Plant hormone manipulation impacts salt spray tolerance, which preempts herbivory as a driver of local adaptation in the yellow monkeyflower, Mimulus guttatus

We would like to thank Ben Blackman and Jack Collichio for germinating 27 seeds at UC Berkeley. This work would not have been possible without the support of staff at the 28 Bodega Marine Lab and Reserve, particularly Jackie Sones, Luis Morales, and Al Carranza, who 29 facilitated field work and use of the greenhouse and lab. Michael Gillogly and Michelle Halbur 30 of Pepperwood Preserve were instrumental to this research through their support of our work at 31 the Preserve, including performing emergency plot-monitoring when the preserve temporarily 32 closed to visitors due to the COVID-19 pandemic. Isabella Johnson and Billie Fraser assisted 33 with chemical sample processing. This work was funded by National Science Foundation 34 Division of Integrative Organismal Systems Grants to DBL and LMH (IOS-1855927) and DBL 35 (IOS-2153100). 36 37 Conflict of Interest statement: We declare no conflict of interest. 38 39 Data Availability Statement: Data and code will be available on Data Dryad upon publication. 40 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint

41 A major challenge in evolutionary biology is identifying the selective agents and phenotypes 42 underlying local adaptation. Local adaptation along environmental gradients may be driven by 43 trade-offs in allocation to reproduction, growth, and herbivore resistance. To identify 44 environmental agents of selection and their phenotypic targets, we performed a manipulative 45 field reciprocal transplant experiment with coastal perennial and inland annual ecotypes of the 46 common yellow monkeyflower (Mimulus guttatus). We manipulated herbivory with exclosures 47 built in the field and exogenously manipulated hormones to shift allocation of plant resources 48 among growth, reproduction, and herbivore resistance. Our hormone treatments influenced 49 allocation to reproduction and phytochemical defense, but this shift was small relative to ecotype 50 differences in allocation. Herbivore exclosures reduced herbivory and increased fitness of plants 51 at the coastal site. However, this reduction in herbivory did not decrease the homesite advantage 52 of coastal perennials. Unexpectedly, we found that the application of exogenous gibberellin 53 increased mortality due to salt spray at the coastal site for both ecotypes. Our results suggest that 54 divergence in salt spray tolerance, potentially mediated by ecotype differences in gibberellin 55 synthesis or bioactivity, is a strong driver of local adaptation and preempts any impacts of 56 herbivory in coastal habitats that experience salt spray. 57 58 Key words: local adaptation, monkeyflower, herbivory, salt spray, gibberellin, Erythranthe 59 guttata 60 61

62 Organisms experience dramatically different environmental conditions throughout their 63 geographic ranges. Spatial gradients in abiotic factors, such as temperature, salinity, and water 64 availability, as well as biotic factors, such as the presence of competitors, predators, and 65 mutualists, can generate divergent natural selection (Kawecki & Ebert, 2004; Maron et al., 66 2014). This divergent selection can in turn lead to evolutionary responses in traits that increase 67 fitness in local environments, and result in the evolution of local adaptation (Clausen et al., 1940; 68 Hereford, 2009; Kawecki & Ebert, 2004; Leimu & Fischer, 2008; Wadgymar et al., 2022). 69 Identifying the causal environmental factors contributing to adaptation is a major challenge 70 because environmental conditions often co-vary and thus, experimental manipulations are 71 necessary to identify the environmental agents of selection (Briscoe Runquist et al., 2020; 72 Hargreaves et al., 2020). Likewise, the phenotypic targets of selection are challenging to identify 73 because traits are often highly correlated, so approaches that minimize trait correlations (e.g., 74 using hybrids) or manipulate trait variation independently of other traits are necessary to identify 75 adaptive traits (Wadgymar et al., 2017, 2022). Despite their importance, experiments that 76 simultaneously manipulate putative environmental selective agents and their phenotypic targets 77 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint are uncommon (Wadgymar et al., 2017, 2022). In this study, we isolate the effect of a putative 78 selective agent, herbivory, at sites that vary in two abiotic factors, salt spray and soil moisture, 79 and manipulate trait variation using hormone applications to identify the environmental and 80 biotic drivers of local adaptation. 81 82 Traits that increase fitness on one end of an environmental gradient can reduce fitness on the 83 opposite end of that gradient, resulting in fitness trade-offs (Kawecki & Ebert, 2004). Trade-offs 84 are often caused by evolutionary changes in the allocation of limited resources to critical 85 biological functions, including growth, reproduction, and defense (Bazzaz et al., 1987; Herms & 86 Mattson, 1992). Theory predicts that resource allocation to herbivore defense should depend on 87 the risk and consequences of herbivory on fitness, and models of the evolution of plant defense 88 assume a cost to the production of herbivore defenses (Rhoades, 1979; Stamp, 2003). Within 89 species, allocation to herbivore resistance frequently trades-off with allocation to reproduction 90 (Agren & Schemske, 1993; Cipollini et al., 2017; Heil & Baldwin, 2002; Stowe & Marquis, 91 2011; Strauss et al., 2002), and increased allocation to herbivore resistance is associated with 92 longer growing seasons. This association could be driven by multiple factors, including a longer 93 period of vegetative growth and resultant longer exposure risk and apparency to herbivores 94 and/or greater herbivore pressure (Feeny, 1976; Hahn & Maron, 2016; Kooyers et al., 2017; 95 Mason & Donovan, 2015; Smilanich et al., 2016). 96 97 The physiology underlying potential trade-offs is still unclear but is likely due to the evolution of 98 plant hormone pathways in response to different environmental conditions. Recent studies have 99 shown that shifts in the allocation of resources from rapid growth to herbivore resistance are 100 made through a set of interacting gene networks (Aerts et al., 2021; Campos et al., 2016; Havko 101 et al., 2016; Huot et al., 2014; Kazan & Manners, 2012; Monson et al., 2022). Jasmonates (JA) 102 are key regulatory hormones involved in the response of plants to herbivore attack (Havko et al., 103 2016; Zhang & Turner, 2008). While JA production increases herbivore defense, it also can 104 inhibit rapid plant growth through interactions with other gene networks (Kazan & Manners, 105 2012; Yan et al., 2007; Yang et al., 2012; Zhang & Turner, 2008). For example, the interactions 106 of JAZ (Jasmonate ZIM-domain) genes with DELLA genes in the signaling pathway of 107 Gibberellin (GA) growth hormones are thought to play a key role in mediating resource 108 allocation (Havko et al., 2016; Hou et al., 2013; Yang et al., 2012). However, evidence that 109 evolutionary changes in the GA pathway lead to changes in the relative allocation of resources to 110 rapid reproduction, long-term growth, and herbivore resistance is still lacking. Further, no study 111 that we are aware of has evaluated the physiological mechanisms underlying the evolution of 112 intraspecific trade-offs driven by allocation to growth, reproduction, and defense that occurs 113 when natural populations adapt to different habitats. Furthermore, phenotypic changes induced 114 by the exogenous application of hormones allow a powerful test linking phenotype to fitness 115 across habitats in carefully controlled field studies. 116 117 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint An excellent system for investigating the mechanisms responsible for the evolution of adaptive 118 trade-offs in growth, reproduction, and resistance are locally adapted ecotypes of the yellow 119 monkeyflower, Mimulus guttatus (syn. Erythranthe guttata). Previous reciprocal transplant 120 experiments showed that the primary environmental factor contributing to local adaptation at 121 inland sites was the onset of summer drought (Hall & Willis, 2006; Lowry et al., 2008), while a 122 combination of above ground factors, including salt spray and herbivory, contributed to 123 adaptation in coastal habitats (Lowry et al., 2009; Popovic & Lowry, 2020). Inland populations 124 of M. guttatus are typically small annuals that allocate resources primarily to reproduction in 125 order to flower prior to the onset of summer drought. Coastal populations, which occur in 126 habitats with year-round soil moisture, are large obligate perennials that allocate resources 127 primarily to long-term growth (Baker et al., 2012; Baker & Diggle, 2011; Hall et al., 2010; Hall 128 & Willis, 2006; Lowry et al., 2008). Coastal populations have higher levels of phytochemical 129 defenses (phenylpropanoid glycosides, PPGs) and experience higher levels of herbivory than the 130 inland annual populations (Holeski et al., 2010, 2013; Lowry et al., 2019). In the greenhouse, 131 coastal populations are more responsive to exogenous applications of gibberellin (GA3) than 132 annuals and respond by recapitulating the elongated growth habit of inland annual populations 133 (Lowry et al., 2019). As a result, we hypothesize that natural variation in allocation to rapid 134 reproduction, long-term growth and resistance is the result of molecular changes that alter the 135 interactions of the gibberellin (GA) and jasmonic acid (JA) pathways. 136 137 In this study, we performed a manipulative reciprocal transplant experiment to test whether 138 trade-offs between allocation to vegetative growth, reproduction, and herbivore resistance 139 contribute to local adaptation at opposite ends of an environmental gradient. We predicted that 140 increased allocation to reproduction (via early flowering) would increase perennial fitness at the 141 inland site, where earlier flowering would rescue fitness for individuals that typically perish 142 before the onset of summer drought, and thus decrease annual homesite advantage. We also 143 expected that increased allocation to vegetative growth (via delayed flowering) and herbivore 144 resistance would increase annual fitness at the coast, where we expected herbivore pressure to be 145 higher. Finally, we predicted that reduction of herbivory via exclosures would rescue annual 146 fitness on the coast, and thus decrease perennial home site advantage. While our study was 147 designed to focus on the role of hormone manipulation on defense against herbivory, we instead 148 discovered that our hormone manipulations had a much larger role in causing susceptibility to 149 stress imposed by oceanic salt spray. This surprise discovery altered our approach to data 150 analysis, which we describe below in the methods and results. 151 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint

152 Study location 153 We performed a reciprocal transplant experiment at two sites – a coastal seep at the Bodega 154 Marine Reserve in Bodega Bay, CA (Latitude: 38.3157, longitude: −123.0686), and an inland 155 seep at the Pepperwood Preserve near Santa Rosa, CA (latitude: 38.5755, longitude: −122.7009). 156 Plant material 157 We used outbred maternal families from a coastal perennial population from Bodega Bay, CA (n 158 = 5 families, BHW: 38.303783, -123.064483) and an inland annual population north of Sonoma, 159 CA (n = 4 families, CAV: 38.342817, -122.4854). Outbred maternal families were generated by 160 crossing field collected maternal families in the greenhouses at Michigan State University. Seeds 161 from these outbred families were sent to UC Berkeley, planted, and placed in a 4°C cold room on 162 January 27, 2020. We staggered perennial and annual germination to synchronize their 163 development (following Popovic & Lowry, 2020). A week after beginning stratification, 164 perennial seeds were moved into a 16-hr day length growth chamber for germination. Two 165 weeks after beginning stratification, annual seeds were moved to the same 16-hr day length 166 growth chamber. Seedlings were transported from UC Berkeley to the Bodega Marine Reserve 167 (BMR) greenhouse on February 20, 2020, and then were transplanted seedlings into individual 168 cell packs over the course of a week. 169 Hormone treatments 170 We altered the allocation phenotypes of each ecotype by manipulating hormone levels of plants 171 with exogenous applications of gibberellin (GA3, a growth hormone), paclobutrazol (a GA 172 inhibitor), and methyl jasmonate (a hormone that induces herbivore resistance and antagonizes 173 GA) to test the role of those hormone pathways in adaptive trade-offs between rapid 174 reproduction versus long-term investment in vegetative growth and herbivore resistance. 175 Following a week of transplanting in the BMR greenhouse, we randomly assigned cell pack trays 176 to one of three hormone treatments or control. We sprayed plants with a 100 µM solution of 177 gibberellic acid (Consolidated Chemical Solvents LLC, following Lowry et al., 2019), 10 mM 178 methyl jasmonate (TCI America, Portland, Oregon, USA), and 14.3 mg/L solution of 179 paclobutrazol (General Hydroponics, Santa Rosa, California, USA). Concentrations of methyl 180 jasmonate and paclobutrazol were chosen after conducting dose response experiments at the 181 MSU greenhouses in winter 2019. These concentrations were chosen based on the minimum 182 concentration needed to elicit a phenotypic change relative to controls without detrimental 183 effects (e.g., leaf damage, stunting, death). Using a spray bottle, we sprayed individual plants 5 184 times, corresponding to 3.5 mL of solution. The control consisted of spraying plants with 3.5mL 185 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint of a 0.25% ethanol solution since a dilute ethanol solution was needed to dissolve methyl 186 jasmonate and all hormones were dissolved in a 0.25% ethanol solution. Hormones were applied 187 once on a single day in the greenhouse prior to field transplanting. All trays were covered with 188 clear plastic domes for 24 hours and moved to different benches to prevent cross contamination. 189 Field planting 190 Prior to transplanting, we removed vegetation from ten (108cm x 84cm) plots at each site. We 191 dug trenches along the edge of each plot to bury the bottom of our control and exclosure 192 structures. At the Pepperwood Preserve, we transplanted 800 seedlings on March 9, 2020, and 193 200 seedlings on March 12, 2020. At the Bodega Marine Reserve, we transplanted 800 seedlings 194 on March 10, 2020, and 200 seedlings on March 13, 2020. Plants were fully randomized within 195 each block (n = 100 seedlings/block) and labeled with a plastic tag. 196 Herbivore exclosures 197 To lessen the effect of herbivory and potentially measure a cost to defense production in the 198 absence of herbivores, we deployed herbivory exclosures on four out of ten plots at each site 199 (Figure 1). A previous reciprocal transplant experiment at our study sites used exclosures that 200 blocked all above-ground factors using agrofabric (Popovic & Lowry, 2020), and thus could not 201 separate the effects of salt spray and herbivory on plant fitness. Thus, we designed exclosures 202 that excluded many herbivores but allowed salt spray to pass through. The exclosures were 108 203 cm long x 84 cm deep x 87 cm tall and constructed of 3/4" pvc pipe covered with fiberglass 204 window-screen (18x16 mesh/inch) that was affixed with fishing line and marine epoxy. Each 205 exclosure had screen doors along both long sides that were attached with velcro to allow access 206 to the plots. The screen extended 4 inches down into the soil around the plots. To control for 207 shading or moisture-collection due to the screen, the remaining six plots at each site were 208 covered with control structures. These structures differed in that only the tops and 30 cm down 209 each side were covered with window screen. 210 Field Censuses 211 After transplanting, we performed regular censuses of our transplant sites recording survival, the 212 presence of herbivore damage, the identity of herbivores (when possible), the presence of salt-213 spray damage, and the presence and number of reproductive structures (buds, flowers, and 214 fruits). In our census, we distinguished damage and death caused by salt spray from herbivory: 215 salt-damaged leaves appeared necrotic and brown and exhibited no sign of herbivore damage 216 (i.e., no missing tissue), when salt damage spread to the entire plant and no green tissue 217 remained, we considered plants to be killed by salt spray. We were prevented from accessing our 218 transplant sites for two weeks at Bodega Marine Reserve and seven weeks at Pepperwood 219 Preserve after transplanting due to the 2020 COVID-19 pandemic lockdowns. Due to site 220 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint differences in growing season length, and restricted access due to the 2020 COVID-19 221 pandemic, we censused each site at different intervals and for different lengths of time 222 (Pepperwood Preserve (inland site):12 censuses over 139 days, Bodega Marine Reserve (coastal 223 site): 22 censuses over 194 days). Since we were unable to access our sites for weeks because of 224 the pandemic, we missed observing the first flower opening for many annual plants. For these 225 plants, we estimated the onset of flowering as the date we first observed any reproductive 226 structures. Our censuses occurred on a roughly weekly basis after we were able to re-access our 227 sites and we continued to estimate the onset of flowering based on the initial observation of a 228 bud, flower, or fruit during each census. 229 Plant chemistry 230 We sampled leaves for chemical analysis 55 to 57 and 59 to 64 days after transplanting at the 231 coastal and inland site, respectively. To minimize the potential effect of diurnal fluctuation in 232 PPGs (phenylpropanoid glycosides), we sampled from 9am until 1pm, and to minimize the effect 233 of leaf position, we sampled 2 leaves from the 3rd node when possible, using leaves from the 4th 234 and 5th nodes if leaves at the 3rd node were damaged. After sampling, leaves were flash-frozen 235 with liquid nitrogen and then freeze-dried. For samples that did not meet the minimum dry mass 236 (3mg), we either pooled them with other low-mass samples (by grouping within all fixed and 237 random factors as discussed below) or excluded them. Our final sample size was 216 from 238 Bodega (perennials only due to high annual mortality at Bodega) and 599 from Pepperwood. To 239 determine the PPG concentrations in sampled leaves, we ground, extracted, and prepped extract 240 aliquots as described in Holeski et al. (2013). We then used high-performance liquid 241 chromatography (HPLC) to quantify PPGs. The HPLC method is described in (Kooyers et al., 242 2017) and was run on an Agilent 1260 HPLC with a diode array detector and Poroshell 120 EC-243 C18 analytical column [4.6/i1 ×/i1 250/i1 mm, 2.7 μ m particle size]; Agilent Technologies). We 244 calculated concentrations of individual PPGs as verbascoside equivalents, using a standard 245 verbascoside solution (Santa Cruz Biotechnology, Dallas, Texas), as described in (Holeski et al., 246 2013, 2014). 