Geographical variation in adult life-history traits of the European cherry fruit fly, Rhagoletis cerasi (Diptera: Tephritidae)

Abstract

To understand the evolution of local adaptation, the interplay between natural selection and gene flow should be considered. Rhagoletis cerasi L. (Diptera: Tephritidae) is a patchily distributed, stenophagous species of the temperate zone, and the geographical structure of its populations reveals substantial variation in gene flow rates across its distribution range. We studied the demographic components of R. cerasi adults from Greek and German populations by estimating the variability in fitness traits among allopatric populations, as well as among geographically discrete populations, with gene flow. Assuming that body size may exert a profound effect on adult fitness, both thorax and head sizes were considered as covariates in our analyses. Our data demonstrated that females were larger than males in all populations, and adult size varied significantly among populations within groups. Significant differences in a suite of life-history traits of R. cerasi adults were detected among populations with gene flow, whereas there were no consistent differences among allopatric populations. Therefore, the genetic differences among R. cerasi populations, driven mainly by geographical isolation, are poor predictors of variation in the life-history traits of adults.

allopatric populations, body size, demography, divergent selection, gene flow

INTRODUCTION

Evolutionary forces, such as natural selection, gene flow and genetic drift, as well as temporal environmental variation and the genetic architecture of underlying traits, are intimately linked to an individual's capacity for local adaptation (Kawecki & Ebert, 2004). Spatial environmental variation gives rise to divergent natural selection, which enhances the intraspecific genetic differentiation (regardless of geographical distance), either directly by acting on functionally important genes and those physically linked to them or indirectly via genetic drift (Schluter, 2001; Rundle & Nosil, 2005; Nosil, Funk & Ortiz-Barrientos, 2009). Gene flow is traditionally considered as a homogenizing force that constrains adaptive divergence (Barton & Partridge, 2000; Lenormand, 2002), but this is not always the case (Garant, Forde & Hendry, 2007). Moreover, populations (especially small ones) that have experienced bottlenecks or founder effects are expected to be particularly susceptible to random differentiation via genetic drift, because natural selection becomes less effective and local adaptation is restricted (Kawecki & Ebert, 2004). Thus, the interplay between divergent selection and other evolutionary forces, particularly gene flow, is expected to shape life-history profiles of populations established in different habitats.

Considering that both survival and reproduction are tied to habitat suitability, variability in fitness traits among geographically isolated populations is a good proxy for understanding the response of a species to divergent selection. Fitness traits can also display fine-scale spatial variation, even among populations with gene flow (Reznick et al., 1997; Koskinen, Haugen & Primmer, 2002), because they might be subjected to rapid evolution (Till-Bottrand, Wu & Herding, 1990; Partridge et al., 1995). Although life-history traits are expected to exhibit low heritability (Fisher, 1958; Mousseau & Roff, 1987; Papaïx et al., 2010; but see Seko et al., 2006), their evolvability remains high (Houle, 1992), enhanced further by high levels of additive genetic variance (Fowler et al., 1997; Kruuk et al., 2000). Understanding the causes of variation in fitness traits is therefore crucial for gaining an insight into the evolution of local adaptation (Ellegren & Sheldon, 2008).

Complex traits, such as life span and reproduction, are strongly affected by increased environmental variance (Price & Schluter, 1991). The size of structures, such as the head and thorax, which are formed by discrete epidermal imaginal discs, has been reported to correlate with life-history traits, such as longevity and fecundity (Ruiz et al., 1991; Chua, 1992; Hasson et al., 1993; Sivinski, 1993; Partridge et al., 1999). Body size, which is directly related to environmental conditions, might serve as a proxy of the environmental contribution to fitness, and should be included as a covariate in the analysis of fitness data towards disentangling environmental from genetic effects (Roff, 1992; Kingsolver & Huey, 2008).

The European cherry fruit fly, Rhagoletis cerasi L. (Diptera: Tephritidae), is a key pest of economic importance for sweet and sour cherry culture, distributed in almost all European countries, Russia (European Territory and Western Siberia) and the temperate regions of west Asia (Fimiani, 1989; White & Elson-Harris, 1992; Mohamadzade Namin & Rasoulian, 2009). It is a stenophagous species infesting fruits of Prunus (Rosaceae; P. cerasus, P. avium, P. mahaleb) and Lonicera (Caprifoliaceae; L. xylosteum and L. tartarica) (White & Elson-Harris, 1992). It undergoes obligatory pupal diapause (Boller & Prokopy, 1976) that ensures the synchronization of adult occurrence at the end of spring or the beginning of summer with fruit availability.

