We assessed changes in the population size, density, and diet composition of wolves inhabiting the Romincka Forest (RF), an area of 480 km² situated along the state border between Poland, Russian Federation (Kaliningrad), and Lithuania. We compared the results of our research in 2020-2021 with data from other projects conducted since 1999. We found that both packs living in RF had transboundary territories. The number of packs was stable over 21 years, the average pack size almost doubled (from 4-4.5 to 7.5-8 wolves per pack), the total wolf number increased 1.8 times, reaching 15-16 wolves, the density increased 1.5 times up to 3.1-3.3 wolves/100 km² in winter 2020/2021. Our analyses of 165 scats revealed that beavers Castor fiber made up 45.6% of food biomass in the wolf diet in 2020, which was 3.4 times more than in 1999-2004 (n=84 scats, 13.4%). Wild ungulates constituted 44.8% of the wolf food biomass in 2020, 1.6 times less than before (71.1%). In our study, among wild ungulates, wolves primarily consumed roe deer Capreolus capreolus (22.6% of food biomass), then wild boars Sus scrofa (13.7%) and red deer Cervus elaphus (5.0%), while moose Alces alces was eaten rarely (0.4%). We also recorded domestic dogs (4.9% of food biomass) and cattle (3.1%). The food niche breadth was wider (B=2.31) than in the earlier period (B=1.84), and the Pianka index showed moderate similarity in food composition between both periods (α=0.816). In November 2022, due to the migration crisis, a 199 km impermeable fence along the state border with Kaliningrad was erected, which blocked access to 48% of the RF area that was regularly used by the resident wolf packs. This may cause wolf numbers to decrease and isolation from the central part of the Baltic wolf population to which they belong, according to our DNA analyses.

Tracking. We tracked wolves by foot or by car, using the regular and dense network of dirt roads, routes, and other linear structures, and the plowed strip of soil along the borderline, across the whole Polish portion of RF, that wolves used for traveling and scent-marking. In snow-free seasons, we found tracks on mud or sand and followed them as far as were visible, usually at distances of 100-300 m, while in winter, snow cover allowed us to follow wolf tracks up to 10 km. Species identification was based on the shape and size of tracks and evidence of animal behavior during scent-marking. Additionally, track identification was verified with genetic analysis of scat and urine samples collected during tracking. In winter, we estimated the number of wolves in the tracked group on snow by counting the number of individual trails when wolves split, which usually happened on road junctions and was associated with intense scent-marking. We measured the length of the footprint of the front paw with claws and the distance between the heels of subsequent footprints in the track to distinguish between adults and pups. To record adult wolves and pups, we used camera traps Browning Spec Ops Advantage and Browning Spec Ops Edge, USA, which were set up to register 30s-long videos with a 1s delay between recordings. Altogether, we set up camera traps in 28 locations, mostly on roads and road junctions, which were defined based on evidence of wolf presence (tracks and scent marks) found during tracking and wolf response to our howl stimulation. Cameras worked for 112 camera days.

Genotyping. To distinguish between wolf family groups in RF, we applied genetic fingerprinting. Because our main goal was to genotype and distinguish between parental pairs and their relatives, which scent-mark the pack territory mostly on roads and junctions, we were looking for non-invasive samples (fresh scats and urine from snow) during wolf tracking on forest roads and trails within the whole forest tract. We put 3-4 cm long fragments of fresh wolf scats into plastic tubes (30 ml) fixed with 96% ethanol and stored at +4°C, while urine samples collected from snow, we mixed with two volumes of 96% ethanol and sodium acetate (100 mM final concentration) and kept at −20°C. All tubes with samples were described with date, GPS coordinates, and, whenever possible, the presumed status (breeder, subadult, male, or female) based on the behaviour of the tracked wolf. DNA isolation was performed in a separate cleanroom to avoid contamination. DNA from scats was isolated either with QIAamp DNA Stool Mini Kit (Qiagen) or Exgene™ Stool DNA Mini kit (GeneAll Biotechnology), while precipitated urine samples were isolated with Exgene™ Tissue SV kit (GeneAll Biotechnology). Genetic fingerprinting was based on 13 polymorphic microsatellite loci: FH2001, FH2010, FH2017, FH2054, FH2087L, FH2088, FH2096, FH2137, FH2140, FH2161, vWF, PEZ17 and CPH5. Additionally, DBX intron 6 and DBY intron 7 were used as sex markers. Details of laboratory analyses are described in the paper of Szewczyk et al. (2019. Sci. Rep.). Furthermore, we assigned the local wolves to the source population, comparing microsatellite genotypes from the study area with reference genotypes from the Central and Eastern European wolf populations (Baltic, Carpathian, and Central European Lowland). From each population, we selected 30 genotypes (from the dataset from Szewczyk et al. 2019, Sci. Rep.) and processed them together with genotypes from RF using STRUCTURE 2.3.4 software. Analyses were performed for a number of clusters (K) ranging from 1 to 10, with 10 iterations per K. The Evanno method applied in STRUCTURE HARVESTER was used to infer the most likely number of clusters.

Wolf diet. The diet composition was assessed by analyzing the content of wolf scats. Scats were collected opportunistically year-round on the dense grid of forest roads and other linear structures spanning the whole Polish RF and during long-distance tracking on snow. Each scat was placed in a paper envelope and described with date and GPS coordinates. Then scats were dried for a minimum of five days at 70°C in a laboratory drier to kill parasites. Socked scats were washed through a 0.5 mm-mesh sieve and then dried. The species eaten by wolves were identified based on hair, bones, teeth, and hooves found in scats using hair or osteological keys and reference material collected during earlier studies on the wolf foraging ecology. In cases where the identification of food items in the scat based on the hair and bones was inconclusive (n=19 samples), we determined the species based on the mtDNA isolated from the hair from each sample. We extracted DNA using Exgene™ Genomic DNA micro kit (GeneAll Biotechnology) with modification of manufacturer protocol with the elongated time of first incubation (until complete dissolution, from 2.5 hours to overnight incubation), with the result of 50 ul of DNA extract. The PCR solution, reaction conditions, and further analyses were conducted as described in Szewczyk et al. (2019, Sci. Rep.) with primers L15995 (Taberlet and Bouvet 1994) and H16498 (Fumagalli et al. 1996) proposed by Pun et al. (2009) for species identification from mixed samples.