The parasitoid Cotesia glomerata (L.) (Hymenoptera: Braconidae) was introduced to the United States as a biological control agent against the invasive vegetable pest Pieris rapae (L.) (Lepidoptera: Pieridae) in 1884 near Washington, District of Columbia (Clausen 1978). Cotesia glomerata is a gregarious endoparasitoid of several species of pierid butterflies. Although C. glomerata established, it was unable to reduce damage from P. rapae larval feeding to a level acceptable to vegetable growers. Cotesia glomerata kills P. rapae larvae at the end of the fifth instar, after most larval feeding has occurred. In fact, larvae parasitized by C. glomerata consume significantly more food during their development than unparasitized ones (Rahman 1970). Thus C. glomerata pest control benefit is limited to intergenerational reduction in P. rapae density, which has not been sufficient to reduce P. rapae to non-pest status in the United States. Also, Cotesia glomerata is not host specific, and has non-target impacts on native pierid butterflies, including Pieris oleracea Harris (formerly Pieris napi oleracea) (Benson et al. 2003).

Cotesia rubecula (Marshall) (Hymenoptera: Braconidae) is a solitary host specific endoparasitoid of P. rapae that attacks first and second instars. Cotesia rubecula not only attacks P. rapae at a high rate (e.g., Van Driesche 2008), but also reduces feeding damage on a per larva basis (Le Masurier & Waage 1993). Cotesia rubecula is successful at reducing feeding damage because it kills P rapae in the fourth instar, before most larval feeding occurs. Also, because C. rubecula is host specific, it rarely attacks native pierids in the field (Van Driesche et al. 2004).

There have been several introductions of C. rubecula into North America since the 1960s. A population of C. rubecula that was not deliberately introduced was detected on Vancouver Island, British Columbia in 1963 (Wilkinson 1966). By the 1980s, this strain had spread as far south as Oregon and displaced C. glomerata there, but did not do so below latitude 44°35' (Biever 1992). The Vancouver strain of C. rubecula was later released in Ontario, Missouri, New Jersey, and South Carolina in the 1960s (Puttier et al. 1970; Williamson 1971, 1972). The Vancouver strain of C. rubecula established in Ontario (Corrigan 1982), but failed to establish in more southern areas, including Missouri (Parker & Pinnell 1972).

It was suggested that this strain failed to establish more southern areas because its diapause requirements were not met (Nealis 1985). To overcome this problem, a strain of C. rubecula from the former Yugoslavia was introduced in the 1980s to Ontario, Missouri, and Virginia (McDonald & Kok 1992). In 1988, the Yugoslavian strain of C. rubecula was recovered in Virginia, but it did not persist. This may have been due either to its diapause requirements not being met or the negative effects of hyperparasitism (McDonald & Kok 1992; Gaines & Kok 1999). In a third attempt to find a climatically adapted population, C. rubecula was collected in Shenyang, China, in 1988,and this strain was released in 17 locations in southern New England (Van Driesche & Nunn 2002), where it established and spread. In the early 1990's, individuals from both the former Yugoslavian and Chinese populations were released in Minnesota and C. rubecula recoveries were made beginning in 2000 (Wold-Burkness et al. 2005; Lee & Heimpel 2005).

Before the release of C. rubecula in New England, the dominant parasitoid of P. rapae was C. glomerata (Van Driesche & Bellows 1988). By 2002, C. rubecula was widely distributed in southern New England, and had become the dominant parasitoid of P. rapae (Van Driesche & Nunn 2002). In western Massachusetts, Ontario, and the western United States C. rubecula has outcompeted and displaced C. glomerata (Corrigan 1982; Biever 1992; Van Driesche 2008).

