Predicting parasitoid accumulation on biological control agents of weeds


Correspondence author. E-mail:


1. Natural enemies may reduce the effectiveness of weed biocontrol agents and can also cause environmental damage, for example to a shared native insect host through apparent competition. Indeed, successful biocontrol may rely on enemy-free space and avoidance of apparent competition in the area where the biocontrol agent is introduced.

2. We surveyed parasitism in 28 insects released for weed biocontrol in New Zealand (NZ). We reviewed the global literature and databases to complement this survey, and to collate records of these insects being parasitized in their area of origin. We also collated records of native insects that feed on weeds targeted for biocontrol in NZ to test Lawton’s (1985) hypothesis that, to find enemy-free space, selected agents should ‘feed in a way that is different’ and ‘be taxonomically distinct’ from native herbivores in the introduced range.

3. We found that 19, mostly native, parasitoid species attack 10 weed biocontrol agents in NZ, of which 15 were confined to five agents that possessed ‘ecological analogues’, defined as a native NZ insect that belongs to the same superfamily as the agent and occupies a similar niche on the target weed. Parasitoid species richness in NZ was positively correlated to richness in the area of origin. However, only agents with ecological analogues contributed significantly to this pattern.

4. A review of NZ weed biocontrol programmes indicated that parasitism is significantly associated with the failure of agents to suppress weed populations.

5.Synthesis and applications. Although our conclusions are based on an unavoidably limited data set, we conclude that biocontrol agents that escape attack from parasitoids are more likely to suppress weed populations and should be less likely to have significant indirect non-target effects in food webs. Biocontrol practitioners can reduce the chance of weed biocontrol agents attracting species-rich parasitoid faunas after introduction by (i) selecting agents that have species-poor parasitoid faunas in their area of origin, and/or (ii) avoiding agents that have ‘ecological analogues’ awaiting them in the introduced range.


Two-thirds of weed biocontrol agents that establish fail to suppress their target weed (McFadyen 1998), and this failure can be a result of natural enemies (Goeden & Louda 1976). These natural enemies, such as parasitoids may cause environmental damage, for example to a shared native insect host through apparent competition (Carvalheiro et al. 2008).

Concerns for environmental safety have triggered tighter regulation of biocontrol world-wide (Sheppard et al. 2003). In New Zealand (NZ), biocontrol introductions are administered by the Environmental Risk Management Authority, which evaluates the risks, costs and benefits of introducing biocontrol agents (Barratt & Moeed 2005), including their potential impacts on food webs. Information regarding the potential parasitoids of a candidate biocontrol agent is, however, often unreliable. For example, the parasitic Hymenoptera of the NZ biogeographic region are poorly known (Berry 2007). Biocontrol practitioners have, therefore, sought to make predictions by identifying factors that influence parasitism. For example, Harris (1991) argued that gall midges should be given a low priority as biocontrol agents because they readily acquire parasitoids, and Hill & Hulley (1995) noted that parasitism varied according to level of concealment. A broad analysis showed that the numbers of primary parasitoid species acquired by insect herbivores in their introduced range are correlated with the numbers of primary parasitoid species attacking them in their area of origin (Cornell & Hawkins 1993). So, should herbivorous insects with a high diversity of natural enemies in their area of origin be passed over as potential weed biocontrol agents? Paradoxically, weed biocontrol practitioners argue that many herbivorous insects are kept rare by natural enemies in their areas of origin and can become excellent weed biocontrol agents in the ‘enemy-free space’ of the introduced range (Strong, Lawton & Southwood 1984; Hunt-Joshi, Root & Blossey 2005). Furthermore, agents that are kept rare in their area of origin may be more effective because their target weeds have had little selection to evolve resistance to them (Myers, Higgins & Kovacs 1989).

Lawton (1985) stated that enemy-free space and avoidance of apparent competition in the introduced range may be critical for weed biocontrol agent success and recommended that selected agents should ‘feed in a way that is different’ and ‘be taxonomically distinct’ from native herbivores in the introduced range.

