Enemy-risk effects and the evaluation of biological control agents
A primary task of biocontrol researchers is evaluating the impact of biological control agents on target and non-target organisms. Evaluations occur during each stage of a biocontrol project, whether the program is classical, augmentative, or conservation biocontrol. First, the initial step in most biocontrol programs is to describe, as quantitatively as possible, the natural enemy community associated with a target pest; for invasive species, this may involve describing food webs in both the native and invaded ranges. Second, as part of classical biological control programs, and in some cases augmentative biological control, candidate agents need to be screened for host/prey specificity to assess the risks of non-target impacts and to identify the most promising agent(s) for mass-rearing and release. Finally, after classical biocontrol agents are released and established, it is important to evaluate efficacy, including effects on targets and non-targets. The methods used in each of these stages of assessment are overlapping, and different methods can be complementary (Barratt et al. 2010; Furlong 2015; Macfadyen et al. 2015; Van Driesche 2016; Lövei & Ferrante 2017). However, as shown in Table 2, many methods, especially increasingly popular sequencing-based methods, capture only consumptive effects, and either partially or completely fail to record enemy-risk effects. Although a narrower focus on CEs is compatible with efforts to describe trophic webs, methods that reflect both CEs and enemy-risk effects will provide better efficacy assessments and measures of non-target effects. As noted in the literature review section, there are relatively few documented cases of enemy-risk effects on non-target species, but this likely reflects a failure to investigate them. The increasing prevalence of methods like immunoassays and sequencing-based approaches to detect direct parasitism or consumption may only exacerbate this lack of documentation. Because there is no guarantee that the magnitudes of CEs are strongly correlated with the magnitudes of risk effects, both must be included in evaluation methods, if not measured separately.
Projecting the non-target impacts of a candidate classical biological control agent is quite challenging. It is difficult to canvas what is often a very broad array of possible non-target species, each of which needs to be brought into quarantine, reared, and tested for vulnerability to attack. Furthermore, it is increasingly acknowledged that not just direct impacts, but also possible indirect effects of an introduction should be assessed, including the potential for competition, intraguild predation, and apparent competition effects (Hajek et al. 2016; Heimpel & Mills 2017). To this already imposing prospect, we add that it may be important to consider enemy-risk effects. In some cases, even the simple in-quarantine host-range testing protocols using small and simplified microcosms and short exposures to natural enemies can reveal some evidence of enemy-risk effects. For example, host-range tests of candidate parasitoid species may reveal elevated mortality of individual hosts that don’t produce parasitoid offspring (e.g., Abram et al. 2016; Bulgarella et al. 2017); in these cases, hosts may die following parasitoid probes without oviposition, or parasitoid oviposition may lead to early mortality of both the host and the parasitoid eggs prior to any consumption of the host. In some cases, such parasitoid-generated host mortality has been found in host species on which parasitoids never successfully produce offspring (Hoddle & Pandey 2014; Valente et al. 2017), emphasizing that parasitism rates alone may not suffice to capture non-target effects (Abram et al. 2019). Depending on response variables measured in target or non-target hosts or prey, including altered movement or microhabitat selection, development rates, feeding behavior, or reproduction, other risk effects could potentially also be detected in a quarantine setting, but current host-range testing generally side-steps the possible importance of these effects. Like indirect effects, however, many possible enemy-risk effects, including those expressed via longer-range movements, are not readily evaluated within a quarantine facility.
More encouragingly, many widely-used protocols for assessing the efficacy of biological control measure either target (or non-target) population density as the primary response variables (Table 2), thereby capturing the combined influences of CEs and NCEs. This is particularly true in studies of conservation biocontrol, which also frequently incorporate larger spatiotemporal scales and whole communities of enemies. Although it may sometimes be of academic interest to separate the roles of CEs and NCEs (but see Box 4), these protocols accomplish the central objective of capturing the full range of pathways through which natural enemies may contribute to herbivore population suppression.