Inter‐ and intraspecific variation of spider mite susceptibility to fungal infections: Implications for the long‐term success of biological control

Abstract Spider mites are severe pests of several annual and perennial crops worldwide, often causing important economic damages. As rapid evolution of pesticide resistance in this group hampers the efficiency of chemical control, alternative control strategies, such as the use of entomopathogenic fungi, are being developed. However, while several studies have focused on the evaluation of the control potential of different fungal species and/or isolates as well as their compatibility with other control methods (e.g., predators or chemical pesticides), knowledge on the extent of inter‐ and intraspecific variation in spider mite susceptibility to fungal infection is as yet incipient. Here, we measured the mortality induced by two generalist fungi, Beauveria bassiana and Metarhizium brunneum, in 12 spider mite populations belonging to different Tetranychus species: T. evansi, T. ludeni, and T. urticae (green and red form), within a full factorial experiment. We found that spider mite species differed in their susceptibility to infection by both fungal species. Moreover, we also found important intraspecific variation for this trait. These results draw caution on the development of single strains as biocontrol agents. Indeed, the high level of intraspecific variation suggests that (a) the one‐size‐fits‐all strategy may fail to control spider mite populations and (b) hosts resistance to infection may evolve at a rapid pace. Finally, we propose future directions to better understand this system and improve the long‐term success of spider mite control strategies based on entomopathogenic fungi.

Both intra-and interspecific variability in host susceptibility to infection may modify epidemiological patterns of parasite in natural host populations (Dwyer, Elkinton, & Buonaccorsi, 1997;Hawley & Altizer, 2011;Read, 1995), thereby altering the efficiency and environmental persistence of biocontrol agents. Moreover, the use of such agents generates a strong selection pressure on the target pests (e.g., Fenner & Fantini, 1999, see also Tabashnik, 1994, Moscardi, 1999 and, in general, variability in host susceptibility to infection may have important consequences for the evolution of host resistance as well as parasite virulence and transmission (Elena, 2017;Sorci, Moller, & Boulinier, 1997;Stevens & Rizzo, 2008). Hence, assessing both intra-and interspecific variability in spider mite susceptibility to infection by different potential biocontrol agents is a prerequisite for the development of efficient and long-lasting control strategies.
We then discuss the possible ecological and evolutionary causes and underlying mechanisms leading to the observed results, as well as their potential consequences for the evolution of both hosts susceptibility to infection and fungi virulence. Finally, we propose future directions to improve long-term success of spider mite control strategies using entomopathogenic fungi.

| Spider mite populations and rearing
Twelve populations of Tetranychid mites were used in this study: three of T. evansi (called BR, GH, and QL), three of T. ludeni (called OBI, Alval, and Assaf), three of the red form of T. urticae (called AlRo, AMP.tet, and FR.tet), and three of the green form of T. urticae (called TOM.rif,LS.tet,and B6JS). Most of these populations were collected in Portugal from 2013 to 2016; FR.tet was collected in France and AlRo in Spain in 2013. The population BR of T. evansi was collected in a greenhouse in Brazil in 2002 (Godinho, Janssen, Dias, Cruz, & Magalhães, 2016;Sarmento et al., 2011), and the population LS.tet of the green form of T. urticae derived from the London strain, which was used to sequence the species genome (Grbic et al., 2011). These populations originated from various plant species in the field, and none of them carried bacterial endosymbionts (i.e., Wolbachia, Cardinium, Rickettsia, Arsenophonus, Spiroplasma), either because they were initially uninfected when collected in the field (Zélé, Santos, Olivieri, et al., 2018a), or following antibiotic treatment (three generations with tetracycline hydrochloride, or one generation with rifampicin; all populations with ".tet" or ".rif" suffix, respectively; Breeuwer, 1997;Gotoh et al., 2005;Li, Floate, Fields, & Pang, 2014). All the information concerning these populations is summarized in (Table 1)

| Entomopathogenic fungi strains and preparation of inoculum
We used the strains V275 (=Met52, F52, BIPESCO 5) of M. brunneum and UPH-1103 of B. bassiana , obtained from Swansea University (UK) and from Siedlce University (Poland), respectively, as they were previously shown to have the potential to suppress T. urticae populations (Dogan et al., 2017). The procedures used for fungal growth, inoculum preparation, and spider mite infection are similar to that described in Dogan et al. (2017). Briefly, the two fungi were grown on Sabouraud Dextrose Agar (SDA) medium at 25°C for 2 weeks.
Conidia were harvested from sporulating cultures with the aid of a spatula, washed with sterile distilled water, and filtered through four layers of gauze to remove any hyphae.