The purpose of this study was to assess the current geographical distribution of C. rubecula and C. glomerata in the northeastern and north central parts of the United States and adjacent parts of Canada in order to determine if C. rubecula has displaced C. glomerata at this scale as it has done locally in New England. We hypothesized there would be a southern limit to the spread of C. rubecula due to an incompatibility between local seasonal day length patterns and diapause cue sensitivity of the parasitoid, as suggested by Nealis (1985). We also hypothesized that C. rubecula would displace C. glomerata over some larger spatial scale given that it has done so in New England, Ontario, Washington, and Oregon (Corrigan 1982; Biever 1992; Van Driesche 2008).

MATERIALS and METHODS

Samples of P. rapae and Cotesia parasitoids were collected from May to late Sep 2011 in 14 states and 2 Canadian provinces, from New England to North Dakota, southward to North Carolina and northward to New Brunswick and Quebec. Samples were collected from various types of cole crops at organic vegetable farms or private gardens. All P. rapae larvae from first to fifth instars, as well as pupae, and cocoons of both species of Cotesia parasitoids (emerged or not) were collected. Collectors were provided with pictures and descriptions of these life stages. Up to 1 h was spent examining crop plants, collecting all of the above life stages until 30 or more “individuals” (one C. glomerata cocoon mass was considered one individual, as it came from one host larva) had been collected. Actual sample numbers per site ranged from 5–103 individuals, depending on local P. rapae density. First and second instar P. rapae larvae may be underrepresented in the survey samples due to their small size. Insects in samples were counted by species and life stage, and all P. rapae larvae were dissected to determine the level of parasitism by each parasitoid species. The only parasitoids observed in dissection were C. glomerata and C. rubecula. All dissections were done by the senior author. The immature stages, including eggs, of these 2 species can be readily separated in dissection by several characteristics. Visible mandibles and an anal hook are present in first instars of C. rubecula, but not in those of C. glomerata (Van Driesche 2008). Also, the number of parasitoid larvae per host is diagnostic (C. rubecula is solitary; C. glomerata is gregarious). No parasitoid eggs were seen in this survey although they can be distinguished by size, shape, and number (Van Driesche & Nunn 2002).

TABLE 1.

RATES OF PARASITISM IN 2011 BY COTESIA WASPS (HYMENTOPTERA: BRACONIDAE) OF PIERIS RAPAE FROM ORGANIC VEGETABLE FARMS IN THE EASTERN UNITED STATES AND CANADA.

In total, 32 samples of P. rapae larvae or pupae and parasitoid cocoons were examined, comprising 1571 individuals. Sample percent parasitism rates for each species were calculated at each location and mapped to look for geographical patterns. Average parasitism rates per species across all sites with any parasitism were also calculated. The percentages were arcsine transformed to better meet the assumption of normality, and then compared with a t test. Hyperparasitism was not examined in this study.

RESULTS

Summed across all 32 samples collected in the survey, 1571 individuals, were obtained and examined (Table 1). From that pool of samples, the only parasitoids recovered were C. rubecula and C. glomerata. Cotesia rubecula was present at 22 of the 32 sample sites (Table 1) and parasitized 20.6 ± 0.02% (95% CI) of the 1571 individuals examined. Cotesia glomerata was present at 12 sites and parasitized 7.3 ± 0.01% (95%CI) of the 1571 individuals. When parasitism was calculated based only on sites where a given parasitoid actually occurred, we found an average parasitism rate of 47 ± 0.03 % (95% CI, n = 1041) for C. rubecula and 25 ± 0.03 % (95% CI, n = 641) for C. glomerata (t-value: 2.748, df: 31, P = 0.0049).

Fig. 1.

Observed pattern of Cotesia parasitism of Pieris rapae in parts of the eastern United States and southeastern Canada in 2011. Parasitism by Cotesia rubecula is shown in gray and Cotesia glomerata in black. The percentage of unparasitized larvae is shown in white.

Spatially, C. rubecula and C. glomerata were largely exclusive in the distribution of their recoveries (Fig. 1). Only at 4 out of the 32 sites sampled was parasitism by both C. rubecula and C. glomerata detected. These 3 of these 4 sites (exclusive of the Charlestown, Rhode Island site) ere on the border of what appears to be a latitudinal point of separation of the regions that each parasitoid now occupies. Cotesia rubecula recoveries were highly concentrated in the north, while C. glomerata was dominant farther south. Cotesia rubecula was not found below latitude N 38° 48′, and is the only parasitoid found in our samples above latitude N 40° 2′. Within the area surveyed, no westward limit was detected for the distribution of C. rubecula (i.e., it was present in the most western of our sample locations [North Dakota]).

