RESEARCH ARTICLES

Cite as: D. Levin et al., Science

10.1126/science.abb5352 (2021).

Diversity and functional landscapes in the microbiota of

animals in the wild
Doron Levin1†, Neta Raab1†, Yishay Pinto1†, Daphna Rothschild2,3,4,5†, Gal Zanir1, Anastasia Godneva2,3,
Nadav Mellul6, David Futorian1, Doran Gal1, Sigal Leviatan2,3, David Zeevi2,7, Ido Bachelet1,6‡, Eran Segal2,3‡*
1Wild Biotech, Rehovot, Israel. 2Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, 7610001 Israel. 3Department of

Molecular Cell Biology, Weizmann Institute of Science, Rehovot 7610001, Israel. 4Department of Developmental Biology, Stanford University, Stanford, CA 94305, USA.
5Department of Genetics, Stanford University, Stanford, CA 94305, USA. 6Augmanity, Rehovot, Israel. 7Center for Studies in Physics and Biology, Rockefeller University,

New York, NY 10065, USA.

†These authors contributed equally to this work. ‡These authors contributed equally to this work. *Corresponding author. Email: eran.segal@weizmann.ac.il

Animals in the wild are able to subsist on pathogen-infected and poisonous food and show immunity to
various diseases. These may be due to their microbiota, yet we have a poor understanding of animal

Downloaded from http://science.sciencemag.org/ on March 26, 2021

microbial diversity and function. We used metagenomics to analyze the gut microbiota of over 180 species
in the wild, covering diverse classes, feeding behaviors, geographies, and traits. Using de novo metagenome
assembly, we constructed and functionally annotated a database of over 5,000 genomes, comprising 1,209
bacterial species of which 75% are unknown. The microbial composition, diversity, and functional content
exhibit associations with animal taxonomy, diet, activity, social structure and lifespan. We identify the gut
microbiota of wild animals as a largely untapped resource for the discovery of therapeutics and
biotechnology applications.

Compared to the human microbiota, which has been exten-
sively studied (1, 2), the microbiota of animals has received
less focus. Microbiome research was pioneered with several
well accepted methodologies; such as metagenomic profiling
applied to several animal species (3–13). Metagenomics anal-
ysis has been combined with 16S ribosomal RNA sequencing
to investigate how diet shapes the gut microbiota within var-
ious host species (14), in particular primates (15, 16). Diet
metabarcoding, in particular, was useful in the study of die-
tary niche partitioning and composition (17–23). 16S riboso-
mal RNA sequencing alone has provided compositional data
and taxonomy-related insights about the microbiota in many
animal species [(24–27) just to cite a few]. Other studies ad-
dressed behavioral aspects and their relation to animal mi-
crobiota [for example (28–33)]. Still, despite these studies, a
comprehensive catalog and unified analysis of the microbiota
of animals across different geographies, feeding behaviors,
and other traits, is still lacking.
With advances in shotgun metagenomic sequencing tech-
nologies and algorithms to build metagenomic assembled ge-
nomes (MAGs), several studies aimed to better capture the
diversity of gut and other microbes by reconstruction of mi-
crobial genomes (34–36). By utilizing metagenomic assembly
methods to build genomes of uncultured bacteria, several
studies successfully reconstructed hundreds of thousands of
MAGs from the human microbiome (37–39). Additional ef-
forts were made to create catalogs of species-level microbial
genomes, resulting in an estimation of over 4,500 distinct mi-
crobial species in the human gut (39, 40). These studies
suggest that despite previous studies of the human gut mi-
crobiome, a large portion of the human microbiome remains
unexplored. Given that less resources have been invested in
wildlife microbiome studies, it can be assumed that the po-
tential for discovery of previously undescribed microbiota is
particularly high.
It is becoming increasingly evident that animal microbi-
omes are a rich source of biological functions that may have
biotechnological impact. For example, recent microbiome
studies have reported the discovery of antimicrobials (41–45),
and immunomodulators (46), as well as industrial enzymes
(47–51). Moreover, animals in the wild exhibit adaptations
such as the safe consumption of rotting, pathogen-infected
meat and poisonous plants (52, 53); production of highly po-
tent toxins (54); bioluminescence (55); specific immunity to
various diseases (56) and microbial pathogens (57); regener-
ative capabilities (58); and, in some species, extreme longev-
ity (59, 60). Some of these adaptations, such as toxin
production and bioluminescence, are conferred, at least in
part, by microbial symbionts living in and on the animal (54,
55). However, despite these examples, a comprehensive view
of the association between animal traits and its microbiota is
still lacking.
Beyond its biotechnological potential, studying animal
microbiomes improves our understanding of host-microbe
ecology and bi-directional interactions. The microbiome af-
fects several factors in host physiology [immunity, digestion,
energy metabolism, development etc. (27, 30)]. In the other
direction, host diet and phylogeny are major forces that

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 1
shape the gut microbiome composition (25, 61–63). This rela-
tionship between host phylogeny and diet ignited studies of
the co-evolution of host-microbiome, but most of these stud-
ies have focused on mammals and/or were limited to differ-
ences in microbiome composition (27).
The microbiota of wild animals is also a natural reservoir
for pathogens of both animals and humans, whose mapping
could elucidate the timing and routes of their transmission
into the human population (64). Recent studies have mapped
pathogen reservoirs in wild animals (64, 65), demonstrating
the potential of this approach in surveillance and forecast of
human pathogen dynamics. Metagenomics have been used to
map antibiotic resistance genes (resistomes) in human,
chicken, pigs, migratory birds, and non-human primates (13,
66, 67). Many pathogenic microbe strains differ from non-
pathogenic strains of the same species at a-priori unknown
single genes (68), highlighting the importance of meta-
genomic sequencing approaches as only they can make such
distinctions, although pathogens are already being moni-
tored using more targeted analytical methods (69–71).
Finally, mapping the microbiota of wild animals could
also help in conservation efforts (72, 73). A recent functional
metagenomic analysis of a captive black rhinoceros revealed
a functional landscape acquired from domesticated livestock
microbiota, which may reflect a nutritional imbalance (74). A
similar conclusion was drawn in a study of the endangered
Tasmanian devil, suggesting that the microbiome shift ac-
quired in captivity negatively influences the ability of animals
reintroduced into the wild to adapt and survive (6). More re-
cent studies have also strengthened the importance of meta-
genomic insights on reintroduction and conservation efforts
of endangered species (75, 76). These studies highlight not
only the need to map wild animal microbiomes, but also po-
tentially to preserve them. A wildlife microbiota sample bank
could be an important resource for probiotics that could pro-
mote the health of animals after reintroduction to the wild.
Results
Constructing an annotated metagenomically-
assembled genome database of animals in the wild
To comprehensively study the microbiota of wild animals, we
assembled teams in five different geographical regions world-
wide, and collected fecal samples from more than 180 taxo-
nomically diverse animal species within these regions.
Animal species were identified from the known species in-
habiting the sampled regions during the sampled seasons, by
observations assisted by local experts. Our sampling plan
aimed to capture biodiversity clusters found in locations as
diverse as possible, and only where sampling and animal
identification could be done reliably and under the appropri-
ate permits. The species sampled and analyzed included the
fossa (C. ferox), Jameson’s Mamba snake (D. jamesoni), and
First release: 25 March 2021 www.sciencemag.org
an air-breathing catfish (C. gariepinus) (Fig. 1, A to C). We
designed a standardized sample collection process, managed
by a custom-designed, publicly-available mobile application
(Zerion Software iFormBuilder). The application was de-
signed by reviewing the collection procedure, to enable effi-
cient documentation of the metadata. The application
documents sampling time and geographical coordinates,
gathers manually-provided details such as animal species
identity, sex, health state, and other traits (when discernible),
and captures photographs of the animal and/or its fecal mat-
ter, when possible. Fecal matter was collected from identified
animals and as close to excretion as possible without disturb-
ing the animal in its natural habitat (table S1). All fecal mat-
ter was collected with the same protocol and using the same
collection kit, and then shipped to our laboratory. We then
Downloaded from http://science.sciencemag.org/ on March 26, 2021
processed and sequenced the metagenomes using the same
standard process.
To develop a stable and robust methodology, we first sam-
pled 121 animals that live in captivity, and then extended the
cohort by sampling 285 samples from animals in the wild,
while carefully implementing the developed collection meth-
odologies. We also assessed known genomic material by con-
structing a comprehensive bacterial reference database of
41,904 genomes obtained from recent microbiome studies
(35, 37–39) and from GenBank (77) (collectively denoted as
the “reference database”) and mapped our metagenomic
reads to this reference.
We obtained a median mapping rate of 7.2% sequencing
reads per sample, suggesting that the genetic material under
study is mostly unknown. To study our animal microbiota da-
taset, we thus adapted a pipeline (39) for de-novo bacterial
genome assembly from metagenomic data, resulting in the
construction of 5,080 genomic bins, which we subsequently
clustered into 1,209 species-level genomic bin clusters (table
S2). We selected a representative single genomic bin (rSGB)
from each bin cluster, used an algorithm that estimates the
relative abundance of these rSGBs across each sample and
annotated the genomes taxonomically (78) (Fig. 1D) (“wild
database”). Notably, this reference microbiota genome da-
taset (“reference database” + “wild database”) significantly
increased the mapping rate by nearly 3-fold to a median map-
ping rate of 21.0% per sample (Fig. 2A and fig. S1) (P<4x10−8,
Wilcoxon signed-rank test).