We addressed these issues by surveying the parasitoid fauna of weed biocontrol agent species in NZ and testing the following hypotheses:

  • 1 Parasitoid richness increases with the time that an agent has been present in the introduced region.
  • 2 Weed biocontrol agents that have rich parasitoid complexes in their area of origin should accumulate more parasitoid species in the introduced range than species which have no or few parasitoids that attack them in their area of origin.
  • 3 Agents that ‘feed in a way that is similar’ and are ‘taxonomically close’ to the native fauna should be susceptible to parasitism.
  • 4 Parasitism has reduced the effectiveness of weed biocontrol agents in NZ.

Materials and methods

Literature review and surveys

Information regarding parasitism (in the area of origin and in NZ) of insect species introduced into NZ for the biological control of invasive weeds, and their ability to suppress weeds in NZ, was obtained by reviewing published literature and conducting field surveys.

Differences in sampling effort may confound comparisons between regions because sampling effort should correlate with parasitoid richness on host species. We, therefore, followed Cornell & Hawkins (1993) example by compiling records from single studies (generally those conducted in weed biocontrol programmes), rather than summary lists, which may be inflated by the inclusion of rare or unusual host-records. For example, Cameron (1935) compiled literature records of 12 primary parasitoids that attack the cinnabar moth Tyria jacobaeae, but only reared seven primary parasitoid species from English material, implying that the other parasitoids were either very rare or only rarely attack T. jacobaeae.

Surveys in NZ were performed during 2003–2009 (1–14 sample localities per agent, depending on availability). The life stages surveyed matched the stages from which parasitoids were reared in the area of origin: larvae (or nymphs) of all species and pupae of most species were collected, while eggs and adults were surveyed less extensively (Table 1). Larvae and pupae of external feeders were hand-collected. Larvae and pupae of gall-formers, leaf-miners, seed and stem-borers were sampled by collecting infested plant parts. With root and rosette-borers, plants were dug up and potted for subsequent rearing. Agents were reared in sealed, ventilated containers that prevented any emerging parasitoids from escaping. The numbers of biocontrol agents and parasitoids that emerged were recorded until emergences ceased. Species which diapause as pupae were stored in a cool room (c. 10 °C and natural day length for 3–4 months), prior to returning them to room temperature and allowing adults and, if present, parasitoids to emerge the following year. Parasitoids were identified to genus or species (Table 2).

Table 1.   Numbers of parasitoid species attacking weed biocontrol agents in NZ (PNZ) and in their area of origin (PAO), and the numbers of individuals reared during field surveys in NZ
Biocontrol agent (and NZ ecological analogues)Agent feeding guildYear of release in NZPNZPAOStages rearedNo. of agents rearedNo. of parasitoids rearedMaximum % mortality due to parasitoids
  1. Ecological analogues are for native insects in the same superfamily as the agent (see text for details); PNZ combines published literature and current surveys. Agent feeding guild: E, external feeder; SB, stem-borer; GF, gall-former; FB, flower-borer; RF, root-feeder; LM, leaf-miner; LR, leaf-roller; RB, rosette-borer. Stages reared: E, eggs; L, larvae; N, nymphs; P, pupae; A, adults. Mortality includes both mortality resulting from parasitism and from host-feeding by parasitoids and is the maximum percentage recorded in the published literature or the current surveys; figures in parentheses are the values recorded in the area of origin, if available.

  2. aStewart (1996); bMaddox (1968); cHill et al. (1995); dVogt et al. (1992); eWatmough (1968); fG. Grosskopf, (personal communication); gHoy (1958, 1960); hMiller (1970); iSyrett et al. (1999); jParnell (1966); kCox (2007); lWilson (1943); mDavies (1928); nSchmitt (1988); oPeterson et al. (2004); pWaloff (1987); qWindig (1991); rGrosskopf & Murphy (1999); sLeen (1995); tQuicke & Shaw (2004); uPaynter et al. (2008a); vHill et al. (2001); wHill (1989); xBess & Haramoto (1959); yMurray et al.(2002); zGassmann & Louda (2001); aaTeulon & Penman (1990); bbR. L. Hill (personal communication); ccWinks et al.(2004); ddScott & Brown (1992); eeBarratt et al. (1997); ffBoldt & Campobasso (1981); ggGaskin (1966); hhHelson (1974); iiCameron (1935); jjSchlumprecht (1989); kkEber & Brandl (1994); llZwölfer & Arnold-Rinehart (1994); mmRedfern et al. (1992); nnSyrett (1989).

