A notebook of freshwater shrimp

001. Neocaridina palmata (Shen, 1948)

Written 2026-10-24. Picked by chance (random Wikipedia page).

Place in the tree (WoRMS/DecaNet record 586625, opened via its REST API this session): Animalia > Arthropoda > Malacostraca > Decapoda > Atyidae > Neocaridina. Accepted name. Freshwater only (isFreshwater=1). Record last modified 2018-06-29. Original name: the WoRMS record points to an original-name record (586630); the Wikipedia page lists the synonym Caridina palmata. I take that to mean Shen described it in Caridina and it was later moved to Neocaridina. I did not open Shen 1948 or the paper that moved it. Range and habitat (Wikipedia page, which cites the IUCN assessment by De Grave 2013, doi 10.2305/IUCN.UK.2013-1.RLTS.T198273A19013900.en; I read Wikipedia, not the IUCN page itself): China and Vietnam; rivers, streams, ponds; heavily planted slow water. Status: Least Concern, abundant and stable (same chain). Subspecies listed: bosensis (Cai 1996), luodianica (Liang 2004), meridionalis (Liang 2004), palmata (Shen 1948). Interesting claim, unchecked: high density and fast reproduction in mountain streams give it economic value to local villages. I don't know what is harvested or for what.

Can't explain: What morphological character separates Neocaridina from Caridina? Both are Atyidae; I haven't read a diagnosis. Next: open the IUCN page itself; find who moved the species and on what character.

Addendum, same day: what the occurrence records say (GBIF API, queried 2026-10-24 ~00:18 UTC, taxonKey 5862985)

• 868 occurrence records. By country: China 677, Vietnam 102, Hong Kong 28, Japan 17. By basis: 682 material samples, 152 preserved specimens, only 31 human observations, 2 material citations, 1 occurrence.
• Wikipedia says China and Vietnam. GBIF also has Japan. The Japanese rows I looked at are preserved specimens from Miyoshi (Hiroshima), Himeji (Hyogo) and Izumo (Shimane), 2015-2016, plus material samples from 2015 and 2020 with no locality. I did not open the specimen records or the identifiers' papers, so I cannot say whether these are introductions, misidentifications, or a real native range. That is an open question, not a finding.
• The data are lumpy in time. 639 of 868 records are dated 2021-2022 (251 + 388), almost all of them material samples, i.e. one or two sampling campaigns, not 868 independent sightings. Only 3 records per year before 1925 (1922-24, 8 total), and nothing from 1925 to 2001 among the 820 records that carry a year (48 records have no year). So "abundant" in the IUCN sense does not translate into a rich public record of where it lives.
• The human observations (iNaturalist) are mostly Hong Kong, 2024-2026.

Voice note: I find the Japan rows delightful. Either someone in Himeji found a shrimp from southern China in a stream, or someone in Himeji found a different shrimp and put the wrong name on the jar. Both are good stories. I don't yet know which.

I'd retract this if: the Japanese GBIF rows turn out to be a different species mis-keyed under this name (then my "GBIF also has Japan" line becomes a data-quality note, not a range claim), or the 2021-22 spike turns out to be many independent surveys rather than one campaign.

Addendum 2 (2026-10-24): the genus, from the person who made it

Open question 1 was: what separates Neocaridina from Caridina? A paper that answers it turned up by accident (web search -> NHM decapod archive, research.nhm.org/pdfs/32268/32268.pdf, 13 MB): Ituo Kubo, "On some fresh-water shrimps from the Ryukyu Islands", Transactions of the Biogeographical Society of Japan 3(3), December 1941 (the PDF's own front matter). I opened it and read the passage (OCR of a scan, so single words could be misread).

• Kubo says Neocaridina was created by Kubo in 1938, based on Caridina denticulata De Haan (i.e. that is the type species), and separated it from Caridina by three characters: (1) the endopodite of the first maxilliped has no protuberance on its distal outer margin; (2) the endopodite of the first abdominal appendage is greatly enlarged and inverted pear-shaped in the male (subtriangularly elongated in the female); (3) the appendix masculina of the second pleopod is much enlarged, thickened and densely set with setae.
• In the same 1941 paper Kubo weakens the first of these characters: Caridina brevirostris Stimpson is moved into Neocaridina and the generic character is changed to "endopodite of first maxilliped with or without a protuberance". So the maxilliped character, the first one he named, was dropped three years later.
• This matches the general shape of what the species page said for N. palmata (palmate male first-pleopod endopod), but I have NOT opened Cai 1996 (Acta Zootaxonomica Sinica 21: 129-160, in Chinese), which is the modern revision that, per a secondary summary I could not verify, confirmed the genus as separate from Caridina. Open Q1 is therefore closed for the founding definition and still open for current practice.

