Assessing indirect impacts of biological control agents on native biodiversity: a community-level approach
L.G. Carvalheiro,1 Y.M. Buckley,2,3 R. Ventim1 and J. Memmott1
Summary The safety of biological control methods is a subject that has received considerable attention for a long time. However, apparent competition (competition due to shared natural enemies) has been neglected when considering possible impacts of biological control agents. One of the reasons for the lack of studies in this area is the difficulty in assessing and predicting indirect effects due to apparent competition. In this paper we outline a methodology to predict and measure non-target impacts of biological control agents due to apparent competition.
Keywords: biological control, methodology.
Underlying rationale such as parasitoids, parasites and pathogens, which can frequently be oligophagous or polyphagous (e.g. Invasive species are one of the main threats to global Hawkins and Goeden, 1984; Memmott et al., 1994). biodiversity (Schmitz and Simberloff, 1997). Classical Therefore, if these natural enemies include such an biological control involves the deliberate introduction abundant food resource in their diet, their population of an alien species and it is viewed as a sustainable, en- abundance can in turn increase, creating a potential for vironmentally friendly form of pest control. The safety apparent competition. of biological control is a subject that has received much Several studies have shown that apparent competi- attention, with particular concerns about the interac- tion can have strong impacts on population dynamics, tions between biological control agents and ‘non-target’ either due to shared parasites (Tompkins et al., 2000), species (Pemberton and Strong, 2000; Thomas and predators (Muller and Godfray, 1997) or parasitoids Willis, 1998; Boettner et al., 2000; Louda et al., 1997). (Morris et al., 2001), as well as on community struc- Non-target species can be affected directly, if an agent ture (e.g. herbivorous communities in Morris et al., attacks a non-target host, or indirectly, for instance, 2004, 2005; aphid-parasitoid communities in Muller when the agent shares natural enemies with native spe- and Godfray, 1999; Muller et al., 1999). However, cies (apparent competition, reviewed by Holt and Law- non-target impacts of an introduced biological control ton, 1994). One of the main criteria for a certain species agent on native species through apparent competition to be considered a safe biological control agent is its is a subject that has not received much attention (Willis high host specificity, reducing its likelihood to directly and Memmott, 2005). affect native species. However, a successfully estab- If a biological control agent is effective in reducing lished biological control agent is an abundant resource weed abundance to low levels, then non-target impacts for natural enemies present in the target ecosystem, due to apparent competition can be minimal. How- ever, very few pre-release studies have predicted the effectiveness of potential biological control agents in reducing target weed abundance (e.g. Buckley et al., 1 School of Biological Sciences, Woodland Road, Bristol BS8 1UG, UK. 2 School of Integrative Biology, University of Queensland, St. Lucia, 2005; Wirfl, 2006). If an introduced agent remains at QLD 4072, Australia. high abundance over a long period of time, the prob- 3 CSIRO Sustainable Ecosystems, Queensland Bioscience Precinct, ability of non-target effects due to apparent competi- 306 Carmody Road, St. Lucia, QLD 4067, Australia. tion is enhanced. Furthermore, non-target impacts are Corresponding author: L.G. Carvalheiro
83 XII International Symposium on Biological Control of Weeds the weed/biological control agent, as they are the most community, which can then be tested using regression likely species to suffer irreversible damage that may models. potentially lead to their extinction. Suggested methodology Community level approach For a correct assessment of the impacts of the abun- Plant–insect interaction systems can be extremely dance of the weed and its biological control agents, complex, involving dense webs of interactions (e.g. two components need to be included in a post-release Waser et al., 1996; Memmott, 1999; Muller et al., 1999; impact assessment programme. The first component is Bascompte et al., 2003). Thus, to fully assess the poten- descriptive, involving the construction of food webs de- tial indirect effects of biological control, community- scribing the patterns of trophic linkages between plants, level surveys are necessary. Food