Plants invest a substantial fraction of their resources into defence against herbivores, with the highest levels of defence often allocated only to the most valuable tissues. Plants in the genus Erysimum (Brassicaceae) have evolved the ability to produce novel cardenolides, in addition to ancestrally conserved glucosinolates. While these plants co‐express both defences, differences in tissue‐specific expression might represent an effective cost‐saving strategy. Larvae of the glucosinolate‐resistant diamondback moth Plutella xylostella occasionally feed on Erysimum cheiranthoides but tend to avoid younger leaves. Here, we predict that caterpillar feeding preference is shaped by variations in cardenolide levels. Thus, we quantified within‐plant variations in defence and nutritional traits of vegetative or early reproductive plants and performed feeding assays to evaluate the relative importance of cardenolides. In accordance with optimal defence theory (ODT), the youngest leaves contained the most nutrients and had highest levels of cardenolides, glucosinolates and trichomes, with more extreme within‐plant differences found in reproductive plants. Caterpillars consistently avoided the well‐defended youngest leaves, both on whole plants and detached leaf discs. Surprisingly, neither experimental addition (external application) nor removal (CRISPR‐Cas9 knockout) of cardenolides significantly affected caterpillar feeding preference. Physical and chemical defences, including cardenolides, co‐vary within E. cheiranthoides to maximize defence of youngest leaves. While P. xylostella clearly responds to some of these traits, the prominent cardenolide defence appears to lack potency against this specialist herbivore. Nonetheless, the careful regulation and re‐mobilization of cardenolides to younger leaves during plant development suggests an important role in plant functioning.

Description of the data and file structure

Raw data provided as .csv files and complete R code used in the analysis of the data and visualization of results.

01_Whole_plant_choice.csv: Data from two whole-plant choice assays. In the first assay (Assay = 1), caterpillars of Plutella xylostella were left to feed on 4-week or 6-week old plants (Plant_age = W4/W6) for 24 hours, after which leaves of assay plants (Plant = 1-18) were numbered from bottom to top of each plant (Leaf_number_b-t), removed and glued to sheets of paper for scanning. Total leaf area with damage was quantified (Leaf_area_damaged), intact leaf area was estimated by masking damage (Leaf_area_whole), and consumed leaf area was calculated as the difference (Area_consumed). All leaf areas have the unit mm2. The same assay was repeated (Assay = 2) comparing 6-week old wildtype plants (Genotype = WT, Plant = 19-21) to plants with a CRISPR/Cas9 knockout in a key cardenolide gene (Genotype = cyp87a126-1, Plant = 22-24).

02_Whole_plant_performance.csv. Data from a clip cage performance assay on whole plants. Caterpillars of Plutella xylostella were placed in clip cages onto leaves of 4-week or 6-week old plants (Plant_age = W4/W6). Using 9 plants per plant age (Plant = 1-18), caterpillars were attached individually to leaves of four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) on each plant. Individual caterpillars (Caterpillar = 1-72) were pre-weighed (Larval_weight_0h) and weighed again after 24h of feeding (Larval_weight_24h), and absolute change (Weightchange_absolute = Larval_weight_24h – Larval_weight_0h) and relative weight change (Weightchange_relative = Weightchange_absolute / Larval_weight_0h) was calculated. Leaves were removed after the assay and scanned to determine the total leaf area with herbivore damage (Leaf_area_damaged), intact leaf area was estimated by masking damage (Leaf_area_whole), and consumed leaf area (Area_consumed) was calculated as the difference. All leaf areas have the unit mm2, while caterpillar weights are in mg. NA values indicate caterpillars which died during the assay.

03_Leaf_disc_choice.csv. Data from herbivore choice assays using cut leaf discs. Leaf discs of four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) from 6-week-old plants (Plant_age = W6) were placed together into Petri dishes (Replicate = 1-20). Plutella xylostella caterpillars were released in the center of each dish, and after 24h of feeding, leaf discs were scanned to determine area with damage (Disc_area_damaged). Intact leaf disc areas were calculated based on the leaf disc diameter (Disc_area_whole), and consumed leaf area (Area_consumed) was calculated as the difference. All leaf areas have the unit mm2.

