Spider mite resistant maize lines, B75 and B96, maintain resistance under water-stress

Climate variability has major implications for agriculture due to the increase in the frequency and intensity of simultaneous abiotic, namely water-stress, and biotic stresses to crops. Plant water-stress alone harms crops but also can attract outbreaks of herbivores with varied host specialization, and plants succumb to further yield losses dealing with multiple stressors. Host plant resistance provides a route to lessen yield losses from herbivory; however, our knowledge of the interactions between water-stress and pest resistance is limited, especially for mite herbivores of maize including the generalist two-spotted spider mite (Tetranychus urticae, TSM) and the specialist Banks grass mite (Oligonychus pratensis, BGM). We conducted parallel greenhouse and field experiments whereby a susceptible line (B73) and two TSM-resistant lines (B75 and B96) were subjected to either optimal irrigation or water-stress [50–60% and 5–10% volumetric water content (VWC), and 25–32% and 10–15% VWC, in the greenhouse and field, respectively] to test whether pest-resistant lines maintain their resistance when exposed to water-stress. We found that under optimal irrigation, TSM and BGM populations increased readily on B73, while B75 and B96 were largely resistant to the TSM but not BGM. While plant water-stress increased the susceptibility of B73 to both mite species, water-stress did not disrupt initial resistance levels of B75 and B96 maize for either mite species. Elevated protease activity was found in B75 and B96 and may contribute to maize resistance. Our findings that B75 and B96 are highly resistant to the TSM, and maintain resistance to both mite species with water-stress, highlight the importance of including the nuances of multiple stressors within the framework of host plant resistance.


Introduction
Climatic variability is predicted to constrain global agricultural production by increasing the frequency and severity of abiotic stresses, especially water limitation (Maxmen 2013;Oerke 2006). Water-stress as a result of suboptimal irrigation in agricultural systems affects diverse plant physiological responses (i.e., leaf temperature, stomatal conductance, and leaf water potential) and overall plant growth (Kramer 1983;Ruckert et al. 2018). The effects of water-stress on plant physiology are well characterized (Fritter and Hay 2012). Apart from the physiological impacts of drought on crops and natural plant communities, abiotic stresses associated 1 3 with drought, including high temperatures, can exacerbate the impact of many arthropod herbivores (Maxmen 2013). For example, agriculturally damaging outbreaks of spider mites, which are chelicerate pests of diverse dicot and monocot crops, have resulted in yield losses as high as ~ 47% in drought-stressed maize (Zea mays ssp. mays) (English-Loeb 1990;Maxmen 2013). While the management of arthropod pests in agriculture has relied heavily on pesticides, the evolution of pesticide resistance has been and remains problematic. The characterization of naturally occurring variation in host resistance provides another route to lessen yield losses from herbivory (Bynum et al. 2004). Nevertheless, for many plant species or varieties, our knowledge of the interactions between abiotic factors such as water-stress and pest resistance is limited or nearly non-existent, especially for mite herbivores of grasses (family Poaceae).
Previous studies of aphids, caterpillars, and flies on various crops have reported idiosyncratic effects of waterstress on plant resistance to herbivory (Grinnan et al. 2013;Sharma et al. 1999;Verdugo et al. 2016Verdugo et al. , 2015. For example, an evaluation of 29 studies focused on aphids across several cropping systems found that resistance in crops exposed to water-stress was either decreased (41.4% of studies), increased (34.5%), showed no change (20.1%), or had conditional effects (3.4%) (Verdugo et al. 2016). A holistic understanding of the impacts of water-stress on plant defenses against pests is further complicated by the fact that plants activate multiple constitutive or induced defense pathways to deter herbivores, and these may themselves be impacted to different degrees by abiotic stresses. Induction of plant defense pathways leads to the activation of many conserved plant defense proteins including chitinase (CHI) and peroxidase (POD) that tend to associate with salicylic acid signaling, and polyphenol oxidase (PPO) and trypsin inhibitor (TI) which associate with the jasmonic acid pathway (Cipollini et al. 2004). CHI may degrade the peritrophic membrane of arthropods, and TI can slow their amino acid acquisition and digestion (Arnaiz et al 2018;Fürstenberg-Hägg et al. 2013). Moreover, plants have decreased nutritional value for arthropods with elevated PPO and POD levels (Mander and Liu 2010). The connection with abiotic stress adds another level of complexity considering that POD activity, for instance, is increased by drought stress alone .
