Kiss (2004) raises several concerns over aspects of our review on the non-target effects of fungi being used as biological control agents of plant diseases (Brimner & Boland, 2003). We welcome the opportunity to provide additional comment and clarification. In our opinion, the most important point raised by Kiss is the assertion that the use of biological control agents ‘is considered, in general, as a safe and environmentally friendly alternative for plant disease control compared to the application of conventional pesticides’ (Kiss, 2004). We agree that this perception is common, but such assumptions are often unsubstantiated by scientific evidence. If biological control products are to live up to this perception, it is important that we confirm such assumptions scientifically, and do not avoid the need for toxicology and environmental fate data that can contribute to the identification of potential environmental and health risk (Whipps & Lumsden, 2001).
There is increasing published scientific evidence that microorganisms being used as biological controls can have significant, measurable effects, both direct and indirect, on non-target organisms. These effects include displacement of non-target microorganisms, allergenicity to humans and other animals, toxigenicity to non-target organisms, and pathogenicity to non-target organisms (Cook et al., 1996; Brimner & Boland, 2003; and references therein). There is a growing awareness for the need to consider these issues when developing biological control products. The commercial development of biological controls for plant diseases is a relatively new field and can arguably be compared with the early stages of pesticide development when often limited evaluations of target and non-target effects preceded more widespread application. It was only through more widespread use that epidemiological and environmental repercussions of using such products were identified. Furthermore, there are well documented examples in other areas of biological control, such as invasive animal and plant species, where introduced biological control strategies have become problematic.
Several aspects of our review appear to have contributed to miscommunication. We used the term ‘environment’ to include both physical and biological aspects of the environment, and the term ‘microorganisms’ to include bacteria, fungi, amoebae, flagellates, nematodes, etc. as is commonly used in the literature. Our use of the term ‘significant’ was, in most cases, consistent with a statistical detection between or among treatments. Similarly, due to the limitations of space, we did not point out the strengths or weaknesses of individual studies underlying the points we made, but did provide citations to the literature for other researchers to consider. Some studies are obviously more substantial and convincing than others, and the apparent shortage of articles on non-target assessment does not dismiss specific experiments if such studies are representative of the available information, or test an important hypothesis. Subsequent studies will support or refute these often exploratory studies. Unfortunately, much of the available evidence on non-target effects of biological controls is not published because of the tendency not to publish the results of experiments that do not show significant differences between or among treatments, or because data are often generated for purposes of commercial registration and are considered confidential. These considerations may produce a bias in the literature towards identification of possible risks.
The host range of parasitic fungi can be particularly difficult to determine and typically relies on compilation of anecdotal records from the literature, by conducting ecological surveys, and/or by inoculations of specific combinations of hosts and parasites. Ampelomyces quisqualis is considered a strict parasite of powdery mildews based on evidence compiled from the literature and, within the Erysiphaceae, has a wide host range (Kiss, 1998). However, specific host–parasite–hyperparasite interactions have seldom been conducted and, at least to some degree, are confounded by taxonomic delineations and a lack of basic ecological data. Inundative biological controls are often used at high rates of application, and may be used both within and outside of the indigenous habitats of the microorganisms being used. Such usage will result in new interactions of biological control microorganisms with non-target organisms, although differences in registration requirements may constrain the use of non-indigenous microorganisms in some regions (Whipps & Lumsden, 2001). The study by Kiss (1998) provides an example of how extensive survey data can contribute to our understanding of the role of mycoparasitism in the natural ecology of fungal communities in general, and to the known host associations of this fungus in particular. Taxonomic delineation of fungal species is another important area of concern in assessing non-target effects of biological control products, and one that is of particular concern to regulatory agencies because of the relationship between the identification of a microorganism and the scientific literature. For example, until recently, strains Th4 and Th2 of Trichoderma harzianum were associated with a green mold disease of Agaricus bisporus, which caused both real and perceived concern regarding the potential use of biological control products containing T. harzianum in surrounding agricultural and horticultural industries. In our article, we specifically noted that the relationship between T. harzianum and A. bisporus was highly specialized and dependent on the strain of the antagonist. More recently, the strains of Trichoderma associated with green mold disease were reclassified as a new species, Trichoderma aggressivum (Samuels et al., 2002), and this report provides an example of how taxonomic clarification can contribute to improved risk assessment, and supports the use of biocontrol strains of Trichoderma spp. that are non-pathogenic to A. bisporus.
