Spider communities were sampled in white oak and related oak forests from six Spanish national parks (Fig. 1), namely Picos de Europa (P), Ordesa y Monte Perdido (O), Aigüestortes i Estany de Sant Maurici (A) (hereafter referred to as the northern parks) Fig. 2), Monfragüe (M), Cabañeros (C) and Sierra Nevada (S) (hereafter referred to as the southern parks) (Fig. 3). The chosen parks fulfilled three conditions: (1) they had representative white oak forests, (2) they represented the main biogeographic areas within the Iberian Peninsula (Atlantic, Alpine and Mediterranean) (European Environment Agency 2012) and (3) they covered a broad latitudinal and elevational gradient within the Iberian Peninsula. The selected parks spanned distances ranging from 80 km apart (A to O) to 720 km (S to A) and elevations from 320 m (Monfragüe) to 1786 m (Sierra Nevada). Sampling was conducted between May and June, when the richness and abundance of adult spiders in Mediterranean habitats are at its maximum (Cardoso et al. 2007), in two consecutive years, 2013 for the northern parks and 2014 for the southern parks. Two replicates (plots) were set up in each park, except in Picos de Europa and Cabañeros, where two different types of oak forest were available and hence two replicates were set up per forest type, resulting in a total of 16 plots (northern parks P=4, O=2, A=2; southern parks M=2, C=4 S=2, respectively). Additional details of the sampling plots are available in Table 1.

Figure 1.  

Map of the Iberian Peninsula with the location of the national parks and the plots where the sampling protocol COBRA was applied. For each park, squares denote the number of plots and the oak forest type (colour code labels in the inset). Northern parks are Picos de Europa (P), Ordesa (O), Aigüestortes (A). Southern parks Monfragüe (M), Cabañeros (C), Sierra Nevada (S). See Table 1 for additional information on plots and parks.

a

b

c

d

Figure 2.

Pictures of collection localities

a: P1 plot Monte Robledo, Quercus petraea forest  
b: P4 plot El Canto, Q. faginea forest (detail leaflitter)  
c: O1 plot O Furno, Q. subpyrenaica forest  
d: A2 plot Sola de Boi, Q. pubescens forest  

a

b

c

d

Figure 3.

Pictures of collection localities:

a: M2 plot Fuente del Frances, Quercus faginea forest  
b: C1 plot Valle Brezoso, Q. pirenaica  
c: C3 plot La Quesera, Q. faginea forest  
d: S2 plot Camarate, Q. pyrenaica  

In each plot, a COBRA 50 sampling protocol was conducted, which is specifically designed to collect 50% of the spider diversity in the sampling area in an optimised manner (Cardoso 2009). This protocol consists of using different sampling methods to obtain the maximum possible number of species. Direct sampling (methods that require the presence of the collector and her/ his active participation in the specimen capture) were foliage beating, vegetation sweeping and aerial hand collection. For foliage beating, a 1 m² beating tray and a wooden pole were used to beat tree branches as high as possible. For vegetation sweeping, a round sweep net with a diameter of 58 cm was used to sweep tall plants and bushes, below the collector‘s waist. Aerial hand collection was done through visual inspection and hand-capture (aided by forceps, pooter or brush, if needed) on the vegetation above knee-level. The maximum possible number of spiders was caught and transferred to a vial with ethanol. Each sampling consisted of one hour of continuous collecting by one collector. In two plots (P1, A1), we conducted two additional ground hand collecting samples but focused on specimens present below knee-level.

Indirect sampling (techniques that do not involve the presence of the collector), consisted in the use of pitfall traps, i.e. vessels 7.5 cm in diameter buried in the ground with the rim at the ground level and filled with propylene glycol, which preserved spiders for both morphologic and genetic analyses. A few detergent drops were added to the liquid to break the surface tension and to allow spiders to sink to the bottom of the vessel. Pitfalls were covered with labelled plastic caps, held about 1 cm above the ground by four short wires anchored to the ground, in order to prevent the fall of debris into the trap and propylene glycol dilution or overflow caused by rainwater.

Direct sampling in each plot consisted of 2 hours of diurnal and 2 hours of nocturnal foliage beating, 2 hours of diurnal and 2 hours of nocturnal vegetation sweeping and 4 hours of nocturnal aerial hand collecting, which totals 12 hours of sampling, equating to 12 man-hours of sampling. Indirect samples were uniformly distributed within each plot in groups of 4 pitfalls, set in squares with 5 m sides. The traps were left active during two weeks. For subsequent analyses, each group of 4 contiguous pitfall traps were combined and considered as a single sample, which totals 12 indirect samples (Carvalho et al. 2011). All in all, the study included 388 samples (24 samples per plot x 16 plots + 2 extra ground samples x 2 plots, P1 and A1, respectively).

