Despite multiple ecological and evolutionary hypotheses that predict patterns of phenotypic relationships between plant growth, reproduction, and constitutive and/or induced resistance to herbivores, these hypotheses do not make any predictions about the underlying molecular genetic mechanisms that mediate these relationships. We investigated how divergent plant life-history strategies in the yellow monkeyflower and a life-history altering locus, DIV1 influence plasticity of phytochemical herbivory resistance traits in response to attack by two herbivore species with different diet breadth. Life-history strategy (annual vs. perennial) and the DIV1 locus significantly influenced levels of constitutive herbivory resistance, as well as resistance induction following both generalist and specialist herbivory. Perennial plants had higher total levels of univariate constitutive and induced defense than annuals, regardless of herbivore type. Annuals induced less in response to generalist herbivory than did perennials, while induction response was equivalent across the ecotypes for specialist herbivory. The effects of the DIV1 locus on levels of constitutive and induced defense were dependent on genetic background, the annual versus perennial haplotype of DIV1, and herbivore identity. The patterns of univariate induction due to DIV1 were non-additive and did not always match expectations based on patterns of divergence for annual/perennial parents. For example, perennial plants had higher levels of constitutive and induced defense than did annuals, but when the annual DIV1 was present in the perennial genetic background induction response to herbivory was higher than for the perennial parent lines. Patterns for multivariate defense arsenals generally echoed those of univariate, with annual and perennial monkeyflowers and those with alternative versions of DIV1 differing significantly in constitutive and induced resistance. Like univariate resistance, induced multivariate defense arsenals were affected by herbivore identity. Our results highlight the complexity of the genetic mechanisms underlying plastic response to herbivory. While a genetic locus underlying substantial phenotypic variation in life-history strategy and constitutive defense also influences defense plasticity, the induction response also depends on genetic background. This result demonstrates the potential for some degree of evolutionary independence between constitutive and induced defense, or induced defense and life-history strategy, in monkeyflowers.

Study System 

Yellow monkeyflower (Erythranthe guttata (DC), GL Nelsom; synonym: Mimulus guttatus DC, Phrymaceae) is a forb that occurs in ephemeral and perennial water sources across western North America. This species produces phenylpropanoid glycosides (PPGs), secondary metabolites which can have negative impacts on a variety of herbivores (Holeski et al. 2013; Keefover-Ring et al. 2014; Rotter et. al. 2018). Yellow monkeyflower is a model system for the study of ecology and evolutionary genetics due in part to the great deal of morphological, genetic, and life-history variation across its range (Hall and Willis 2006; Wu et al. 2008; Flagel et al. 2014; Friedman et al. 2014; Monnahan and Kelly 2017; Kooyers et al. 2023). The plants used in this study are derived from natural populations in Northern California that differ in life-history strategy across a seasonal soil-water availability gradient (Lowry and Willis 2010). Populations growing near the coast live in year-round seeps and are perennials that invest heavily in vegetative growth, including lateral stolons, and are slow to flower. In contrast, populations growing further inland live in ephemeral seeps which dry up each year with the onset of regular summer drought. These plants are short-lived annuals that invest little in vegetative growth and reproduce quickly.

Herbivores

To examine differences in induction due to different herbivores, we selected two caterpillar species that feed upon natural populations of M. guttatus in Northern California where our source plant populations occur.  Common buckeye, Junonia coenia (Nymphalidae), specializes on plants that contain iridoid glycosides or PPGs and can sequester both (Bowers and Collinge 1992; Rothwell, Keefover-Ring, and Holeski, unpublished data). Cabbage looper, Trichoplusia ni (Noctuidae), is a common agricultural pest and extreme dietary generalist that feeds on wild species across a broad range of plant families.

