Beside the River Avon at Saltford, between Bath and Bristol, Will Duckworth has kept a running ledger of one patch: every bird met on every visit, written into a single grid. Between January 2020 and November 2026 there were 2,271 visits, nine days in every ten. Read closely, the grid describes a place that is getting wetter, busier with gulls and garden birds, and quieter in its woods. It also shows the beavers arriving.
What one recorder wrote down, how often, and who turns up.
Each cell below is one day. A coloured cell is a visit, and its shade is how many of the 92 regular species were written down that day. The grid is almost solid. The few gaps are short: the longest break in seven years was 15 days, in early 2025.
National monitoring schemes visit a survey square a couple of times a year. Here the same ground was checked almost daily, so a migrant's arrival can be pinned to the day and a lost winter flock is noticed within a week. That density is what makes the patterns in this page detectable at all.
Goldfinch and Magpie were noted on 97% of visits, Song Thrush and Chaffinch on 92%. Fourteen regulars turn up on more than four visits in five. At the other end, 19 "regulars" appear on fewer than one visit in twenty. Nine everyday birds never appear in the counts at all: Robin, Wren, Blackbird, Dunnock, Blue and Great Tits, Woodpigeon, Jackdaw and Carrion Crow are present throughout and deliberately not counted.
The common birds carry the trend analysis, because for them a missing tick is real information. The long tail carries the surprises. Both matter, but they answer different questions, and the rarest regulars give the noisiest trends.
The first year produced 117 species. After that the curve flattens, but it never stops: 6 new species in 2021, 8 in 2022, 5 in 2023 and a handful since. The latest additions include Common Coot (April 2023), Black-crowned Night Heron (November 2023), Great Crested Grebe (October 2024), Marsh Tit back after years away (April 2025), White-fronted Goose (December 2025) and Richard's Pipit (31 October 2026). Each year holds between 109 and 117 species. 94 species were recorded in all seven years, and 12 in just one.
A well-watched patch saturates fast. Additions after the first year are either true rarities or genuine colonists. Coot is the clearest colonist: it was not recorded at all in 2020–22, and by 2025 it was breeding.
The patch breathes in and out with the seasons, and the timing is shifting.
Here every regular species gets a row, and each row is a year, week by week, averaged over all seven years. The patterns sort themselves. There are 38 residents, present all year. Then come 10 summer visitors that vanish in winter, 17 winter visitors that vanish in summer, and 25 passage and partial migrants that pulse in spring and autumn.
The calendar of a single site is a precise instrument. Chiffchaff and Blackcap sort as residents or partial migrants because some now spend the winter here. Little Grebe and Coot show as "partial", because they arrived part-way through the period, not because they come and go.
The first Reed Warbler of spring was on 26 April in 2020 and 25 April in 2021. Since 2024 it has been 6–9 April. Swallows have crept from early April to late March (about 1.6 days earlier a year), and Willow Warblers likewise (1.5 days a year). In the other direction, Golden Plover now return in autumn about a month later than they did (27 August in 2020, 30 September in 2025), and the last Redwing of spring leaves about five days earlier each year.
Earlier springs for long-distance migrants match what is seen across Britain as springs warm. A first date also depends on how many birds there are, though. Wheatear's first date has moved later (+2.9 days a year) and Golden Plover's autumn return is later too, and both species are declining here. When fewer birds come, the first one is found later.
About one bird in eighteen of all those counted was passing over. There were 2,299 overhead records totalling 49,318 birds. Meadow Pipits (13,703), Redwings (8,935) and Swallows (5,750) lead. The direction tells the season. In autumn, 72% of directional migrants were heading due south and 81% somewhere between south-east and south-west. In spring, 87% headed north-west to north-east.
The weather explains when they are seen. Joining each autumn visit to that day's wind shows a clear pattern. Southbound passage was heaviest on days with southerly and south-westerly winds, about 74 and 66 birds per visit, and lightest under northerlies, about 16 per visit.
Daytime migrants fly low into a headwind and high with a tailwind. A south-westerly in October brings pipits and thrushes down to where they can be seen and counted, while a northerly carries them past too high to notice. The overhead counts therefore measure visible passage, a mix of real movement and how easy it is to see.
