Present address: H. R. Dambroski, USDA/ARS/MWA, Cereal Disease Research Unit, 1551 Lindig Street, St Paul, MN 55108, USA
Jeffrey L. Feder, Department of Biological Sciences, PO Box 369, Galvin Life Science Center, University of Notre Dame, Notre Dame, IN 46556-0369, USA. Tel.: (574) 631 4159; fax: (574) 631 7413; e-mail: email@example.com
Variation in the overwintering pupal diapause of Rhagoletis pomonella appears to adapt sympatric populations of the fly to seasonal differences in the fruiting times of their host plants, generating ecological reproductive isolation. Here, we investigate what aspects of diapause development are differentially affected (1) by comparing the propensities of apple vs. hawthorn-infesting host races of R. pomonella to forgo an initially deep diapause and directly develop into adults, and (2) by determining the chronological order that R. pomonella races and sibling species break diapause and eclose when reared under standardized environmental conditions. The results imply that factors affecting initial diapause depth (and/or differential mortality during the prewintering period) and those determining the timing of diapause termination or rates of post-diapause development are both under differential selection and are to some degree genetically uncoupled in flies. The modular nature of diapause life history adaptation in Rhagoletis suggests that phenology may involve multiple genetic changes and represent a stronger ecological barrier separating phytophagous specialists than is generally appreciated.
Differences in life history timing appear to be particularly important ecological barriers separating many host plant-specific taxa of temperate insects (Smith, 1988; Wood & Keese, 1990; Feder et al., 1993; Craig et al., 1993; Pratt, 1994; Itami et al., 1998; Groman & Pellmyr, 2000). Novel hosts often differ in seasonality from natal plants. Successful colonization of these new temporal resource islands therefore requires that a univoltine insect shifts its life cycle to coincide with the timing of host availability. Most temperate zone insects possess a physiological diapause mechanism that allows them to optimize the timing of their life cycles to seasonal events (Tauber & Tauber, 1981), with reproductive isolation arising as a pleiotropic by-product of seasonal shifts.
Despite the general observation that phytophagous insects have life histories that closely match host availability, it is not always clear what specific aspect(s) of the diapause phenotype is responding to environment conditions to generate synchrony for insects with their plants. Selection may be acting directly on the timing of diapause termination during winter or rates of post-diapause development and thermal thresholds following winter. Alternatively, eclosion time could mainly be a function of the initial depth that insects enter diapause. If this is the case, then differences in eclosion time accompanying host shifts may largely be adaptive by-products of environmental selection pressures exerted by novel plants on insects prior to winter. It is also conceivable that life history shifts involve little underlying genetic change at all and represent plastic phenotypic responses to varied environmental conditions (e.g. egg hatch in Enchenopa binotata treehoppers is cued to spring sap flow in host trees; Wood et al., 1990).
Resolving the basis for diapause life history shifts has important implications for assessing the strength of phenology as an ecological barrier to gene flow. If life history shifts are mainly epigenetic and the developmental consequences of exposure of insects to different host-related environmental conditions, then phenology may not be a strong reproductive isolating barrier. Given local variation in the seasonal distribution of plants and laxness in host fidelity, gene flow is likely to be substantial between insect populations unless strong performance trade-offs exist between plants. However, if life history shifts require adaptation across several stages of diapause development, then a shift to a new plant could pose a serious phenological challenge to an insect and generate reproductive isolation that goes beyond just allochronic mating isolation.
The Rhagoletis pomonella sibling species complex provides an opportunity to test for multifaceted diapause adaptation accompanying host plant shifts. Bush (1966, 1969, 1992) has hypothesized that the various species and host races of the R. pomonella group arose via sympatric host shifts. Each fly in the complex infests a unique set of host plants that overlap broadly in their geographical distributions, but mainly differ in their fruiting times (Bush, 1966; Berlocher, 2000). For example, blueberry and deerberry (Vaccinium spp.), the hosts for R. mendax, generally fruit 2–3 weeks earlier in the season than domesticated apple (Malus pumila) varieties favoured by the apple host race of R. pomonellaPayne & Berlocher, 1995; Feder et al., 1998). This is followed c. 3–4 weeks later by preferred hawthorn species such as Crataegus mollis attacked by the hawthorn race of R. pomonella (Feder et al., 1993). Finally, Cornus florida, the host for the undescribed ‘flowering dogwood’ fly, generally fruits 2–3 weeks later than hawthorn (Smith, 1988; Berlocher, 1999).