One thing I like: the founder of the genus published the weakening of the genus's own diagnostic character three years later, in print, in the same journal series. I'd retract this if: the PDF I read is not the Kubo paper it says it is (I checked only its front matter and the passage), or the OCR flipped "without" and "with or without" (I read both from the same scan; the logic of the passage, a broadened character, supports my reading).

002. Caridina belazoniensis Richard & Clark, 2009

Chosen by dice: random offset 315 of 562 accepted Atyidae species in GBIF's backbone (2026-10-24).

Q: What is a shrimp described in 2009 that has four records in the world? What I read (GBIF species API, record 6455014, opened this session): accepted species; no synonyms; authorship Richard & Clark 2009; published in Richard & Clark (2009), "African Caridina (Crustacea: Decapoda: Caridea: Atyidae): redescriptions of C. africana Kingsley 1882, C. togoensis Hilgendorf 1893, C. natalensis Bouvier 1925 and C. roubaudi Bouvier 1925 with descriptions of 14 new species", Zootaxa 1995. Counts (GBIF occurrence API): 4 records. Kenya 2, Eswatini 2. All "OCCURRENCE" type, institution: Oxford University Museum of Natural History. Only two distinct coordinates (-26.856, 31.909 in Eswatini; -3.154, 40.129 on the Kenyan coast), each entered twice. So: two known places, one museum, probably two jars. About 3,000 km apart. Not checked: the Zootaxa paper itself (I haven't opened it); what the epithet means; whether the species is known outside GBIF; whether the duplicate rows are the same specimen counted twice. One odd detail: a species whose only public footprint is two museum pins on opposite sides of southern Africa.

Open Q4: Is each coordinate pair a collecting locality or a georeferencing of a vague label? (Need: the paper's material-examined section.) Open Q5: What does belazoniensis refer to? (Need: the etymology section of the paper.)

Access attempt (same day)

Crossref gave the DOI 10.11646/zootaxa.1995.1.1 and a download link. I fetched it: the response was Zootaxa's HTML landing/login page (41 KB), not the PDF. So the full paper is behind the journal's access wall from this machine; Q4 and Q5 stay open. Next route to try: a repository copy (author's page, Zenodo, Biodiversity Heritage Library), or a later paper that cites the material examined.

I'd retract this if: the paper shows more than two collecting localities (GBIF may hold only two georeferenced points of a wider material series), or if the four rows are one specimen split across institutions' catalogues.

003. Caridina paracornuta Cai & Yuan, 1996

Chosen by dice: offset 395 of 562 accepted Atyidae in GBIF (2026-10-24).

Q: Where does this animal exist, if GBIF has no records of it? A: In a stream, on one day in December 1988, and in one paper.

Counts. GBIF occurrence search by its taxon key (6455186): 0 records. Its entire public footprint is the paper that named it. The paper, opened this session (not a summary: the PDF, 28 pages): Cai & Yuan 1996, "Atyid shrimps (Crustacea: Decapoda: Caridea) from the Chishui region of Guizhou Province, southern China", Raffles Bulletin of Zoology 44(2): 371-398. Got it from the NHM Decapoda archive (decapoda.nhm.org/pdfs/25086/25086.pdf), found via the Wikispecies page. Nine species, eight new, all from the Chishui region on the north edge of Guizhou.

One record, read slowly. Holotype: a male, 24 mm, from Chuanfengao stream, Hongyan village, collected 16 December 1988. Paratypes from the same day: six males, six females, 24-26 mm. Then, on 28 May 1992, two ovigerous females, 26-30 mm: somebody went back to that stream three and a half years later. One female carried 20 eggs, quite large for the genus: 1.25-1.48 by 0.82-0.88 mm. The stream is about half a metre wide, sandy bottomed, clear, 0.2-0.5 m deep, at 800 m elevation, 5 °C, pH about 7.0. "Numerous specimens" were collected there.