For every rSGB, we calculated the highest similarity that
it had to any species level genome bin in the reference data-
base. As the genome similarity measure, we used average nu-
cleotide identity, or ANI (79). The maximum ANI values were
used for novelty categorization as follows: rSGBs with maxi-
mum ANI values above 95% were defined as known species,
values below 83% as novel species, and values in between as
intermediate. Notably, less than 25% of the rSGBs in all ani-
mal classes were known, implying that our database contains
(Page numbers not final at time of first release) 2
previously unsampled bacterial species (Fig. 2B). As a control,
we applied the same metagenome assembly pipeline to 240
human gut microbiota samples that were not used in the con-
struction of the comparison reference database (80), and
found that over 99% of the 962 species-level genomic bins
that we constructed from these control human gut samples
were classified as known (Fig. 2B).
To assess the completeness of our database, we performed
a taxa accumulation analysis. Namely, we randomly ordered
our samples as well as the above control human gut microbi-
ome samples, and computed the total number of bacterial ge-
nomes that accumulate in these sample sets as a function of
samples added in this random order. As expected, the accu-
mulation rate of bacterial species of mammalian origin was
higher than that from human samples, likely because the
group of Mammalia is more diverse than human subjects
(Fig. 2C).
Among our sampled species, Aves (birds) and Osteich-
thyes (bony fish) accumulate new bacterial species at a lower
rate than Mammalia (Fig. 2C). Taking into account only novel
microbial species (ANI<83% relative to the reference data-
base), we found that those that originate in Mammalian,
Avian or Osteichthyan guts accumulate in nearly linear rate,
even when all samples are added. In contrast, microbiota
samples taken from humans rarely add novel bacterial spe-
cies to the existing reference set (Fig. 2D). These results sug-
gest that the gut microbiota of animals collected in the wild
represents a reservoir of yet unknown bacterial species. Ad-
ditional sampling from such hosts, as well as unsampled taxa,
should yield the discovery of novel genetic material.
This result may also explain why, despite mapping to our
bacterial genome database from animals in the wild, the
mapping rate was still relatively low (21%). From the above
accumulation curves, we hypothesize that the remaining
reads map to other genetic elements (e.g., bacteriophages and
plasmids) and to additional bacterial species of which cover-
age in our samples was too low to assemble from meta-
genomic sequencing, and require greater coverage depth or
samples from more animals. Comparing the accumulation
rate of bacterial species between wild and captive animals,
we found that the rate of accumulation of species was similar
between groups, with animals in the wild contributing more
novel microbial species (Fig. 2, E and F).
The phylogeny of the constructed genomes
We constructed a maximum-likelihood phylogenetic tree
with PhyloPhlAn 3.0 (81) for all 1,209 rSGBs, representing the
evolutionary landscape of wildlife microbiota (Fig. 3A and fig.
S2). Notably, the 1,209 wild microbial species were spread
across the prokaryotic tree of life (fig. S3). Additionally, the
novel species identified in this study contribute unknown va-
riety to the 18 phyla with 41 rSGBs that were not mapped to
First release: 25 March 2021 www.sciencemag.org
any known phylum (Fig. 3A). The number of novel bacteria
varies across different phyla (Fig. 3B), with Verrucomicrobia
being the most enriched (1/27 known rSGBs). Fusobacteria
exhibit a unique distribution of maximum ANI values, imply-
ing a taxonomic continuity between species and genus levels.
As an illustrative example, we examined a specific clade
of 11 novel species (star, Fig. 3C) that belong to the Verru-
comicrobia phylum, for which we reconstructed a distance-
based phylogenetic tree (Fig. 3C) and performed a Kyoto en-
cyclopedia of genes and genomes (KEGG) pathways enrich-
ment analysis (82). We found that this clade is enriched with
a unique subset of pathways, providing an initial framework
for understanding the role of this unknown bacterial clade.
This demonstrates the importance of applying comparative
methods to describe novel clades, for which taxonomic anno-
Downloaded from http://science.sciencemag.org/ on March 26, 2021
tation is limited. Another example is the reconstructed dis-
tance-based phylogenetic tree of the Spirochaetes phylum
(Fig. 3D). A particular clade of interest (star, Fig. 3D) belongs
entirely to the Treponema genus, which is known to contain
various human pathogens, such as the syphilis-causing T. pal-
lidum. Our database includes 13 new species in this genus,
which are also enriched with several pathways. Such findings
may assist in the discovery of weaker variants of known path-
ogens and serve as a basis for developing new vaccines by
identifying new antigenic proteins from these species.
Microbiota composition and co-existing bacterial
clusters
We next studied the relationship between animal microbiota
and host taxonomy and traits. To this end, we focused exclu-
sively on wild animals that are expected to have such associ-
ations, while avoiding possible biases that may result from
living in captivity. Starting with a coarse-grained view of
these relationships, we demonstrated gradually decreasing
pairwise Bray-Curtis distances between samples as a function
of the taxonomic group of reference (P<0.05 between species
and genus and P<0.0001 between other groups, Mann-Whit-
ney U test, Fig. 4A). For instance, animals within the same
order tend to have less similar microbiomes than animals
within the same family or genus, consistent with previous re-
ports (25).
We constructed a heat-map representation of the abun-
dance of the rSGBs in all samples from the wild. Even though
the presented abundance matrix is sparse, indicating the high
variability of microbial composition across the samples, small
clusters are evident (Fig. 4B). To identify these clusters, we
constructed the correlation matrix between abundance vec-
tors of all rSGBs (Fig. 4C). After clustering, by visual inspec-
tion, we identified several clusters of co-existing bacteria,
four of which are shown in Fig. 4D. Notably, visual examina-
tion identified these bacteria as belonging to various phyla,
with tendency to exist in hosts of the same (or mostly the
(Page numbers not final at time of first release) 3
same) class. For example, as seen in clusters 1, 2 and 4 (Fig.
4D), the clusters are associated with Mammalia, Reptilia and
Aves, respectively, and belong to at least two bacterial phyla.
Such a strong host association suggests that these clusters
may have a role in host-microbiome interactions. Cluster 1,
for example, consists of the bacterial phyla Firmicutes and
Bacteroidetes, consistent with a co-occurrence network re-
ported in a recent study (25), but also contains members of
other, less common, phyla. In some cases, as demonstrated in
cluster 3, the members of the cluster are found in various an-
imal classes, but all belong to the same bacterial phyla.
Microbiota composition associates with host class and
traits
We next investigated how microbiota composition is associ-
ated with various animal traits such as dietary adaptations,
activity hours, social structure, lifespan and weight, which we
carefully hand-curated for the entire cohort. First, we studied
how microbiota richness (evaluated using Shannon’s diver-
sity index) was related to host traits. We found that the mi-
crobiota of herbivores is significantly more diverse than that
of carnivores or omnivores, while controlling for phyloge-
netic confoundment (P<10−3, phyloANOVA, Fig. 5A), which is
consistent with previous studies (61).
To go beyond this coarse-level analyses, and to identify
specific bacterial species and particular taxonomic groups
that associate with animal class and traits, we studied the dif-
ferential abundance of rSGBs and various microbial taxo-
nomic ranks. First, we compared microbial differential
abundances across host classes (which are phylogenetically
disjoint), and found that 243 of 1,209 (20%) bacterial species
are significantly enriched for some classes [P<0.05, false dis-
covery rate (FDR) corrected for multiple hypotheses, Fig. 5B].
For example, rSGB 63, which represents the vitamin B12 pro-
ducing species Cetobacterium somare, is highly abundant in
Osteichthyes (average abundance of 23%, compared with less
than 0.5% abundance in the other classes). This is in agree-
ment with a previous study, reporting that C. somare is pre-
dominant in several freshwater fish that do not require
vitamin B12 supplementation (83).
Further, we studied the association between microbial
abundance and animal traits. Due to the dependence be-
tween various traits, which can be related to phylogenetic
confoundment (e.g., the mammalian order Carnivora, which
consists mostly of carnivores), we aimed to quantify and com-
pare the contribution of each trait/taxonomy to the differen-
tial abundance of each rSGB. We hypothesized that some
rSGBs have the strongest association with taxonomy, while
others associate with lifestyle traits. To examine this hypoth-
esis, we first calculated the pairwise effect size [specifically
the common language effect size of the Mann-Whitney U test
(84)] of each trait and host class on the bacterial abundance
First release: 25 March 2021 www.sciencemag.org
of each rSGB for all Aves and Mammalia from the wild (Fig.
5C). Indeed, some rSGBs exhibit stronger association with the
diet rather than with the host class. For example, rSGB 435,
an unknown species of the bacterial class Clostridia, which
we found exclusively in carnivores, is known to be related to
toxin production (85). This association may reveal mecha-
nisms of toxin metabolism in the guts of carnivores. In con-
trast, rSGB 292, an unknown species of the bacterial phylum
Firmicutes, was found solely in mammals, while having no
effect size related to the examined lifestyle traits. It was con-
structed from the microbiome of a southern sea lion (O. fla-
vescens), but was also found in lower abundance in several
species from the Carnivora, Primates and Artiodactyla orders,
having very different lifestyles and habitats. Interestingly,
such microbial species may have a strong interaction with a
Downloaded from http://science.sciencemag.org/ on March 26, 2021
specific taxonomic group, independent of the behavior or nu-
trition, indicative of a physiologically important host-micro-
biota interaction.
To further explore the discussed relations, we calculated
the effect size for microbial phyla (while taking the sum of
relative abundances of all corresponding rSGBs) (Fig. 5D).
For most of the reported phyla, the trait-related effect size is
higher than the taxonomical. Fusobacteria, for example, is
rarely found in herbivores (carnivores vs. herbivores abun-
dance fold change of 24.5, P<1.5x10−18, Mann-Whitney U test),
consistent with the potential role of Fusobacteria in protein
degradation and the high abundance of Fusobacteria in car-
rion eaters (10, 27, 86). Interestingly, the Fibrobacteres phy-
lum, which has an important role in cellulose degradation
(87), has the strongest association with diet, and was specifi-
cally found to be enriched in herbivores in comparison to
both carnivores and omnivores (P<1.3x10−7 and P<9x10−3, re-
spectively, FDR corrected Mann-Whitney U test). To address
effects that are confined at the order level, we repeated the
same analysis within Mammalia and Aves (see table S3).