Previous work has implied that the timing of the overwintering pupal diapause is a key life history adaptation to the seasonal differences in host fruiting phenology that reproductively isolates R. pomonella flies (Feder et al., 1993, 1994). Rhagoletis has a univoltine life cycle and a limited adult life span (c. 1 month) closely tied to host fruiting time (Fig. 1a). After feeding within host fruit, larvae emerge from abscised fruit, burrow into the soil, and form a puparium. Flies then enter a facultative pupal diapause in which they overwinter, eclosing the following summer as adults. Field and semi-controlled laboratory rearing studies have shown that R. pomonella host races and sibling species eclose at different times matching the fruiting phenologies of their respective host plants (Smith, 1988; Feder et al., 1993, 1994, 1998; Feder, 1995; Fig. 1b).
It has been hypothesized that R. pomonella phenology is predicated on a gene × environment interaction between temperature and diapause depth occurring prior to winter (Feder et al., 1997a,b; Filchak et al., 2000). The facultative nature of the R. pomonella pupal diapause is the central feature of this ‘prewinter’ hypothesis. When exposed to warm temperature for an extended period of time, a significant proportion of fly pupae will forgo a prolonged diapause and immediately develop into adults (Prokopy, 1968; Feder et al., 1997a; Note: temperature is a more important factor than photoperiod for direct development because pupae reside at a depth of 1 in or more in the soil where they are not exposed to light.). Such ‘nondiapausing’ second generation flies are at a severe fitness disadvantage in nature because they eclose at times in the late fall when little, if any, host fruit are available. Thus, it was postulated that flies attacking hosts such as apple and blueberry that fruit earlier in the field season are selected for a deeper initial pupal diapause to withstand the longer periods of warmer temperature they experience before winter (Fig. 1b). In contrast, flies infesting later fruiting hosts like hawthorn and dogwood would have faster pupal development rates to quickly attain the proper overwintering state in the face of colder fall temperatures (Fig. 1b). Carry-over effects from the initial prewinter conditions were argued to account for the subsequent eclosion time differences among flies the following summer (Feder et al., 1998; Feder & Filchak, 1999). Specifically, the warmer temperature conditions experienced by apple and blueberry pupae prior to winter were hypothesized to compensate for their genetic disposition for slower development, advancing the timing of diapause termination for these flies to an extent that they eclose as adults prior to hawthorn and dogwood flies.
Several lines of evidence provide indirect support for the prewinter diapause hypothesis. First, field data from Grant, MI indicate that the net developmental period from pupation to eclosion is actually about a week longer for apple than hawthorn flies (Feder et al., 1993). Secondly, six allozyme loci displaying significant frequency differences between R. pomonella host races and sibling species all correlate with eclosion time in a manner consistent with the prewinter hypothesis (Feder et al., 1993); the apple race possesses higher frequencies of alleles associated with later eclosion, and by inference deeper diapause, than the hawthorn race at sympatric sites in the eastern USA (Feder et al., 1993, 1997a,b; see Fig. 2 for map of fly ranges). Thirdly, laboratory experiments in which environmental rearing conditions were manipulated to emulate earlier and later fruiting plants selected for allozymes in predicted directions in both nondiapause and diapausing apple and hawthorn flies (Feder et al., 1997a,b; Filchak et al., 2000). Finally, the six allozymes associated with diapause display latitudinal variation in both host races that match variation in local thermal conditions (Feder & Bush, 1989a; Feder et al., 1990; Berlocher, 2000), with southern populations having increased frequencies of delayed development (‘apple-race’) alleles.
The current data imply, but are not a direct test for whether apple and hawthorn flies differ in their propensities for nondiapause development, as they do not indicate whether under standardized warm temperature conditions, a greater proportion of apple than hawthorn pupae remain in diapause. The studies also do not resolve whether life history adaptation is multifaceted and extends beyond initial diapause depth. Indeed, previous eclosion studies were not completely controlled for host fruit or for rearing conditions to properly access whether environmental variation is the major contributor to emergence time differences (Smith, 1988; Feder et al., 1997a,b; Filchak et al., 2000).
