Assembling an illustrated family-level tree of life for exploration in mobile devices

Abstract

Since the concept of the tree of life was introduced by Darwin about a century and a half ago, a considerable fraction of the scientific community has focused its efforts on its reconstruction, with remarkable progress during the last two decades with the advent of DNA sequences. However, the assemblage of a comprehensive tree of life for its exploration has been a difficult task to achieve due to two main obstacles: i) information is scattered into a plethora of individual sources and ii) practical visualization tools for exceptionally large trees are lacking. To overcome both challenges, we aimed to synthetize a family-level tree of life by compiling over 1400 published phylogenetic studies, ensuring that the source trees represent the best phylogenetic hypotheses to date based on a set of objective criteria. Moreover, we dated the synthetic tree by employing over 550 secondary-calibration points, using publicly available sequences for more than 5000 taxa, and by incorporating age ranges from the fossil record for over 2800 taxa. Additionally, we developed a mobile app (Tree of Life) for smartphones in order to facilitate the visualization and interactive exploration of the resulting tree. Interactive features include an easy exploration by zooming and panning gestures of touch screens, collapsing branches, visualizing specific clades as subtrees, a search engine, a timescale to determine extinction and divergence dates, and quick links to Wikipedia. Small illustrations of organisms are displayed at the tips of the branches, to better visualize the morphological diversity of life on earth. Our assembled Tree of Life currently includes over 7000 taxonomic families (about half of the total family-level diversity) and its content will be gradually expanded through regular updates to cover all life on earth at family-level.

Full text

1
Assembling an illustrated family-level tree of life for
exploration in mobile devices
Andrés A. Del Risco, Diego A. Chacón, Lucia Ángel, David A. García
Corresponding Author: Andrés A. Del Risco. E-mail: andresdelrisco29@gmail.com
ABSTRACT
Since the concept of the tree of life was introduced by Darwin about a century and a half ago,
a considerable fraction of the scientific community has focused its efforts on its reconstruction,
with remarkable progress during the last two decades with the advent of DNA sequences.
However, the assemblage of a comprehensive tree of life for its exploration has been a difficult
task to achieve due to two main obstacles: i) information is scattered into a plethora of
individual sources and ii) practical visualization tools for exceptionally large trees are lacking.
To overcome both challenges, we aimed to synthetize a family-level tree of life by compiling
over 1400 published phylogenetic studies, ensuring that the source trees represent the best
phylogenetic hypotheses to date based on a set of objective criteria. Moreover, we dated the
synthetic tree by employing over 550 secondary-calibration points, using publicly available
sequences for more than 5000 taxa, and by incorporating age ranges from the fossil record for
over 2800 taxa. Additionally, we developed a mobile app (Tree of Life) for smartphones in
order to facilitate the visualization and interactive exploration of the resulting tree. Interactive
features include an easy exploration by zooming and panning gestures of touch screens,
collapsing branches, visualizing specific clades as subtrees, a search engine, a timescale to
determine extinction and divergence dates, and quick links to Wikipedia. Small illustrations of
organisms are displayed at the tips of the branches, to better visualize the morphological
diversity of life on earth. Our assembled Tree of Life currently includes over 7000 taxonomic
families (about half of the total family-level diversity) and its content will be gradually
expanded through regular updates to cover all life on earth at family-level.
Keywords: Tree of Life, Phylogeny, Visualization, Mobile App, Exploration, Evolution
INTRODUCTION
The fact that all life on Earth is
connected through common ancestry
across a single genealogical tree has
fascinated scientists for over one and a
half century (Darwin, 1859). Since then,
reconstructing the tree of life has become
one of the most challenging goals in the
biological sciences, prompting the
emergence of phylogenetics as a
fundamental subdiscipline of
evolutionary biology.
Even though the scientific
community has made substantial
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

2
advances towards this goal, especially
during the last two decades following the
advent of molecular systematics (Pennisi
2013), two major obstacles still stand tall
in the way of providing a unified and
comprehensive tree of life to browse
through the evolutionary relatedness of
all life on Earth. Firstly, phylogenetic
reconstructions are usually focused on
very small fractions of the whole tree of
life, so the available fragments of
information are widely scattered across
thousands of scientific publications which
in turn are buried in the often paywalled
primary literature. Secondly, due to the
high number of terminal branches, having
a fully assembled tree of life would
require effective and practical ways to
navigate and visualize a phylogenetic
diagram of such large dimensions.
Some important efforts have been
undertaken to tackle both obstacles
(although separately) with varying
degrees of success. Early projects aiming
to reconstruct the tree of life involved the
compilation of phylogenies of specific
groups of taxa into sectional, hyperlinked
webpages (Maddison et al. 2007) or books
(Lecointre & Le Guyader 2006; Hedges &
Kumar 2009), but these were eventually
superseded by the most comprehensive
initiative to synthesize phylogenetic and
taxonomic information into a single
source tree: The Open Tree of Life
(hereafter, OTOL: Hinchliff et al. 2015).
Although the establishment of the OTOL
project represents a huge step in
overcoming the first obstacle by the
extensive compilation of phylogenies, it is
not devoid of shortcomings. Firstly, the
construction of its constantly updating
synthetic tree mainly depends upon
published phylogenies in the form of tree
files, which are acquired from either
public data repositories or by direct
submission by authors. Unfortunately,
less than a fifth of the studies provide
such files (Drew 2013), meaning that most
of the potentially valuable phylogenetic
information that is presented in the form
of figures in published articles are yet to
be incorporated into its synthetic tree.
Secondly, the OTOL synthetic tree does
not include the temporal component
(time-proportional branch lengths) of the
Tree of Life, so this information must be
consulted elsewhere (e.j., Kumar et al.
2017). Finally, the OTOL project currently
aims to reconstruct the evolutionary
relationships among living taxa, so extinct
taxa known from the vast fossil record has
yet to be incorporated into the synthesis.
On the other hand, while many
programs that allow users to visualize and
navigate tree files have been available for
over a decade (Page 2011), only two
recently developed web-based tools have
attained the goal of providing a complete
visualization and exploration of large
phylogenies: Onezoom (Rosindell &
Harmon 2012) and LifeMap (de Vienne
2016). Both allow users to smoothly
explore the tree of life by employing
zooming features, and by being also
available in the form of mobile apps, they
take advantage of touch screens for a
more intuitive exploration. These tools
were solely developed for visualization,
and hence depend on external datasets to
explore a given phylogeny, relying on the
synthetic tree compiled by the OTOL or
the NCBI taxonomic tree. The main
limitation of these approaches is the
visual formats in which trees are
presented; even though they are
aesthetically appealing, they greatly
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

3
deviate from the conventional diagrams
in which phylogenies are presented
(cladograms, phylograms, timetrees, etc.)
and thus hinder the proper inference of
the topological relationships depicted in
the tree. Additionally, another
consequence of these distinctive
visualization formats is their inability to
incorporate a temporal dimension in the
form of branch lengths, which is necessary
to determine extinction and divergence
dates.
With the aim of overcoming both
obstacles while avoiding some of the
mentioned pitfalls of previous projects,
we introduce the Tree of Life App: a novel
tool to visualize and interactively explore
a thoroughly compiled, time-calibrated
family level-tree of life through a mobile
app. We aim to establish this tool as a
reference resource for scientific
consultation, educational purposes, and
even casual exploration for the general
public. Tree of Life App is publicly
available in the Google Play Store and
iPhone AppStore, and it is constantly
being updated as new phylogenetic
studies are published.
METHODS
The general procedure for the assemblage
of the family-level tree of life consisted
mainly in a tree grafting approach, in
which higher-level phylogenies are pasted
into lower-level phylogenies. The whole
procedure, from the initial lists of taxa to
the time-calibrated tree file, is depicted in
Figure 1 and is thoroughly explained in the
sections below.
Taxonomic Input
To assemble the family-level tree of life,
the first step was establishing
comprehensive taxonomic lists for large
groups of organisms based on clade-
specific, published taxonomic backbones
covering at least family-ranked taxa. We
employed the comprehensive family-level
classification of living organisms of
Ruggiero (2014) as a starting point and as
the default taxonomic backbone for
groups without specialized proposals.
Although different taxonomic proposals
or treatments may exist for a given group,
we chose what we consider to be the best
taxonomic backbone based on their
congruence with current phylogenetic
hypotheses (i.e., recognition of mostly
monophyletic families). Each family-level
taxonomic backbone is then cited in and
displayed in the secondary information
window of its corresponding supra-
familial taxon. For any given group, a
taxonomic family list may or may not be a
composite of more than one source. Such
is the case for groups that have a rich
fossil record, as are most major
vertebrate and arthropod clades, among
others. In these cases, taxonomic lists are
usually composed of a source for extant
families, and another one for extinct
families (which are often excluded in the
former). For most groups, fossil families
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