In this context, herbivore host range may also be important. For instance, generalist herbivores, which are defined as having hosts in many plant families, may rely on broad detoxification of plant defenses they encounter across multiple hosts (Ali and Agrawal 2012; Grbić et al. 2011). Specialist herbivores, on the other hand, have a narrow set of plant hosts on which they persist and have often evolved ways to counter the plant defenses of their specific hosts (Ali and Agrawal 2012; Glas et al. 2014). Moreover, specialists and generalists may elicit differential plant responses (Ali and Agrawal 2012). This was the case for generalist and specialist spider mite species where the generalist increased CHI and TI activity on water-stressed plants, an effect not observed by the specialist mite (Gill et al. 2020). Therefore, even for generalist and specialist herbivores in the same feeding guild, the combinatorial effect of herbivory and abiotic stress may impact herbivore species differently, and hence lead to differences in plant fitness. These interactions can have major implication for the development of host plant resistance across multiple herbivore species.
Maize is one of the most important cereal crops and is damaged by spider mite herbivores in outbreaks that are typically associated with drought conditions (plant waterstress and high temperatures) (Archer and Bynum 1993;English-Loeb 1990). Although a number of spider mite species feed on maize, major economic damage is typically associated with the generalist two-spotted spider mite (Tetranychus urticae, TSM) that feeds on a wide range of plant species (> 1100), as well as the specialist Banks grass mite (Oligonychus pratensis, BGM) that is found predominantly on species in the Poaceae family (Archer and Bynum 1993). In hot and dry conditions, and especially on more mature plants (post-seedling stages), spider mite populations can rapidly increase within weeks to cause severe yield losses (Archer and Bynum 1993;Bynum et al. 2004). Further, both species, but especially the TSM, have extremely high rates of the evolution of pesticide resistance, limiting control options during outbreaks (Van Leeuwen et al. 2010).
The challenges of controlling spider mites on waterstressed maize have generated substantial interest in the identification of resistant maize varieties (Bui et al. 2021;Gill et al. 2020;Ruckert et al. 2021). In a number of studies, variation in resistance among maize inbred lines has been reported, especially to the TSM (Archer 1987; Bui et al. 2021;Bynum et al. 2004;Kamali et al. 1989;Tadmor et al. 1999). For example, the maize inbred line B96 (formerly called 41.2504B) has been shown in multiple studies to be highly resistant to the TSM and the carmine spider mite Tetranychus cinnabarinus (Bui et al. 2021;Kamali et al. 1989;Tadmor et al. 1999), and the latter is thought to be a color morph of the TSM. Variation in resistance to BGM has also been reported (Owens et al. 1976). Recently, we systematically screened 38 maize lines at a post-seedling stage (the 8 th leaf) for resistance to both TSM and BGM by antibiosis (that is, the reduction in pest reproductive rates) (Bui et al. 2021). We replicated the finding that B96 is exceptionally resistant to the TSM and identified two additional lines with resistance (B49 and B75). In contrast, most of the remaining maize lines were TSM susceptible, including B73, an important maize line used for the development of elite maize varieties (Romay et al. 2013), and none of the 38 lines were even moderately resistant to the BGM.
Nonetheless, the screen of the 38 lines was performed under optimal irrigation (Bui et al. 2021), so it remains unknown whether these lines maintain resistance to the TSM and are comparably sensitive to BGM under water-stress. However, given that both of these species reach economic damage thresholds for field-grown maize and that susceptible B73 is differentially modulated (Gill et al. 2020), additional research within this system is critically important. In the current study, we performed replicate greenhouse and field trials to assess how water-stress and herbivory by both the TSM and BGM interact to impact mite resistance levels and plant defenses of the B96 and B75 (TSM-resistant) maize lines as compared to B73 (TSM susceptible).

Greenhouse TRIAL 1 and TRIAL 2
We conducted a 2 × 3 × 2 factorial design experiment using two levels of water (optimal irrigation and water-stress), three levels of mite (Control (no mite), TSM, and BGM) and two levels of maize resistance (TRIAL 1: susceptible B73 and resistant B75; TRIAL 2: susceptible B73 and resistant B96) with a time course (1, 3, and 7-days) in the greenhouse. These experiments were performed at Utah State University's Research Greenhouse in Logan, UT, USA.
Experimental units consisted of 144 buckets of 18.9 L each filled with potting soil (Sunshine Mix #3, Sun Gro Horticulture, MA, USA) arranged in a complete randomized design. We planted two maize plants per bucket representing a single line (TRIAL 1: B73 or B75; TRIAL 2: B73 or B96), where each bucket was a replicate and each plant was a subsample. Each bucket received a single treatment (water × mite × resistance) which was replicated twelve times. Maize plants were grown in greenhouse-controlled conditions (25 ± 2 °C, 60 ± 5% RH, 16:8 h (L:D) photoperiod) and fertigated at a rate of 4.8 kg/100L of 21 N-5P-20 K using Peters Excel Water Soluble Fertilizer mixture (ICL Specialty Fertilizers, SC, USA). We used drip tape (DIG Corporation, CA, USA, 12.7 mm and 6.35 mm diameter tubing with 3.8 L/hr compensating emitters) to irrigate the plants throughout the experiment.