DISCUSSION

The Cotesia spp. distribution patterns observed in our survey (Fig. 1) are not explained by the history of these species. Cotesia glomerata, now largely absent in the northern portion of our survey area, was once widely present there (Fig. 2) and likely still occurs there at very low levels (e.g., in Massachusetts, Van Driesche 2008). Similarly, the absence of C. rubecula in the southern portion of our survey area is not due to failure to release the parasitoid there, since releases were made in both Virginia and Missouri (Fig. 2). The absence of C. rubecula in the southern portion of our survey area is consistent with previous studies on the diapause needs of this species (Nealis 1985). Diapause in C. rubecula is induced by short day length. Cool temperatures during diapause are believed to preserve the insect's fat supply and coordinate post-diapause development (Nealis 1985). Nealis (1985) further suggested that the mechanism for poor establishment of some populations of C. rubecula in southern locations was the premature induction of diapause, caused by short daylength, before seasonal temperatures had declined. Temperatures above 15 ^°C on average have been hypothesized to be lethal to diapausing prepupae of C. rubecula (Nealis 1985). Another potential explanation for the failure of C. rubecula to establish in some areas of the United States is the effect on C. rubecula densities of high rates of mortality to its immature stages due to hyperparasitism, as observed in Virginia (McDonald & Kok 1992; Gaines & Kok 1998); however, there is no evidence in the literature that hyperparasitism rates in southern states are higher than in other areas. We did not examine hyperoparasitism rates in our survey.

Fig. 2.

USA States and Canadian Provinces where the Pieris rapae parasitoid, Cotesia glomerata, was previously (pre-1990) dominant (crosses), and where releases of the Chinese strain of Cotesia rubecula (open circles) or the Yugoslavian strain (black circles) were made.

The absence of C. glomerata in samples from the northern portion of our survey area, where it was formerly widespread, is likely related to competitive displacement by C. rubecula. The phenomenon of parasitoid displacement has been well documented in other systems (e.g., DeBach & Sundby 1963; Le Brun et al. 2009). Our study suggests that such displacement of C. glomerata by C. rubecula has occurred at this larger spatial scale, as was previously observed for these species at the state/province level in Massachusetts (Van Driesche 2008), Quebec (Godin & Boivin 1998), and Oregon and Washington (Biever 1992). Cotesia rubecula is now widespread in the northeastern and north central United States and parts of southeastern Canada. The lack of such displacement in Europe, where both Cotesia species coexist, is likely due to the presence of the specific host of C. glomerata, Pieris brassicae L. and which is not attacked by C. rubecula.

The increase in prevalence and dominance of C. rubecula provides benefits both by increasing the level of control of the imported cabbageworm (P. rapae) and lessening the damage to non-target native pierids from C. glomerata. The displacement of C. glomerata in the northern United States by the more host-specific C. rubecula should allow some native pierids such as P oleracea (Benson et al. 2003; Van Driesche et al. 2004) and Pontia protodice Boisduval and Leconte (Dave Wagner, University of Connecticut, pers. comm.), whose ranges collapsed in some regions due to attack by C. glomerata, to recolonize areas from which they were extirpated, providing a benefit to protection of native biodiversity.

Also vegetable producers will benefit from this change in parasitoid species. Cotesia rubecula, which is now the dominant parasitoid of P. rapae in the northern part of our survey area, causes high levels of mortality to P rapae (47 ± 0.03%). and kills individual larvae before most of their feeding occurs. Although we cannot say with certainty which strain of C. rubecula is now found at particular sites, the introduction of C. rubecula in North America appears to be at least a partially successful biological control program that has met its objectives.