4
Figure 1. Summary workflow of the tree assembling process. Major steps are enclosed in gray circles, which
encompass (1) the establishment of the taxonomic backbones from the primary literature, (2) the obtention
of phylogenetic studies, (3) ranking phylogenetic hypotheses given a set of criteria to select the most suitable
trees, (4) the transcription of the selected higher ranked phylogenies (and the placement of missing taxa from
others) to a Newick file, and (5) the assignation of branch lengths based on a comprehensive tree-dating
analysis with BEAST2, resulting in the final synthesized tree.
are taken from the PaleoBiology Database
(hereafter PBDB; http://paleobiodb.org)
taxonomy, unless otherwise specified in
secondary information windows. Taxa
that are currently unassigned to a family-
level group, such as genera incertae sedis
or fossil taxa that do not always employ
Linnaean ranks are not included in the
procedure.
Phylogenetic arrangements
The second step in the procedure is
arranging taxa into cladograms by
manually writing text files in Newick tree
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

5
notation. These topological arrangements
are direct transcriptions of phylogenies
published as figures in the primary
literature that were identified and
collected after a comprehensive search
with the Google Scholar engine. By using
figures rather than tree files, we avoid the
overdependence of available datasets in
public repositories, which are lacking for
over 80% of the published phylogenetic
knowledge (Drew et al. 2013). Family-
level phylogenies were separately built in
this way for lower-level clades and were
then progressively condensed into larger
and higher-level trees, whose backbone
topologies were inferred and manually
written following the same approach
described previously. Because the
operational taxonomic units (OTUs) of
published phylogenies are mostly species-
level, terminals are tagged with its
corresponding family-level taxon (or
higher rank if the cladogram of a deep
level phylogeny is being constructed)
prior to the transcription of the figures to
the Newick trees, and suprafamilial clade
names were also assigned to their
corresponding nodes.
During the transcription of
phylogenies to Newick notation, poorly
supported nodes were collapsed to
polytomies when support values are
below commonly employed thresholds
(70 for bootstrap, 0.95 for Bayesian
posterior probabilities, and 3 for Bremer
support [Siddall 2002]).
Ranking phylogenies
The presence of multiple phylogenetic
hypotheses for a given set of taxa seems
to be the rule rather than the exception in
the scientific literature, especially for
biologically well studied taxa. Different
phylogenetic trees are found either in
different published sources, or even
within the same study, in which more
than one approach is often used to
produce a phylogeny. Considering that
such hypotheses are not equally accurate,
we assigned a set of criteria for ranking
phylogenies for ultimately displaying
what we consider the best available tree.
Such criteria are based on whether the
tree was reconstructed through a cladistic
analysis or not, the underlying data for
tree reconstruction (type and number of
characters), and the inference method, as
summarized in Table 1 and as described
below.
First, we prioritize trees produced
through any character-based cladistic
analysis, followed by tentative
cladograms proposed by authors. In the
absence of any published phylogeny for a
given group, taxonomy is then used as a
proxy for phylogeny.
Second, character-based cladistic analysis
are ranked according to the type of
characters employed, in which molecular
characters are prioritized over
morphological ones given their practical
and methodological advantages (Scotland
et al. 2003; Vogt et a. 2010). Within
molecular analyses, we favor nucleotide
characters over aminoacids due to the
usually greater informativeness of the
former (Townsend et al. 2008). Within
morphological phylogenetic analyses, we
favor those implementing molecular
scaffolds (Lee & Palci 2015) over
unconstrained ones.
Third, phylogenies are prioritized based
on the number of characters employed
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

6
for their reconstruction, thus favoring the
recent phylogenomic analyses over
traditional single-locus or multi-locus
approaches. Although it has been shown
that a higher amount of characters does
not always translate into a more accurate
species tree (e.g., Zhang et al. 2021) and
are still prone to different sources of
systematic error (Philippe et al. 2017;
Reddy et al. 2017), we believe that the
advantages of larger datasets outweight
their drawbacks in most cases, especially
considering that recent phylogenomic
studies are thoroughly analyzing their
data with scrutiny to account for potential
misleading issues (e.g., Arcila et al. 2017;
Betacur-R et al. 2019; Redmong &
McLysaght 2021; Zhang et al. 2021).
Higher taxon sampling is used as a
tiebreaker in instances of equal number of
characters, due to its positive impact in
breaking long branches and improving
model selection (Nabhan & Sarkar 2012).
Fourth, phylogenies are ranked according
to the methodological approach for tree
inference, in which coalescent-based
approaches are favored over
concatenation approaches, due to the
former accounting for variation in gene
histories by confounding processes such
as incomplete lineage sorting (Kubatko &
Degnan 2007; Rock & Steel 2015; Jiang et
al. 2020). Within concatenation
approaches, model-based methods
(Bayesian inference and maximum
likelihood) are prioritized (Gadagkar &
Kumar 2005; Philippe et al. 2005; Puttick
et al. 2019), followed by maximum
parsimony and distance methods
(neighbor joining, minimum evolution).
Additionally, analyses that account for
further artefacts or systematic errors
detected in previous reconstructions are
also favored (e.g., compositional
heterogeneity across sites in Puttick et al.
2018; long branch attraction in Puttick et
al. 2014).
1. OVERALL
APPROACH
2. CHARACTER TYPE
3. DATA QUANTITY
4. TREE INFERENCE METHOD
1.1. Clastistic
(character-based)
analysis
2.1. Molecular
2.1.1. Nucleotides
3.1. Number of characters
4.1. Coalescent approaches
2.1.2. Aminoacids
4.2. Concatenation
approaches
4.2.1. Model based approaches
(Maximum likelihood & Bayesian
inference)
2.2 Morphological
2.2.1. Molecular
scaffold
3.2. Number of taxa
(tiebreaker when number
of characters are equal)
4.2.2. Maximum parsimony
and distance methods
2.2.2. Unconstrained
1.2. Tentative phylogeny
1.3. Taxonomy
Table 1. Ranking system for selection of phylogenetic hypotheses. The numerical arrangement of each
attribute reflects the priorities employed for the selection, starting from the overall approach for the depiction
of the phylogeny (in which cladistic character-based analysis are preferred over tentative phylogenies and
taxonomic hierarchies), to the last step consisting in the tree inference methodology.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

7
Figure 2. Approach for placing missing taxa in higher ranking topologies. The first scenario consists in (a) a
lower-ranking tree which includes the missing taxon X and its sister clade enclosed in gray, and (b) a higher-
ranking tree (without taxon X) with this same sister clade also enclosed in gray. In the resulting tree (c), the
missing taxon X is then added to the higher ranked topology as sister to the gray enclosed clade. Note that
the gray clade is in (a) and (b) is the same in terms of its circumscription (includes a, b, c and d), although its
internal topology is different. The second scenario also consists of (d) a lower-ranking tree which includes the
missing taxon X and its sister clade enclosed in gray, but in the (e) higher ranking phylogeny, this clade does
not exist, as its members (a, b and c) are distantly related, conforming a polyphyletic group. In these instances,
the node representing the most recent common ancestor between members of the gray-enclosed clade in
the lower tree is spotted in the higher tree (gray arrow). Then, the missing taxon X is placed as a direct
descendant of such node in the final topology (f), resulting in a polytomy.
Placement of unsampled taxa
Complete taxon sampling is seldom
achieved for a single reconstructed
phylogeny, which is especially the case for
analyses encompassing high amounts of
characters such as phylogenomic ones.
Consequently, the exclusion of several
taxa would be mostly prevalent in the
higher ranked phylogenies according to
the criteria for choosing a phylogenetic
hypothesis. To tackle this issue, we looked
for the lower-ranked phylogenies which
included each of the missing taxa in the
highest-ranked one to infer their
phylogenetic placement and incorporate
them in the transcribed Newick trees. This
step is not straightforward since the
topology of such phylogenies may differ
from the higher ranked one. In the
simplest scenario, to assign the placement
of a missing taxon we first identified its
sister clade in the lower-ranked
phylogeny. Then we looked for such clade
in the higher-ranked phylogeny and
situated the missing taxon as its sister in
the Newick tree (Figure 2a). More
complex scenarios arise when the sister
clade of such a taxon is absent (i.e., its
constituent taxa does not conform a
monophyletic group) in the higher-ranked
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

