3.3. Revised coding in Necrolestes and Cronopio

advertisement
Journal of Mammalian Evolution
Supplementary Material
Were there Miocene meridiolestidans? Assessing the phylogenetic
placement of Necrolestes patagonensis and the presence of a 40
million year meridiolestidan ghost lineage.
Rachel N. O’Meara1 and Richard S. Thompson1*
1
The University Museum of Zoology, Downing Street, Cambridge, CB2 3EJ
*corresponding author: rst32@cam.ac.uk
Contents
1. Estimating the length of the meridiolestidan ghost lineage
2. Coding additional taxa
2.1. Materials
3. Recoding
3.1. Coding methods
3.2. Revised character state definitions
3.3. Revised coding in Necrolestes and Cronopio
4. Tree lengths from the alternate parsimony analyses
5. Alternate Yalkaparidon analyses
6. Bayes factor analyses
6.1. Using stepping stone analyses to compare topologies
6.2. Setting up the stepping stone analysis in MrBayes v3.2.2
6.3. Results: Marginal Likelihoods and Bayes factors
7. Optimizations
7.1. Parsimony analysis: optimization of character states related to zalambdodonty
7.2. Parsimony analysis: optimization of character states across the meridiolestidan
lineage
7.3. Parsimony analysis: optimization of character states related to tribospheny
8. Bayesian analysis: ancestral character state reconstruction
9. References
1. Estimating the length of the meridiolestidan ghost lineage
Necrolestes patagonensis is known from the Miocene Santa Cruz Formation, which is dated
between approximately 19 and 14 Ma, although in the Atlantic coastal regions this
formation spans a shorter temporal interval of between approximately 18 and 16 Ma
(Perkins et al. 2012). The most recent known meridiolestidan, other than Necrolestes, is
Peligrotherium of the ‘Banco Negro Inferior’ of the Salamanca Formation (early Selandian)
(Gelfo et al. 2001, 2009). The ghost lineage of Necrolestes may therefore range between
approximately 40 and 48 Ma. We have conservatively referred to the length of the ghost
lineage of Necrolestes as ‘approximately 40 million years’ throughout this paper.
2. Coding additional taxa
The phylogenetic analysis of Rougier et al. (2012) was expanded to include three additional
taxa: Notoryctes typhlops, Naraboryctes philcreaseri, and Yalkaparidon (coded as composite
of Y. coheni and Y. jonesi).
2.1. Materials:
Notoryctes typhlops material:
UMZC A5.1/1 - Complete postcranium. Skull with damage to right auditory bulla. Jaw with
complete dentition.
UMZC A5.1/5 – Skull with jaw, but missing zygomatic arches and with damage to both
auditory bullae. Dentition incomplete: missing right ultimate premolar and right first molar.
AMNH 202107 - Animated 3D renderings and ‘dynamic cutaway’ views of complete skull
and jaw from Digital Morphology website, University of Texas, Austin (Rodgers 2008).
Further information for coding Notoryctes characters was provided by figures and text from
the following publications: Szalay (1994); Long et al. (2002); Asher and Sánchez-Villagra
(2005); Asher et al. (2007); Ladevèze et al. (2008); Ungar (2010). In particular, figures of CT
scans of a Notoryctes petrosal from Ladevèze et al. (2008) were used in coding petrosal
characters.
Naraboryctes philcreaseri material:
Coded from figures and text of Archer et al. (2011). Naraboryctes is known from two left
dentaries, fragments of maxilla, upper and lower dentition, a left humerus, and a left ulna.
Yalkaparidon material:
The character states of Yalkaparidon used in the main analyses of the manuscript are coded
from figures and text of Archer et al. (1988), Beck (2009) as the most detailed peer reviewed
published articles on this taxon at the start of the project. Most coding of palatal, orbital,
zygomatic, mandibular, and dental characters was taken from images of a skull and its
associated dentary, and of upper and lower molars of Yalkaparidon coheni. An image of a
right dentary and third lower premolar of Yalkaparidon jonesi was also used (Archer et al.
1988). Some maxillary characters and characters relating to the upper dentition were coded
using images of isolated maxillae referred to Yalkaparidon sp. (Beck 2009).
Alternate codings are also presented, which account for the novel interpretations of
Yalkaparidon in Beck et al. (2013). Beck et al. (2013) revised the interpretation of the dental
formula of Yalkaparidon, identifying a single premolar (P3) and four molars (contra Archer
1988). Furthermore, they suggested uncertainty over the identity of the distolingual lower
molar cusp, raising the possibility that this represents the remnant of a talonid basin rather
than a metaconid. This latter hypothesis is tested in a separate analysis. Our alternative
codings of Yalkaparidon can be found in our nexus file, along with variants of all three
codings (Archer 1988 and two alternatives from Beck et al. 2013), which incorporate
uncertainty over the identity of the primary lingual cusp of Yalkaparidon. The names of the
taxa in the nexus file are explained below.
