Study sites and water sampling
We sampled altogether 120 lakes in triplicates (3 samples per lake, referred here on as ‘replicates’) from the most northernmost Finnish Lapland, except for one lake from which only one sample was taken (available filters for sampling had been used up at this point) (Fig. 1). One lake was sampled in both 2021 and 2022. Out of the total of 120 lakes, 50 lakes were sampled during 28.6.-1.7.2021 and 70 lakes during 18.6.-21.6.2022. The selection of the sampled lakes was based on historical breeding data of LWfG in Finnish Lapland from the time the species was still breeding there (see Fig. 1). The sampling covered a large area of the previous breeding grounds to ensure broad geographical sampling coverage.
Water samples were collected in 500 ml sterile water sampling bottles (VWR). A helicopter was used for sampling trips and water sample transportation, because the potential breeding sites of the LWfG are in remote roadless wilderness areas. Between 6 to 19 lakes were sampled per day, the water samples were kept cold by cooling blocks. Filtering took place after the landing of the helicopter for the day. A field negative control was taken on each day of sampling (n = 4 in 2021 and 2022). The negative control consisted of sterile water transported to the field in sterile bottles, following the same transportation process as the samples in the helicopter, and it was filtered the same manner as the samples. These field negative controls ensured that the equipment transported to the field were not contaminated. Disposable gloves were used when collecting water samples, with a new pair of gloves being changed between each replicate and each lake.
Sampled water was collected using a sterile 50 ml syringe with a luer-lok tip (VWR) and filtered through Sterivex-HV Pressure Filter Units with 0.45 µm pore size and Male nipple outlet (Merck/Millipore). Water was passed through the Sterivex Filter Unit until the filter become clogged (< 500 ml of water). The residual water from the Sterivex Filter Unit was removed by injecting air to the Filter Unit using the same 50 ml syringe, after which the outlet was plugged with a mouldable silicone earplug (from local pharmacy; sterilized with 10% bleach and UV-radiation).
To store the filters, 2.5 ml of absolute ethanol (Spens et al. 2016) was passed through the Sterivex Filter Unit using 3 ml sterile syringe with a luer-lok tip (VWR). The inlet was sealed with a Cole-Parmer Animal Free Male Luer Lock Plug (Cole-Parmer), and a piece of parafilm was used to seal both the inlet and outlet ends. The Filter Units prepared this manner were placed in Minigrip bags for storage and transportation. The filters were stored in 4 ⁰C prior to DNA extraction. Freezing was avoided at this point as freeze-thaw cycles could be detrimental to DNA. All equipment used in the field were either sterile or had been sterilized with 10% bleach and rinsed with deionized water and 70% ethanol. Table surfaces used for filtering of the samples were wiped with 10% bleach and rinsed with deionized water and 70% ethanol.
To control for the methods, we also collected samples from lakes or puddles where LWfG was visually observed. These samples were either from a spring staging site in Northern Ostrobothnia, Finland (n = 1; 2021) or from lakes at a breeding site in Norway (n = 4; 2022) (Fig. 1). These samples serve as “positive controls” since the presence of the target species was known in advance. The sample from the spring staging site was collected from several puddles in a field in which LWfGs had been feeding earlier that day. It was filtered with one Sterivex-filter. Sampling was conducted after the geese had left the feeding site to a roosting site, ensuring that no geese were disturbed during sampling. The Norwegian samples were collected in triplicates (n = 2), duplicates (n = 2) or once per lake (n = 1) depending on the presence of the LWfGs, as the goal was not to disturb the breeding geese. The sampling process for the positive controls was similar to that used for the other samples, except Sterivex-GP Pressure Filter Units with 0.22 µm pore size were used and the water was collected directly with a 50 ml syringe from the water body, rather than being collected in a bottle. Additionally, a field negative control (sterile water transported to field) was used.
DNA extraction
The Filter Units were wiped with 10% bleach on the outside before DNA extraction. The storage ethanol was removed from the Filter Unit by opening both the outlet and the inlet end and pushing air through the Filter Unit with a sterile 3 ml syringe while the outlet was placed within an Eppendorf tube. The emptied Filter Unit was cut open from the outlet end using washed and flame-sterilized PVC pipe cutters (obtained from a local hardware store) following the “Open Sterivex”- method by Cruaud et al. (2017). The filter was cut into small pieces on a petri dish using a sterile single-use scalpel and flame-sterilized tweezers, as described in Cruaud et al. (2017). The ethanol was allowed to evaporate, and the filter pieces were placed into sterile 2 ml screw cap microcentrifuge tubes.
