PlanetCatfish.com

Wilson J. E. M. Costa, Elisabeth Henschel & Axel M. Katz. [Paper dated 2019 but published in 2020]. Multigene phylogeny reveals convergent evolution in small interstitial catfishes from the Amazon and Atlantic forests (Siluriformes: Trichomycteridae).
Zoologica Scripta, 2019;00:1-15. (There are also four supporting files for the paper). First published: 03 January 2020.

Abstract
Interstitial trichomycterid catfishes of the Glanapteryginae and Sarcoglanidinae with uncommon morphology have been known for long time from taxa endemic to the Amazon. In most recent decades, two genera, Listrura and Microcambeva, respectively, placed in Glanapteryginae and Sarcoglanidinae on the basis of morphological characters, have been described from the Atlantic Forest of eastern South America, about 1,500 km from the area inhabited by those Amazon taxa. Herein, we first test the phylogenetic positioning of Listrura and Microcambeva using a multigene data set, including two nuclear and three mitochondrial genes for nine species of Listrura and Microcambeva and 11 species representing all closely related subfamilies (TSVSG‐clade), as well as five used as outgroups. The phylogenetic analyses generated a robust tree with high support values in all nodes, where monophyly of Glanapteryginae and Sarcoglanidinae is refuted. In contrast, Listrura and Microcambeva form a highly supported clade, herein formally described as a new subfamily, sister to a clade containing taxa representing the Glanapteryginae, Sarcoglanidinae, Stegophilinae, Tridentinae and Vandelliinae. This study also indicates that Microcambeva and Listrura exhibit divergent evolutionary trends in ecological and morphological attributes. Species of Microcambeva inhabit patches of loose sand and possess morphological traits that were convergently acquired by Amazon sand‐dwelling sarcoglanidines, including loss of body pigmentation and long maxilla. Species of Listrura live burrowed inside small stretches of dense leaf litter and have morphological traits that were convergently acquired by Amazon glanapterygines inhabiting leaf litter bottom rivers, including elongate body, with numerous vertebrae and loss or reduction of all fins.



https://onlinelibrary.wiley.com/doi/abs ... /zsc.12403

@CharlieM9, multigene analysis.

Cheers, Eric

Hi Eric,

Sorry, but I don't understand the message. Please clue me in. Thanks.

Lee

multigene analysis

The easiest way of producing a tree of relatedness is study one tree. However, coincidence, or luck, can have a rather large influence.

If one samples more than one gene, the cances of luck having an influence will decrease with the number of genes analyzed.

I think CharlieM9 has once pointed this out to Eric

Bas Pels wrote: ↑07 Jan 2020, 14:23I think CharlieM9 has once pointed this out to Eric

You're explanation of the significance of using multiple genes is the correct point, but the directionality of the reason is backwards. I was with Charlie just three days ago and I was sharing with him about a scientist who uses lots of genes and (I believe) microarray technology to build more reliable phylogenetic trees than those obtained using only one or two genes.

Cheers, Eric

Thanks for the double explanation of the (to me, at least) cryptic post...it is appreciated.

Lee

Here's the paper I was refering to. This team uses a full 1% of the entire genome of the fish when doing comparisons, not just one or two genes, and tests anywhere from 60 to over 200 fish specimens per species. https://journals.plos.org/plosone/artic ... ne.0189417

Cheers, Eric

It's increasingly common now to use entire genomes for phylogenetic reconstructions. But what always staggers me is how well the crappy old methods can still perform. Take the recent loricariid phylogeny by Roxo et al. (2019) using 328,330 bp of ultraconserved element DNA ... I get much the same topology using 600 bp of one mitochondrial gene!

The danger with large amounts of data, is that for various reasons, completely the wrong answer can become extremely well supported. Systematic biases, if not adequately modelled, can be quite misleading.

racoll wrote: ↑07 Jan 2020, 19:36The danger with large amounts of data, is that for various reasons, completely the wrong answer can become extremely well supported. Systematic biases, if not adequately modelled, can be quite misleading.

Absolutely true, although I balance that with the exact opposite posture too, as based on the reasoning of the Baumsteiger paper - Making phylogenetic assumptions based on a small amount of data can likewise be very misleading.

Cheers, Eric

bekateen wrote:Making phylogenetic assumptions based on a small amount of data can likewise be very misleading.

With a small amount of data those incorrect relationships tend not to be well supported, and so the gut feeling is that with more data they will be resolved. They often are resolved with 100% support ... and yet they are still wrong.

Check out Table 5 of this paper for a terrifying reminder of this: https://dx.doi.org/10.1093/sysbio/syx089.

racoll wrote: ↑07 Jan 2020, 21:31Check out Table 5 of this paper for a terrifying reminder of this: https://dx.doi.org/10.1093/sysbio/syx089.

Wow. For as much as I think I understand this stuff when I read one paper, all it takes is a paper like this to remind me of how little I understand regarding what goes into this type of analysis. Thanks, racoll.

Cheers, Eric