More Callichthyinae than I can imagine? (But no loricariids)

[Jools]

OK, so a good way to sensibly fix this is to introduce a "corrected type locality" lat/long field. In 99 out of 100 cases it will be blank. However, where a type locality is wrong, misformatted or otherwise duff, then this can be used and will take precedence. If the issue persists, it means fresh (duff) data will not wipe out good work but can be periodically removed if the source data is improved. May also add a notes field to assist.

Leave it with me to implement that!

Jools

[bekateen]

Thank you.

And the notes field is a great idea. I recall typing some comments about the lat and long for the Astroblepus I fixed. I didn't want it to go in the General field, so I modified the Distribution field. As I did so, I wondered if that would get overwritten later by a bot. That is the kind of info that belongs in a Notes or Comments type of field.

The idea of a secondary pair of fields for corrected lat and long seems wise. Just tell your code to ignore the official values if secondary (Corrected) are typed in.

Cheers, Eric

[bekateen]

All the species described by Fowler (1913) are based on fish collected previously by someone else and the specimens were donated to the museum:

The fishes described in the present paper were collected by Mr.Edgar A. Smith, in 1912 and 1913. They were secured at various localities along the Madeira River or in its tributaries during Mr. Smith's last trip to Brazil.

This applies to 5). Farlowella smithi, 6). Hyphessobrycon agulha and 8 ).Nannostomus digrammus.

The type locality for these three species is nonsensical. Looking at the original article by Fowler (1913), the type locality is reported accurately - he actually wrote of the type localities, "Madeira River, about 200 miles east of W. Long. 62°20', Brazil." But if you drop a pin on the middle Rio Madeira (near Ipixuna region), you'll see that the longitude is exactly 62°20' W. So how do you go 200 miles east of that and also be on the Madeira river? The only way is to move northeast along the Madeira to Nova Olinda do Norte. This does keep you on the Madeira and you are about 200 miles east of 62°20'W, but that is an awfully circular way to describe what could have been an easy place to characterize, especially given that for all three of these species, the paratypes are collected in the vicinity of Porto Velho - therefore I'd expect the type localities are all further south east than Nova Olinda do Norte (unless Edgar Smith obtained the fish in a fish market upstream). But using the paratype info, even to say "Rio Madeira in the vicinty of Porto Velho" would be a more helpful location than what we have to go on.

2). ** Stored Lat/Long doesn't match type locality **
Platydoras armatulus, type locality=Río Paraná, but not below 27°30'S, South America.
Parsed Lat=-27.5, long=
Stored Lat=0, long=0
Also a lost cause. Nothing to do here.

1). ** Stored Lat/Long doesn't match type locality **
Neosilurus pseudospinosus, type locality=Rocky pool of Ord River on Old Lissadel Station, Kimberley District, Western Australia, about 16°40'S, 128°83'E.
Parsed Lat=-16.666666666667, long=
Stored Lat=0, long=0
Already addressed in a prior post with coordinates proposed. Make longitude = 128°43'E

3). ** Stored Lat/Long doesn't match type locality **
Corydoras(ln9) paragua, type locality=Río Paragua, northwestern bank, river mouth area, río Itenez system, Provincia J.M. Velasco, Departamento Santa Cruz, Bolivia, 13°31.842'S, 61°90.015'W.
Parsed Lat=-13.5307, long=
Stored Lat=-13.53, long=-62.5
Already addressed in a prior post with coordinates proposed: Make longitude 61°50.25'W.

4). ** Stored Lat/Long doesn't match type locality **
Micropanchax ehrichi, type locality=Mali, southwest of Bamako near Samalofila [Samalofira], 12°15'N, 8°72'W [coordinates corrected].
Parsed Lat=12.25, long=
Stored Lat=0, long=0
Samalofila is located at 12.268017, -8.558788. If you accept 12°15'N for the latitude, then you're left with a longitude around 8°33'W (I don't wish to be more precise since I'm not sure how you go from 72 to 33 or anything like it). So perhaps 12.25000, -8.56000 is a good approximation of type locality?