DNA was extracted using the E.Z.N.A. Tissue DNA Kit (Omega Bio-Tek) with the tissue protocol with modifications from Spens et al. (2016) as follows. We added 720 µl of TL buffer and 80 µl of proteinase K and incubated the samples in a rocking platform at 55 ºC within a hybridization oven overnight. The samples were vortexed for 15 s and centrifuged at 6000 g for 5 min. The amount of liquid was measured (650 µl) and pipetted into a 2 ml screw cap microcentrifuge tube. An equal amount of buffer BL (650 µl) was added, and the samples were vortexed and incubated in 70 ºC for 10 min. Six hundred fifty microliters of ice-cold ethanol was added, samples were vortexed and 650 µl of the mixture was pipetted into HiBind DNA Mini Column. The column was then centrifuged for 1 min at 6,000 g and the filtrate was discarded. This step was repeated until all liquid had passed through the column. Subsequently, the Kit’s manual was followed, with the exception that the centrifugal speeds were 6,000 g, and the empty column was centrifuged at 16,200 g for 4 min. Moreover, the DNA was eluted in 50 µl of 70 ºC of Elution Buffer, the column was let stand in room temperature for 5 min and centrifuged in 17,000 g for 1 min. This elution process was repeated a second time, yielding a total of 100 µl of DNA. From this, 20 µl of the DNA was aliquoted into another tube. and only this smaller aliquot was subjected to freeze-thaw cycles while the rest of the DNA was kept in -20 ºC, because repeated cycles can be harmful for DNA preservation. The positive controls (samples from the spring staging site and Norwegian breeding site) and field negative controls were extracted similarly.
All DNA extractions were performed in a dedicated room where no PCR products are handled, specifically designed for samples containing low-quantity and -quality DNA. Filter tips were used in all DNA extraction procedures. This room is physically separated from other molecular biology laboratory areas and the was subjected to UV-light treatment before DNA extractions. Extraction negative controls were also processed, containing only the DNA extraction reagents, to ensure that the reagents used were not contaminated.
PCR and sequencing
Preparations of the PCR reactions were similarly conducted in the room dedicated to low-quantity and -quality DNA samples, and the room was subjected to UV-light treatment before the work. The PCR was performed using primers AdCR2-F and AdCR2-R developed for ancient Anser geese (Honka et al. 2018) and validated as suitable for eDNA with taiga bean goose (A. fabalis fabalis; Honka et al. 2023). Performance of the primers on the target species, the LWfG, was assessed through a PCR and Sanger sequencing using DNA extracted from a LWfG feather and a scat sample. The effectiveness of the primers on eDNA was evaluated using the spring staging site eDNA sample. The feather DNA, the scat DNA and the spring staging eDNA sample produced a PCR band of correct size (155 bp) and were Sanger sequenced (see Supplementary Fig. 1 for the eDNA sample). Upon comparison with a custom Anser-geese sequence database downloaded from GenBank (see details of the sequence database from Sequence analyses section below), all the obtained sequences were identified as belonging to the LWfG.