Here, we conduct a series of laboratory-based experiments to directly test the prewinter diapause hypothesis by (1) comparing the propensities of apple vs. hawthorn flies to undergo direct development, and (2) determining the chronological order of eclosion for R. pomonella host races and sibling species reared under standardized environmental and host fruit conditions. The prewinter hypothesis predicts that pupae from the earlier fruiting host apple and from southern populations of both host races should be less likely to undergo direct development than hawthorn or northern flies. The hypothesis also predicts that flies from early fruiting hosts and southern populations will eclose later when reared under standardized environmental conditions. If, however, the chronological order of eclosion corresponds to host phenology and apple flies are most recalcitrant to direct development, then this would imply that life history adaptation is multifaceted across different diapause stages.
Materials and methods
Previous studies investigating the eclosion characteristics of R. pomonella have revealed several features of the diapause phenotype important for understanding the current experiments. In particular, rearing flies under controlled lab conditions separated both apple and hawthorn flies into three categories of diapause development (see Fig. 3a for results for hawthorn flies reared from Grant, MI in 1989). One group of flies was found to immediately eclose as adults 25–35 days following puparia formation. Twenty-five days is near the lower limit that R. pomonella is capable of completing development (J.L. Feder, personal observation). We therefore categorize these early eclosing flies as direct developing pupae that do not enter diapause, hereafter designating them ‘ND’ for ‘nondiapausing’.
A second group of flies also does not require chilling before they eclose. However, unlike ND flies these individuals display delayed eclosion times, emerging from 37 to 65 days post-puparia formation (Fig. 3a). We hypothesize that these pupae begin to initiate diapause, but that this diapause is not deep and is overridden by continued rearing at warm temperatures. We refer to these individuals as ‘SD’ for ‘shallow diapausing’ flies.
A third group of pupae apparently does not eclose unless they experience at least a brief period of chilling at < 9 °C (Fig. 3a). We propose that these flies have entered a deep pupal diapause that requires a shift to lower temperature to terminate. None of these flies eclosed following 150 days of heating in our earlier experiments, but a proportion did emerge after chilling for 4 months at 5 °C in a refrigerator. We refer to these individuals as CD flies, for ‘chilling-dependent diapause’.
Collecting sites and predictions
To test the prewinter hypothesis, we conducted two separate experiments. The first experiment quantified the propensity for apple vs. hawthorn flies to remain in diapause when exposed to warm temperature conditions conducive to direct development (hereafter designated DDE for ‘Direct Development Experiment’). The prediction for the DDE is that the proportions of ND, SD and CD category flies will vary in a consistent host-dependent manner among the latitudinally arrayed sites (more SD and CD category flies in the apple race and southern populations vs. more ND flies in the hawthorn race and northern sites). The second experiment determined the chronological order of adult eclosion for R. pomonella host races and sibling species reared under standardized conditions (hereafter designated ETE for ‘Eclosion Time Experiment’). The prewinter hypothesis predicts that blueberry and apple flies should eclose later than hawthorn and flowering dogwood flies in the ETE.
Flies used for the DDE were collected from three pairs of ‘sympatric’ apple and hawthorn populations at Grant, MI, Fennville, MI and Urbana, Illinois (IL) (see Table 1; Fig. 2). Apple and hawthorn populations at these three sites were separated by distances < 2 km and have been shown to differ significantly from each other in allozyme frequencies (Feder & Bush, 1989a). The Grant, MI and Urbana, IL sites approximately bracket the northern and southern ends, respectively, of the latitudinal range of overlap between the apple and hawthorn host races in the mid-western USA (Bush, 1966; Fig. 2).
Table 1. Location, latitude, and collection dates for eight study sites. Site numbers correspond to those shown in Fig. 2.
DDE, Direct Development Experiment; ETE, Eclosion Time Experiment.
Brazos Bend, TX
R. pomonella n.r.
South Bend, IN
R. pomonella n.r.
R. pomonella n.r.
Flies used in the ETE came from a total of five apple and hawthorn populations of R. pomonella, as well as three populations of flowering dogwood flies and one population of blueberry flies, R. mendax (Table 1; Fig. 2). Four of the R. pomonella populations represented the same two pairs of sympatric apple and hawthorn sites from Grant, MI and Urbana, IL used in the DDE. An additional hawthorn fly population was studied from Brazos Bend State Park outside of Houston, Texas (TX). The Brazos population is representative of the southern range of the hawthorn race in the United States, which extends beyond the area of overlap with the apple race into the deep South (Fig. 2). Three populations of flowering dogwood flies from the northern [Granger, Indiana (IN)], middle [Raleigh, North Carolina (NC)] and southern [Byron, Georgia (GA)] portions of the taxon’s distribution were sampled. Only a single population of R. mendax was analysed from Sawyer, MI, which is close to the northern extreme of the fly’s range.