247 Statistical analyses 248 We performed all statistical analysis in R version 4.3.1 (R core team 2023). We addressed the 249 following main questions within each transplant site: Do annuals and perennials differ in 250 allocation to reproduction, allocation to herbivore resistance, and fitness? Do hormones and 251 herbivores influence allocation to reproduction, allocation to herbivore resistance, and local 252 adaptation? 253 254 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Measurements of allocation & adaptation 255 To compare allocation to reproduction, we measured the onset of flowering for each plant. 256 Earlier-flowering plants invest in reproductive tissues at a time when other plants are allocating 257 all energy to growth and defense. To compare allocation to herbivore resistance, we measured 258 the presence or absence of herbivore-attack for each plant and the concentration and composition 259 of the defensive compounds PPGs in leaves. Finally, to determine adaptation in each 260 environment, we measured survival across the season, the presence of flowers, and seasonal fruit 261 production. For annuals, these measures indicate lifetime fitness, whereas perennials that 262 survived the season have the potential to reproduce in subsequent years. 263 264 Univariate analysis 265 Within each transplant site, we fit mixed effect models for each analysis that included ecotype, 266 hormone treatment, and exclosure type as interactive fixed factors. The response variables were 267 flowering time, the presence or absence of herbivory, survival, total PPG concentration (summed 268 concentration for all PPG compounds), whether an individual produced a reproductive structure 269 (buds, flowers, or fruit), or the number of fruits produced by plants that flowered at the end of 270 the season. All models also included two random effects for maternal family and experimental 271 plot. We fit all mixed models except for the survival model with the R package glmmTMB 272 (Brooks et al., 2017), and modeled survival using the R package coxme (Therneau, 2022). We 273 identified the best fitting error distributions by evaluating model diagnostics with the R package 274 DHARMa (Hartig, 2022). We fit mixed models for flowering time with gaussian error 275 distributions, mixed models for herbivory and flowering probability with binomial error 276 distributions, and mixed models for log-transformed total PPGs with gamma distributions. We 277 modeled survival using a mixed effect Cox Proportional Hazards model, and modeled fruit 278 number with a zero-inflated negative binomial error distribution at the coastal site and a negative 279 binomial error distribution at the inland site. To prevent model overfitting, we used an analysis 280 of deviance (Wald χ 2 test) to assess the significance of model terms and sequentially removed 281 unsupported model terms (R package car, (Fox & Weisberg, 2018). We compared fits of 282 complex versus reduced models using likelihood ratio tests (LRT) to find the minimum adequate 283 model for each response variable in each site (Tables S1 and S2). We compared treatment groups 284 using post-hoc tests on the minimum adequate model with the R package emmeans (Lenth et al., 285 2020). No contrasts were performed on predictor variables that were not in the minimum 286 adequate model. We predicted the mean and 95% confidence intervals for each response variable 287 from our models using the R package ggeffects (Lüdecke, 2018). For all non-binary response 288 variables, we predicted confidence intervals via bootstrapping (n=500 iterations). We plotted raw 289 data and predictions in ggplot2 (Wickham, 2016) and combined plots with patchwork in R 290 (Pedersen, 2019). 291 292 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Multivariate analysis 293 Within each transplant site, we modeled the concentration of all nine different PPGs (the PPG 294 arsenal) using mixed effect models for each analysis that included ecotype (at Pepperwood only), 295 hormone treatment, and exclosure type as interactive fixed factors and block as a random factor. 296 We fit all models with PERMANOVA with Bray-Curtis distance using the adonis2 function 297 from the R package vegan (Oksanen, 2016). We dropped all non-significant factors for the 298 minimum adequate model (Table S3). To test for homogeneity of variance among treatment 299 groups, which can influence inference, we used the betadisper function from the vegan package. 300 The only factor that had heterogeneity of variance among levels was ecotype. Due to the strength 301 of the signal for ecotype, and confirmation from other studies that annuals and perennials have 302 different PPG arsenals (Holeski et al., 2013), we are confident that differences due to ecotype are 303 not attributable only to heterogeneity of variance. We compared treatment groups using post-hoc 304 tests on the minimum adequate model with the function pairwise.adonis2 from the R package 305 pairwiseAdonis (Arbizu, 2019). To visualize how multivariate PPG composition is influenced by 306 our factors, we used non-metric multidimensional scaling (NMDS) (MetaMDS function in vegan 307 package with Bray-Curtis distance to determine dissimilarity) and added standard-error ellipses 308 at 95% confidence around the centroid of each cluster (function ordiellipse from package vegan). 309 310

311 Do annuals and perennials differ in allocation to reproduction and 312 vegetative growth (through differences in the onset of flowering)? Do 313 hormone treatments or herbivore exclosures affect allocation? 314 At both sites, annuals had greater allocation to reproduction, flowering significantly earlier than 315 perennials. Hormones did influence this allocation slightly, though only for annuals; annuals 316 treated with GA (at both sites) and paclobutrazol (at the coast only) showed delayed flowering 317 relative to controls. Herbivory (in control structures vs exclosures) did not influence allocation to 318 reproduction. 319 320 At the coastal site, gibberellic acid (GA) and paclobutrazol slightly, but significantly, delayed 321 annual flowering time relative to control annuals (plants sprayed with 0.25% ethanol). 322 Paclobutrazol-treated and GA-treated annuals flowered 10-17 days later than controls (Fig. 2A, 323 Tukey post-hoc tests: Table S3). However these effects were small relative to ecotype 324 differences in flowering time: all annuals flowered 46 to 64 days earlier than their corresponding 325 hormone treated perennials, all significant differences (Table S4). 326 327 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint At the inland site, in the control structures only, GA slightly, but significantly, delayed annual 328 flowering time relative to the control. GA-treated annuals in control structures flowered 9 days 329 later than control annuals in control structures (Fig. 2B, Tukey post-hoc tests: Table S4). Again, 330 this effect was small relative to ecotype differences in flowering time: all annuals flowered 39 to 331 62 days earlier than their corresponding hormone-treated perennials in both the exclosures and 332 control structures, all significant differences (Table S5). 333 334 Hormone treatments had no effect on perennial flowering time relative to controls at either 335 transplant site (Tables S4 & S5). Exclosures had no effect on flowering time at either transplant 336 site (Tables S1, S2, S4, S5). 337 Do annuals and perennials differ in allocation to herbivore resistance 338 (via changes in the probability of herbivore attack)? Do hormone 339 treatments or herbivore exclosures affect allocation? 340 At both sites, and contrary to expectations, perennials were more likely to experience herbivory 341 than annuals (excluding the exclosures at the coast, in which both ecotypes experienced 342 equivalent chances of herbivory). GA was the only hormone to influence the probability of 343 herbivory, and only at the coast, where, again contrary to expectations, it reduced the probability 344 of herbivory for both ecotypes. This is likely due to an interaction with salt-spray resistance 345 rather than allocation to herbivore resistance. 346 347 Perennials were significantly more likely to be damaged by herbivores than annuals in the 348 control structures at the inland site, and in both control structures and exclosures at the coastal 349 site, but the difference between ecotypes was smaller in the exclosures (Table S6 and S7). At the 350 inland site, herbivores damaged 42% (126/300) of perennials and 13% (38/300) of annuals in the 351 control structures and 33% (65/200) of perennials and 13% (26/200) of annuals in the exclosures. 352 At the coastal site, herbivores damaged 74% (221/300) of perennials and 4% (12/300) of annuals 353 in the control structures and 48% (96/200) of perennials and 9% of annuals (17/200) in 354 exclosures. However, these numbers are somewhat misleading at the coastal site, since annuals 355 perished quickly due to salt spray and had less time to encounter herbivores and accrue 356 herbivory. 357 358 The mesh-size of screen used in our exclosures, while necessary to allow salt spray to enter, did 359 allow some small insects, including leaf miners and weevils, to enter the exclosures (or they 360 were present when the exclosures were erected) and damage plants mildly. As a result, our 361 herbivore exclosures did not significantly reduce the probability of insect herbivory for 362 perennials at either transplant site, although they were highly effective at reducing herbivory 363 from deer and voles that removed flowering stalks from plants, greatly impacting fecundity. 