Yalkaparidon_met_Archer – Coded following Archer et al. (1988) and Beck (2009) with
metacone zalambdodonty.
Yalkaparidon_uncertain_Archer - Coded following Archer et al. (1988) and Beck (2009) with
uncertain zalambdodonty.
Yalkaparidon_met_Beck - Coded following Beck et al.’s (2013) dental formula with
metacone zalambdodonty.
Yalkaparidon_uncertain_Beck - Coded following Beck et al.’s (2013) dental formula with
uncertain zalambdodonty
Yalkaparidon_met_Becktalonid - Coded following Beck et al.’s (2013) dental formula with
metacone zalambdodonty and interpreting the distolingual lower molar cusp as vestigial
talonid basin.
Yalkaparidon_uncertain_Becktalonid - Coded following Beck et al.’s (2013) dental formula
with uncertain zalambdodonty and interpreting the distolingual lower molar cusp as vestigial
talonid basin.
The results of these alternate analyses are detailed in Section 5.
3. Recoding
3.1. Coding Methods
Throughout, we have followed the coding strategy of Rougier et al. (2012) as faithfully as
possible in order to maintain the consistency of the analysis. As regards the coding of
inapplicable and unknown character states, we note that in the original Rougier et al. (2012)
matrix, there are several instances where one character state is dependent on the presence
of another character. For example, character 90:
(90. Entoconid. (0) absent (1) present, but far from hypoconulid, at least equal to one cusp
length (2) present and twinned with hypoconulid.)
…is dependent on character 85:
(85. Talonid Basin. (0) Absent. (1) Present).
If character 85 is coded as absent (0), then logically, character 90 should be coded as
inapplicable (-): the entoconid is a talonid cusp, and is not recorded without the presence of
a fully basined talonid. However, under the Rougier et al. (2012) coding strategy, the
entoconid is scored as absent, effectively up-weighting the absence of the talonid in the
analysis. We have followed this original coding strategy in these instances. We are acutely
aware of the difficulties of assigning tooth cusp homologies across Mesozoic
mammaliaforms, particularly those distal cusps that could be associated with the talonid. It
would be interesting to recode the entire matrix to account for the contingency of these
characters, so as to prevent up-weighting the absence of certain features such as the talonid
basin, although this is clearly a complex endeavor.
When coding additional taxa by the Rougier et al. (2012) character list, it was assumed that
when tooth cusp identity is designated by letter, rather than by name, that this refers to the
relative position of the cusp, rather than its strict homology, e.g. ‘cusp a’ indicates the labialmost cusp of the trigonid of the lower molar rather than a cusp homologous with the
protoconid.
Character 113 (see below for example) would be coded as (1) in both paracone- and
metacone zalambdodonts since ‘upper primary cusp b’ refers to a cusp in the same relative
position in both forms of zalambdodonty, rather than strictly to the paracone.
113. Topographic Relationships of Wear Facets to Main Cusps. (0) Lower cusps a and c
support two different wear facet (1 and 4) that contact the upper main cusp a. (1) Lower
cusps a and c support a single wear facet (4) that contacts the upper primary cusp b.
However, if a cusp is designated solely by name, and not by letter, then it was assumed that
this refers to its homology, and not its relative position. Thus, character 130 (Metacone. 0 =
Present; 1 = Absent) would be coded (1) if Necrolestes is a paracone zalambdodont and (0) if
it is a metacone zalambdodont.
Archer et al. (1988) stated that: “…although Yalkaparidon molars are zalambdodont there is
no trace of… paraconids, metaconids or stylar cusps that are prominent features in molars of
Notoryctes”, but that the lower molars “…have [a] single large lingual cusp which we
interpret as [the] protoconid from which emanate two lingually directed crests. Although
the corners at which these crests terminate could be homologs of paraconid and metaconid,
they do not form discrete cusps.”
However, the protoconid and the two lingually directed crests are the only prominent
eminences of the lower molar of Yalkaparidon and occupy the same relative positions as the
protoconid, paraconid, and metaconid in most therians. Since character definitions relating
to the trigonid cusps are phrased by designating cusps by letter (i.e., by relative position)
rather than by name (i.e., by homology), the mesio-lingual and disto-lingual crests and the
labial cusp of the lower molars of Yalkaparidon have been coded, in the main analysis, as if
they were the paraconid, metaconid, and protoconid respectively. Alternative codings for
the lower molar cusps of Yalkaparidon are explored in the analyses
“Yalkaparidon_uncertain_Becktalonid” and “Yalkaparidon_met_Becktalonid” in order to
investigate the possible alternative lower cusp homologies suggested by Beck et al. (2013).