webs have been herbivores and parasitoids in communities invaded by suggested as the appropriate way of analysing possible weeds. The second component involves statistical test- non-target interactions in biological control (Henneman ing of the effects of the weed and the biological control and Memmott, 2001; Strong, 1997), since food webs agent abundance on native communities’ abundance enable us to ask how a biological control agent can and species richness. influence native communities (Memmott et al., 2007). Using food webs as predictive tools in conservation biology has, until recently, been considered an unat- Sampling tainable goal, as at first sight they appear very labor Selection of ten to 20 plots covering all habitats that intensive to make and statistically difficult to analyse are threatened by the weed, and covering a gradient (Memmott et al., 2007). However, community-level of abundance of the weed and the biological control ecology has developed to a stage where we are capable agent, is required. Plot size should be selected in order of sampling, visualizing and analysing complex food to include the maximum number of plant species of the web interactions at community-level scale (Memmott field site (a suggested size of the plot is 40× 40 m), and et al., 2004; Dunne et al., 2002; Sole et al., 2001; Ber- the plots should be at least 500 m apart, so they can be sier, et al., 2002; Banasek-Richter et al., 2004; Cattin considered independent. et al., 2004). Ideally, all ecological niches would be studied, but Some studies have already used a community-level it is more practical to focus on the most likely ecologi- approach to look for non-target effects of biological cal niche to be affected, this being the one that includes control agents. For example, Louda et al., (1997) used the biological control agent in focus (e.g. seed preda- this approach to highlight the ability of biological con- tors, leaf miners, herbivores) and its parasitoids. Fur- trol agents to disrupt communities. They demonstrated thermore, assessing parasitism has another advantage, that an exotic seed-feeding biological control agent was since it may also be highly relevant to the success and displacing native seed feeders associated with non- impact of the biological control programme. target plants. Henneman and Memmott (2001) used this Community-level sampling requires a high amount approach to show that in a remote area of Hawaii, 83% of effort. Based on a pilot project, we estimate that of parasitoids reared from native moths were biological it will take approximately four weeks to sample 20 control agents. Nowadays, this type of non-target im- field sites, with two full-time people. Repeated- sam pact (due to lack of host specificity) is avoided by us- pling over time is needed during the seasons of higher ing the current safety regulations governing biological abundances of the biological control agent to include control (e.g. Fowler et al., 2000; Sheppard et al., 2005). the maximum number of species. The plots should be However, indirect non-target impacts are much harder sampled for plants, herbivores and parasitoids monthly. to predict and avoid. Willis and Memmott (2005) re- The sampling and rearing methods have been described vealed that the biological control agent, Mesoclanis po- in previous literature: seed predators and their parasit- lana (Munro) (Diptera: Tephritidae) had the potential oids (Memmott and Godfray, 1994); leaf herbivores to disrupt the native food web structure due to appar- and their parasitoids (Memmott et al., 1994; Lewis et ent competition, mediated by shared native parasitoids, al., 2002); and aphids and their parasitoids (Muller and whose population abundances exponentially increased Godfray, 1997; Muller et al., 1999). Rearing time can following the population outbreak of M. polana. Howev- vary with the biology and geographical region of the er, this study did not clearly test for impacts of the weed species involved. As an example, a pilot study with and the biological control agent on abundance and/or seed predators in Australia involved ten weeks of rear- species richness of native communities. To test for such ing after samples were collected. effects, repeated sampling in sites with different abun- dances of weed and biological control agent is needed. In this paper we propose that food webs provide a Determining species links protocol that can quantify the impact of both the alien It is relatively straightforward to determine trophic plant and its biological control agent upon the natural links between herbivorous insects and plant species.
