Biological Control 26 (2003) 270–278 www.elsevier.com/locate/ybcon Plant size preference of Agapeta zoegana L. (Lepidoptera: Tortricidae), a root-feeding biological control agent of spotted knapweed L. Smitha,* and J.M. Storyb a USDA-ARS, Western Regional Research Center, 800 Buchanan Street, Albany, CA 94710, USA b Montana Agricultural Experiment Station, Western Agricultural Research Center, Corvallis, MT 59828, USA Received 26 December 2001; accepted 22 October 2002 Abstract Agapeta zoegana L. (Lepidoptera: Tortricidae) is an oligophagous herbivore that was introduced to North America as a bio- logical control agent of spotted knapweed, Centaurea stoebe L. subsp. micranthos (Gugler) Hayek (often called Centaurea maculosa Lam.). Spotted knapweed is a perennial plant that usually increases in size each year. A previous field study reported that more larvae were found on larger plants and that infested plants tended to be larger than uninfested ones. Precisely quantifying the size- specific attack rate can help us model the impact of this agent on the weed population and better understand the interspecific in- teractions to improve the effectiveness of biological control. Field data were analyzed to determine the relative preference of attack for each size class of the host plant. Plants were classified based on root diameter at 2 cm below the root crown. Although small plants (<3 mm root diameter) were more abundant in the field population, the highest infestation rates occurred in large plants. ChessonÕs electivity index was generally positive for root diameters >3.5 mm, indicating preferential attack of large plants. Because of its host-size preference, A. zoegana is expected to primarily affect large plants, which is contrary to previous expectations. Quantifying the insectÕs direct impact is a difficult challenge, which may require several field seasons of measuring accumulated damage on individual plants. In order to complement the biological control agents already established, foreign exploration should focus on finding an agent that attacks young knapweed plants. These results also indicate that the efficiency of sampling roots in the field to detect the presence of A. zoegana can be improved by choosing only the largest plants. Published by Elsevier Science (USA). Keywords: Biological control; Herbivore; Preference; Spotted knapweed 1. Introduction suspected to have arrived in North America as a con- taminant of alfalfa seed from Asia Minor, especially Spotted knapweed is an important rangeland weed in Turkmenistan, or from Germany (Maddox, 1982; the western United States (Sheley et al., 1998, 1999) that Ochsmann, 2000). Twelve species of introduced insect has been a target for biological control (Lang et al., biological control agents of this weed have become es- 2000; Schroeder, 1985; Smith, 2001; Story, 1995). This tablished in North America (Rees et al., 1996), and some plant has often been reported in North America as are now being evaluated in the field for impact (Smith, Centaurea maculosa Lam. (Asteraceae), but should more 2001). properly be called Centaurea stoebe L. subsp. micranthos The root-feeding moth, Agapeta zoegana L. (Lepi- (Gugler) Hayek (Ochsmann, 2000) because it is peren- doptera: Tortricidae), was introduced from Austria and nial, polycarpic, and tetraploid, whereas C. maculosa is Hungary (Story et al., 1991) and has become well es- biennial, monocarpic, and diploid. Spotted knapweed is tablished in western Montana. Several attempts have been made to measure its impact on spotted knapweed * Corresponding author. Fax: +510-559-5737. plants (Callaway et al., 1999; Clark et al., 2001; Muuller-€ E-mail address: [email protected] (L. Smith). Schaarer,€ 1991; Steinger and Muuller,€ 1992; Story et al., 1049-9644/02/$ - see front matter. Published by Elsevier Science (USA). doi:10.1016/S1049-9644(02)00169-X L. Smith, J.M. Story / Biological Control 26 (2003) 270–278 271 2000). However, impact of this agent has been difficult Although there is a substantial literature on host size to measure, even in laboratory studies (Muuller,€ 1989). preference for insect parasitoids and predators (e.g., Also, it has been suggested that mycorrhizal interactions Godfray, 1994; Jervis and Kidd, 1996; Smith, 1993), between spotted knapweed and neighboring grasses little has been reported for herbivores. However, some could cause this agent to harm competing perennial herbivorous insects have been shown to exercise size grasses more than spotted knapweed, on which it feeds preference when ovipositing in flowers (Fondriest and (Callaway et al., 1999; Marler et al., 1999). Nevertheless, Price, 1996), fruits (Messina, 1990), stems (Pires and Story et al. (2000) measured a 43% reduction in above Price, 2000), leaves (Stuart and Sridhar, 1998), or plants ground biomass per plant and a 43% reduction in the (Langan et al., 2001; Tinney et al., 1998). There is some number of capitula per plant by comparing field plots variation in the