After six weeks, plants were switched from fertigation to irrigation to establish optimal irrigation or water-stress levels. Acclima 315 L soil sensors (Acclima, ID, USA) were used to monitor the volumetric water content (VWC) of soil. All replicates were irrigated evenly at field capacity of potting soil (50-60% VWC) for seven weeks after sowing. When plants were 8 weeks old, buckets were randomly assigned to either optimal irrigation (maintaining field capacity) or water-stress (reducing irrigation to 5-10% VWC) above permanent wilting point (Suppl Fig. S1). Water-stress or optimal irrigation levels were quantified by measuring stomatal conductance (mmolm −2 s −1 ) and leaf temperature (°C) using a leaf porometer (Model SC-1, Meter Group, WA, USA), leaf water potential (bar) using a pressure chamber (Model 615, PMS Instrument Company, OR, USA), and stem height (cm) by using a ruler. Leaf temperature, stomatal conductance and stem height were measured at 3 and 7-days post-mite introduction, while leaf water potential was measured after sample collection at 7-days post-mite introduction (Suppl Table 1 and 2). At an 8-week stage, Tanglefoot (Scotts Miracle-Gro Company, OH, USA) non-phytotoxic wax arenas were created on the 8 th leaves of plants after Bui et al. (2018). Within arenas, twenty adult female mites (mated BGM or TSM) from laboratory colonies reared on B73 maize (28 ± 2 °C, 50 ± 5% RH, 16:8 h (L:D) photoperiod) were introduced. Mite transfer to arenas was accomplished by vacuuming mites into filter pipette tips that were then anchored within leaf arenas following Bui et al. (2021). Mites exited from the pipette tip within approximately an hour to settle on the undersides of the leaves.
At 1, 3, and 7-days post-mite introduction, leaf samples (leaf areas inside the Tanglefoot arenas) from eight plants of four randomly selected replicates (2 plants/replicate) were collected, flash-frozen using liquid nitrogen and stored in a freezer (−20 °C) until processing. Each sample was processed by counting the number of eggs and all mite stages and by performing defense protein bioassays after leaf samples were frozen (see 2.2. Plant Defense Bioassay section).

Field SEASON 1 and SEASON 2
In the 2018 field season (SEASON 1), we conducted a 2 × 3 × 3 factorial design experiment using two levels of water (optimal irrigation and water-stress), three levels of mite (Control, TSM, and BGM) and three levels of maize resistance (susceptible B73, and resistant B75 and B96) with a time course (1, 3, and 7-days). In 2019 (SEASON 2), the experimental design was similar with the exception that there were only two levels of mite (Control and TSM). In SEASON 2, BGM was not evaluated following the analysis of SEASON 1 because it was less sensitive to maize resistance (see 3.2 Results section). These experiments were conducted at the Greenville Research Station at Utah State University, Logan, UT, USA.
Six plants representing each respective maize inbred line (B73, B75, and B96) were grown in a Lumite cage (1.8 m L × 1.8 m W × 1.8 m H) (Lumite, GA, USA), the experimental unit. Cages were arranged in a randomized complete block design (RCBD) within the varied water treatments. Each cage received a single treatment (water × mite × resistance) which was replicated four times in SEASON 1. A total of 72 cages in SEASON 1 were arranged in 12 rows (6 cages/row), with each row spaced 4 m apart to establish distinct irrigation treatments with adjacent rows and cages within a row (i.e., with the same irrigation level) spaced 2 m apart. In SEASON 2, there were a total of 60 cages with each treatment replicated five times and BGM not evaluated.
When the plants were 6 weeks old, water treatments were established by maintaining half the number of cages at 25-32% and the other half at 10-15% VWC (Suppl Fig. 1). Similar to the greenhouse, stomatal conductance, leaf temperature and stem height were measured at 3-and 7-days post-mite introduction, while leaf water potential was measured at 7-days post-mite introduction, to assess water-stress (Suppl Table 1 and 2).
Mirroring the greenhouse protocol, the 8th leaves of 8-week old plants were selected for the establishment of Tanglefoot arenas. Twenty adult female mites, BGM or TSM (recall that BGM was not included in SEASON 2), were introduced to the leaf arenas using the pipette tip method previously described. After 1, 3, and 7-days post-mite introduction, leaf samples (leaf area inside the Tanglefoot arena) from two plants per replicate were collected for each of the three time periods, stored and processed as described for the greenhouse trials.