In contrast to the statement by Kiss, there have been numerous studies on the potential use of Trichoderma spp. for the management of Sclerotinia spp., and Trichoderma spp. have a recognized role in the natural ecology of sclerotia of Sclerotinia spp. (Willetts & Wong, 1980; Adams & Ayers, 1979; Zhou & Boland, 1998). Our review did not focus on the results of biological control efficacy to target organisms but did include brief mention of the study by Gerlagh et al. (1999) because of the unexpected result where applications of the biocontrol agent resulted in an increased production of apothecia – an unexpected and possible non-target effect in itself. In our view, this was not a ‘small-scale experiment’ (Kiss, 2004) as it examined 12 biological control and crop treatments in a factorial design in 5 × 5 m plots with 10 replications over a five year period. The results made a substantial contribution to the in situ comparative efficacy of the biological controls that were compared. The Trichoderma treatment was dropped from the experiment after three years due to lack of efficacy although there were relatively few significant differences among the number of apothecia or disease in any of the treatments assessed during these years. In at least two plots, including bean and chicory, that had been treated with Trichoderma spp., diseased crops developed up to 2.6 times more apothecia than the untreated control. The authors advanced no reason for these unexpected results and additional experimentation would be required to determine if this was a spurious effect, as suggested by Kiss (2004), or an initial observation of a more substantial biological phenomenon. In our view, it is too early to discriminate between these possibilities.
Perhaps one of the most challenging aspects of risk assessment for biological control organisms is to discriminate between significant, measurable effects on other organisms and important, long-lasting perturbations of other species. This has become an important discussion in several arenas of risk assessment and often requires considerable study and discussion to resolve. Because of the often inadequate amount of published information on risk assessment of biological controls for plant diseases to date, we consider it premature to distinguish effectively between what are significant vs. important non-target effects, whether this effect be in agricultural, forestry or natural ecosystems. There is increasing recognition that above-ground–below-ground communities can be important factors in terrestrial ecosystems, with both positive and negative feedbacks on species abundance and diversity (Wardle et al., 2004). In contrast to the view of Kiss (2004), non-target effects in any of these environments can potentially be important, and it seems evident that biological control microorganisms can move from areas of application to surrounding areas. As pointed out by Whipps & Lumsden (2001), the key determinants of whether biological controls are advantageous or not are influenced by scientific facts or observations, public opinion and perception, and commercial or financial considerations.
In conclusion, Kiss draws attention to ‘decision makers’ and the possible repercussions of non-target assessments on pest control product registration policies. We agree that ‘this is not a reason to stop using them … whenever a careful scientific investigation has shown that this is a feasible method and has no major non-target effects’ (Kiss, 2004). The challenge, of course, is for those of us involved in biological control research to provide scientific evidence to support these assumptions. Indeed, it was through discussions with regulatory agencies in Canada that one of the authors (GJB) became aware of many of the issues summarized in our article. The regulatory approval process for many countries requires substantial documentation of potential effects to non-target organisms, and on the environmental fate and persistence of the biological control microorganism. Such data can be challenging to collect and there is a need for more discussion on the most appropriate protocols for assessing the safety of such products. The letter by de Jong (1992) summarizes one example of how quantitative risk assessment in plant pathology and biological control can contribute to effective risk analysis and policy development.
We remain optimistic that an increased understanding of these questions can contribute to the successful understanding and development of microorganisms as effective and environmentally benign biological controls for plant disease, and we believe that our review raises important issues that must be considered when developing or using biological control products.