364 365 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint At the coastal site, GA-treatment reduced the probability of herbivore attack in both annuals and 366 perennials, likely due to GA effects on survival and salt spray sensitivity detailed below. The 367 only effect of exclosure was increasing herbivory probability for GA-treated annuals, though 368 again this is likely due to an interaction with salt-spray (Figure 2C, Tukey post-hoc tests: Table 369 S6). At the inland site, hormone-treated annuals and perennials did not significantly differ from 370 their respective controls and exclosures did not influence the probability of herbivory for either 371 ecotype (Tukey post-hoc tests: Table S7). 372 Do annuals and perennials differ in allocation to herbivore resistance 373 (via changes in PPGs)? Do hormone treatments or herbivore exclosures 374 affect allocation? 375 Perennials showed greater allocation to herbivore resistance (via PPG concentration) than 376 annuals at the inland site, while annual mortality at the coast prevented this comparison. 377 Herbivory (in control structures vs exclosures) had limited impacts on PPGs, moderating the 378 effects of hormone treatments at the coast and influencing multivariate PPG arsenals inland. GA 379 influenced PPG allocation at both sites (negatively at the coast and positively inland) and methyl 380 jasmonate increased PPG allocation inland. 381 382 At the inland site, perennials had significantly higher total PPG concentration than annuals 383 (Tukey post-hoc tests: Table S9) and annuals and perennials differed in their multivariate PPG 384 arsenals. The effect of ecotype was generally stronger than any hormone effects. We were unable 385 to compare annuals and perennials at the coastal site due to high annual mortality. 386 387 At both sites, exclosures had no effect on total PPG concentration (Table S1), though exclosures 388 did moderate the effect of hormone treatment at the coastal site (Table S2). Exclosure did not 389 influence the multivariate PPG arsenal at the coastal site but did at the inland site (Table S3). At 390 the coast, the only effect of hormone treatment was that GA reduced total PPG concentration of 391 perennials in the control plots (Figure 3a, Tukey post-hoc tests, Table S8) and caused the PPG 392 arsenal to differ from control plants (Figure 3c, PERMANOVA pairwise, Table S10). While this 393 impact of GA is consistent with our predictions that GA downregulates defense-allocation, it is 394 also possible that the decrease in total PPG is due to increased salt-stress experienced by GA-395 treated plants. Inland, hormone treatments did not influence PPGs in perennials (Figure 3b,c, 396 Table S9). In annuals at the inland site, GA and MeJa increased the total concentration of PPGs 397 (Figure 3b, Table S10) and caused the PPG arsenal to differ (Figure 3d, PERMANOVA 398 pairwise, Table S11). While we expected MeJa to increase allocation to defense, we expected 399 GA to decrease it. However, the increase in total PPG is consistent with an increase in days to 400 flowering in GA-treated annuals at the inland site (these traits positively covary in annuals, 401 (Kooyers et al., 2020), though the mechanism for this shift is unknown. 402 403 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Do annuals and perennials differ in fitness components (survival, the 404 probability of flowering, and fruit number)? Do hormone treatments or 405 herbivore exclosures affect fitness components and homesite advantage? 406 Perennials survived significantly longer than annuals at both sites. At the coast, while annuals 407 and perennials were equally likely to flower, the vast majority of annuals were killed by salt 408 before producing fruits. Inland, annuals were more likely to flower and produce fruits than 409 perennials, though perennials that did flower produced as many fruits as annuals. Herbivory (in 410 control structures vs exclosures) influenced the probability of flowering and fruit production only 411 at the coast where herbivory resulted in reproductive failure of perennials outside of exclosures. 412 In general, GA had a negative effect on fitness at both sites, though impacts varied by fitness 413 component and ecotype across sites. At the coast, GA reduced survival relative to controls by 414 increasing susceptibility to salt spray and reduced the probability of flowering for both ecotypes. 415 Inland, GA and MeJa reduced the probability of flowering for perennials and GA reduced fruit 416 production in both ecotypes. 417 GA-treatment reduced survival due to oceanic salt spray at the coastal site 418 At both transplant sites, perennials had significantly higher survival by the end of the experiment 419 than annuals (coastal site: 1% (4/500) of annuals and 77% (383/500) of perennials survived; 420 inland site: 2% (10/500) of annuals and 21% (106/500) of perennials survived, Tukey post-hoc 421 tests: Table S12 and S13). At the coastal site, salt spray was the only source of mortality for 422 annuals and 93% (109/117) of the perennials that died. For the remaining 8 perennial plants, the 423 source of mortality was attributed to herbivory. The main source of mortality at the inland site 424 was the onset of summer drought. 425 426 At both transplant sites, exclosures had no effect on survival (Figure 4, Tables S1, S2 and S12). 427 Hormone treatments had no effect on survival at the inland site (Table S1), but GA treatment 428 significantly reduced survival for both ecotypes relative to their respective controls at the coastal 429 site (Figure 4, Table S12). GA treatment reduced survival for both ecotypes at the coastal site by 430 increasing susceptibility to salt spray. GA-treated perennials were also more upright compared to 431 prostrate controls, and elongated their stems early in development like annuals which may have 432 increased exposure to salt spray (Figure 5). 433 434 GA-treatment reduced flowering probability at the coastal site 435 Despite high mortality due to salt spray at the coastal site, annuals and perennials did not 436 significantly differ in the probability of flowering (Tukey post-hoc tests: Table S14). Due to their 437 rapid phenology, 40% (198/500) of annual transplants were able to flower prior to dying of salt 438 spray, although death occurred quickly after flowering so very few annuals produced fruit 439 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint (detailed below). Herbivore exclosures significantly increased the probability of flowering for 440 perennials (control structures: 27% (80/300) of perennials and 26% (77/300) of annuals 441 flowered; exclosures: 74% (148/200) of perennials and 61% (121/200) of annuals flowered, 442 Table S14), which may be due to the reduction of large mammalian herbivory and/or minor 443 buffering of salt spray from condensation collecting on mesh screens. In addition, GA treatment 444 reduced the probability of flowering for both ecotypes relative to their respective controls, likely 445 due to the effect of GA on sensitivity to salt spray (Figure 6A, Table S14). 446 447 At the inland site, annuals had a significantly higher probability of flowering than perennials, and 448 exclosures had no effect on the probability of flowering for either ecotype (control structures: 449 39% (117/300) of perennials and 92% (277/300) of annuals flowered; exclosures: 66% 450 (133/200) of perennials and 100% (200/200) of annuals flowered, Tukey post-hoc tests: Table 451 S15). GA and MeJa treatment reduced the probability of flowering in perennials, but hormone 452 treatment did not affect the probability of flowering in annuals (Figure 6B, Table S15). 453 Herbivore exclosures drastically increased fruit production at the coastal site 454 At the coastal site, only plants protected by exclosures successfully produced fruit by the end of 455 the season (annuals: 0% (0/300) produced fruit in control structures and 4% (7/200) produced 456 fruits in the herbivore exclosures; perennials: 0% (0/300) produced fruit in control structures and 457 60% (119/200) produced fruits in the herbivore exclosures). The reason that none of the plants 458 outside of the exclosures produced fruits was because of complete herbivory of the 459 inflorescences of these plants by mule deer (Odocoileus hemionus). Since no plants produced 460 fruit outside of the exclosures, and few annuals produced fruit inside the exclosures (annuals in 461 exclosures: n=3 controls, n=1 GA-treated, n=3 paclobutrazol-treated), we analyzed only the 462 effect of hormone treatments on perennial fruit production inside the exclosures at the coastal 463 site (Figure 6C). Fruit production in perennial plants that flowered in exclosures at the coastal 464 site was not significantly associated with hormone treatment (Table S1). 465 466 At the inland site, 88% (265/300) of annuals and 29% (86/300) of perennials produced fruit in 467 the control structures, while 100% (200/200) of annuals and 53% (105/200) of perennials 468 produced fruit in the exclosures. The majority of plants that flowered produced fruit: 96% 469 (265/277) of annuals and 74% (86/117) of perennials that flowered produced fruit in the control 470 structures, while 100% (200/200) of annuals and 79% (105/133) of perennials that flowered 471 produced fruit in the exclosures. However, annuals and perennials that flowered did not 472 significantly differ in fruit production in either control structures (mean fruit number for 473 flowering annuals: 4.6, flowering perennials: 5.7) or exclosures (mean fruit number for flowering 474 annuals: 7.6, flowering perennials: 7.8; Tukey post-hoc tests: Table S16). GA treatment 475 significantly reduced fruit production in both ecotypes that flowered relative to their respective 476 controls (difference between control and GA-treatment for perennials: 5.8 fruit in exclosures, 2.3 477 in controls; and for annuals: 3.3 fruit in exclosures, 0.7 in controls; Figure 6; Table S16). 478 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 479

