Many plant viruses and other vector-borne plant pathogens can affect host plant traits as well as vector preferences and performance in ways that facilitate transmission (Ingwell et al., 2012; Mauck et al., 2014; Eigenbrode et al., 2018). However, while many plant viruses have multiple hosts and are transmitted by generalist arthropod vectors (Power and Flecker, 2003), most of the theory and empirical work on host–vector–pathogen interactions has been built upon studies involving a single host species (Eigenbrode et al., 2018). The ability of vectors to move between host species can affect virus prevalence by facilitating interspecific transmission between transient hosts, or by ensuring the maintenance of high vector populations over time even when hosts are scarce (Duffus, 1971; Swei et al., 2011; Ashby et al., 2014; McLeish et al., 2018). Yet, relatively few studies have assessed how viruses may directly or indirectly affect vector preferences or performance across multiple host species.

Many insect herbivores, including vector species, develop specialized races to specific hosts due to natural selection and genetic drift (Drès and Mallet, 2002; Loxdale et al., 2011). Specialized insect populations exhibit higher performance on their associated hosts and prefer these hosts to others in their potential host range. Similarly, adaptation of plant viruses to specific host immune systems often leads to specialization via antagonistic pleiotropies that limit viruses’ ability to overcome host defenses across plant taxa (García-Arenal and Fraile, 2013). On one hand, plant pathogens and vectors experience selection pressure towards host specialization, but specialization can become maladaptive when hosts are transient or pathogens are restricted to a small number of insect vectors.

The pea aphid, Acyrthosiphon pisum Harris (Hemiptera: Aphididae), is a generalist vector that feeds on hundreds of legume species (Fabaceae) (Eastop, 1971; Peccoud and Simon, 2010). However, populations of pea aphids often associate with specific hosts (Via et al., 2000; Peccoud et al., 2008; Eigenbrode et al., 2016). The pea aphid is also a vector for over 30 viruses such as Pea enation mosaic virus (PEMV) and Bean leafroll virus (BLRV), pathogens that infect several legumes (Rashed et al., 2018). Studies show that PEMV and BLRV affect host attractiveness and vector performance (Hodge and Powell, 2010; Wu et al., 2014; Davis et al., 2017), but it is not known whether these viruses affect performance or preferences of pea aphids with ingrained host associations. We predict that if viruses require interspecific transmission for persistence, vector host preference and performance across hosts may be altered by viruses (Chesnais et al., 2019).

Here we tested the idea that host preferences and performance of pea aphids from colonies with distinct host associations will be altered when hosts are infected with viruses. While host preferences and performance seem to be deeply imprinted in pea aphids (Via, 1991; Eigenbrode et al., 2016), whether pathogens persist in transient host populations may cause an evolutionary conflict with consequences for the evolution of host–virus–vector pathosystems. In the Pacific Northwest US (PNW), pea aphids often associate with either alfalfa or pea crops (Eigenbrode et al., 2016), but perennial legumes such as hairy vetch, red clover, and alfalfa can sustain early pea aphid infestations in spring and are reservoirs for PEMV or BLRV (Rashed et al., 2018; Clark et al., 2023). We hypothesize that plants infected with PEMV or BLRV alter pea aphid performance and preference in such a way that interspecific transmission is favored. Here, we used a series of experiments to test whether aphid colonies with specific host associations had altered host preferences and performance when hosts were infected with viruses as compared to uninfected. Overall, our study is among the first to test the notion that viruses manipulate host preferences and performance for multiple hosts in ways that promote interspecific transmission.

Plant virus infection of hosts can alter performance and preference of pea aphid, consistent with manipulation (Wu et al., 2014; Davis et al., 2017). In this study, we predicted that pea aphid host preferences and performance would depend on virus infection status across host species. We found that the effect of virus infection on aphid performance was not consistent across all tested host species. Plants infected with BLRV affected aphid performance, with higher aphid abundance on BLRV-infected alfalfa and lower on BLRV-infected vetch. While we did not find any changes in aphid performance among PEMV-infected hosts, across all host species, aphids were slightly more abundant on PEMV-infected plants. However, no changes to aphid preferences among infected or sham plants were found, supporting the assertion that viruses may indirectly affect vectors by altering their fitness similarly across host species (Mauck et al., 2018). Without concomitant behavioral changes, we only have partial support for effects of viruses on vector host preferences, although evidence suggests aphids tend to prefer plants where they perform better, regardless of virus-infection status.

Although most studies of vector-borne plant virus transmission focus on a single host plant species, in field settings many plant viruses often rely on transmission among multiple hosts, especially when some hosts are annual or transient (Wilke et al., 2006; Elena et al., 2009). This phenomenon is important for annual crop systems (such as pea, lentil, and garbanzo bean) where vectors and pathogens must use perennial hosts when crop plants are not present. Consequently, it may be adaptive for viruses to facilitate interspecific transmission by affecting vector host preferences and performance. For example, we predicted that host plants infected with PEMV would trigger different aphid host preferences compared to sham-inoculated plants; specifically, we predicted that host associations would be weaker when viruses were present if viruses break down specific plant defenses. However, we did not find evidence that host preferences among aphids differed when presented with PEMV-, BLRV- or sham-infected plants.