The wide geographical distribution of R. cerasi implies that individuals of isolated populations are bound to experience different ecological conditions (habitat structure and climate), which potentially differentiate their life-history traits (Stearns, 1992). This is further enhanced by the limited dispersion activity of R. cerasi and the patchy distribution of its hosts (Phillips & Dirks, 1933; Jones & Wallace, 1955; Boller & Prokopy, 1976; Fletcher, 1989; Kneifl, Paprštein & Kňourková, 1997). However, to date, there are sparsely collected data regarding fitness-related traits for R. cerasi adults (Boller, 1966a; Boller & Prokopy, 1976), and an absolute lack of comparative demographic studies of populations obtained from different geographical areas. However, recent genetic analyses using microsatellite markers have revealed genetic structure of R. cerasi populations driven mainly by geographical isolation (Augustinos et al., 2011; A. Augustinos, unpubl. data). Given that genetic isolation arising mainly in allopatry is considered to be fundamental to phenotypic divergence, the current study, by estimating the demographic parameters of adults of R. cerasi from a wide geographical range, aims to determine whether genetic isolation driven mainly by geographical distance serves as a good predictor of divergence in adult life-history traits. We tested the hypothesis that adult life-history traits differ among genetically isolated populations across a wide geographical range, whereas, in the face of gene flow, divergence in life-history traits is rather negligible.

MATERIAL AND METHODS

Study populations

Five R. cerasi populations originating from four Greek and one German locality were used in the current study (Fig. 1). The four Greek populations were obtained from Dafni (Kozani), Pertouli (Trikala), Kato Lechonia (Magnesia) and Chania (Crete), and the German one from Stecklenberg (Harz). The genetic structure of such populations has been studied recently using microsatellite markers. Populations from Dafni, Pertouli and Kato Lechonia form one cluster with moderate rates of gene flow, whereas populations from Chania and Stecklenberg are quite differentiated, both from each other and from the previously described cluster (Augustinos et al., 2011; A. Augustinos, unpubl. data).

Figure 1.

Map showing the geographical distribution of the populations used.

Dafni and Pertouli are highland areas, whereas Kato Lechonia is coastal (Table 1). Stecklenberg is characterized by a temperate climate in the transition zone of oceanic and continental influence, with cold winters and warm, humid summers with frequent precipitation (Table 2). Summers in Chania are fresh and dry, followed by mild but rainy winters. In Pertouli and Dafni, summers are also fresh and dry, followed by cold and rainy winters. By contrast, warm and dry summers characterize the coastal area of Kato Lechonia, followed by mild and wet winters. Patterns of the ripening time of cherry cultivars and other minor hosts (P. mahaleb and Lonicera sp.; Table 3) are sound predictors of the phenology of R. cerasi adults in each area.

Table 1.

Geographical areas and fruit collection date for the five different Rhagoletis cerasi populations

Population	Habitat	Location	Altitude (m)	Fruit collection date
Dafni	Greek highland	40°17′08″N, 21°08′53″E	1050	30 June 2007
Pertouli	Greek highland	39°32′19″N, 21°27′58″E	1121	2 July 2008
Kato Lechonia	Greek coastal	39°19′49″N, 23°02′17″E	41	21 May 2007
Chania	Greek island	35°51′01″N, 24°01′07″E	450	13 June 2008
Stecklenberg	Foreland of the mountains Harz	51°73′00″N, 11°08′00″E	264	10 July 2007

Table 2.

Climatic data from the five areas in which populations were obtained

Month	Temperature [mean (min–max)] (°C)*					Mean precipitation (mm)
	Dafni	Pertouli	Kato Lechonia	Chania	Stecklenberg	Dafni	Pertouli	Kato Lechonia	Chania	Stecklenberg
Jan.	4.5 (−3.5–18.7)	1.55 (−1–4.4)	8.7 (6.1–11.7)	12.0 (9.3–15.7)	2.0 (−0.7–4.8)	36.3	204.2	39.8	96.8	48.4
Feb.	4.0 (−4.3–17.6)	0.8 (−3–4.7)	9.9 (6.5–13.6)	11.7 (8.5–15.6)	1.2 (−1.8–4.3)	20.9	216.7	29.2	85.7	43.9
Mar.	7.0 (0.3–20.6)	3.5 (−0.9–8.5)	12.9 (8.9–17.5)	13.9 (10.3–18.5)	4.1 (0.1–8.5)	56.0	177.1	23.4	34.4	49.1
April	11 (4.0–22.2)	7.7 (2.6–12.2)	15.9 (11.7–20.6)	16.3 (12.6–20.7)	9.8 (3.8–15.3)	42.3	90.0	17.0	29.1	38.2
May	15.8 (6.8–27.5)	12.9 (7.0–19.7)	21.3 (16.8–26.3)	20.0 (15.8–24.5)	13.7 (7.9–19.2)	44.9	85.4	31.1	22.6	87.6
June	18.9 (11.2–27.5)	15.8 (9.6–22.4)	25.9 (21.4–31.4)	24.9 (20.4–30.3)	17.1 (11.1–22.1)	39.5	56.9	24.8	0.1	60.7
July	23.3 (13.8–31.8)	18.1 (11.5–25.8)	28.5 (23.8–33.6)	27.0 (22.2–31.4)	19.6 (13.8–25.2)	0.0	57.0	7.4	0.0	82.5
Aug.	23.2 (15.0–34.3)	18.9 (11.7–24.4)	28.3 (23.3–32.7)	26.7 (22.1–31.4)	16.5 (11.6–21.5)	3.7	22.5	28.4	0.2	49.6
Sept.	17.5 (9.3–30.6)	13.2 (9.3–19.0)	22.6 (18.5–26.6)	23.3 (19.3–28.0)	15.4 (10.2–21.1)	43.7	101.3	34.5	14.5	82.2
Oct.	15.5 (7.2–28.9)	9.2 (5.7–14.9)	18.0 (14.6–22.5)	19.9 (16.2–24.8)	11.0 (7.0–15.6)	77.1	343.4	66.0	43.7	27.1
Nov.	13.0 (6.4–22.9)	7.7 (1.9–12.8)	14.4 (10.3−17.5)	16.8 (13.3–20.7)	6.2 (3.2–9.3)	146.1	160.8	46.5	38.0	40.9
Dec.	3.3 (−5.1–16.6)	3.7 (2.5–8.1)	10.3 (7.2−13.3)	13.5 (10.4–16.7)	3.2 (0.7–5.8)	62.1	177.3	99.6	108.9	33.1

*

Reference period: Dafni, Chania and Kato Lechonia, 2007–2010; Pertouli, 2009–2010; Stecklenberg, 2005–2007.

Table 3.

Seasonal patterns of fruit availability for each population. Grey horizontal bars denote the period of fruit availability

On the basis of genetic differentiation, geographical distance and habitat characteristics (patterns of fruit availability and climate), we formed two groups of populations. The first group (hereafter ‘allopatric populations’) included populations from Dafni, Chania and Stecklenberg, which are geographically distant and show minimal (by neutral markers) gene flow rates. The population from Dafni (representative of mainland Greece) was randomly selected from the above three mainland Greek populations. Differences in life-history traits (see Results) among allopatric populations were consistent irrespective of which population (Dafni, Petrouli or Kato Lechonia) was included in the group of allopatric populations. The second group consists of populations from Dafni, Pertouli and Kato Lechonia, which exhibit moderate rates of gene flow (hereafter ‘populations with gene flow’) despite landscape heterogeneity.

Experimental protocol

The experiments were conducted in the Laboratory of Entomology and Agricultural Zoology of the University of Thessaly from April 2007 to July 2009, at 25 ± 1 °C, 65 ± 5% relative humidity and a 14 h : 10 h light : dark cycle. Light intensity, ranging from 1500 to 2000 Lux inside the test cages, was provided by daylight fluorescent tubes and by natural light from four windows.

Rhagoletis cerasi pupae obtained from field-infested sweet cherries (P. avium L.) were collected from abandoned fields or from wild growing trees of the aforementioned areas of Greece and Germany. Infested fruits from Greece were taken to the Laboratory of Entomology and Agricultural Zoology in Volos and placed in plastic containers over a layer of dry sand (1 cm thick). Mature larvae and pupae were collected weekly. Infested cherries from Stecklenberg were taken to the Julius Kühn Institute in Dossenheim, and treated in a similar manner before being sent to Greece. Pupae were then maintained at 25 ± 1 °C for 2.5 months. At the end of this period, cohorts of 400 pupae of each population were placed in an incubator set at 4 ± 1 °C for a period of 6–8 months for diapause completion (C. Moraiti & N. Papadopoulos, unpubl. data). Finally, pupae were returned to 25 ± 1 °C until adult emergence.

A recently emerged male and female of each population were placed in individual cages containing water, adult diet (mixture of yeast hydrolysate, sugar and water at a ratio of 1 : 4 : 5, respectively) and five 18-mm, hollow, ceresin hemispheres of black colour as oviposition devices (Prokopy & Boller, 1970). Daily egg production and female and male ages at death were recorded under the above constant laboratory conditions. We ran 33–50 replicates for each of the five populations.

Morphometrics

The effect on demographic parameters of potential variation in larval diet was controlled by measuring the thorax and head sizes of every adult (female and male) and using adult body size as a covariate in data analyses. Thorax length was measured from the anterior margin of the thorax to the posterior tip of the scutellum, and the thorax and head widths were taken at their widest points. All measurements were carried out using a stereomicroscope (Leica MZ12) fitted with an ocular micrometer (1 mm = 2 ocular units).

Statistical analyses

The normality of the data was assessed with the Kolmogorov–Smirnov test. Kruskall–Wallis and Mann–Whitney tests were used to determine whether population and sex had an effect on thorax length, thorax width and head width within the two groups of examined populations (allopatric and populations with gene flow). The Cox proportional hazard model h_i(t) = h₀(t)e^ηi (model 1) (Collett, 2003) was used to assess the effect of population, sex and adult body size (thorax length, thorax width and head width) on survival patterns, as well as the effect of population and adult body size on the length of the female reproductive periods (pre-oviposition, oviposition and post-oviposition periods) within the above two groups of populations. h_i(t) and h₀(t) denote the hazard rates conditional on the observed covariates and baseline hazard, respectively. Significant factors were entered into a multifactorial Cox regression model using a forward stepwise procedure for model selection. Nonsignificant factors (Wald's test) were excluded from the final model. For example, considering the group of populations with gene flow, the estimated linear component of the Cox proportional hazard model, with population as covariate, was computed as η_i = β₁population₁ + β₂population₂ with population₁ equal to unity when an adult fly originated from Dafni, and zero otherwise, and population₂ equal to unity when an adult fly originated from Pertouli, and zero otherwise. The Kato Lechonia population was tested as baseline. The β coefficients in the linear component of the model reveal a population effect on either lifespan or duration of the reproductive periods.