480 Across environmental gradients, shifts in allocation between reproduction, growth, and defense 481 have been found to follow predictable patterns, suggesting that these shifts underlie local 482 adaptation (Bazzaz et al., 1987; Hahn & Maron, 2016; Züst & Agrawal, 2017). However, 483 multiple abiotic and biotic factors co-vary across environmental gradients and multiple traits 484 often differ between locally adapted populations, making the identification of selective agents 485 and their phenotypic targets a major challenge (Wadgymar et al., 2017, 2022). In this study, we 486 used a manipulative reciprocal transplant experiment to test the hypothesis that herbivory and 487 divergence in allocation to reproductive timing, vegetative growth, and defense against 488 herbivores contributes to local adaptation across a coastal to inland environmental gradient. 489 Growing seasons are shorter in inland environments, which generates selection for earlier 490 reproduction. At our coastal site, herbivore exclosures dramatically increased fecundity of local 491 coastal perennials, but contrary to our predictions, did not contribute to local adaptation. This is 492 likely due to the abiotic effect of salt-spray pre-empting the impacts of herbivory on annuals. Our 493 hormone treatments slightly shifted allocation between vegetative growth, reproduction and 494 defense in each ecotype, but did not recapitulate the full effect size of differences previously 495 observed in controlled greenhouse conditions (Lowry et al., 2019). Nevertheless we observed 496 dramatic effects of our hormone treatments on survival and fecundity across our transplant sites. 497 Despite delaying flowering, the GA application caused obviously earlier bolting and taller 498 heights in the perennial transplants. This earlier bolting, and possibly other physiological 499 changes, may have been responsible for the increased mortality due to salt spray on the coast in 500 both ecotypes, and salt spray was the primary (>99%) source of mortality for transplants at our 501 coastal site. Our results suggest that divergence in salt spray tolerance, potentially mediated by 502 ecotype differences in gibberellin synthesis/sensitivity, is an important driver of local adaptation 503 to coastal habitats. 504 Role of biotic interactions in local adaptation 505 The organisms a plant interacts with vary across the landscape, causing different selective 506 pressures (Friberg et al., 2019; Thompson, 2005; Urban, 2011). Given the differences in the 507 abiotic environment at our two sites - cool and foggy on the coast, hot and dry inland - the 508 communities of organisms which our plants interact with differ substantially. The moist coastal 509 environment has far more molluscan herbivores (snails, slugs), and a rare leaf-mining fly 510 (Eiseman et al., 2023), which we did not see at the much drier Pepperwood Preserve. Voles and 511 deer also contribute to herbivory at the coastal site only. Given the differences in communities, 512 differences in defense-levels, and prior research suggesting differences in intensity of herbivory 513 on the coast and inland (Holeski, 2007), we predicted herbivory would be a driving factor in 514 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint local adaptation. Remarkably, we found no effects of herbivory on local adaptation at these two 515 specific sites; nevertheless, we stress that insect-plant interactions regularly occur in a complex 516 mosaic across the landscape and vary temporally (Rotter et al., 2022). 517 High rate of herbivory at the coastal site did not contribute to local adaptation 518 At the coast, we predicted that high rates of herbivore attack would result in herbivory playing a 519 strong role in local adaptation. Although we observed high rates of herbivory, reducing 520 herbivory with exclosures did not increase local adaptation because of the effect of an abiotic 521 factor, oceanic salt spray. Annuals transplanted on the coast quickly exhibited necrosis from salt 522 spray before dying; the window that they could have received herbivory was short, and they were 523 likely poor quality host plants during that time. The perennials, in comparison, were larger, 524 healthier, and had many more days in which to encounter an herbivore and receive damage. 525 Ephemeral plants are more likely to escape herbivory (Feeny, 1976), and all reproductive 526 herbivory at the coastal site came after the median death date of our annual plants. This pattern 527 highlights the importance of the timing of selective events, particularly for local adaptation of 528 ecotypes with differing life-history strategies. The importance of fecundity versus survival are 529 likely to differ between ecotypes (DeMarche et al., 2016), and early-season factors (like coastal 530 salt spray) that impact survival might disproportionately contribute to fitness differences between 531 populations relative to a late season factors that influence fecundity (such as herbivory) (Crone, 532 2001; Wadgymar et al., 2022). 533 534 While this study suggests that herbivory is preempted from playing a role in keeping annuals out 535 of coastal environments, it does not mean it is unimportant. In the control structures on the coast, 536 perennials completely failed to reproduce due to deer herbivory of inflorescences. By virtue of 537 allocating growth to clonal expansion and non-reproductive tissue, perennials are likely 538 increasing both tolerance (Stevens et al., 2008) and temporally escaping herbivory. Some 539 populations are completely sterilized (i.e., all inflorescences are completely consumed by 540 herbivores) in certain years (Toll, pers. obs.), and thus herbivory may be an extremely strong 541 selective pressure in the morphology, allocation to clonal growth, and reproductive timing of 542 these coastal perennials. The results of this study were also clearly influenced by the close 543 proximity of our coastal field site to the open ocean (within 50 meters of the shoreline) While it 544 is common for coastal perennials to grow in close proximity to the ocean, where they are 545 impacted by high levels of salt spray, it is also common for them to grow slightly further inland, 546 where salt spray is greatly reduced (Barbour, 1978; Boyce, 1954; Du & Hesp, 2020). 547 Life history contributed to differences in herbivore attack at the inland site 548 Our finding that herbivory did not influence local adaptation inland is somewhat less surprising, 549 as there is evidence that herbivore damage is generally less extensive there (Holeski, 2007). It 550 was unexpected, however, that perennials were also more likely to be attacked by herbivores 551 than annuals at the inland site, as we predicted that perennials would be more resistant to 552 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint herbivory due to ecotype differences in phytochemical defenses (PPGs). At our inland site, 553 however, perennials were more likely to be attacked even during periods of time when both 554 ecotypes were alive in the same site. The general (though non-significant) trend for the homesite 555 advantage of annuals to decrease in the exclosures relative to the control structures, may suggest 556 that herbivory, in some years, does contribute to local adaptation inland. Higher attack rates for 557 perennials could be due to differences in apparency caused by differences in plant size (Feeny, 558 1976), or herbivore preference due to nutritional differences or water content. In addition, while 559 PPGs are feeding deterrents to generalists, some can be feeding stimulants for specialist 560 herbivores (Holeski et al., 2013; Rotter et al., 2018), and thus perennials may be more likely to 561 get attacked by specialists. Our presence-absence measure of herbivory also may have missed 562 differences in degree of herbivory among plants that were attacked, which may have greater 563 impacts on fitness. 564 Hormone pathways underlying local adaptation 565 Oceanic salt spray sensitivity increased with gibberellin treatment 566 The most surprising result of our experiment was how dramatically GA3 application decreased 567 survival of coastal perennial genotypes at the coastal field site. Based on the patterns of damage, 568 necrosis of plant tissue, we attributed this mortality primarily to oceanic salt spray. There are two 569 non-mutually exclusive ways that GA3 could have decreased fitness in the coastal environment 570 with regard to salt spray. First, the addition of GA3 increased plant height, as evident by 571 increased internode elongation of plants (Lowry et al., 2019). Increased plant height could put 572 the aboveground portions of these plants more directly in the path of prevailing wind delivering 573 the salt spray (Zambiasi & Lowry, 2023). A second hypothesis is that the GA3 treatment may 574 directly increase susceptibility of tissues to salt spray independent of changes in plant height. 575 The second hypothesis is particularly intriguing, as it is the opposite of what would be expected 576 based on the soil salinity literature. For example, previous experiments in rice (Rodríguez et al., 577 2006), wheat (Iqbal & Ashraf, 2013), apple (X. Wang et al., 2019), cucumber (Y. Wang et al., 578 2020), and sorghum (J. Liu et al., 2023) have all found that the application of GA3 increases 579 yields under saline conditions. The conflicting results of those studies and our experiment make 580 it clear that findings from the soil salinity literature cannot be directly extrapolated to what is 581 experienced by plants growing in coastal environments, where salt spray is a major source of 582 stress on plant aboveground tissues (Boyce, 1954; Du & Hesp, 2020; Itoh et al., 2024). The exact 583 mechanisms by which GA3 increases salt spray susceptibility are still not clear, but are an active 584 focus of our current research. One possibility is that the addition of GA3 increases stomatal size 585 and/or opening (X. Liu & Hou, 2018; Nir et al., 2017; Shohat, Cheriker, et al., 2021; Shohat, 586 Eliaz, et al., 2021), which allows for more salt spray to enter leaves. 