84 Assessing indirect impacts of biological control agents on native biodiversity: a community-level approach
Determining the parasitoids of most herbivore spe- ‘pattern’, allowing the assessment of the total magni- cies can be also straightforward. Immature stages of tude of the effects of the biological control agent. the host insect are reared in isolation until either adult The approach presented here has recently been ap- hosts or parasitoids emerge (Memmott, 1999; Mem- plied by the authors to test for indirect impacts due mott and Godfray, 1994). Determining parasitoids of a to apparent competition of a highly specific biologi- given endophagous herbivore species (e.g. seed preda- cal control agent, Mesoclanis polana Munro, recently tor) is not as simple, as the seed predators themselves introduced in Australia (1996) to control an invasive develop inside the seed. However, for many plant spe- weed, Chrysanthemoides monilifera (L.) T. Norlindh, cies there are only a few pre-dispersal seed predators spp rotundata (Carvalheiro et al., 2008). and information in the literature on the food habits of the parasitoid species may be enough to identify the Acknowledgements host. Plant–herbivore–parasitoid webs are taxonomi- cally complex; therefore, taxonomic input is essential, We would like to thank Kate Henson for her comments although it can be time consuming and costly. on the manuscript and Fundação para a Ciência e Tec- nologia for funding. Testing for apparent competition Effects of the weed and biological control agent References on native communities of herbivores/seed predators, Banasek-Richter, C., Cattin, M.F. and Bersier, L.F. (2004) parasitoids and plants can be tested by using general- Sampling effects and the robustness of quantitative and ized linear models (GLMs) where all possible combi- qualitative food-web descriptors. Journal of Theoretical nations of the relevant variables (e.g. habitat, latitude, Biology 226, 23–32. weed abundance, biological control agent abundance) Bascompte, J., Jordano, P., Melian, C.J. and Olesen, J.M. will be tested. By ranking all possible models, the best (2003) The nested assembly of plant–animal mutualistic model can be selected (Zuur et al., 2007). If the effect networks. Proceedings of the National Academy of Sci- of the biocontrol agent is strong enough, a significant ences of the United States of America 100, 9383–9387. effect will be detected over and above the effect of the Bersier, L.F., Banasek-Richter, C. and Cattin, M.F. (2002) weed abundance. This allows differentiating which Quantitative descriptors of food-web matrices. Ecology native community patterns are significantly related to 83, 2394–2407. the weed abundance and/or to the biological control Boettner, G.H., Elkinton, J.S. and Boettner, C.J. (2000) Ef- fects of a biological control introduction on three nontar- agent abundance. For example, if a model including get native species of saturniid moths. Conservation Biol- the biological control agent abundance is selected as ogy 14, 1798–1806. the best model (e.g. native herbivores species richness Buckley, Y.M., Rees, M., Sheppard, A.W. and Smyth, M.J. ~ habitat*biological control agent abundance), and if (2005) Stable coexistence of an invasive plant and bio- the contribution of the biological control agent abun- logical control agent: a parameterised coupled plant– dance is significant to the fit of the model, we can con- herbivore model. Journal of Applied Ecology 42, 70–79. clude that the analysed variable is being affected by the Carvalheiro, L.G., Buckley, Y.M., Ventim, R., Fowler, S.V. and agent. Memmott, J. (2008) Apparent competition can compromise the safety of highly specific biocontrol agents.Ecology Let- ters 11, 690–700. Conclusions Cattin, M.F., Bersier, L.F., Banasek-Richter, C., Baltensperger, R. and Gabriel, J.P. (2004) Phylogenetic constraints and ad- Insects form numerous key links with other species, aptation explain food-web structure. Nature 427, 835–839. leading to complex networks of interactions. To fully Dunne, J.A., Williams, R.J. and Martinez, N.D. (2002) Net- assess the post-release impacts of an introduced biolog- work structure and biodiversity loss in food webs: ro- ical control agent, community-level studies involving bustness increases with connectance. Ecology Letters 5, quantitative data are needed. The recent practical and 558–567. theoretical advances made in food-web construction Fowler, S.V., Syrett, P. and Hill, R.L. (2000) Success and and analysis allows wider applications in the field of safety in the biological control of environmental weeds in conservation biology, such as the assessment of biolog- New Zealand. Austral Ecology 25, 553–562. ical control impacts. The food-web approach suggested Hawkins, B.A. and Goeden, R.D. (1984) Organization of in this work will provide a post-release impact assess- a Parasitoid Community Associated with a Complex of ment in an understandable, applicable form for both Galls on Atriplex Spp in Southern-California. Ecological Entomology 9, 271–292. biological control practitioners and site managers. In Henneman, M.L. and Memmott, J. (2001) Infiltration of a addition, although the methodology proposed here al- Hawaiian community by introduced biological control lows the assessment of post-release impacts, it is advis- agents. Science 293, 1314–1316. able that community-level studies are also done before Holt, R.D. and Lawton, J.H. (1994) The Ecological Conse- the release of the biological control agents. This would quences of Shared Natural Enemies. Annual Review of reveal the native community’s pre- and post-invasion Ecology and Systematics 25, 495–520.
