Figure 1.
Study system background and experimental design.
(A) Sketches of the two host plants addressed in this study: Pedicularis semibarbata (Psem, Orobanchaceae), a perennial herb with basal rosette morphology, and Collinsia torreyi (Ctor, Plantaginaceae), a small upright annual herb. (B) Map of California highlighting E. editha populations that use Ctor (blue circles), Psem (yellow circles), or other plants (grey dots) as primary hosts (see [39] for a description of patterns of host use in this species overall). This study specifically addresses the four populations in bold: Leek Spring (LK), Tamarack Ridge (TR), Colony Meadow/Rabbit Meadow metapopulation (CM/RM), and Piute Mountain (PI). (C) Schematic of the crosses performed to produce insects examined in this study. Colored circles represent populations from which parents were sampled. Each black arrow represents a specific type of cross and leads from the native population of the female to the native population of her male mate. Bold black two-letter symbols above or below arrows symbolize the offspring produced by the cross, the first and second letters corresponding to the host affiliations of the mother and father, respectively (P, Psem; C, Ctor). Thus, the left-most arrow indicates that females from the Psem-adapted CM population were mated to males from the Psem-adapted PI population to produce PP offspring. Likewise, the middle arrows show the crosses used to produce PC offspring (with a Psem-adapted mother and Ctor-adapted father) and CP offspring (reciprocal of the previous). Grey circular arrows represent within population crosses that were not conducted by us but occurred naturally prior to the capture of mated females. These crosses produced “pure” offspring symbolized as simply P or C.
Figure 2.
Summary of traits addressed in this study.
E. editha populations adapted to Collinsia torreyi (top row) and Pedicularis semibarbata (bottom row) have diverged in the six traits illustrated—two expressed in larvae and four expressed in adult females [40]. Larval performance: Both types of larvae grow and survive well on Ctor, but only Psem-adapted larvae are able to survive on Psem. Larval foraging height: Ctor- and Psem-adapted larvae tend to feed at the top and base of their hosts, respectively. Female oviposition preference: Adult females prefer to lay eggs on the host to which they are adapted. Moreover, when forced to lay on Ctor, Ctor- and Psem-adapted females prefer individual plants at earlier and later phenological stages, respectively. Oviposition site height: Ctor-adapted females tend to lay their eggs at the top of their host near the point where they first land. Psem-adapted females invariably drop to explore the basal leaves of their host and lay near the ground. Clutch size: Ctor-adapted females lay 1–20 eggs per clutch while Psem-adapted females lay 30–100 eggs per clutch. This does not translate into a difference in either daily or lifetime fecundity since Ctor-adapted females lay more frequently than Psem-adapted females. No major phenotypic differences have been described in pupae or adult males.
Figure 3.
Different-host hybrid larvae forage at intermediate heights, slowing growth on both host plants.
(A) Mean foraging height on Ctor plants (expressed as a percentage of the height of the plant). Different-host hybrids foraged lower than larvae with 2 Ctor-adapted parents (C/CC) but higher than larvae with 2 Psem-adapted parents (P/PP). Data analyzed using an ANOVA that examined differences among major groups (C/CC versus CP/PC versus P/PP) and among cross types within groups (C versus CC, CP versus PC, P versus PP). The former effect was significant (p<0.0001); the latter was not (p = 0.3). Circles show means ± SEM. Capital letters at top of panel show Tukey's HSD groupings. Data on pure types (C and P) were previously published in [40]. (B) Mean foraging height on Psem leaves. C and CC larvae were not included since they do not accept Psem as a host. Data analyzed as in (A). (C) Mean weight of hybrid larvae allowed to forage freely on mature Ctor plants for 10 d. Variation was significant (ANOVA p<0.0001; Table S3). Capital letters at top of panel indicate Tukey's HSD groupings. (D) Growth rates of pure larvae reared on basal, middle, and top sections of mature Ctor plants in small cups. Larvae grew more quickly when fed top sections than when fed middle/basal sections. LS means come from ANOVAs described in Table S4. All means were different from one another using Tukey's HSD. (E) Growth rates of pure P larvae placed on low versus mid-high leaves of naturally growing Psem plants in the field. Lines connect data from paired groups placed on the same plant. Larvae on lower leaves grew more rapidly than those placed on mid-high leaves (Table S5).
Figure 4.
Different-host hybrid females exhibit intermediate oviposition behavior.
Colored circles show means ± SEM and small grey circles show raw data. (A) Host Preference for Ctor versus Psem. Preferences are quantified in terms of the minimum discrimination phase (d-phase), whose absolute value and sign reflect preference strength and direction, respectively. The dashed line at zero indicates no preference. Variation among cross types was significant (ANOVA p<0.0001). Capital letters at top of panel show Tukey's HSD groupings. (B) Host preference for budding versus blooming Ctor. Positive and negative d-phases indicate preferences for budding and blooming plants, respectively. Typical examples of such plants are shown in the photographs. The difference between CC and CP/PC was significant (t test p = 0.03) and is reflected by capital letters at top of panel. (C) Oviposition site height on Psem. All females were placed on leaves above 5 cm (white portion of plot); females choosing oviposition sites below 5 cm (grey portion of plot) dropped towards the ground during their exploration of the plant. Variation among cross types was significant (ANOVA p<0.0001). Capital letters at top of panel show Tukey's HSD groupings. (D) Clutch size. Variation was significant with [PC = CP]>CC on Ctor (ANOVA p<0.0001) and [PC = CP]<PP on Psem (ANOVA p = 0.003).
Figure 5.
Intermediate oviposition behavior reduces offspring fitness.
(A) Effects of clutch size and host phenology on Ctor. LS mean weights and survival ± SEM of neonate larvae left on blooming plants (black circles) and senescing plants (grey circles) in groups of various size for 10 d. Both growth and survival were higher on the blooming plants than on the senescing plants. Survival also decreased with increasing group size. LS means come from ANOVAs described in Table S6. The grey line in the right-hand panel represents a monotone cubic spline fit to the combined data from blooming and senescing plants. (B) Effects of clutch size on Psem: LS mean weights and survival ± SEM of neonate larvae left on plants in groups of various size for ∼10 d. Larvae in larger groups survived better than those in smaller groups. LS means come from ANOVAs described in Table S7. The grey line in the right-hand panel represents a monotone cubic spline fit to the data. (C) Effects of oviposition site height on egg development time on Psem: Each line connects data from low and mid/high clutches laid on the same plant. Eggs on low leaves developed more rapidly than those on mid/high leaves (Table S8).
Figure 6.
Divergence in host preference and clutch size among Psem- and Ctor-adapted populations generates strong extrinsic postzygotic isolation.
Each dot represents a hypothetical CC (blue), CP/PC (green), or PP (yellow) female. Different-host hybrids fall in an adaptive valley between the two peaks occupied by the two types of same-host hybrids, signaling the presence of significant EPI. The phenotypes of these hypothetical females were drawn from the distributions that best fit our observed data (see Methods). The scatter of the dots thus represents the biological variation and experimental noise present in our data. The shape of the surface was estimated based on field experiments. The small red arrows point to the curved edges of the surface that illustrate the relationship between offspring survival and clutch size on Ctor (foreground curve) and Psem (background curve). These curves were estimated directly from the data shown in Figure 5A and 5B and are represented in that figure as light grey lines. The front to back axis represents host preference. Thus, for females with large intrinsic clutch sizes (right-hand side of the plot), expected offspring survival goes from being relatively low for females that are most likely to lay on Ctor (foreground = positive d-phases) to relatively high for females that are most likely to lay on Psem (background = negative d-phases).