It is possible that transient virus infections affect aphid performance, but longer-term virus infection might be required to alter host preferences. Enzyme linked immunosorbent assay (ELISA) tests carried out before the experiments confirmed the presence of PEMV and BLRV in 41.55% and 45.45% of the test plants that were inoculated with infectious aphids, respectively. Although the virus detection did not reach the expected levels, we rely on the high transmission efficiency of both viruses by pea aphids (Fargette et al., 1982; Skaf and Makkouk, 1988) to ensure that inoculated plants had a progressing infection. In fact, the occurrence of false negatives in plant virus detection is common when the disease has not progressed sufficiently due to the irregular distribution of plant virus titers within infected hosts (Hipper et al., 2013; Lacroix et al., 2016). Regardless of the virus detection rate, while pea aphids can detect a wide range of plant volatiles, facilitating host recognition (Robertson et al., 2019; Shih et al., 2023), these cues and others from infected plants may not be altered sufficiently to modify aphid preference until later stages of disease progression than we tested here. We carried out the whole data analysis only using the plants that tested positive for the inoculated viruses and the results were similar, except for the reduced statistical power due to the reduced sample size ( Tables S4 , S5 ; Figure S2 ). Finally, our study examined plant-mediated mechanisms rather than comparing behavioral differences in viruliferous compared to non-viruliferous vectors (Chesnais et al., 2020). For example, beet leafhoppers carrying Beet curly top virus have increased probing behavior (Lee et al., 2022), while bird-cherry oat aphids carrying Barley yellow dwarf virus prefer perennial grasses relative to other hosts. Additional work in pea aphids should consider comparisons of host preferences and performance for viruliferous and non-viruliferous aphids (e.g., Chesnais et al., 2020).

Many generalist insect herbivores exhibit higher performance on single species of hosts, and these often are host species they could have fed on for multiple generations (Garrido et al., 2012). Such mechanisms have long been predicted to drive evolution of dietary specialization, generated by selective pressures to adapt to host resources present in specific environments (Van Zandt and Mopper, 1998). Pea aphids are a globally-distributed species, but their hosts vary worldwide, and unique populations form clonal associations with both crop and non-crop hosts, with some populations exhibiting local adaptation (Caillaud and Via, 2000). Host specialization implies that there will be performance trade-offs that often match host preferences and reduce the diversity of plants that herbivores can exploit (Berlocher and Feder, 2002). For example, our study supports prior work that shows “pea colonies” have dramatically reduced performance on alfalfa compared to competing “alfalfa colonies” (Eigenbrode et al., 2016); however, we did not find evidence for a similar trade-off for “alfalfa colonies” aphids with respect to pea.

We also predicted that virus-infection status of host plants would alter host associations in aphids. For example, our hypothesis was that if viruses increased performance on alternative hosts, that vector host associations might weaken when viruses were present. In other words, if viruses facilitate greater feeding breadth by aphids by increasing performance across hosts, aphid preferences for particular hosts should diminish. However, neither of our factorially-designed bioassays show a disruptive effect of virus-infection status on host preferences by pea aphids, suggesting that indirect effects of viruses on vector host preferences and performance may indeed be operating independently. To resolve this issue, further work could explore how plant responses to infection, and resulting effects on vectors, operate over longer time periods.

Correlations between aphid host preferences, performance, and reproductive isolation have been established as a function of local adaptations or recent co-evolution (Via et al., 2000; Hawthorne and Via, 2001; Queller and Strassmann, 2018). By facilitating higher reproductive output on specific host species, strong host associations can exacerbate outbreaks of insect herbivores in agroecosystems or increase the likelihood of outbreaks occurring in the future. Locally adapted herbivore populations may be more capable of breaking down the defenses of specific hosts that are abundant in the local environment, but not those of hosts that are rare (Scriber, 2002; Garrido et al., 2012). We did indeed observe some additive effects between aphid colonies associated with alfalfa and viral infection. BLRV-infected alfalfa plants had higher pea aphid abundance, and “alfalfa colonies” performed better on alfalfa hosts. While only isolated to a single host species in our bioassays, the pattern suggests synergy between local adaptation and vector-borne pathogens which could exacerbate economic damage from pea aphids. However, host manipulation by viruses may slow viral epidemics within crops by encouraging viruliferous vectors to move into non-crop plants (Shoemaker et al., 2019) or by causing vectors to orient towards otherwise inferior host plants (Mauck, 2016). Overall, our study provides evidence that gauging risk of outbreaks of vectors and associated plant pathogens requires an integrative approach that considers how viruses may alter vector performance and host preferences across a range of locally-relevant host species.

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: OSF, 10.17605/OSF.IO/KNYTA.

The manuscript presents research on animals that do not require ethical approval for their study.

SE originated the ideas. RC and SE completed bioassays. RC performed analyses under advisement by DC and SE. All authors contributed to the article and approved the submitted version.