7). ** Stored Lat/Long doesn't match type locality **
Hyphessobrycon langeanii, type locality=Kilometer 476.3 of Ferronorte railroad, 17°25'08"S, 53°13'60"W, córrego Mosquito, município de Alto Araguaia, Mato Grosso, Brazil.
Parsed Lat=-17.418888888889, long=
Stored Lat=-17.418888888889, long=-53.233333333333
You'll notice that the "stored lat" of -53.23333333 corresponds to 53°14' (which is what you get if you convert 13' and 60'' to all minutes). This coordinate (-17.418889, -53.233333) is also what is reported for Lima and Moreia, for a collection they have on file at GBIF. So I'd suggest it is the accurate interpretation of the description. In support of that, on Google Maps, that coordinate drops the pin on railroad tracks, which is consistent with the reference to a specific kilometer location along the Ferronorte railroad.

9). ** Stored Lat/Long doesn't match type locality **
Fundulopanchax amieti, type locality=Rainforest creek on the road from Koupongo, 500 meters west of Somakak, Sanaga system, southwestern Cameroon, 3°58'N, 10°89'E.
Parsed Lat=3.9666666666667, long=
Stored Lat=0, long=0
Cross-checking these numbers with the records at GBIF, they do have collections at 3°58'N, 11°29'E, which would coincide with 10°89'E if you subtract off 60 seconds and add 1 minute. So I'd go with that: 3.966667, 11.483333

10). ** Stored Lat/Long doesn't match type locality **
Trichomycterus aguarague, type locality=Provice Gran Chaco, National Park Aguarague, Quebrada Timboycito, a Río Caiguami tributary, Tarija, Bolivia, 21°30'S, 63°60'W, elevation 700-800 meters.
Parsed Lat=-21.5, long=
Stored Lat=0, long=0
This one is not easily resolvable. As with other species in this dilemma, you can imagine that the longitude is between 63°0'W and 64°0'W, but that doesn't correspond to any specific little stream which can be identified on Google Maps, and there are no townships or roads to reference. If you look at the original publication, they display a map showing all the collection sites, but they don't distinguish the type locality from those of paratypes. Rough eye-balling the map, I can pretty safely say the collection sites are distributed from -21.25 to -21.6 Lat, and -63.3 to -63.6 Long. Actually, I see on Google Maps that if you choose between -63.5 and -63.6 Long, you'll get inside the National Park Aguarague (mentioned for the type locality) on a mountain side. That is consistent with these being collected in mountain streams. So I'll arbitrarily say that your best record may be: -21.5, -63.6.

11). ** Stored Lat/Long doesn't match type locality **
Ancistrus tolima, type locality=Quebrada El Pascado, 3.599306°N, -74.854556°W, vereda San Pablo, municipio de Dolores, departamento de Tolima, Colombia.
Parsed Lat=3.599306, long=-74.854556
Stored Lat=0, long=0
The parsed Lat/Long is correct. But it's confused by the combination of "-" in front of the number and "°W" after. Normally wouldn't add the minus sign if your designating west. Just go with 3.599306, -74.854556.

12). ** Stored Lat/Long doesn't match type locality **
Glyptothorax kurdistanicus, type locality=Serdesht, at Little Zab (36°N), River Bané basin, Iran, elevation 1500 meters.
Parsed Lat=36, long=
Stored Lat=0, long=0
Based on description, Serdesht is a community approx. 2.5 km west of the Little Zab river, about 36°10'N Lat. If a horizontal line is drawn from Serdesht to the Little Zab, the coordinates can be approximated to be 36.159916, 45.534451. But without a longitude value or more precision on the latitude, this is merely an estimate.

13). ** Stored Lat/Long doesn't match type locality **
Amphilius ruziziensis, type locality=Lufiro River, upper reaches of Matchuza, Ruzizi River drainage, ca. 2°44'00"S, 29°02'99"E, Rwanda.
Parsed Lat=-2.7333333333333, long=
Stored Lat=0, long=0
There is a small widening of the river in this area, and anything between 29°02'E and 29°03'E longitude drops the pin in the same area of the river. The seconds appear to be inconsequential in this case. Checking GBIF for records, they have 2°44'00''S, 29°02'E on file for specimens collected; that corresponds to which is -2.7333333, 29.033333

14). ** Stored Lat/Long doesn't match type locality **
Chaetostoma trimaculineum, type locality=Santiago River drainage, Morona-Santiago Province, confluence of the Changachangasa and Tutanangosa rivers and upstream in both, 02°35'51.18''S, 78°11'1038''W.
Parsed Lat=-2.59755, long=
Stored Lat=-2.59755, long=-78.186216666667
It's the 1038''W that are throwing us off here - can't have more than 59.9999''. GBIF has a data point reported by the authors who described the species as -2.59755, -78.186217, the same as what you have in "Stored Lat". I'd say that's good.