3.2. Revised character state definitions
Some character definitions required revision in order to accommodate new taxa or the
assumption of a different form of zalambdodonty within the existing list of character states.
Original character definitions from Rougier et al. (2012) are shown in regular typeface
below, with revisions shown in red. Each character statement is discussed in italics, after its
definition.
2. Position of the posteriormost mental foramen. (0) Below the canine and anterior
premolariform region. (1) Below the penultimate premolar. (2) Below the ultimate
premolar. (3) Posterior to the ultimate premolar.
Revised character state 3 definition since, in Yalkaparidon, the posteriormost mental
foramen is situated below the second molar, not between the ultimate premolar and the first
molar as indicated by the previous character state.
115. Upper molar—development of facet 1 and preprotocrista (or
paracrista, or premetacrista) on upper molars. (0) Facet 1 (prevallum crest) is
short, and does not extend to the stylocone area. (1) Wear facet 1 extends beyond
into the hook-like area near the stylocone. (2) Long preprotocrista (or paracrista or
premetacrista) is added to the prevallum shear and extends labially beyond
paracone (or metacone in metacone zalambdodonts).
Revised character definition to accommodate different forms of zalambdodonty. Original
coding for Necrolestes (1) and Cronopio (1) retained when assuming metacone
zalambdodonty.
285. Maximum vertical depth of zygomatic arch relative to length of skull (0) Between
10% and 20%. (1) Between 5% and 7%.
Maximum vertical depth of zygomatic arch relative to length of skull is 7.6% and 8.1% for
Notoryctes and Yalkaparidon respectively. Since both these values are closer to 7% than 10%,
both species are coded (1) and the character state definitions redefined to: (0) Between 8.5%
and 20%. (1) Between 5% and 8.5%.
3.3. Revised coding in Necrolestes and Cronopio
Characters were recoded in two different ways, assuming either that the lingual-most cusp
of the upper molars of Necrolestes and Cronopio is the metacone (metacone
zalambdodonty) or that the identity of this cusp is uncertain (uncertain zalambdodonty). Ten
characters were recoded under the assumption of metacone zalambdodonty; these changes
are described individually below. Under the assumption of uncertain zalambdodonty these
characters were recoded as (?) unless otherwise stated.
58. Development of postprotocrista on upper molars for double rank postvallum shear.
(0) Postprotocrista is short and does not extend labially beyond the metacone. (1)
Postprotocrista is long and extends labially beyond the metacone.
Assuming metacone zalambdodonty in Necrolestes and Cronopio, both protocone and
postprotocrista absent in upper molars. Coding in Necrolestes and Cronopio therefore
revised to inapplicable (-).
102. Relative height and size of paracone (cusp b) and metacone of upper molars. (0)
Paracone higher and larger than metacone. (1) Metacone is higher and larger than
paracone.
Assuming metacone zalambdodonty in Necrolestes and Cronopio, the most lingual cusp is
the metacone, and the paracone is absent. Coding in both species therefore revised to (1).
122. Bifid metastyle. (0) Absent. (1) Present.
The identity of the metastyle is ambiguous if metacone zalambdodonty is assumed in
Cronopio. Coding in Cronopio is therefore revised to (?).
125. Position of stylocone in posterior molariforms. (0) Along buccal edge.
(1) Separated.
The identity of the stylocone is ambiguous if metacone zalambdodonty is assumed in
Necrolestes and Cronopio. Coding in both species is therefore revised to (?).
126. Stylocone relationship in triangular teeth. (0) Stylocone connected to paracrista or
mesial to its end. (1) Stylocone distal to labial ending of paracrista. (2) Stylocone
detached of preparacrista occupying central position on crown.
The identity of the stylocone is ambiguous if metacone zalambdodonty is assumed in
Necrolestes and Cronopio. Coding in both species is therefore revised to (?).
127. Stylocone size in triangular teeth. (0) Absent. (1) Small stylar cusp. (2) Prominent
cusp subequal or larger than paracone.
The identity of the stylocone is ambiguous if metacone zalambdodonty is assumed in
Necrolestes and Cronopio. Coding in both species is therefore revised to (?).
129. Paracone orientation. (0) Erect. (1) Recumbent. (2) Procumbent.
Assuming metacone zalambdodonty in Necrolestes and Cronopio, the most lingual cusp is
the metacone, and the paracone is absent. Coding in both species therefore revised to
inapplicable (-).
130. Metacone. (0) Present. (1) Absent.
Assuming metacone zalambdodonty in Necrolestes and Cronopio, the most lingual cusp is
the metacone, and the paracone is absent. Coding in both species therefore revised to (0).
131. Paracone-metacone labiolingually aligned. Consider only in subtriangular upper
teeth: (0) Absent, the metacone is labial to paracone. (1) Present, the metacone is
approximately aligned mesiodistally with the paracone.