| Spider mite infection and survival
The experiment was conducted in a growth chamber under standard conditions (25 ± 2°C, 80% RH, 16/8-hr L/D). Roughly 2 weeks prior to the experiment, age cohorts were created for each spider mite population by collecting ca. 100 females from each mass culture, allowing them to lay eggs during 4 days on detached bean leaves placed on water-soaked cotton. The offspring from these cohorts was used in the experiment.
One day prior to the onset of this experiment, 20 adult mated females with similar age were randomly collected from each cohort and placed on a 9-cm 2 bean leaf disk on top of wet cotton (to ensure the leaf remained hydrated) with the abaxial (underside) surface facing upwards. On the first day of the experiment, the surface of the leaf disks was sprayed using a hand sprayer with 2.5 ml of a spore Tween-20 at 1 × 10 7 conidia/ml, or, as control, with 0.03% aqueous Tween-20 only. Subsequently, female survival was monitored every 24 hr during 10 days by counting both dead and alive individuals.

| Statistical analysis
The analyses were carried out using the R statistical package (version 3.5.3). Survival data were analyzed using Cox proportional hazards mixed-effect models (coxme, kinship package). Spider mite species, or populations within each species, and infection treatment (sprayed with B. bassiana , with M. brunneum, or with Tween-20 only as control) were fit in as fixed explanatory variables, whereas disks nested within population, population (in the case of interspecific variation only), and block were fit as random explanatory variables. Hazard ratios (HR) were obtained from these models as an estimate of the difference between the rates of dying (i.e., the instantaneous rate of change in the log number of survivors per unit of time; Crawley, 2007) between the controls of each species/population (by changing the intercept of the model) and the B. bassiana or M. brunneum treatments.
Maximal models, including all higher-order interactions, were simplified to establish a minimal model by sequentially eliminating nonsignificant terms and interactions (Crawley, 2007). The significance of the explanatory variables was established using chi-squared tests (Bolker, 2008

| Interspecific variation of spider mite susceptibility to infection by Beauveria bassiana and Metarhizium brunneum
The statistical analyses revealed a significant interaction between treatments (females sprayed with either Tween-20 only as control, Moreover, while the two fungi induced similar mortality in T. evansi and in T. ludeni, infection with B. bassiana led to higher mortality than M. brunneum in the red form of T. urticae, while the reverse was found in the green form of T. urticae. Note, however, that survival in the T. evansi control was higher than in that of the three other species ( Figure 1d and Table 2).

| Intraspecific variation of spider mite susceptibility to infection by Beauveria bassiana and Metarhizium brunneum
We also found a statistically significant interaction between infection treatment and population on spider mite survival within each of the species studied (in the green form of T. urticae: X 2 4 = 79.60, p < .0001; Figure 2; in the red form T. urticae: X 2 4 = 12.12, p < .02; Figure 3; in T. ludeni: X 2 4 = 17.41, p < .002; Figure 4; in T. evansi: X 2 4 = 106.72, p < .0001; Figure 5). Indeed, although the two fungi induced a similar mortality in most populations within each species (e.g., the populations LS.tet, FR.tet, AlRo, OBI, Assaf, and all populations of T. evansi; see Table 3 Table 3d). Note, however, that QL control had a much lower survival than that of the two other populations (Figure 5d; Table 3d).