We also found that the abundance of 18 and 14 microbial
species is associated with body mass or lifespan of Mammalia
(Fig. 5E) and Aves (Fig. 5F), respectively [calculated with phy-
logenetic generalized least-squares model (PGLS), P<0.05,
FDR corrected]. For Aves, we found association for higher
taxonomic ranks as well. Notably, in Mammalia, several
rSGBs show a similar association with both body mass and
lifespan (independent of phylogeny), in contrast to Aves, con-
sistent with the lower mass-lifespan correlation previously
reported for Aves (88). Interestingly, some rSGBs show inde-
pendent association to one of the discussed traits. While cau-
sation cannot be deduced, these results may suggest that
some members of the microbiota may play a role in the lon-
gevity of the host.
To conclude, we found various patterns related to the as-
sociation between microbiome composition and the taxon-
omy and/or lifestyle and physiological traits. While the
(Page numbers not final at time of first release) 4
confoundment between these traits and taxonomy is a key
difficulty for properly analyzing these associations, we
showed that in some cases (both at the microbial species and
the phylum level) the abundance is associated exclusively
with either the taxonomy or a trait. These findings suggest
that some microbial niches may have a specific role in the
adaptation to environmental conditions (regardless of the
taxonomy). In contrast, other microbes may have a co-evolu-
tionary role in specific taxonomy groups regardless of life-
style.
The functional landscape of the microbiota of animals
in the wild
To obtain a deeper functional view of the wild animal micro-
biota, we studied its gene functional content by performing
gene-calling on the derived bins, resulting in 12,125,877 puta-
tive genes. We functionally annotated them using the egg-
NOG mapper and specifically the KEGG database (82, 89).
Out of all putative genes, 1,880,067 belong to bins of medium-
quality or higher [with completeness of at least 50% and con-
tamination of less than 5%, in accordance with previous stud-
ies (39, 90)]. Of these, 355,688 (18.9%) did not receive any
cluster of orthologous genes (COG) annotation (89), while
other genes were classified into COG categories (Fig. 6A). We
built a similar profile of COG categories based on published
human dataset (37) and found that such a profile from hu-
man gut samples is significantly different from the profile
constructed for wild animals (P<10−30, chi-square).
We hypothesized that microbial functions are associated
with host characteristics, focusing on Mammalia from the
wild, since it is the largest class in the cohort. To this end, we
constructed sample-level profiles of KEGG annotations, by
counting the number of KEGG orthologs for each pathway.
We performed principal coordinates analysis (PCoA) based
on the Euclidean distance matrix followed by a phylogenetic-
corrected Mantel test, and found that the microbial func-
tional landscape is significantly correlated with diet (R2=0.35;
P<1.3x10−3, Fig. 6B), activity hours (R2=0.31; P<1.3x10−3, Fig.
6C), social structure (R2=0.24; P<8x10−3, Fig. 6D), body mass
(R2=0.36; P<1.3x10−3, Fig. 6E) and lifespan (R2=0.41;
P<1.3x10−3, Fig. 6F) (all P-values FDR corrected; for phy-
loPCA, see fig. S4). These results imply that the functional
landscape of the microbiome of an animal is tightly related
to physiological and lifestyle traits, independent of the phy-
logeny.
Motivated by diet-driven differences in Fig. 6B and Fig. 5,
C and D, we further hypothesized that specific functions are
associated with diet and specifically with metabolizing die-
tary toxins or handling diet-associated pathogens (such as
meat-borne pathogens). To examine this, we determined the
existence of KEGG orthologs in each sample, and performed
enrichment tests while controlling for purely phylogenetic
First release: 25 March 2021 www.sciencemag.org
effects. We found numerous orthologs that are enriched in
specific diet groups (Fig. 6G).
Next, we aimed to examine whether the enriched
orthologs correspond to specific physiological processes. To
this end, we tested for overrepresentation of these orthologs
in KEGG pathways. Our analysis revealed several intriguing
associations. For example, carnivore and omnivore microbi-
omes exhibited very significant enrichment in components of
the ABC transporter group (P=2.5x10−3 and P=3.4x10−23, re-
spectively, FDR corrected Fisher exact test for the enriched
orthologs in these groups relative to herbivores, Fig. 6H), and
the phosphotransferase system (PTS) (P=2.5x10−49 and
P=6.5x10−12, Fig. 6I), in agreement with previous reports (91,
92). Specifically, carnivore/omnivore-enriched orthologs in-
cluded ABC transporters for amino acids (e.g., methionine,
Downloaded from http://science.sciencemag.org/ on March 26, 2021
arginine, taurine, aspartic acid, etc.), phosphonates, phospho-
lipids, and ribose, reflecting the more diverse diet of these
hosts.
Enrichment in additional specific orthologs in carni-
vores/omnivores compared with herbivores suggests some in-
teresting functions of the microbiota of these hosts. For
example, we found enrichment in orthologs associated with
the biosynthesis of metabolites with potential antibiotic and
cytotoxic activity such as monobactam (K05375, an MbtH
protein, P=2x10−3, FDR corrected, in carnivores vs. herbi-
vores), which is consistent with recent studies (41–45). Also
interesting is the observed enrichment in antibiotic re-
sistance-associated orthologs in carnivores, which represents
multiple mechanisms, e.g., class C (K19215, P=0.02, FDR cor-
rected, in carnivores vs. herbivores) and D beta-lactamases
(K18793, P=0.02, FDR corrected, in carnivores vs. herbivores),
efflux pumps (K18147, P=0.007, FDR corrected, in carnivores
vs. herbivores), and penicillin-binding proteins. This plausi-
bly suggests that the microbiome is resistant (partially or
fully) to antibiotic compounds synthesized by some of its
members. As another example, we found the microbiota of
carnivores to be enriched in orthologs associated with cell
motility, such as flagellar assembly (K02386, flgA, P=6x10−4,
FDR corrected, in omnivores vs. herbivores. P=2x10−3, FDR
corrected, in carnivores vs. herbivores, but with phylogenetic
P=0.08) and bacterial chemotaxis (K02556, motA, P=0.05,
FDR corrected, in omnivores vs. herbivores). Finally, we
found that the carnivore microbiota exhibits significant en-
richment in orthologs associated with degradation of multi-
ple xenobiotics, such as toluene (K15763, P=0.007, FDR
corrected, in carnivores vs. herbivores), aminobenzoate
(K20712, P=3x10−4, FDR corrected in omnivores vs. herbi-
vores. P=9x10−4, FDR corrected in carnivores vs. herbivores,
but phylogenetic P=0.055), and other aromatic compounds.
Overall, these results unravel numerous potential functional
associations between microbiome-host interactions in wild
animals. Further studies are likely to strengthen these
(Page numbers not final at time of first release) 5
associations and may produce biological insights into the
contributions to the hosts that are made by microbiomes.
Discovery of novel toxin-metabolizing genes in carrion
eaters
Finally, as a proof of concept for the potential of our animal
microbiota resource to serve as a reservoir of bacterial func-
tions, we searched for genes that metabolize bacterial toxins
within the microbiota of carrion eaters. Our underlying hy-
pothesis was that microbial symbionts within the gut micro-
biota of carrion eaters confer at least part of the ability to
neutralize toxins in the food of their hosts. This hypothesis
was previously tested in new world vultures (10). For the val-
idation study, we focused on microbiota from the gut of grif-
fon vultures (Gyps fulvus), which we sampled weekly from
hatching until adulthood, over the course of 120 days. From
the assembled genomes, we identified a list of 693 proteases
[average length 268 amino acids (aa), range 39-973 aa, Fig.
7A], and annotated them using eggNOG (89) to identify pro-
tease types. We found a few large groups with known anno-
tations (e.g., 201 Serine endopeptidases and 194
Aminopeptidases) and 40 proteins without any annotation
(Fig. 7B), from which we selected for further experimentation
15 proteases which had the lowest similarity scores (Fig. 7, C
and D).
To test the ability of these 15 candidate proteases to break
down toxins, we synthesized these proteases, expressed them
as His-tagged proteins in E. coli at a 0.5 L production scale,
and purified them on a Ni2+ affinity resin. We then compiled
a list of 12 toxins, all with solved structures available from the
protein data bank (PDB), which are produced by bacteria and
are likely to exist in carrion (Fig. 7E). From these structures,
we computationally isolated 120 peptide sequences which
could function as substrates for proteolytic cleavage, based
on their position on the toxin and solvent accessibility. We
then incubated the synthetized proteases with the peptide
substrates on a peptide microarray. Notably, we identified
significant activity (P<0.01, FDR corrected one-sample Kol-
mogorov-Smirnov) in 50 of the 1,800 protease-substrate pairs
tested (Fig. 7F). We further validated these hits kinetically in
solution, using the putative peptide substrates synthesized as
Forster resonance energy transfer (FRET) substrates (Fig.
7G).
Taken together, these results experimentally validate the
presence of proteases, capable of metabolizing bacterial tox-
ins, in the gut microbiota of vultures. More generally, they
demonstrate the potential of finding functions within our an-
imal gut microbiota database and in particular, that such
functions may be unraveled by a discovery pipeline, in which
searches for specific functions are complemented by a syn-
thetic approach.
First release: 25 March 2021 www.sciencemag.org
Discussion
Organism-associated microbial communities have a com-
plex and profound impact on the biology of their hosts. They
critically contribute to host adaptation to its habitat, and are
linked in many cases, as has been recently shown in humans,
to the health and disease state of the host. However, our un-
derstanding of these phenomena in non-human hosts is
much less developed. Metagenomic studies of animal micro-
biota typically probed animals in captivity, which has likely
been altered compared to the microbiota of counterpart ani-
mals from their natural habitat. Other works have focused
more on microbial composition, using such methods as 16S
ribosomal RNA sequencing, providing mainly taxonomy-re-
lated insights (24, 25). Yet, a large-scale view at the meta-
genomic level is missing. Since the metagenomic approach
Downloaded from http://science.sciencemag.org/ on March 26, 2021
has significant advantages in the context of discovery of novel
organisms, genes, and functions, the motivation for this study
is particularly high.