587 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Oceanic salt spray preempted herbivory 588 Our hormone treatments altered the probability of herbivore attack at the coastal site, however, 589 this was not likely due to an increase in allocation to resistance. GA-treated perennials were less 590 likely to be attacked by herbivores (Figure 2), but were also less salt spray tolerant (Figure 4) 591 than control perennials at the coastal site. GA treatment was not associated with Total PPG 592 concentrations in the exclosures at the time of tissue sampling, but GA-treated perennials had 593 lower Total PPG concentrations than control perennials in the control structures at the coastal 594 site (Figure 3). Thus, the observed decrease in herbivory was likely due to a decrease in tissue 595 quality induced by salt spray stress. GA-treated plants also senesced and died faster in the control 596 structures at the coastal site; the median death date in the control structures was 53 days 597 compared to 96 days in the exclosures (Figure 4). Coastal fog sometimes condensed on the 598 screens that we used to exclude herbivores, which may have slightly decreased the transmission 599 of salt spray into exclosures. Eventually, individuals in both the exclosures and controls showed 600 evidence of salt spray damage and death, but the lag in the onset of damage may also partially 601 explain why we observed a reduction in total PPGs in the control structures but not the 602 exclosures at the time of sampling. 603 Hormone effects were attenuated in the field 604 Aside from salt spray tolerance, the effects of our hormone applications on measured phenotypes 605 were markedly weaker than we expected from greenhouse experiments. The limited impact on 606 flowering time and the notable impact on growth habit are consistent with a previous greenhouse 607 study (Lowry et al., 2019). In that same greenhouse study, however, daily spraying of GA on 608 perennial monkeyflowers reduced the concentration of PPGs. In our study, GA only reduced the 609 concentration of PPGs in the control structures at the coast (Figure 3), though it did alter the PPG 610 arsenal of perennials at both sites, albeit not dramatically. We also expected a greater impact of 611 MeJa, an antagonist of GA that induces plant defense (Baldwin, 1998; Kessler & Baldwin, 612 2002). This may be due in part to methodological constraints imposed by field studies. The 613 difficulty of preventing cross-contamination of nearby plants prevented us from repeatedly 614 treating our transplants with hormones after field planting, which may have weakened and/or 615 attenuated the effects compared with long-term applications (Hummel et al., 2009). Also, 616 interactions with environmental conditions in the field (e.g., short days, cold nights, and greater 617 temperature variation relative to greenhouse conditions) may have impacted our measured traits 618 more than the hormone treatments. For example, temperature interacts strongly with GA-619 pathways to control phenology and development (Penfield, 2008) and PPG production in 620 monkeyflowers is influenced by temperature and day-length (Blanchard et al. in review). Thus, 621 while our finding that GA impacts local adaptation via salt-tolerance supports the value of field-622 based hormone treatments, our study also suggests the need for field-based preliminary trials to 623 determine field-relevant doses. 624 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint

625 Interactions among hormone pathways mediate differences in allocation to growth, reproduction, 626 and resistance, but few studies have investigated how evolutionary changes in hormone 627 pathways contribute to local adaptation (James et al., 2023; Wilkinson et al., 2021). Evidence 628 that GA application reduced allocation to resistance in greenhouse experiments led us to 629 hypothesize that selection by herbivores drove the evolution of GA-suppression in coastal 630 perennials. Unexpectedly, we found that GA application reduced local adaptation of perennials at 631 the coast by making them more susceptible to salt spray and that coastal salt spray killed all 632 annuals. This suggests a strong role for an abiotic factor, salt spray, in selecting for differences in 633 GA pathway genes in coastal populations. Additionally, herbivory had a dramatic impact on 634 perennial fecundity at the coast, though it was precluded from contributing to local adaptation by 635 the salt spray induced mortality of all annuals at the coast. While our study shows how hormone 636 applications can be used to investigate the mechanisms underlying local adaptation, our results 637 also stress the importance of considering the interaction and timing of selective agents. 638 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Figure Legends 639 Figure 1. Photographs depicting transplant sites and structures used in the reciprocal transplant 640 experiment. For our reciprocal transplant experiment, each site had ten plots, each with 100 641 plants. Each site had six control plots (with mesh tops and partially-mesh sides) and four 642 exclosure plots (fully enclosed with mesh on all sides). A control and exclosure plot are shown 643 side by side in the foreground of the inland site. 644 645 Figure 2. Allocation to reproduction and defense: flowering time and probability of herbivory of 646 annuals (circles) and perennials (triangles) treated with gibberellic acid (GA, yellow), 647 paclobutrazol (Paclo, blue), and methyl jasmonate (MeJa, green), and the controls (0.25% 648 ethanol, black) in control structures and herbivore exclosures at the coastal site, Bodega Marine 649 Reserve (A & C), and the inland site, Pepperwood Preserve (B & D). Larger symbols in the 650 foreground are the mean predictions and 95% confidence intervals from the minimum adequate 651 mixed effect models, smaller and lighter symbols in the background are the raw data. Results of 652 Tukey post-hoc contrasts within each site are indicated above each prediction; shared letters 653 indicate that groups do not significantly differ, while non-overlapping letters indicate that groups 654 significantly differ within each site. Exclosure type was not plotted for flowering time on the 655 coast (A) because the minimum adequate model did not include exclosure type as a fixed effect. 656 657 Figure 3. Allocation to chemical defense: total concentration of all PPGs and differences in 658 multivariate PPG arsenals of annuals (circles) and perennials (triangles) treated with gibberellic 659 acid (GA, yellow), paclobutrazol (Paclo, blue), and methyl jasmonate (MeJa, green), and the 660 controls (0.25% ethanol, black) in control structures and herbivore exclosures at the coastal site, 661 Bodega Marine Reserve (A & C), and the inland site, Pepperwood Preserve (B & D). In the Total 662 PPG figures (A & B), larger symbols in the foreground are the mean predictions and 95% 663 confidence intervals from the minimum adequate mixed effect models, smaller and lighter 664 symbols in the background are the raw data. Exclosure type was not plotted for Total PPG at the 665 inland site (B) because the minimum adequate model did not include exclosure type as a fixed 666 effect. PPG arsenal figures (C & D) use non-metric multidimensional scaling (NMDS) with 667 Bray-curtis distance (and 95% confidence interval ellipses) to visualize multivariate differences 668 among plants. Exclosure type was either not a significant factor in the multivariate model (C ) or 669 did not interact with other fixed effects (D) and was therefore not included in these plots. 670 671 672 Figure 4. GA application decreased survival at the coastal site. Survival probabilities for annual 673 (solid line) and perennial (dashed line) transplants at the coastal site, Bodega Marine Reserve 674 (A), and the inland site, Pepperwood Preserve (B). Survival probabilities and 95% confidence 675 intervals for control (black), GA (yellow), methyl jasmonate (blue) and paclobutrazol (green) 676 treatments were predicted from Cox Proportional Hazards models. At the inland transplant site 677 (B), survival probabilities and 95% confidence intervals were only plotted for ecotypes (grey) 678 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint because the minimum adequate model did not include exclosure or hormone treatment as a fixed 679 effect. Results of Tukey post-hoc contrasts within each site are indicated above each final 680 predicted survival; shared letters indicate that groups do not significantly differ, while non-681 overlapping letters indicate that groups significantly differ within each site. 682 683 Figure 5. GA application on perennials resulted in stem-elongation relative to controls. The 684 plants pictured (on day 16 after transplantation) are from the same family and were grown in the 685 same plot at the coastal site. 686 687 Figure 6. Herbivore exclosures tended to increase, while GA applications tended to decrease 688 components of fecundity: probability of flowering and fruit number of annuals and perennials 689 that flowered treated with gibberellic acid (GA, yellow squares), paclobutrazol (Paclo, blue 690 diamond), and methyl jasmonate (MeJa, green triangle), and the controls (0.25% ethanol, black 691 circles) in control structures and herbivore exclosures at the coastal site, Bodega Marine Reserve 692 (A & C), and the inland site, Pepperwood Preserve (B & D). Larger symbols in the foreground 693 are the mean predictions and 95% confidence intervals from mixed effect models, smaller and 694 lighter symbols in the background are the raw data. Results of Tukey post-hoc contrasts within 695 each site are indicated above each prediction; shared letters indicate that groups do not 696 significantly differ, while non-overlapping letters indicate that groups significantly differ within 697 each site. Predictions are not plotted for fruit production at Bodega Marine reserve because no 698 fixed factors were in the minimum adequate model. To improve visualization, one outlier that 699 produced 164 fruit was not plotted at the Bodega Marine Reserve. 700 701 702 Figure 1. 703 d e .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 704 Figure 2. 705 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 706 Figure 3. 707 708 Figure 4. 709 710 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 711 Figure 5. 712 713 714 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 715 Figure 6. 716 717 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint

718 Aerts, N., Pereira Mendes, M., & Van Wees, S. C. M. (2021). Multiple levels of crosstalk in 719 hormone networks regulating plant defense. The Plant Journal: For Cell and Molecular 720 Biology, 105(2), 489–504. 721 Agren, J., & Schemske, D. W. (1993). The cost of defense against herbivores: an experimental 722 study of trichome production in Brassica rapa. The American Naturalist, 141(2), 338–350. 723 Arbizu, P. M. (2019). pairwiseAdonis: pairwise multilevel comparison using adonis. 2017. R 724 Package Version 00, 1. 725 Baker, R. L., & Diggle, P. K. (2011). Node-specific branching and heterochronic changes 726 underlie population-level differences in Mimulus guttatus (Phrymaceae) shoot architecture. 727 American Journal of Botany, 98(12), 1924–1934. 728 Baker, R. L., Hileman, L. C., & Diggle, P. K. (2012). Patterns of shoot architecture in locally 729 adapted populations are linked to intraspecific differences in gene regulation. The New 730 Phytologist, 196(1), 271–281. 731 Baldwin, I. T. (1998). Jasmonate-induced responses are costly but benefit plants under attack in 732 native populations. Proceedings of the National Academy of Sciences of the United States of 733 America, 95(14), 8113–8118. 734 Barbour, M. G. (1978). Salt spray as a microenvironmental factor in the distribution of beach 735 plants at point reyes, California. Oecologia, 32(2), 213–224. 736 Bazzaz, F. A., Chiariello, N. R., Coley, P. D., & Pitelka, L. F. (1987). Allocating Resources to 737 Reproduction and Defense. Bioscience, 37(1), 58–67. 738 Boyce, S. G. (1954). The Salt Spray Community. Ecological Monographs, 24(1), 29–67. 739 Briscoe Runquist, R. D., Gorton, A. J., Yoder, J. B., Deacon, N. J., Grossman, J. J., Kothari, S., 740 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Lyons, M. P., Sheth, S. N., Tiffin, P., & Moeller, D. A. (2020). Context Dependence of 741 Local Adaptation to Abiotic and Biotic Environments: A Quantitative and Qualitative 742 Synthesis. The American Naturalist, 195(3), 412–431. 743 Brooks, M. E., Kristensen, K., & Van Benthem, K. J. (2017). glmmTMB balances speed and 744 flexibility among packages for zero-inflated generalized linear mixed modeling. The R. 745 https://www.research-collection.ethz.ch/handle/20.500.11850/239870 746 Campos, M. L., Yoshida, Y., Major, I. T., de Oliveira Ferreira, D., Weraduwage, S. M., 747 Froehlich, J. E., Johnson, B. F., Kramer, D. M., Jander, G., Sharkey, T. D., & Howe, G. A. 748 (2016). Rewiring of jasmonate and phytochrome B signalling uncouples plant growth-749 defense tradeoffs. Nature Communications, 7, 12570. 750 Cipollini, D., Walters, D., & Voelckel, C. (2017). Costs of resistance in plants: From theory to 751 evidence. In Annual Plant Reviews online (pp. 263–307). John Wiley & Sons, Ltd. 752 https://doi.org/10.1002/9781119312994.apr0512 753 Clausen, J., Keck, D. D., & Hiesey, W. M. (1940). Experimental studies on the nature of species. 754 I. Effect of varied environments on western North American plants. Carnegie Institution of 755 Washington Publication No. 520. 756 Crone, E. E. (2001). Is survivorship a better fitness surrogate than fecundity? Evolution; 757 International Journal of Organic Evolution, 55(12), 2611–2614. 758 DeMarche, M. L., Kay, K. M., & Angert, A. L. (2016). The scale of local adaptation in Mimulus 759 guttatus: comparing life history races, ecotypes, and populations. The New Phytologist, 760 211(1), 345–356. 761 Du, J., & Hesp, P. A. (2020). Salt Spray Distribution and Its Impact on Vegetation Zonation on 762 Coastal Dunes: a Review. Estuaries and Coasts, 43(8), 1885–1907. 763 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Eiseman, C. S., Namayandeh, A., Linden, J. V. A. N. D. E. R., & Palmer, M. W. (2023). 764 Metriocnemus erythranthei sp. nov. and Limnophyes viribus sp. nov. (Diptera: 765 Chironomidae: Orthocladiinae): leafminers of monkeyflowers, speedwells, and other 766 herbaceous plants, with new observations on the ecology and habitats of other leaf-mining 767 Chironomidae. Zootaxa, 5249(1), 41–68. 768 Feeny, P. (1976). Plant Apparency and Chemical Defense. In J. W. Wallace & R. L. Mansell 769 (Eds.), Biochemical Interaction Between Plants and Insects (pp. 1–40). Springer US. 770 Fox, J., & Weisberg, S. (2018). An R Companion to Applied Regression. SAGE Publications. 771 Friberg, M., Schwind, C., Guimarães, P. R., Jr, Raguso, R. A., & Thompson, J. N. (2019). 772 Extreme diversification of floral volatiles within and among species of Lithophragma 773 (Saxifragaceae). Proceedings of the National Academy of Sciences of the United States of 774 America, 116(10), 4406–4415. 775 Hahn, P. G., & Maron, J. L. (2016). A Framework for Predicting Intraspecific Variation in Plant 776 Defense. Trends in Ecology & Evolution, 31(8), 646–656. 777 Hall, M. C., Lowry, D. B., & Willis, J. H. (2010). Is local adaptation in Mimulus guttatus caused 778 by trade-offs at individual loci? Molecular Ecology, 19(13), 2739–2753. 779 Hall, M. C., & Willis, J. H. (2006). Divergent selection on flowering time contributes to local 780 adaptation in Mimulus guttatus populations. Evolution; International Journal of Organic 781 Evolution, 60(12), 2466–2477. 782 Hargreaves, A. L., Germain, R. M., Bontrager, M., Persi, J., & Angert, A. L. (2020). Local 783 Adaptation to Biotic Interactions: A Meta-analysis across Latitudes. The American 784 Naturalist, 195(3), 395–411. 785 Hartig, F. (2022). DHARMa: Residual Diagnostics for Hierarchical (Multi-Level/Mixed) 786 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Regression Models (R Package Version 0.4 5). 2022. 787 Havko, N. E., Major, I. T., Jewell, J. B., Attaran, E., Browse, J., & Howe, G. A. (2016). Control 788 of Carbon Assimilation and Partitioning by Jasmonate: An Accounting of Growth–Defense 789 Tradeoffs. Plants, 5(1), 7. 790 Heil, M., & Baldwin, I. T. (2002). Fitness costs of induced resistance: emerging experimental 791 support for a slippery concept. Trends in Plant Science, 7(2), 61–67. 792 Hereford, J. (2009). A quantitative survey of local adaptation and fitness trade-offs. The 793 American Naturalist, 173(5), 579–588. 794 Herms, D. A., & Mattson, W. J. (1992). The Dilemma of Plants: To Grow or Defend. The 795 Quarterly Review of Biology, 67(3), 283–335. 796 Holeski, L. M. (2007). Quantitative trait evolution in Mimulus guttatus (yellow monkeyflower). 797 https://search.proquest.com/openview/391cdb5a832736c89d6509c7e9cb3565/1?pq-798 origsite=gscholar&cbl=18750 799 Holeski, L. M., Chase-Alone, R., & Kelly, J. K. (2010). The genetics of phenotypic plasticity in 800 plant defense: trichome production in Mimulus guttatus. The American Naturalist, 175(4), 801 391–400. 802 Holeski, L. M., Keefover-Ring, K., Bowers, M. D., Harnenz, Z. T., & Lindroth, R. L. (2013). 803 Patterns of phytochemical variation in Mimulus guttatus (yellow monkeyflower). Journal of 804 Chemical Ecology, 39(4), 525–536. 805 Holeski, L. M., Monnahan, P., Koseva, B., McCool, N., Lindroth, R. L., & Kelly, J. K. (2014). A 806 high-resolution genetic map of yellow monkeyflower identifies chemical defense QTLs and 807 recombination rate variation. G3 , 4(5), 813–821. 808 Hou, X., Ding, L., & Yu, H. (2013). Crosstalk between GA and JA signaling mediates plant 809 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint growth and defense. Plant Cell Reports, 32(7), 1067–1074. 810 Hummel, G. M., Schurr, U., Baldwin, I. T., & Walter, A. (2009). Herbivore-induced jasmonic 811 acid bursts in leaves of Nicotiana attenuata mediate short-term reductions in root growth. 812 Plant, Cell & Environment, 32(2), 134–143. 813 Huot, B., Yao, J., Montgomery, B. L., & He, S. Y. (2014). Growth-defense tradeoffs in plants: a 814 balancing act to optimize fitness. Molecular Plant, 7(8), 1267–1287. 815 Iqbal, M., & Ashraf, M. (2013). Gibberellic acid mediated induction of salt tolerance in wheat 816 plants: Growth, ionic partitioning, photosynthesis, yield and hormonal homeostasis. 817 Environmental and Experimental Botany, 86, 76–85. 818 Itoh, M., Fukunaga, K., & Osako, T. (2024). Local adaptation in parapatric and sympatric mosaic 819 coastal habitats through trait divergence of Setaria viridis. The Journal of Ecology, 112(4), 820 784–799. 821 James, M. E., Allsopp, R. N., Groh, J. S., Kaur, A., Wilkinson, M. J., & Ortiz-Barrientos, D. 822 (2023). Uncovering the genetic architecture of parallel evolution. Molecular Ecology, 823 32(20), 5575–5589. 