Study site
Small passerines were caught using a 13 m mist net placed between trees and shrubs on the edge of the village of Tarbet, Argyll & Bute (56.21 N 4.71 W), adjacent to a large forestry plantation. Siskins can forage up to at least 5 km from their nest during the breeding season5,18, and the area within 5 km was therefore considered likely to include breeding siskins that would move through the catching site. Plantation forestry species, age, and area were determined from maps from the Forestry Commission compartment data base. There were 1152 ha of plantation forestry within 5 km of the catching site, comprising 79% Sitka spruce, 7% Norway spruce Picea abies, 13% larch and < 1% Scots pine Pinus sylvestris and lodgepole pine Pinus contorta. Most of the area was planted between 1952 and 1976, and most of the Sitka spruce was therefore of cone-bearing age when the study started in 200536. Tree age has been shown to have a strong influence on seed production, even in mature trees17, and given that very little felling occurred in the area up to 2020, cone production was likely to have increased throughout the study period (2005–2020). Other major habitats within this radius include freshwater (Loch Lomond), ancient woodland dominated by oak, regenerating woodland dominated by birch, open moorland with scrub, and rural village gardens. Deciduous woodland and rural village gardens may provide alternative foraging opportunities for siskins, and in particular many gardens contain bird feeders providing seeds and peanuts, which can be used by siskins and some other small passerines12 and so may attract siskins and other birds. A total of 12,513 ha of plantation forestry (76% Sitka spruce) was located within a wider radius (20 km) from the catching site, with about 2500 ha planted each decade from 1950 to 2000. Plantation forestry is thus an important habitat both locally and over the wider area. Daily rainfall data were provided by the Met Office from a weather station 4 km north of the study site (Sloy Power Station, Inveruglas, 56.25 N 4.71 W).
Cone scores and seed shedding
Sitka spruce cone production was assessed using a 10-point scale developed by Petty et al.26. The score was based on a visual assessment of the relative abundance of ripe cones in late autumn/early winter throughout the forestry plantation. Cone crop was scored from zero (no cones visible on the tree canopy when viewed with binoculars from a vantage point) to 10 (exceptionally large crop, the entire upper tree canopy appearing brown with cones when viewed from vantage points). This is essentially the same scale used by Shaw18 in SW Scotland, but divided into 10 rather than five categories. Scoring was generally carried out over a few days in autumn before the cones had fully ripened, to discount any remaining old cones from the previous years, by viewing through binoculars all of the areas of plantation forestry from overlooking vantage points. Bird data were related to the cone crop in the preceding autumn/winter, because seeds remain within the cones for up to 12 months10, and the breeding of siskins in year X would therefore be expected to be influenced by the cone crop in year X − 1. The cone score was only recorded for Sitka spruce as the main conifer in the study area, but the Norway spruce cone crop has been shown to correlate with the Sitka spruce crop, and cone crop is synchronous across the whole of the UK13, so the cone scores derived in this study apply also to other parts of the UK.
In order to assess whether Sitka spruce seed was available to siskins as food throughout their breeding season, we carried out a cone ripening study to measure seed loss from cones. Seed retention was estimated by sampling Sitka spruce cones (one per tree) from a sample of 10–30 trees at approximately 1-month intervals from 1 October 2006 to 31 July 2007 (a fairly typical year in terms of rainfall according to Met Office data but a year with high cone crop) from two locations, Argyll Forest Park near Tarbet (within 1 km of the bird-catching site) and West Lamberkine Wood, Perthshire. The latter site was chosen to represent an area of Scotland with generally drier weather conditions compared with Argyll, to determine if seeds were shed faster in a drier climate, as found in Alaska7,8. Individual cones were cut using a 4 m pruning pole and handled carefully to avoid seed loss due to cutting. Cones were stored individually in polythene bags in the field and then placed in a dry room for several days to allow the cones to open. The seeds were then extracted by tapping the cones over a sheet of paper, and the number of scales per cone and the number of extracted seeds were counted. Each scale initially holds two seeds, and the maximum number of seeds was calculated as: (number of scales) × 2. The number of seeds extracted was then expressed as a percentage of the maximum number of seeds. Monthly sampling of cones was not carried out in other years, so we cannot be certain that seed loss is the same in years of low cone crop.
