The effect of the coccinellid Harmonia axyridis (Coleoptera: Coccinellidae) on transmission of the fungal pathogen Pandora neoaphidis (Entomophthorales: Entomophthoraceae)

. The coccinellid Harmonia axyridis is a recent arrival in the UK and is an intraguild predator of the entomopathogenic fungus Pandora neoaphidis . Harmonia axyridis entirely consumes P. neoaphidis -sporulating cadavers and this may have a negative effect on the epizootic potential of P. neoaphidis . Here we assessed within plant transmission, and between plant vectoring, of P. neoaphidis in the presence of either H. axyridis or Coccinella septempunctata , a native coccinellid that only partially consumes fungal cadavers. Transmission was greater in the presence of coccinellids, with 21% of aphids becoming infected with the fungus whilst only 4% were infected in the control. However, there was no significant effect of coccinellid species or sex on fungal transmission.


INTRODUCTION
Aphid populations are regulated by predators, pathogens and parasitoids. Interactions between aphid enemies may have positive, neutral or negative effects on pest control (Rosenheim et al., 1995;Ferguson & Stiling, 1996;Straub et al., 2008). It is difficult to predict the outcome of intraguild interactions on pest suppression, especially when a non-native polyphagous predator enters the guild. Harmonia axyridis (Pallas) (Coleoptera: Coccinellidae), known as the multicoloured Asian lady beetle and the harlequin ladybird, is a polyphagous coccinellid species that is not native to Britain but established in 2004 and is predicted to have a negative impact on biodiversity Roy et al., 2006;Brown et al., 2008;. Although H. axyridis is a dominant competitor over many native coccinellid species, the interactions between H. axyridis and noncoccinellid aphid natural enemies are less well studied . Here we present data on the interactions between H. axyridis and an aphid-specific pathogen.
Pandora neoaphidis (Remaudière & Hennebert) Humber is an aphid-specific entomopathogenic fungus in the order Entomophthorales that can cause epizootics in field populations (Feng et al., 1992;Pell et al., 2001;Barta & Cagan, 2006). Transmission of P. neoaphidis occurs via conidia that are forcibly ejected from dead, infected aphid cadavers and can remain infective for up to 14 days (Brobyn et al., 1985). These conidia are either deposited on aphid hosts, on the substrate surrounding the P. neoaphidis-sporulating cadavers or may pass through the boundary layer of the plant and passively disperse on wind currents to new host populations (Brobyn et al., 1985, Hemmati et al., 2001. The presence of foraging native predators, parasitoids and extraguild co-occurring arthropods all increase local transmission of P. neoaphidis by increasing aphid movement and, therefore, the encounter rate with infective conidia (Pell et al., 1997;Feuntes-Contreras et al., 1998;Roy et al., 1998Roy et al., , 2001Baverstock et al., 2008). The native aphid predator Coccinella septempunctata L. (Coleoptera: Coccinellidae) also vectors P. neoaphidis to previously uninfected aphid colonies on different plants, thereby enhancing dispersal (Roy et al., 2001).
Coccinella septempunctata is an intraguild predator of P. neoaphidis-sporulating cadavers, but rarely entirely consumes these cadavers in Petri dish experiments (Roy et al., 1998. Although conidia production from partially consumed P. neoaphidis-sporulating cadavers is reduced there is no significant reduction in transmission of the fungus (Roy et al., 1998). In contrast, H. axyridis completely consumes P. neoaphidis-sporulating cadavers in Petri dish arenas and does not discriminate between dead uninfected aphid prey and sporulating cadavers . The intraguild interactions between P. neoaphidis and H. axyridis may therefore be different to those with C. septempunctata. Here we compare the effect of H. axyridis and C. septempunctata on the transmission of P. neoaphidis within an Acyrthosiphon pisum (Harris) colony and investigate the potential of the cocci-Abstract. The coccinellid Harmonia axyridis is a recent arrival in the UK and is an intraguild predator of the entomopathogenic fungus Pandora neoaphidis. Harmonia axyridis entirely consumes P. neoaphidis-sporulating cadavers and this may have a negative effect on the epizootic potential of P. neoaphidis. Here we assessed within plant transmission, and between plant vectoring, of P. neoaphidis in the presence of either H. axyridis or Coccinella septempunctata, a native coccinellid that only partially consumes fungal cadavers. Transmission was greater in the presence of coccinellids, with 21% of aphids becoming infected with the fungus whilst only 4% were infected in the control. However, there was no significant effect of coccinellid species or sex on fungal transmission. Between plant vectoring occurred infrequently in the presence of both species of coccinellid. The effect of H. axyridis on P. neoaphidis transmission is, therefore, likely to be similar to that of the native coccinellid C. septempunctata. nellids to vector P. neoaphidis between A. pisum colonies.

