Evidence · Birds & biodiversity

Birds, fish & wildlife of the Teddington–Richmond reach

What the national biological-records archive holds for this stretch of the tidal Thames — waterbirds, a Critically Endangered eel, a near-absent clean-water fish, and an invasive mussel. Every count here reflects how much recording was done, not how many animals live in the river.

NBN Atlas occurrence records Teddington Lock → Richmond bbox 51.41–51.47 N, 0.29–0.34 W Accessed 2026-07-31 RAW-PULL

NBN Atlas · occurrence records in the reach

The headline figures

These figures come from a single live query to the NBN Atlas occurrence web service, bounded to the Teddington-to-Richmond reach. They are a useful indicator of what lives here and who has been watching — but they are a record of observations, not a census. Read them alongside the caveats at the foot of the page.

142,070
occurrence records of all kinds in the reach — birds 95,080, mammals 4,121, molluscs 1,133, fishes just 98.
NBN Atlas occurrence search, species-group facet RAW-PULL
19,469
waterbird records across 23 core species — led by Black-headed Gull (1,826), Mallard (1,823) and Canada Goose (1,817).
NBN Atlas, 23-species waterbird set RAW-PULL
98
fish records only — against 95,080 for birds. The eel and smelt evidence base here is thin and stops ~2017.
NBN Atlas fishes facet RAW-PULL · PROVEN GAP
2014
documented first record of the invasive quagga mussel at Richmond — yet NBN's earliest here is 2015, so the aggregator lags the ground truth.
NBN 10 records 2015–2017 vs documented 2014 PROVEN GAP

What is recorded — and what is barely recorded

Split the 142,070 records by broad species group and one thing dominates: birds. Fish — the group most directly exposed to water quality and sewage discharge — return just 98 records for the whole reach. That is an evidence gap, not evidence that fish are absent: fish are simply surveyed far less, and less openly, than birds.

Occurrence records by species group

All NBN Atlas records in the Teddington–Richmond bounding box, by broad taxon

NBN Atlas occurrence records by species group, Teddington to Richmond reach Horizontal bar chart of 142,070 occurrence records by species group. Birds dominate with 95,080 records. Every other species group not itemised here totals 41,638. Mammals 4,121, molluscs 1,133, and fishes only 98 — a striking under-representation of fish that reflects an evidence gap, not an absence of fish from the river. 0 20,000 40,000 60,000 80,000 Occurrence records in the reach Birds: 95,080 records All other species groups (not itemised): 41,638 records Mammals: 4,121 records Molluscs: 1,133 records Fishes: only 98 records Birds Other groups Mammals Molluscs Fishes 95,080 41,638 4,121 1,133 98 — fish evidence gap
Source: NBN Atlas occurrence web service (records-ws.nbnatlas.org), species-group facet, bounding box lat 51.41–51.47 N / lon 0.29–0.34 W, accessed 2026-07-31. "Other groups" is every species group other than the four named above, not individually itemised in this pull. The fishes bar (98) is flagged as an evidence gap. Provenance: RAW-PULL.

Why rising record counts are not rising populations

It is tempting to read a climbing number of records as a recovering river. It is not. Waterbird records rose roughly sevenfold from the 2000s to the 2010s (1,691 → 11,766) — but records of all birds rose in lockstep (23,744 → 45,147). When both climb together, the driver is recording effort — more surveyors, apps and uploaded datasets — not a change in the animals themselves. This is the single most important thing to understand about these data.