Catching and processing birds
Mist netting was carried at intervals of about one week. However, as mist netting can only be done at low wind speeds and under dry conditions, the interval between catches varied according to the weather conditions. Mist netting was carried out on a total of 568 days between 2005 and 2020, with annual totals of 28–44 days (Table 1), resulting in the capture of 50,282 birds, including 16,553 siskins. The mist net was kept under continuous observation and birds were normally removed from the net as soon as they were caught. On a few occasions, the net had to be closed to prevent a build-up of birds if the catch rate was exceptionally high or because weather conditions changed (usually rain showers), but catching was usually continued throughout the day, from about 0700 to about 1600 GMT. Each bird was identified to species, ringed (or its ring number was recorded if had been ringed previously), classified according to sex when possible43, weighed (to the nearest 0.01 g) using an electronic balance, and wing length (maximum chord) was measured (to the nearest mm). The time of capture was noted to the nearest 30 min. Age of siskins was determined based on the shape of the tail feathers and the presence or absence of a colour contrast between the inner and outer greater coverts43. Adult female siskins were checked for brood-patch score (as in Shaw 199018) and adult siskins were checked for primary moult score27. Any feathers shed by juvenile birds caught before their post-juvenile moult (i.e. in the short period immediately following fledging), or any tail feathers shed by juveniles were stored dry in a labelled paper envelope for stable isotope analysis. All birds were released, usually within a few minutes after initial capture. Mist net capture and ringing of birds was carried out under the supervision of R.W.F., British Trust for Ornithology (BTO) bird ringing permit number S/2282, following BTO guidelines and regulations.
Stable isotope analysis
Sitka spruce seeds were sampled from cones in Argyll Forest Park in 2005, 2006, 2017, 2018, 2019 and 2020. Pooled samples of seeds from about 30 cones were used for stable isotope analysis, with approximately equal representation of each of the years in which seeds were sampled. Seed samples were milled to a fine powder in a Retsch MM200 ball mill before weighing. Ground seed samples (1.5 ± 0.1 mg) were weighed into tin capsules (8 × 5 mm). Up to 10 samples of feathers were collected from different juvenile siskins from each year for which feathers were available: six from 2001, and 10 from each of 2007, 2009, 2011, 2016, 2017, 2018, 2019 and 2020 (86 birds in total). Feathers from other juvenile small passerines that had not yet entered post-juvenile moult were sampled for the same years: 35 chaffinches, 21 coal tits, 10 blue tits, nine great tits, eight robins, six blackbirds Turdus merula, six dunnocks, six goldfinches and two willow warblers. Feather samples were prepared by finely chopping with surgical scissors and 1 ± 0.1 mg samples were weighed into tin capsules (8 × 5 mm).