Plant defense bioassays
The activity of four plant defense proteins-polyphenol oxidase (PPO), trypsin inhibitor (TI), peroxidase (POD), and chitinase (CHI)-was assayed following methods from Gill et al. (2020). Briefly, each leaf sample (500 mg) was pulverized in liquid nitrogen and suspended in 1 mL of 0.05 M sodium phosphate buffer. Following centrifuging at 12,000 RPM for 12 min, the cell lysate (the supernatant containing soluble proteins) was reserved for enzymatic assays. The activities of POD, PPO, and CHI were analyzed using a microplate reader (Biotek, EPOCH, VT, USA), while the activity of TI was analyzed by using radial diffusion techniques. PPO and POD were quantified as ΔAbs 470nm min −1 mg extract protein −1 , CHI was quantified as ΔAbs 405nm mg extract protein −1 , and TI was quantified as µg TI mg extract protein −1 .

Statistical analysis
All statistical analyses were performed using SAS 9.4 M4 University Edition. Mite (TSM and BGM) population sizes, assessed as the sum of eggs, nymphs and adults, and defense protein activity measurements from greenhouse trials were analyzed using a general linear model (Proc Glimmix). Here, analyses for mite populations consisted of two levels each of water (optimal irrigation and water-stress), mites (TSM and BGM) and maize resistance (TRIAL 1: B73 and B75; TRIAL 2: B73 and B96), with time course (1, 3, and 7-days post-mite introduction). However, for defense protein assays, analyses consisted of two levels of water, three levels of mite (Control, TSM and BGM), and two levels of maize resistance. The assays and analysis of defensive proteins, which can respond rapidly to herbivory, were only performed at 1 day post-mite introduction following the results of our previous study (Gill et al. 2020). Square-root transformation was used for both mite population growth and defensive protein activities (POD, PPO, CHI and TI) data to conform to the assumption of normality and heteroscedasticity.
When interactions were not significant, differences within significant main effects were determined using Tukey's HSD post hoc test. When multi-factor interactions were significant, the LSMESTIMATE statement (Proc Glimmix) with Tukey-Kramer adjustment was used for further analysis. For instance, when defense protein activity revealed a three-way interaction (water × mite × resistance), each mite species was independently analyzed comparing resistance in maize inbred lines and water treatments.
SEASON 1 analyzes for mite (TSM and BGM) population sizes and defense protein activity used a general linear model as described with greenhouse trials (Proc Glimmix). Here, three levels of maize resistance (B73, B75 and, B96) along with each respective level of water, mite and time course as described for greenhouse trials were analyzed. Data transformations (i.e., square-root), as well as additional analyses for interactions (i.e., LSMESTIMATE with Tukey-Kramer), were performed following the greenhouse trials.
For SEASON 2, recall the BGM mite treatment was not included. Therefore, analyses consisted of only TSM, three levels of maize resistance, and two levels of water with time course as described for SEASON 1. Additionally, defense protein assays for 1 day post-mite introduction were analyzed instead with two levels of mite (Control and TSM), and the same levels described for each factor (i.e., resistance and water). Data transformations as well as further analyses of significant interactions were performed as described previously.

Effect of water-stress on plant physiological measures
In greenhouse TRIAL 1, plant physiological measurements including leaf water potential and leaf temperature significantly increased due to water-stress, and these effects were similar between the maize lines (susceptible B73 and resistant line B75) (Suppl Table 1). Specifically, waterstressed plants increased leaf water potential and leaf temperature by 3.87 ± 0.85 bar and 1.32 ± 0.34 °C compared to optimally irrigated plants (Suppl Table 2). Water-stress also significantly reduced stomatal conductance and stem height of plants by as much as 59.17 ± 8.38 mmolm −2 s −1 and 17.22 ± 2.81 cm compared to optimally irrigated plants (Suppl Table 2).
TRIAL 2 plants (susceptible B73 and resistant line B96) exposed to water-stress conditions similarly increased leaf water potential by 3.87 ± 0.85 bar and decreased stomatal conductance and stem height by 128.28 ± 18.11 mmolm −2 s −1 and 17.07 ± 1.51 cm compared to optimally irrigated plants (Suppl Table 2).

Effect of water-stress on mite population growth
For TRIAL 1, mite population growth was significantly impacted by the interactions of water × resistance × time (P < 0.01, Table 1, Fig. 1a-b) and mite × resistance × time (P = 0.05, Table 1, Fig. 1a-b). The water × resistance × time interaction appeared to be driven by mite populations remaining the same on resistant plants (B75) exposed to water-stress over the 7-day period compared to optimally irrigated plants (P < 0.01, Table 1, Fig. 1a-b). To assess the interaction in more detail, we analyzed mites on individual maize inbred lines by comparing water treatments in time. We found that water-stress susceptible plants (B73) had increased mites (P < 0.05, LSMESTIMATE Tukey adjustment), while resistant plants (B75) exposed to waterstress did not differ in mite populations (P > 0.05) compared to optimally irrigated plants. Further, a significant mite × resistance × time interaction appeared to be driven by specialist BGM not being impacted by maize resistance over the 7-day period compared to TSM. To further understand this interaction, we analyzed population growth for TSM and BGM independently, comparing maize inbred lines, at each time point. Here, TSM was decreased on resistant B75 plants compared to susceptible B73 plants at 7-days (P < 0.05). However, BGM grew equally well on both B73 and B75 throughout the experiment (P > 0.05, Fig. 1b).