To meet parametric assumptions of normality, fecundity data of populations with gene flow were sqrt(x) + 1 transformed. Analysis of covariance (ANCOVA) was employed to determine the factors that affect fecundity, with thorax length and thorax width, as well as head width, taken as covariates and population as fixed factor. Comparison of pairwise oviposition distributions was performed using receiver operating characteristic (ROC) curve analysis along the lines described by Alonzo et al. (2009). Statistical analyses were performed using SPSS 17.0 (SPSS Inc., Chicago, IL, USA) and R 2.10 (http://www.r-project.org).

RESULTS

Morphometrics

Mann–Whitney tests on adult body size estimates revealed that males were significantly smaller than females for both allopatric populations (thorax length: U = 1759.5, N = 261, P > 0.001; thorax width: U = 2280, N = 261, P > 0.001; head width: U = 2864, N = 261, P > 0.001) and populations with gene flow (thorax length: U = 1944.5, N = 295, P > 0.001; thorax width: U = 3229, N = 295, P > 0.001; head width: U = 3441, N = 295, P > 0.001) (Fig. 2).

Figure 2.

Thorax length, thorax width and head width (mm ± SE) of Rhagoletis cerasi adults originating from allopatric populations and populations with gene flow. Filled and open circles denote males and females, respectively. For comparative purposes, the data of the Dafni population are given in both columns.

There were no significant differences among allopatric populations in thorax length of males (Kruskal–Wallis test; χ² = 4.852, d.f. = 2, P = 0.088) or females (Kruskal–Wallis test; χ² = 1.633, d.f. = 2, P = 0.435). In contrast, both thorax width and head width varied significantly among populations for both males (Kruskal–Wallis test; thorax width: χ² = 42.970, d.f. = 2, P > 0.001; head width: χ² = 26.522, d.f. = 2, P > 0.001) and females (Kruskal–Wallis test; thorax width: χ² = 11.835, d.f. = 2, P = 0.003; head width: χ² = 12.473, d.f. = 2, P = 0.002).

Significant differences among populations with gene flow were found for both males and females in adult body size, except for thorax length in males (males, thorax length: χ² = 3.556, d.f. = 2, P = 0.169; thorax width: χ² = 57.202, d.f. = 2, P > 0.001; head width: χ² = 20.140, d.f. = 2, P > 0.001; females, thorax length: χ² = 10.277, d.f. = 2, P = 0.006; thorax width: χ² = 36.960, d.f. = 2, P > 0.001; head width: χ² = 16.935, d.f. = 2, P > 0.001). Adults from Pertouli were the largest relative to those from Kato Lechonia and Dafni.

Survival and longevity

The average longevity of males and females from both allopatric populations and populations with gene flow is given in Table 4. In allopatric populations, Cox regression analysis (testing each factor separately) revealed that neither population (χ² = 4.298, d.f. = 2, P = 0.117) nor sex (χ² = 3.319, d.f. = 1, P = 0.068) had a significant effect on adult longevity, as opposed to body size (thorax length: χ² = 5.032, d.f. = 1, P = 0.025; thorax width: χ² = 6.366, d.f. = 1, P = 0.012; head width: χ² = 7.277, d.f. = 1, P = 0.007). The subsequent forward stepwise procedure revealed that head width was the only significant factor. Thus, head width adequately explained the variation in longevity.

Table 4.

Average longevity and reproductive rates (net and gross fecundity) of Rhagoletis cerasi adults originating from five different R. cerasi populations

Population	Average longevity (days ± SE)		Fecundity (eggs/female)
	Males	Females	Average ± SE	Gross
Dafni	45.98 ± 2.44	57.34 ± 3.21	278.84 ± 22.3	384.16
Pertouli	60.48 ± 2.61	62.38 ± 2.87	116.84 ± 11.1	152.20
Kato Lechonia	47.58 ± 2.59	43.54 ± 3.07	138.88 ± 18.0	266.38
Chania	48.10 ± 3.40	48.20 ± 2.92	156.68 ± 16.3	220.73
Stecklenberg	42.94 ± 2.80	44.64 ± 3.59	191.67 ± 22.5	284.19

x, age interval in days; α, age at the start of reproduction; b, age at the end of reproduction; M_x, total number of eggs laid by the average female at age x; gross fecundity = (Carey, 1993).