  3. Numbers of agents reared are from the current survey except: oopublished literature only; pppublished literature and current surveys combined.

Agasicles hygrophilaE19820a1bELPA91600
Agonopterix umbellanaE1990010cL43100 (62)c
Arcola malloiSB198206dL4800 (38·5)d
Arytainilla spartiophilaE199300eNA20000 (0)e
Aulacidea subterminalisGF199902fLP15300 (42·9)f
Botanophila jacobaeaeFB19360g7hLP2842pp0pp0 (51)h
Bruchidius villosusSB19871i4jELPA731351118·4 (56·4)j
Chrysolina hypericiE194305kLA22600
Chrysolina quadrigeminaE196302lEL12100 (40)l
Cydia succedanaSB19920?L61300
Exapion ulicisSB19311i6mLP75792·3 (82·3)m
Lema cyanellaE198302nLA12100
Lochmaea suturalisE19970o2pELPA311pp0pp0 (53·2)p
Longitarsus jacobaeaeRF198301qLP105400
Macrolabis pilosellaeGF200201rLP9900 (80·3)r
Pempelia genistella (Uresiphita polygonalisS)E199811tL6011·6
Phytomyza vitalbae (Phytomyza clematadiU)LM19968u13vLP83810358 (‘high’)v
Procecidochares utilisGF19581w6xLP83126100
Procecidochares alaniGF20011?LP21610568·2
Rhinocyllus conicusFB19731y7zALP2849016·8y (71)z
Sericothrips staphylinus (Thrips obscuratusAA)E199000bbNA150000 (0)bb
Tortrix s.l. sp. ‘chrysanthemoides’ (‘Cnephasia’ jactatana; Ctenopseustis obliquanaCC)LR200736ddEL3032314·7
Trichosirocalus horridusRB19790ee4ffLPA21000 (<1)ff
Tyria jacobaeae (Nyctemera annulataGG)E19293hh7iiLP265078 (60)ii
Urophora carduiGF199404jjLP34200 (100)kk
Urophora solstitialisGF199002llLP59000
Urophora stylataGF199903mmLP63800 (29·2)mm
Zeuxidiplosis giardiGF19611nn5lLP39017540·8 (87)l
Table 2.   Parasitoid species reared from NZ weed biocontrol agents
  1. aRegarded as native by taxonomists, but probably recently self-introduced from Australia (see text for details).

Pteromalus sequesterAccidental introductionExapion ulicis; Bruchidius villosus
Aucklandella sp.NativePempelia genistella
Chrysocharis sp.NativePhytomyza vitalbae
Diglyphus isaeaBiocontrol agentPhytomyza vitalbae
Neochrysocharis‘sp. 1’NativePhytomyza vitalbae
Neochrysocharis‘sp. 2’NativePhytomyza vitalbae
Opius sp. ‘PC’NativePhytomyza vitalbae
Opius cinerariaeNativePhytomyza vitalbae
Pnigalio soemiusAccidental introductionPhytomyza vitalbae
Proacrias‘sp.’NativePhytomyza vitalbae
Megastigmus‘sp.’NativeaProcecidochares utilis; P. alani
Microctonus aethiopoidesBiocontrol agentRhinocyllus conicus
Trichogrammatoidea sp.
Dolichogenidea tasmanica
Tortrix s.l. sp. chrysanthemoides Tortrix s.l. sp. chrysanthemoides
Trigonospila brevifasciesBiocontrol agentTortrix s.l. sp. chrysanthemoides
Echthromorpha intricatoriaNativeaTyria jacobaea
Pales castaNativeTyria jacobaea
Pales nyctemerianaNativeTyria jacobaea
Torymoides‘sp.’NativeZeuxidiplosis giardi

Definition of an ecological analogue

We defined an ecological analogue as a native NZ insect that (i) is taxonomically related to the agent (see ‘Taxonomy’ below); (ii) has a similar lifestyle niche; and (iii) feeds on the target weed. We used the feeding guild and taxonomic criteria of Lawton (1985) plus a third criterion: the native analogue had to have been recorded feeding on the target weed. Lawton (1985) used this third criterion implicitly as well because the target weed he discusses is itself a native plant and he is referring to analogues in the native insect herbivore fauna on that plant. To identify ecological analogues, we consulted the Plant-SyNZTM data base ( and NZ weed faunal surveys (Syrett 1993; Winks, Fowler & Smith 2004) to locate records of native NZ insects attacking introduced weeds in NZ.