Bird records by decade — all birds vs waterbirds

The two series rise and fall together, so the trend is effort, not abundance · 2020s partial (to 2026)

Bird occurrence records by decade, all birds versus waterbirds Grouped bar chart from the 1990s to the 2020s. All-bird records: 1,958 in the 1990s, 23,744 in the 2000s, 45,147 in the 2010s, and 23,450 in the partial 2020s. Waterbird records rise and fall in step: 193, then 1,691, then 11,766, then 5,612. Because both series climb and fall together, the pattern reflects recording effort, not a change in bird populations. 0 10,000 20,000 30,000 40,000 Occurrence records per decade All bird records Waterbird records (a subset of all birds) 1990s: 1,958 all-bird records 2000s: 23,744 all-bird records 2010s: 45,147 all-bird records 2020s (partial): 23,450 all-bird records 1990s: 193 waterbird records 2000s: 1,691 waterbird records 2010s: 11,766 waterbird records 2020s (partial): 5,612 waterbird records 1,958 23,744 45,147 23,450 193 1,691 11,766 5,612 1990s 2000s 2010s 2020s Decade (NBN Atlas records in the reach · 2020s partial)
Source: NBN Atlas occurrence records by decade, waterbird set vs all birds, same bounding box. The waterbird share of all bird records did climb — from about 7% in the 2000s to about 26% in the 2010s — which is suggestive but indicative only. Effort comparison DERIVED from RAW-PULL counts.

Because effort dominates the signal, these counts cannot demonstrate a pollution- or discharge-driven decline, and rising records must never be presented as a recovering river. Correlation with any water-quality series would not be causation. For a defensible waterbird trend the correct source is the request-only WeBS counts (see caveats).

Three species worth watching

Behind the bird totals sit three species that tell a sharper story: a Critically Endangered European eel, a clean-water smelt that is almost absent from the record, and an invasive quagga mussel that arrives in the mid-2010s. All three are sparsely recorded, and the fish records stop around 2017 — so this is a story about how little is known as much as about the animals.

Key indicator species — individual records by year

European eel, smelt and quagga mussel · dot size ∝ number of records that year

Key indicator species occurrences by year: European eel, smelt and quagga mussel Timeline of individual NBN records from 1993 to 2017. European eel appears in 17 records between 1993 and 2017, then stops. Smelt has only three records ever — two in 1995 and one in 2011. Quagga mussel, an invasive, appears only from 2015 to 2017 (3 then 7 records) — later than the documented 2014 Richmond first-record. Dot size is proportional to the number of records in that year. 1993 1998 2003 2008 2013 2018 Year of record (NBN Atlas) European eel 17 records · IUCN CR Smelt 3 records · indicator Quagga mussel 10 records · invasive 2014 documented Richmond first-record European eel — 1993: 4 records European eel — 1994: 1 record European eel — 1995: 3 records European eel — 1996: 2 records European eel — 2000: 1 record European eel — 2005: 1 record European eel — 2007: 2 records European eel — 2012: 1 record European eel — 2015: 1 record European eel — 2017: 1 record Smelt — 1995: 2 records Smelt — 2011: 1 record Quagga mussel — 2015: 3 records (NBN earliest here) Quagga mussel — 2017: 7 records 4 3 2 2 2 3 7
Source: NBN Atlas indicator-species query (target_species_records.csv), drawn from the EA "Protected and Invasive Species Records 1995–2021", the "Database for the Atlas of Freshwater Fishes" and "England Non-Native Species records 1965–2017". The dashed line marks the documented 2014 Richmond first-record for quagga mussel, which pre-dates NBN's earliest capture (2015). Provenance: RAW-PULL.
NBN Atlas indicator species in the Teddington–Richmond reach. Counts and date ranges are live index totals; fish records stop around 2017. Provenance: RAW-PULL.
Species NBN records Date range Status / role Principal data sources
European eel (Anguilla anguilla) 17 1993–2017 IUCN Critically Endangered; catadromous migrant EA Survey 1995–2021; Atlas of Freshwater Fishes
Smelt (Osmerus eperlanus) 3 1995 (×2), 2011 Estuarine / clean-water indicator EA Survey 1995–2021; Atlas of Freshwater Fishes
Quagga mussel (Dreissena rostriformis bugensis) 10 2015–2017 Invasive non-native; NBN earliest 2015 lags documented 2014 Richmond first-record England Non-Native Species 1965–2017; EA Survey 1995–2021

Invasive context. The quagga mussel is one of several non-natives now established in the reach. Among molluscs, the most abundant records are the Asian clam Corbicula fluminea (192), the New Zealand mudsnail Potamopyrgus antipodarum (147) and the zebra mussel Dreissena polymorpha (82). Invasive pressure is a real ecological signal — but it is not attributable to any water company, and it is not evidence of a discharge-driven decline. RAW-PULL, molluscs facet.