Assuming metacone zalambdodonty in Necrolestes and Cronopio, the most lingual cusp is
the metacone, and the paracone is absent. Coding in both species therefore revised to
inapplicable (-).
147. Position of lingual upper root. (0) Under paracone. (1) Under protocone or trigon.
(2) Under metacone.
If metacone zalambdodonty is assumed in Necrolestes and Cronopio, the lingual upper root
is positioned under the metacone, requiring addition of character state 2. Coding in
Necrolestes and Cronopio revised to new character state: (2) Under metacone.
If uncertain zalambdodonty is assumed, the lingual root could be positioned under either the
paracone or the metacone, but not under the protocone or trigon. Coding was, therefore,
revised to (0,2) in the case of uncertain zalambdodonty.
4. Tree lengths from the alternate parsimony analyses
Rougier et al. (2012) l = 1154
Including
Cronopio
Without Cronopio
Including new
metatherians
Without new
metatherians
1242
1153
1239
1150
1221
1132
1218
1129
Metacone
zalambdodonty
Uncertain
zalambdodonty
Metacone
zalambdodonty
Uncertain
zalambdodonty
5. Alternate Yalkaparidon analyses
Coding strategy
Following Archer et
al. (1988) and Beck
(2009)
Following Beck
(2008)
Following Beck’s
(2008) talonid
retention
hypothesis
Metacone zalambdodont
TL
MPT Topology
Uncertain zalambdodont
TL
MPT Topology
1242
37
Fig 1
1242
38
Identical to
Fig 1.
1246
66
Collapse nodes 49
& 50 of Fig 1.
1246
62
Collapse nodes
49 & 50 of Fig 1.
1241
60
Collapse nodes 49
& 50 of Fig 1.
1244
60
Collapse nodes
49 & 50 of Fig 1.
The a priori assumptions of primary lingual cusp homology in Yalkaparidon do not affect its
phylogenetic placement in this dataset, and only affect tree length if we assume the
presence of a vestigial talonid on the lower molars. Variant interpretations of Yalkaparidon
only serve to lessen resolution, but do not otherwise impact upon the tree.
6. Bayes factor analyses
6.1. Using stepping stone analyses to compare topologies
Bayes factors are calculated from the marginal likelihood, which was estimated here using
stepping stone analyses. In order to compare different hypotheses for the placement of
both Necrolestes and the Australosphenida, it was necessary to impose constraints upon our
analyses. By applying constraints to a Bayesian analysis, the amount of tree space available
to search is reduced by the prior on tree topology. This means that the posterior probability,
and thus marginal likelihood, of a constrained topology will be higher than an unconstrained
one, simply because there are fewer potential topological outcomes. Therefore, in order to
make a fair comparison of two different topological hypotheses, the tree should be fixed in
both instances, so that both analyses have a single possible topological outcome. If the
model used for each analysis is then identical, apart from the fixed topologies, the marginal
likelihoods should be comparable, and Bayes factors meaningful.
In order to perform such comparisons between the hypothesis that Necrolestes is a
meridiolestidan vs. the hypothesis that Necrolestes is a metatherian, and the different
hypotheses for the placement of Australosphenida, it was therefore essential to establish a
fully resolved topology. The topologies used for each analysis are detailed below:
Meridiolestidan Necrolestes: The most likely topology (found in the TreeProbs file) sampled
during an unconstrained Bayesian analysis was used as the fully resolved fixed topology for a
stepping stone analysis. This was consistent with the majority rule consensus tree presented
in Figure 3.
Metatherian Necrolestes: A Bayesian analysis was performed in which Necrolestes was
constrained to the Metatheria. The most likely topology (found in the TreeProbs file)
sampled during this analysis was used as the fully resolved fixed topology for a stepping
stone analysis.
Bayesian placement of the Australosphenida: The most likely topology (found in the
TreeProbs file) sampled during an unconstrained Bayesian analysis was used as the fully
resolved fixed topology for a stepping stone analysis. This was consistent with the majority
rule consensus tree presented in Fig. 3.
Parsimony placement of the Australosphenida: One of the 37 MPTs produced from a
parsimony analysis of the dataset (Fig. 1) was used as the fixed topology in this analysis.
6.2. Setting up the stepping stone analysis in MrBayes v3.2.2
Analyses used 31 million generations, with a sample frequency of 1000. The stepping stone
burnin was set to 1000 samples, producing a burnin run of 1,000,000 generations, sufficient
for all runs to reach convergence in the independent analyses. The stepping stone analysis
took 40 steps, meaning that each step lasted 750,000 generations. The settings are shown
below.
