Eduardo Freitas / Lab / Hidden Lineages

Lab notebook · ECO-02 · Southern Atlantic Forest

Some fish may be two species under one name.

A DNA barcode is a short stretch of one gene that works like a species fingerprint. When animals carrying the same name turn out to have two clearly different fingerprints, the name may be hiding a species nobody has described. This project reads every public barcode from Paraná, Santa Catarina and Rio Grande do Sul and lists the names most likely to be hiding one.

12,228
records from the three states
6,336
usable DNA barcodes
74
names split across groups
7
fish candidates
Plate 1. Every site in PR, SC and RS where a usable barcode was collected. Sites of the seven candidates are marked in blue.
How it works

From public records to a short list

  1. 01

    Barcodes

    BOLD Systems

    Public DNA barcodes, each with a name, a place and a sequence.

  2. 02

    Groups

    BINs

    BOLD clusters near-identical barcodes into groups that roughly match species.

  3. 03

    Splits

    one name, two groups

    Names whose barcodes fall into two or more groups the name itself dominates. Mislabelled records are set aside.

  4. 04

    Second opinion

    ASAP

    An independent method checks whether there is a real gap between the groups, or only noise.

  5. 05

    Candidates

    for a specialist

    What survives, with the evidence next to it.

Fig. 2. Each step removes a kind of false alarm.
The candidates

Seven fish worth a second look

Ordered by how strongly the second method backs the split. None of these is a new species yet: each is a question for someone who can examine the fish.

Salminus brasiliensis

The dourado, or golden dorado: a large river predator and one of the most sought-after fish in sport fishing.

Its barcodes fall into two groups that differ by 6.5%. That is larger than the difference between many pairs of fish already recognised as separate species.

For specialists

BINs: BOLD:AAZ3803 (7), BOLD:AAD2790 (4). ASAP finds a gap at the BIN boundary (p = 6e-04).

The two largest groups were collected at a shared site (6 and 3 sites).

Published: Rosso et al. 2012, Mol Ecol Resour 12:999-1011. Over 3% between southern populations and the rest of the range.

largest groupsecond group

Phalloceros harpagos

A barrigudinho: a small livebearing fish of streams and ditches, a relative of the guppy.

Its barcodes fall into two groups that differ by 2.2%. That sits at the edge where biologists start to suspect two species.

For specialists

BINs: BOLD:ABY8844 (40), BOLD:ACE6602 (3), BOLD:AAB5570 (1). ASAP finds a gap at the BIN boundary (p = 1e-05).

The second group is known from 1 site, 88 km from the first.

Published: Pereira et al. 2013, BMC Genetics 14:20, Table 3. 1.7 to 2.2% within the species in the Upper Paraná.

largest groupsecond group

Phalloceros titthos

Another barrigudinho, a close relative of the one above.

Its barcodes fall into two groups that differ by 1.4%. That is small: it may be ordinary variation inside one species, so the case is weaker.

For specialists

BINs: BOLD:AAB5570 (13), BOLD:AGE7052 (3). ASAP finds a gap at the BIN boundary (p = 0.03).

The two largest groups were collected apart (8 and 2 sites, closest 83 km).

Nothing published found. Its main BIN also holds one P. harpagos record.

largest groupsecond group

Synbranchus marmoratus

The muçum, or marbled swamp eel: an eel-shaped fish that breathes air and can wait out a dry spell in the mud.

Its barcodes fall into three groups that differ by 7.3%. That is larger than the difference between many pairs of fish already recognised as separate species.

For specialists

BINs: BOLD:AAB6824 (4), BOLD:AAB6826 (3), BOLD:ADI2293 (1). ASAP finds a gap at the BIN boundary (p = 4e-05).

The second group is known from 1 site, 176 km from the first.

Published: Pereira et al. 2013, Table 3; Rosso et al. 2012. 2.9 to 8.0%, a known species complex.

largest groupsecond group

Piabina argentea

A piaba: a small silvery fish of rivers and streams.

Its barcodes fall into two groups that differ by 1.7%. That is small: it may be ordinary variation inside one species, so the case is weaker.

For specialists

BINs: BOLD:AAA5781 (16), BOLD:ADJ7993 (10). ASAP's gap at the BIN boundary is weak (p = 0.26).

The second group is known from 1 site, 11 km from the first.

Published: Pereira et al. 2013, Table 3. 1.9 to 5.6%, 6 subclusters in the Upper Paraná.

largest groupsecond group

Pimelodus maculatus

The mandi-amarelo: a spotted catfish common in the Paraná and Uruguay basins.

Its barcodes fall into two groups that differ by 1.7%. That is small: it may be ordinary variation inside one species, so the case is weaker.

For specialists

BINs: BOLD:AAB6504 (86), BOLD:AFE7458 (8). ASAP's gap at the BIN boundary is weak (p = 0.27).

The two largest groups were collected at a shared site (11 and 1 sites).

Published: Biology 2024, 13:162, doi:10.3390/biology13030162. Several MOTUs across basins.

largest groupsecond group

Apareiodon affinis

The canivete: a small fish that scrapes algae off river stones.

Its barcodes fall into two groups that differ by 5.8%. That is larger than the difference between many pairs of fish already recognised as separate species.

For specialists

BINs: BOLD:ABW1944 (5), BOLD:AAD2370 (3). ASAP's gap at the BIN boundary is weak (p = 0.19).

The second group is known from 1 site, 479 km from the first.

Published: PMC6726146. Two genetic groups and an undescribed species.

largest groupsecond group
Read this first

What this is not

  • These are candidates, not new species. One gene can be misleading: hybridisation, old shared ancestry or a mislabelled specimen can all produce two groups. Naming a species takes more genes and a close look at the animals.
  • The data are thin. Only fish and insects have enough barcodes in the south, and most species have fewer than five. Frogs, where the idea started, have none that qualify.
  • It finds what is already known. Tested against published cases written down before the run, it found them. How often it raises a false alarm is not yet measured.
  • Insects are not reviewed yet. 34 insect names pass the same checks and wait for the same reading the fish got.
Set aside

Set aside

Names whose extra barcodes sit in another species' group, or split by a distance usual between genera with another genus in the same group. These look more like labelling problems than hidden species, and are listed for whoever holds the specimens.

  • Astyanax laticeps: BOLD:ABU7559 / BOLD:AAB3435, 16.8%, Deuterodon langei. Southern barcodes of the complex do support splitting P. laticeps, but a 17% gap with a Deuterodon record in the same BIN fits misidentified vouchers better than a cryptic Astyanax.
  • Corydoras paleatus: 2 → BOLD:AAC7264 (Corydoras ehrhardti). Rosso et al. report over 3% between southern C. paleatus and the rest of the range, so these records may be a lineage rather than a misidentification.
  • Rhamdia quelen: 5 → BOLD:AAA6323 (Rhamdia voulezi), 4 → BOLD:ACF1690 (Rhamdia branneri). Ribolli et al. work on Rhamdia lineages with BINs on SC material; either the vouchers are R. voulezi and R. branneri, or the three names do not separate in COI.
Wantedtaxonomists who know southern fish

Which of these are worth a look? If you work on any of these groups, a one-line answer helps decide what to examine next. A candidate that holds up could become a co-authored description.

Write about Hidden Lineages

Data: public records from BOLD Systems, downloaded 2026-10-01. Species delimitation with ASAP (Puillandre et al. 2021). State outlines from IBGE. Generated 2026-10-02.