Considering populations with gene flow, Cox regression analysis (testing each factor separately) revealed a significant effect of population (χ² = 21.003, d.f. = 2, P > 0.001), sex (χ² = 4.311, d.f. = 1, P = 0.038) and body size (thorax length: χ² = 23.552, d.f. = 1, P > 0.001; thorax width: χ² = 23.627, d.f. = 1, P > 0.001; head width: χ² = 20.766, d.f. = 1, P > 0.001) on adult longevity. Significant factors were entered into a multifactorial Cox regression model using a forward stepwise procedure for model selection. Both population (χ² = 15.159, d.f. = 2, P = 0.001) and thorax length (χ² = 17.885, d.f. = 1, P > 0.001) were significant predictors of adult longevity (Table 5). Specifically, adults originating from Pertouli were the longest lived. The hazard rate of an adult fly originating from Pertouli decreased by a factor e^−0.475 (≈ 0.622), on average, or 62%, compared with that of Kato Lechonia (baseline).

Table 5.

Effects of explanatory variables of the Cox proportional hazards model on the longevity of Rhagoletis cerasi adults from populations with gene flow. B is the model coefficient and exp(B) shows the relative risk of adult longevity at any given time

Source of variance	B ± SE	Exp(B)	P
Population			0.001
Dafni	−0.265 ± 0.146	0.767	0.070
Pertouli	−0.574 ± 0.148	0.563	0.000
Thorax length	−2.030 ± 0.480	0.131	0.000

The age-specific survival schedules for both females and males originating from allopatric populations or populations with gene flow are given in Figure 3. For females from allopatric populations, age-specific mortality rates were very low up to adult day 20, increasing progressively for older ages. However, mortality rates for females from Dafni were lower (but not significantly) than those of females from Chania and Stecklenberg. For males, survival curves for Dafni and Stecklenberg were almost identical, characterized by a rapid increase in age-specific mortality rates after the 20th day, whereas that for Chania followed a similar general pattern.

Figure 3.

Age-specific survival schedules for males and females originating from allopatric populations and populations with gene flow. Left column: thick black, thin black and grey lines denote Dafni, Chania and Stecklenberg, respectively. Right column: thick black, thin black and grey lines denote Dafni, Pertouli and Kato Lechonia, respectively. For comparative purposes, the data of the Dafni population are given in both columns.

In populations with gene flow, females from Dafni and Pertouli exhibited higher age-specific survival rates after day 10 of age than those from Kato Lechonia. For males, age-specific mortality rates were very low up to approximately adult day 20 for Dafni and Kato Lechonia populations, whereas males from Pertouli exhibited high survival rates until day 40.

Reproduction

Oviposition was initiated by day 7 and mean post-oviposition lifespan was calculated at approximately 3 days for all three allopatric populations tested (Fig. 4). Mean oviposition periods ranged from 37 to 48 days among females from allopatric populations. Specifically, females from Dafni exhibited a longer oviposition period than females from Chania and Stecklenberg. Cox regression analysis revealed that neither population (χ² = 1.326, d.f. = 2, P = 0.515) nor body size (thorax length: χ² = 0.354, d.f. = 1, P = 0.552; thorax width: χ² = 0.546, d.f. = 1, P = 0.460; head width: χ² = 0.219, d.f. = 1, P = 0.640) were significant predictors of the duration of the pre-oviposition period. Likewise, no significant effects of population or body size were detected for the duration of the post-oviposition period (Cox regression analysis; population: χ² = 2.367, d.f. = 2, P = 0.306; thorax length: χ² = 0.000, d.f. = 1, P = 0.993; thorax width: χ² = 0.537, d.f. = 1, P = 0.463; head width: χ² = 0.253, d.f. = 1, P = 0.615). Nonetheless, population (χ² = 14.633, d.f. = 1, P = 0.001) was a significant predictor of the duration of the oviposition period, as opposed to body size (thorax length: χ² = 0.137, d.f. = 1, P = 0.711; thorax width: χ² = 0.015, d.f. = 1, P = 0.902; head width: χ² = 0.081, d.f. = 1, P = 0.777).

Figure 4

Bar length (time line) depicts the average pre-oviposition (white), oviposition (black) and post-oviposition (grey) periods of females from allopatric populations and populations with gene flow. For comparative purposes, the data of the Dafni population are given in both graphs.

In populations with gene flow, oviposition was initiated by day 7 for females from Dafni, whereas it took more than 9 days for females from Kato Lechonia (Fig. 4). Moreover, the oviposition period of females from both Dafni and Pertouli was longer than that of females from Kato Lechonia. Cox regression analysis revealed that population (χ² = 8.195, d.f. = 2, P = 0.017) was a significant predictor of the duration of the pre-oviposition period, as opposed to body size (thorax length: χ² = 0.119, d.f. = 1, P = 0.730; thorax width: χ² = 0.065, d.f. = 1, P = 0.799; head width: χ² = 0.024, d.f. = 1, P = 0.878). Similar results were obtained for the duration of the post-oviposition period (population: χ² = 11.930, d.f. = 2, P = 0.003; thorax length: χ² = 2.671, d.f. = 1, P = 0.102; thorax width: χ² = 3.685, d.f. = 1, P = 0.055; head width: χ² = 0.364, d.f. = 1, P = 0.547). Cox regression analysis, testing each factor separately, revealed significant effects of population (χ² = 11.059, d.f. = 2, P = 0.004) and thorax length (χ² = 6.487, d.f. = 1, P = 0.011) on the duration of the oviposition period, as opposed to thorax width (χ² = 2.746, d.f. = 1, P = 0.097) and head width (χ² = 2.697, d.f. = 1, P = 0.101). Significant factors were included in a final Cox regression analysis, which revealed that both population (χ² = 8.502, d.f. = 2, P = 0.014) and thorax length (χ² = 4.223, d.f. = 1, P = 0.040) were significant predictors of the duration of the oviposition period.