All adults were identified, when possible, to species level. Amongt a wide spectrum of taxonomic literature, the “Araneae: Spiders of Europe” database was used to identify most of the known species found in the samples (Nentwig et al. 2017). Identifications were made mainly with the use of a ZEISS Stemi 2000 stereomicroscope. Images were taken with a Leica DFC 450 camera attached to a Leica MZ 16A stereomicroscope, using the software Leica Application Suite v4.4. After collection, specimens were stored in 95% ethanol and kept at -20ºC in Falcon vials until these were sequentially sorted and identified (materials from the northern parks were collected and sorted before the materials from the southern parks), from which they were moved to smaller vials of 2 ml and returned to -20ºC, for subsequent genetic analyses.

For each species, we provided the number of male and female specimens identified by plot (see abbreviations in Table 1) and collecting technique, namely foliage beating (beating), vegetation sweeping (sweeping), aerial hand collection and pitfall trapping. Unidentified morphs and putative new species were provisionally labelled using the genus name and a sequential numeration (e.g. Brigittea sp04). Distributions were based on information available in public databases (Nentwig et al. 2017, World Spider Catalog 2018).

DNA barcodes were obtained from all sampled species – five individuals were analysed per morpho-species and per plot when possible, as many species collected without taxonomic targeting are usually found in singletons or doubletons. Legs were used for DNA extraction and the rest of the individual was kept as a voucher, although for small species, the entire specimen was used. In these cases, the extractions were non-destructive (i.e. specimens were not ground up) and specimens were recovered as vouchers after the lyses of soft internal tissues. Total genomic DNA was extracted using the REDExtract-N-Amp™ Tissue PCR Kit Protocol from Sigma-Aldrich, following the manufacturer’s protocol and performed in 96 well-plates. The primers used for amplification are listed in Table 2. The LCOI1490/HCO2198 was the preferred combination, while Nancy was used as a replacement for HCOI2198 and Ron as replacement for LCOI1490 (in that order). For problematic amplifications, we used the internal primers mlCOIintF/jgHCOI2198. The polymerase chain reaction (PCR) was performed in 96-well plates using 8 µl REDExtract-N-Amp™ PCR ReadyMix from Sigma-Aldrich, primers forward and reverse, 4 µl of diluted DNA and ultrapure, distilled water up to a total reaction volume of 20 µl. PCR conditions were as follows: initial denaturing step at 95°C for 5 min, 35 amplification cycles (94°C for 30 s, 45°C for 35 s, 72°C for 45 s) and a final step at 72°C for 5 min. In some cases a Touchdown protocol was used, consisting of 16 cycles of annealing temperature starting at 62°C and decreasing 1°C each cycle and 25 additional cycles of annealing at 46°C. PCR products were cycle-sequenced in both directions at Macrogen Inc. (Seoul, South Korea).

Raw chromatograms were assembled, edited and further manipulated using the software Geneious v7.1.9 (Kearse et al. 2012).

Although DNA barcodes were obtained for all species, we decided to investigate a step further with several families that presented us with cases of incongruence between morphology-based identification and genetic-based identification, namely the Dictynidae, Gnaphosidae, Linyphiidae and Philodromidae. Alignments were obtained by combining all DNA barcodes of focal species (see Results) for each family. We inferred the maximum-likelihood tree for each alignment by finding the best partition scheme first (Kalyaanamoorthy et al. 2017), followed by tree inference using the edge-linked partition model in the IQ-TREE software v1.6.1 (Chernomor et al. 2016, Nguyen et al. 2015). The best tree was then used to delimit putative species using the mPTP algorithm (Kapli et al. 2017). The mPTP method allows identifying species boundaries based on branch lengths obtained from a single-locus, without the need for an ultrametric tree and has been shown to generate more stable outputs than alternative approaches (Blair and Bryson 2017). Genetic distances within and between the clusters identified by mPTP were estimated using the Kimura 2 parameter model (Kimura 1980) in MEGA v.6 (Tamura et al. 2013). One DNA barcode per genetic cluster was further used for automatic identification using BOLD (Ratnasingham and Hebert 2007).

The delimited and identified species were subsequently grouped in four groups, namely "Cosmopolitan", "Palearctic", "Mediterranean" and "Iberian", based on the distribution information available at the World Spider Catalog 2018, further refined with our own species presence data (see Suppl. material 1. We note that species found only in Iberia were considered Iberian and not Mediterranean and the species recorded only in countries of the Mediterranean basin (including north-African countries) were considered Mediterranean and not Palearctic. Percentage of each of the four groups per plot were estimated and visualised using R (R Development Core Team 2017).