15). ** Stored Lat/Long doesn't match type locality **
Trichomycterus pirabitira, type locality=Small stream between Carmo do rio Claro and Conceição da Aparecida, tributary of rio Grande, rio Paraná basin, ca. 21°8'53"S, 46°14'95"W, Estado de Minas Gerais: Município de Conceição da Aparecida, Brazil, elevation ca. 960 meters.
Parsed Lat=-21.148055555556, long=
Stored Lat=-21.148056, long=-46.235972
Your stored longitude treats 46°14'95"W as 46°14'9.5"W, which is probably close and I'd recommend keeping that. Even so, the narrative problem with this species' type locality is that if you use the coordinates given, and even if you allow the authors some leeway on the latitude (e.g., instead of exactly 46°14'95"W, use a range of values between 46°14'W and 46°15'35"W (assuming that 95'' can also be translated as another 1'35''), then you can get a relatively narrow location just beside the rio Claro. BUT, that location is not "between Carmo do rio Claro and Conceição da Aparecida" as described by the authors. Rather the Lat/Long combination is southwest of Conceição da Aparecida, whereas Carmo do rio Claro is northeast of it. There are no locality pins on file at GBIF for this to cross check, and the original paper doesn't provide a map to cross-check.

16). ** Stored Lat/Long doesn't match type locality **
Chiloglanis kabaensis, type locality=Mongo River, east of Marela, on N2, Kaba River (Little Scarcies) drainage, Guinea, 10.17023°N, 11.
Parsed Lat=10.17023, long=
Stored Lat=0, long=0
Already addressed in a prior post with coordinates proposed: 10.170238N, 11.399358W

17). ** Stored Lat/Long doesn't match type locality **
Pseudomystus sobrinus, type locality=Stream about 15 kilometers south of Tuy Hoa, 12°53'05"N, 109°23'70"E, Phu Yen Province, Vietnam.
Parsed Lat=12.884722222222, long=
Stored Lat=0, long=0
At GBIF, one of the authors who described this species has specimens on file with the coordinates 12°53'05.0"N 109°23'42.0"E. That corresponds to 12.8847, 109.395. I'd recommend using that.

[Jools]

A small step has been made in so far as the computed lat/long which is refreshed from co-ords in the type locality field every time a data submission is made is complemented by a "regular" type locality lat/long field pair which can be manually input.

I am still tweaking how this data is used, but it means, I think, that data entered by a human will not get "nuked" by computed data. However, we will fall back to using the computed data if no human data (typically corrections) exist.

Cheers,

Jools

[Jools]

Corydoras(ln9) paragua is fixed up. Will work through the others, slowly at first as this is fiddly, and there may be a few more bugs to squish...

Jools

[bekateen]

Corydoras(ln9) paragua is fixed up. Will work through the others, slowly at first as this is fiddly, and there may be a few more bugs to squish...
Jools

Thanks for the update. Will the new system deposit a type locality starred pin on the map? Or is that lost when revised coordinates are used? I don't see them on the Corydoras(ln9) paragua map.

Cheers, Eric

[Jools]

It should once I fix it!

Jools

[Jools]

I think this is more or less fixed. However, I have a few other things I need to attend to before coming back to clean up the data. Will report back here when progress is being made! Meantime, I think it's safe to try out.

Cheers,

Jools

[bekateen]

Okay, thanks

[Jools]

I've now updated the checker to ignore data where there is a manually entered type locality. It will only store a computed type locality lat/long if one doesn't already exist. The output after I fix a few entries is as follows:

TL02 - Stored Lat/Long doesn't match type locality
Neosilurus pseudospinosus, type locality=A rocky pool of Ord River on Old Lissadel Station, Kimberley District, Western Australia, about 16°40'S, 128°83'E.
Parsed from above Lat=-16.666666666667, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

2). TL02 - Stored Lat/Long doesn't match type locality
Platydoras armatulus, type locality=Río Paraná, but not below 27°30'S, South America.
Parsed from above Lat=-27.5, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

3). TL02 - Stored Lat/Long doesn't match type locality
Farlowella smithi, type locality=Rio Madeira, about 200 miles east of 62°20'W, Amazonas State, Brazil.
Parsed from above Lat=62.333333333333, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

4). TL02 - Stored Lat/Long doesn't match type locality
Hyphessobrycon agulha, type locality=Madeira River, about 200 miles east of 62°20'W, Brazil.
Parsed from above Lat=62.333333333333, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