For TRIAL 2, mite population growth was also significantly affected by the interactions of water × resistance × time (P < 0.01, Table 1, Fig. 1c-d) and mite × resistance × time (TRIAL 2: P = 0.01, Table 1, Fig. 1c-d). Similar to TRIAL 1, the water × resistance × time interaction was driven by mites remaining low on resistant B96 plants exposed to waterstress after 7-days compared to control plants. As in TRIAL 1, we evaluated each maize line (B73 and B96) to compare water treatments across time. Again, water-stress increased mites in B73 at 7-days (P < 0.05, LSMESTIMATE). However, resistance for B96 was not affected by water-stress as mites remained low over 7-days (P > 0.05). Furthermore, as in TRIAL 1, the mite × resistance × time interaction appeared to be driven by an unresponsiveness of BGM to B96, while this resistant maize line appeared to decrease TSM throughout the experiment. To confirm, each mite was analyzed independently, to compare B73 and B96 across time. At 3 Table 1 ANOVA results of the effect of water treatments (optimal irrigation and water-stress) on population growth of mites (TSM and BGM) on B73 and B75 maize inbred lines at time ( Fig. 1d).

Effect of water-stress and mite herbivory on the activity of plant defense proteins in each greenhouse TRIAL
In TRIAL 1, POD (peroxidase) was significantly affected by water × mite (P < 0.01) and mite × resistance interactions (P < 0.04, Table 2). To understand the interactions, we independently analyzed plant responses to mite conditions comparing water treatments. Here, only BGM combined with water-stress resulted in a 0.53-fold (from 50.90 ± 6.26 to 27.31 ± 6.05 ∆Abs 470nm min −1 mg extract protein −1 , Fig. 2a and b) decrease in POD compared to optimally irrigated plants (P < 0.05, LSMESTIMATE). We analyzed the mite × resistance interaction by independently analyzing plant responses to mite conditions, comparing maize lines. B75 plants (Fig. 2b) had reduced POD activities compared to B73 plants when exposed to TSM (P < 0.05; Fig. 2a and b), an effect not seen with BGM or control. With the water × mite × resistance interaction for PPO (polyphenol oxidase; P = 0.01, Table 2), mite treatment was analyzed independently, to compare water treatments and maize lines. PPO decreased from 3.95 ± 0.54 to 1.5 ± 0.1 ∆Abs 470nm min −1 mg extract protein −1 in control B73 plants exposed to water-stress compared to optimally irrigated plants (P < 0.05, LSMESTIMATE, Fig. 2c). In addition, TSM combined with water-stress in susceptible B73 increased PPO from 0.27 ± 0.09 to 3.0 ± 0.44 ∆Abs 470nm min −1 mg extract protein −1 compared to optimally irrigated plants with TSM (P < 0.05, Fig. 3c). The water × mite × resistance interaction for CHI (chitinase) was significant (P < 0.01, Table 2). Water-stress in control B73 plants decreased CHI from 39.64 ± 7.71 to 14.35 ± 1.78 ∆Abs 405nm mg extract protein −1 (P < 0.05, LSMESTIMATE, Fig. 2e). In addition, the combination of water-stress and BGM decreased CHI from 100.68 ± 24.58 to 32.93 ± 8.77 ∆Abs 405nm mg extract protein −1 in B73 plants compared to plants exposed to BGM alone (no water-stress) (P < 0.05, Fig. 2e). TI (trypsin inhibitor) also had a significant interaction of water × mite × resistance (P = 0.02, Table 2) and mites were analyzed independently.