824 Kawecki, T. J., & Ebert, D. (2004). Conceptual issues in local adaptation. Ecology Letters, 7(12), 825 1225–1241. 826 Kazan, K., & Manners, J. M. (2012). JAZ repressors and the orchestration of phytohormone 827 crosstalk. Trends in Plant Science, 17(1), 22–31. 828 Kessler, A., & Baldwin, I. T. (2002). Plant responses to insect herbivory: the emerging molecular 829 analysis. Annual Review of Plant Biology, 53, 299–328. 830 Kooyers, N. J., Blackman, B. K., & Holeski, L. M. (2017). Optimal defense theory explains 831 deviations from latitudinal herbivory defense hypothesis. Ecology, 98(4), 1036–1048. 832 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Kooyers, N. J., Donofrio, A., Blackman, B. K., & Holeski, L. M. (2020). The Genetic 833 Architecture of Plant Defense Trade-offs in a Common Monkeyflower. The Journal of 834 Heredity, 111(4), 333–345. 835 Leimu, R., & Fischer, M. (2008). A meta-analysis of local adaptation in plants. PloS One, 3(12), 836 e4010. 837 Lenth, R., Buerkner, P., Herve, M., Love, J., Riebl, H., & Singmann, H. (2020). emmeans: 838 Estimated marginal means, aka least-squares means (1.5. 2-1)[Computer software]. 839 Liu, J., Wu, Y., Dong, G., Zhu, G., & Zhou, G. (2023). Progress of Research on the Physiology 840 and Molecular Regulation of Sorghum Growth under Salt Stress by Gibberellin. 841 International Journal of Molecular Sciences, 24(7). https://doi.org/10.3390/ijms24076777 842 Liu, X., & Hou, X. (2018). Antagonistic Regulation of ABA and GA in Metabolism and 843 Signaling Pathways. Frontiers in Plant Science, 9, 251. 844 Lowry, D. B., Hall, M. C., Salt, D. E., & Willis, J. H. (2009). Genetic and physiological basis of 845 adaptive salt tolerance divergence between coastal and inland Mimulus guttatus. The New 846 Phytologist, 183(3), 776–788. 847 Lowry, D. B., Popovic, D., Brennan, D. J., & Holeski, L. M. (2019). Mechanisms of a locally 848 adaptive shift in allocation among growth, reproduction, and herbivore resistance in 849 Mimulus guttatus. Evolution; International Journal of Organic Evolution, 73(6), 1168–850 1181. 851 Lowry, D. B., Rockwood, R. C., & Willis, J. H. (2008). Ecological reproductive isolation of 852 coast and inland races of Mimulus guttatus. Evolution; International Journal of Organic 853 Evolution, 62(9), 2196–2214. 854 Lüdecke, D. (2018). Ggeffects: Tidy data frames of marginal effects from regression models. 855 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Journal of Open Source Software, 3(26), 772. 856 Maron, J. L., Baer, K. C., & Angert, A. L. (2014). Disentangling the drivers of 857 context/i1 dependent plant–animal interactions. The Journal of Ecology, 102(6), 1485–1496. 858 Mason, C. M., & Donovan, L. A. (2015). Does investment in leaf defenses drive changes in leaf 859 economic strategy? A focus on whole-plant ontogeny. Oecologia, 177(4), 1053–1066. 860 Monson, R. K., Trowbridge, A. M., Lindroth, R. L., & Lerdau, M. T. (2022). Coordinated 861 resource allocation to plant growth-defense tradeoffs. The New Phytologist, 233(3), 1051–862 1066. 863 Nir, I., Shohat, H., Panizel, I., Olszewski, N., Aharoni, A., & Weiss, D. (2017). The Tomato 864 DELLA Protein PROCERA Acts in Guard Cells to Promote Stomatal Closure. The Plant 865 Cell, 29(12), 3186–3197. 866 Oksanen, J. (2016). Vegan: ecological diversity. R Project. 867 http://mirror.linux.duke.edu/cran/web/packages/vegan/vignettes/diversity-vegan.pdf 868 Pedersen, T. L. (2019). Package “patchwork.” R Package http://CRAN. R-Project. 869 Org/package= Patchwork. Cran. https://cloud.r-870 project.org/web/packages/patchwork/patchwork.pdf 871 Penfield, S. (2008). Temperature perception and signal transduction in plants. The New 872 Phytologist, 179(3), 615–628. 873 Popovic, D., & Lowry, D. B. (2020). Contrasting environmental factors drive local adaptation at 874 opposite ends of an environmental gradient in the yellow monkeyflower (Mimulus 875 guttatus). American Journal of Botany, 107(2), 298–307. 876 Rhoades, D. F. (1979). Evolution of plant chemical defenses against herbivores. Herbivores-877 Their Interaction with Secondary Plant Metabolites, 3–48. 878 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Rodríguez, A. A., Stella, A. M., Storni, M. M., Zulpa, G., & Zaccaro, M. C. (2006). Effects of 879 cyanobacterial extracellular products and gibberellic acid on salinity tolerance in Oryza 880 sativaL. Saline Systems, 2(1), 7. 881 Rotter, M. C., Christie, K., & Holeski, L. M. (2022). Climate and the biotic community structure 882 plant resistance across biogeographic groups of yellow monkeyflower. Ecology and 883 Evolution, 12(11), e9520. 884 Rotter, M. C., Couture, J. J., & Rothwell, E. M. (2018). Evolutionary ecology of plant resistance 885 traits across the herbivore diet spectrum: a test in the model plant Mimulus guttatus. 886 Evolutionary. https://www.evolutionary-ecology.com/abstracts/v19/3151.html 887 Shohat, H., Cheriker, H., Kilambi, H. V., Illouz Eliaz, N., Blum, S., Amsellem, Z., Tarkowská, 888 D., Aharoni, A., Eshed, Y., & Weiss, D. (2021). Inhibition of gibberellin accumulation by 889 water deficiency promotes fast and long-term “drought avoidance” responses in tomato. The 890 New Phytologist, 232(5), 1985–1998. 891 Shohat, H., Eliaz, N. I., & Weiss, D. (2021). Gibberellin in tomato: metabolism, signaling and 892 role in drought responses. Molecular Horticulture, 1(1), 1–12. 893 Smilanich, A. M., Fincher, R. M., & Dyer, L. A. (2016). Does plant apparency matter? Thirty 894 years of data provide limited support but reveal clear patterns of the effects of plant 895 chemistry on herbivores. The New Phytologist, 210(3), 1044–1057. 896 Stamp, N. (2003). Out of the quagmire of plant defense hypotheses. The Quarterly Review of 897 Biology, 78(1), 23–55. 898 Stevens, M. T., Kruger, E. L., & Lindroth, R. L. (2008). Variation in Tolerance to Herbivory Is 899 Mediated by Differences in Biomass Allocation in Aspen. Functional Ecology, 22(1), 40–900 47. 901 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint Stowe, K. A., & Marquis, R. J. (2011). Costs of defense: correlated responses to divergent 902 selection for foliar glucosinolate content in Brassica rapa. Evolutionary Ecology, 25(4), 903 763–775. 904 Strauss, S. Y., Rudgers, J. A., Lau, J. A., & Irwin, R. E. (2002). Direct and ecological costs of 905 resistance to herbivory. Trends in Ecology & Evolution, 17(6), 278–285. 906 Therneau, T. M. (2022). Mixed Effects Cox Models [R package coxme version 2.2-18.1]. 907 https://cran.ms.unimelb.edu.au/web/packages/coxme/ 908 Thompson, J. N. (2005). The Geographic Mosaic of Coevolution. University of Chicago Press. 909 Urban, M. C. (2011). The evolution of species interactions across natural landscapes. Ecology 910 Letters, 14(7), 723–732. 911 Wadgymar, S. M., DeMarche, M. L., Josephs, E. B., Sheth, S. N., & Anderson, J. T. (2022). 912 Local adaptation: Causal agents of selection and adaptive trait divergence. Annual Review 913 of Ecology, Evolution, and Systematics, 53(1), 87–111. 914 Wadgymar, S. M., Lowry, D. B., & Gould, B. A. (2017). Identifying targets and agents of 915 selection: innovative methods to evaluate the processes that contribute to local adaptation. 916

in Ecology and Evolution / British Ecological Society. 917 https://besjournals.onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.12777 918 Wang, X., Chen, X., Wang, Q., Chen, M., Liu, X., Gao, D., Li, D., & Li, L. (2019). MdBZR1 919 and MdBZR1-2like Transcription Factors Improves Salt Tolerance by Regulating 920 Gibberellin Biosynthesis in Apple. Frontiers in Plant Science, 10, 1473. 921 Wang, Y., Gong, X., Liu, W., Kong, L., Si, X., Guo, S., & Sun, J. (2020). Gibberellin mediates 922 spermidine-induced salt tolerance and the expression of GT-3b in cucumber. Plant 923 Physiology and Biochemistry: PPB / Societe Francaise de Physiologie Vegetale, 152, 147–924 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 156. 925 Wickham, H. (2016). Data Analysis. In H. Wickham (Ed.), ggplot2: Elegant Graphics for Data 926 Analysis (pp. 189–201). Springer International Publishing. 927 Wilkinson, M. J., Roda, F., Walter, G. M., James, M. E., Nipper, R., Walsh, J., Allen, S. L., 928 North, H. L., Beveridge, C. A., & Ortiz-Barrientos, D. (2021). Adaptive divergence in shoot 929 gravitropism creates hybrid sterility in an Australian wildflower. Proceedings of the 930 National Academy of Sciences of the United States of America, 118(47). 931 https://doi.org/10.1073/pnas.2004901118 932 Yang, D.-L., Yao, J., Mei, C.-S., Tong, X.-H., Zeng, L.-J., Li, Q., Xiao, L.-T., Sun, T.-P., Li, J., 933 Deng, X.-W., Lee, C. M., Thomashow, M. F., Yang, Y., He, Z., & He, S. Y. (2012). Plant 934 hormone jasmonate prioritizes defense over growth by interfering with gibberellin signaling 935 cascade. Proceedings of the National Academy of Sciences of the United States of America, 936 109(19), E1192–E1200. 937 Yan, Y., Stolz, S., Chételat, A., Reymond, P., Pagni, M., Dubugnon, L., & Farmer, E. E. (2007). 938 A downstream mediator in the growth repression limb of the jasmonate pathway. The Plant 939 Cell, 19(8), 2470–2483. 940 Zambiasi, T., & Lowry, D. B. (2023). A cline within an ecotype of the yellow monkeyflower, 941 Mimulus guttatus. In bioRxiv (p. 2023.07.24.550335). 942 https://doi.org/10.1101/2023.07.24.550335 943 Zhang, Y., & Turner, J. G. (2008). Wound-induced endogenous jasmonates stunt plant growth by 944 inhibiting mitosis. PloS One, 3(11), e3699. 945 Züst, T., & Agrawal, A. A. (2017). Trade-Offs Between Plant Growth and Defense Against 946 Insect Herbivory: An Emerging Mechanistic Synthesis. Annual Review of Plant Biology, 68, 947 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint 513–534. 948 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.23.595619doi: bioRxiv preprint