Diana Pilson

INCREASED PROBABILITY OF EXTINCTION DUE TO DECREASED GENETIC EFFECTIVE POPULATION SIZE: EXPERIMENTAL POPULATIONS OF CLARKIA PULCHELLA

We established replicated experimental populations of the annual plant Clarkia pulchella to evaluate the existence of a causal relationship between loss of genetic variation and population survival probability. Two treatments differing in the relatedness of the founders, and thus in the genetic effective population size (Ne), were maintained as isolated populations in a natural environment. After three generations, the low Ne treatment had significantly lower germination and survival rates than did the high Ne treatment. These lower germination and survival rates led to decreased mean fitness in the low Ne populations: estimated mean fitness in the low Ne populations was only 21% of the estimated mean fitness in the high Ne populations. This inbreeding depression led to a reduction in population survival: at the conclusion of the experiment, 75% of the high Ne populations were still extant, whereas only 31% of the low Ne populations had survived. Decreased genetic effective population size, which leads to both inbreeding and the loss of alleles by genetic drift, increased the probability of population extinction over that expected from demographic and environmental stochasticity alone. This demonstrates that the genetic effective population size can strongly affect the probability of population persistence.

A Bt TRANSGENE REDUCES HERBIVORY AND ENHANCES FECUNDITY IN WILD SUNFLOWERS

Gene flow from transgenic crops can introduce novel traits into related spe- cies, but the ecological importance of this process is unknown. Here, we report the first empirical evidence that wild plants can benefit from a bacterial transgene under uncaged, natural conditions. Cultivated sunflower ( Helianthus annuus) is known to hybridize fre- quently with wild sunflower ( H. annuus) in the western and midwestern United States. We studied a crop-developed Bacillus thuringiensis(Bt) transgene, cry1Ac, in backcrossed wild sunflower populations. Lepidopteran damage on transgenic plants was strongly reduced relative to control plants at our two study sites, while damage by several weevil and fly species was unaffected. Our results suggest that reduced herbivory caused transgenic plants to produce an average of 55% more seeds per plant relative to nontransgenic controls at the field site in Nebraska. A similar but nonsignificant trend was seen at the site in Colorado (14% more seeds per plant). In a greenhouse experiment the transgene had no effect on fecundity, suggesting that it was not associated with a fitness cost. If Bt sunflowers are released commercially, we expect that Bt genes will spread to wild and weedy populations, limit damage from susceptible herbivores on these plants, and increase seed production when these herbivores are common.

Herbivory and natural selection on flowering phenology in wild sunflower, Helianthus annuus

Abstract Plant fitness is strongly affected by flowering phenology, and there are several ecological factors that are thought to shape the distribution of flowering times. One relatively underexamined factor is the timing and intensity of attack by herbivores that feed on flowers or developing seeds. This study tests the hypothesis that herbivores that feed on developing seeds of wild sunflower, Helianthus annuus (Asteraceae), impose selection on flowering phenology. First, the study population was found to contain genetic variation for mean date of flowering, so this trait could evolve if natural selection were operating. Next, the phenological pattern of abundance of five seed-feeding herbivores was documented. Damage by three herbivores, Haplorhynchites aeneus (Cucurlionidae), the head-clipping weevil, Homoeosoma electellum (Lepidoptera: Pyralidae), the sunflower moth, and Suleima helianthana (Lepidoptera: Tortricidae), the sunflower bud moth, was highest early in the flowering season, and declined as the season progressed. Damage by one herbivore, the seed fly Gymnocarena diffusa (Diptera: Tephrididae), was lowest early in the flowering season and increased as the season progressed. Finally, damage by two seed weevils, Smicronyx fulvus and S. sordidus (Curculionidae), whose damage was not distinguished, was constant through the flowering period. Third, damage by Haplorhynchites, Homoeosoma, and Suleima was found to be detrimental to plant fitness, suggesting that plants that flower when these herbivores are not abundant should have higher fitness. Finally, two phenotypic selection analyses were performed. The first included damage by Homoeosoma and Suleima, as well as flowering date, leaf area, and inflorescence diameter, as characters predicting plant fitness. In this analysis directional selection was found to act to decrease damage by the two herbivores, but did not act on flowering date. The second selection analysis was identical except that damage by the two herbivores was not included. In this analysis significant directional selection was found to favor later-flowering plants. Comparison of these two analyses suggests that all selection on flowering phenology is attributable to damage by Homoeosoma and Suleima: plants that flower later avoid damage by these two herbivores. While other influences on flowering phenology, such as pollination, mate availability, and seasonality, have been well documented, this study is one of few to demonstrate natural selection on flowering phenology that is a direct consequence of insect attack.

The evolution of plant response to herbivory: simultaneously considering resistance and tolerance in Brassica rapa.