85 XII International Symposium on Biological Control of Weeds
Lewis, O.T., Memmott, J., Lasalle, J., Lyal, C.H.C., White- Muller, C.B., Adriaanse, I.C.T., Belshaw, R. and Godfray, foord, C., Godfray, H.C.J. (2002) Structure of a Diverse H.C.J. (1999) The structure of an aphid–parasitoid com- Tropical Forest Insect–Parasitoid Community. The Jour- munity. Journal of Animal Ecology 68, 346–370. nal of Animal Ecology 71 (5), 855–873. Pemberton, R.W. and Strong, D.R. (2000) Safety data crucial Louda, S.M., Kendall, D., Connor, J. and Simberloff, D. for biological control insect agents. Science 290, 1896– (1997) Ecological effects of an insect introduced for the 1897. biological control of weeds. Science 277, 1088–1090. Schmitz, D.C. and Simberloff, D. (1997) Biological inva- Memmott, J. (1999) The structure of a plant-pollinator food sions: A growing threat. Issues Science and Technology web. Ecology Letters 2, 276–280. 13, 33–40. Memmott, J. and Godfray, H.C.J. (1994) The use and con- Sheppard, A.W., Shaw, R.H., and Sforza, R. (2005) Top 20 struction of parasitoid webs. In: Hawkins, B.A. and Shee- environmental weeds for classical biological control in han, W. (eds) Parasitoid Community Ecology. Oxford Europe: a review of opportunities, regulations and other University Press, Oxford, pp. 300–318. barriers to adoption. Weed Research 46, 93–117. Memmott, J., Godfray, H.C.J. and Gauld, I.D. (1994) The Sole, R.V. and Montoya, J.M. (2001) Complexity and fra- Structure of a Tropical Host Parasitoid Community. Jour- gility in ecological networks. Proceedings of the Royal nal of Animal Ecology 63, 521–540. Society of London Series B-Biological Sciences 268, 2039– Memmott, J., Waser, N.M. and Price, M.V. (2004) Tolerance 2045. of pollination networks to species extinctions. Proceed- Strong, D.R. (1997) Ecology—Fear no weevil? Science 277, ings of the Royal Society of London Series B-Biological 1058–1059. Sciences 271, 2605–2611. Strong, D.R., Lawton, J.H., and Southwood, T.R.E. (1984) Memmott, J., Gibson, R.H., Carvalheiro, L.G., Heleno, R., Insects on Plants. Blackwell Scientific Publications, Ox- Henson, K.S.E., Lopezaraiza, M.E. and Pearce, S. (2007) ford, 313 pp. The Conservation of Ecological Interactions. In: Stew- Thomas, M.B. and Willis, A.J. (1998) Biocontrol—risky but art, A.J.A., Lewis, O.T. and New, T.R. (eds) Insect Con- necessary? Trends in Ecology & Evolution 13, 325–329. servation Biology. CABI Publishing, Wallingford, UK, Tompkins, D.M., Draycott, R.A.H. and Hudson, P.J. (2000) pp. 226–244. Field evidence for apparent competition mediated via the Morris, R.J., Muller, C.B. and Godfray, H.C.J. (2001) Field shared parasites of two gamebird species. Ecology Letters experiments testing for apparent competition between 3, 10–14. primary parasitoids mediated by secondary parasitoids. Waser, N.M., Chittka, L., Price, M.V., Williams, N.M. and Journal of Animal Ecology 70, 301–309. Ollerton, J. (1996) Generalization in pollination systems, Morris, R.J., Lewis, O.T. and Godfray, H.C.J. (2004) Experi- and why it matters. Ecology 77, 1043–1060. mental evidence for apparent competition in a a tropical Willis, A.J. and Memmott, J. (2005) The potential for indirect forest food web. Nature 428, 310–313. effects between a weed, one of its biocontrol agents and Morris, R.J., Lewis, O.T. and Godfray, H.C.J. (2005) Apparent native herbivores: A food web approach. Biological Con- competition and insect community structure: towards a spa- trol 35, 299–306. tial perspective. Annales Zoologici Fennici 42, 449–462. Wirf, L. (2006) Using simulated herbivory to predict the Muller, C.B. and Godfray, H.C.J. (1997) Apparent competi- efficacy of a biocontrol agent: the effect of manual de- tion between two aphid species. Journal of Animal Ecol- foliation and Macaria pallidata Warren (Lepidoptera: ogy 66, 57–64. Geometridae) herbivory on Mimosa pigra seedlings Aus- Muller, C.B. and Godfray, H.C.J. (1999) Indirect interactions tralian Journal of Entomology 45, 324–326. in aphid–parasitoid communities. Researches on Popula- Zuur, A.F., Ieno, E.N. and Smith, G.M. (2007). Analysing tion Ecology 41, 93–106. Ecological Data. Springer, 680 pp.
86