ANCOVA, with population as a fixed factor and adult body size traits as covariates, showed that population and thorax length were significant predictors of lifetime fecundity of females from allopatric populations, as opposed to thorax width and head width (population: F = 12.772, d.f. = 2, 124, P > 0.001; thorax length: F = 5.880, d.f. = 1, P = 0.017; thorax width: F = 0.456, d.f. = 1, P = 0.501; head width: F = 0.526, d.f. = 1, P = 0.470). Females from Dafni laid more eggs than those from Chania and Stecklenberg (Table 4). In addition, females from Dafni had larger oviposition distributions than those from Chania (ROC analysis, P > 0.001) and Stecklenberg (P > 0.011). The oviposition distributions of females from Chania and Stecklenberg were similar (P = 0.511).

In populations with gene flow, females from Dafni were more fecund than those from Pertouli and Kato Lechonia (Table 4). ANCOVA, with adult body traits (thorax length, thorax width, head width) as covariates and population as a fixed factor, demonstrated that only population was a significant predictor of the lifetime fecundity (population: F = 26.530, d.f. = 2, 140, P > 0.001; thorax length: F = 0.835, d.f. = 1, P = 0.362; thorax width: F = 0.685, d.f. = 1, P = 0.409; head width: F = 2.086, d.f. = 1, P = 0.151). However, the distribution of oviposition days was similar between females from Dafni and Pertouli (ROC analysis, P = 0.978), Dafni and Kato Lechonia (P = 0.206), and Pertouli and Kato Lechonia (P = 0.235).

Age-specific patterns of egg laying revealed subtle differences among allopatric populations (Fig. 5). Peak oviposition rates were recorded at five eggs per female per day between days 20 and 27. Soon after peaking, there was a gentle decrease in egg production for both the Dafni and Chania populations, and a steeper decline for females from Stecklenberg. Details of the age-specific oviposition rates are depicted in the event history graphs (Carey et al., 1998) in Figure 6.

Figure 5.

Age-specific schedules of reproduction for females from allopatric populations and populations with gene flow. For comparative purposes, the data of the Dafni population are given in both columns.

Figure 6.

Event history diagram of females from allopatric populations and populations with gene flow: green, 0 eggs/day; yellow, 1–5 eggs/day; red, 5 eggs/day; 33–50 individuals were tested in each treatment.

Substantial differences among populations with gene flow were reported for the peak and location of the age-specific oviposition schedules (Fig. 5). Oviposition rates peaked between days 20 and 27 for females from Dafni, but showed no well-defined peak for females originating from Pertouli and Kato Lechonia. Differences in age-specific oviposition rates among populations with gene flow are also highlighted in the event history graphs in Figure 6. Oviposition effort for females from Dafni was more intensive and continuous than that for those from Kato Lechonia, whereas females from Pertouli exhibited rather sporadic and irregular oviposition.

DISCUSSION

The current study provides an extensive analysis of adult demographic traits of several populations of the European cherry fruit fly. Our results demonstrate significant variation in demographic traits among both allopatric populations and populations with gene flow. Significant differences were also detected in adult size. Within-population comparison showed that males were smaller than females; however, average longevity was similar for the two sexes. Differences among allopatric populations were detected in lifetime fecundity and the duration of the oviposition period, whereas variation in adult longevity was controlled by the variation in body size (head width). There were no differences in the duration of pre- and post-oviposition periods. In the face of gene flow, adult longevity, lifetime fecundity rates and the duration of pre-oviposition, oviposition and post-oviposition periods varied significantly among populations. Therefore, habitat heterogeneity (duration of fruit availability) seems to be a better predictor of the divergence in life-history traits of adults of R. cerasi populations than is genetic isolation caused by geographical distance.

Adult size

Morphological traits respond to environmental variation (de Jong, 1995; Debat & David, 2001; Liefting, Hoffman & Ellers, 2009), and developmental plasticity enhances body size differences among individuals that experience different temperature, nutritional (Davidowitz, D'Amico & Nijhout, 2004; Salazar-Ciudad, 2007; Shingleton, 2010) and host size (Lafferty & Kuris, 2002; Bonal & Muñoz, 2009) regimes. In our study, once the infested fruits had been collected and transferred to the laboratory, temperature was controlled; thereby, the experimental noise was substantially reduced during the last larval instar, which is an important stage for traits related to adult size. Nonetheless, the fruits of different sweet cherry cultivars might vary in sugar/amino acid ratio caused by both genetic and environmental factors (Serrano et al., 2005; Wang, 2006; Karlidag et al., 2009). As reported earlier (Boller, 1966b), cherry fruits with low sugar and high amino acid concentrations provoke slow larval growth rates in R. cerasi. Consequently, plastic adjustments of larval growth rates to the nutritional status of the host fruit may be a source of variation in body size of R. cerasi adults.