literature regarding usage of the terms that had high versus low infestation rates of this agent. ‘‘preference’’ and ‘‘electivity’’ (Singer, 2000), but here we However, rosette densities did not decrease during the follow definitions used by Manly et al. (1972) and 2.5 year study. Chesson (1983). Preference reflects the relative proba- Measuring impact of this agent is difficult because the bility that a host of a particular type is attacked when all direct damage to mature plants generally appears to be types are equally abundant. Preference can include dif- sublethal (Muuller,€ 1989). The plant is perennial and has ferent components, such as active searching behavior by the potential to recover from damage. Environmental ovipositing females, ability of larvae to enter the host, variation and the complexities of ecological interactions and of larvae to survive until the time of sampling. In a make it difficult to measure impact on individuals in field study, the relative proportion of host plants of a field studies (e.g., reduction of seed production or sur- given size that are infested is a function of the relative vivorship). Reduction of knapweed plant densities can abundance of plants of that size and the preference of reduce plant competition causing an increase in size and the insect for that plant size. The electivity index fecundity of individual plants (Myers et al., 1990). (Chesson, 1983) adjusts for unequal numbers of plants Furthermore, seeds buried in the soil can persist up to 7 of the different size classes. Thus, the electivity index years (Davis et al., 1993), causing a lag before reduction provides an estimate of the relative preference for each of seed production affects seedling recruitment. Multi- host plant size class that is unbiased by the relative year, field population studies are confounded by dis- abundance of the size classes. persal of agents into check sites or by the effects of cages The purpose of this study is to analyze field data from or insecticides used to exclude them (Luck et al., 1999). Story et al. (2000) to describe the host plant size pref- The more we can learn about the components of the erence of spotted knapweed plants attacked by A. zo- plant–insect interactions, the better we can understand egana. Size preference is important because if insects the impact of the biological control agents. Efforts to preferentially attack plants of a particular size (e.g., model these interactions have begun to provide a basis small plants) then the direct impact will be limited to for understanding the effects of these interactions on such plants. Understanding which plants are most likely population dynamics (Cloutier and Watson, 1990; to be infested will improve our ability to: (1) monitor Jacobs and Sheley, 1998; Myers and Risley, 2000). direct impact and (2) predict the ultimate impact on Relatively little is known about the field ecology of A. population demographics of the target weed. zoegana. Adults live only a few days, and females ovi- posit on spotted knapweed leaves and stems and on nearby vegetation (Muuller€ et al., 1988). Larvae attack 2. Materials and methods the root crown, initially feed on the root cortex, and eventually burrow into the center of the root, especially 2.1. Experimental procedures on small plants. They overwinter as small larvae, com- plete development the following spring, and emerge as The study was conducted in an 8-ha, level field on the adults in summer. Story et al. (2000) discovered in a field Teller Wildlife Refuge near Corvallis, Montana (see study that the number of larvae per root was not sig- Story et al., 2000 for more description). The field was an nificantly different from that predicted by random at- abandoned pasture dominated by spotted knapweed tack (binomial model), suggesting that there was no (59% of vegetation cover), grass species (35%), and other tendency for the moth to avoid plants already infested. forbs (6%). Spotted knapweed density was 383 Æ 24 (SE) However, there was a positive relationship between the plants/m2 (excluding seedlings). A total of 788 and 2080 number of larvae per infested plant and root diameter, A. zoegana adults were released at the site in 1988 and regardless of whether the plant was in bolted or rosette 1989. All the knapweed plants were collected in each of stage. These results suggest that the insect may prefer to 16 randomly placed plots (50 Â 50 cm) during each of attack large plants, but no data were presented regard- five periods: June and October of 1992 and 1993, and in ing the proportion of plants attacked for different plant June 1994 (i.e., total of 80 plots) to study establishment sizes. and impact of the insect on the knapweed population 272 L. Smith, J.M. Story / Biological Control 26 (2003) 270–278 (Story et al., 2000). Data were collected on the root those >8 mm were combined into one class. The average diameter (measured 2 cm below the root crown) and number of plants in a class was 134 (range: 32–374).