Lepidoptera

Single linkage clustering is performed on the aligned sequence data. This approach ordinarily requires the generation of a distance matrix for all pairs of sequences followed by a clustering step where sequences are grouped based on a pre-selected distance threshold [45] (link), [46] . RESL performs distance calculations and clustering concurrently, employing the transitive property to avoid distance determinations for sequences that are certain to possess a divergence above the threshold. This strategy is implemented by flushing all clusters to disk, and retaining one or more representative sequences, depending on the diameter of the cluster, for each cluster and inter-cluster distance statistics in active memory, excepting those clusters whose members show high variability (max intra-cluster distance >2.2%, see below). The sequence divergence between each new sequence and the representative(s) of all existing clusters is then calculated. If its distance to any existing cluster is more than twice the threshold [>4.4%], it is recognized as the founder of a new cluster. If, on the other hand, it shows lower divergence, all members of the closest cluster(s) are retrieved from disk to enable more detailed analysis of sequence variation. This approach considerably reduces computational requirements without compromising accuracy, and analysis is further expedited by moving clusters to disk when they have seen no activity ( =  gained new members) for a number of cycles.
The implementation of single linkage clustering requires the selection of a threshold parameter, t, which represents the level of sequence divergence for the designation of OTUs. Early work [13] suggested that a threshold value of 2% was effective because most specimens showing more than this level of divergence represented different species, while those with less divergence were usually conspecific. However, this issue was examined in more detail by inspecting the patterning of OTU recovery with variance in the distance threshold for eight datasets (Table 1). Sixty single linkage cluster analyses were generated for each dataset by stepping the distance threshold parameter by an increment of 0.1% across the range from 0.1%–6.0%. The OTUs recovered at each threshold were subsequently evaluated for their concordance with recognized species boundaries (Figure 2). These analyses revealed that maximal concordance was achieved by thresholds that varied from a low of t = 0.7% (in North American birds) to a high of t = 1.8% (in Bavarian moths). It also showed that performance, as measured by the number of correctly recognized species, dropped steeply when the threshold deviated on either side of optimality. Thresholds higher than optimal inflated the number of cases where members of different species were merged in a single OTU, while thresholds lower than the optimal value increased the cases where members of what are thought by current taxonomy to be a single species were split into two or more OTUs. Based on these analyses, a threshold (t) of 2.2% was adopted as it represents the upper 99% confidence limit for the optimal thresholds in the eight test datasets SD  = 0.40). Its adoption will lead to the merger of some distinct clusters, but such cases are addressed in the third step of the analysis.

DNA was extracted using DNeasy Blood and Tissue kits (Qiagen, Hilden, Germany) as
well as a BioSprint 96 extraction robot (Qiagen). Whole individuals were first
soaked overnight in the extraction buffer with proteinase K at 56°C, leaving
the gut and chitinous body parts intact. This treatment preserves the
morphological characteristics of the individuals, which can be easily mounted
for microscope identification. Extracted DNA, individuals and photographs are
deposited at the Museum of Zoology, Lausanne, Switzerland. We amplified a 658-bp
fragment of mitochondrial protein-coding cytochrome c oxidase subunit I
(cox1), extensively used in species identification (e.g.
DNA barcoding) and delimitation, using LCO1490 and HCO2198 primers [48] (link). We also
amplified ca. 540 bp of nuclear protein-coding
phosphoenolpyruvate carboxykinase (PEPCK) using newly designed primers Flv13
(5′-CTAACAGCACCAACCCCATT) and Rlv45 (5′-ACCTTGTGCTCKGCTGCT). Flv13 and
the newly designed Rlv4 (5′-CTCATTGCTGCTCCAACAAA) PEPCK primer were used to
amplify an individual of Cinygmula (Heptageniidae) as an
outgroup. These PEPCK primers were designed from sequences first obtained using
19.5 dF and 22.5 drc primers for Lepidoptera [49] (link). Polymerase Chain
Reaction (PCR) was conducted with a denaturation temperature of 94°C for 30
sec, an annealing temperature of 48°C for 30 sec (cox1) or
ranging between 58°C and 62°C for 30 sec (PEPCK), and an elongation
temperature of 72°C for one min for a total of 40 cycles, followed by a
final extension for 10 min at 72°C.
All PCR products were visualized after agarose gel electrophoresis to verify
amplicon size and detect possible contamination using negative controls. PCR
products were purified using QIAquick PCR purification kits (Qiagen), and
cycle-sequenced in both directions using BigDye v. 3.1 (Applied Biosystems,
Foster City, CA). Sequences were analyzed using an ABI 3100 capillary sequencer
(Applied Biosystems) at the Center for Integrative Genomics (CIG) at the
University of Lausanne. Forward and reverse sequencing reads were assembled and
edited using CodonCode Aligner v. 3.0.1 (CodonCode Corporation, Dedham, MA). The
PEPCK heterozygous sites, typically identified as double peaks within the
chromatograms, were coded according to the IUPAC code. Initial alignments were
performed using ClustalW [50] (link) as implemented in Jalview v. 2.4 [51] (link). Amino acid
translation was then used to distinguish between coding (exon) and non-coding
(intron) PEPCK regions, with intron boundaries identified using the GT-AG rule.
A subsequent alignment of the PEPCK intron section was done using MAFFT v. 5
[52] (link) in
Jalview.

We assembled a taxon set of 771 species across 94 out of 109 recognized extant families (sensu Huber¹⁸ , with modifications by Chen et al.^{79 (link)}, Pilgrim et al.^{80 (link)}, and Sann et al.^{81 (link)}), belonging to all 22 recognized superfamilies within the Hymenoptera^{18 ,80 (link)}, and six non-hymenopteran outgroups. Our taxon sampling aimed for the representation of major lineages within families while sampling across the respective root nodes on the family level, covering between 0.06–50% (=1–150 representatives) of the described species diversity. While we generated UCE sequence data de novo for most taxa, some sequences have already been published in other studies by some of us: 126 aculeate wasps^{38 (link),82 (link),83} , 25 chalcidoids^{84 –86 (link)}, 76 cynipoids^{87 (link)}, 26 Ichneumonidae^{88 (link)–90 (link)} and 142 Braconidae^{91 (link)}. We further included six representatives of other insect orders as outgroups by mining UCEs in silico from published genomes: Coleoptera (Agrilus planipennis), Diptera (Aedes albopictus), Lepidoptera (Papilio glaucus), Hemiptera (Homalodisca vitripennis), Psocodea (Pediculus humanus corporis), and Blattodea (Blattella germanica). Supplementary Data 1 list voucher information and NCBI accession numbers for all sequences, while more detailed specimen data is provided for sequences newly released in this article. All specimens were collected with the required permits and in accordance with local regulations at the time of their collection, and vouchers have been deposited in major collections.

This study has complied with relevant institutional, national, and international guidelines and legislation. This study does not contain any studies with human participants or animals performed by any of the authors. A laboratory strain of S. littoralis was reared on castor bean leaves, Ricinus communis L., under constant conditions of 27 ± 2 °C and 65 ± 5% relative humidity (RH), in the Insect Physiology Laboratory, Department of Applied Entomology and Zoology, Faculty of Agriculture, Alexandria University, Egypt. Moths were provided with Nerium oleander L. leaves for egg laying. Moreover, as the field strain, egg masses of S. littoralis were collected from cotton fields at El-Beheira Governorate, Egypt, and maintained in the laboratory under the aforementioned conditions.