wider range of habitats may provide an additional insight on diapause intensity response to winter warming in univoltine insects.

Acknowledgments

We thank A.D. Diamantidis (University of Thessaly) and S.A. Papanastasiou (University of Thessaly) for assisting sweet cherries collection. Thanks are extended to H. Vogt (Julius Kühn–Institute, Dossenheim) and K. Köppler (Centre for Agriculture and Technology, Plant Protection in Fruit Crops, Stuttgart) for supplying R. cerasi pupae and climatic data from Dossenheim, Germany. We also thank P. Xyptera for English proofreading.

Ancillary