Plant materials

To test the impact of the DIV1 locus on the induction of PPGs, we utilized near-isogenic lines (NILs) where annual and perennial DIV1 ecotypes were reciprocally introgressed into the opposite genetic background. These lines were created using an inland annual population (LMC) and a coastal perennial population (SWB) that occur in habitats that differ dramatically in soil-water availability during summer months (Lowry and Willis 2010). As described in more detail by Lowry and Willis (2010), 3 lines from each ecotype were crossed to create heterozygous F1s. These were then backcrossed to the parental lines of each ecotype. After genotyping using two markers outside of DIV1 and one within DIV1, heterozygous individuals were backcrossed again to both parental ecotype lines. After 3-4 generations of backcrosses, plants that were selected that were homozygous for the annual or perennial orientation of DIV1. This resulted in 4 lines, one with the annual genetic background and the annual orientation of DIV1 (Annual control), one with the annual genetic background and the perennial orientation of DIV1 (Annual with perennial DIV1), one with the perennial background and the perennial orientation of DIV1 (Perennial control), and one with the perennial background and the annual orientation of DIV1 (Perennial with annual DIV1). These lines were then self-fertilized 4 times. This crossing design is illustrated in Figure S1. Annual and perennial parent lines (Annual parent and Perennial parent) were also self-fertilized. The control lines and lines with the DIV1 haplotype that have the same genetic background (e.g., annual control versus annual with perennial DIV1) are expected to be, on average, equivalent across all regions of the genome except at the DIV1 locus. Note, however, that some genes from the alternative ecotype are also introgressed, at random, across the genome of all of these individuals. For example, both the annual control lines and the annual with perennial DIV1 lines have some perennial genes introgressed.

Induction Experiment

We used caterpillar herbivores to experimentally induce changes in PPGs in the four NILs and in the annual and perennial parent lines. In the winter of 2021, we cold-stratified seeds in the dark at 4°C on soil for one week. We then transferred the seeds to a grow room, under full-spectrum T5-grow lights, at 21°C. We germinated seeds under humidity domes in the grow room under 16-hr day-length for one week. We then transplanted all seedlings to individual 4" pots and continued to grow them at 21°C with 13-hr day-length, randomizing their position within the grow racks once a week. The 13-hour day length is similar to the early growth conditions for the parent populations in their natural habitat. We bottom-watered the plants once a day for 30 minutes and fertilized with a 10-30-20 fertilizer (Peters’ Professional Base Formulation) once a week. We used two soil-mixes; for germination, we used 65% peat moss, 20% perlite, 15% vermiculite, 3 tablespoons dolomite lime, 3 tablespoons wetting agent, and 3 tablespoons silica; for the rest of the experiment, we used 55% peat moss, 30% bark, 15% perlite, 3 tablespoons dolomite lime, 3 tablespoons wetting agent, and 3 tablespoons silica.

We tested induction in the lines discussed above (annual parent, annual control, annual with perennial DIV1, perennial parent, perennial control, perennial with annual DIV1), factorially crossed with two caterpillar species (buckeyes and cabbage loopers).  These two variables resulted in 12 unique groups. For each group, we assigned 20 plants to induction treatment (feeding by caterpillars; 20 plants x 2 caterpillar species) and 10 to the control group (no herbivory; 10 plants x 2 caterpillar species). The cabbage looper and buckeye treatments (and the timing of affiliated plants) were offset by one month.

At day 35, we applied six neonate caterpillars to each treatment plant, three to one leaf at the 2nd node and three to one leaf at the 3rd node. To keep caterpillars in place, we used green mesh bags with drawstrings closed around the feeding leaves. As added protection for the leaves that we intended to sample for PPG concentration, in case of any caterpillar escape, we also placed these bags around the opposite 2nd-node leaf from the 2nd-node feeding leaf. The control plants had the same three leaves bagged, but without any caterpillars. For 10 days, we checked these plants daily, replacing any deceased caterpillars (this occurred six times in the first three days and then not after) with newly-hatched neonates. On day 10 of the experiment, we removed all caterpillars and mesh bags from all plants. We then used a hole punch to apply additional leaf-area removal to any herbivory treatment leaves that did not reach 50% removal so that the proportion of tissue removed was standardized. On day 11, we sampled both the opposite 2nd-node leaf and the 4th-node leaf on the same side of the plant as the 2nd node feeding leaf for chemical analysis. We flash froze leaves with dry-ice and then lyophilized, after which we stored them in a freezer at -20°C until processing for chemical analysis.

Quantification of PPGs

To determine the PPG concentrations in sampled leaves, we ground, extracted, and prepped extract aliquots as described in Holeski et al. (2013). We then used high-performance liquid chromatography (HPLC) to quantify PPGs. The HPLC method is described in Kooyers et al. (2017) and was run on an Agilent 1260 HPLC with a diode array detector and Poroshell 120 EC-C18 analytical column [4.6 × 250 mm, 2.7 μm particle size]; Agilent Technologies). We measured six different PPGs: Calceolarioside A, conandroside, verbascoside, mimuloside, unknown PPG 10 and unknown PPG 16 (Keefover-Ring et al. 2014; Kooyers et al. 2017). We calculated concentrations of PPGs as verbascoside equivalents, using a standard verbascoside solution (Santa Cruz Biotechnology, Dallas, Texas), as described in Holeski et al. (2013, 2014).