In this study, we constructed a large metagenomic data-
base from over 180 distinct host species sampled in four dif-
ferent continents, including over 90 host species whose
microbiota was not previously profiled. We found that more
than 75% of the genomes in our database represent previ-
ously undescribed bacterial species. Notably, the rate at
which genomes were discovered is far from asymptote, moti-
vating further expansion of this collection. The animals that
we profiled cover diverse animal classes, dietary patterns, ac-
tivity hours, social structures, and lifespan, and we uncover
numerous associations between these traits and their micro-
biota composition and functional gene content. While some
of these associations may be correlative, others may point to
a role of the microbiota in conferring these traits, and repre-
sent many avenues for further research. To begin to uncover
some of the potential functions represented by this genetic
material, we computationally identified proteases within the
microbiota of carrion eaters and experimentally validated
their ability to cleave bacterial toxins.
The uniqueness of this study lies in the diverse set of sam-
ples from across the globe followed by state-of-the-art meta-
genomics methods to construct a database of genes and
genomes. This database allowed us to confirm that the wild
animal microbiome is extensively unexplored, with more
than 75% of the genetic material from wild species uncharac-
terized. We demonstrate patterns related to microbial com-
position and animal traits. Some of these associations are
consistent with the previous findings, but others have not
been reported to our knowledge. For example, we identified
bacteria of a previously unidentified species that appears to
be specific to mammals and was found across mammals with
diverse characteristics. Further, we hypothesized that micro-
biome function is related to animal traits and provides in-
sight about the co-evolution of the host and its microbiota.
(Page numbers not final at time of first release) 6
Finally, our experimental study identified that much poten-
tial remains for discovery inside animal microbiomes.
Our results enable a glimpse into potentially functional
roles of gut microbiomes in the biology of their hosts. For ex-
ample, we found enrichment in orthologs associated with an-
tibiotic biosynthesis and bacterial motility. This combination
highlights the possibility that carnivore microbiomes are flex-
ible functionally and anatomically, having been adapted to
rapidly responding to the active and dynamic environment in
which they subsist. This pattern is absent from herbivore mi-
crobiomes, as their hosts’ gastrointestinal tracts are signifi-
cantly longer and structurally complex, and their diet is less
likely to be contaminated with animal pathogens. The ob-
served enrichment in orthologs associated with xenobiotic
detoxification may, hypothetically, reflect the bioaccumula-
tion gradients of environmental pollutants observed in wild-
life food chains (93, 94), deriving from the trophic transfer of
pollutants from plants to herbivores, omnivores, and finally
to carnivores, where they accumulate in the highest concen-
trations. As many of these compounds are man-made con-
taminants, the presence of degradation pathways could
suggest that these pathways were acquired relatively recently,
possibly as a result of habitat contamination by industrial
waste. This is speculative, as some ubiquitous detoxification
mechanisms, such as glutathione, have a much earlier origin.
Our study has several limitations. First, our assembly-
based approach relies on sufficient coverage for both assem-
bly and binning. Thus, low-abundance bacteria, as well as mi-
crobes that have larger genomes, are under-represented in
our database. Furthermore, the presented wildlife cohort
does not uniformly cover the entire animal tree of life. The
high diversity of our wildlife cohort along with the tight con-
foundment between traits and phylogeny, made phylogenetic
correction difficult. For example, we were only able to com-
pare effect sizes within the same order, while confoundment
at lower taxonomy may exist. In addition, methods for phylo-
genetically corrected dimensionality reduction methods [e.g.,
principal component analysis (PCA)] are still lacking or lim-
ited (95–97). Lastly, the short read-based approach has some
limitations, such as difficulties in reconstruction of low com-
plexity genomic regions. Future studies can overcome this
difficulty and improve the overall assembly by relying on
complementary long-read sequencing approaches to provide
a crude genomic landscape of the sequenced sample.
The database described here could yield insights into the
ecology of sampled animal species. An interesting example
could be the reconstruction of food chains using foreign ani-
mal DNA found in the gut of sampled animals, defining them
as prey or predator species. Another example could be the
inference of interaction, proximity, or social structure in wild
animals from shared microbial species. Recent reports high-
light this as a feasible direction [(31, 98, 99) to name a few
First release: 25 March 2021 www.sciencemag.org
among various studies].
Mapping the microbiota of animals in the wild could shed
light on the natural reservoir of microbial pathogens. In par-
ticular, it could enable the surveillance and tracking of the
movements of specific pathogens from their animal hosts
into the human population, and enable early warning of po-
tential outbreaks. In an initial screen, we identified an esti-
mated 100 bacterial species that are known human
pathogens, such as Klebsiella, Enterobacter, Shigella, and
Clostridium species, including several pathogens of plants,
such as Pantotea. Further studies should elucidate the feasi-
bility of reservoir mapping relying on our database, for the
benefit of public health.
Our study demonstrates the importance and potential of
complementing the computational work with large synthetic
Downloaded from http://science.sciencemag.org/ on March 26, 2021
gene libraries for validation purposes. Techniques for high-
throughput synthesis of genomes or gene libraries have been
recently described, some of them for the explicit purpose of
functional protein testing (100, 101). Such gene library scale
is important not only because it improves the probability of
identifying real functions, but also because it allows the syn-
thesis and biochemical interrogation of entire pathways. Fur-
ther work on larger pools, on the order of 1,000 genes, can
now be done using the same approach. For example, an initial
search that we conducted, unraveled an estimated 2,000 sys-
tems of clustered regularly interspaced short palindromic re-
peats (CRISPR)/Cas, of which approximately one-third did
not belong to any described type.
In summary, our work describes a large-scale, annotated
metagenomic database of the microbiota of animals in the
wild, and demonstrates that it is a promising resource for the
discovery of novel biological functions and technologies. The
methods we describe could be implemented in the future con-
struction of metagenomic databases and their investigation.
Since even our large wildlife microbiome database is far from
saturation, further expanding it to cover additional species
may unravel unknown roles for the microbiota across diverse
habitats and extraordinary traits.
Methods summary
The cohort consists of 406 samples representing 184 unique
species. Fresh fecal samples were collected using a non-inva-
sive method. Collection was performed by the authors, ac-
companied by local guides or rangers in each country. Animal
species were identified from the known species inhabiting
the sampled regions during the sampled seasons, by observa-
tions assisted by local experts. Species-specific traits were
manually curated. All samples were sequenced using Illu-
mina NextSeq 500, following a QC and pre-processing proce-
dures. An assembly-based approach was taken, including de-
novo reads assembly, binning and construction of rSGBs.
Phylogenetic relationships were inferred by reconstructing
(Page numbers not final at time of first release) 7
the phylogenetic tree of all rSGBs. Novelty of rSGBs was as-
sessed by comparison to a collection of publicly available ref-
erences. For each sample, relative abundance of each rSGB
was estimated and followed by diversity, co-existence and dif-
ferential abundance analyses. Functional annotation (includ-
ing COG and KEGG) was inferred to each gene. To avoid
phylogenetic confoundment, various phylogenetic correction
methods were applied. For synthetic validation, genes were
selected, expressed in E. coli and purified using standard
methods. Proteases were screened for cleavage activity using
peptide arrays containing 120 solvent-accessible peptide se-
quences, ranging from 2 to 6 amino acids, derived from 12
bacterial toxins with solved structures in the PDB. All statis-
tical analyses were performed using hypothesis testing and
the resulting P-values were corrected to account for multiple
testing (FDR).
REFERENCES AND NOTES
1. The Human Microbiome Project Consortium, Structure, function and diversity of the
healthy human microbiome. Nature 486, 207–214 (2012).
doi:10.1038/nature11234 Medline
2. J. Qin, R. Li, J. Raes, M. Arumugam, K. S. Burgdorf, C. Manichanh, T. Nielsen, N.
Pons, F. Levenez, T. Yamada, D. R. Mende, J. Li, J. Xu, S. Li, D. Li, J. Cao, B. Wang,
H. Liang, H. Zheng, Y. Xie, J. Tap, P. Lepage, M. Bertalan, J.-M. Batto, T. Hansen,
D. Le Paslier, A. Linneberg, H. B. Nielsen, E. Pelletier, P. Renault, T. Sicheritz-
Ponten, K. Turner, H. Zhu, C. Yu, S. Li, M. Jian, Y. Zhou, Y. Li, X. Zhang, S. Li, N. Qin,
H. Yang, J. Wang, S. Brunak, J. Doré, F. Guarner, K. Kristiansen, O. Pedersen, J.
Parkhill, J. Weissenbach, P. Bork, S. D. Ehrlich, J. Wang, MetaHIT Consortium, A
human gut microbial gene catalogue established by metagenomic sequencing.
Nature 464, 59–65 (2010). doi:10.1038/nature08821 Medline
3. F. Warnecke, P. Luginbühl, N. Ivanova, M. Ghassemian, T. H. Richardson, J. T. Stege,
M. Cayouette, A. C. McHardy, G. Djordjevic, N. Aboushadi, R. Sorek, S. G. Tringe,
M. Podar, H. G. Martin, V. Kunin, D. Dalevi, J. Madejska, E. Kirton, D. Platt, E. Szeto,
A. Salamov, K. Barry, N. Mikhailova, N. C. Kyrpides, E. G. Matson, E. A. Ottesen, X.
Zhang, M. Hernández, C. Murillo, L. G. Acosta, I. Rigoutsos, G. Tamayo, B. D.