Fly collecting and rearing methods for the Direct Development Experiment and Eclosion Time Experiment
All flies used in the DDE and ETE were derived from natural populations collected as larvae feeding within host fruit. The infested fruit were transported back to the laboratory where they were held at 21 °C in a constant temperature chamber (15 : 9 L : D) on wire mesh racks positioned over plastic collecting trays. Emerging larvae leaving host fruit and forming puparia were collected on a daily basis from the trays. The pupae were transferred to petri dishes containing moist vermiculite and the dishes held for an additional 10 days at 21 °C before being over-wintered at 5 °C in a refrigerator. We have found that this set of rearing conditions causes minimal selection on flies. After a 5-month overwintering period at 5 °C in the refrigerator, fly pupae were removed from the cold and placed in a 21 °C incubator, set for a 14 : 10 light dark (L : D) cycle. Newly eclosing, virgin adults were collected on a daily basis and moved into 1 m (L) × 1 m (W) × 0.5 m (H) Plexiglas mass-mating cages in the incubator.
For the DDE, two mass mating cages were set up for each of the six parental populations sampled in the study (two apple × apple fly and two hawthorn × hawthorn fly cages each for Grant, Fennville and Urbana sites), as well as for two reciprocal crosses between Grant, MI apple × Grant, MI hawthorn flies. The mating cages were maintained over a 4-month period from May through August, 2003. Each mating cage contained a minimum of 20 males and 20 females, with flies replenished as they died using newly emerging adults. Each mating cage was supplied with water, food (sugar cubes and 1 : 1 slurry of autolysed brewers yeast and brown sugar) and four red delicious variety apples for female oviposition. Apples were replaced in the cages every third day. Following removal from the cages, the apples were held on wire racks over plastic collecting trays in a constant temperature chamber set at 24 °C, 15 : 9 L : D (note: hotter temperature and longer photoperiod were used in the DDE to create prewinter conditions more conducive for direct development, like those faced by flies in earlier fruiting hosts). The collecting trays were checked daily for puparia. Puparia were counted and their emergence date from fruit recorded, after which the pupae were placed into small, clear plastic SoloTM dental cups (1 oz.) (Solo Cup Company, Chicago, IL, USA) containing moist vermiculite. The cups were held for a period of 120 days at 24 °C, 15 : 9 L : D. Fly emergence was monitored on a daily basis during the 120-day prewintering period, with the number and sex of each eclosing adult recorded. After 120 days, the vermiculite in the cup was sifted using a fine, wire mesh screen to recover surviving pupae that had yet to initiate post-diapause development. Overall survivorship (the total percentage of eclosing and viable pupae at the end of the 120 prewinter treatment) and the eclosion data for populations were tested for statistical significance by Fisher exact tests. For statistical analysis, we classified individuals eclosing prior to 36 days as ND flies and before 95 days as SD flies (note: no individual eclosed in the study between 95 and 120 days post-puparium formation). The remaining viable pupae were categorized as CD flies, consistent with the results from our previous diapause studies. Because we could not nondestructively assess pupal mortality during the course of the DDE and could only determine viability based on eclosion or the dissection of puparia after the 120-day prewinter treatment, we could not estimate the proportions of ND, SD and CD flies in populations at the time of pupation. Rather, our estimates reflect the proportion of flies in these diapause categories that survived the prewinter treatment. This difference can have important consequences for interpreting of the results, allowing for the possibility that other factors related to differential mortality, in combination with diapause depth, could contribute to the distribution of ND, SD and CD phenotypes observed in the study. We enumerate on these possibilities in the Discussion section.
For the ETE, one mating cage was established and maintained over a 4-month period for each of the nine parental populations. With the exception of the Brazos Bend, TX cross, each mating cage contained a minimum of 10 males and 10 females, with dead flies being replenished daily by newly emerging adults. For the Brazos Bend cross, a total of only five males and five females could be used because of the limited size of the sample from this site. Food, water and apples in each cage were handled in the same manner as described above for the DDE. Following removal from the mating cages, the apples were held in a second constant temperature chamber set at 21 °C, 14 : 10 L : D. The collecting trays were checked daily for puparia. Puparia from each mass cross were counted, the emergence date from fruit recorded, and the daily samples of pupae from each population placed into a separate Petri-dish that contained moist vermiculite. The Petri-dishes were held in the incubator for a period of 10 days before being overwintered in a refrigerator for 20 weeks at 5 °C. Following chilling, the Petri-dishes were returned to the incubator at 21 °C, 14 : 10 L : D and adult emergence was monitored on a daily basis over a 6-month period, with the number and sex of each eclosing fly recorded.
Direct Development Experiment
General features of eclosion curves
Eclosion curves for the six parental mass cross populations were similar in overall shape to those observed in previous studies (see Fig. 3b for results of the Grant, MI apple population). As before, three different diapause classes of flies could be distinguished for each of the eclosion curves: ND flies eclosing from between 27 and 35 days post-puparium formation, SD flies eclosing from between 37 and 95 days and CD flies remaining in pupal diapause. Consequently, no feature of diapause or adult eclosion was host or population specific.
Comparisons between the host races
Paired apple and hawthorn populations showed significant differences in the distributions of flies belonging to ND, SD and CD diapause classes. Consistent with the prewinter hypothesis, the absolute proportion of CD flies surviving the 120-day prewinter treatment and remaining in diapause was significantly higher for the apple than hawthorn race at all three paired sites (Grant apple flies = 16.8%, n = 381 vs. hawthorn flies = 6.3%, n = 932; Fennville apple flies = 17.6%, n = 791 vs. hawthorn flies = 11.3%, n = 775; Urbana apple flies = 34.0%, n = 667 vs. hawthorn flies = 15.6%, n = 921; one-tailed Fisher exact tests, P < 0.0003 in all cases). These comparisons were confounded, however, by overall survivorship being higher for apple than hawthorn flies at the Grant and Fennville, MI sites (Fig. 4). To compensate for host fruit-related survivorship effects, relative percentages of CD to eclosing ND and SD flies were also compared between paired apple and hawthorn-infesting populations. Once again, a significantly higher relative percentage of surviving apple than hawthorn flies belonged to the CD class at each of the three sites (Fig. 5a) and lower relative percentage of ND flies (Fig. 5b). In addition, for those flies eclosing during the 120-day prewinter treatment (i.e. SD and ND flies), a significantly higher relative proportion of apple than hawthorn flies were SD compared with ND flies at all three sites (Fig. 5c).
Geographical pattern of variation
The latitudinal pattern of eclosion variation was consistent with the predictions of the prewinter hypothesis, as the relative percentage of CD flies increased with decreasing latitude among both apple and hawthorn populations (Fig. 5a). The same was also true with respect to the proportion of eclosing flies that belonged to SD vs. ND classes (Fig. 5c). In contrast, the relative percentage of ND flies positively varied with latitude within both host races (Fig. 5b).
Comparisons of F1 apple × hawthorn flies
Overall survivorship of F1 pupae was significantly higher than that for hawthorn flies, but lower than that for apple flies at the Grant, MI site (Fig. 6a). There was a slight, but not significant (two-tailed Fisher exact test, P = 0.0753), trend for higher survivorship in the female apple × male hawthorn fly reciprocal cross (Fig. 6a). F1 hybrids also tended to be intermediate between parental types with respect to the percentages of flies displaying CD and ND diapause development (Figs. 6b–d). However, the relative proportion of CD flies in the female apple × male hawthorn fly cross was slightly higher (30.2%) than that for apple flies (27.1%), rather than being intermediate between the host races (Fig. 6C).
