The dataset is from a study examining the effects of host plant water stress on the developmental and transcriptomic responses of its specialist Lepidopteran herbivore. The study combines host plant metabolic profiling with development assays and full-transcriptome sequencing of herbivore larvae. First, we profiled metabolic differences between well-watered and water-limited ribwort plantain (Plantago lanceolata) using proton nuclear magnetic resonance spectroscopy (¹H-NMR). Second, we tested how performance of developing Glanville fritillary (Melitaea cinxia) larvae was affected by host plant water limitation experienced at different larval developmental stages. Third, we examined larval gene regulatory responses to water limited host plants by sequencing full transcriptomes of 77 female larvae (RNA seq). Finally, to examine intrapopulation variation in the responses of the larvae, we compared the phenotypic and transcriptomic responses across full-sib families originating from different parts of the metapopulation. In this dataset, we provide data for the P. lanceolata metabolite responses to water limitation and developmental responses of the M. cinxia larvae to feeding on water limited P. lanceolata. The transcriptomic data are available from NCBI's Gene Expression Omnibus, with the accession number GSE159376.

We collected seeds from a natural population in the Åland Islands (60.196° Lat., 20.704° Lon.) and after germination planted 360 plants in 0.75 litre pots (two saplings each). We reared the plants for three months in controlled greenhouse conditions (ca. 40 ml water / pot daily, 15L:9D photoperiod with 26:18 °C temperature cycle) before initiating the water limitation treatment. We exposed 240 plants to a water limitation treatment in which daily watering was reduced by 50% compared to controls (20ml per pot).

Each morning prior to watering (9-10 am), we randomly harvested P. lanceolata leaves from control and water limited plants and cut them into 2.25cm2 pieces, discarding the basal and tip parts of the leaves. We used these pieces to feed larvae during the experiment (see below), and selected a random subset of six pieces from both treatments for metabolomics assays. For the assays, we recorded the fresh biomass of each piece of leaf, snap froze the pools in liquid nitrogen and stored them in -80 °C. We then freeze dried the samples for 48 hours after which we measured their dry weight, estimated relative water content in the sample [i.e. (fresh mass – dry mass) / fresh mass], ground the dried pool samples in dry ice using a sterile pestle and prepared the finely ground tissue powder for proton nuclear magnetic resonance spectrometry (1H-NMR) following the protocol described by Kim et al. (2010). The 1H-NMR spectra of the P. lanceolata pool samples was recorded at the Finnish Biological NMR Center (Institute of Biotechnology, University of Helsinki, http://www.biocenter.helsinki.fi/bi/) using a Bruker Avance III HD NMR spectrometer (Bruker BioSpin, Germany) operated at 1H frequency of 850.4 MHz equipped with a cryogenic probe head. The 1H-NMR spectra were aquired at a 25 °C temperature with a “zgpr” pulse sequence.

After obtaining the spectra, we corrected the baseline, used automatic peak detection, binned the observed peaks into 0.04 ppm bins for all pool samples and normalized them according to the TSP signal variation and dry mass of the sample using MNOVA v.10.0.2 software (Mestrelab research S.L., Spain). We further annotated the peaks within the binned intervals based on signals of pure compounds for Aucubin, Catalpol and Verbascoside or from published characteristic signals (Agudelo-Romero et al., 2014; Gogna et al., 2015; Kim et al., 2010).

M. cinxia larval families, treatments and developmental assays

We created experimental full-sib larval family groups by mating individuals originating from local populations across the natural Åland islands metapopulation. From nine mated females, we picked two larval groups of a minimum of eighty larvae each to enter the experiment. On the day after hatching, we divided the larval groups into four smaller groups of twenty larvae each and placed them on separate petri dishes (9 cm diameter, 1.5 cm deep) lined with filter paper. We then randomly assigned each of the dishes to one of four different treatments mimicking different temporal exposures to drought stressed host plants. In addition to a control treatment, in which we fed the larvae with control reared host plants only, the larvae experienced water limited host plant at different stages during pre-diapause development: (1) at late pre-diapause development during 3rd and 4th larval instars, (2) at early pre-diapause development during 1st and 2nd larval instars, and (3) throughout their pre-diapause development from the 1st to the 5th larval instar. In all treatments, we fed the larvae daily with pieces of host plant leaf tissue corresponding to the treatment. We provided ad libitum food such that – in order to avoid feeding on old leaf tissue with potentially altered phytochemistry – the larvae consumed most but not all of the leaf tissue during the next 24 hours after provisioning.

During the experiment, we recorded development time, body mass at diapause and mortality during development. Once the last larva on a petri dish had entered diapause, we allowed them to spend another four days in normal rearing temperature and photoperiod, after which we measured their body mass and placed them in climate chambers (+5 °C, 95% air humidity) for diapause. We allowed the larvae to diapause for six months, after which we woke them up and recorded overwintering mortality.

The data are accompanied by a README.txt file describing all data fields in all data files.

Note that although larval development time and body mass related data is provided at the individual level, the records do not have individual level IDs, Therefore the development time and diapause body mass data cannot be combined with each other to examine their association within each individual. This is because the larvae spend their pre-diapause development and overwintering in full-sib larval groups making it very challenging to track individual larvae across their development.

Note also that the data also contains some outlier cases we suggest removing before any analyses. Larvae in the control treatment of the first replicate larval group of family F-5 (clutch ID 17) had a mortality of 55%. As the larvae in this group were developing poorly in general, we concluded it to be an outlier case, potentially suffering from a disease or some other unknown agent. For transparency we provide these cases with the data, but recommend excluding this larval group from any further analyses. In all data files regarding M. cinxia larvae, records for this group have been marked as outlier cases with an "x" in the data column "outlier".