The ability of generalist predators to switch among different prey species often is exploited in conservation biological control (Symondson et al., 2002). For example, plantings of wildflowers or other perennial refuges can provide pollen, nectar, and habitat for prey that can help build predator abundance and diversity (Patt et al., 1997; Blitzer et al., 2012; Balzan et al., 2016; Gurr et al., 2017). However, to be effective, predators must willingly leave the refuge and move into adjacent crop fields (Blaauw and Isaacs, 2012). For example, Middleton and MacRae (2021) found that several kilometers of wildflower plantings around potato (Solanum tuberosum L.) fields yielded a dramatic >50% increase in predator abundances in the refuge. However, these natural enemies did not readily leave the refuge for the cropping fields such that predation of Colorado potato beetle (Leptinotarsa decemlineata Say) eggs was not increased (Middleton and MacRae, 2021). A seemingly simpler approach would be to increase abundance of prey other than pests in the cropping field itself, so that predators can be conserved in-place (Agustí et al., 2003). Settle et al. (1996) found that plant thatch and reduced insecticide applications allowed generalist predator populations to build in Indonesian rice paddies, feeding on detritus-feeding prey, before plants emerged and were colonized by herbivores; the predators then switched to attacking pests as detritivores declined and herbivores increased (see also Brust, 1994; Stoner et al., 1996; Johnson et al., 2004). However, here too success is not guaranteed. For example, Halaj and Wise (2002) found that adding straw mulch to cucurbit plantings built densities of detritus-feeding prey and greatly enhanced generalist predator abundance, but pest control was not improved because the predators never switched to attacking herbivores. These examples highlight that alternative prey can indirectly enhance the biocontrol effectiveness of generalists in some situations, but disrupt it in others (Eubanks and Denno, 2000a,b; Harmon and Andow, 2004; Koss and Snyder, 2005; Symondson et al., 2006).

Another complexity when considering generalist predators as biocontrol agents, is that they often feed on one another in addition to pests (Rosenheim, 1998; Paul et al., 2020). Intraguild predation is most disruptive when a predator both infrequently feeds on the pest and heavily attacks the pest’s key natural enemy (Finke and Denno, 2004; Ives et al., 2005). It is important to note that biological control can be disrupted even when predators do not commonly feed on one another, if a predator species reduces its foraging activity to avoid becoming a victim of intraguild predation (Preisser et al., 2005, 2007). Often, prey and predator abundance and diversity interact to determine how often intraguild predation occurs (Finke and Denno, 2002). When predator abundance is relatively high and herbivores are uncommon, intraguild predation offers a way for predators to escape food limitation (Hironori and Katsuhiro, 1997). However, when prey is relatively plentiful, generalists might often encounter and kill herbivorous or detritus-feeding prey rather than natural enemies (Lucas et al., 1998). More broadly, a high diversity of other prey might allow predators to move into separate feeding niches that lead to fewer predator-predator encounters, and thus less intraguild predation (Schmitz et al., 1997; Letourneau et al., 2009; Schmitz, 2009; Dainese et al., 2017; Jonsson et al., 2017; Greenop et al., 2018). Of course, in the field, abundances of alternative prey and intraguild predators might widely vary from site to site and throughout the year, leading to complex indirect effects on pest suppression by generalists (Snyder, 2019).

Here, we use molecular gut-content analysis to track predation of herbivorous western flower thrips (Frankliniella occidentalis) by the predatory bugs Geocoris sp. and Nabis sp. [molecular identification failed to reveal a confident species determination for either predator; Krey et al. (2021)] in potato (S. tuberosum) fields. The crops were managed by growers using organic or conventional management practices, which creates site-to-site differences in predator and prey communities (Koss et al., 2005; Crowder et al., 2010; Krey et al., 2021). Both predator taxa are generalists that presumably feed on a broad diversity of arthropods, with thrips typically being among the most abundant herbivores at our study sites (Krey et al., 2021). Our central hypotheses were that (1) greater arthropod abundance and/or diversity would increasingly draw attacks away from thrips, reducing the probability of detecting thrips predation by Geocoris and Nabis, but that (2) this could be counteracted by reduced predator-predator interference in fields with more robust arthropod communities, indirectly enhancing foraging efficiency by the generalists.

Our project had three complementary components. First, we developed a species-specific PCR primer that allowed detection of DNA of F. occidentalis. Second, we surveyed densities of Nabis and Geocoris predators, thrips, and other arthropods that might serve as prey, in organic and conventional potato fields managed by cooperating commercial growers (see Krey et al., 2021). Third, during these arthropod community surveys we used molecular gut content analysis to test a subsample of Nabis and Geocoris adults for the presence of F. occidentalis DNA, using model fitting to attempt to link detection of thrips DNA to management and arthropod community metrics.