04_Leaf_disc_performance.csv. Data from herbivore performance assays using cut leaf discs. Single leaf discs of four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) from 6-week-old plants (Plant_age = W6) were placed into wells of 12-well plates (Plate = 1-4, Position = A1-C4). Individual Plutella xylostella caterpillars (Caterpillar = 1-48) were pre-weighed (Larval_weight_0h) and added to each well. Caterpillars were weighed again after 24h of feeding (Larval_weight_24h) and absolute differences were calculated (Weightchange_absolute = Larval_weight_24h – Larval_weight_0h). Leaf discs were scanned after the assay to determine the total leaf area with damage (Disc_area_damaged). Intact leaf areas were calculated based on the diameter of the leaf disc, or estimated by masking damage when whole leaves were used instead of leaf discs (Disc_area_whole). Consumed leaf area (Area_consumed) was calculated as the difference. All leaf areas have the unit mm2, while caterpillar weights are in mg. NA values indicate caterpillars which died during the assay.

05_Leaf_extract_choice.csv. Data from herbivore choice assays using Erysimum leaf extracts from 6-week-old plants painted onto broccoli leaf discs. Broccoli discs were painted with methanolic leaf extracts of four leaf age classes (Treatment = L1 [youngest] - L4 [oldest]) or with a MeOH control (Treatment = Control) and placed together into Petri dishes (Replicate = 1-12). Individual Plutella xylostella caterpillars were released in the center of each dish, and after 24h of feeding (Time = 24h), leaf discs were scanned to determine the total leaf area with damage (Disc_area_damaged). Intact leaf area was estimated by masking damage (Disc_area_whole), and consumed leaf area (Area_consumed) was calculated as the difference. Caterpillars were then moved to second set of Petri dishes with fresh painted leaf discs and left to feed for an additional 24h (Time = 48h), after which consumed leaf area was quantified again. All leaf areas have the unit mm2.

06_Leaf_extract_performance.csv. Data from herbivore performance assays using Erysimum leaf extracts painted onto broccoli leaf discs. Broccoli discs were painted with methanolic leaf extracts of four leaf age classes (Treatment = L1 [youngest] - L4 [oldest]) or with a MeOH control (Treatment = Control), and placed into wells of 12-well plates (Plate = 1-5, Position = A1-C4). In a first time interval (Time_interval = 1), individual caterpillars of Plutella xylostella were pre-weighed (Larval_weight_0h), added to each well, and weighed again after 24h of feeding (Larval_weight_24h). Caterpillars were then transferred to a second set of plates with identical, freshly painted broccoli leaf discs and left to feed for a second time interval (Time_interval = 2). Both set of leaf discs were scanned to determine the total leaf area with damage (Disc_area_damaged), intact leaf areas were calculated based on the diameter of the leaf disc (Disc_area_whole), and consumed leaf area (Area_consumed) was calculated as the difference. All leaf areas have the unit mm2, while caterpillar weights are in mg. NA values indicate caterpillars which died during the assay.

07_Pure_compound_choice.csv. Data from herbivore choice assays using pure glucosinolate or cardenolide compounds (Compound_class = Glucosinolate/Cardenolide, Compound = Sinigrin/Glucocheirolin/Digitoxin/Convallatoxin) painted onto broccoli leaf discs. Compounds were applied at concentrations corresponding to concentrations found in leaves of four leaf age classes from 6-week-old plants (Treatment = L1 [youngest] - L4 [oldest]). Broccoli discs with four concentrations were placed together into Petri dishes (Replicate = 1-80), and single Plutella xylostella caterpillars were released in the center of each dish. After 24h of feeding, leaf discs were scanned to determine the total leaf area with damage (Disc_area_damaged), intact leaf areas were calculated based on the diameter of the leaf disc (Disc_area_whole), and consumed leaf area (Area_consumed) was calculated as the difference. All leaf areas have the unit mm2. NA values indicate caterpillars which died during the assay.

08_Leaf_nutrients.csv. Nutrient concentrations of leaves from four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) of six 6-week-old plants (Plant = 1-6). Concentrations of soluble glucose, fructose and sucrose were determined enzymatically. Starch was first digested and then quantified as glucose equivalents. Total hydrolysable protein was extracted from leaves using NaOH, precipitated, and then quantified using Bradford Dye. All nutrients have the unit of ug per mg dry weight.