The feeding guild and taxonomic criteria broadly reflect knowledge regarding host selection by parasitoids: most parasitoids are oligophagous, attacking related taxa of hosts (Strand & Obrycki 1996), whereas others attack hosts in a range of higher level taxonomic groups that share feeding niches/guilds (Messing 2001). The third criterion of feeding on the target host encompasses several parasitoid behaviour and spatial co-occurrence factors. Many parasitoids locate hosts by responding to visual and chemical cues provided by the host plants of their prey species (Eben et al. 2000); occurrence of analogues on the same individual host plants may just give native parasitoids close physical proximity to introduced herbivores; at larger scales, if the analogue’s utilization of the target weed is widespread then native parasitoids are likely to have overlapping geographic ranges with the introduced biocontrol agents. Specifications included in our definition of an ecological analogue are:

  • 1 Feeding guild: To allow comparison with previous studies, we classified insects into guilds based on the larval feeding habits using Cornell & Hawkins (1993) categories, plus one additional category (root-feeder).
  • 2 Taxonomy: Only one agent was congeneric with a native insect species. Therefore, we tested whether defining the ecological analogue within the same family, superfamily or order as the agent, explained the most variance in our analyses (below).


We used data from the literature review and field surveys to test the hypothesis that introduced herbivores should acquire parasitoids over time by performing a Spearman’s rank correlation to investigate whether the parasitoid richness (i.e. number of parasitoid species reared per species) of weed biocontrol agents in NZ (PNZ) increased according to their residence times (time since the biological control agent was introduced). To test the hypothesis that species should be more likely to accumulate parasitoids in NZ if they (i) have a high diversity of parasitoids in their area of origin, and (ii) possessed an ecological analogue, we performed an analysis of covariance by correlating parasitoid richness in the area of origin (PAO) with parasitoid richness in NZ (PNZ) and also using the grouping factor ‘ecological analogue present’, with two values, yes or no. Two agents for which native range parasitism data was lacking (Table 1) were excluded from this analysis.


Literature review and field surveys

We found literature records for and/or we surveyed 28 of the 30 insects established in NZ for the biological control of weeds, as of December 2008 (the exceptions were Cleopus japonicus Wingelmüller and Platyptilia isodactyla Zeller that have only very recently established). The details are summarized in Table 1 and below.

Literature records of parasitism in New Zealand

Syrett et al. (1999) reported that the broom seed-beetle Bruchidius villosus and gorse seed-weevil Exapion ulicis were parasitized by an accidentally introduced parasitoid Pteromalus sequester, which also parasitizes both species in their native European ranges.

Barratt et al. (1997) investigated non-target attack by Microctonus aethiopoides, a parasitoid introduced into NZ for biological control of the pasture pest weevil Sitona discoideus, and found that it attacked adult thistle receptacle weevils Rhinocyllus conicus, but did not attack the thistle crown weevil Trichosirocalus horridus.

The devil weed gall-fly Procecidochares utilis pupae were reported to be parasitized by a wasp Megastigmus sp. in 1972, 14 years after the introduction of P. utilis (Hill 1989). It is the same species reported to parasitize P. utilis in Australia, the assumed origin of Megastigmus sp. populations in NZ (Hill 1989).

Three parasitoids attack T. jacobaeae in NZ (Helson 1974); an ichneumonid Echthromorpha intricatoria attacks the pupae and two native tachinid flies Pales (Cerosomyia) casta and P. nyctemeriana attack as much as 53–78% of the larvae (Miller 1970). Echthromorpha intricatoria was first recorded in NZ in 1915 (Gourlay 1964) and is native to Australia. Both P. casta and E. intricatoria also attack the related native moth Nyctemera annulata (Helson 1974), which occupies a similar niche to T. jacobaeae on native Senecio spp. and also attacks ragwort Jacobaea vulgaris Gaertn. in NZ.

