PEST MANAGEMENT Compatibility of Seed Head Biological Control Agents and Mowing for Management of Spotted Knapweed 1 JIM M. STORY, JANELLE G. CORN, AND LINDA J. WHITE Montana Agricultural Experiment Station, Western Agricultural Research Center, 580 Quast Ln., Corvallis, MT 59828 Environ. Entomol. 39(1): 164Ð168 (2010); DOI: 10.1603/EN09123 ABSTRACT Field studies were conducted at two sites in western Montana during 2006 and 2008 to assess the compatibility of mowing with Þve seed head insect species introduced for biological control of spotted knapweed, Centaurea stoebe Lamarck subsp. micranthos. In 2006, mowing of spotted knapweed plants at the bolting and ßower bud stages resulted in the development of new seed heads that contained signiÞcantly more seeds and signiÞcantly fewer larvae of each insect species than in seed heads in unmowed controls. No seed heads were produced in the plots mowed at the ßowering stage. Seed numbers per seed head in 2008 were also signiÞcantly higher in plots mowed at the bolting stage than in unmowed controls, but between-treatment differences in insect numbers were more variable. The seed head insects Larinus spp. and Urophora affinis Frauenfeld were the primary cause of the reduced knapweed seed numbers per seed head in 2006. Spotted knapweed should not be mowed at the bolting and ßower bud stages if large populations of seed head insects are present because mowing can result in the formation of new seed heads that are free from the insectsÕ attack, thus allowing greater seed production. Mowing of spotted knapweed at the ßowering stage and later can be conducted without a subsequent increase in seed production, but the mowing may cause mortality of the insect larvae. KEY WORDS Centaurea stoebe, seed production, mowing, Urophora affinis, Larinus spp. Spotted knapweed, Centaurea stoebe Lamarck subsp. (Story et al. 1989, 1991, 2006, 2008; Smith and Mayer micranthos (Gugler) Hayek (formerly C. maculosa 2005; Seastedt et al. 2007). These insects include two Lamarck) (Ochsmann 2001), is a perennial plant from seed head ßies, Urophora affinis Frauenfeld and U. Eurasia that has become a serious weed on rangelands quadrifasciata (Meigen) (Diptera: Tephritidae), a of the northwestern United States. First reported in seed head moth, Metzneria paucipunctella Zeller (Lep- North America in 1893 (Groh 1944), the plant now idoptera: Gelechiidae), and two seed head weevils, infests Ͼ3,000,000 ha of rangeland and pasture in 14 Larinus obtusus Gyllenhal and L. minutus Gyllenhal states and 2 Canadian provinces (Lacey 1989, Sheley (Coleoptera: Curculionidae). A combination of U. af- et al. 1998). Spotted knapweed reduces livestock and finis and the two Larinus spp. caused a 96% reduction wildlife forage (Thompson 1996, Watson and Renney in knapweed seed production in western Montana 1974), increases surface water runoff and soil sedi- (Story et al. 2008). mentation (Lacey et al. 1989), and reduces plant di- Mowing has limitations but has been used with some versity (Tyser and Key 1988). success against noxious weeds. The method can pre- Many control options are effective against spotted vent seed production, reduce carbohydrate reserves, knapweed, including herbicides, grazing, tillage, and give competitive advantages to desirable peren- mowing, and biological control (Sheley et al. 1998; nial grasses (DiTomaso 2000, Jacobs 2007). Sheley Duncan et al. 2001, Jacobs 2007). However, chemical (2002) reported that mowing of spotted knapweed at and many cultural practices are often very expensive, the ßowering stage decreased mature plant density by temporary, and usually limited to accessible areas. Ϸ85%. Watson and Renney (1974) reported that the Considerable work has been conducted on the use number of seed-producing spotted and diffuse knap- of host-speciÞc natural enemies to biologically control weed plants was signiÞcantly reduced by mowing at spotted knapweed. Twelve Eurasian insect species the ßower bud stage, the ßowering stage, or at both the have been introduced into Montana and the PaciÞc ßower bud and ßowering stage. Seed germination was Northwest for biological control of the plant. Of these, signiÞcantly reduced by mowing at the ßowering stage Þve species of seed-head insects are causing signiÞcant (Watson and Renney 1974). However, under favor- reductions in spotted knapweed seed production able conditions, spotted knapweed will produce new ßower buds after mowing. Story et al. (2008) reported 1 Corresponding author, e-mail: [email protected]. that, because ßower buds initiated after mowing gen- 0046-225X/10/0164Ð0168$04.00/0 ᭧ 2010 Entomological Society of America February 2010 STORY ET AL.: COMPATIBILITY OF INSECTS AND MOWING FOR KNAPWEED MANAGEMENT 165 erally develop after oviposition by