For each winter visit (December–February) we took the average temperature of that day and the two before it. When it averaged below 2 °C, Lapwing were recorded on 42% of visits. Above 8 °C they were recorded on 6%. Snipe (46% against 16%) and Golden Plover (31% against 10%) follow the same pattern. These are cold-weather movers, pushed off frozen ground elsewhere to the river's softer margins.
Winters are getting softer. Dec–Feb frost days were 29 in 2020/21 and 33 in 2022/23, then 17, 14 and only 9 in 2025/26. A milder run of winters would by itself mean fewer cold-weather visits from plovers and Lapwing, whatever is happening to their populations.
Seven years is short, but at this density the patch has measurably changed.
An average visit in 2020 produced 32.7 regular species. By 2026 it was 39.7, a steady rise of about 1.3 species a year. Some of that is real: there are new birds to find. Some may come from longer visits, sharper ears or a changed route. The raw numbers can't tell these apart, and they bias every trend upwards. Without correction, 64 species appear to increase and only 26 to decline.
The fix is a standard one, list-length analysis. For each species a model asks: on visits with the same number of species recorded, in the same month, did this bird become more or less likely to be noted? The correction is conservative. Where a visit is longer because new wetland birds have arrived, it explains away a little of a real gain. Both versions are shown from here on.
After correction, the biggest gains are in water birds and gulls. Coot went from absent to recorded on a third of visits. Gadwall, Little Grebe, Water Rail and Reed Bunting rose too, and Lesser Black-backed and Herring Gulls went from about two visits in three to nearly every visit. Greenfinch rose from 68% to 96% of visits, against a steep national decline from disease. The losses are quieter. Corrected for effort, Bullfinch fell from about half of visits to under one in ten, and Nuthatch from a quarter to 3–7%. Starling, Buzzard, Sparrowhawk, Coal Tit and Jay are all slipping.
The split is ecological, not random. Gains cluster on open water, reedy margins and around houses. Losses cluster in woodland and hedgerow, among birds of prey, and among the winter flocks of open fields. The next two charts follow that thread.
Every regular species, plus Cetti's Warbler, which has outgrown its "rarity" row. The solid line is the effort-corrected share of visits each year; the faint dashed line is the raw share. Filter by habitat or search for a bird.
Grouping the regulars by the landscape they depend on turns 92 noisy lines into six clear ones. Wetland and river birds held level from 2020 to 2023, then nearly doubled in 2025. Village and garden birds rose steadily. Gulls climbed sharply. Woodland and hedgerow birds lost about 30%. Farmland birds dipped and partly recovered, and the small birds-of-prey group fell until 2025.
The patch is moving along two axes at once: towards water, and towards people. The woodland decline survives every correction we tried, and it is the one most likely to be a real, wider loss rather than an artefact. The wetland jump is so recent and so abrupt that it points to a local cause.
The first Beaver was recorded on 7 November 2024. By November 2026 there were 57 sightings and 44 records of fresh field signs: gnawed trunks, chew marks, felled stems. Over the same months the water birds surged. Coot was recorded on 1.5% of visits before and 33% after, and bred in 2025 and 2026. Reed Bunting rose from 24% to 64%, Little Grebe from 18% to 36%, Water Rail from 4% to 14% and Gadwall from almost nil to 11%.
The timing fits beavers as engineers: dams, pools and opened-up banks make exactly the habitat Coot, grebes, rails and buntings use. But this is a comparison over time, not an experiment. Cetti's Warbler surged from July 2024, before the first Beaver. Reed Bunting had been rising since 2020, and Little Grebe stepped up and bred in 2024. Something on the ground may have changed first, with the beavers part of that change rather than its cause. Water levels, a new scrape or new reedbed would all leave this signature. The recorder will know which.
Counts tell a sharper story than ticks. Golden Plover averaged 30 birds per winter visit in 2021/22 and under 1 by 2025/26, with the peak flock falling from 360 to 75. Fieldfare fell from 34 a visit to 5 over the last two winters. The great Starling flocks of 2020/21 (one count of 3,120) have not returned. Meanwhile Mallard went from 9 to 53 a visit, Cormorant from 6 to 26, Common Gull from 10 to 88, and Snipe tripled.
