Re: gephyra or agassizii?

[Randy or Deb Carey <carey@spacestar.net>]

Marco Lacerda wrote:

> Hi!
>
> <snip>

> That's all ok, but recently I had problems to identify fishes from upper
> Rio Negro (Igarape Urupau). The features of fishes from that locality
> are (now that the fishes are about eight months old in aquarium):
>
> a) very long tip of last dorsal fin ray, fitting very well A. agassizii;
> b) very short tip of pelvic fin, FITTING VERY WELL A. GEPHYRA (when
> compared to A. agassizii from other localities, i.e., lower Amazon
> populations, very very short pelvic fins).
> c) strongly lanceolated caudal fin, fitting very well A. agassizii;
> d) NO POINTS OR MARKINGS on both upper and lower parts of caudal fin,
> but instead of it, orange-reddish color. There is just an edge for both
> upper and lower parts of caudal fin.
> e) about body depth difficult to say exactly without pickling the
> fishes.
>
> Does anybody or RANDY himself have any opinion about it?
> The Rio Negro is known to shelter both A. agassizii and A. gephyra.
>
> All the best, Marco.

I feel obligated to respond in some way.

Identifying fish is hard enough from photo's, so all I can do is comment.

I'm assuming that in all other features the fish look like agasizzi and gephyra.
So aside from feature 'b', it could qualify as agasizzi.  A rule I believe in is:

   "the presence of a feature carries more weight than does the absence of a
feature."

In this case, 'a' and 'c' are features which are "present" and suggest
agassizii.  Feature 'd' argues agaisnt gephyra due to an absence of gephyra's
caudal markings.  The color of the caudal in feature 'd' sounds acceptable for
agassizii.   Feature 'b' argues against agassizii due to the absence (or late
development) of a feature.  I'd say the stated evidence leans strongly for
agassizii, but with some doubt that perhaps it is a third species (or a deviant
population).  Afterall, new species crop up fairly often.

Romer's new book (due next May?)  is supposed to have info indentification info.
I hope it is not the typical binary key which is standard in zoology.  That's
where the key tells you to check a feature and go 'left' or 'right' depending on
the answer.  The problem is, if you can't quite determine any one of the required
features (maybe a young specimen or a difficult feature to read), then you have
no hope of coming to the right conclusion.   That method was the best they could
come up with a couple hundred years ago, but with computers we can do much
better.  The user should be able to respond to as many features/questions as
possible and ignore uncertain responses.  A computer can evaluate the responses
against the features of all candidate species.  Well, that's much of how I view
ID'ing a species from visible features.

- --Randy