the seed head in- mowed on 30 June ßower bud stage, Þve were mowed sects is completed, those ßower buds escape attack on 21 July (ßowering stage), and Þve plots were not and thus produce a normal complement of seeds. mowed. Fifty seed heads were collected per unmowed The objective of this study was to determine control plots on 29 July, whereas available seed heads whether mowing could be timed so that any ßower from the plots mowed in the bolting stage were col- buds produced after mowing would not escape attack lected on 1 and 4 August. No ßower buds were pro- by the seed head insects. We predicted that ßower duced in the plots mowed at the ßower bud and buds produced on plants mowed at the bolting, ßower ßowering stages. bud, or ßowering stages would escape insect attack Unrue. The Unrue site was a natural spotted knap- and thus produce a full complement of seeds. weed infestation on a south-facing slope, located 3.2 km east of Corvallis. In 2006, Þve 7 by 1-m strips were Materials and Methods mowed at a 15-cm height with hedge shears on 3 July at the ßower bud stage. Fifty seed heads from each of The study was conducted at two sites in western Þve unmowed strips were randomly collected on 4 Montana during 2006 and 2008. The study was con- August, whereas 50 seed heads from each of the Þve ducted at the Montana State UniversityÐWestern Ag- mowed plots were similarly collected on 28 August. ricultural Research Center (WARC), Corvallis, MT, The seed heads were processed in the laboratory using during both years. The study was also conducted at a the above-mentioned procedures. This site received Þeld site in 2006 (Unrue, Corvallis, MT). only natural precipitation. WARC. The WARC site, located 3.2 km northeast of Statistical Analysis. Spotted knapweed seed num- Corvallis, was a level, arable Þeld. In 2006, treatment bers and insect numbers per seed head were analyzed plots were established within a large (64 by 41 m) by analysis of variance (ANOVA) procedures, and spotted knapweed garden that received some irriga- means were compared using least signiÞcant differ- tion during the growing season. Each plot consisted of ence (LSD; Statistix 8 2003). The data were trans- a 7-m portion of a row in the knapweed garden, with formed by ͱ͑y ϩ 0.5) before the ANOVA to improve each plot randomly located in a different row. Spotted normality (Sokal and Rohlf 1981). Multiple regression knapweed plants were mowed in Þve plots on one of three dates representing different developmental analysis was used to describe the relationship between stages: 8 June (bolting stage), 29 June (ßower bud individual insect species numbers and seed numbers stage), and 20 July (ßowering stage), for a total of 20 per seed head. plots (including 5 unmowed control plots). Mowing at the bolting stage occurred when Ͼ80% of the plants were bolting, whereas mowing at the ßower bud and Results and Discussion ßowering stages was conducted when Ͼ50% of the plants contained ßower buds or ßowers, respectively. WARC. In 2006, spotted knapweed seed numbers Mowing consisted of cutting the plants 15 cm above per seed head in plots mowed at the bolting stage (8 June) and the ßower bud stage (29 June) were sig- ground with hedge shears. Fifty spotted knapweed ϭ seed heads were randomly collected from each of the niÞcantly higher than in unmowed controls (F ϭ Ͻ Þve unmowed controls on 3 August, just before the 111.0; df 2,12; P 0.001; Table 1). Seed numbers per seed heads opening for seed dispersal. The two nearest seed head in the plots mowed at the bolting stage were seed heads on every fourth plant were collected along signiÞcantly higher than in plots mowed at the ßower a transect through the plots until a total of 50 seed bud stage. As previously mentioned, no ßower buds heads had been collected. Similarly, because of the were produced in the plots mowed at the ßowering mowing-caused delay in development of the seed stage. The numbers of each of the four insect species heads, 50 seed heads were randomly collected on 15 and all four insect species combined (total insects) per August from each of the Þve plots mowed at the seed head were signiÞcantly higher in the unmowed bolting stage and on 18 August from the Þve plots control plots than in the mowed plots(for U. affinis, mowed at the ßower bud stage. No ßower buds were F ϭ 38.8; df ϭ 2,12; P Ͻ 0.001; for U. quadrifasciata, F ϭ produced in the plots mowed at the ßowering stage. 8.7; df ϭ 2,12; P ϭ 0.005; for Larinus spp., F ϭ 17.4; df ϭ The collected seed heads were individually stored in 2,12; P Ͻ 0.001; for M. paucipunctella, F ϭ 7.4; small envelopes in the laboratory until October, when df ϭ 2,12; P ϭ 0.008; and for total insects, F ϭ 52.5; df ϭ the seed heads were examined under a dissecting 2,12; P Ͻ 0.001).