bPTP cut-off values, with outgrup and without.

53 views
Skip to first unread message

Gillmax Ferreira

unread,
Apr 7, 2025, 6:41:07 AM4/7/25
to PTP and GMYC species delimitation
Hi guys!
I recently discovered this community. I read some of the questions answered and realized that an observation I had seen in my data has already been raised here.
The cutoff values for bPTP are slightly different without outgrup and with.

without outgrup : Max likilhood partition Species 1 (support = 0.588) LC323030,LC323029,LC323031,LC323028 Species 2 (support = 0.704) LC323018,AB996722,AB996695,AB996726 Species 3 (support = 0.585) LC323033,LC323008,LC323007,LC323034 Species 4 (support = 0.550) LC323010,LC323009 Species 5 (support = 0.594) OP038895,KR916843,OP038871 Species 6 (support = 0.678) OP038706,EU300755,OP038683,EU300741 Species 7 (support = 0.452) EU300773,OP038723,EU300774,PV449856_Spain,OP038769,FJ030991,PV449855_Italy,OP038847 Species 8 (support = 0.801) OP038821,EU300637,EU300651,OP038681 Species 9 (support = 0.658) OP038771,OP038721,OP038863,KF737343

with outgrup : Max likilhood partition
outgrup
Species 1 (support = 1.000)
     MN812982

Species 2 (support = 1.000)
     NC082301

Species 3 (support = 1.000)
     KU992689

my species
Species 4 (support = 0.760)
     LC323030,LC323029,LC323031,LC323028

Species 5 (support = 0.840)
     LC323018,AB996722,AB996695,AB996726

Species 6 (support = 0.728)
     LC323033,LC323008,LC323007,LC323034

Species 7 (support = 0.654)
     LC323010,LC323009

Species 8 (support = 0.735)
     OP038895,KR916843,OP038871

Species 9 (support = 0.808)
     OP038706,EU300755,OP038683,EU300741

Species 10 (support = 0.771)
     EU300773,OP038723,EU300774,PV449856_Spain,OP038769,FJ030991,PV449855_Italy,OP038847

Species 11 (support = 0.916)
     OP038821,EU300637,EU300651,OP038681

Species 12 (support = 0.748)
     OP038771,OP038721,OP038863,KF737343


my tree used for bPTP:

#NEXUS
begin trees;
tree tree_1 = [&R] ((((((((((EU300773:2.0E-6,OP038723:1.8E-5):2.0E-6,EU300774:2.0E-6):0.00152,PV449856_Spain:0.003101):2.0E-6,(OP038769:2.0E-6,FJ030991:2.0E-6):2.0E-6):0.001687,(PV449855_Italy:2.0E-6,OP038847:2.0E-6):0.009508):0.027736,((((OP038821:7.7E-4,EU300637:2.0E-6):4.3E-5,EU300651:0.002989):2.04E-4,OP038681:0.010433):0.017967,(((OP038771:0.0,OP038721:0.0):2.0E-6,OP038863:2.0E-6):0.001531,KF737343:2.0E-6):0.025245):0.012411):0.010109,(((OP038706:0.001317,EU300755:0.009503):2.07E-4,OP038683:0.010876):0.005083,EU300741:2.0E-6):0.034272):0.007823,((OP038895:2.0E-6,KR916843:2.0E-6):0.004609,OP038871:2.0E-6):0.04308):0.021379,(((((LC323030:0.001705,LC323029:2.0E-6):0.00523,LC323031:0.005283):0.007021,LC323028:2.0E-6):0.106174,(((LC323018:2.0E-6,AB996722:0.003146):2.0E-6,AB996695:2.0E-6):0.001565,AB996726:2.0E-6):0.110677):0.052149,((((LC323033:0.003537,LC323008:2.0E-6):2.0E-6,LC323007:0.001755):0.007131,LC323034:2.0E-6):0.04585,(LC323010:2.0E-6,LC323009:0.008939):0.047802):0.043087):0.049943):0.177458,((MN812982:0.282046,NC082301:0.126871):0.059124,KU992689:0.187967):0.05851);
end;

I'm working with polychaetes. bPTP recovers all 9 species, but the support is low regardless of whether they are with or without outgrup. Demilitarizers by distance also recover 9 species with good scores and the species have different morphologies.
I wonder if the number of specimens, the COI gene can for this model is the reason for the low support.
and if there is any reference to justify something in this line of thought, please refer me to it.

Das Grimm

unread,
Apr 8, 2025, 7:59:07 AM4/8/25
to PTP and GMYC species delimitation
Hej,

Already from the tree alone the assignment to nine species would be obvious (see pic). So yes, the relatively(!) low probabilities probably relate to a generally somewhat limited differentiation signal in the underlying cox1 data. 

However, there isn't really a problem, is there? You get nine species, visually apparent from and perfectly matching the tree, which are also phenotypically different. That's a result pretty much justifying itself. The bPTP (or any other test) are in a case like this just methodological necessities to put decimals to something already obvious.
One only needs to think about justifying results, looking for high probabilities, when e.g. different methods come to different conclusions or the algorithms spill out something unexpected.
This result here is report & publish.

/G

PS If the reviewers complain about the "low" support, just point that out: "We applied two methods, bPTP and distance-based, both find nine species with support between xxx and xxx. While support may be suboptimal, these nine species are obvious from the ML tree topology as well, where they form highly coherent clades of near-identical individuals (long, prominent roots, flat terminal subtrees)."

polychaets.png

Gillmax Ferreira

unread,
Apr 8, 2025, 8:08:33 AM4/8/25
to Das Grimm, PTP and GMYC species delimitation

Thank you very much. I put it in this perspective "that the results were current with the ML topology, although the bPTP presents low support values"

Grateful,


--
You received this message because you are subscribed to the Google Groups "PTP and GMYC species delimitation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ptp-species-delimi...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ptp-species-delimitation/87731a67-730e-4562-b85b-081eb5fe16d9n%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages