Study area and collection of fecal samples
The study was carried out in two mountain areas in northwestern Spain, Natural Park Os Montes do Invernadeiro and its surroundings (Ourense prov.; 42°07′52″N / 7°19′09″O; 880-1700 m.a.s.l.) and Sierra de la Culebra (Zamora prov.; 41°53′54″N / 6°20′01″O; 800-1200 m.a.s.l). Vegetation in these areas is mainly composed by predominant mixed scrub with different species of heath (Erica spp.) and brooms (Genista florida, Cytisus scoparius), and other bushes, and small oak forests (Quercus pyrenaica) with birch (Betula celtiberica) and holly (Ilex aquafolium), often replaced by reforestation of conifers (Pinus pinaster and P. sylvestris). The areas have a high density of wild ungulates (red deer, roe deer, wild boar) and carnivores, specially holding the highest density of wolves in the Iberian Peninsula and throughout Western Europe.
We collected fecal samples from five wolf-breeding groups; four in the Sierra de la Culebra and one in the Natural Park Os Montes do Invernadeiro. In each group, fresh feces of adult wolves were collected monthly, from May 2007 to December 2008, along forest tracks and firebreaks that were frequently transited by wolves. The fecal samples were collected without handling the animals. To discriminate which group the fecal samples belonged to, the pathways prospected to collect the feces were established in each group in the vicinity of the rendezvous sites, since the wolves defend territorial groups and there is no overlap between them [35, 36]. The rendezvous sites were located at the beginning of the study and were located in the center of the territory where there was a great activity of pups and adults (footprints, excrements, bony remains of prey, trampling of vegetation, tracks, etc.), which facilitated the collection of fresh feces and decreased the likelihood of confusion with the excrements of other carnivores. Nevertheless, to discriminate the feces of wolves from those of other species of sympatric carnivores (fox, wildcat and European marten) their size and shape were taken into account, not collecting samples from feces with a diameter of less than 2.5 cm and a length less than 25 cm. Moreover, despite all these precautions, the feces collected were analyzed by molecular techniques to identify the species and sex (see below).
The transects were inspected in an off-road vehicle twice a day, one at dawn and the other at dusk, so the time from deposition to collection of feces was less than 12 h. In addition, during the study only very fresh feces were collected, those that had a mucosal cuticle (it appears only in the newly deposited excrements and dried quickly after the deposition), a strong smell and showed no signs of dehydration. Moreover, wolves show their peaks of greatest activity at dawn and dusk . Thus, the collection of fecal samples at this time ensured that the time elapsed since the deposition was short, thus minimizing the losses of volatile compounds due to exposure to environmental factors .
Feces found were classified into two groups, a) feces deposited with a possible function of marking in intraspecific visual and chemical communication and b) feces deposited by the wolves as simple excretions. Feces were considered to have a marking function when deposited on conspicuous substrates (plants, rocks, trunks, etc.), above ground level, at crossroads and/or over feces of conspecifics (over-marking). We considered that a substrate was conspicuous when it was the most outstanding of all within a 2 m radius circle around the excrement [12, 15, 20, 24]. The rest of the substrates were considered non-conspicuous. In addition, feces were considered as marking cues if they occurred on a substrate >4 cm above ground level [15, 38] and at an intersection where two or more trails crossed . It was considered that there was over-marking or re-marking when the wolves defecated over one or several previous older fecal marks . We collected around 10 g of each fresh excrement and stored it in a portable refrigerator plugged into the lighter of the car and loaded with ice until reaching the laboratory freeze where it was kept at -20ºC until further analyses.
Specific and sexual identification using molecular techniques
In order to reliably verify that the visually identified samples correspond to wolves, we conducted a species identification step consisting of sequencing mitochondrial DNA (mtDNA) control regionon a subsample of the faecal samples collected in the field. These analyses ensured the origin of the fecal samples collected and avoided confusion with feces of other sympatric carnivores.
We collected samples of each excrement in tubes with 96% ethanol and stored them at -20º C until processed. The extraction of DNA from fecal samples was carried out using an extraction kit based on silica membranes and adapted to non-invasive samples (QIAamp DNA Stool Mini Kit, Qiagen), following the manufacturer’s guidelines.
To determine the specific origin of the fecal samples, a 440 bp fragment of the mitochondrial DNA control region was sequenced following the methodology described in Vilá et al. . The experimental part consisted in the amplification of DNA using the PCR technique (Polymerase Chain Reaction) and the use of primers Thr-L 15926 and DL-H 16340  and in its subsequent sequencing through the application of the commercial kit dRhodamine Terminator Cycle Sequencing Ready Reaction (Applied Biosystems), in an automatic sequencer ABI PRISM Model 3130 (Applied Biosystems). The success of the DNA amplification was verified by gel electroforesis. The cleaning and purification of the amplified product was carried out according to the combined method of alkaline phosphatase and exonuclease I (ExoSAP-IT®) developed by Amersham Biosciences, to eliminate the primers and the excess of deoxynucleotides that could interfere in the subsequent sequencing reaction.
Species assignment was made thanks to the comparison of the sequences obtained with reference sequences of dogs and wolves obtained in previous studies [40-45] and with those deposited in the GenBank databases for different mammalian species (http://www.ncbi.nlm.nih.gov/) and using the BLAST 2.0 program (http://www.ncbi.nlm.nih.gov/BLAST/).
To determine the sex of the samples identified as produced by wolves, we used the method described by Seddon , designed specifically for sexual determination in fecal samples. To this end, two specific canine markers were amplified using the PCR technique: the DBX intron6 (249 bp), which identifies the X chromosome in males and females, and the DBY intron7 (118 bp) that identifies the Y chromosome in males. The success of the DNA amplification was verified by the electrophoretic migration of the amplified product in 1.5% agarose gels. We identified as males those samples that presented the bands corresponding to the X and Y chromosomes, and as females those samples that exclusively presented the band corresponding to the X chromosome. As there are several problems associated with the low quantity and quality of DNA extracted from scats, all samples were processed in duplicate. Samples whose identification by agarose gel was doubtful and all female samples were also genotyped with two replicates using an automatic sequencer (ABI PRISM 3130, Applied Biosystems). For the visualization and detection of the fragments corresponding to the X and Y chromosomes, the program GENEMAPPER version 4.0 (Applied Biosystems) was used.
Chemical analyses of volatiles in feces
We transferred a small amount of each fecal sample to a chromatography glass vial to which 250 μl of n-hexane was added (Sigma, capillary GC grade). Each vial was closed with a Teflon-lined stopper before mixing the solution for 1 min using a vortex. Thereafter, the vial was placed in a fridge for 10 min to rest until the solid material that was not dissolved precipitated at the bottom of the vial. We extracted the supernatant clear liquid phase with a glass syringe and transferred it to a clean vial closed with a Teflon-lined stopper. We also made blank control vials using the same procedure, but without adding fecal material, to compare with the wolf samples. Thus, we were able to exclude contaminants from the handling procedure or from the environment and to detect potential impurities in the solvent.
To analyze samples, we used a Finnigan-ThermoQuest Trace 2000 gas chromatograph (GC) fitted with a poly (5% diphenyl/95% dimethylsiloxane) column (Supelco, Equity-5, 30 m length, 0.25 mm ID, 0.25 mm film thickness) and a Finnigan-ThermoQuest Trace 2000 mass spectrometer (MS) as detector. We used helium at 30 cm/s as the carrier gas. We injected 2 µl of each sample in splitless mode with an inlet temperature of 250 ºC. The oven of the GC was programmed so that the temperature was kept initially at 45 ºC for 15 min, and then increased at a rate of 5 ºC/min until a final temperature of 280 ºC, which was kept for 15 min. Ionization by electron impact (70 eV) was carried out at 250 ºC. We did not record mass spectral fragments below m/z = 39.
The initial tentative identification of the volatile compounds in the fecal samples was carried out by comparing the fragmentation patterns (i.e., mass spectra) of the compounds detected in the samples with those available in the NIST/EPA/NIH 2010 mass spectral library. When possible, the identification was confirmed by comparing the spectra and retention times with those obtained under the same conditions of the analysis using authentic standards (from Sigma-Aldrich Chemical Co). Impurities identified in the control vial samples are not reported.
All fecal samples collected were of unknown origin with respect to the individual that had produced. To minimize pseudoreplication and avoid bias in the study due to a small number of different prospective individuals, five wolf breeding groups whose group sizes ranged between 6 and 14 individuals (I. Barja, data unpublished) were followed. The group size was obtained by direct observation of the groups at dusk and dawn. Likewise, the alpha pair of each group is the only one that reproduces, and the rest of the members collaborate in the breeding, providing food for the pups and the female when they are in the den and in the rendezvous sites (cooperative breeding) [36, 47]. Therefore, in the access lanes to these zones, we can often find several fresh excrements belonging to different individuals, thus ensuring that the collection of samples does not distort our results.
The relative amount of each chemical compound was determined as the percentage of the area of its peak in the chromatogram in relation to the total area occupied by all the peaks (TIC area), excluding contaminants. For this, the integration capacity of the peak areas available in the software Xcalibur (Finningan Co.) was used. For statistical analyses, the relative proportions of each compound were transformed following the formula: log[(proportion)/(1- proportion)], to correct the problem of non-independence between proportions .
The software PRIMER V6.1.13  and PERMANOVA + V1.0.3  were used to test for differences between the chemical profiles. We calculated the Euclidean distances between every pair of individual samples and produced a resemblance matrix that was the basis of further analyses. We used permutational multivariate variance analyses (PERMANOVA) based on the Euclidean resemblance matrix, using 999 permutations, to analyse whether chemical profiles of the fecal samples varied between sexes, reproductive status of the individuals (not reproductive vs. reproductive vs. breeding) and in relation to the presumable marking function of feces. Pairwise post-hoc comparisons were made with permutation tests. Differences were further investigated using canonical analyses of principal coordinates (CAP, ).To determine which compounds differed between categories (sex, reproductive condition, marking function), we used the transformed areas of the compounds that appeared in at least five samples to make a principal component analysis (PCA) with a varimax normalized rotation. The extracted principal components (PCs) were used as new variables to compare categories using one-way analyses of variance (ANOVA). Post-hoc multiple comparisons were made using Tukey's tests . We further used discriminant analyses to test whether a given sample could be assigned to a given category based on the compounds which PC scores differed significantly between these categories. Statistical analyses were performed using the software Statistica 7.0 (StatSoft Inc., Tulsa, OK).