Life-history traits

To date, a large amount of information has been accumulated regarding the life-history evolution of the multivoltine, polyphagous species of Tephritidae (Carey, 1984, 2011; Carey, Harris & McInnis, 1985; Papadopoulos, Katsoyannos & Carey, 2002; Joachim-Bravo et al., 2003; Diamantidis et al., 2009; Duyck et al., 2010). However, little is known about the demography of multivoltine oligophagous species (Brèvault, Duyck & Quilici, 2008) and, particularly, univoltine stenophagous species (Wiesmann, 1933; Leski, 1963; Boller, 1966a; Boller & Prokopy, 1976; Kasana & AliNiazee, 1994). With regard to R. cerasi, our data showed that the average adult longevity ranges from 42 to 62 days among the populations tested, being approximately 10 days longer than the estimated average longevity in the wild (Wiesmann, 1933; Böhm, 1949). Lifespan was equal for both sexes in R. cerasi, similar to other Rhagoletis species (Kasana & AliNiazee, 1994). In contrast, other species of the family Tephritidae exhibit a rather male-biased adult longevity (Foote & Carey, 1987; Carey & Liedo, 1995; Carey et al., 1995; Vargas et al., 2000; Diamantidis et al., 2009). For example, in the medfly (Ceratitis capitata), longer male longevity has been attributed to the sex-specific reproductive biology, suggesting that the polygynous males increase their mating opportunities by ‘extending’ their lifespan, whereas this is not the case for the oligo-androus females (Carey & Liedo, 1995; Carey et al., 1995). In R. cerasi, both males and females mate throughout their life, and it seems that there is no sex-specific mating advantage that could account for selection for longer male longevity. Lifetime fecundity rates of R. cerasi females ranged from 117 to 279 eggs per female. Similar fecundity rates have been reported earlier (Boller, 1966a). Nonetheless, the fecundity rates are expected to decrease dramatically (30–100 eggs per female) under field conditions (Wiesmann, 1933; Leski, 1963). Our data showed that R. cerasi, similar to other stenophagous Tephritids, has a lower reproductive output than polyphagous species (Papadopoulos et al., 2002; Carey et al., 2005; Ekesi, Nderitu & Rwomushana, 2006; Brèvault et al., 2008; Diamantidis et al., 2009). The short availability of patchily distributed host fruit seems to account for the lower reproductive outputs of stenophagous tephritids compared with polyphagous species that are capable of flying longer distances to find alternative hosts (Fletcher, 1989).

With regard to the group of allopatric populations, differences in adult longevity were related to differentiation in head width. Population origin was not a significant predictor of adult longevity. Environmental canalization on life-history traits caused by stabilizing selection, which has been demonstrated previously in Drosophila melanogaster (Stearns & Kawecki, 1994), may weaken differentiation among populations. Differences in fecundity rates and oviposition period were found only among females from Dafni and those from Stecklenberg and Chania. Differentiation between Chania and Dafni populations is in line with the hypothesis that island populations are phenotypically differentiated from their mainland conspecifics (Grant, 1998). However, fitness traits did not differ significantly between adults from Chania and Stecklenberg, despite genetic and geographical distance. Differentiation in fitness traits between the above two populations remained insignificant even when populations from mainland Greece (genetically distinct from both Chania and Stecklenberg), other than the Dafni population, were included in the analyses (data not shown). It is therefore plausible to argue that other evolutionary force(s) that interplay(s) with divergent selection (at different habitats) ultimately determine the fitness output of R. cerasi adults. Future studies should be implemented to precisely address the evolutionary forces that restrict life-history divergence between Chania and Stecklenberg populations.

Considering the group of populations with gene flow, we found considerable differences not only in reproductive periods and fecundity rates, but also in longevity. Adults obtained from pupae collected from Pertouli were the longest lived and, in addition, the oviposition periods of females from Pertouli and Dafni were longer than that of those from Kato Lechonia. In the area of Pertouli, wild growing sweet cherry and P. mahaleb trees with overlapping fruiting patterns extend fruit availability throughout, and even well beyond, the R. cerasi flight period. The availability of oviposition resources, which is the longest in Pertouli, may account for differences in longevity. However, plastic responses to larval diet (see variation in thorax length) may additionally account for differences in both adult longevity and oviposition period within populations. Spatial variation in time periods from female emergence to the ripening of sweet cherries or wild growing sweet cherries (oviposition sources) may determine the variation in the pre-oviposition period, which was shorter for females from Dafni than for those from Pertouli and Kato Lechonia. Taking into account the fact that post-oviposition life is a life-history trait without any plausible effects on fitness (Reznick, Bryant & Holmes, 2006), the extended post-oviposition period of females from Pertouli remains to be explained. An extended post-oviposition period characterizes individuals that allocate energy in favour of maintenance over reproduction. In this sense, flies from Pertouli that exhibited the longer post-reproductive lifespan and the lowest lifetime fecundity might have reached their reproductive potential under the benign laboratory conditions. Considering fecundity rates, females from the highland Dafni and Pertouli areas exhibited the highest and lowest lifetime fecundity, respectively. Either way, the inconsistency of the fecundity patterns of the females from Pertouli, in terms of the observed false starts and the numerous zero-egg days, indicates insufficiency in reproductive resources (Novoseltsev, Novoseltseva & Yashin, 2003).