Eclosion Time Experiment
Host race and sibling species differences
The chronological order of adult eclosion for R. pomonella flies reared under standardized conditions was the reverse from that predicted by the prewinter hypothesis (Fig. 7). The blueberry maggot population from Sawyer, MI eclosed the earliest, with a mean time to emergence of 58.2 ± 2.741 SE days. Apple fly populations from Grant, MI and Urbana, IL took slightly longer to eclose than R. mendax (59.6 ± 0.973 SE days and 65.2 ± 0.886 SE days respectively). The hawthorn race generally eclosed next, although there was pronounced geographical variation among different hawthorn fly populations. The mean time to eclosion was 72.6 ± 0.623 SE days for the Grant, MI hawthorn population, 79.1 ± 0.997 SE days for the Urbana, IL population, and 125.5 ± 6.062 SE days for the Brazos Bend, TX population. Nevertheless, hawthorn flies eclosed from 12 to 14 days later than apple flies at each of the two paired study sites (Grant, MI and Urbana, IL), bracketing the approximate latitudinal range of overlap of the host races in the Midwest. The 12- to 14-day difference observed between the host races under controlled conditions in the current study was therefore in line with the 8- to 13-day (mean = 10) difference previously estimated from field emergence traps at the Grant site (Feder et al., 1993; Feder,1995). Dogwood flies tended to take the longest time to eclose, although again there was significant geographical variation among dogwood fly populations (Fig. 7). The mean emergence time was 97.0 ± 1.736 SE days for the Granger, IN flowering dogwood fly population, 119.5 ± 0.725 SE days for the Raleigh, NC population and 126.7 ± 1.627 SE days for the Byron, GA population. The eclosion times of the four taxa tested therefore mirrored the sequence of fruiting phenology of their respective host plants from earliest to latest in the season of blueberry < apple < hawthorn < flowering dogwood.
Sex-related eclosion difference
Within each population, female flies tended to eclose slightly earlier than males (data not shown). However, this sex-related difference was significant for only apple flies at Urbana, IL (females = 61.1 ± 1.31 days SE, n = 144; males = 66.4 ± 1.19 days SE, n = 115; P < 0.020) and was subsumed within the difference between the host races.
The geographical pattern of adult eclosion conformed to the predictions of the prewinter hypothesis. A general linear relationship existed among apple, hawthorn and dogwood populations for flies from southern sites to take longer to eclose following heating than flies from more northern sites (Fig. 7). The difference in mean eclosion time between hawthorn populations from near the northern and southern extremes of the fly’s distribution in the United States was quite pronounced (c. 54 days), as was the case for dogwood flies (c. 30 days). Host-related diapause differences therefore appear superimposed on a large-scale spatial pattern of latitudinal variation within taxa.
The results from the Diapause Depth Experiment (DDE) were consistent with the predictions of the prewinter hypothesis. Fewer apple than hawthorn flies eclosed as ND adults at sympatric sites and higher percentages of apple flies were viable CD pupae after the 120-day prewinter period than hawthorn flies. A similar pattern was observed within the races between southern and northern populations. In a related study using field collected blueberry flies from Sawyer, MI, only c. 5% of R. mendax (n = 200) directly developed into ND adults (J.L. Feder, unpublished data) in accord with blueberry being the earliest fruiting host. Direct development rates remain to be quantified for dogwood flies, but are predicted to be high, as is seen for hawthorn flies.
Other factors in addition to diapause depth may have influenced the distribution of ND, SD and CD phenotypes in the DDE, however. For example, if hawthorn fly larvae are not as physiologically well adapted to feeding in the standardized apple host fruit used in the study as apple larvae, then hawthorn fly pupae could, in general, have been smaller and less-well provisioned for withstanding stressful prewinter conditions in the DDE than apple pupae. As a result, hawthorn pupae may have experienced increasing levels of mortality with time (number of heating days) in the DDE. Such differential mortality would account for the lower level of overall survivorship observed for hawthorn flies in the study and help generate the pattern of greater relative proportions of ND than CD hawthorn flies.
Unfortunately, because we could not nondestructively determine pupal viability in the DDE and did not measure pupal mass prior to the prewinter treatment (a difficult task because of fluctuations in moisture uptake and content of pupae), we cannot directly assess the pupal mass hypothesis at the present time. We note, however, that host-related feeding effects cannot account for the geographical trend observed within the host races for decreasing proportions of ND and increasing percentages of CD flies in southern latitudes. In addition, at the Urbana, IL site overall survivorship was similar between the apple and hawthorn races, yet hawthorn flies still displayed higher absolute and relative proportions of ND and fewer CD individuals than the apple race. Thus, it is unlikely that differential mortality because of variation in pupal mass was the sole driver of the patterns seen within and between the host races in the DDE. Moreover, previous studies have found little evidence that apple and hawthorn larvae are differentially adapted to any chemical or nutritional difference between their respective host fruits (Prokopy et al., 1988; Feder et al., 1995). It is therefore possible that the discrepancy in survivorship in the current study was because of a random, uncontrolled factor in fly rearing that happened to disproportionately affect the hawthorn race. However, the DDE measured survivorship from pupal to adult life history stages, whereas the previous studies were primarily based on egg to pupal survivorship (Reissig & Smith, 1978; Prokopy et al., 1988). The detrimental effects of feeding in alternate host fruit such as apple could therefore be more acute for hawthorn flies as pupae than as larvae. If apple-feeding hawthorn flies have reduced energy reserves for overwintering, then this could have interacted with these flies possessing shallower diapause depths to heighten their mortality and the difference in the relative proportions of ND vs. CD phenotypes in the study. The DDE may therefore have revealed a previously unrecognized relationship between host fruit quality and diapause success differentially selecting on Rhagoletis flies. Moreover, the intermediate survivorship and diapause characteristics of F1 hawthorn × apple hybrids suggest that this physiological interaction has a genetic basis, either directly or indirectly, a hypothesis requiring further testing.
