Table 3.

Splits supported

Split	Support for splits
	LD	NJP	NJK	NJD	MP	ML	QP
i	1	100	100	100	100	100	100
m	2	100	100	100	100	100	100
k	3	100	100	100	100	100	100
e	4	100	100	100	100	100	100
j	5	100	100	100	100	100	100
a	6	100	100	100	83	–	98
b	7	100	100	100	100	100	100
d	8	100	100	100	94	98	100
l	9	100	100	100	100	100	100
f	10	100	86	98	–	–	–
h	11	96	99	99	79	89	96
c	12	98	70	83	–	–	–
g	13	93	84	95	81	–	66
o	14	–	–	–	–	85	65
p	15	–	–	–	–	–	–
r	16	–	–	–	–	–	–
n	–	–	–	–	98	99	–
q	–	–	–	–	–	64	–

^*Splits a–q in Fig. 4. Except columns LD and QP, numbers indicate the average bootstrap proportion across the four data sets with the given method. Average values <50 are indicated with a dash.

^† Splits listed in order of strength as determined through spectral analysis, e.g., branch “i” was the strongest split detected with LD.

^‡ Average of resampling estimated log-likelihood (RELL) bootstrap proportions.

^§ Average of quartet puzzling support values.