This doesn't totally answer your query as it won't keep the species name, but a safe way to shorten the names will be to use Biopython:

E.g. if you current file was called long.phy:

>>>  from Bio import AlignIO
>>> AlignIO.convert('long.phy', 'phylip-relaxed', 'short.phy', 'phylip')

Which will give:

 15 60
KM894618.1 ---------- ---------- ---------- --TCTTTGCA TTTATTACGG
KU508975.1 AAATTCTTCG ATATTGGCTG AAAGATCCCT CTTCTTTGCA TTTATTACGA
KC747175.1 AAACTCTCCG ATACTGGTTG AAAGATGCTT CTTCTTTGCA TTTATTACGA
KF632783.1 AAGTTCTGCA AGGCTGGATA CAAGATGTTC CGTCTTTACA TTTATTGCGG
JQ587494.1 ---------- ---------- ---------- ---------- ----------
KR735146.1 ---------- ---------- ---------- ---------- ----------
GU135030.1 AAACTCTCCG ATACTGGTTG AAAGATGCTT CTTCTTTGCA TTTATTACGA
JF953164.1 ---------- ---------- ---------- ---------- --------GA
HM989726.1 AGGCTCTTCG CTATTGGATA AAAGATGCTT CCTCTTTGCA TTTATTAAGA
KF163819.1 ---------- ---------- ---GATGTTC CGTCTTTGCM TTTATTGCGA
MG225316.1 ---------- ---------- ---------T CCTCTTTGCA TTTATTAAGA
MF063987.1 ---------- ---------- ---------- ---------- -------CGA
JQ412229.1 ---------- ---------- ---------- ---------- ----------
JN895697.1 ---------- ---------- -------CTT CCTCTTTGCA TTTATTAAGA
MF350103.1 ---------- ------GGTA AAAGACGCCT CCTCTTTGTA TTTATTAAGA

           TTCTCTCTCT
           CTCTTTCTTC
           TTCTTTCTTT
           TTCTTTCTCC
           ----------
           ----------
           TTCTTTCTTT
           TACTTTCTTT
           TTCTTTCTCC
           TTCWTTCTCC
           TTCTTTCTCC
           TTCTTTCTTT
           -TCTTTCTCA
           TTCTTTCTCC
           CTTTTTCTTT

Many phylip tools use the strict format still which only allows for 10 character names. As I mentioned in the comments, it would be much easier to make your input sequence or downstream files compliant than to make the Phylip compliant.

If you created a map file of accession numbers -> names, you could use simple sed work later on in any file if you wanted to switch between them at will.

Such as:

Abutilon_oxycarpum           KM894618.1
Acalypha_australis           KU508975.1
Achyranthes_bidentata        KC747175.1
Acorus_calamus               KF632783.1
Aeschynomene_americana       JQ587494.1
Ageratum_conyzoides          KR735146.1
Alternanthera_philoxeroides  GU135030.1
Amaranthus_tricolor          JF953164.1
Artemisia_argyi              HM989726.1
Arthraxon_hispidus           KF163819.1
Aster_alpinus                MG225316.1
Bassia_scoparia              MF063987.1
Cynodon_dactylon             JQ412229.1
Bidens_tripartita            JN895697.1
Boehmeria_nivea              MF350103.1

If you had made a tree from the alignment for example:

(((JQ587494.1:0.0,KR735146.1:0.0):5.990225341,((JQ412229.1:0.118111457,((KF632783.1:0.079049581,KF163819.1:0.068672610)0.944:0.180951505,(MG225316.1:0.000000005,(HM989726.1:0.000000005,JN895697.1:0.000000005)0.393:0.000000005)0.894:0.088130659)0.890:0.047666678)0.323:0.000000005,KM894618.1:0.084141325)0.002:0.025703744)0.000:0.020579960,(JF953164.1:0.075087560,((KC747175.1:0.0,GU135030.1:0.0):0.000000005,MF063987.1:0.000000005)0.423:0.000000005)0.219:0.027420688,(MF350103.1:0.197253279,KU508975.1:0.074619017)0.701:0.065875748);

This general form could be used to switch out the names:

while read name accession ; do sed -i "s/$accession/$name/g" mytree.tree ; done < accession_mapfile.txt

to give:

(((Aeschynomene_americana:0.0,Ageratum_conyzoides:0.0):5.990225341,((Cynodon_dactylon:0.118111457,((Acorus_calamus:0.079049581,Arthraxon_hispidus:0.068672610)0.944:0.180951505,(Aster_alpinus:0.000000005,(Artemisia_argyi:0.000000005,Bidens_tripartita:0.000000005)0.393:0.000000005)0.894:0.088130659)0.890:0.047666678)0.323:0.000000005,Abutilon_oxycarpum:0.084141325)0.002:0.025703744)0.000:0.020579960,(Amaranthus_tricolor:0.075087560,((Achyranthes_bidentata:0.0,Alternanthera_philoxeroides:0.0):0.000000005,Bassia_scoparia:0.000000005)0.423:0.000000005)0.219:0.027420688,(Boehmeria_nivea:0.197253279,Acalypha_australis:0.074619017)0.701:0.065875748);

If you really are hell-bent on shortening the names, the closest/most sensible thing I can come up with, without truncating the string, is to form abbreviated latin binomials like so:

sed -i 's/[a-z]*_/\./g' mapfile.txt

But some of these are still too long and I still advise against doing that in the phylip, unless you come up with some relable shortening system.

Hello.

Here is a solution using sed.

$ sed -r 's/^[^_]*_([^_]*_[^_]*)[^ \t]*(.*)$/\1\2/' phylip_alignment > shorten

What we are trying to do here, is to replace each line with substrings of the line. To realize this, one have to define the pattern of the string in each line and mark those substrings we want to keep. This is done with ^[^_]*_([^_]*_[^_]*)[^ \t]*(.*)$. The whole line is than replaced with the substrings that matches the patten in (...). The replacment is done with \1\2.

fin swimmer