PCRs were carried out in 10 µl reaction volumes with 1 x QIAGEN Multiplex PCR Master Mix, 0.2 µM of F- and R-primers (AdCR2-F and -R), RNAse-free water (Qiagen) and 1 µl of the extracted eDNA. The QIAGEN Multiplex PCR Kit was determined to be the most suitable PCR protocol for eDNA (Honka et al. 2023). The thermal profile consisted of 95 ºC for 15 min, followed by 45 cycles of 94 ºC for 30 s, 59 ºC for 90 s and 72 ºC for 90 s with a final extension of 72 ºC for 10 min. All PCRs also included negative controls in which instead of DNA, an equal amount of RNAse-free water was used. The success of the PCR was assessed by electrophoresis of the PCR products on a 2% agarose gel with 0.5 x TBE and 3 µl of Midori Green Advance DNA stain (Nippon Genetics), for 50 min at 115 volts. The success of the PCR was determined from gel as follows. Samples that did not produce a PCR band of the correct size (155 bp, including the primers) were classified as failed samples (Supplementary Fig. 1). Most samples produced a PCR band of a very large size due to nonspecific priming, probably caused by a lack of the target DNA (Supplementary Fig. 1). PCR products which contained both, a band of correct size (155 bp) and unspecific products (Supplementary Fig. 1), were extracted from the gel by cutting the band of the correct size on a UV-table and purifying the gel slice with the GeneJET Gel Extraction Kit (Thermo Fisher Scientific) following the manufacturer’s instructions. The samples which produced a single PCR band of correct size (155 bp) (Supplementary Fig. 1) were direct-sequenced. The PCR products were enzymatically purified with Fast-AP (Thermo Fisher Scientific) and ExoI (Thermo Fisher Scientific) and sequenced in both directions with the PCR primers using BigDye Terminator v.3.1 (Applied Biosystems) chemistry and the reactions were run on an ABI 3730 (Applied Biosystems).
In a few samples (n = 3) the species was not resolved due to a failure to sequence the reactions with the F-primer. These samples were precipitated to increase the DNA concentration and to remove PCR-inhibitors using 10 µl of DNA, 5 µl of Linear acrylamide (Thermo Fisher Scientific) and 60 µl of ice-cold 70% ethanol as in Maixner et al. (2021) but with an additional ethanol wash of the pellet as follows. After removing the supernatant, 200 µl of 70% ethanol was added and the samples were centrifuged for 5 min in 16,000 g. The supernatant was removed, and the leftover ethanol was allowed to evaporate in a laminar flow hood with lid open in a 60 ºC heat block. The pellet was dissolved in 20 µl of water. However, this precipitation method proved effective for only one replicate of one sample. Additionally, we tested precipitating samples (including replicates) which had been successfully sequenced (n = 4), failed samples (n = 10), failed replicates of positive samples (replicate n = 4) and field negative controls (n = 4) but the results proved to be very sporadic, as samples from which a sequence was obtained without precipitation failed, except for one replicate. The precipitation was tested on a larger scale with other eDNA samples (Olli et al. 2023, unpublished manuscript) and it was found that it did not improve the sequencing results and hence we did not perform precipitation beyond this small test set.
Sequence analyses
The obtained sequences were manually edited using the program CodonCode Aligner v.4.0.4. (CodonCode Corporation). Haplotypes were phased using DnaSP v5. (Librado and Rozas 2009) because some of the samples exhibited heteroplasmic-looking sites, i.e. more than one haplotype present. We employed a custom sequence database for phasing. We downloaded control region sequences of all Anser species from the GenBank to obtain the custom sequence database for sequence comparisons. This database included the following sequences by their GenBank accession numbers: lesser white-fronted goose AF159955-AF159956 (Ruokonen, Kvist and Lumme 2000) and AY072580 (Paxinos et al. 2002), greater white-fronted goose (A. albifrons) AF159957-AF159959 (Ruokonen, Kvist and Lumme 2000), bean goose (A. fabalis) EU186805-EU186812 (Ruokonen et al. 2008), AF159951 (Ruokonen, Kvist and Lumme 2000) and MH491806-MH491819 (Honka et al. 2017), pink-footed goose (A. brachyrhynchus) AF159952-AF159954 (Ruokonen, Kvist and Lumme 2000), greylag goose (A. anser) AF159961-AF159963 (Ruokonen, Kvist and Lumme 2000), swan goose (A. cygnoid) AY072581 (Paxinos et al. 2002), snow goose (A. caerulescens) FJ905228, FJ905257 and FJ905285 (Humphries et al. 2009), Ross’s goose (A. rossii) AY072582 (Paxinos et al. 2002), emperor goose (A. canagica) AY072583 (Paxinos et al. 2002) and bar-headed goose (A. indicus) KM455570 (Mu and Xu 2014). The program BioEdit 7.2.5 (Hall 1999) was used to align the Anser-species sequences and eDNA sequences. The program PopART (Leigh and Bryant 2015) was used to construct a median-joining network (Bandelt et al. 1999). The presence of LWfG or other Anser geese species was confirmed based on both the median-joining network and the BLAST search tool against sequences in GenBank.