Mcmcp ngen=31000000 temp=0.25 samplefreq=1000;
ssp burninss=1000 nsteps=40;
6.3. Results: Marginal Likelihoods and Bayes factors
Model testing:
Table of Marginal likelihoods
Model
Run 1
Run 2
MK+G
-5052.65
-5052.51
MK
-5126.06
-5127.67
Table of Bayes factors (expressed as their natural logarithm)
MK+G run 1 MK+G run 2 MK run 1 MK run 2
MK+G run 1
0.14
73.41
75.02
MK+G run 2
73.55
75.16
MK run 1
1.61
MK run 2
Note that Bayes factors are reported in the manuscript as twice the natural logarithm.
Hypothesis testing:
Table of Marginal likelihoods
Topology
Run 1
Run 2
Unconstrained -4668.18
-4668.19
Metatherian
-4907.81
-4907.95
Parsimony
-4869.75
-4869.68
Table of Bayes factors (expressed as their natural logarithm)
Uncon.
Uncon.
Meta. Meta. Pars.
run 1
run 2
run 1
run 2
run 1
Unconstrained run 1 -0.01
239.63 239.62 201.57
Unconstrained run 2 239.77 239.76 201.5
Metatherian run 1
0.14
-38.06
Metatherian run 2
-38.2
Parsimony run 1
-
Pars.
run 2
201.56
201.49
-38.13
-38.27
-0.07
Parsimony run 2
-
-
-
-
-
-
Note that Bayes factors are reported in the manuscript as twice the natural logarithm.
7. Optimizations
7.1. Parsimony analysis: optimization of character states related to zalambdodonty
This table shows unambiguous optimization of character states related to zalambdodonty
(described in Section 3.3) across nodes A to D of the parsimony topology assuming uncertain
zalambdodonty (topology and node numbering identical to Fig. 1). In some cases, where
unambiguous optimization to a particular node could not be obtained, ACCTRAN and
DELTRAN node optimizations are given. These are marked * and presented in the format
ACCTRAN/DELTRAN. These optimizations suggest that loss of the metacone unites all taxa of
Node B, and hence, is consistent with the hypothesis that Necrolestes is paracone
zalambdodont.
Character
State
Node
58
No optimizations for this character
state across nodes A to D
-
102
122
125
126
127
129
130
131
147
No optimizations for this character
state across nodes A to D
0 - Bifid metastyle absent
1 - Stylocone positioned along
buccal edge of posterior molariform
2 - Stylocone detached of
preparacrista occupying central
position on crown
2 - Stylocone subequal or larger
than paracone
No optimizations for this character
state across nodes A to D
1 - Metacone absent
No optimizations for this character
state across nodes A to D
No optimizations for this character
state across nodes A to D
C/Leonardus*
A
B/B*
B
B
-
7.2. Parsimony analysis: optimization of character states across the
meridiolestidan lineage
This table shows unambiguous optimization of all character states (dental and non-dental)
across nodes A to D (Fig. 2) of the parsimony topology, assuming metacone zalambdodonty
in both Necrolestes and Cronopio. Additional characters optimized at these nodes by
ACCTRAN and DELTRAN optimization are also shown. Those characters which optimize
across these nodes in both ACCTRAN and DELTRAN optimization are highlighted in red and
blue for non-dental and dental characters, respectively.
Meridiolestidan Node (Fig. 2)
A
Non-dental
Characters
Unambiguous
Optimization
17:1;
5:2; 218:0;
285:0
Dental
Characters
46:1; 55:2;
108:1; 125:1
42:1; 48:1;
55:1; 60:2;
81:0; 82:1;
86:0; 111:1;
127:2; 137:1
Non-dental
Characters
1:0; 12:1;
171:1; 181:0;
188:0; 201:0;
219:0; 223:0;
247:1; 250:1;
262:1; 281:0;
291:1; 303:0
1:1; 2:1;
Dental
Characters
93:2; 106:1;
126:1; 140:1
34:1; 47:0;
52:2; 93:0;
126:2;
130:1; 152:1
ACCTRAN
Optimization
DELTRAN
Optimization
B
78:2; 108:2;
153:1;
154:1
9:3; 23:1;
246:0;
273:0;
288:1;
289:2;
296:0;
299:2
41:1;
122:0;
136:2;
139:1;
146:0;
156:1
12:1; 231:2;
247:1;
250:1; 312:1
Non-dental
Characters
Dental
Characters
C
76:0; 120:2;
123:1; 141:0
126:2;
47:0; 52:2;
140:1;
152:1
D
102:1;
134:0;
135:0;
147:2
2:0;
106:0;
130:0
9:3; 226:1;
246:0;
291:1;
299:2
41:1; 136:2;
139:1;
146:0
7.3. Parsimony analysis: optimization of character states related to tribospheny
This table shows unambiguous optimization of character states related to the acquisition of
tribospheny across our parsimony topology (see Fig 1). The node at which each particular
character state optimizes is presented. We found that none of these characters optimize
basal to the Australosphenida-Boreosphenida divergence, which is consistent with
independent acquisition of tribospheny in these two lineages. Therefore, the optimized
characters are presented in two columns, representing the separate acquisition of
tribospheny in the boreosphenidan and australosphenidan lineages. Where unambiguous
optimization to a particular node could not be obtained, ACCTRAN and DELTRAN node
optimizations are given. These are marked * and presented in the format
ACCTRAN/DELTRAN.