8
phylogeny. In such instances, we first
tagged the constituent OTUs of such a
sister clade in the higher-ranked
phylogeny, and then we mapped the node
representing their most recent common
ancestor. The missing taxon is then placed
as a direct descendant of such node,
resulting in a polytomy (Figure 2b). This is
largely the case for assessing the
phylogenetic position of fossil taxa, which
are mostly derived from morphological
phylogenies whose topologies are
sometimes drastically different from
molecular ones.
Non-monophyletic families
One of the consequences of synthesizing
supra-specific level phylogenies is dealing
with non-monophyletic terminals. This is
not uncommon, since for some groups of
organisms’ taxonomic classifications have
not yet been amended to reflect current
knowledge on phylogeny. Non-
monophyletic terminals (families in this
case) are problematic due to the difficulty
of representing accurate topological
relationships. Here, we approach this
issue with what we consider to be the
most conservative solution, as illustrated
in Figure 3. For paraphyletic families in
which other families of clades are nested
within them, we simply allocate such
family as sister to their nested lineages
(Figure 3a). While the resulting topology
would be indistinguishable from one in
which all terminals are monophyletic, the
paraphyletic status of the family is
denoted in the Remarks field of the taxon
metadata window (see Taxon Metadata
for further details)
For polyphyletic families whose members
are distantly related, we pseudo-
replicated the source phylogeny N times,
where N is the number of distantly related
clades belonging to the family. In each
pseudo-replicate, only one of such clades
is represented, while the others are
pruned from the tree. A strict consensus
is then applied between all pseudo-
replicates (Figure 3b). It is worth noting
that this approach may lead to highly
polytomic trees when lineages of a
polyphyletic family are very distantly
located from each other, even when the
original phylogeny is highly resolved. This
is the case for a few parts of our current
synthesized tree such as Microsporidia, in
which the highly polyphyletic family
Thelohaniidae “breaks up” some of the
resolved relationships among other
families. This highlights the urgent need
to carry out further systematic studies
with broader taxon sampling to establish
proper taxonomic arrangements. As with
paraphyletic families, such instances of
polyphyly are also denoted in the taxon
metadata text file (see corresponding
section below).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

9
Figure 3. Topological placement of non-monophyletic terminals. a) Source phylogeny with
paraphyletic terminal “D”, which in the resulting topology is simply accommodated as the
sister lineage of the terminal nested within its grade, which in this case is the E terminal. b)
Source phylogeny with polyphyletic terminal “D”. Two pseudo-replicate trees are
generated, each with an isolated “D” lineage in one of its two topological positions. A strict
consensus is applied for the pseudo-replicates, resulting in a final topology in which “D” is
placed in a polytomy. Note the collapse of other nodes in the resulting consensus trees due
to the distant position of the “D” lineages”.
Time calibration
Dating the entire tree of life poses as an
even more difficult challenge than the
assemblage of a solely topological tree.
The relative scarcity of published time-
calibrated phylogenies is the main reason
behind such difficulty, leading to a patchy
distribution of divergence dates and clade
ages estimates across the tree of life.
Moreover, the high degree of variation in
age estimates across analyses and studies
for a given clade also makes this task even
more difficult; there are instances in
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

10
which the age estimate for a clade in a
study may be older than the age estimates
of its parent clades in another study,
which would result in negative branch
lengths if assembled in a single tree. This
is a serious problem when attempting to
synthesize a unified time-calibrated tree
of life by compiling estimates from
different sources.
To partially overcome such difficulties, we
assigned branch length values to our
synthesized tree by following two main
steps: 1) compiling node age estimates for
all possible clades present in such tree by
consulting the TimeTree database
(Hedges et al. 2009; Hedges et al. 2015;
Kumar et al. 2017), and 2) estimating
dates of the remaining nodes by using
TimeTree data as secondary calibrations,
incorporating taxon age ranges from the
fossil record and DNA sequences.
We find the TimeTree database to be a
suitable source for extracting node age
estimates due to (i) its extensive
collection of data from the literature
under properly selective criteria, (ii) the
mean values reported for divergence
dates between different sources, and (iii)
the consistency tests performed to avoid
negative branch lengths. Data collection
was manually performed between June
2017 and December 2019 by querying
taxon names, as the whole database is not
available for download. Despite the strict
selection criteria employed by the authors
to assemble the database, we
nonetheless filtered out node ages that
were younger than the oldest occurrence
of the current fossil record.
To estimate branch lengths and
divergence dates of the remaining nodes
not covered in the TimeTree database, we
compiled age range estimates from the
fossil record for each family by consulting
the PBDB between 2014 and 2021. Age
ranges obtained from alternate sources
were cited for each taxon in the taxon
metadata text file (see next Methods
subsection) for their eventual display in
secondary information windows. We
carefully excluded some spurious records
for taxa whose ages dramatically
expanded their ranges (i.e., were too old
or too young) by revising the recent
literature to address whether those
records represent dubious taxonomic
assignations. The resulting age ranges for
each taxon were then used to filter out
some of the node ages compiled in the
previous step.
Nucleotide sequences were also obtained
from GenBank for every family in the
synthetic tree present in the database.
Because there is little overlap of
homologous loci between different major
groups across the tree of life, we
partitioned the tree in 20 subsections,
with some of them nested within others.
For each partition, we selected the set of
loci that maximizes taxon coverage, so for
some partitions more than one locus was
necessary (Table 2). The procedures
described hereafter were performed
separately for each tree partition.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

11
Table 2. Tree partitions and their respective loci employed for the phylogeny-dating analysis in BEAST2.
Nested partitions are indicated within parentheses. Accession numbers for family-level taxa are listed in
Supplementary Table 2. *Metazoa and Hexapoda have nested clades that were treated as separate partitions
(Porifera and Endopterygota, respectively) despite the use of the same set loci. This was necessary due to
their high number of terminal taxa and the consequent difficulty in running the analysis with large data
matrices.
Sequences were aligned and
concatenated (when presenting more
than one locus) with Geneious® v5.4
(Drummond et al., 2011) using the
MUSCLE algorithm (Edgar 2004), and
conflicts were manually resolved by
visually inspecting the resulting
alignments. For a total evidence dating
analysis of tree partitions, alignments
were imported to the utility program
BEAUTi for setting the parameters of
BEAST2 (Bouckaert et al. 2014) utilizing a
fossilized birth-death model (Heath et al.
2014). To add taxa with reported age
ranges from the fossil record and no
molecular data, we edited the resulting
alignments for manually adding such
taxon names with empty sequences. Ages
for fossil taxa were set (in millions of
years) as the younger bound values of
Tree partition
Loci
Prokaryotes (Bacteria and the "archaean" grade)
16s
Eukaryota (excluding Chlorophyta, Streptophyta, Amoebozoa, Fungi and Metazoa)
18s
Metazoa* (excluding Porifera, Xenocarida, Hexapoda and Vertebrata)
18s
Porifera
18s
Xenocarida
COI, H3
Hexapoda* (excluding Endopterygota)
COI, 18s
Endopterygota
COI, 18s
Vertebrata (excluding Chondrichthyes, Actinopterygii, Squamata and Passeriformes)
Cytb
Chondrichthyes
Cytb, NADH2
Actinopterygii (excluding Characiphysi)
COI
Characiphysi
Cytb, COI, 12s
Squamata
Cytb, 16s
Passeriformes
Cytb, COI, RAG1, Myo
Chlorophyta
rbcL, 18s
Streptophyta (excluding Bryophyta and Malvales)
rbcL
Bryophyta
rbcL, nad5
Malvales
rbcL, 18s
Fungi (excluding Ascomycota and Chytridiomycota)
18s, 28s
Ascomycota
18s, 28s
Amoebozoa
18s, Actin
Abbreviations: 12S=12S ribosomal RNA, 16S= 16S ribosomal RNA, 18S= 18S ribosomal RNA, 28S=28S ribosomal RNA, COI= cytochrome oxidase
subunit I, Cytb=cytochrome b, H3= Histone 3, Myo=myoglobin, NADH2=NADH dehydrogenase subunit 2, nad5=NADH dehydrogenase subunit 5,
RAG1=recombination activating protein 1, rbcL=RuBisCo large subunit,
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