The results from the ETE did not concur with the predictions of the prewinter hypothesis. Specifically, eclosion time was not found to be a simple function of initial diapause depth, as mediated by prewinter temperature. Although apple flies and blueberry flies may be more difficult to coax into nondiapause development, they nevertheless eclosed as adults prior to hawthorn and dogwood flies when they were reared under standardized conditions in the ETE. Southern populations of the same host race or species did, however, take longer to emerge than flies from northern sites. This finding is consistent with the observation that host plants tend to fruit later in the field season in southern latitudes (S. Lyons-Sobaski and S.H. Berlocher, unpublished data). The results imply that southern flies are faced (1) with having to withstand both longer, warmer prewintering periods as pupae (selecting for greater recalcitrance to direct, nondiapause development and possibly greater energy storage) and (2) with an increased number of degree days in the spring preceding fruit ripening (selecting for delayed adult eclosion) than northern flies. Apple and blueberry flies similarly have to contend with nondiapausing conditions as larvae and pupae. However, in contrast to southern hawthorn flies, apple and blueberry flies must develop quickly and eclose early the next season to maximize fruit availability.
When taken together, the results from the DDE and ETE imply that we have yet to identify a key genetic change(s) in the apple race that is acting to cause early adult eclosion. The existence of such an element(s) would help to explain several anomalies in the pattern of allozyme differentiation between the host races. For example, although all six allozyme loci displaying host-related frequency differences correlate with diapause, alleles more common to apple than the hawthorn race at sympatric sites are associated with later eclosion time, and by inference deeper diapause (Feder et al., 1993). This relationship can help explain why apple flies enter an initially deeper pupal diapause and appear more resistant to direct development during the prewinter period than hawthorn flies. However, it cannot account for the result in the ETE that apple flies eclose earlier than hawthorn flies; this finding is likely due to the action of a different gene(s) in the apple race causing these flies to terminate diapause earlier and/or develop faster or at lower temperatures post-diapause than hawthorn flies. The existence of such a locus can also account for why the allozymes explain less of the phenotypic variation in eclosion time for the apple race (14.6–17.4% among different trees at the Grant, MI site) compared with the hawthorn race (34.7–40.5% among different trees) in a semi-controlled, laboratory experiment (Feder et al., 1993). In addition, an early eclosion apple gene can account for why latitudinal frequency clines (Feder & Bush, 1989a; Feder et al., 1990); local tracking of environmental conditions at sites (Feder et al., 1993, 1998) and responses to selection in laboratory experiments (Feder et al., 1993, 1997a,b) are all less pronounced for the allozymes in the apple than the hawthorn race.
We must emphasize that our results represent the general pattern of diapause differentiation among flies. Variation in diapause can be found among local populations on regional and microgeographical scales (Feder et al., 1993; Feder, 1995). Indeed, even within a given host population collected from a single tree, variation exists among individual flies in their eclosion time. Thus, genetic variation in eclosion time exists for flies to adapt to local vagaries in the fruiting times of their host plants, as well as to differences in potential novel hosts. But the extent of this variation is subsumed by host-related and geographical differences among populations (Smith, 1988; Feder et al., 1993). For example, Smith (1988) documented significant and consistent eclosion time differences between apple and hawthorn flies from sites through the state of Illinois (mean time to eclosion for apple populations = 46.7 ± 0.52 days, range 44.5–49.3 days, n = 7 sites vs. mean time to eclosion for hawthorn populations = 72.9 ± 0.52 days, range 67.8–75 days, n = 8 sites). A similar pattern was observed within an old field near Grant, MI, in which flies from several apple trees eclosed before those from hawthorn trees by an average of 6–13 days (Feder et al., 1993; Feder, 1995).