The open-field winter flocks are shrinking in step with milder winters, and the wetland flocks are growing in step with the wetter ground. Six winters can't separate cause from coincidence, but both lines point the same way as the habitat index.
Age codes in the grid (juveniles, downy young) show breeding. Coot young were seen on 62 visits in 2025 and 32 in 2026, the first breeding of the period. Little Grebe bred in 2024. Mute Swan cygnets were seen on 40–42 visits a year in 2020–21 but only 8 in 2025. Moorhen chicks almost vanished in 2023–24, then recovered a little. Mallard ducklings appear every year, usually from the first half of April.
Colonisation shows up first as presence and then as breeding, and Coot has now done both. The swan record is a reminder that a single pair's success or failure can swing a whole row. These are counts of sightings of young, not of broods.
Insects, mammals, the rare and the lost, and what the patch teaches.
The same notebook holds 25 butterfly and 20 dragonfly species with eight or more records each, and their seasons stack neatly through the year. 2026 was exceptional. Clouded Yellow was recorded 34 times, against at most 5 in any earlier year, and Hummingbird Hawk-moth 24 times, against at most 1. Willow Emerald Damselfly appeared for the first time in 2025 and Lesser Emperor in 2026, both southern dragonflies spreading north. Jersey Tiger moths are now recorded every year.
Insect records here are incidental rather than systematic, so year-to-year totals mostly reflect attention and weather. First records and mass arrivals are robust, though, and they point the same way as the birds: a warming climate brings southern species to the Avon.
Otter was recorded on 2–8% of visits until 2024, then 17% in 2025 and 13% in 2026, alongside the Beaver's arrival. Brown Hare fell from about 18% of visits to 8% in 2024–25. Hedgehog peaked in 2023 (10%) and has nearly disappeared. Badger and Muntjac edge up. Rabbits dipped sharply in 2025. Daubenton's Bat appears from 2025, more likely a sign of new interest or kit than of new bats.
44 species from the rarities list were recorded on the patch, 11 of them only once: Hoopoe, Richard's Pipit and Night Heron among them. Others are slowly becoming regular. Jack Snipe and Green Sandpiper appear every year, and Cetti's Warbler, with 234 records, is now a resident. The sheet also keeps a list of former regulars: birds that once belonged here.
Six former regulars have no wild record in seven years of near-daily watching (Grey Partridge appears only as released birds). They include Turtle Dove, Tree Sparrow, Lesser Spotted Woodpecker and Nightingale, the same farmland and woodland birds lost across lowland England. Corn Bunting still clings on with 58 records. The patch's losses came before this dataset began, and the woodland trend suggests they are not finished.
Read together, the charts make an argument about how to watch a place.
Seven more species per visit would have turned a balanced picture (30 up, 31 down) into a rosy one (64 up, 26 down). Any long personal record should be read with its own effort curve beside it.
Only one habitat group rose sharply after correction, and it rose within months of the Beavers. Coot, Little Grebe, Water Rail, Gadwall and Reed Bunting all moved together. A small change in water can reorganise a patch faster than decades of land-use change.
Bullfinch, Nuthatch, Coal Tit, Jay and Sparrowhawk: no single crash, but a 30% slide in the woodland index that survives correction. Of all the trends here, this is the one most likely to be a real loss.
Reed Warblers 18 days earlier. Frost days down from 33 to 9. Plover and Fieldfare flocks collapsing in the mild winters. Clouded Yellows and Willow Emeralds arriving from the south. Each alone is anecdote; together they make one pattern.
National schemes would average the Saltford Beaver effect into nothing, and would report Greenfinch falling where here it rose. Near-daily attention to one place catches causes at the scale they act.
"A patch is a place you agree to keep looking at. Look long enough and it starts to answer."
What changed on the ground in 2024: water levels, a new scrape, scrub cleared, reeds grown in? Did the visit route or length change around 2022–24? Are overflying gulls counted the same way in every season? What does "l" mean in the age codes? The answers would turn several "consistent with" statements here into firm ones.
Caveats: one recorder, one site, seven years. 2026 is incomplete (to 5 November). Habitat groupings are ours. Insect and mammal recording is incidental. Overflying gulls are noted in the sheet as "not counted in normal seasons", so gull counts reflect that rule. Before/after comparisons show association in time, not cause.