Effect of water-stress on plant physiological measures
In SEASON 1, a significant interaction of water × resistance (P = 0.017, Suppl Table 1) revealed that leaf water potential was significantly lower for optimally irrigated resistant B96 plants (2.57 ± 0.25 bar) compared to optimally irrigated susceptible B73 plants (3.33 ± 0.37 bar) (P < 0.05, LSMESTI-MATE Tukey adjustment). Also, water-stress increased leaf water potential in susceptible B73 as well as in resistant B75 and B96 by 3.8 ± 0.45 bar, 2.93 ± 0.46 bar and 5.62 ± 0.47, respectively (P < 0.05, LSMESTIMATE Tukey adjustment). For stomatal conductance, a significant interaction between resistance × time (P = 0.002, Suppl Table 1) revealed that resistant B75 and resistant B96 plants had 0.66-fold and 0.54-fold lower stomatal conductance compared to susceptible B73 plants, respectively (P < 0.01). The significant main effect of water-stress in plants also reduced stomatal conductance by 68.29 ± 16.85 mmolm −2 s −1 , increased leaf temperature by 0.9 ± 0.25 °C, and reduced stem height by 11.09 ± 2.24 cm compared to optimally irrigated plants (water: P < 0.01, Suppl Table 2). Also, the main effect of resistance was significant for stem height, which showed that susceptible B73 (82.41 ± 2.43 cm) and resistant B75 (82.60 ± 1.97 cm) plants had higher stem height compared to resistant B96 (52.38 ± 2 cm) in the field (resistance: P < 0.01, Suppl Table 2). Similar to SEASON 1, a significant effect of waterstress in SEASON 2 also increased leaf water potential and leaf temperature by 8.48 ± 0.5 bar and 1.55 ± 0.39 °C, respectively, compared to optimally irrigated plants (Suppl Table 2). In water-stressed plants, stem height and stomatal conductance were significantly decreased by 14.66 ± 2.96 cm and 146.82 ± 24.17 mmolm −2 s −1 compared to optimally irrigated plants (Suppl Table 2). Optimal irrigation and waterstress are indicated by white and gray bar, respectively. Statistical analyses are presented in Table 2. Here, interaction terms were significant; the LSMES-TIMATE statement (Proc Glimmix) with Tukey-Kramer adjustment was used for further analyses of specific contrasts (P < 0.05) as indicated by horizontal lines. α represents significant differences between control plant responses to water treatments c,e. β and γ represent significant difference between plant responses for water treatment and exposure to BGM e and TSM c,g, respectively Fig. 3 Effect of water treatment and mite herbivory on the activity of B73 and B96 plant defense proteins (left and right panels, respectively) for POD a-b, PPO c-d, CHI e-f, and TI g-h in greenhouse TRIAL 2. Treatments include BGM and TSM for mites, with un-infested control plants, CNT (means ± SEs). Optimal irrigation and water-stress are indicated by white and gray bar, respectively. Statistical analyses are presented in Table 2. Here, interaction terms were significant; the LSMESTIMATE statement (Proc Glimmix) with Tukey-Kramer adjustment was used for further analyses of specific contrasts (P < 0.05) as indicated by horizontal lines. α represents significant differences between control plant responses to water treatments (h). γ represents significant difference between plant responses for water treatment and exposure to TSM g,h

Effect of water-stress on mite population growth
In SEASON 1, mite × resistance × time and water × time interactions significantly affected mite population growth (P < 0.01, Table 2, Fig. 1e-f). The mite × resistance × time interaction appeared to be driven more by lower TSM throughout the experiment than by BGM, which were significantly lower only at 7-days on resistant plants compared to susceptible B73 plants. This was confirmed by evaluating each mite species to compare maize inbred lines across time. The resistant B75 and B96 plants had lower TSM compared to susceptible B73 plants throughout the experiment (P < 0.05, LSMESTIMATE). Lower BGM were found on resistant B75 (at 1 and 7-days) and B96 (at 7-days) compared to susceptible B73 plants (Table 1, Fig. 1f). Interestingly, resistant B96 plants also had lower BGM compared to resistant B75 at 3 days (P < 0.05). Further, a water × time interaction revealed that water-stress increased both mite populations in all maize inbred lines at 7-days (P < 0.05).
In SEASON 2, for which only TSM was investigated, a significant resistance × time interaction (P < 0.01) revealed lower TSM on resistant B75 (at 7-days) and resistant B96 (at 3 and 7-days) compared to susceptible B73 plants (Table 1, Fig. 1g). Also, resistant B96 plants had lower TSM than resistant B75 Fig. 4 Effect of water treatment and mite herbivory on the activity of B73, B75, and B96 plant defense proteins (left, middle, and right panels, respectively) for POD a-c, PPO d-f, CHI g-i, and TI j-l in field SEASON 1. Treatments include BGM and TSM for mites, with un-infested control plants, CNT (means ± SEs). Optimal irrigation and water-stress are indicated by white and gray bar, respectively. Statistical analyses are presented in Table 2 plants at 7-days (P < 0.05). With the exception of B73, mites on resistant lines B75 and B96 were not impacted by water-stress (Fig. 1g).