Although the evolution of plant response to herbivory can involve either resistance (a decrease in susceptibility to herbivore damage) or tolerance (a decrease in the per unit effect of herbivory on plant fitness), until recently few studies have explicitly incorporated both of these characters. Moreover, theory suggests these characters do not evolve independently, and also that the pattern of natural selection acting on resistance and tolerance depends on their costs and benefits. In a genotypic selection analysis on an experimental population of Brassica rapa (Brassicaceae) I found a complex set of correlational selection gradients acting on resistance and tolerance of damage by flea beetles (Phyllotreta cruciferae: Chrysomelidae) and weevils (Ceutorhynchus assimilis: Curculionidae), as well as directional and stabilizing selection on resistance to attack by weevils. Evolution of response to flea beetle attack is constrained by a strong allocation cost of tolerance, and this allocation cost may be caused by a complex correlation among weevil resistance, weevil tolerance, flea beetle resistance, and flea beetle tolerance. Thus, one important conclusion of this study is that ecological costs may involve complex correlations among multiple characters, and for this reason these costs may not be detectable by simple pairwise correlations between characters. The evolution of response to weevil attack is probably constrained by a series of correlations between weevil resistance, weevil tolerance, and fitness in the absence of weevil damage, and possibly by a cost of tolerance of weevil damage. However, the nature of these constraints is complicated by apparent overcompensation for weevil damage. Because damage by both flea beetles and weevils had non-linear effects on plant fitness, standard measures of tolerance were not appropriate. Thus, a second important contribution of this study is the use of the area under the curve defined by the regression of fitness on damage and damage-squared as a measure of tolerance.

TWO HERBIVORES AND CONSTRAINTS ON SELECTION FOR RESISTANCE IN BRASSICA RAPA

Although most plants experience herbivory by several insect species, there has been little empirical work directed toward understanding plant responses to these simultaneous selection pressures. In an experiment in which herbivory by flea beetles (Phyllotreta cruciferae) and diamondback moths (Plutella xylostella) was manipulated in a factorial design, I found that selection for resistance to these herbivores is not independent in Brassica rapa. Specifically, the effect of flea beetle damage on B. rapa fitness depends on the amount of diamondback moth damage a plant experiences: damage by these herbivores has a nonadditive effect on plant fitness. When diamondbacks are abundant, plants that sustain high levels of damage by flea beetles are favored by natural selection, but when diamondbacks are rare, a low level of damage by flea beetles is favored. However, resistance to the later‐feeding diamondback moth is not affected by the presence or absence of damage by early‐feeding flea beetles. Thus, there are no plant‐mediated ecological interactions between these herbivores that affect the outcome of selection for resistance. Because these herbivores do not independently affect plant fitness, neither is likely to develop a pairwise coevolutionary relationship with its host. Instead, coevolution is diffuse.

COMPENSATION FOR HERBIVORY IN WILD SUNFLOWER: RESPONSE TO SIMULATED DAMAGE BY THE HEAD-CLIPPING WEEVIL

Herbivore damage is generally detrimental to plant fitness, and the evolu- tionary response of plant populations to damage can involve either increased resistance or increased tolerance. While characters that contribute to resistance, such as secondary chem- icals and trichomes, are relatively well understood, characters that contribute to a plants ability to tolerate damage have received much less attention. Using Helianthus annuus (wild sunflower) and simulated damage of Haplorhynchites aeneus (head-clipping weevil) as a model system, we examined morphological characters and developmental processes that contribute to compensatory ability. We performed a factorial experiment that included three levels of damage (none, the first two, or the first four inflorescences were clipped with scissors) and eight sires each mated to four dams. We found that plants compensated fully for simulated head-clipper damage and that there was no variation among plant families in compensatory ability: seed production and mean seed mass did not vary among treat- ments, and sire X treatment interactions were not significant. Plants used four mechanisms to compensate for damage: (1) Clipped plants produced significantly more inflorescences than unclipped plants. Plants produced these additional inflorescences on higher order branches at the end of the flowering season. (2) Clipped plants filled significantly more seeds in their remaining heads than did unclipped plants. (3) Clipped plants, because they effectively flowered later than unclipped plants, were less susceptible to damage by seed- feeding herbivores other than Haplorhynchites. (4) In later heads, seed size was greater on clipped plants, which allowed mean seed size to be maintained in clipped plants. Although there was genetic variation among the families used in this experiment for most of the characters associated with compensation for damage (seed number, mean seed size, mean flowering date, length of the flowering period, and branching morphology), in analyses of these characters, no sire X treatment interactions were significant indicating that all of the families relied on similar mechanisms to compensate for damage.