Carbon and nitrogen isotope ratios of feather and seed samples were analysed by elemental analysis—isotope ratio mass spectrometry. Samples and reference materials were weighed into tin capsules, sealed, and loaded into an auto-sampler on a Europa Scientific elemental analyser. Combusted gases were swept in a helium stream over a combustion catalyst (Cr2O3), copper oxide wires (to oxidize hydrocarbons), and silver wool to remove sulphur and halides. The resultant gases were swept through a reduction stage of pure copper wires held at 600 °C.. Water was removed using a magnesium perchlorate chemical trap. Nitrogen and carbon dioxide were then separated using a packed column gas chromatograph held at a constant temperature of 65 °C. The resultant nitrogen peak entered the ion source of the Europa Scientific 20–20 IRMS and was ionized and accelerated. Nitrogen gas species of different masses were separated in a magnetic field and measured simultaneously using a Faraday cup collector array to measure the isotopomers of N2 at 28, 29 and 30 m/z. After a delay, the carbon dioxide peak entered the ion source and was ionized and accelerated, and carbon dioxide gas species of different masses were separated in a magnetic field and measured simultaneously using a Faraday cup collector array to measure the isotopomers at 44, 45 and 46 m/z. Both references and samples were converted to N2 and CO2 and analysed using this method. The analysis was carried out in a batch process involving analysis of a reference, followed by several samples and then another reference. The reference material used for δ13C and δ15N analyses of feather and seed samples was IA-R068 (soy protein, δ13CV-PDB = − 25.22‰, δ15NAIR = 0.99‰). IA-R068, IA-R038 (L-alanine, δ13CV-PDB = − 24.99 ‰, δ15NAIR = − 0.65‰), IA-R069 (tuna protein, δ13CV-PDB = − 18.88‰, δ15NAIR = 11.60‰) and a mixture of IAEA-C7 (oxalic acid, δ13CV-PDB = − 14.48‰) and IA-R046 (ammonium sulphate, δ15NAIR = 22.04‰) were run as quality control samples. IA-R068, IA-R038 and IA-R069 were calibrated against and traceable to IAEA-CH-6 (sucrose, δ13CV-PDB = − 10.449‰) and IAEA-N-1 (ammonium sulphate, δ15NAIR = 0.40‰). IA-R046 was calibrated against and traceable to IAEA-N-1. IAEA-C7, IAEA-CH-6 and IAEA-N-1 were inter-laboratory comparison standards distributed by the International Atomic Energy Agency, Vienna. These are standard methods for stable isotope analysis, and are similar to those described elsewhere44,45.
Data analysis
Analysis was performed using the dataset of birds caught from 2005 to 2020. Although birds were captured in previous years, ringing was less frequent and the sample sizes were smaller. Linear regression analyses were performed in R46 and Excel47, and ANOVA analyses were performed in R46 with continuous predictors (except in the case of ‘month’ in the analysis of fledgling siskin weights and wing lengths, which was a categorical variable). Siskin adult and juvenile annual ringing totals for Britain and Ireland for 2005–2019 were extracted from Robinson et al.24.
We used several metrics of timing of breeding. Numbers of adult females caught was used as one index of timing of breeding because female siskins incubate eggs whereas males do not, so when incubation starts the proportion of females in catches declined. Once chicks have hatched, the proportion of adult females increases again, so this metric indicates timing of breeding, as does the presence of a brood patch on adult females (which develops as eggs are laid but re-feathers after breeding, whereas adult males do not develop a brood patch). Arrival of fledged young in mist net catches also indicated timing of breeding.
Knotted function analysis was used to fit a curve to certain variables, and differences in the shapes of the curve between years could be quantified to measure changes in the behaviour of the siskin population between years. The curves were fit using the ‘get_natural_cubic_spline_model’ function in the Basis Expansions Python module48 within a custom Python function (Supplementary Eq. 1), within a custom Python script (Supplementary Eq. 2). After fitting the curve to the data, the script adjusted the position of the curve independently for each year to minimize the misfit between the predicted and observed values in each year, and recorded the magnitude and direction of shift in each year as a metric of the ‘earliness’ or ‘lateness’ of the siskin population dynamics in that year. Knotted function analysis excluded captures with < 15 valid birds on the basis that captures with few individuals might contain relatively unbalanced sex or age ratios purely by chance. Birds caught after the 220th day of the year were also excluded on the basis that siskins became very rare around this time each year, and the graph-fitting process would therefore be unreliable. This filtering left 208 data points to construct the fitted curve of sex ratios, and 260 data points to construct the fitted curve of age ratios.
Filtering of data and conversion from the records of individual birds into records of individual ringing sessions involved the use of three custom Python scripts (Supplementary Eqs. 3–5): Script 3 was used to analyse the frequency of fledgling siskins, relative to all siskins; Script 4 was used to analyse the frequency of adult female siskins, relative to all adult siskins; and Script 5 was used to analyse the frequency of 1st year adult siskins relative to all adult siskins. The number of juvenile siskins, relative to all siskins, in June and July was analysed by sorting the list of relevant birds by age and counting the number of captures in each category. The proportion of juvenile and/or female siskins were measured rather than absolute numbers in order to remove any influence of variable ringing effort or weather on catch.
Source: Ecology - nature.com