TI was significantly impacted by a water × mite interaction (P < 0.01; Table 2). The interaction was driven by an increase in TI due to TSM and an opposite effect due to BGM on water-stressed plants. Across maize lines, plants exposed to a combination of TSM and water-stress increased TI from 0.85 ± 0.21 to 3.87 ± 1.5 µg TI mg extract protein −1 compared to TSM alone at 1 day (P < 0.05) (Fig. 4j-l). In contrast, plants exposed to a combination of BGM and water-stress decreased TI from 4.56 ± 2.27 to 1.76 ± 0.4 µg TI mg extract protein −1 compared to BGM alone at 7-days (P < 0.05). In addition, the significant main effect of resistance showed that B96 had 3.91-fold and 4.49-fold higher TI compared to B73 (1.48 ± 0.49 µg TI mg extract protein −1 ) and B75 (1.28 ± 0.18 µg TI mg extract protein −1 ), respectively (P < 0.01).

Discussion
Consistent with earlier findings (Gill et al. 2020), TSM increased readily on B73 under optimal irrigation, and at 7-days, mites increased even further on water-stressed plants. In contrast, across all greenhouse and field trials, and under both water conditions, B96 and B75 were largely immune to supporting TSM. While the responses for B96 and B75 were similar, in SEASON 1, and more so in SEA-SON 2, TSM did tend to be slightly elevated on B75 as compared to B96 under both water conditions, although this Effect of water treatment and mite herbivory on the activity of B73, B75, and B96 plant defense proteins (left, middle, and right panels, respectively) for CHI a-c and TI d-f in field SEASON 2. Treatments include TSM for mites, with un-infested control plants, CNT, (means ± SEs). Optimal irrigation and water-stress are indicated by white and gray bars, respectively. Statistical analyses are presented in Table 2 finding requires additional studies. Nevertheless, this observation is consistent with Bui et al. (2021) who identified two large-effect quantitative trait loci (QTL) on chromosomes 1 and 6 for TSM resistance in B96, but only one large-effect resistance QTL on chromosome 6 in B75 (the chromosome 6 QTL in B75 likely originated from B96, which is in the pedigree for B75) (Bui et al. 2021;Portwood et al. 2019). It is important to note that in greenhouse and field trials that included TSM-resistant B96 and B75, as well as TSMsusceptible B73, we confirmed that water-stress was induced in our experimental design as revealed by expected changes in plant leaf water potential, stomatal conductance, leaf temperature, and plant stem height (Suppl Table 1 and 2) (Isoda and Shahenshah 2010;Ruckert et al. 2018).
Although the selection of B96 and B75 for the current study was motivated by these lines' high resistance to the TSM, we also assessed how plant water-stress impacted resistance of B96 and B75 to the grass specialist BGM. The BGM is not as broadly distributed as the TSM, but is found on multiple continents (Migeon et al. 2011). As observed for the TSM, BGM outbreaks and agricultural damage on maize are typically associated with drought conditions ). If TSM resistance under water-stress in B96 and B75 was to be counter-associated with heightened susceptibility to BGM, the use of B96 or B75 in breeding efforts to enhance spider mite resistance would be confounded. Mirroring our findings for the TSM, and consistent with earlier studies of BGM on several maize varieties (Gill et al. 2020;Ruckert et al. 2018), plant water-stress led to increases in BGM as compared to optimal watering on B73 in both the greenhouse and field. Given that Bui et al. (2021) found that B73, B96, and B75 were all susceptible to BGM when optimally watered, it might have been expected that under waterstress, the responses for B96 and B75 would have mirrored that of B73 (i.e., mites would have increased more quickly). However, with the possible exception of B75 in SEASON 1, this pattern was not apparent, as plant water-stress did not enhance BGM as compared to optimal watering. Also, in several instances (greenhouse TRIAL 2 and field SEASON 2), and especially for B96, a trend for higher BGM resistance as compared to B73 was observed under both water conditions. Additional work is desirable to confirm this trend, and it should be noted that the relative effect (if true) was minor in magnitude as compared to the respective highly significant difference observed for the TSM.
Our studies conducted here add to a growing literature that has revealed that abiotic stresses can have varied species-or varietal-specific effects on resistance to arthropod herbivores. For instance, as reviewed by Verdugo et al. (2016), abiotic resistance of aphids to several abiotic stresses was either ameliorated, enhanced, or unchanged in a variety of plant systems. Although less is known about how abiotic stresses impact plant resistance to mites as compared to insects, the impact of plant water-stress on spider mite resistance has been investigated in several crops. In common bean, water-stress was reported to have nonlinear effects on TSM resistance (English-Loeb 1990), and for a barley cultivar and in tomato, TSM performance on water-stressed plants was elevated compared to well-watered ones (Santamaria et al. 2018;Ximénez-Embún et al. 2017). Moreover, a study of four drought-adapted tomato lines found that when tomato plants were subjected to moderate drought stress, the performance of Tetranychus evansi, a sister species of the TSM, was enhanced for two lines but not for two others (Ximénez-Embún et al. 2018). Such variation, including at the intra-specific level, was reflected in our study, as depending on the inbred line, plant water-stress decreased resistance (B73: TSM and BGM), or alternatively resistance was (largely) not changed (B96 and B75: TSM and BGM). Interestingly, five maize genotypes resistant to the European corn borer (Ostrinia nubilalis) retained their resistance levels across eleven different environments (Willmot et al. 2009). Although these environments did not highlight water-stress independently, they suggest that some insectresistant maize lines can maintain resistance in the face of substantial abiotic variation. In the context of this study, it should be noted that both B96 and B75 are highly resistant to O. nubilalis (Kamali et al. 1989;Portwood et al. 2019), with Kamali et al. (1989) noting that B96 is nearly "immune" to herbivory by this major lepidopteran pest. Whether the stability of high-level resistance to the TSM under water-stress also translates to resistance to the European corn borer in B96 and B75 is not known, but warrants future investigation given the potential value of lines that exhibit resistance to multiple arthropods.