A simple comparison of predators collected from organic versus conventional fields did not show any significant difference in the probability of detection of western flower thrips DNA in either Geocoris or Nabis (Figure 1). However, this did not mean that farming system had no impact. Our model fitting efforts that considered several aspects of arthropod community structure alongside management, revealed several interesting interactions. Geocoris foraging in conventional fields were more likely to have thrips DNA detections with increasing total arthropod abundance (Figure 2B). The opposite pattern was found in organic fields, however, with greater arthropod abundance correlated with a lower probability of detection of thrips predation by Geocoris. One possible explanation is that, at the relatively low prey abundance and diversity typical of conventional potato fields (Krey et al., 2021), increases in prey trigger greater Geocoris foraging activity that leads a greater chance that they will find and eat thrips. In contrast, the plentiful prey in organic fields may instead draw attacks away from thrips as Geocoris feast on other prey. Another possibility is that very low arthropod abundances in conventional fields correlated with recent insecticide applications, with sublethal effects reducing overall predator foraging (Stark et al., 1995; Biondi et al., 2013). In either case, observations of Geocoris foraging behavior under low and high prey conditions, and in the presence versus absence of insecticide residue, would be needed to discern between these possible explanations. For Geocoris we also observed an interaction between western flower thrips abundance and the probability of detecting thrips DNA (Figure 2A). However, greater thrips abundance correlated with greater probability of thrips DNA detection, with the interaction perhaps simply resulting from a steeper relationship in organic than conventional fields (Figure 2A). In future work, it may be helpful to separate thrips collected in suction samples into life stages, or conduct open-field observations of predator foraging, to further delineate which thrips life stages were present and being attacked by which predator species. A final possibility is that organic fields were weedier, which might have complicated predator foraging to alter feeding relationships (e.g., Blubaugh et al., 2021) in organic versus conventional potato fields.

Interestingly, for both predator species, the probability of detection of thrips predation generally decreased in fields with higher Nabis abundance. This is consistent with greater predator-predator interference where Nabis was more abundant, leading either to reduced overall foraging or a switch away from predation on thrips. Previous work suggests that either explanation is possible. Nabis is an effective intraguild predator of Geocoris (Snyder et al., 2006), and also is highly cannibalistic (Takizawa and Snyder, 2011), such that predators might face heightened risk when foraging where Nabis is abundant. Geocoris in these fields do appear to feed more heavily on detritus-feeding S. pallida flies, rather than attacking aphids, in fields where other predator species are relatively more abundant; this suggests a feeding-niche shift when the threat of intraguild predation is higher (Krey et al., 2021). Altogether, these findings suggest another case where the contribution of generalist predators to biocontrol is reduced by altered foraging to reduce the risk of intraguild predation (e.g., Prasad and Snyder, 2006; Hosseini et al., 2021).

Molecular gut content analysis allows the inference of predation patterns under open field conditions, where predator-prey interactions naturally occur, without the constraints of caging or other artificial manipulations (King et al., 2008). However, the method does have its limitations that must be acknowledged. We could not discern how many thrips of what stages were consumed, if they were alive when consumed, or if the predator had eaten another natural enemy that had itself eaten thrips. Of course, scavenging or intraguild predation do not contribute to thrips suppression and might well weaken it (Juen and Traugott, 2005). All of the results reported here result from models that look for correlations among factors that differ among sites, but were not directly manipulated. It then remains uncertain whether the correlations reported here reflect true cause-effect relationships. Additionally, arthropod community metrics are often highly correlated (Supplementary Figure 1), making it difficult to isolate individual effects. Clearly, additional experimental work, ideally alongside observations of predator foraging behavior, are needed to further elucidate possible feeding relationships in this arthropod community. This would be particularly powerful if enough different fields could be sampled to construct and test the fit of Structural Equation Models, which could include explicit examination of indirect interactions suggested above (e.g., Blubaugh et al., 2021). Unfortunately, logistical constraints prevented us from sampling more fields in the study presented here.

Organic farming relies on natural processes, wherever possible, as an alternative to chemical interventions to control pests. This approach consistently leads to higher arthropod diversity in organic than conventional fields, including among natural enemies (Bengtsson et al., 2005; Hole et al., 2005; Crowder et al., 2010, 2012). Yet, pest abundance also generally is higher in organic fields, and greater natural pest suppression is not always apparent (Hilbeck and Kennedy, 1996; Macfadyen et al., 2009a,b; Schmidt et al., 2014; Muneret et al., 2018; Cloyd, 2020). The findings presented here suggest ecological complexities that might contribute to these general patterns. First, at the higher arthropod abundances typical of organic fields, growing abundance of possible prey correlated with reduced probability of detecting thrips DNA in Geocoris. Second, greater abundance of Nabis generally correlated with reduced probability of detection of thrips DNA in both Geocoris and Nabis. Higher predator abundance in organic fields, then, might lead to greater predator-predator interference that defuses any gains for pest suppression. So, robust arthropod communities may not necessarily translate into more effective biological control.

The effectiveness of generalist predators as biological control agents has long been questioned (DeBach and Rosen, 1991; Symondson et al., 2002). This is because the same polyphagy that allows predators to build their populations on detritus-feeding or other non-pest prey, can sometimes also distract them from attacking key herbivores (Harmon and Andow, 2004). Likewise, predators that feed heavily on other natural enemies might disrupt, rather than strengthen, net pest suppression (Rosenheim, 1998; Venzon et al., 2001; Finke and Denno, 2004; Janssen et al., 2007). We found evidence that these two disruptive interactions might reinforce one another, as detection of thrips DNA in predators was reduced both in the presence of abundant arthropod prey and with increasing abundance of predators perhaps drawn to those prey. This reinforces the complexity of feedbacks that might be seen in open field situations, where prey and predator abundance interact with one another in complex ways (Paul et al., 2020). Molecular gut content analysis, despite its limitations, may be a particularly powerful tool to detect these relationships against the high background arthropod diversity typical of real agricultural fields.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/genbank/, MZ677017 – MZ677025, https://www.ncbi.nlm.nih.gov/genbank/, MZ677026 – MZ677035, and https://www.ncbi.nlm.nih.gov/genbank/, MZ677036 – MZ677047. Raw arthropod abundance and diversity, and gut content, data are provided in the Supplementary Material of this article.

OS and WS led writing of the initial draft. EC and JH led molecular work. OS and MC led data analysis. DC, ZF, KK, CL, and GS led field work. All authors contributed to manuscript editing and revision.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.