Although gene flow is traditionally thought to be antagonistic to population differentiation (Kirkpatrick & Barton, 1997), recent studies have demonstrated that it can also facilitate adaptation to local environmental conditions (Gandon & Nuismer, 2009; Ribeiro, Llord & Bowie, 2011). Gene flow among populations experiencing different environmental conditions can preserve relatively large amounts of standing variation, and thus replenishes local genetic variability, providing new material for selection pressures to act upon (Alleaume-Benharira, Pen & Ronce, 2006; Barrett & Schluter, 2008). Therefore, the moderate rates of gene flow enhance the adaptive potential of R. cerasi populations (occupying different habitats in a fragmented landscape) to local habitats, and therefore prevent them from extinction by genetic processes (Lynch, Conery & Burger, 1995; Crnokrak & Roff, 1999).

Gene flow may override genetic differentiation at all loci, except those that are directly under selection or are linked to selection (Via, 2009). In polygenic characters (with many loci of small effects), such as fitness-related traits, strong divergent selection can promote rapid adaptation through a spread of small allele frequency shifts across many loci. However, polygenic adaptation does not produce classical signatures of selective sweeps and, accordingly, it usually goes undetected using conventional methods (Pritchard & Di Rienzo, 2010; Pritchard, Picknell & Coop, 2010). Therefore, understanding the adaptive potential of R. cerasi populations is critical to assess the genomic basis of adult life-history traits. Studies that combine quantitative genetics, such as quantitative trait locus (QTL), and population genomics, such as next-generation sequencing (NGS), would provide a powerful tool to identify genes controlling recent adaptive change in R. cerasi (Stapley et al., 2010).

Overall, our results suggest that spatial patterns in genotypic variation in neutral loci are not related to spatial patterns of life-history traits of R. cerasi adults. Not only were differences in life-history traits among allopatric populations of R. cerasi adults inconsistent, there was also substantial variation in a suite of traits among populations with gene flow. In the context of ecological adaptation, divergence in life-history traits of R. cerasi adults in the absence of detectable differentiation in neutral markers indicates divergent selection trajectories as a result of interactions with local habitats. Nevertheless, other evolutionary forces, such as genetic drift, might hinder divergence in adult life-history traits among allopatric populations. Hence, our results are in line with the theoretical prediction that neutral genetic differentiation may not reflect biologically meaningful differences (Hedrick, 1999; McKay & Latta, 2002). This is further supported by several examples in plants (Karhu et al., 1996; Gonzalo-Turpin & Hazard, 2009), frogs (Palo et al., 2003; Knopp et al., 2007; Richter-Boix et al., 2010), fishes (Conover et al., 2006; Hutchings et al., 2007; Hankinson & Ptacek, 2008; Nielsen et al., 2009), birds (Ballentine & Greenberg, 2010) and insects (Andersen et al., 2008).

Overall, this is the first study that has assessed systematically the adult demographic traits of R. cerasi populations from a wide range of temperate regions, and revealed differences in lifespan and reproduction. Estimation of the life table parameters of different R. cerasi populations provides fundamental information for the construction of area-specific population models (Carey, 2001). To completely map the geography of variation in life-history traits of R. cerasi adults, populations from more temperate regions should be examined, and other biological traits, such as diapause intensity, should be considered. It is also worth noting that temporal variation in gene flow during the early stages of adaptive divergence has been reported recently for spring-spawning salmonid populations from different habitats (Junge et al., 2011). Hence, it is strongly recommended that the spatial structure of the examined R. cerasi populations based on neutral markers should be assessed for a few consecutive years in order to detect potential fluctuating levels of gene flow, indicating initial stages of adaptive divergence. However, the relationship between genotype and phenotype can only be adequately assessed if we reveal the genomic signatures underlying ecologically relevant adaptation.

ACKNOWLEDGEMENTS

We thank Freerk Molleman (University of Tartu, Estonia) and Antonis Augustinos (University of Patra, Greece) for their constructive comments on an earlier draft of the manuscript, and Alexandros Diamantidis (University of Thessaly, Greece; and University of California Davis, USA) and Penelope Mavragani-Tsipidou (Aristotle University of Thessaloniki, Greece) for assisting in sweet cherry collection from Dafni, Kato Lechonia and Chania. Thanks are also extended to E. Kaiktsi (University of Thessaly, Greece) for technical assistance in laboratory experiments. The comments of three anonymous reviewers greatly improved the manuscript. This study was partially supported by an IKYDA (Programme IKYDA, Greek State Scholarship Foundation and German Academic Exchange Service) grand awarded to Nikos Papadopoulos (University of Thessaly, Greece) and Heidrun Vogt (Julius Kühn Institute, Germany).