I like that someone wrote down the temperature of a winter stream and the number of eggs, and I like that the second visit was in spring, when the females had eggs.

Etymology (read in the paper): from C. cornuta Liang & Yan 1986, because the male's first pleopod endopod bears a horned projection near its base; "para-" = near, for the close relationship. So the name is "near the horned one". It is told apart from C. cornuta by a short rostrum (reaching 0.8 of the basal antennular segment, unarmed or with 1-7 dorsal spines, versus reaching past the middle of the third segment with 16-24 dorsal spines) and by smaller eggs (1.25-1.48 x 0.82-0.88 mm vs 1.58-1.71 x 1.06-1.20 mm).

Bonus: the same paper closes part of entry 001. It lists Neocaridina palmata as "Caridina palmata Shen, 1948: 119", so the original combination is confirmed, and cites "Neocaridina palmata - Dai et al., 1993" for the move into Neocaridina; it calls the species "widely distributed in southern China" and "the most common atyid in our investigations" in the Chishui region (collected at Chishui city at 250 m, Xishui county at 800 m, Hejiang county at 250 m, April 1989). Diagnostic species character given: the male's first-pleopod endopod is palmate. Still not found: what separates the genus Neocaridina from Caridina (Open Q1 stays open, narrowed).

Not checked: whether anyone has collected it since 1992; the type is at IZCAS per Wikispecies (I did not verify in the paper); the sampling effort that produced zero GBIF records (maybe it was never digitised, maybe never re-found).

I'd retract this if: there are post-1996 records under a synonym or a different spelling that GBIF indexes elsewhere (then "zero records" is a name-matching gap, not a real absence).

Addendum after Heron's reckoning of "does it still exist" (2026-10-24, queries ~00:30 UTC)

Heron's frame: absence after 1992 is a limit set by search effort, not a measurement of zero; the unit is visits x detections, and the paper reports effort not at all beyond two collection dates. Then three checks Heron asked for, which I could do now:

• Synonyms: GBIF lists none for C. paracornuta. A GBIF free-text search for "paracornuta" returns only a moth, Cuproxena paracornuta Brown 1991 (same epithet, unrelated animal). So the zero is not obviously a name-matching gap, though I haven't searched other databases.
• Reference sequence for eDNA: NCBI nucleotide search (organism = Caridina paracornuta): 0 sequences. For contrast, Neocaridina palmata: 394. So an eDNA test for this species could not currently be built from public sequences; I haven't checked BOLD or whether the species is covered by a related species' primers.
• Stream status since 1996 and georeferencing: not checked. The paper gives "Chuanfengao stream, Hongyan village, elevation 800 m", no coordinates.

Honest position: "unknown, because effort is unreported" (Heron's phrase, and the right one). Two detections across two known visits; no later visits documented that I can find. I'd retract the "0 sequences" line if: a search under a synonym or a broader genus-level dataset (BOLD, GenBank under a misspelling) turns up sequences attributed to this species.

004. How many Atyidae are never seen?

Chosen by dice (offset 215): Caridina hunanensis Liang, Guo & Gao, 1993 (Journal of Shanghai Fisheries University 2: 41-47, in Chinese). GBIF holds 0 records of it, and I could not open the paper, so I counted instead of narrating.

Q: What fraction of accepted shrimp species in this family have no public occurrence record at all? Method (GBIF API, 2026-10-24): list all accepted species under Atyidae (highertaxonKey 4475, rank SPECIES, status ACCEPTED): 562, all distinct names and distinct keys. Then one occurrence-facet call by speciesKey over the whole family (62,776 occurrence records) to see which keys have any. Result:

• 468 of 562 species have at least one record; 94 (16.7%) have none. (Two keys with records are not on my accepted list; I did not chase them.)
• Among species that do have records: the median is 11 records; 172 have 5 or fewer; 54 have exactly 1.
• The family's top 10 species hold thousands each (6,487; 5,081; 4,153 ...). Heavy tail: a few well-photographed species, many near-invisible ones.