12
their ranges, and an uncorrelated
lognormal relaxed clock was used as the
clock model. Monophyly constrains were
applied to match the topology to that of
the synthesized tree, and the compiled
node ages were then applied as secondary
calibrations, setting them as mean values
under a normal distribution with a
standard deviation of 0.0001, to constrain
the node ages of the resulting tree to the
original TimeTree values as much as
possible.
Because models that account for
stratigraphic ranges (rather than
uncertainty ranges) have not yet been
implemented for phylogenetic analysis,
we could not directly establish the older
bounds of familial age ranges.
Consequently, to incorporate these older
bounds in an indirect way, we used
uniform calibration priors on nodes
whose direct descendants include at least
one taxon presenting age range
information. The minimum bound of the
distribution was set as the older bound of
the taxon’s age range, and the maximum
bound was set as the age of the nearest
normally distributed, secondarily
calibrated parent node (Figure 4). If more
than one of the descendants of a node
had age range data, the minimum bound
of the uniform distribution was set as the
oldest value of such age ranges.
Because of the constant need to re-run
these analyses each time the synthesized
tree is updated, we elected a GTR+I+G
substitution model for all analyses (to
avoid time expenditure by testing for
multiple models), as it is one of the
models that most frequently fits real data
(Arenas et al. 2015). Each analysis was run
for 1.000.000 generations, sampling every
1000 trees. The maximum credibility tree
was generated using mean values with
the program TreeAnnotator in the BEAST
package, with a 10% burn-in cutoff. The
resulting time-calibrated trees were then
used to manually assign branch lengths to
the synthesized Newick tree, rounding
values up or down to the closest integer.
Despite all the collected data to obtain
age estimates for every node in the
synthesized tree, there are several family-
level taxa with neither molecular data nor
stratigraphic age ranges from the fossil
record. These taxa were excluded from
the previously described procedures, and
the age of their parent nodes were
manually calculated by equally dividing
the distance between the nearest parent
and daughter dated nodes, which is an
identical approach to the one employed in
the BladJ algorithm (Webb et al. 2002).
Branch lengths were then assigned in
accordance with these interpolated node
ages.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

13
Figure 4. General approach for dating the phylogeny in BEAST2. In this hypothetical phylogeny, all six extant
terminals exhibit sequence data and age ranges from the fossil record, spanning from the present (0) to their
oldest fossil occurrence in the past. Additionally, two of all five nodes have reported age estimates from the
literature (V and W). These estimates are employed as secondary calibration points, setting them as the mean
value of normal distribution priors (with very narrow standard deviations). Additional uniform distribution
priors are set for the nodes whose direct descendants are terminal taxa with reported age ranges (Y and Z).
The lower bound of the uniform distribution is set as the oldest bound of the ranges of its daughter terminal
taxa: (29 for node Y and 17 for node Z, from C and F terminals, respectively). The upper bound of the uniform
distribution is set as the nearest secondarily calibrated parent node, which in this case is W (age=65) for both
cases. Note that in this hypothetical scenario, age ranges for taxa A and B are not utilized because they are
direct descendants of secondarily calibrated nodes (and whose maximum age values do not exceed such
secondary calibration ages anyways). Moreover, age ranges from taxon D and E are not employed either
because their respective sister taxa have older age values in their ranges, which are ultimately employed for
the uniform distribution priors.
Taxon metadata
For each of the families and supra-familial
taxa in the resulting synthetic tree, we
gathered information on their common
names and taxonomic synonyms. Most
common names were taken from the
NCBI taxonomy (Federhen 2012),
Catalogue of Life (2021), Encyclopedia of
Life (Parr et al. 2014), and Christenhusz et
al. (2017; specifically for vascular plants).
Synonyms were taken from the NCBI
taxonomy, PBDB, World Register of
Marine Species, Wilson & Reeder (2005;
for mammals), Stevens (2016; for vascular
plants), and Kluge (2010; for supra-
familial insect taxa). Supporting sources
for phylogeny and taxonomy were also
annotated for each taxon, along with the
non-monophyletic status of some
families. Taxon metadata was stored into
two separate text files, one for common
names, and another for the rest of the
attributes that are eventually displayed in
primary (common names) and secondary
(sources, synonyms, and others)
information windows (see Mobile App
Overview section in Results).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

14
Selection of pre-collapsed clades
In order to exhibit a backbone of the tree
of life to easily illustrate the relationships
between major groups of organisms, we
decided to collapse a set of higher-level
taxa so the initial view is rendered with 88
terminals. These higher-level taxa were
not necessarily at the same taxonomic
rank, but where rather chosen arbitrarily
to match what we consider some large
and popularly known groups or
organisms, and whose names were listed
in a plain text file as one of the input files
for the tree rendering code (see this taxon
list in Supplementary File 1).
Taxonomic coverage
The methodological procedures described
up to this point were applied over 68 of
the 88 terminals shown in the initial
backbone tree, as specified in the
previous section (Selection of pre-
collapsed clades). The 20 clades that have
not been covered yet are listed in a plain
text file (see list in Supplementary File 2)
that is used as an input for the code of the
application, whose purpose is to denote
its current unavailability to users. These
taxa also appear as pre-collapsed clades,
but unlike others, are unable to be
uncollapsed by the Open button (see
Mobile App Overview section in Results).
Illustrations
Family-level images were produced by
eight different artists (listed in the Credits
section of the mobile app) and were
loosely based on a set of real-life
photographs and/or artwork material
provided by the authors of this study. Due
to the high number of images needed and
the limited storage of mobile devices, we
opted for rather simple but faithful artistic
representations of the organisms, instead
of overly detailed illustrations. Each
illustration was based on a representative
member of the family, which was often a
species of its type genus. Because of the
lack of specific paleontological knowledge
of the artists, illustrations of fossil taxa
were always based on pre-existing artistic
reconstructions. Consequently, we
excluded taxa based on very fragmentary
remains that have not been artistically
reconstructed to the date of this
publication.
When possible, we attempted to depict
the same sex and life stage within groups
of closely related taxa. For instances of
marked sexual dimorphism, we chose the
sex that mostly retains the general body-
plan of closely related taxa. Using the case
of coccoid insects (Hemiptera: Coccoidea)
as an example, in which females are
wingless and highly paedomorphic
individuals, we chose the winged males
for the easier visual comparison of
homologous traits with other hemipteran
families and insects in general. We also
chose adult stages for plants and animals,
the life stage that produces fruiting bodies
for multicellular fungi, and trophozoites
for apicomplexans. Finally, although it is
common practice for illustrations of
modular organisms such as plants to
depict focused parts of the individual, we
instead attempted to depict the entire
organism for those presenting large,
woody growth habits such as trees,
shrubs, and hedges.
Images for supra-familial clades were
generated by transforming an image of
one of its daughter families into a gray
silhouette. Each representative family
was chosen based on what the authors
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

15
consider to be the most ancestral
morphology for the whole group, as very
specialized morphologies may not be a
good representation of an entire clade.
App Development
A mobile app for interactively exploring
and visualizing the Newick tree was
written in Dart with Flutter 2.2
(https://flutter.dev/), an open-source
software development kit employed to
program cross platform applications
(incluging Android and iOS). The code for
tree rendering was loosely based on that
of Figtree v1.3.1 (Rambaut 2012). The
resulting code employs a set of text files
as input, which include the Newick tree
file, the Common names file, the Taxon
Metadata File, the Pre-Collapsed Groups
file, and the Unavailable groups file, as
well as an image folder containing the
whole set of taxon icons in Portable
Network Graphics (PNG) format.
RESULTS
As of May of 2021, the assembled tree
comprises 7335 terminal taxa (Figure 5;
for a detailed view, see Supplementary
Figure 1), with 7315 of them being
families, while the remaining 20 being
higher level groups whose family-level
phylogenies have not been assembled yet
(Annelida, Echinodermata, Brachiopoda,
Branchiopoda, Bryozoa, Cnidaria,
Chelicerata, Ciliophora, Dinozoa,
Endomyxa, Filosa, Heterokontophyta,
Multicrustacia, Myriapoda, Oligostraca,
Mollusca, Nematoda, Plathelminthes,
Retaria and Rhodophyta). Moreover,
1999 suprafamilial clades were included
and annotated as node labels in the
Newick tree.
At the time of publication, a total of 1479
bibliographic sources have been
employed to assemble the final time-
calibrated tree (see Reference section in
the mobile app or Supplementary File 3),
excluding those that were revised but
eventually discarded because of the
availability of higher ranked phylogenies.
The time calibrated tree comprises 559
secondarily calibrated nodes obtained
from the TimeTree database
(Supplementary File 4), while age ranges
from the fossil record were acquired for
2859 taxa (Supplementary File 5) and
molecular data for at least one locus were
obtained for 5060 taxa among all tree
partitions (Supplementary File 6). Neither
molecular data nor stratigraphic age
ranges from the fossil record were
available for 333 families. Of the 4888
nodes in the resulting synthetic tree, the
age of 4084 of them were estimated
through the BEAST2 analyses, while the
remaining ones were fixed as secondarily
calibrated nodes (from TimeTree) and
nodes whose age values were assigned by
interpolation.
Regarding taxon metadata, common
names were obtained for 2389 taxa, and
797 taxa presented at least one
taxonomic synonym. Moreover, 126
families were found to be non-
monophyletic as currently circumscribed
according to the revised source studies.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

16
Figure 5. Current version of the family-level synthesized tree with six major groups of organisms coded by
color, with non-monophyletic groups enclosed in quotation marks. Branches that do not reach the outer rim
of the circular phylogeny represent fossil families, which are especially evident in animals and absent in
bacteria, archaeans and protists.
In addition to the code for tree rendering
and its interactivity features, the resulting
APK file of the developed mobile
application includes the synthesized
Newick tree file, the common names text
file, the additional taxon metadata text
file (phylogenetic and taxonomic sources,
synonyms, and remarks), the pre-
collapsed clade list text file, the
unavailable clade list text file, and the
9334 illustrated icons for both families
and supra familial clades, totaling up to
128 megabytes. The app is freely available
for download in Android devices in
https://play.google.com/store/apps/deta
ils?id=com.pappcorn.tol and in iOS
devices in
https://apps.apple.com/co/app/tree-of-
life-tol-app/id821256110
Mobile App Overview
The main menu of Tree of Life App
offers five main options: Explore, Tutorial,
References, Frequently Asked Questions
(F.A.Q.), and Credits (which lists all the
staff who participated in the project).
Most of the following contents is focused
in the Explore option, which revolves
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

17
around the exploration of the large
phylogeny assembled here.
The Explore section (Figure 6) leads
users to a circular phylogeny representing
the family-level tree of life. Exploration is
performed by the standard zooming and
panning using the standard gestures on
touch screens of mobile devices. Instead
of exposing all the family level content,
the initial view of the tree consists in
several large clades of organisms
displayed as collapsed branches, so users
can easily visualize the higher-level
relationships across the tree of life before
straying in an overwhelmingly large tree
diagram (Figure 6a). These pre-collapsed
branches do not represent any specific
taxonomic rank but were rather arbitrarily
selected to represent large and well-
known groups. Users may either select
these collapsed branches individually to
unfold the desired family-level
relationships or employ the “open all”
button at the upper right of the screen to
unfold the entire family-level tree (Figure
6b and c).
In contrast to other efforts to
compile the tree of life in which only
topological information is shown, we
provide a time-calibrated phylogeny that
might be used to visualize divergence
dates among lineages, as well as
extinction dates of fossil taxa. This is
achieved with the aid of the radial ring
pattern on the background and the
horizontal labeled timescale at the right
hemisphere of the circumference (Figure
6a).
Furthermore, unlike previous tools,
Tree of Life App does not depend on
internet connection for its main
functionalities as all data is stored within
the device once the app is downloaded.
Consequently, users might explore and
consult the tree of life in remote locations
(i.e., during field work) or situations in
which internet connection is scarce or
lacking.
Branch tips
One of the most distinctive features
of the phylogeny displayed in Tree of Life
App is that rather than plain text,
illustrated icons of organisms are
displayed at the branch tips of the tree.
Displaying images instead of text can be
advantageous for several reasons, not
only because they make the tree more
visually appealing, but also because an
image of the organism is arguably more
informative than the taxon name when
the latter is mostly unknown for an
average user. Consequently, we argue
that an image-rich tree of life is the best
way of visually exhibiting the evolutionary
history of life, which is in line with the
recent ongoing trend of displaying tip-
illustrated phylogenies in the figures of
published studies (e.g., Hasegawa 2017).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

18
Figure 6. Basic features of Tree of Life app. a) Initial view of the tree with major groups of organisms pre-
collapsed into single branches. b) A full view of the family-level tree of life after activating the “open all”
button. c) A zoomed in portion of the tree focused on part of the paleognath and galloanseran birds. c) Primary
information window displayed after tapping on one of the pictures at the tips. b) Secondary information
window after tapping on the “i” button.
Two types of icons can be seen at
branch tips: color icons and gray
silhouettes. Color icons correspond to
taxonomic families, which are OTUs at
which this tree was constructed. The
organism displayed in an icon correspond
to a representative species of each
taxonomic family. As one of the several
interactive features of this app, these
icons can be tapped on to access what we
call a primary information window (Figure
6d), which displays the enlarged icon, the
taxon name, and its associated common
names, if any. Furthermore, a link to
Wikipedia is also present at the lower
right corner of the window for users that
intend to get further information of the
selected taxon, as well as a button that
leads to a secondary information window
(Figure 6e) containing the sources
supporting the phylogenetic position of
such taxon as well as other bits of
information (see further details in the
References section of this article). The
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

19
second type of icons are gray silhouettes,
which in contrast to the colored ones,
have their corresponding taxon name
placed over the image. Rather than
taxonomic families, these icons
correspond to supra-familial taxa that are
either the pre-collapsed groups shown in
the initial view of the tree, or manually
collapsed clades obtained by using the
branch options described in the next
section. Users may also tap on these
silhouette icons, but the options shown in
their corresponding pop-up windows
differ from those of colored icons as
described below.
Internal Node Options
In addition to the family-level
branch tips, internal nodes are also
labelled on the phylogeny denoting supra-
familial taxa or clade names. Such labelled
nodes are displayed as black dots adjacent
to their text label (Figure 7a). Users may
tap on the black dots to display a primary
information window which, as in the case
of their branch tip family-level
homologues, display a gray silhouette
icon denoting a representative organism
of the clade, its associated taxon name, a
link to Wikipedia and the References
section, and two additional options
exclusive to supra-familial taxa: Subtree
and Collapse (Figure 7b).
Subtree — Users may employ this
option to visualize specific clades of the
tree by excluding everything outside the
selected clade. In such cases, smaller
circular trees may be displayed without
the need of visualizing them by
thoroughly zooming in a specific location
of the large tree of life (Figure 7b,c).
Collapse —This option enables the
user to collapse the entire clade denoted
by the selected node into a single branch.
The collapsed clade or branch will then be
displayed as a red line with a gray
silhouette at its end, along with a text
label displaying its taxon name over the
icon (Figure 7d,e). This feature allows the
user to customize the tree at will by hiding
undesired details of the tree for any
specific purpose, (e.g., visualizing the tree
at higher, supra-familial levels by
selectively collapsing different clades). To
re-open a collapsed branch, the silhouette
icon at the tip must be tapped, and then
an almost identical pop-up window will
appear, with an “open” button replacing
the previous “collapse” button must be
tapped on to unfold the contents of the
collapsed clade.
Highlighting Branches - Highlighting
single branches or even entire clades is
possible by tapping pictures at the tip or
black dots at the nodes and then closing
the primary information window. The
selected branch will then appear
highlighted in blue. This may be useful to
keep a lineage marked without losing
sight of it in a highly dense tree, to easily
identify the sister group of a lineage by
tracking the selected branch to its parent
node, or to view the family-level diversity
of a clade relative to the rest of the
displayed tree.
Secondary information windows
Users may access secondary
information windows by tapping on the
“i” buttons in each of the primary pop-up
windows that are accessed by tapping on
icons at the branch tips or labelled nodes.
These windows contain three information
fields: a first one for the citations of
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

20
sources supporting the taxonomy,
phylogeny, or phylogenetic position of the
taxon, a second one for taxonomic
synonyms, and a third one for additional
remarks such as the phyletic status of a
family (Figure 6b).
Search Function
As the family-level tree of life
displayed is considerably large, users may
have a hard time finding specific
organisms in such a big tree. To solve this
problem, a search function is available so
users may query specific taxa by typing
either taxon (both family names and
higher, supra-familial clade names) or
common names. A green arrow tip will
then appear indicating the location of the
matched results. In the current version of
the app, the search function is unable to
find queried taxa if they are within a
collapsed group. Given that the initial
view of the tree of life consists in large
pre-collapsed groups, this limitation
implies that users must have an a priori
knowledge regarding the higher-level
taxonomic affiliation of the queried taxon,
and then open its corresponding branch
before using the search function.
However, users may use the “open all”
button (Figure 6a) to unfold all the pre-
collapsed branches shown in the initial
view to visualize the whole family-level
tree. This way, the search function will
successfully find any taxon included in the
current version of the phylogeny.
Figure 7. Some examples of branch options. a) A zoomed in portion of the tree focused on the deep nodes
around anurans (frogs). b) Primary information window displayed after tapping on the Anura node, displaying
both branch options at the bottom of the window. c) Full view of the anuran tree after selecting the Subtree
option. d) Another view of a primary information window, this time selecting the Hyloidea node within the
anuran tree. e) Appearance of the anuran tree after selecting the “Collapse” option on the Hyloidea node.
Note that Hyloidea is now represented by a single red branch with a gray silhouette and text label.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

21
References Section
The tree of life shown in the app was
synthetized from over 1400 sources that
provided phylogenetic and/or taxonomic
information. Each of these sources are
properly cited in the secondary
information windows and listed in the
References section of the main menu,
which are arranged alphabetically and
divided into six separate pages with their
respective alphabetical ranges.
DISCUSSION
By a thorough revision of the scientific
literature, a set of rigorous criteria for
source selection, and a collection of
divergence dates, fossil stratigraphic
records and sequence data, we
assembled an up-to-date family-level tree
of life. Additionally, we developed an app
for mobile devices to interactively explore
and visualize the large, synthesized tree.
To our knowledge, this is the only project
that have approached data compilation
and visualization issues at the same time,
as other similar initiatives have relied on
others for one or the other (LifeMap and
OneZoom with the NCBI taxonomy and
the OTOL synthetic tree). In the following
subsections, we discuss the strengths,
novelties, limitations and pitfalls of the
performance, interactivity features, and
phylogenetic information exhibited in this
project.
Strengths and novelties
Fossil taxa
A relevant novelty introduced in our
synthesized tree is the inclusion of fossil
taxa, considering that large-scale
phylogenetic data sources employed by
other tree of life visualizers (the OTOL by
LifeMap and OneZoom) are currently
devoid of extinct organisms. The inclusion
of fossil taxa into phylogenies have long
been considered crucial for the
understanding of overall evolutionary
processes (Prothero 2007; Lyson et al.
2010; Slater et al. 2012). By breaking long
branches, the incorporation of fossil taxa
provides a clearer visualization of the
gradual evolution of traits because of the
transitional morphology of some extinct
organisms. This is further exploited in our
mobile app by the representation of
branch terminals with images. In this way,
users can easily observe gradual
transitions between disparate
morphologies, as well as understanding
the close relationships between some
extant taxa that would otherwise be
surprising (e.g., birds to other non-avian
“reptiles”, cetaceans to hippopotami,
terrestrial vertebrate to fish).
Branch lengths
Another novelty strongly tied to the
inclusion of fossil taxa is the
implementation of a temporal axis and
time-proportional branch lengths.
Visually inferring branch lengths and
divergence dates provide valuable
information that complements
topological visual cues; while the latter
indicates plain relatedness, the former
allows user to visualize the degree of such
relatedness. While this is also lacking in
the OTOL synthetic tree, it is nevertheless
unable to be visualized by the formats
displayed by OneZoom and LifeMap. It is
worth mentioning that OneZoom
provides ages for many of its nodes with
numerical labels, but the age of such
clades cannot be visually inferred by the
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

22
proportionality of branch lengths. The
TimeTree project also assembled a large
time calibrated tree, but such a project is
mostly employed as a divergence date
database rather than as a topological tree
assemblage or for tree exploration
purposes, due to their restricted coverage
and the narrow range of eligible studies
(only time-calibrated phylogenies, which
are a small fraction of the overall
published phylogenetic knowledge).
Images
Partially in accordance with what was
previously mentioned about visualizing
fossil taxa in phylogenies, we consider
that one of the most solid strengths of our
version of the tree of life is the
representation of terminal taxa in the
form of illustrated icons. We argue that
this feature not only enhances the
aesthetic appeal of the tree, but also
provides valuable visual information to
users who are not familiarized with
certain groups of organisms. This could
even prove attractive enough for the non-
academic public who may not even be
able to properly read phylogenies but
might very well be engaged to visually
explore the morphological diversity of life
arranged by evolutionary relatedness.
The incorporation of images in a
phylogeny is not particularly a novelty, as
it is not uncommon to find publications
with organisms’ illustrations or
photographs in their phylogenetic figures.
In fact, the OneZoom visualizer
incorporates some images within its
leaves and internal nodes, consisting
mostly in circularly cropped photographs.
Although photographs have a huge
advantage over illustrations because they
faithfully represent the real-life image of
an organism, we find this approach as a
disadvantage in other aspects. First,
photographs are usually presented as a
geometrical area (squares, rectangles, or
circles in the case of OneZoom) whose
contents also include a background
behind the depicted organism. This is
problematic because the true shape and
outline of the organism itself is gradually
lost and becomes progressively
indistinguishable from the rest of the
photograph while zooming out, and
zooming out is ironically crucial for a
broad comparison of multiple organisms
in a phylogeny. Additionally, photographs
may include only specific parts of the
organisms and some of those may not
even be homologous with the parts
displayed in other photographs, so a
proper comparison is not always
achieved. By presenting images without a
background —as is the case with the
illustrated icons presented here— the
shape of the organism is easily maintained
over a broad range of zoom levels,
enabling its identification and comparison
with its neighboring taxa. As our icons also
depict the whole organism and mostly
homologous life stages, a proper
comparison is achieved while visualizing
many related organisms at once within
the phylogenetic framework.
Literature coverage
Incorporating as much of the relevant
literature as possible is pivotal for
assembling a tree of life that best
represents most of the generated
knowledge on the subject. However,
prevailing large-scale projects rely on
specific data formats for the inclusion of
studies into their synthetic tree, limiting
the number of eligible sources and
consequently, representing only a
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

23
fraction of all the available knowledge on
phylogeny. The OTOL is notorious for this
restriction because its infrastructure was
designed to exclusively incorporate
phylogenetic data in the form of tree files,
but the overall presence of this data
format in publicly available repositories is
low, and requests to authors for such files
had a low response rate (Hinchcliff et al.
2015). Because of this, the placement of
98% of the tips in their synthetic tree is
based on taxonomy rather than
phylogenetic studies at the time of its
formal publication, although this number
may have been reduced since then. On
the other hand, the TimeTree project is
focused on providing a summary on
divergence time estimates and as such, it
incorporates a small fraction of the
phylogenetic knowledge, which consist in
time-calibrated trees. Moreover, it is also
dependent on tree files obtained through
public repositories or by directly
contacting authors, presumably suffering
from the same difficulties mentioned for
the OTOL project.
Contrastingly, we bypassed these
obstacles by manually transcribing
phylogenies available as figures to Newick
notation, allowing us to incorporate
virtually any published study into the tree
assembling process. Hence, we argue that
the tree of life displayed in the app is a
comprehensive representation of the
available knowledge on phylogeny in the
scientific literature.
Tree layout
Rectangular or diagonal tree formats are
the most commonly used layouts for
publishing phylogenies, and circular
formats are often used for large
phylogenies. Despite this, current tree of
life explorers —namely OneZoom and
LifeMap— employ unfamiliar tree layouts
to display their extremely large
phylogenies. These layouts may facilitate
exploration by deep zooming, but the
reading of topological relationships is
troublesome and the overall shape of the
tree at shallow zoom levels is hardly
perceived (i.e., the relative size of clades
and tree asymmetry).
We attempted to circumvent these
pitfalls by representing our synthesized
tree in a traditional and conventional
format: a circular phylogeny. This layout is
useful for optimizing space for large trees
and provide the additional advantage of
preventing the usual interpretation of
phylogenies as evolutionary ladders,
rather than a branching process. As with
other conventional formats, this layout
also allows the incorporation of a linear
timescale, which in this case displays
time-proportional branch lengths.
Internet connection
Lastly, another relevant novelty of our
tree of life explorer is its independence of
internet connection, which contrasts with
other prevailing explorers in their reliance
to web servers. The tree file, taxon
metadata, reference list and organisms’
icons are all stored within the mobile
device once the app is downloaded. The
only features that require internet
connection are the tutorial (which
redirect users to a YouTube video) and the
links to Wikipedia for each taxon, which is
shown in primary information windows.
Nevertheless, the local storage of all these
files (especially images) has a relevant
drawback. The independence of a web
server combined with the limiting
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

24
processing power of mobile devices (in
contrast to computers) leads to a relative
loss of performance for features such as
the smoothness of zooming and panning
gestures. This performance difference
issues are especially evident in older
devices when compared with the superior
exploration capacities of the OneZoom
and LifeMap explorers.
Limitations
Current taxonomic coverage
Although the tree assembling procedures
described here allows a broad sampling of
the available phylogenetic knowledge,
these thorough processes have only been
carried out over selected group of
organisms due to the time-consuming
methodologies. As such, with over 7200
terminal taxa, around half of the total
family-level diversity is represented at the
moment (Ruggiero 2014; PBDB). Some of
the large and well-known groups among
those currently unavailable sections of
the tree are chelicerates (arachnids and
allies), multicrustaceans (the largest clade
of the now paraphyletic Crustacea),
mollusks, annelids, cnidarians, and red
algae, along with other smaller or less
popular groups.
Additionally, within the large, covered
groups, some taxa are absent because
either they have not been formally
assigned to a family-level taxon or no
photograph, image, or artistic
reconstruction exists for such taxa. The
former case is especially evident in some
fossil taxa whose researchers have largely
abandoned Linnean ranks, such as current
systematics of sauropod dinosaurs. For
representing those groups in our tree, we
included some older, family-level
described taxa, but newly described taxa
from the last two decades are
unfortunately unrepresented. This is also
the case for some family-unaffiliated taxa
denominated as incertae sedis (most
usually found in microscopic fungi and
algae, as well as some unicellular
eukaryotes), although these organisms
represent a minuscule fraction of the
described biodiversity. On the other hand,
the lack of imagery is notorious for fossil
taxa based on very fragmentary remains,
which makes the depiction of the whole
organism difficult. This is also the case for
candidate taxa that are only known from
environmental DNA samples, leading to
the exclusion of a vast amount of the
known prokaryote diversity in the tree of
life (Hug et al. 2016). Although these taxa
could be incorporated as branches with
only text labels in their tips, we consider
this approach would not fit within this
project’s scope of visualizing a fully
illustrated tree of life. For this reason, we
refrain for including such taxa until more
complete fossils are discovered, or some
of the candidate taxa become cultivated
and formally described (e.g., Imachi et al.
2020).
Taxonomic resolution
The fact that the goal of this project is to
produce an illustrated, family-level tree of
life implies that a broad variation of the
diversity of life is to some extent reduced
to single terminals representing clusters
of closely related species. Hence,
phylogenetic relationships below family
level are out of the scope of this project
and should be consulted in other sources
encompassing a higher taxonomic
resolution, such as the OTOL.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

25
Divergence date estimates
Unlike reconstructing cladograms or
phylograms from sequence data, dating
phylogenies is a more complex and
difficult process, as it requires
considerable a priori knowledge on the
detailed fossil record of the specific clade
being treated and the tentative
systematic position of fossil taxa within
the tree. Due to our lack of this specialized
expertise, we attempted to use as many
secondary calibration points as possible,
so several of the node ages in the
resulting tree represent estimates from
other studies. However, we did attempt
to estimate many other nodes in between
those secondary calibrations, and
although we obtained (and to some
extent curated) stratigraphic range data
from the PBDB for a wide array of taxa,
several inaccuracies from such database
might have been overlooked. Moreover,
the fact that the BEAST analyses we
carried out were run for relatively few
generations, along with the lack of chain
convergence for some runs and the use of
a single substitution model for all tree
partitions without prior model testing,
implies that some node ages estimated
here should be interpreted with caution.
Future projections
Besides the constant incorporation of
novel and better ranked phylogenies
published in the literature, our most
relevant short-term goal involves
covering the currently unavailable groups
for displaying a fully comprehensive tree
of life.
Several other additional features for the
mobile app are also under plans, such as
additional taxon associated metadata
such as number of described species; the
incorporation of a geologic time scale
with all its conventional subdivision,
parallel to the already implemented
numerical timescale; a function to find
and highlight the path and the divergence
date between two selected taxa; an
educational and didactic section for
enhancing tree-thinking skills, and a trivia
game section in the main menu where a
random set of taxa is drawn from the tree
and users must establish their proper
topological relationships. Finally, a web
app is also under consideration so users
can access this tool from their laptops and
desktop computers.
FUNDING
This project was funded by the authors’
own resources and the funding members of
PappCorn S.A.S., who aided in the programming
and graphical design of the app. Because of this,
we want to highlight the scarce opportunities
offered by Colombian institutions for funding
scientific projects, which is probably hindering
hundreds of similar initiatives.
ACKNOWLEDGEMENTS
This project could not have been
completed without all the talented professionals
who were essential for the development of this
project, most notably the artists Katie Walker,
Vanessa Reina, Oscar Builes, Camilo Delgado, Lina
Beltrán, Luisa Orozco and Carolina Salazar, the
developers Mario García, Janet Brumbaugh, Alex
Lamb, Yang Xiaoming and Nicolas Contreras, and
the graphic designer Ricardo Mejía. We also thank
the PappCorn S.A.S. team for its overall guidance
into the world of mobile app development. Finally,
we are very grateful to Adolfo Jara, Maria Sole
Calbi, Francisco Fajardo, Sebastian Gonzalez,
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

26
Julián León, Pablo Tovar, Natalia Contreras and
Ángela Rodriguez for their valuable comments on
the early version of the manuscript and the mobile
app.
REFERENCES
Appeltans, W., Decock, W., Vanhoorne, B.,
Hernandez, F., Bouchet, P., Boxshall, G.,
Fauchald, K., Gordon, D. P., Poore, G. C.
B., van Soest, R., Stöhr, S., Walter, C.,
Costello, M. J. (2011). The World Register
of Marine Species: an authoritative,
open-access web-resource for all marine
species.
Arcila, D., Ortí, G., Vari, R., Armbruster, J. W.,
Stiassny, M. L., Ko, K. D., Sabaj, M. H.,
Lundberg, J., Revell, L. J., & Betancur-R, R.
(2017). Genome-wide interrogation
advances resolution of recalcitrant
groups in the tree of life. Nature Ecology
& Evolution, 1(2), 1-10.
Arenas, M. (2015). Trends in substitution models
of molecular evolution. Frontiers in
genetics, 6, 319.
Betancur‐R, R., Arcila, D., Vari, R. P., Hughes, L. C.,
Oliveira, C., Sabaj, M. H., & Ortí, G. (2019).
Phylogenomic incongruence, hypothesis
testing, and taxonomic sampling: The
monophyly of characiform
fishes. Evolution, 73(2), 329-345.
Bouckaert, R., Heled, J., Kühnert, D., Vaughan, T.,
Wu, C. H., Xie, D., Suchard, M. A.,
Rambaut, A., & Drummond, A. J. (2014).
BEAST 2: a software platform for Bayesian
evolutionary analysis. PLoS Comput
Biol, 10(4), e1003537.
Catalogue of Life, et al. (2021). Species 2000 & ITIS
Catalogue of Life, 2021-04-05. Digital
resource at www.catalogueoflife.org.
Species 2000: Naturalis, Leiden, the
Netherlands. ISSN 2405-8858.
Christenhusz, M. J., Fay, M. F., & Chase, M. W.
(2017). Plants of the world: an illustrated
encyclopedia of vascular plants.
University of Chicago Press.
Darwin, C.R. (1859). On the Origin of Species by
Means of Natural Selection, or the
Preservation of Favoured Races in the
Struggle for Life. John Murray, London.
de Vienne, D. M. (2016). Lifemap: exploring the
entire tree of life. PLoS biology, 14(12),
e2001624.
Drew, B. T., Gazis, R., Cabezas, P., Swithers, K. S.,
Deng, J., Rodriguez, R., Katz, L. A.,
Crandall, K. A., Hibbett, D. S., & Soltis, D.
E. (2013). Lost branches on the tree of
life. PLoS Biology, 11(9), e1001636.
Drummond, A.J., Ashton, B., Buxton, S., Cheung,
M., Cooper, A., Duran, C., Field, M.,
Heled, J., Kearse, M. and Markowitz, S.
(2011). Geneious v5. 4, Available
from http://www.geneious.com/
Edgar, R. C. (2004). MUSCLE: multiple sequence
alignment with high accuracy and high
throughput. Nucleic acids
research, 32(5), 1792-1797.
Federhen, S. (2012). The NCBI taxonomy
database. Nucleic acids research, 40(D1),
D136-D143.
Gadagkar, S. R., & Kumar, S. (2005). Maximum
likelihood outperforms maximum
parsimony even when evolutionary rates
are heterotachous. Molecular biology
and evolution, 22(11), 2139-2141.
Hasegawa, M. (2017). Phylogeny mandalas for
illustrating the Tree of Life. Molecular
phylogenetics and evolution, 117, 168-
178.
Hedges, S., & Kumar, S. (2009). The Timetree of
Life. Oxford: Oxford University Press.
Hedges, S. B., Marin, J., Suleski, M., Paymer, M., &
Kumar, S. (2015). Tree of life reveals
clock-like speciation and
diversification. Molecular biology and
evolution, 32(4), 835-845.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

27
Imachi, H., Nobu, M. K., Nakahara, N., Morono, Y.,
Ogawara, M., Takaki, Y., Takano, Y.,
Uematsu, K., Ikuta, T., Ito, M., Matsui, Y.,
Miyasaki, M., Murata, K., Saito, Y., Sakai,
S., Song, C., Tasumi, E., Yamanaka, Y.,
Yamaguchi, T., Kamagata, Y., Tamaki, H.,
& Takai, K. (2020). Isolation of an
archaeon at the prokaryote–eukaryote
interface. Nature, 577(7791), 519-525.
Jiang, X., Edwards, S. V., & Liu, L. (2020). The
multispecies coalescent model
outperforms concatenation across
diverse phylogenomic data
sets. Systematic biology, 69(4), 795-812.
Kluge, N. J. (2010). Circumscriptional names of
higher taxa in
Hexapoda. Bionomina, 1(1), 15-55.
Kubatko, L. S., & Degnan, J. H. (2007).
Inconsistency of phylogenetic estimates
from concatenated data under
coalescence. Systematic biology, 56(1),
17-24.
Kumar, S., Stecher, G., Suleski, M., & Hedges, S. B.
(2017). TimeTree: a resource for
timelines, timetrees, and divergence
times. Molecular biology and
evolution, 34(7), 1812-1819.
Lyson, T. R., Bever, G. S., Bhullar, B. A. S., Joyce, W.
G., & Gauthier, J. A. (2010). Transitional
fossils and the origin of turtles. Biology
Letters, 6(6), 830-833.
Heath, T. A., Huelsenbeck, J. P., & Stadler, T.
(2014). The fossilized birth–death process
for coherent calibration of divergence-
time estimates. Proceedings of the
National Academy of Sciences, 111(29),
E2957-E2966.
Hinchliff, C. E., Smith, S. A., Allman, J. F., Burleigh,
J. G., Chaudhary, R., Coghill, L. M., ... &
Gude, K. (2015). Synthesis of phylogeny
and taxonomy into a comprehensive tree
of life. Proceedings of the National
Academy of Sciences 112, 12764-12769.
Hug, L. A., Baker, B. J., Anantharaman, K., Brown,
C. T., Probst, A. J., Castelle, C. J.,
Butterfield, C. N., Hernsdorf, A. W.,
Amano, Y., Ise, K., Suzuki, Y., Dudek, N.,
Relman, D. A., Finstad, K. M., Amundson,
R., Thomas, B. C., & Banfield, J. F. (2016).
A new view of the tree of life. Nature
microbiology, 1(5), 1-6.
Lecointre, G., Le Guyader, H. (2006). The tree of
life: a phylogenetic classification. Harvard
University Press.
Lee, M. S., & Palci, A. (2015). Morphological
phylogenetics in the genomic
age. Current Biology, 25(19), R922-R929.
Maddison, D.R., K.-S. Schulz, W.P. Maddison.
2007. The Tree of Life Web Project.
Zootaxa 1668:19-40.
Nabhan, A. R., & Sarkar, I. N. (2012). The impact of
taxon sampling on phylogenetic
inference: a review of two decades of
controversy. Briefings in
bioinformatics, 13(1), 122-134.
Page, R. D. (2012). Space, time, form: viewing the
Tree of Life. Trends in ecology &
evolution, 27(2), 113-120.
Parr, C. S., Wilson, M. N., Leary, M. P., Schulz, K. S.,
Lans, M. K., Walley, M. L., Hammock, J. A.,
Goddard, A., Rice, J., Studer, M., Holmes,
J. T. G., & Corrigan, Jr, M. R. J. (2014). The
encyclopedia of life v2: providing global
access to knowledge about life on
earth. Biodiversity data journal, (2):
e1079.
Pennisi, E. (2013). Large-Scale Gene Comparisons
Boost Tree of Life Studies. Science
342(6164):26-27.
Philippe, H., de Vienne, D. M., Ranwez, V., Roure,
B., Baurain, D., & Delsuc, F. (2017). Pitfalls
in supermatrix phylogenomics. European
Journal of Taxonomy, (283).
Philippe, H., Zhou, Y., Brinkmann, H., Rodrigue, N.,
& Delsuc, F. (2005). Heterotachy and
long-branch attraction in
phylogenetics. BMC evolutionary
biology, 5(1), 1-8.
Prothero, D. R. (2007). Evolution: what the fossils
say and why it matters. Columbia
University Press.
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint

28
Puttick, M. N., O'Reilly, J. E., Pisani, D., &
Donoghue, P. C. (2019). Probabilistic
methods outperform parsimony in the
phylogenetic analysis of data simulated
without a probabilistic
model. Palaeontology, 62(1), 1-17.
Rambaut, A. (2012). FigTree v 1.3.1. Available
online
at: http://tree.bio.ed.ac.uk/software/figt
ree/
Reddy, S., Kimball, R. T., Pandey, A., Hosner, P. A.,
Braun, M. J., Hackett, S. J., Han, K.-L.,
Harshman, J., Huddleston, C. J., Kingston,
S., Marks, B. D., Miglia, K. J., Moore, W. S.,
Sheldon, F. H., Witt, C. C., Yuri, T., &
Braun, E. L. (2017). Why do phylogenomic
data sets yield conflicting trees? Data
type influences the avian tree of life more
than taxon sampling. Systematic
Biology, 66(5), 857-879.
Roch, S., & Steel, M. (2015). Likelihood-based tree
reconstruction on a concatenation of
aligned sequence data sets can be
statistically inconsistent. Theoretical
population biology, 100, 56-62.
Rosindell, J., & Harmon, L. J. (2012). OneZoom: a
fractal explorer for the tree of life. PLoS
Biol, 10(10), e1001406.
Rothwell, G. W., Escapa, I. H., & Tomescu, A. M.
(2018). Tree of death: The role of fossils
in resolving the overall pattern of plant
phylogeny. American journal of
botany, 105(8), 1239-1242.
Ruggiero, M. A. (2014). Families of All Living
Organisms, Version 2.0.a.15, (4/26/14).
Expert Solutions International, LLC,
Reston, VA. 420 pp.
Scotland, R. W., Olmstead, R. G., & Bennett, J. R.
(2003). Phylogeny reconstruction: the
role of morphology. Systematic
Biology, 52(4), 539-548.
Siddall, M. E. (2002). Measures of support. In
DeSalle, R., Giribet, G., & Wheeler, W.
(Eds.). Techniques in molecular
systematics and evolution. Springer
Science & Business Media.
Slater, G. J., Harmon, L. J., & Alfaro, M. E. (2012).
Integrating fossils with molecular
phylogenies improves inference of trait
evolution. Evolution: International
Journal of Organic Evolution, 66(12),
3931-3944.
Stevens, P. F. (2016). Angiosperm Phylogeny
Website. Version 13. Angiosperm
Phylogeny Website. Version 13.
Struck, T. H., Wey-Fabrizius, A. R., Golombek, A.,
Hering, L., Weigert, A., Bleidorn, C.,
Klebow, S., Iakovenko, N., Hausdorf, B.,
Petersen, M., Kück, P., Herlyn, H.,
Hankeln, T. (2014). Platyzoan paraphyly
based on phylogenomic data supports a
noncoelomate ancestry of
Spiralia. Molecular biology and
evolution, 31(7), 1833-1849.
Townsend, J. P., López-Giráldez, F., & Friedman, R.
(2008). The phylogenetic informativeness
of nucleotide and amino acid sequences
for reconstructing the vertebrate
tree. Journal of Molecular
Evolution, 67(5), 437-447.
Vogt, L., Bartolomaeus, T., & Giribet, G. (2010).
The linguistic problem of morphology:
structure versus homology and the
standardization of morphological
data. Cladistics, 26(3), 301-325.
Webb, C. O., Ackerly, D. D., McPeek, M. A., &
Donoghue, M. J. (2002). Phylogenies and
community ecology. Annual review of
ecology and systematics, 33(1), 475-505.
Wilson, D. E., & Reeder, D. M. (Eds.).
(2005). Mammal species of the world: a
taxonomic and geographic
reference (Vol. 1). JHU Press.
WoRMS Editorial Board (2021). World Register of
Marine Species. Available from
http://www.marinespecies.org at VLIZ.
Zhang, D., Rheindt, F. E., She, H., Cheng, Y., Song,
G., Jia, C., Qu, Y., Alström, P., & Lei, F.
(2021). Most genomic loci misrepresent
the phylogeny of an avian radiation
because of ancient gene flow. Systematic
Biology (in press).
.CC-BY-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 6, 2021. ; https://doi.org/10.1101/2021.08.04.454988doi: bioRxiv preprint