Sample collection
Samples of insects for the analyses of NAs were collected in four basins of Sierra Nevada National Park, southern Spain (Supplementary Fig. S1). Sample sites covered different spatial and temporal scales of investigation: three sampling stations across an elevational gradient and two sampling periods in spring and autumn of 2015. Given that all samples were collected from similar environments, the effect of abiotic conditions was not considered crucial for testing NAs in insects. Samples of aquatic insects were collected using a kick sampler (250 µm mesh size) by removing the substrate from at least 20 sample units (total area of 2.5 m2) taken on each station and date and distributed randomly in proportion to the occurrence of major stream habitats (i.e. rapid and slow flow, gravel, sand, zones proximal to and distant from shore and vegetated areas). All samples from each station were pooled and individuals representing six hemimetabolous and six holometabolous taxa sorted specifically for the analysis of nucleic acids (Table 1). We refer to taxa as a generic term to designate a group of one or more populations of organisms that were identified to the lowest taxonomic level possible by eye. Thus, most taxa were identified to the species (Dinocras cephalotes, and Perla marginata) or genus level (Baetis sp., Ecdyonurus sp., Epeorus sp., Rhithrogena sp., Hydropsyche sp., and Rhyacophila sp.), except for Lepidostomatidae, Limnephilidae, Brachycentridae, and Simuliidae that were identified to the family level. Although taxonomic resolution in the identification varied, taxa identified at the species and genus level represented the majority of the samples in this study. In addition, morphologically similar animals were selected for all supraspecific taxa in order to represent similar morphospecies for each taxon. When possible, up to 20 individuals per taxa that covered the full size spectrum available for each taxon were sorted into 10-mL vials containing RNAlater (Ambion Inc., Austin, Texas, USA), and transported inside a cooler to the laboratory. There, all insect samples were stored at − 80 °C until prepared for analysis. Before processing the insects, we measured body length to the nearest half millimetre under a stereoscopic microscope and verified the insect’s identity. In total, 639 individuals of 12 different taxa (six hemimetabolans and six holometabolans) were measured and analysed for NA content.
Nucleic acid analysis
NA analyses largely followed the methods by Wagner et al.13 with a number of recommendations by Gorokhova & Kyle25 and Bullejos et al.26. Analyses were carried out on insect legs and/or heads except for Simuliidae, where entire individuals were analysed. Preliminary analyses using legs and heads for a given individual showed that the coefficient of variation in RNA and DNA content rarely exceeded 5% (Supplementary Table S1). For the calculation of dry weight of insects where legs (one to three) were analysed, the opposite legs and the remaining body parts were separately weighed to estimate total body dry mass (total body weight = legs dry weight * 2 + remaining body parts dry weight). For the estimation of dry weight of insects where heads or the entire body were analysed, body length–weight relationships were specifically developed for each taxon in this study (Supplementary Table S2). Dry-weight was estimated by drying samples to constant weight in preweighed aluminium capsules and reweighing them with a Mettler UMT2 microbalance (± 0.1 µg; Mettler Toledo, Im Langacher, Switzerland).
NAs were measured using a microplate fluorimetric high-range assay Ribo-Green assay (Initrogen, Carlsbad, California, USA) after N-laurylsarcosine extraction and RNase digestion, as described in Gorokhova & Kyle25. We used the following reagents: RiboGreenTM RNA Quantitation Kit (Invitrogen Corporation, Carlsbad, California, USA); RNase DNasefree (working solution 5 mg mL21; Q-biogen, Weston, Massachusetts, USA); N-lauroysarcosine (Sigma-Aldrich, Saint Louis, Missouri, USA); Tris-EDTA buffer (Q-biogene). Fluorescence measurements were performed using a FLUOstar Optima fluorometer (microplate reader, filters: 485 nm for excitation and 520 nm for emission; BMG Labtechnologies, Ortenberg, Germany) and black solid flat-bottom microplates (Greiner Bio-One GmbH, Frickenhausen, Germany). The plate was scanned with a 0.2-s well measurement time, making 10 measurements per well, before and after RNase digestion (30 min under dark conditions at 37 °C). Fluorescence measurements were converted into RNA and DNA concentrations (pg) by using standard curves for RNA (16S and 23S from Escherichia coli, component C of the RiboGreen Kit) and DNA (calf thymus; Sigma-Aldrich), and expressed as a percentage of body dry mass (%RNA and %DNA).
Animal genome size database
To test the generality of our hypothesis that DNA size varied between insect metamorphosis modes across taxa and environments (terrestrial and aquatic), we incorporated the Animal Genome Size Database by Gregory16 in our analysis. The dataset covers a variety of insect groups (including 140 families and 20 orders of hemimetabolous and holometabolous insects) with a representation of most functional feeding groups, life cycles, and trait-based morphologies, comprising a total of 336 hemimetabolous and 999 holometabolous insect records.
Statistical analysis
Testing for differences in NAs between metamorphosis modes, we found that data were not normally distributed (Shapiro–Wilk’s W test) and could not be transformed to fit a normal distribution, so differences in NAs were tested using generalized linear mixed effects models (GLMM). Models included body length and metamorphosis mode as fixed factors, and insect taxa nested within order as random factors to account for variability within taxa subgroups. The significance of the interaction between body length and metamorphosis mode was used to test whether NA allocation to RNA differed during the ontogenetic development of animals. To examine whether insect genome size (C-value) varied between holo- and hemimetabolans using Gregory’s genome size dataset, a GLMM was also used with metamorphosis mode set as a predictor and taxa nested within order as a random variable. Before performing the models, the data were standardized (Deconstand function in R) to provide meaningful estimates of main effects in models with interaction terms27 and the best GLMM was selected according to deviance information criteria28. GLMM analyses were conducted using the ‘glmer’ function in the package ‘lme4’29. Finally, because NA data for taxon subsets were normally distributed after a log-transformation, linear-regression models were used to test the relationship between RNA and body length for each taxon. All statistical analyses were made in R30.
Source: Ecology - nature.com