St John’s wort gall midge Zeuxidiplosis giardi larvae were parasitized by Torymoides sp. (Syrett 1989) that is probably native (J. A. Berry, pers. comm.). Torymid wasps attack native NZ gall midges, e.g. Dasyneura hebefolia is a host of Dimeromicrus (=Torymoides) sp. (Valentine 1967).


No new cases of parasitism were discovered on previously surveyed biocontrol agent species. However, four biocontrol agents, not previously surveyed, were found to be parasitized. An ichneumonid wasp (Aucklandella sp.) was reared from silken tents containing gorse hard shoot moth Pempelia genistella Duponchel larvae. Little is known about Aucklandella sp., although Berry (1990) reared Aucklandella sp. from the silken tents of a native moth Hierodoris atychioides. Mist flower gall-fly Procecidochares alani Steyskal was parasitized by Megastigmus sp., which also attacks P. utilis in NZ (Hill 1989). Boneseed leafroller Tortrix s.l. sp. chyrsanthemoides eggs were parasitized by a wasp Trichogrammatoidea sp. Two native species of Trichogrammatoidea are known from NZ including T. bactrae, which has been reared from Epiphyas postvittana eggs, while other, unidentified material has been reared from eggs of several moths including Planotortrix excessana (Noyes & Valentine 1989). Tortrix s.l. sp. chyrsanthemoides larvae were parasitized by Trigonospila brevifacies Hardy and a braconid wasp Dolichogenidea tasmanica (Cameron). Trigonospila brevifacies was deliberately introduced from Australia to control tortricid orchard pests (Thomas 1975) and attacks various non-target hosts (Munro & Henderson 2002); D. tasmanica is self-introduced (Munro & Henderson 2002). Finally, six native (eulophid and braconid) and two exotic eulophid parasitoids were reared from old man’s beard leaf-miner Phytomyza vitalbae larvae and pupae (previously reported in Paynter et al. 2008a). Five of the native parasitoids also attack a closely related native leaf-miner Phytomyza clematadi (Paynter et al. 2008a). The exotic parasitoids were: Pnigalio soemius (misidentified previously as P. pectinicornis), which was accidentally introduced c. 1950 with its host, the oak leaf-miner (Thomas & Hill 1989) and Diglyphus isaea, which was deliberately introduced to control exotic leaf-miner pests (McGregor 1989). Parasitism of P. vitalbae was relatively low (6%; Paynter et al. 2008a). However, more larvae are killed by female eulophids host-feeding than by parasitism (Hill, Wittenberg & Gourlay 2001) and c. 50% of P. vitalbae mines were subject to predation (Paynter et al. 2008a).

No obligate hyperparasitoids were reared or mentioned in literature records of parasitism in NZ. Parasitoids assumed to be rearing contaminants are listed in Table S1 (Supporting information).

Analysis of factors associated with parasitism

Nineteen, mostly native, parasitoid species were reared from 10 of the 28 biocontrol agent species surveyed. PNZ was not correlated to years since agent introduction (Spearman’s rank correlation, rs = 0·064, d.f. = 26, = 0·748).

Defining an ecological analogue as a native insect in the same superfamily as a biological control agent, resulted in the recognition of five agents with ecological analogues (Table 1) that were attacked by 15 of the 19 parasitoids identified (Table 2). We assume that species that are not parasitized in the area of origin are unlikely to be parasitized in the introduced range, regardless of the presence of an analogue. A Fisher’s exact test, therefore, examined the difference between the categories ‘analogue present AND parasitized in area of origin’ (Y/N) and parasitized in NZ (Y/N) and indicated the associations between these categories was non-random (= 0·0088). All four species that possessed both an analogue and were parasitized in their area of origin were parasitized in NZ, compared with only 5 of the remaining 22 species [two agents for which native range parasitism data was lacking (Table 1) were excluded from this analysis].

The analysis of covariance explained 92·4% of the variance in parasitoid richness in NZ (PNZ), which was positively correlated to richness in the area of origin (PAO; ancovaF1, 22 = 145·5, < 0·001), but only agents with ecological analogues contributed significantly to this pattern (Fig. 1; ancova on effect of analogue presence: F1, 22 = 91·77, < 0·001; PAO/analogue interaction; F1, 22 = 69·93, < 0·001). The same relationships were significant when we omitted the potentially influential P. vitalbae data point: ancova PAO: F1, 21 = 18·87, < 0·001; effect of analogue presence: F1, 21 = 49·40, < 0·001; PAO/analogue interaction: F1, 21 = 21·57, < 0·001.