Stem and Crown
Australosphenida
Stem and Crown Boreosphenida
Character State
60
64
65
3 - Part of the talonid
occludes with the lingual
face of the upper molar
1 - Cristid obliqua present
2 - Cristid obliqua short
and pointed anteriorly
between metaconidprotoconid notch and the
protoconid.
-
66
1 - Talonid pre-entocristid
crest is in alignment with
the metaconid or with the
postmetacristid if the
latter is present
83
1 - Molar trigonid/talonid
width ratio: wide (talonid
is 40 - 70% of trigonid)
84
85
86
87
88
89
90
2 - Talonid is equal or
wider than trigonid (over
70% of width of trigonid)
1 - Lower molar hypoflexid
deep (but less than 50% of
talonid width)
Node
31
40
State
3 - Part of the talonid
occludes with the lingual
face of the upper molar
1 - Cristid obliqua present
Node
16
16
48
1 - Cristid obliqua
hypertrophied and directed
to posterior part of
metaconid
19/21*
-
2 - Cristid obliqua short and
pointed anteriorly between
metaconid-protoconid
notch and the protoconid.
20/20*
43
2 - Talonid pre-entocristid
crest is offset from the
metaconid and is lingual to
the base of the metaconid
16
42
2 - Talonid is equal or wider
than the trigonid (above
70% of the width of the
trigonid)
16
48
-
-
40
-
-
1 - Talonid basin present
2 - Rear portion of
molariform present as a
heel (with at least one
functional cusp)
4 - Rear portion of
molariform rimmed with
three major cusps
1 - Hypoconulid (=cusp d)
is elevated above the
cingulid level
1 - Hypoconid present
1 - Hypoconulid cusp tip
recumbent
1 - Entoconid present but
far from hypoconulid
42
1 - Lower molar hypoflexid
deep (but less than 50% of
talonid width)
2 - Very deep (>60% of
talonid width)
1 - Talonid basin present
16
21
16
31
3 - Present as a transverse
V-shaped basin with two
major cusps
21
43
-
-
1 - Hypoconulid (=cusp d) is
40
elevated above the cingulid
level
40
1 - Hypoconid present
Character not optimized in
44/44*
this lineage
1 - Entoconid present but
44
far from hypoconulid
15
16
16/17*
91
92
93
94
97
98
99
100
116
118
124
1 - Entoconid height
subequal to the
hypoconulid
1 - Paraconid, metaconid
and entoconid aligned
2 - Aspect ratio in occlusal
view of functional talonid
basin: wider than long
1Hypoconulid/protoconid
height ratio between 25%
and 35%
Character not optimized in
this lineage
1 - Functional lingual
protocone that grinds
against basin on lowers
present
Character not optimized in
this lineage
1 - Distance from
protocone apex >0.60 of
total width
1 - Moderate
development of lingual
region of upper molars
1 - Differentiation of wear
facet 3 and wear facet 4
present
Wear pattern on talonid
present
Character not optimized in
this lineage
2 - Present and twinned
with hypoconulid
1 - Entoconid height
44
subequal to the
hypoconulid
1 - Paraconid, metaconid
47
and entoconid aligned
2 - Aspect ratio in occlusal
48/50* view of functional talonid
basin: wider than long
20/20*
19/20*
19/20*
15
40
1 - Hypoconulid/protoconid
height ratio between 25%
and 35%
16
-
3 - Hypoconulid/protoconid
height ratio: 50% or higher
19
42
Character not optimized in
this lineage
-
-
Character not optimized in
this lineage
-
45
Character not optimized in
this lineage
-
45
Character not optimized in
this lineage
-
40
43
-
1 - Differentiation of wear
facet 3 and wear facet 4
present
Character not optimized in
this lineage
1 - Multiple ridges within
talonid basin
16
20
10. Bayesian analysis: ancestral character state reconstruction
These tables show reconstruction of ancestral character states related to acquisition of
tribospheny at each of four nodes (W, X, Y, and Z) in our Bayesian topology (Fig. 3). Where
reconstructed character states differ from those at Node W, these are shown in red.