Aphid distribution and the evolution of goldenrod resistance

Although there is considerable evidence indicating that herbivory is detrimental to plant fitness, some recent studies of the evolution of plant resistance have concluded that insects do not impose selection on their host plants. A previously untested assumption that underlies most studies of the evolution of plant resistance is that insect distribution patterns are controlled directly by the effects of plant genotype on insect preference and performance. The experiments described here explicitly tested this assumption using the specialist herbivore Uroleucon tissoti (Homoptera: Aphididae) and its host plant Solidago altissima (Asteraceae). Measures of aphid preference and performance were used to predict aphid distribution patterns, and then the predicted distribution patterns were compared with the natural distribution pattern. Although goldenrod genotype had a strong effect on aphid distribution, aphid distribution was not controlled directly by the effect of goldenrod genotype on aphid preference and performance. Instead, a second experiment demonstrated that aphid and spittlebug (Philaenus spumarius and Lepyronia quadrangularis Homoptera: Cercopidae) distribution is controlled largely by genetic variation for resistance to a suite of “branch‐causing” herbivores. These herbivores induce branching and aphids and spittlebugs are more abundant on branched plants than unbranched plants. These results indicate that any natural selection imposed by aphids and spittlebugs on goldenrod will depend on the presence or absence of branch‐causing herbivores. Thus, selection for plant resistance may depend as much on the assemblage of insect species present as on the identity of each individual species.

Virus infections in wild plant populations are both frequent and often unapparent.

PREMISE OF THE STUDY Pathogens are thought to regulate host populations. In agricultural crops, virus infection reduces yield. However, in wild plants little is known about the spatial and temporal patterns of virus prevalence. Thus, pathogen effects on plant population dynamics are unclear. Prevalence data provide necessary background for (1) evaluating the effects of virus infection on plant population size and dynamics and (2) improving risk assessment of virus-resistant transgenic crops. METHODS We used ELISA and RT-PCR to survey wild Cucurbita pepo populations over 4 years for five viruses, aphid-transmitted viruses of the genus Potyvirus as a group and PCR to survey for virus-resistance transgenes. In addition, we surveyed the literature for reports of virus prevalence in wild populations. KEY RESULTS In 21 C. pepo populations, virus prevalence (0-74%) varied greatly among populations, years, and virus species. In samples analyzed by both ELISA and RT-PCR, RT-PCR detected 6-44% more viruses than did ELISA. Eighty percent of these infections did not cause any visually apparent symptoms. In our samples, the virus-resistance transgene was not present. In 30 published studies, 92 of 146 tested species were infected with virus, and infection rates ranged from 0.01-100%. Most published studies used ELISA, suggesting virus prevalence is higher than reported. CONCLUSIONS In wild C. pepo, the demographic effects of virus are likely highly variable in space and time. Further, our literature survey suggests that such variation is probably common across plant species. Our results indicate that risk assessments for virus-resistant transgenic crops should not rely on visual symptoms or ELISA and should include data from multiple populations over multiple years.

PLANT HYBRID ZONES AND INSECT HOST RANGE EXPANSION

The hybrid bridge hypothesis suggests that plant hybrids “bridge” the genetic gap between actual and potential host species, and that, for this reason, herbivorous insects are more likely to evolve an expanded host range in the presence of hybrids. While intuitively appealing, the hypothesis has two implicit assumptions: that phenotypic gaps between potential hosts limit host range, and that characters controlling host use are additively inherited in plant hybrids. Evaluation of these assumptions suggests that operation of the hybrid bridge hypothesis is relatively uncommon. In addition, the hypothesis has not been well integrated into existing theoretical and empirical work on the evolution of host range in herbivorous insects. Proper evaluation of the hypothesis will require information on the effect of plant hybridization on both insect preference and insect performance. Ecological and genetic factors affecting range expansion in both hybrids and novel parents also require evaluation.

Relative resistance of goldenrod to aphid attack changes through the growing season

Genetic variation for resistance to insect attack has been documented in many natural plant populations (Berenbaum et al., 1986; Fritz et al., 1987; Karban, 1987; Maddox and Root, 1987; Rausher and Simms, 1989), and herbivory is generally thought to be detrimental to plant fitness (Mattson and Addy, 1975; Morrow and LaMarche, 1978; Janzen, 1979; Rausher and Feeny, 1980; Marquis, 1984; Sacchi et al., 1988; but see Paige and Whitham, 1987; Maschinski and Whitham, 1989). These two observations suggest that insects may impose natural selection on their host plants (e.g., Rausher and Simms, 1989). Plants that are more resistant suffer less herbivory and have higher fitness, and these fitness differences lead to the evolution of resistance in plant populations. Nonetheless, the importance of natural selection imposed by insects on plants as a determinant of extant plant-insect associations is currently being debated (Bemays and Graham, 1988; Rausher, 1988). Some authors argue that insects are usually rare in space and time, and therefore herbivory is rarely an important determinant of plant fitness (Jermy, 1984). Others argue that even if insects are usually rare, when they do become common herbivory can be so devastating to plant fitness that they impose strong selection for resistance to herbivory in their hosts. These points of view might be reconciled by closer examination of plant resistance to herbivory and the effect of herbivory on plant fitness. For example, relative resistance to herbivory might change through the growing season. If so, which plant genotypes suffer the most herbivory will vary through the growing season, and net fitness differences, measured as seed set or cumulative growth at the end of the season, might be small. Moreover, conclusions about selection imposed by insects might depend on when resistance was measured. Suppose that the plant genotypes most resistant to herbivory early in the season tend to be less resistant late in the season, but that averaged over the whole season plants most resistant early are most resistant overall. Also suppose that herbivory is detrimental to