MATERIAL AND METHODS
Single plant arenas, each consisting of one sixteen day old Vicia faba plant (L. Cultivar: The Sutton) contained in a lamp glass (1.4 litre capacity) were used. Each plant was infested with twenty adult A. pisum aphids and maintained at 18°C (16L : 8D), as described by Roy et al. (1998). Eight treatments were prepared: (1) no enemy control, (2) % H. axyridis, (3) & H. axyridis, (4) P. neoaphidis, (5) % H. axyridis + P. neoaphidis, (6) & H. axyridis + P. neoaphidis, (7) % C. septempunctata + P. neoaphidis and (8) & C. septempunctata + P. neoaphidis. Where required, P. neoaphidis was added as rehydrated pairs of P. neoaphidis-sporulating cadavers on a water agar plug (isolate X4, from the Rothamsted Research collection, original host = A. pisum). One cadaver pair was placed on the adaxial surface of the four largest top leaves. The experiment was split into two consecutive parts, (a) within plant transmission and (b) between plant vectoring. (a) Where required, a single adult coccinellid (sex recorded) that had been starved for 24 h was added to arenas. Coccinellids were removed after 4 h followed by P. neoaphidis and the aphids. Aphids were counted and transferred to new single plant arenas previously free from aphids. (b) Coccinellids were transferred to new single plant arenas (1 per arena) containing twenty uninfected adult A. pisum and were allowed to forage for 16 h. All arenas were sealed with cling film for the first 24 h of the experiment to ensure a high relative humidity to allow the fungus to germinate, after this time the cling film was replaced with muslin. The numbers of uninfected and P. neoaphidis-infected aphids were assessed after a further five days. The experiment was set up as a completely randomized block design with three blocks of ten arenas, and was repeated on three occasions. Each coccinellid treatment was tested once in each block and the coccinellid absent treatments twice. In total, each coccinellid treatment was replicated nine times and each coccinellid-absent treatment 18 times. The proportions of aphids recovered after the initial 4 h foraging period in the presence or absence of H. axyridis (treatments 1-7) were analysed using logistic regression (generalized linear model with binomial error and logit link) in GenStat (Payne et al., 2009). For the fungus-present treatments, the proportions of P. neoaphidis infected aphids recovered at the end of the experiment were also analysed using logistic regression. The overall treatment effect was partitioned into contrasts representing a comparison between the coccinellid-absent treatment (treatment 4) and all coccinellid-present treatments combined (treatments 5-8), and the main effects of coccinellid species, coccinellid sex and their interaction within the coccinellid-present combinations. Over dispersion was accounted for by comparing ratios of deviances to the residual mean deviance against critical values of the F distribution. Means and 95% confidence intervals shown are back-transformed from the logistic scale.