Node W
Character
Reconstructed
Description
State
60
3
64
1
65
0
66
0
83
0
84
1
85
Equivocal
86
2
87
88
89
1
1
Equivocal
90
Equivocal
91
Equivocal
92
Equivocal
93
Equivocal
94
1
97
0
98
99
Equivocal
0
100
Equivocal
116
118
123
124
130
1
0
0
0
0
Part of the talonid occludes with the lingual face (or any
part) of the upper molar)
Cristid obliqua present
Orientated to or lingual to the metaconid-protoconid
notch
Talonid lacks medial and longitudinal crest
Molar trigonid/talonid width ratio: narrow or absent
(talonid <40 % of trigonid)
Lower molar hypoflexid deep (but less than 50% of talonid
width)
Talonid basin presence equivocal
Rear portion of molariform present as a heel (with at least
one functional cusp)
Hypoconulid (=cusp d) is elevated above the cingulid level
Hypoconid present
Hypoconulid orientation is equivocal
Entoconid presence and/or distance from hypoconulid is
equivocal
Height of entoconid in comparison with other talonid
cusps is equivocal
Alignment of paraconid, metaconid and entoconid is
equivocal
Aspect ratio in occlusal view of functional talonid basin is
equivocal
Hypoconulid/protoconid height ratio between 25% and
35%
Functional lingual protocone that grinds against basin on
lowers absent
Trigon basin presence is equivocal
Transverse width of protocone narrow
Anteroposterior development of lingual region of upper
molars is equivocal
Differentiation of wear facet 3 and wear facet 4 present
Wear pattern on talonid absent
Distal metacristid present
Smooth surface on the talonid (or on cusp d)
Metacone present
Node X
Character
Reconstructed
Description
State
60
3
64
1
65
0
66
83
0
Equivocal
84
1
85
Equivocal
86
2
87
88
89
1
1
Equivocal
90
Equivocal
91
Equivocal
92
Equivocal
93
Equivocal
94
1
97
1
98
99
1
0
100
Equivocal
116
118
123
124
130
1
0
0
0
0
Part of the talonid occludes with the lingual face (or any
part) of the upper molar)
Cristid obliqua present
Orientated to or lingual to the metaconid-protoconid
notch
Talonid lacks medial and longitudinal crest
Molar trigonid/talonid width ratio equivocal
Lower molar hypoflexid deep (but less than 50% of talonid
width)
Talonid basin presence equivocal
Rear portion of molariform present as a heel (with at least
one functional cusp)
Hypoconulid (=cusp d) is elevated above the cingulid level
Hypoconid present
Hypoconulid orientation is equivocal
Entoconid presence and/or distance from hypoconulid is
equivocal
Height of entoconid in comparison with other talonid
cusps is equivocal
Alignment of paraconid, metaconid and entoconid is
equivocal
Aspect ratio in occlusal view of functional talonid basin is
equivocal
Hypoconulid/protoconid height ratio between 25% and
35%
Functional lingual protocone that grinds against basin on
lowers present
Trigon basin present
Transverse width of protocone narrow
Anteroposterior development of lingual region of upper
molars is equivocal
Differentiation of wear facet 3 and wear facet 4 present
Wear pattern on talonid absent
Distal metacristid present
Smooth surface on the talonid (or on cusp d)
Metacone present
Node Y
Character
Reconstructed
Description
State
60
3
64
1
65
0
66
1
83
1
84
1
85
1
86
4
87
88
89
1
1
Equivocal
90
Equivocal
91
Equivocal
92
Equivocal
93
Equivocal
94
1
97
1
98
99
1
0
100
Equivocal
116
118
123
124
130
1
1
0
0
0
Part of the talonid occludes with the lingual face (or any
part) of the upper molar)
Cristid obliqua present
Orientated to or lingual to the metaconid-protoconid
notch
Talonid pre-entocristid crest is in alignment with the
metaconid or with the postmetacristid if the latter is
present
Molar trigonid/talonid width ratio: wide (talonid is 40 70% of trigonid)
Lower molar hypoflexid deep (but less than 50% of talonid
width)
Talonid basin present
Rear portion of molariform rimmed with three major
cusps
Hypoconulid (=cusp d) is elevated above the cingulid level
Hypoconid present
Hypoconulid orientation is equivocal
Entoconid presence and/or distance from hypoconulid is
equivocal
Height of entoconid in comparison with other talonid
cusps is equivocal
Alignment of paraconid, metaconid and entoconid is
equivocal
Aspect ratio in occlusal view of functional talonid basin is
equivocal
Hypoconulid/protoconid height ratio between 25% and
35%
Functional lingual protocone that grinds against basin on
lowers present
Trigon basin present
Transverse width of protocone narrow
Anteroposterior development of lingual region of upper
molars is equivocal