Green, C. Chang, E. M. Rubin, E. J. Mathur, D. E. Robertson, P. Hugenholtz, J. R.
Leadbetter, Metagenomic and functional analysis of hindgut microbiota of a
wood-feeding higher termite. Nature 450, 560–565 (2007).
doi:10.1038/nature06269 Medline
4. J. M. Brulc, D. A. Antonopoulos, M. E. B. Miller, M. K. Wilson, A. C. Yannarell, E. A.
Dinsdale, R. E. Edwards, E. D. Frank, J. B. Emerson, P. Wacklin, P. M. Coutinho, B.
Henrissat, K. E. Nelson, B. A. White, Gene-centric metagenomics of the fiber-
adherent bovine rumen microbiome reveals forage specific glycoside hydrolases.
Proc. Natl. Acad. Sci. U.S.A. 106, 1948–1953 (2009).
doi:10.1073/pnas.0806191105 Medline
5. R. Isaacson, H. B. Kim, The intestinal microbiome of the pig. Anim. Health Res. Rev.
13, 100–109 (2012). doi:10.1017/S1466252312000084 Medline
6. Y. Cheng, S. Fox, D. Pemberton, C. Hogg, A. T. Papenfuss, K. Belov, The Tasmanian
devil microbiome-implications for conservation and management. Microbiome 3,
76 (2015). doi:10.1186/s40168-015-0143-0 Medline
7. F. Kong, J. Zhao, S. Han, B. Zeng, J. Yang, X. Si, B. Yang, M. Yang, H. Xu, Y. Li,
Characterization of the gut microbiota in the red panda (Ailurus fulgens). PLOS
ONE 9, e87885 (2014). doi:10.1371/journal.pone.0087885 Medline
8. L. Zhu, Q. Wu, J. Dai, S. Zhang, F. Wei, Evidence of cellulose metabolism by the giant
panda gut microbiome. Proc. Natl. Acad. Sci. U.S.A. 108, 17714–17719 (2011).
doi:10.1073/pnas.1017956108 Medline
9. K. R. Amato, C. J. Yeoman, A. Kent, N. Righini, F. Carbonero, A. Estrada, H. R.
Gaskins, R. M. Stumpf, S. Yildirim, M. Torralba, M. Gillis, B. A. Wilson, K. E. Nelson,
B. A. White, S. R. Leigh, Habitat degradation impacts black howler monkey
(Alouatta pigra) gastrointestinal microbiomes. ISME J. 7, 1344–1353 (2013).
doi:10.1038/ismej.2013.16 Medline
10. M. Roggenbuck, I. Bærholm Schnell, N. Blom, J. Bælum, M. F. Bertelsen, T.
First release: 25 March 2021 www.sciencemag.org
Sicheritz-Pontén, S. J. Sørensen, M. T. P. Gilbert, G. R. Graves, L. H. Hansen, The
microbiome of New World vultures. Nat. Commun. 5, 5498 (2014).
doi:10.1038/ncomms6498 Medline
11. J. Quan, Z. Wu, Y. Ye, L. Peng, J. Wu, D. Ruan, Y. Qiu, R. Ding, X. Wang, E. Zheng, G.
Cai, W. Huang, J. Yang, Metagenomic characterization of intestinal regions in pigs
with contrasting feed efficiency. Front. Microbiol. 11, 32 (2020).
doi:10.3389/fmicb.2020.00032 Medline
12. J. L. Chun, S. Y. Ji, S. D. Lee, Y. K. Lee, B. Kim, K. H. Kim, Difference of gut
microbiota composition based on the body condition scores in dogs. J. Anim. Sci.
Technol. 62, 239–246 (2020). Medline
13. T. P. Campbell, X. Sun, V. H. Patel, C. Sanz, D. Morgan, G. Dantas, The microbiome
and resistome of chimpanzees, gorillas, and humans across host lifestyle and
geography. ISME J. 14, 1584–1599 (2020). doi:10.1038/s41396-020-0634-2
Medline
14. B. D. Muegge, J. Kuczynski, D. Knights, J. C. Clemente, A. González, L. Fontana, B.
Henrissat, R. Knight, J. I. Gordon, Diet drives convergence in gut microbiome
functions across mammalian phylogeny and within humans. Science 332, 970–
974 (2011). doi:10.1126/science.1198719 Medline
Downloaded from http://science.sciencemag.org/ on March 26, 2021
15. S. Manara, F. Asnicar, F. Beghini, D. Bazzani, F. Cumbo, M. Zolfo, E. Nigro, N.
Karcher, P. Manghi, M. I. Metzger, E. Pasolli, N. Segata, Microbial genomes from
non-human primate gut metagenomes expand the primate-associated bacterial
tree of life with over 1000 novel species. Genome Biol. 20, 299 (2019).
doi:10.1186/s13059-019-1923-9 Medline
16. K. R. Amato, J. G Sanders, S. J. Song, M. Nute, J. L. Metcalf, L. R. Thompson, J. T.
Morton, A. Amir, V. J McKenzie, G. Humphrey, G. Gogul, J. Gaffney, A. L Baden, G.
A O Britton, F. P Cuozzo, A. Di Fiore, N. J Dominy, T. L Goldberg, A. Gomez, M. M.
Kowalewski, R. J Lewis, A. Link, M. L Sauther, S. Tecot, B. A White, K. E Nelson, R.
M Stumpf, R. Knight, S. R Leigh, Evolutionary trends in host physiology outweigh
dietary niche in structuring primate gut microbiomes. ISME J. 13, 576–587 (2019).
doi:10.1038/s41396-018-0175-0 Medline
17. T. R. Kartzinel, P. A. Chen, T. C. Coverdale, D. L. Erickson, W. J. Kress, M. L.
Kuzmina, D. I. Rubenstein, W. Wang, R. M. Pringle, DNA metabarcoding illuminates
dietary niche partitioning by African large herbivores. Proc. Natl. Acad. Sci. U.S.A.
112, 8019–8024 (2015). doi:10.1073/pnas.1503283112 Medline
18. M. De Barba, C. Miquel, F. Boyer, C. Mercier, D. Rioux, E. Coissac, P. Taberlet, DNA
metabarcoding multiplexing and validation of data accuracy for diet assessment:
Application to omnivorous diet. Mol. Ecol. Resour. 14, 306–323 (2014).
doi:10.1111/1755-0998.12188 Medline
19. F. Pompanon, B. E. Deagle, W. O. C. Symondson, D. S. Brown, S. N. Jarman, P.
Taberlet, Who is eating what: Diet assessment using next generation sequencing.
Mol. Ecol. 21, 1931–1950 (2012). doi:10.1111/j.1365-294X.2011.05403.x Medline
20. C. M. Lopes, M. De Barba, F. Boyer, C. Mercier, P. J. S. da Silva Filho, L. M.
Heidtmann, D. Galiano, B. B. Kubiak, P. Langone, F. M. Garcias, L. Gielly, E.
Coissac, T. R. O. de Freitas, P. Taberlet, DNA metabarcoding diet analysis for
species with parapatric vs sympatric distribution: A case study on subterranean
rodents. Heredity 114, 525–536 (2015). doi:10.1038/hdy.2014.109 Medline
21. B. E. Deagle, A. C. Thomas, J. C. McInnes, L. J. Clarke, E. J. Vesterinen, E. L. Clare,
T. R. Kartzinel, J. P. Eveson, Counting with DNA in metabarcoding studies: How
should we convert sequence reads to dietary data? Mol. Ecol. 28, 391–406 (2019).
doi:10.1111/mec.14734 Medline
22. V. A. Mata, H. Rebelo, F. Amorim, G. F. McCracken, S. Jarman, P. Beja, How much
is enough? Effects of technical and biological replication on metabarcoding
dietary analysis. Mol. Ecol. 28, 165–175 (2019). doi:10.1111/mec.14779 Medline
23. A. Alberdi, O. Razgour, O. Aizpurua, R. Novella-Fernandez, J. Aihartza, I. Budinski,
I. Garin, C. Ibáñez, E. Izagirre, H. Rebelo, D. Russo, A. Vlaschenko, V. Zhelyazkova,
V. Zrnčić, M. T. P. Gilbert, DNA metabarcoding and spatial modelling link diet
diversification with distribution homogeneity in European bats. Nat. Commun. 11,
1154 (2020). doi:10.1038/s41467-020-14961-2 Medline
24. S. J. Song, J. G. Sanders, F. Delsuc, J. Metcalf, K. Amato, M. W. Taylor, F. Mazel, H.
L. Lutz, K. Winker, G. R. Graves, G. Humphrey, J. A. Gilbert, S. J. Hackett, K. P.
White, H. R. Skeen, S. M. Kurtis, J. Withrow, T. Braile, M. Miller, K. G. McCracken,
J. M. Maley, V. O. Ezenwa, A. Williams, J. M. Blanton, V. J. McKenzie, R. Knight,
Comparative analyses of vertebrate gut microbiomes reveal convergence
between birds and bats. mBio 11, e02901-19 (2020). doi:10.1128/mBio.02901-19
Medline
(Page numbers not final at time of first release) 8
25. N. D. Youngblut, G. H. Reischer, W. Walters, N. Schuster, C. Walzer, G. Stalder, R.
E. Ley, A. H. Farnleitner, Host diet and evolutionary history explain different
aspects of gut microbiome diversity among vertebrate clades. Nat. Commun. 10,
2200 (2019). doi:10.1038/s41467-019-10191-3 Medline
26. K. J. McDermid, R. P. Kittle III, A. Veillet, S. Plouviez, L. Muehlstein, G. H. Balazs,
Identification of gastrointestinal microbiota in Hawaiian green turtles (Chelonia
mydas). Evol. Bioinform. Online 16, 1176934320914603 (2020).
doi:10.1177/1176934320914603 Medline
27. T. J. Colston, C. R. Jackson, Microbiome evolution along divergent branches of the
vertebrate tree of life: What is known and unknown. Mol. Ecol. 25, 3776–3800
(2016). doi:10.1111/mec.13730 Medline
28. G. Bennett, M. Malone, M. L. Sauther, F. P. Cuozzo, B. White, K. E. Nelson, R. M.
Stumpf, R. Knight, S. R. Leigh, K. R. Amato, Host age, social group, and habitat
type influence the gut microbiota of wild ring-tailed lemurs (Lemur catta). Am. J.
Primatol. 78, 883–892 (2016). doi:10.1002/ajp.22555 Medline
29. M. Gacias, S. Gaspari, P.-M. G. Santos, S. Tamburini, M. Andrade, F. Zhang, N.
Shen, V. Tolstikov, M. A. Kiebish, J. L. Dupree, V. Zachariou, J. C. Clemente, P.
Casaccia, Microbiota-driven transcriptional changes in prefrontal cortex override
genetic differences in social behavior. eLife 5, e13442 (2016).
doi:10.7554/eLife.13442 Medline
30. E. Münger, A. J. Montiel-Castro, W. Langhans, G. Pacheco-López, Reciprocal
interactions between gut microbiota and host social behavior. Front. Integr.
Neurosci. 12, 21 (2018). doi:10.3389/fnint.2018.00021 Medline
31. A. H. Moeller, S. Foerster, M. L. Wilson, A. E. Pusey, B. H. Hahn, H. Ochman, Social
behavior shapes the chimpanzee pan-microbiome. Sci. Adv. 2, e1500997 (2016).
doi:10.1126/sciadv.1500997 Medline
32. V. O. Ezenwa, N. M. Gerardo, D. W. Inouye, M. Medina, J. B. Xavier, Animal behavior
and the microbiome. Science 338, 198–199 (2012). doi:10.1126/science.1227412
Medline
33. P. Luczynski, K. A. McVey Neufeld, C. S. Oriach, G. Clarke, T. G. Dinan, J. F. Cryan,
Growing up in a bubble: Using germ-free animals to assess the influence of the gut
microbiota on brain and behavior. Int. J. Neuropsychopharmacol. 19, pyw020
(2016). doi:10.1093/ijnp/pyw020 Medline
34. S. C. Forster, N. Kumar, B. O. Anonye, A. Almeida, E. Viciani, M. D. Stares, M. Dunn,
T. T. Mkandawire, A. Zhu, Y. Shao, L. J. Pike, T. Louie, H. P. Browne, A. L. Mitchell,
B. A. Neville, R. D. Finn, T. D. Lawley, A human gut bacterial genome and culture
collection for improved metagenomic analyses. Nat. Biotechnol. 37, 186–192
(2019). doi:10.1038/s41587-018-0009-7 Medline
35. D. H. Parks, C. Rinke, M. Chuvochina, P.-A. Chaumeil, B. J. Woodcroft, P. N. Evans,
P. Hugenholtz, G. W. Tyson, Recovery of nearly 8,000 metagenome-assembled
genomes substantially expands the tree of life. Nat. Microbiol. 2, 1533–1542
(2017). doi:10.1038/s41564-017-0012-7 Medline
36. R. D. Stewart, M. D. Auffret, A. Warr, A. W. Walker, R. Roehe, M. Watson,
Compendium of 4,941 rumen metagenome-assembled genomes for rumen
microbiome biology and enzyme discovery. Nat. Biotechnol. 37, 953–961 (2019).
doi:10.1038/s41587-019-0202-3 Medline
37. A. Almeida, A. L. Mitchell, M. Boland, S. C. Forster, G. B. Gloor, A. Tarkowska, T. D.
Lawley, R. D. Finn, A new genomic blueprint of the human gut microbiota. Nature
568, 499–504 (2019). doi:10.1038/s41586-019-0965-1 Medline
38. S. Nayfach, Z. J. Shi, R. Seshadri, K. S. Pollard, N. C. Kyrpides, New insights from
uncultivated genomes of the global human gut microbiome. Nature 568, 505–510
(2019). doi:10.1038/s41586-019-1058-x Medline
39. E. Pasolli, F. Asnicar, S. Manara, M. Zolfo, N. Karcher, F. Armanini, F. Beghini, P.
Manghi, A. Tett, P. Ghensi, M. C. Collado, B. L. Rice, C. DuLong, X. C. Morgan, C. D.
Golden, C. Quince, C. Huttenhower, N. Segata, Extensive unexplored human
microbiome diversity revealed by over 150,000 genomes from metagenomes
spanning age, geography, and lifestyle. Cell 176, 649–662.e20 (2019).
doi:10.1016/j.cell.2019.01.001 Medline
40. A. Almeida, S. Nayfach, M. Boland, F. Strozzi, M. Beracochea, Z. J. Shi, K. S.
Pollard, E. Sakharova, D. H. Parks, P. Hugenholtz, N. Segata, N. C. Kyrpides, R. D.
Finn, A unified catalog of 204,938 reference genomes from the human gut
microbiome. Nat. Biotechnol. 39, 105–114 (2021). doi:10.1038/s41587-020-
0603-3 Medline
41. S. M. Pimentel-Elardo, L. Grozdanov, S. Proksch, U. Hentschel, Diversity of
nonribosomal peptide synthetase genes in the microbial metagenomes of marine
First release: 25 March 2021 www.sciencemag.org
sponges. Mar. Drugs 10, 1192–1202 (2012). doi:10.3390/md10061192 Medline
42. N. Akbar, R. Siddiqui, K. A. Sagathevan, N. A. Khan, Gut bacteria of animals/pests
living in polluted environments are a potential source of antibacterials. Appl.
Microbiol. Biotechnol. 103, 3955–3964 (2019). doi:10.1007/s00253-019-09783-
2 Medline
43. M. G. Chevrette, C. M. Carlson, H. E. Ortega, C. Thomas, G. E. Ananiev, K. J. Barns,
A. J. Book, J. Cagnazzo, C. Carlos, W. Flanigan, K. J. Grubbs, H. A. Horn, F. M.
Hoffmann, J. L. Klassen, J. J. Knack, G. R. Lewin, B. R. McDonald, L. Muller, W. G.
P. Melo, A. A. Pinto-Tomás, A. Schmitz, E. Wendt-Pienkowski, S. Wildman, M.
Zhao, F. Zhang, T. S. Bugni, D. R. Andes, M. T. Pupo, C. R. Currie, The antimicrobial
potential of Streptomyces from insect microbiomes. Nat. Commun. 10, 516
(2019). doi:10.1038/s41467-019-08438-0 Medline
44. J. G. Ganley, G. Carr, T. R. Ioerger, J. C. Sacchettini, J. Clardy, E. R. Derbyshire,
Discovery of antimicrobial lipodepsipeptides produced by a Serratia sp. within
mosquito microbiomes. ChemBioChem 19, 1590–1594 (2018).
doi:10.1002/cbic.201800124 Medline
45. J. L. Ochoa, L. M. Sanchez, B.-M. Koo, J. S. Doherty, M. Rajendram, K. C. Huang, C.
A. Gross, R. G. Linington, Marine mammal microbiota yields novel antibiotic with
Downloaded from http://science.sciencemag.org/ on March 26, 2021
potent activity against Clostridium difficile. ACS Infect. Dis. 4, 59–67 (2018).
doi:10.1021/acsinfecdis.7b00105 Medline
46. M. Yin, X. Yan, W. Weng, Y. Yang, R. Gao, M. Liu, C. Pan, Q. Zhu, H. Li, Q. Wei, T.
Shen, Y. Ma, H. Qin, Micro Integral Membrane Protein (MIMP), a newly discovered
anti-inflammatory protein of Lactobacillus plantarum, enhances the gut barrier
and modulates microbiota and inflammatory cytokines. Cell. Physiol. Biochem.
45, 474–490 (2018). doi:10.1159/000487027 Medline
47. K. Rashamuse, W. Sanyika Tendai, K. Mathiba, T. Ngcobo, S. Mtimka, D. Brady,
Metagenomic mining of glycoside hydrolases from the hindgut bacterial
symbionts of a termite (Trinervitermes trinervoides) and the characterization of a
multimodular β-1,4-xylanase (GH11). Biotechnol. Appl. Biochem. 64, 174–186
(2017). doi:10.1002/bab.1480 Medline
48. C. Song, B. Wang, J. Tan, L. Zhu, D. Lou, Discovery of tauroursodeoxycholic acid
biotransformation enzymes from the gut microbiome of black bears using
metagenomics. Sci. Rep. 7, 45495 (2017). doi:10.1038/srep45495 Medline
49. R. C. de Freitas, H. I. F. Marques, M. A. C. da Silva, A. Cavalett, E. J. Odisi, B. L. D.
Silva, J. E. Montemor, T. Toyofuku, C. Kato, K. Fujikura, H. Kitazato, A. O. S. Lima,
Evidence of selective pressure in whale fall microbiome proteins and its potential
application to industry. Mar. Genomics 45, 21–27 (2019).
doi:10.1016/j.margen.2018.11.004 Medline
50. L. Ufarté, E. Laville, S. Duquesne, D. Morgavi, P. Robe, C. Klopp, A. Rizzo, S. Pizzut-
Serin, G. Potocki-Veronese, Discovery of carbamate degrading enzymes by
functional metagenomics. PLOS ONE 12, e0189201 (2017).
doi:10.1371/journal.pone.0189201 Medline
51. K. M. Singh, B. Reddy, D. Patel, A. K. Patel, N. Parmar, A. Patel, J. B. Patel, C. G.
Joshi, High potential source for biomass degradation enzyme discovery and
environmental aspects revealed through metagenomics of Indian buffalo rumen.
BioMed Res. Int. 2014, 267189 (2014). doi:10.1155/2014/267189 Medline
52. H. Huang, S. Yie, Y. Liu, C. Wang, Z. Cai, W. Zhang, J. Lan, X. Huang, L. Luo, K. Cai,
R. Hou, Z. Zhang, Dietary resources shape the adaptive changes of cyanide
detoxification function in giant panda (Ailuropoda melanoleuca). Sci. Rep. 6,
34700 (2016). doi:10.1038/srep34700 Medline
53. J. Lanszki, M. W. Hayward, N. Nagyapáti, Feeding responses of the golden jackal
after reduction of anthropogenic food subsidies. PLOS ONE 13, e0208727 (2018).
doi:10.1371/journal.pone.0208727 Medline
54. V. Pratheepa, V. Vasconcelos, Microbial diversity associated with tetrodotoxin
production in marine organisms. Environ. Toxicol. Pharmacol. 36, 1046–1054
(2013). doi:10.1016/j.etap.2013.08.013 Medline
55. E. A. Widder, Bioluminescence in the ocean: Origins of biological, chemical, and
ecological diversity. Science 328, 704–708 (2010). doi:10.1126/science.1174269
Medline
56. A. Seluanov, C. Hine, J. Azpurua, M. Feigenson, M. Bozzella, Z. Mao, K. C. Catania,
V. Gorbunova, Hypersensitivity to contact inhibition provides a clue to cancer
resistance of naked mole-rat. Proc. Natl. Acad. Sci. U.S.A. 106, 19352–19357
(2009). doi:10.1073/pnas.0905252106 Medline
57. L. P. Villarreal, Viruses and the Evolution of Life (ASM Press, 2005).
58. L. Alibardi, Tail regeneration in Lepidosauria as an exception to the generalized
(Page numbers not final at time of first release) 9
 lack of organ regeneration in amniotes. J. Exp. Zool. B Mol. Dev. Evol. 336, 145–
164 (2021). doi:10.1002/jez.b.22901 Medline
59. V. Briggs-Gonzalez, C. Bonenfant, M. Basille, M. Cherkiss, J. Beauchamp, F.
Mazzotti, Life histories and conservation of long-lived reptiles, an illustration with
the American crocodile (Crocodylus acutus). J. Anim. Ecol. 86, 1102–1113 (2017).
doi:10.1111/1365-2656.12723 Medline
60. V. Quesada, S. Freitas-Rodríguez, J. Miller, J. G. Pérez-Silva, Z.-F. Jiang, W. Tapia,
O. Santiago-Fernández, D. Campos-Iglesias, L. F. K. Kuderna, M. Quinzin, M. G.
Álvarez, D. Carrero, L. B. Beheregaray, J. P. Gibbs, Y. Chiari, S. Glaberman, C. Ciofi,
M. Araujo-Voces, P. Mayoral, J. R. Arango, I. Tamargo-Gómez, D. Roiz-Valle, M.
Pascual-Torner, B. R. Evans, D. L. Edwards, R. C. Garrick, M. A. Russello, N.
Poulakakis, S. J. Gaughran, D. O. Rueda, G. Bretones, T. Marquès-Bonet, K. P.
White, A. Caccone, C. López-Otín, Giant tortoise genomes provide insights into
longevity and age-related disease. Nat. Ecol. Evol. 3, 87–95 (2019).
doi:10.1038/s41559-018-0733-x Medline
61. R. E. Ley, M. Hamady, C. Lozupone, P. J. Turnbaugh, R. R. Ramey, J. S. Bircher, M.
L. Schlegel, T. A. Tucker, M. D. Schrenzel, R. Knight, J. I. Gordon, Evolution of
mammals and their gut microbes. Science 320, 1647–1651 (2008).
doi:10.1126/science.1155725 Medline
62. M. Groussin, F. Mazel, J. G. Sanders, C. S. Smillie, S. Lavergne, W. Thuiller, E. J.
Alm, Unraveling the processes shaping mammalian gut microbiomes over
evolutionary time. Nat. Commun. 8, 14319 (2017). doi:10.1038/ncomms14319
Medline
63. A. H. Nishida, H. Ochman, Rates of gut microbiome divergence in mammals. Mol.
Ecol. 27, 1884–1897 (2018). doi:10.1111/mec.14473 Medline
64. M. A. Duarte, J. M. F. Silva, C. R. Brito, D. S. Teixeira, F. L. Melo, B. M. Ribeiro, T.
Nagata, F. S. Campos, Faecal virome analysis of wild animals from Brazil. Viruses
11, 803 (2019). doi:10.3390/v11090803 Medline
65. L. A. Ramírez-Martínez, E. Loza-Rubio, J. Mosqueda, M. L. González-Garay, G.
García-Espinosa, Fecal virome composition of migratory wild duck species. PLOS
ONE 13, e0206970 (2018). doi:10.1371/journal.pone.0206970 Medline
66. Y. Wang, Y. Hu, F. Liu, J. Cao, N. Lv, B. Zhu, G. Zhang, G. F. Gao, Integrated
metagenomic and metatranscriptomic profiling reveals differentially expressed
resistomes in human, chicken, and pig gut microbiomes. Environ. Int. 138, 105649
(2020). doi:10.1016/j.envint.2020.105649 Medline
67. J. Cao, Y. Hu, F. Liu, Y. Wang, Y. Bi, N. Lv, J. Li, B. Zhu, G. F. Gao, Metagenomic
analysis reveals the microbiome and resistome in migratory birds. Microbiome 8,
26 (2020). doi:10.1186/s40168-019-0781-8 Medline
68. E. V. Sokurenko, V. Chesnokova, D. E. Dykhuizen, I. Ofek, X. R. Wu, K. A. Krogfelt,
C. Struve, M. A. Schembri, D. L. Hasty, Pathogenic adaptation of Escherichia coli
by natural variation of the FimH adhesin. Proc. Natl. Acad. Sci. U.S.A. 95, 8922–
8926 (1998). doi:10.1073/pnas.95.15.8922 Medline
69. J. Brindha, K. Chanda, M. M. Balamurali, Biosensors for pathogen surveillance.
Environ. Chem. Lett. 16, 1325–1337 (2018). doi:10.1007/s10311-018-0759-y
70. S. Ali, A. Hassan, G. Hassan, C.-H. Eun, J. Bae, C. H. Lee, I.-J. Kim, Disposable all-
printed electronic biosensor for instantaneous detection and classification of
pathogens. Sci. Rep. 8, 5920 (2018). doi:10.1038/s41598-018-24208-2 Medline
71. J. Vidic, M. Manzano, C.-M. Chang, N. Jaffrezic-Renault, Advanced biosensors for
detection of pathogens related to livestock and poultry. Vet. Res. 48, 11 (2017).
doi:10.1186/s13567-017-0418-5 Medline
72. F. Wei, Q. Wu, Y. Hu, G. Huang, Y. Nie, L. Yan, Conservation metagenomics: A new
branch of conservation biology. Sci. China Life Sci. 62, 168–178 (2019).
doi:10.1007/s11427-018-9423-3 Medline
73. B. K. Trevelline, S. S. Fontaine, B. K. Hartup, K. D. Kohl, Conservation biology needs
a microbial renaissance: A call for the consideration of host-associated microbiota
in wildlife management practices. Proc. Biol. Sci. 286, 20182448 (2019).
doi:10.1098/rspb.2018.2448 Medline
74. K. M. Gibson, B. N. Nguyen, L. M. Neumann, M. Miller, P. Buss, S. Daniels, M. J. Ahn,
K. A. Crandall, B. Pukazhenthi, Gut microbiome differences between wild and
captive black rhinoceros – implications for rhino health. Sci. Rep. 9, 7570 (2019).
doi:10.1038/s41598-019-43875-3 Medline
75. Y. Sun, Y. Sun, Z. Shi, Z. Liu, C. Zhao, T. Lu, H. Gao, F. Zhu, R. Chen, J. Zhang, R.
Pan, B. Li, L. Teng, S. Guo, Gut microbiota of wild and captive alpine musk deer
(Moschus chrysogaster). Front. Microbiol. 10, 3156 (2020).
doi:10.3389/fmicb.2019.03156 Medline
First release: 25 March 2021 www.sciencemag.org
76. W. Guo, S. Mishra, C. Wang, H. Zhang, R. Ning, F. Kong, B. Zeng, J. Zhao, Y. Li,
Comparative study of gut microbiota in wild and captive giant pandas (Ailuropoda
melanoleuca). Genes 10, 827 (2019). doi:10.3390/genes10100827 Medline
77. E. W. Sayers, M. Cavanaugh, K. Clark, J. Ostell, K. D. Pruitt, I. Karsch-Mizrachi,
GenBank. Nucleic Acids Res. 48, D84–D86 (2020). Medline
78. F. A. B. von Meijenfeldt, K. Arkhipova, D. D. Cambuy, F. H. Coutinho, B. E. Dutilh,
Robust taxonomic classification of uncharted microbial sequences and bins with
CAT and BAT. Genome Biol. 20, 217 (2019). doi:10.1186/s13059-019-1817-x
Medline
79. C. Jain, L. M. Rodriguez-R, A. M. Phillippy, K. T. Konstantinidis, S. Aluru, High
throughput ANI analysis of 90K prokaryotic genomes reveals clear species
boundaries. Nat. Commun. 9, 5114 (2018). doi:10.1038/s41467-018-07641-9
Medline
80. D. Zeevi, T. Korem, N. Zmora, D. Israeli, D. Rothschild, A. Weinberger, O. Ben-
Yacov, D. Lador, T. Avnit-Sagi, M. Lotan-Pompan, J. Suez, J. A. Mahdi, E. Matot, G.
Malka, N. Kosower, M. Rein, G. Zilberman-Schapira, L. Dohnalová, M. Pevsner-
Fischer, R. Bikovsky, Z. Halpern, E. Elinav, E. Segal, Personalized nutrition by
prediction of glycemic responses. Cell 163, 1079–1094 (2015).
