INTRODUCTION

A diverse group of insect species exhibit environmentally determined morphotypes, or polyphenisms, for dispersal that take the form of flight-capable (typically winged) or flightless individuals (Braendle et al., 2006; Hochkirch & Damerau, 2009; Xu & Zhang, 2017). These dispersal polyphenisms are generally thought to be driven by trade-offs between potential reproduction and survival in temporally variable environmental conditions (Harrison, 1980; McPeek & Holt, 1992; Roff, 1986; Simpson et al., 2011; Zera & Denno, 1997). When environmental conditions are optimal in a patch of habitat, maximising reproductive output is favoured over dispersal if flightless individuals exhibit higher fecundity than winged individuals. As environmental factors decline, dispersal is favoured if it can lead to occupation of better habitat. This can result in species using changes in one or several environmental factors to provide cues about when dispersal morphs will likely increase fitness.

Depending on how variable individual responses are, environmental cues can induce potentially large shifts in the phenotypic structure of a population, from flightless to flight-capable or vice versa (Crossley et al., 2022; Lagos-Kutz et al., 2020). These shifts can affect the population dynamics of the polyphenetic species by changing immigration or emigration between subpopulations in the surrounding area. A species with dispersal polyphenism could experience sporadic rates of gene flow, colonisation of patches, disease transmission, and overall population growth depending on how spatially variable the environmental cues are (Hanski & Woiwod, 1993; Lagos-Kutz et al., 2020; Lin et al., 2018; Mazzi & Dorn, 2012; Simpson & Sword, 2008). Understanding what environmental factors are used for cues can help forecast the population dynamics of polyphenetic species and how the surrounding community will be affected (Sun et al., 2015; Wennersten & Forsman, 2012; Williams & Dixon, 2007).

The pea aphid (Acyrthosiphon pisum Harris) is a sap-feeding hemipteran that is an agricultural crop pest on several species in the Fabaceae family including peas, alfalfa, beans, and clover (Holman, 2009). During their parthenogenetic stage throughout the growing season, populations can double in size in as few as 3 days due to high fecundity and short development time (Harmon et al., 2009; Hutchison & Hogg, 1984, 1985). Pea aphids, along with most other aphid species, exhibit a polyphenetic wing dimorphism in their adult stage. The wing dimorphism is found only in parthenogenetic females during the summer growing season and is developmentally determined before birth by the mother depending on environmental cues she experiences (Müller et al., 2001; Sutherland, 1969a).

Environmental factors associated with changes in the propensity of pea aphid mothers to produce alates (winged offspring) has been intensively studied (Braendle et al., 2006). Crowding due to high pea aphid densities has been shown to increase alate production (proportion of offspring with wings) via an increase in tactile stimulation (antennal contact with other aphids) (Sutherland, 1969a). An increase in alate production in response to increasing host plant maturity has also been observed. This increase has been suggested to be a result either of increased movement leading to increased tactile stimulation or directly from the nutritional quality of the host plant (Sutherland, 1969b). Although the effects of temperature on alate production in pea aphids have not been directly investigated, studies conducted on other aphid species have shown an increase in alate production at lower temperatures. Much like the case for host plant quality, it is unclear whether the mechanism for alate response to temperature is direct or indirectly caused by increased tactile stimulation through changes in movement (Johnson, 1966; Schaefers & Judge, 1971). Natural enemies (predators and parasitoid wasps) have also been shown to increase pea aphid alate production, with natural enemy presence having a greater effect than factors associated with prior natural enemy presence (e.g., eggs, frass, and kairomones) (Dixon & Agarwala, 1999; Purandare et al., 2014). A wide range of natural enemy species has been investigated and results suggest that increased alate production can be attributed to the release of an aphid alarm pheromone ((E)-β-farnesene) causing an increase in aphid movement and inducing crowding effects (Hatano et al., 2010; Kunert et al., 2005; Podjasek et al., 2005; Sloggett & Weisser, 2002; Weisser et al., 1999). These effects reported for natural enemies are distinct from the ability of some parasitoid species to subvert wing development in parasitized juvenile aphids (Christiansen-Weniger & Hardie, 2000). Pea aphid adults infected with fungal pathogens and colonies with fungus-infected individuals may have higher alate production due not only to increased movement and tactile stimulation but also to physiological cues expressed by infected individuals (Hatano et al., 2012). Finally, pea aphids on Pisum sativum have been shown to produce more alates when feeding on plant tissue infected with pea enation mosiac virus (Hodge & Powell, 2010). Although a diversity of factors can affect the propensity of pea aphid mothers to produce alates, crowding effects through tactile stimulation seem to play the largest role, with many other factors operating by changing the degree of tactile stimulation experienced.

Although alate production in pea aphids has been intensively studied, work has largely been confined to controlled conditions in laboratory settings where individual or colony responses to separate environmental factors are recorded. Little work has been conducted linking the results from these studies to alate production in the field, and therefore there is little evidence that cues affecting alate production in the lab have sufficient variation in the field to explain natural fluctuations in alate abundance. However, the wide range of factors shown to affect alate production creates a challenge in linking the results of lab studies to field observations, because many of the factors may be correlated in the field; for example, pea aphid abundance and natural enemy abundance are likely to be positively correlated, making it difficult to separate these two possible environmental factors explaining variation in alate production. Nonetheless, the challenge presented by correlated explanatory variables can be addressed with appropriate statistical approaches.

In this study, we identified which of the cues that have been shown to affect pea aphid alate production in the lab can also explain variation in alate production in the field. Specifically, we analysed the effects of pea aphid abundance, alfalfa height (as a measure of plant maturity), temperature, predators, parasitism, fungal infection, and herbivores on alate production in 5–9 alfalfa fields over 3 years. We included herbivore abundance even though no prior study had investigated their effects on alate production; this allowed us to compare herbivore abundance with predator abundance to see if competition or predation risk had differing effects. We used statistical methods designed to account for collinearity between predictor variables (environmental cues) and for spatiotemporal correlation between observations across the samples taken twice weekly from each field. Because previous studies showed or suggested crowding to be a major cue for alate production, we expected pea aphid abundance to be the strongest predictor of alate production in the field. This is the first study of which we know investigating the roles of multiple environmental factors on pea aphid alate production in field populations.