| D ISCUSS I ON
In this study, we found both intra-and interspecific variability in the  B. bassiana occurs naturally in more than 700 host species (Inglis, Goettel, Butt, & Strasser, 2001), and this range is likely underestimated as prevalence estimates are usually done in arthropod species that are crop pests or predators and parasitoids used as biocontrol agents (Meyling & Eilenberg, 2007). Moreover, differences in virulence between the two fungi shown here suggest population-specific Decreased host susceptibility to infection may be the result of two different (albeit nonexclusive) mechanisms (Boots et al., 2009;Read, Graham, & Raberg, 2008): resistance (i.e., reduction in parasite load) and/or tolerance (i.e., reduction of the damage incurred by a parasite). Differential host resistance to fungal infection might be due, for instance, to variability in different cuticular barriers. Such  and lipids) on the cuticle surface, but also the cuticle thickness, its degree of hardening by sclerotization, its resistance to enzymatic degradation, and its permeability (reviewed in Hajek & St. Leger, 1994).
Subsequently, when a fungus bypass cuticular barriers, variability in systemic immunity may also lead to differential host resistance responses. This may include differential activation of the Toll and JAK/ STAT pathways, which converge into the transcriptional activation of genes involved in phagocytosis, encapsulation, and humoral responses (e.g., Dong, Morton, Ramirez, Souza-Neto, & Dimopoulos, 2012 Indeed, whereas host resistance is predicted to select for increased parasite virulence (e.g., Gandon & Michalakis, 2000), host tolerance does not reduce parasite fitness and, therefore, will not lead to antagonistic counter-adaptation by pathogens (Raberg, Sim, & Read, 2007;Rausher, 2001). Still, depending on the nature of the tolerance mechanism, it may lead to the evolution of more virulent and transmissible parasites (Miller et al., 2006), with potentially serious implications for nontolerant populations (Boots et al., 2009), including nontarget species such as crop auxiliaries or spider mite predators. Finally, although increased mortality due to infection should lead to a reduction in oviposition duration, spider mites may evolve the ability to compensate infection-driven fitness costs by changing the timing of their reproductive efforts (i.e., "fecundity compensation"; Parker, Barribeau, Laughton, Roode, & Gerardo, 2011;Vezilier, Nicot, Gandon, & Rivero, 2015), thereby limiting the efficiency of fungi applications for population control. Hence, assessing which of these evolutionary outcomes is more likely is timely. In particular, it is likely that the level of intraspecific variation in susceptibility to infection found in our study is recapitulated within populations and is, at least partly, genetically determined. If this is the case, then this trait may evolve at a rapid pace.
In conclusion, our results show both intra-and interspecific variability in spider mite susceptibility to fungi-induced mortality using two generalist fungi, B. bassiana and M. brunneum. To our knowledge, this is the first study investigating the effect of entomopathogenic fungi on the survival of multiple spider mite populations belonging to different species within a single full factorial experiment. In line with laboratory virulence tests that are not necessarily well correlated with field effectiveness (Roberts & Leger, 2004), our results highlight the importance of studying several host populations/genomes when assessing the efficiency of a given biocontrol agent. These results also draw caution on Note: Hazard ratios of infection by each fungus were estimated relative to the control within each population. * p-value < .05, ** p-value < .01, *** p-value < .001.

TA B L E 3 (Continued)
the development of single strains as biocontrol agents, as hosts resistance to infection may evolve at a rapid pace.

ACK N OWLED G M ENTS
We thank Diogo Godinho and Miguel Cruz for their help in some parts of the experiment, as well as Marta Palma for technical support. We also thank all members of the SM laboratory for useful discussions and suggestions. This work was funded by an FCT-Tubitak agreement (FCT-TUBITAK/0001/2014 and TUBITAK TOVAG 115O610) to IC and SM, by the cE3c FCT Unit UID/BIA/00329/2020 to FZ and SM, and by Adnan Menderes University Research Foundation (ZRF-17055) to IC. FZ was funded through an FCT Post-Doc fellowship (SFRH/BPD/125020/2016). Funding agencies did not participate in the design or analysis of experiments.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
Full dataset has been deposited in the Dryad Data Repository (https ://doi.org/10.5061/dryad.gmsbc c2j4).