5). TL02 - Stored Lat/Long doesn't match type locality
Nannostomus digrammus, type locality=Rio Madeira about 200 miles east of 62°20'W, Brazil.
Parsed from above Lat=62.333333333333, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

6). TL02 - Stored Lat/Long doesn't match type locality
Fundulopanchax amieti, type locality=Rainforest creek on the road from Koupongo, 500 meters west of Somakak, Sanaga system, southwestern Cameroon, 3°58'N, 10°89'E.
Parsed from above Lat=3.9666666666667, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

7). TL02 - Stored Lat/Long doesn't match type locality
Glyptothorax kurdistanicus, type locality=Serdesht, at Little Zab (36°N), River Bané basin, Iran, elevation 1500 meters.
Parsed from above Lat=36, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

. TL02 - Stored Lat/Long doesn't match type locality
Amphilius ruziziensis, type locality=Lufiro River, upper reaches of Matchuza, Ruzizi River drainage, ca. 2°44'00"S, 29°02'99"E, Rwanda.
Parsed from above Lat=-2.7333333333333, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

9). TL02 - Stored Lat/Long doesn't match type locality
Chiloglanis kabaensis, type locality=Mongo River, east of Marela, on N2, Kaba River (Little Scarcies) drainage, Guinea, 10.17023°N, 11.
Parsed from above Lat=10.17023, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

10). TL02 - Stored Lat/Long doesn't match type locality
Pseudomystus sobrinus, type locality=Stream about 15 kilometers south of Tuy Hoa, 12°53'05"N, 109°23'70"E, Phu Yen Province, Vietnam.
Parsed from above Lat=12.884722222222, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0

11). TL02 - Stored Lat/Long doesn't match type locality
Thymallus arcticus, type locality=Sob River tributaries, lower Ob River below 60°20'N.
Parsed from above Lat=60.333333333333, Long=
Previously computed Lat=0, Long=0
Manually stored Lat=0, Long=0


Jools

[Jools]

This means I will work through the post of Oct 6th (Thanks Eric) and apply the co-ords manually and they will "stick". This is confirmed by the entry for Neosilurus pseudospinosus which works as planned.

Jools

[bekateen]

Hi Jools, I believe I've submitted now the lat, long updates for all the catfishes.

May also add a notes field to assist.

One thing you mentioned before was a field for correction comments in order to include explanation/documentation for readers going forward that these corrections are just that - corrections of ambiguous data. I think it's a really valuable feature and a note of integrity that we display the fact that some type localities are not necessarily correct but are best efforts to interpret ambiguous data. Given how the "General remarks" field is used in so many other species, it's my feeling that this data correction should not go in that field. I suppose it can go in the "Occurrences" field, unless that field is also autopopulated by external database info (I think it is). How do you wish to handle this?

Cheers, Eric

[Jools]

I am not a fan of notes fields, but you're right, so much of this good work will be forgotten or go unexplained if we don't do something. How about a Type Locality notes field?


Jools

[bekateen]

Yes that works. Or perhaps more generically an "occurrence notes" field, since often occurrences get reported here or in literature without being part of the official occurrence data. That would allow us to comment on type locality and also things like "also collected in rio xyz". Thanks.

[Jools]

Working on adding those new fields now, one other question - I'm looking at "Rio Madeira, about 200 miles east of 62°20'W, Amazonas State, Brazil." for Farlowella smithi. Do you think that's 200 miles downriver (northeast) or the point the river cross 200 miles east line of latitude?

Jools

[bekateen]

Working on adding those new fields now, one other question - I'm looking at "Rio Madeira, about 200 miles east of 62°20'W, Amazonas State, Brazil." for Farlowella smithi. Do you think that's 200 miles downriver (northeast) or the point the river cross 200 miles east line of latitude?

Jools

I tried to take it literally. As you could tell by my narrative, it didn't make sense.

Cheers, Eric

[Jools]

It doesn't. That is unless you take "east" to mean "downriver", then it does work and Nova Olinda do Norte is the locality? Think it would be OK to do that once I add the explanation field.

Jools

[bekateen]

Exactly, you can get there using the description, but as I mentioned above, that is inconsistent with the paratypes. But that's okay, I'd prefer to stay as true to the original paper as possible. I agree with your plan.

Cheers, Eric

[Jools]

Good stuff, I have added the type locality notes feature. Would you like to "guinea pig it" via the usual data submission process? I've tested it, but good to know I've not missed something.

Jools

[bekateen]

Okay, will do.

Cheers, Eric