Understanding the genetics of insect phenology is just one aspect of the diapause problem, however. It is also important to resolve how these genetic differences alter and interact with insect physiology to affect diapause development. In this regard, it is known that there are several distinct subperiods during diapause when insects are differentially sensitive to ecdysteroid induction of adult development (Denlinger et al., 1988). It is therefore possible that natural selection is acting independently on hormone titre sensitivity during these different diapause modules to produce the observed complex phenotypes in Rhagoletis. In addition, to stretch limited nutrient reserves, most insects suppress metabolism as part of the diapause programme. The degree of metabolic suppression varies among species and is greater in less active stages, such as eggs and pupae (Chaplin & Wells, 1982; Wolda & Denlinger, 1984; Wipiking et al., 1995;Irwin & Lee, 2002). Although metabolism is suppressed, warmer temperatures during diapause increase oxygen consumption and decrease nutrient reserves more quickly (Chaplin & Wells, 1982; Wipiking et al., 1995; Irwin & Lee, 2002). Warmer temperatures during diapause have also been shown to affect diapause length, survival through diapause, and post-diapause fecundity (Han & Bauce, 1998; Irwin & Lee, 2001; Ellers & van Alphen, 2002; Williams et al., 2003). These works collectively imply that the degree of metabolic suppression and temperature dependence of suppression play a role in regulating the diapause programme. They also admit the potential for strong selection on energy acquisition during the larval feeding stage both within Rhagoletis taxa and accompanying shifts to novel fruit. Further testing is therefore required to determine if life history adaptation involves apple flies undergoing greater metabolic suppression at high temperatures and larval feeding specialization to mitigate the costs of the warmer prewinter period they experience.
In conclusion, the diversification of phytophagous insect specialists and their life histories is intimately tied to their associations with host plants. Two things need to happen for an insect to radiate onto a new host plant. First, the insect must be able to utilize the nutrients in the novel host plant to support growth and reproduction. Secondly, the insect must be able to coordinate its lifecycle with the host plant. Attention has been paid to the nutritional physiology and mechanisms used to surmount host plant defences (Ehrlich & Raven, 1964; Berenbaum, 1983; Futuyma et al., 1995; Becerra, 1997). However, details underlying lifecycle synchronization and their relationship with feeding specialization and nutritional physiology have been relatively unexplored. Here, we have presented evidence for R. pomonella (1) that there may be rapid evolution of diapause physiology to accompany differences in host phenology through time and space and across host species, (2) that factors affecting the initiation and depth of diapause are likely uncoupled from those determining its termination and post-diapause development (phenology is multifaceted) and (3) that factors effecting pupal energy reserves, such as larval feeding specialization, may interact with diapause traits to affect pre- and overwintering survivorship. The first finding appears to be a common occurrence for phytophagous specialists (Berlocher & Feder, 2002). If the second finding of multifaceted diapause adaptation is also a general feature for insects, then phenology could represent a stronger ecological barrier to gene flow facilitating sympatric divergence than is currently appreciated. The third inference for a potential interaction between host-specific larval feeding adaptation and diapause was not anticipated for R. pomonella. This hypothesis requires further study to confirm and to assess the physiological and genetic nature of the relationship (e.g. directly or indirectly metabolically tied because of pleiotropy or linkage). If true, however, then it would imply that whole suites of host-related traits from feeding to life history are intertwined during shifts to novel plants, enhancing the potential for ecological reproductive isolation and speciation.
The authors wish to thank the following individuals for their assistance, moral support, and/or conversational input: S. Berlocher, G. Bush, D. Denlinger, K. Filchak, A. Forbes, D. Funk, D. Hahn, A. Michel, J. Smith, U. Stolz, S. Velez, F. Wang, F. Wang Jr, J. Wise, X. Xie, the USDA/APHIS/PPQ facility at Niles, MI, the Trevor Nichols Research Station of Michigan State University at Fennville, MI and several anonymous reviewers. This research was supported, in part, by grants from the National Science Foundation, a National Research Initiative grant from the United States Department of Agriculture and the 21st Century Fund of the state of Indiana to JLF.