Waterbirds of the reach — 23 species, 19,469 records

The waterbird set is the richest part of this archive. Gulls, common wildfowl and the naturalised Egyptian Goose lead the list; two species — the Kingfisher and Grey Wagtail — are worth singling out as riverine indicators, birds tied to clean, insect-rich flowing water (highlighted below).

NBN Atlas waterbird records in the reach, all 23 core species (total 19,469). Riverine indicator species highlighted. Provenance: RAW-PULL.
Species Records Species Records
Black-headed Gull1,826Lesser Black-backed Gull764
Mallard1,823Great Crested Grebe647
Canada Goose1,817Red-crested Pochard494
Mute Swan1,714Grey Wagtail — riverine indicator431
Coot1,517Common Gull404
Moorhen1,279Mandarin Duck334
Egyptian Goose1,259Gadwall236
Grey Heron1,231Common Tern212
Herring Gull1,008Kingfisher — riverine indicator193
Tufted Duck918Great Black-backed Gull190
Cormorant917Little Grebe131
Greylag Goose124

The two indicator species are recorded but not abundant in this archive (Grey Wagtail 431, Kingfisher 193). As with every count on this page, these are observation totals, shaped by where and how often people surveyed — not population estimates. RAW-PULL.

How to read this

Caveats an ecologist would press — stated up front

  • Effort, not population. This is the master caveat. NBN occurrence counts rise with the number of surveyors, apps and datasets uploaded — not with animal numbers. Waterbird and all-bird records rose together (see the decade chart), so no count on this page can demonstrate a pollution- or discharge-driven change in wildlife. Treat every trend here as recording activity.
  • Indicative, not complete. The fine-scale data that could support a defensible trend — the WeBS Wetland Bird Survey (BTO) counts and full-resolution GiGL London records — are request-only and not on NBN at full resolution. What you see here is the open subset. PROVEN GAP
  • Fish evidence is thin and stops ~2017. Just 98 fish records for the whole reach; smelt has only three records ever; the Critically Endangered eel is present but barely monitored, with nothing on NBN after 2017. A genuine data gap for any fish- or ecology-based decision. PROVEN GAP
  • Aggregators lag the ground. The quagga mussel was documented at Richmond in 2014, but NBN's earliest record here is 2015 — a reminder that official aggregators trail first-hand observation.
  • Coarse geography. Localities were returned as "Not supplied"; the reach match is by coordinate, and some records are grid-blurred, so fine placement within the reach is approximate.
  • Not the water company's ledger. Invasive-species pressure is a real signal but is not attributable to any water company, and correlation with any water-quality series would not be causation. NBN alone cannot prove a discharge-driven decline — read this as context, not proof.

Sources & provenance

  • RAW-PULL  NBN Atlas occurrence web service — https://records-ws.nbnatlas.org/occurrences/search, WKT bounding box lat 51.41–51.47 N, lon −0.34 to −0.29 W (Teddington/Kingston/Ham/Petersham/Richmond), accessed 2026-07-31. Live index totals may shift slightly as datasets reload.
  • Underlying datasets: EA "Protected and Invasive Species Records Collected Through Environment Agency Survey 1995–2021"; "Database for the Atlas of Freshwater Fishes"; "England Non-Native Species records 1965–2017".
  • PROVEN GAP  BTO / WeBS fine-resolution Wetland Bird Survey counts and GiGL full-resolution species records are request-only and not carried on NBN at full resolution — the correct sources for a defensible trend.
  • Working dataset: /tmp/thames_datastore/biodiversity/ — waterbird_species_counts.csv, waterbird_by_year.csv, birds_by_decade.csv, target_species_records.csv, and the pull/chart scripts.