Figure 1.

 Number of parasitoid species attacking weed biocontrol agents in New Zealand (PNZ) vs. in their areas of origin (PAO) for agents categorized as having a native analogue in same insect superfamily (see text for details) either present in NZ (circles, solid line; PNZ = 0·577*PAO-0·116) or absent (triangles, dashed line; PNZ = 0·068*PAO-0·022).

Defining an ecological analogue as a native insect within the same family as the biocontrol agent excluded P. genistella/Uresiphita polygonalis (Felder) (Table 1) and explained marginally less variance (91·9%). The same relationships were significant: PNZ was positively correlated to PAO (ancovaF1, 22 = 135·58, < 0·001; effect of analogue presence F1, 22 = 78·85, <0·001; PAO/analogue interaction F1, 22 = 70·33, < 0·001).

Defining an ecological analogue as a native insect within the same order as the biocontrol agent recognized 6 agents with ecological analogues (the ‘superfamily analogues’ listed in Table 1 plus Agonopterix umbellana (F.) due to the presence of several native tortricid leafroller species that feed on gorse in NZ) and explained much less variance (57·5%). PNZ was again positively correlated to PAO (ancovaF1, 22 = 26·03, < 0·001), as was the effect of analogue presence (ancovaF1, 22 = 7·23, < 0·05), but the PAO/analogue interaction was not significant (ancovaF1, 21 = 3·63, NS).

Impact of parasitism

We reviewed the success of biocontrol programmes and of individual agents in NZ, in relation to parasitism (Table 3). Excluding programmes that are too recent to assess, nine insect agents are classified as successful, or partially successful, of which only one is parasitized, while 8 of the 15 insects classified as unsuccessful are parasitized (Table 3). A one-tailed Fisher’s exact test confirmed the hypothesis that agent success was negatively associated with parasitism (= 0·048).

Table 3.   Success of weed biocontrol agents confirmed established in NZ by December 2008
Target weedAgent(s) to which success attributedOther agent(s) releasedNotes
  1. Successful programmes: no other control methods required where biocontrol agents are well established; partially successful programmes: substantial biocontrol impacts, but other control options still required. Long-term programmes: first agents released >10 years ago; programmes too early to assess success: first agents released <10 years ago.

  2. aParasitized; bParasitized and ecological analogue present. Entyloma ageratinae, Cercospora eupatorii and Phoma clematidina are fungal pathogens, Tetranychus lintearius is a mite. Note that Cleopus japonicus and Platyptilia isodactyla are excluded as they were not surveyed for parasitoids.

  3. cLandcare Research, unpublished data; dBarton et al. (2007); eGourlay et al. (2008); fDymock (1985); gMiller (1970); hFowler et al. (2000); iSyrett (1989); jRoberts & Sutherland (1989); kHill (1989); lShea & Kelly (2004); mGourlay (2004); nPaynter et al. (2008b); oPeterson et al. (2000); pPaynter et al. (2006).