RESULTS AND DISCUSSION
Foraging by H. axyridis decreased the proportion of aphids recovered ( 2 = 47.44, df = 1, p < 0.001) ( Table  1). There was no difference in the proportion of aphids eaten by H. axyridis in the presence or absence of P. neoaphidis ( 2 = 1.08, df = 1, p = 0.345) or by male and female H. axyridis ( 2 = 1.68, df = 1, p = 0.20) (Table 1). There was no interaction between these treatments ( 2 = 0.55, df = 1, p = 0.463). Significantly more aphids became infected with P. neoaphidis in treatments that contained a foraging coccinellid compared to those that did not (F1,41 = 33.42, p < 0.001) with infection levels at 21% (95% CI: 15-28%) and 4% (95% CI: 2-8%) respectively. However, transmission in the presence of H. axyridis and C. septempunctata was not significantly different (F1,41 = 0, p = 0.997) with a mean of 22% (95% CI: 7-51%) and 21% (95% CI: 7-49%) P. neoaphidisinfected aphids recovered in the presence of H. axyridis and C. septempunctata respectively (Fig. 1). There was no significant effect of coccinellid sex on transmission (F1,41 = 0, p = 0.973) nor was there an interaction between species and sex (F1,41 = 1.06, p = 0.309) (Fig. 1). The complete consumption of a P. neoaphidis-sporulating cadaver by a coccinellid was only observed on one occasion and was by a female H. axyridis. Partial consumption of cadavers may have occurred but this could not be determined after sporulation. Vectoring occurred on three occasions. Two C. septempunctata males vectored P. neoaphidis, with a single adult aphid becoming infected on each occasion. One female H. axyridis vectored P. neoaphidis, infecting three nymphs. There were no P. neoaphidis-infected aphids in treatments that did not contain P. neoaphidis.  Under natural conditions H. axyridis, C. septempunctata and P. neoaphidis co-occur on plants such as bean and common nettle (unpubl. data). The transmission of P. neoaphidis was similar in the presence of both coccinellid species despite H. axyridis being a greater intraguild predator of P. neoaphidis in previous Petri dish studies . The positive effect on transmission in the presence of this non-native species is therefore similar to that found in the presence of native intra-guild and extra-guild species (Roy et al., 1998;Baverstock et al., 2008Baverstock et al., , 2009. Similar increases in P. neoaphidis transmission as a result of foraging by C. septempunctata were found by Roy et al. (1998) andEkesi et al. (2005). Lower levels of transmission were found in this study, possibly because aphids were transferred to clean plants after being exposed to foraging coccinellids thereby preventing further transmission of conidia from the plant surface. Alternatively, the lower levels of transmission observed in this study may have been due to the age-dependent susceptibility of A. pisum to P. neoaphidis (Milner, 1982(Milner, , 1985Lizen et al., 1985). Indeed, Roy et al. (1998) assessed transmission of P. neoaphidis to 4 th instar A. pisum which may have been more susceptible to P. neoaphidis than the adult aphids assessed here.

88
In the current study, no fungal cadavers were consumed by C. septempunctata and only one fungal cadaver was completely consumed by H. axyridis within four hours. This was surprising given that H. axyridis showed no discrimination between dead uninfected aphids and P. neoaphidis-sporulating cadavers in previous Petri dish experiments whereas C. septempunctata showed a preference for uninfected aphids . An increase in the proportion of uninfected prey will result in a reduced encounter rate between the coccinellid predator and fungal cadavers and, if the predator does not show a preference between prey types, this may decrease the predation of fungal cadavers. In addition, the presence of P. neoaphidis did not affect the predation of uninfected aphids by H. axyridis, therefore, P. neoaphidis may not disrupt aphid suppression by H. axyridis at the single plant scale.
Vectoring of conidia by non-host insects has been observed for entomopathogenic fungi in both the Hypocreales and the Entomophthorales and has been investigated in three collembolan species (Dromph, 2003), a black ant (Bird et al., 2004) and a bug (Down et al., 2009). In this study H. axyridis and C. septempunctata vectored P. neoaphidis at a similar rate. However, vectoring was very infrequent and experiments at a more realistic scale are required to determine the importance of vectoring for P. neoaphidis. Roy et al. (2001) showed that C. septempunctata could vector P. neoaphidis and that vectoring events are highly irregular, with no correlation between initial level of pathogen exposure and the number of aphids that became infected. In addition, C. septempunctata adults that had foraged on non-crop plants found in field margins such as nettle, knapweed or birds foot trefoil containing P. neoaphidis were able to vector the pathogen to aphids on bean plants (Ekesi et al., 2005). In contrast to the passive dispersal of infective conidia on wind currents, the dispersal of P. neoaphidis by coccinellids is targeted as the pathogen is directly transported between aphid colonies. When P. neoaphidisinfected cadavers are low in number or the habitat is diverse, targeted dispersal by coccinellids could be the most important mode of dispersal (Roy et al., 2001;Baverstock et al., 2010). However, Roy et al. (2003) found that C. septempunctata inoculated with P. neoaphidis were only able to vector the pathogen within 4 h of inoculation and conidia vectored onto plants by foraging C. septempunctata were only infective up to 24 h post conidia dispersal (Roy et al., 2003). Nonetheless, vectoring is seen as an important form of dispersal and methods of manipulating this in augmentative and conservation biocontrol strategies are being investigated (Furlong et al., 1995;Bird et al., 2004;Ekesi et al., 2005;Down et al., 2009;Baverstock et al., 2010).
In conclusion, H. axyridis increases within plant transmission of P. neoaphidis and can vector the pathogen between aphid populations on different plants. The effect H. axyridis could have on P. neoaphidis epizootiology is, therefore, likely to be similar to that of the native coccinellid C. septempunctata. Experiments at larger temporal and spatial scales under more realistic conditions are needed to determine the effect H. axyridis will have on P. neoaphidis at the landscape level.