We also investigated how water-stress impacts several plant defense protein activities including POD, PPO, CHI and TI that can directly and negatively affect herbivores (Arnaiz et al. 2018;Cipollini et al. 2004;English-Loeb et al. 1997;Fürstenberg-Hägg et al. 2013;Thipyapong et al. 2004). These activities also serve as readouts for the broader activation of phytohormone-mediated pathways for defense against biotic challenges (i.e., jasmonic acid and salicylic acid signaling), although they may also respond to other stimuli. For instance, POD and PPO are broadly conserved proteins that can be induced by herbivory, mechanical wounding, and water-stress (Cao et al. 2015;Minibayeva et al. 2015;Schnable et al. 2009;Suzuki et al. 2012). In our study, POD and PPO activities were impacted by herbivory, water-stress, or combinations of both, albeit somewhat sporadically by maize genotype and trial (given this variation, POD and PPO were not evaluated in SEASON 2). Similar patterns were observed for CHI, which can degrade exoskeletons or peritrophic membranes (at least in insects), and TI, a protease inhibitor that can disrupt herbivores' acquisition of amino acids (Arnaiz et al. 2018;Cipollini et al. 2004;Fürstenberg-Hägg et al. 2013). It was notable that for several trials (i.e., greenhouse TRIALS 1 and 2), B73 plants exposed to combinations of water-stress and BGM herbivory resulted in either decreased TI activity or no change, a pattern that contrasted to that observed for TSM herbivory. This suggests some specificity in the response of maize to the generalist TSM versus the specialist BGM and confirms the prior report from Gill et al. (2020) that water-stressed maize (B73) plants had elevated TI activities when exposed to TSM, but not BGM, herbivory.
Despite our extensive quantification of defensive protein activities among genotypes, clear associations to the large differences in resistance that we observed among lines were often either not obvious or were tenuous. For instance, while TI activity was higher in resistant B96 versus susceptible B73 in the field (SEASONS 1 and 2), TI activity levels for B75 more closely resembled B73, but both B96 and B75 are highly resistant to TSM herbivory compared to B73 (Bui et al. 2021 and this study). One interpretation of our findings is that, although the defensive activities we assayed may modulate the observed variation in resistance levels, other maize defenses against herbivores may have more important roles. For instance, B96 and B75 have both been shown to have high constitutive levels of the specialized benzoxazinoid compound DIMBOA at post-seedling stages (Barry et al. 1994;Bing et al. 1990), while B73 does not (Zheng et al. 2015). Benzoxazinoid compounds are known to confer resistance to many lepidopteran species, including the European corn borer (Wouters et al. 2016). Further, on mutant maize seedlings that lack benzoxazinoids, TSM populations expanded more quickly than on wildtype plants, although BGM population growth was high on both (Bui et al. 2018).
The chromosome 6 QTL interval for TSM resistance shared between B96 and B75 does not encode known benzoxazinoid biosynthetic enzymes, although an impact on benzoxazinoid levels by an unknown mechanism cannot be excluded (Bui et al. 2021). Regardless, these observations, coupled with our current work, highlight the importance of genetic dissection (e.g., by fine mapping) of the large-effect TSM resistance QTLs identified in B96 and B75 in order to determine specific resistance loci, alleles, and attendant molecular mechanisms. Our observation is that B96 and B75 remain resistant to the TSM, and that BGM resistance is not lessened, when plants are exposed to water-stress provides compelling motivation to pursue such studies. Indeed, continued efforts to address climate variability impacts on pest management are needed, especially for a specialist like BGM that outbreaks under such conditions and appears to evade maize resistance. Recent work highlights the complexity of host plant resistance as the development of drought-tolerant corn hybrids to alleviate impacts of drought appear to also lessen herbivory of BGM when the crop has limited irrigation (Ruckert et al. 2021). These findings and those in this current work expand the importance of measuring the effect of drought on the resistance of the crop to a pest and the possibility for discovering new pest management approaches in the face of climate change.