Successful programmes
 Heather Calluna vulgaris (L.) HullLochmaea suturalis Thomson Considered an incipient success: heather patches are being killed by L. suturalis outbreaks at several localitiesc
 Mist flower Ageratina riparia R.M. King & H. Rob.Entyloma ageratinae Barreto and EvansProcecidochares alaniaE. ageratinae destroyed mist flower stands before P. alani impact was measuredd
 Ragwort Jacobaea vulgaris Gaertn.Longitarsus jacobaeaeBotanophila jacobaeae
Tyria jacobaeae
In most regions, L. jacobaeae controls ragworte. Asynchrony with host plant limits B. jacobaeaef. T. jacobaeae outbreaks are not sustained due to parasitism and predationg
 St John’s wort Hypericum perforatum L.Chrysolina hyperici
C. quadrigemina
Zeuxidiplosis giardiaComplete control due to Chrysolina beetlesh. Z. giardi is highly localized around Nelson (South Island)i
Partially successful programmes
 Alligator weed Alternanthera philoxeroides Griseb.Agasicles hygrophila
Arcola malloi
 Good control of weed floating in still water, but not terrestrial weedj
 Mexican Devil weed Ageratina adenophora (Spreng.) King & H. Rob.Cercospora eupatorii Peck
Procecidochares utilis*
 Relative importance of agents is unknown but P. utilis has declined due to parasitismk
 Nodding thistle Carduus nutans L.Trichosirocallus horridus
Urophora solstitialis
Rhinocyllus conicusaModelsl indicate that T. horridus and U. solstitialis are responsible for partial successh
Long-term programmes considered unsuccessful to date
 California thistle Cirsium arvense (L.) Scop. Lema cyanella
Urophora cardui
L. cyanella is confined to one localitym. Palatable U. cardui stem galls are eaten by sheep and cattle, limiting its abundancem
 Gorse Ulex europaeus L. Agonopterix umbellana
Cydia succedana
Exapion ulicis
Pempelia genistellab
Sericothrips staphylinus
Tetranychus lintearius Dufour
C. succedana and E. ulicis are ineffective because they do not attack the autumn/winter seed cropn. T. lintearius outbreaks do not persist due to predationo; P. genistella is rare and highly localized. A. umbellana and S. staphylinus are abundant at some localities, but their impacts are unknown
 Old Man’s Beard Clematis vitalbae L. Phytomyza vitalbaeb
Phoma clematidina Thüm.) Boerema
Insufficient damage for controlp
 Scotch broom Cytisus scoparius (L.) Link Arytainilla spartiophila
Bruchidius villosus
A. spartiophila impact is minor, possibly due to predation. B. villosus destroys c. 70% of seed at early release sites, which is insufficient to control broomc
Ongoing programmes too early to assess success
 Boneseed Chrysanthemoides monilifera (L.) Norlindh Tortrix s.l. sp. chrysanthemoidesbPatchy establishment on the North Island; no establishment on the South Island despite multiple releases
 Hieracium Pilosella officinarum (L.) F.W.Schultz & Sch.Bip. Aulacidea subterminalis
Macrolabis pilosellae
Impact of agents not measured, they are common at some early release sites
 Scotch thistle Cirsium vulgare (Savi) Ten. Urophora stylataImpact not measured


The lack of correlation between parasitoid richness and residence time indicates that, as previously noted from South Africa (Hill & Hulley 1995), susceptible weed biocontrol agents rapidly accumulate parasitoids. Our analyses indicate that ecological analogues are a major source of parasitoids capable of rapidly colonizing novel biocontrol agent hosts. Agents that are heavily parasitized in their areas of origin may, therefore, find enemy-free space after introduction if they lack ecological analogues in the herbivore fauna they are introduced into. If analogues exist, however, then weed biocontrol practitioners may do well to avoid agents that are parasitized in their areas of origin: these species are likely to accumulate parasitoids, potentially reducing their performance and creating the possibility of indirect non-target impacts in food webs (e.g. Willis & Memmott 2005). Analogues could be identified without substantially increasing the operating costs of a weed biocontrol programme because target weeds should be surveyed in the introduced range during the early stages of a biological control programme (Wapshere, Delfosse & Cullen 1989).

Compared with successful agents, a high proportion of agents that have failed to suppress their target weeds in NZ are parasitized. Our conclusions, however, are based on an unavoidably limited data set and the relative importance of parasitism as a key factor for biocontrol failure requires further study. For example, Hawkins, Thomas & Hochberg (1993) noted that parasitoids that kill fewer than c. 40% of their hosts rarely control that host and few NZ weed biocontrol agents were parasitized to that extent (Table 1). Our surveys, however, were designed to determine parasitoid diversity. Percentage parasitism samples are unreliable indicators of total parasitoid impact per host generation and the meticulous methods required for accurate estimates (Van Driesche et al. 1991) were beyond the scope of this study.

In addition to their relevance to biocontrol, our findings have implications for assessing the biosecurity risk posed by accidental introductions of insect herbivores. We hope that the results from this study will encourage researchers to test our findings.


We thank Drs Darren Ward and Jocelyn Berry for parasitoid identifications and two anonymous referees for helpful comments. We acknowledge Foundation for Research, Science and Technology (Contracts C09X0504 and CO9X0905) funding.