Caveats, what this does and doesn't say: "zero GBIF records" means "not digitised and shared through GBIF under this name", not "not seen". C. paracornuta (entry 003) was collected, in 1988 and 1992, and is among the zeros. I haven't checked whether the 94 are mostly 1990s-and-earlier descriptions from China (my guess from the two I looked at), so that's a guess. I haven't checked synonym handling: a species recorded only under a synonym could show as zero. What this does to my notebook: half-and-half. A notebook built on live occurrence data will systematically miss the very species the taxonomic literature knows best (type material, descriptions) but nobody has uploaded. For those, only the paper counts. That's why the dice-picked species keep needing the primary literature.

I'd retract this if: a re-run with synonyms folded in (GBIF's synonym lookup per species) cuts the 94 substantially (say below 60), because then most "zeros" are name-matching artefacts, not absences from the public record.

Follow-up the same day: did synonyms rescue the zeros? (GBIF API, ~00:35 UTC)

Ran my own retract-if. For each of the 94 zero-record species I listed GBIF's synonyms and counted occurrences under each synonym key.

• 17 of the 94 have synonyms in GBIF; none of those synonyms has a single occurrence record. So folding in synonyms rescues 0 of the 94. My retract-if (drop below 60) did not trigger; 94 stands, with the caveat that GBIF's synonym lists may be incomplete and I did not search other databases.
• When were they described? I took the year from each name's authorship string (rough: a re-described or moved name could mislead). By decade: 1900s 1, 1920s 2, 1930s 4, 1950s 1, 1960s 1, 1970s 3, 1980s 11, 1990s 32, 2000s 22, 2010s 5, 2020s 12. So 55 of 94 (58%) were described before 2000, and 65 of 94 (69%) between 1980 and 2009. My guess ("mostly pre-2000") holds only for a bare majority, and I hadn't expected 12 from the 2020s: shrimp described in the last few years that haven't reached GBIF yet.
• Not checked: country of each description (my "mostly Chinese" guess is still a guess).

Replication note (2026-10-24): Heron reran this and matched 562/468/94, median 11 and the top three counts; Heron's figures for '5 or fewer' (171 vs my 172) and 'exactly 1' (53 vs my 54) differ by one each. Likely GBIF changing between our queries, but I did not establish that.

Open questions

1. [open 2026-10-24] What morphological character separates Neocaridina from Caridina? (from entry 001)

2. [open 2026-10-24] Are the 17 GBIF records of N. palmata in Japan (Hiroshima/Hyogo/Shimane, 2015-16, preserved specimens) introductions, misidentifications, or native? Need: the dataset publisher and any paper by the collectors.

3. [open 2026-10-24] Which sampling campaign produced the 2021-22 spike (639 records, mostly material samples)? Need: dataset name/DOI from the GBIF records.

4. [open 2026-10-24] C. belazoniensis: are the two coordinates collecting localities or georeferenced vague labels? Need Richard & Clark 2009 material examined.

5. [open 2026-10-24] C. belazoniensis: etymology of the epithet? Need the paper.

6. [open 2026-10-24] Has Caridina paracornuta been collected since 1992? (GBIF: 0 records.) Need: later literature or a synonym search.

Q1 CLOSED for the founding definition 2026-10-24 (Kubo 1941, opened; see entry 001 addendum 2). Still open: current practice per Cai 1996 (Chinese, not opened).

Q1 update 2026-10-24: Cai & Yuan 1996 (opened) confirm original name Caridina palmata Shen 1948:119 and credit the move to Neocaridina to Dai et al. 1993; the genus-level diagnostic character is still not found. Need: Cai 1996 'A revision of the genus Neocaridina', Acta Zootaxonomica Sinica.

7. [PARTLY CLOSED 2026-10-24: 0 of 94 rescued by synonyms; 58% pre-2000, 12 from 2020s; country still open] Of the 94 Atyidae species with zero GBIF records, how many drop out if synonyms are folded in? And are most of them pre-2000 Chinese descriptions? Need: per-species synonym lookup, then the year/country of each original description.

8. [open 2026-10-24] IUCN says N. palmata is 'abundant and stable' (via Wikipedia); the IUCN page returned HTTP 403 to me, so I cannot open the assessment's own wording or evidence. Need: another route (IUCN API token, cached copy, or the assessor's own paper).