09_Trichome_quantification.csv. Trichome density quantified from high-resolution photographs of leaves from four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) of three 6-week-old plants (Plant = 1-3). Both sides (Side = Adaxial/Abaxial) were photographed and quantified for each leaf. Trichomes were counted (Trichome_count) in areas of variable size chosen for each photograph (Count_area [mm2]) to calculate the number of trichomes per mm2 (Trichome_density).

10_Defence_compounds_qtof.csv. Quantified mass peak areas from the analysis of leaf extracts by UPLC-QTOF. Extractions were carried out in two separate batches (Batch = B1/B2) using 4-week or 6-week old plants (Plant_age = W4/W6). On six plants per plant age (Plant = 1-12, PlantID = P01-P12), leaves of four leaf age classes (Leaf_age = L1 [youngest] - L4 [oldest]) were harvested. On each harvested leaf (LeafID), small leaf discs (Disc, DiscID) were cut out and extracted separately as technical replicates. Leaf extracts (Sample_type = Analyte) were analyzed on a UPLC-QTOF instrument in a randomized order (Sample_order). At regular intervals, a quality control sample (Sample_type = QC) consisting of pooled plant extracts was inserted into the analysis sequence to test and correct for shift in signal intensity over time. For each leaf extract, glucosinolates (Compound_type = Glucosinolate) and cardenolides (Compound_type = Cardenolides) were identified by separate methods using negative and positive ionization, respectively. For cardenolides, 15 compounds (Compound, Compound_number = c01-c15) and two internal standards (Compound_number = IS1/IS2) were quantified, while for glucosinolates, two compounds (Compound, Compound_number = c16-c17) and one internal standard (Compound_number = IS3) were quantified. For each compound, the molar mass (M), the quantified ion (Quant_ion), and the integrated peak area (Area_MS) is provided. NA values in integrated peak areas indicate mass signals below a threshold of detection.

11_Defence_compounds_UV.csv. Quantified mass and UV peak areas from the repeat analysis of leaf extracts by HPLC-UV-MS. For cardenolides (Compound_type = Cardenolides), five randomly selected extracts (Sample_type = Analyte, DiscID) were re-analysed together with a concentration series of the cardenolide standard digitoxin (Sample_type = Standard, Concentration_ugml = 20/100/200/300). For 14 plant cardenolides (Compound, Compound_number = c01-c14), both UV (Area_UV) and mass signals (Area_MS) could be quantified. For glucosinolates (Compound_type = Glucosinolates), a larger set of extracts (Sample_type = Analyte, DiscID) was re-analysed together with a concentration series of the glucosinolate standard sinigrin (Sample_type = Standard, Concentration_ugml = 5/25/50/100/200). For each plant glucosinolate (Compound, Compound_number = c16-c17), both UV (Area_UV) and mass signals (Area_MS) could be quantified. Different glucosinolate compounds are known to vary in UV absorption, requiring the use of published response factors (UV_response_factor = 0.9-1.2) for accurate quantification. For cardenolides, equimolar absorbtion for each compound was assumed (UV_response_factor = 1).

R script_herbivores.R. Complete R code to analyze herbivore choice and herbivore performance in response to leaf and plant age, using whole plant assays, cut leaf disc assays, extract assays, and pure compound assays.

R script_plant traits.R. Complete R code to perform absolute quantification of glucosinolates and cardenolides from mass spectrometry data, and to analyze defense, nutrient, and trichome data as a function of leaf and plant age.

Code/Software

All R scripts were run in version 4.3.1

Within‐plant variation in chemical defence of Erysimum cheiranthoides does not explain Plutella xylostella feeding preference

Data files

Abstract

Description of the data and file structure

Code/Software

Within‐plant variation in chemical defence of Erysimum cheiranthoides does not explain Plutella xylostella feeding preference

Data files

Abstract

README: Within-plant variation in chemical defence of Erysimum cheiranthoides does not explain Plutella xylostella feeding preference

Description of the data and file structure

Code/Software