Differentiation of wear facet 3 and wear facet 4 present
Wear pattern on talonid present
Distal metacristid present
Smooth surface on the talonid (or on cusp d)
Metacone present
Node Z
Character
Reconstructed
Description
State
60
3
64
1
65
0
66
2
83
2
84
1
85
1
86
4
87
88
89
1
1
0
90
Equivocal
91
Equivocal
92
0
93
Equivocal
94
1
97
Equivocal
98
99
Equivocal
Equivocal
100
Equivocal
116
118
123
124
130
1
0
Equivocal
0
Equivocal
Part of the talonid occludes with the lingual face (or any
part) of the upper molar)
Cristid obliqua present
Orientated to or lingual to the metaconid-protoconid
notch
Talonid pre-entocristid crest is offset from the metaconid
and is lingual to the base of the metaconid
Talonid is equal or wider than the trigonid (above 70% of
the width of the trigonid)
Lower molar hypoflexid deep (but less than 50% of talonid
width)
Talonid basin present
Rear portion of molariform rimmed with three major
cusps
Hypoconulid (=cusp d) is elevated above the cingulid level
Hypoconid present
Hypoconulid cusp tip erect or procumbent
Entoconid presence and/or distance from hypoconulid is
equivocal
Height of entoconid in comparison with other talonid
cusps is equivocal
Paraconid, metaconid and entoconid not aligned
Aspect ratio in occlusal view of functional talonid basin is
equivocal
Hypoconulid/protoconid height ratio between 25% and
35%
Presence of functional lingual protocone that grinds
against basin on lowers is equivocal
Trigon basin presence is equivocal
Transverse width of protocone equivocal
Anteroposterior development of lingual region of upper
molars is equivocal
Differentiation of wear facet 3 and wear facet 4 present
Wear pattern on talonid absent
Distal metacristid presence is equivocal
Smooth surface on the talonid (or on cusp d)
Metacone presence equivocal
11. References
Asher RJ, Horovitz I, Martin T, Sánchez-Villagra M (2007) Neither a rodent nor a platypus: a
reexamination of Necrolestes patagonensis Ameghino. Am Mus Novitates 3546:1-40
Asher RJ, Sánchez-Villagra M (2005) Locking yourself out: diversity among dentally
zalambdodont therian mammals. J Mammal Evol 12:265-282
Archer M, Beck R, Gott M, Hand S, Godthelp H, Black K (2011) Australia’s first fossil
marsupial mole (Notoryctemorphia) resolves controversies about their evolution and
palaeoenvironmental origins. P R Soc B 278:1498–1506
Archer M, Hand S, Godthelp H (1988) A new order of Tertiary zalambdodont marsupials.
Science 239:1528-1531
Beck RMD (2009) Was the Oligo-Miocene Australian metatherian Yalkaparidon a
‘mammalian woodpecker’? Biol J Linn Soc 97:1–17
Beck RMD, Travouillon KJ, Aplin KP, Godthelp H, Archer M (2013) The osteology and
systematics of the enigmatic Australian Oligo-Miocene metatherian Yalkaparidon
(Yalkaparidontidae; Yalkaparidontia; ?Australidelphia; Marsupialia). J Mammal Evol DOI
10.1007/s10914-013-9236-3
Gelfo JN, Goin FJ, Woodburne MO, Muizon C de (2009) Biochronological relationships of the
earliest South American Paleogene mammalian faunas. Palaeontology 52:251-269
Gelfo JN, Pascual R (2001) Peligrotherium tropicalis (Mammalia, Dryolestida) from the early
Paleocene of Patagonia, a survival from a Mesozoic Gondwanan radiation. Geodiversitas 23:
369–379
Ladevéze S, Asher RJ, Sánchez-Villagra MR (2008) Petrosal anatomy in the fossil mammal
Necrolestes: evidence for metatherian affinities and comparisons with the extant marsupial
mole. J Anat 213:686-697
Long JA, Archer M, Flannery T, Hand S (2002) Prehistoric Mammals of Australia and New
Guinea: One Hundred Million Years of Evolution. Johns Hopkins University Press, Baltimore
Perkins ME, Fleagle JG, Heizler MT, Nash B, Bown TM, Tauber AA, Dozo MT (2012)
Tephrochronology of the Miocene Santa Cruz and Pinturas Formations, Argentina. In:
Vizcaíno SF, Kay RF, Bargo MS (eds) Early Miocene Paleobiology in Patagonia. Cambridge
University Press, Cambridge
Rodgers J (2008) Notoryctes typhlops Digital Morphology
http://digimorph.org/specimens/Notoryctes_typhlops/
Rougier GW, Wible JR, Beck RMD, Apesteguia S (2012) The Miocene mammal Necrolestes
demonstrates the survival of a Mesozoic nontherian lineage into the late Cenozoic of South
America. Proc Natl Acad Sci USA 109:19871-19872
Szalay F (1994) Evolutionary History of the Marsupials and an Analysis of Osteological
Characters. Cambridge University Press, Cambridge
Ungar PS (2010) Mammal Teeth: Origin, Evolution and Diversity. Johns Hopkins University
Press, Baltimore
Download