Downloaded from http://science.sciencemag.org/ on March 26, 2021
doi:10.1016/j.cell.2015.11.001 Medline
81. F. Asnicar, A. M. Thomas, F. Beghini, C. Mengoni, S. Manara, P. Manghi, Q. Zhu, M.
Bolzan, F. Cumbo, U. May, J. G. Sanders, M. Zolfo, E. Kopylova, E. Pasolli, R. Knight,
S. Mirarab, C. Huttenhower, N. Segata, Precise phylogenetic analysis of microbial
isolates and genomes from metagenomes using PhyloPhlAn 3.0. Nat. Commun.
11, 2500 (2020). doi:10.1038/s41467-020-16366-7 Medline
82. M. Kanehisa, S. Goto, KEGG: Kyoto encyclopedia of genes and genomes. Nucleic
Acids Res. 28, 27–30 (2000). doi:10.1093/nar/28.1.27 Medline
83. C. Tsuchiya, T. Sakata, H. Sugita, Novel ecological niche of Cetobacterium
somerae, an anaerobic bacterium in the intestinal tracts of freshwater fish. Lett.
Appl. Microbiol. 46, 43–48 (2008). Medline
84. K. O. McGraw, S. P. Wong, A common language effect size statistic. Psychol. Bull.
111, 361–365 (1992). doi:10.1037/0033-2909.111.2.361
85. L. Baldassi, Clostridial toxins: Potent poisons, potent medicines. J. Venom. Anim.
Toxins Incl. Trop. Dis. 11, 391–411 (2005). doi:10.1590/S1678-
91992005000400002
86. S. W. Keenan, A. S. Engel, R. M. Elsey, The alligator gut microbiome and
implications for archosaur symbioses. Sci. Rep. 3, 2877 (2013).
doi:10.1038/srep02877 Medline
87. E. Ransom-Jones, D. L. Jones, A. J. McCarthy, J. E. McDonald, The Fibrobacteres:
An important phylum of cellulose-degrading bacteria. Microb. Ecol. 63, 267–281
(2012). doi:10.1007/s00248-011-9998-1 Medline
88. B. Mayne, O. Berry, C. Davies, J. Farley, S. Jarman, A genomic predictor of lifespan
in vertebrates. Sci. Rep. 9, 17866 (2019). doi:10.1038/s41598-019-54447-w
Medline
89. J. Huerta-Cepas, D. Szklarczyk, D. Heller, A. Hernández-Plaza, S. K. Forslund, H.
Cook, D. R. Mende, I. Letunic, T. Rattei, L. J. Jensen, C. von Mering, P. Bork,
eggNOG 5.0: A hierarchical, functionally and phylogenetically annotated
orthology resource based on 5090 organisms and 2502 viruses. Nucleic Acids
Res. 47, D309–D314 (2019). doi:10.1093/nar/gky1085 Medline
90. R. M. Bowers, N. C. Kyrpides, R. Stepanauskas, M. Harmon-Smith, D. Doud, T. B.
K. Reddy, F. Schulz, J. Jarett, A. R. Rivers, E. A. Eloe-Fadrosh, S. G. Tringe, N. N.
Ivanova, A. Copeland, A. Clum, E. D. Becraft, R. R. Malmstrom, B. Birren, M. Podar,
P. Bork, G. M. Weinstock, G. M. Garrity, J. A. Dodsworth, S. Yooseph, G. Sutton, F.
O. Glöckner, J. A. Gilbert, W. C. Nelson, S. J. Hallam, S. P. Jungbluth, T. J. G.
Ettema, S. Tighe, K. T. Konstantinidis, W.-T. Liu, B. J. Baker, T. Rattei, J. A. Eisen,
B. Hedlund, K. D. McMahon, N. Fierer, R. Knight, R. Finn, G. Cochrane, I. Karsch-
Mizrachi, G. W. Tyson, C. Rinke, A. Lapidus, F. Meyer, P. Yilmaz, D. H. Parks, A. M.
Eren, L. Schriml, J. F. Banfield, P. Hugenholtz, T. Woyke, Genome Standards
Consortium, Minimum information about a single amplified genome (MISAG) and
a metagenome-assembled genome (MIMAG) of bacteria and archaea. Nat.
Biotechnol. 35, 725–731 (2017). doi:10.1038/nbt.3893 Medline
91. J. G. Sanders, A. C. Beichman, J. Roman, J. J. Scott, D. Emerson, J. J. McCarthy, P.
R. Girguis, Baleen whales host a unique gut microbiome with similarities to both
carnivores and herbivores. Nat. Commun. 6, 8285 (2015).
doi:10.1038/ncomms9285 Medline
92. L. Zhu, Q. Wu, C. Deng, M. Zhang, C. Zhang, H. Chen, G. Lu, F. Wei, Adaptive
(Page numbers not final at time of first release) 10
 evolution to a high purine and fat diet of carnivorans revealed by gut microbiomes
and host genomes. Environ. Microbiol. 20, 1711–1722 (2018). doi:10.1111/1462-
2920.14096 Medline
93. C. C. Rimmer, E. K. Miller, K. P. McFarland, R. J. Taylor, S. D. Faccio, Mercury
bioaccumulation and trophic transfer in the terrestrial food web of a montane
forest. Ecotoxicology 19, 697–709 (2010). doi:10.1007/s10646-009-0443-x
Medline
94. C. R. Dromard, Y. Bouchon-Navaro, S. Cordonnier, M. Guéné, M. Harmelin-Vivien,
C. Bouchon, Different transfer pathways of an organochlorine pesticide across
marine tropical food webs assessed with stable isotope analysis. PLOS ONE 13,
e0191335 (2018). doi:10.1371/journal.pone.0191335 Medline
95. L. J. Revell, Size-correction and principal components for interspecific
comparative studies. Evolution 63, 3258–3268 (2009). doi:10.1111/j.1558-
5646.2009.00804.x Medline
96. J. C. Uyeda, D. S. Caetano, M. W. Pennell, Comparative analysis of principal
components can be misleading. Syst. Biol. 64, 677–689 (2015).
doi:10.1093/sysbio/syv019 Medline
97. P. D. Polly, A. M. Lawing, A. Fabre, A. Goswami, Phylogenetic principal components
analysis and geometric morphometrics. Hystrix 24, 33–41 (2013).
doi:10.4404/hystrix-24.1-6383
98. J. Tung, L. B. Barreiro, M. B. Burns, J.-C. Grenier, J. Lynch, L. E. Grieneisen, J.
Altmann, S. C. Alberts, R. Blekhman, E. A. Archie, Social networks predict gut
microbiome composition in wild baboons. eLife 4, e05224 (2015).
doi:10.7554/eLife.05224 Medline
99. A. H. Moeller, T. A. Suzuki, M. Phifer-Rixey, M. W. Nachman, Transmission modes
of the mammalian gut microbiota. Science 362, 453–457 (2018).
doi:10.1126/science.aat7164 Medline
100. S. Kosuri, N. Eroshenko, E. M. Leproust, M. Super, J. Way, J. B. Li, G. M. Church,
Scalable gene synthesis by selective amplification of DNA pools from high-fidelity
microchips. Nat. Biotechnol. 28, 1295–1299 (2010). doi:10.1038/nbt.1716
Medline
101. C. Plesa, A. M. Sidore, N. B. Lubock, D. Zhang, S. Kosuri, Multiplexed gene
synthesis in emulsions for exploring protein functional landscapes. Science 359,
343–347 (2018). doi:10.1126/science.aao5167 Medline
102. D. Levin, N. Raab, Y. Pinto, D. Rothschild, I. Bachelet, E. Segal, Diversity and
functional landscapes in the microbiota of animals in the wild. Zenodo (2021);
https://doi.org/10.5281/zenodo.4419908.
103. J. Suez, T. Korem, D. Zeevi, G. Zilberman-Schapira, C. A. Thaiss, O. Maza, D.
Israeli, N. Zmora, S. Gilad, A. Weinberger, Y. Kuperman, A. Harmelin, I. Kolodkin-
Gal, H. Shapiro, Z. Halpern, E. Segal, E. Elinav, Artificial sweeteners induce glucose
intolerance by altering the gut microbiota. Nature 514, 181–186 (2014).
doi:10.1038/nature13793 Medline
104. Babraham Bioinformatics, FastQC: A quality control tool for high throughput
sequence data; www.bioinformatics.babraham.ac.uk/projects/fastqc/.
105. A. Mikheenko, V. Saveliev, A. Gurevich, MetaQUAST: Evaluation of metagenome
assemblies. Bioinformatics 32, 1088–1090 (2016).
doi:10.1093/bioinformatics/btv697 Medline
106. D. D. Kang, J. Froula, R. Egan, Z. Wang, MetaBAT, an efficient tool for accurately
reconstructing single genomes from complex microbial communities. PeerJ 3,
e1165 (2015). doi:10.7717/peerj.1165 Medline
107. D. D. Kang, F. Li, E. Kirton, A. Thomas, R. Egan, H. An, Z. Wang, MetaBAT 2: An
adaptive binning algorithm for robust and efficient genome reconstruction from
metagenome assemblies. PeerJ 7, e7359 (2019). doi:10.7717/peerj.7359 Medline
108. B. Langmead, S. L. Salzberg, Fast gapped-read alignment with Bowtie 2. Nat.
Methods 9, 357–359 (2012). doi:10.1038/nmeth.1923 Medline
109. SourceForge, BBMap; https://sourceforge.net/projects/bbmap/.
110. A. Criscuolo, A fast alignment-free bioinformatics procedure to infer accurate
distance-based phylogenetic trees from genome assemblies. Res. Ideas
Outcomes. 5, e36178 (2019). doi:10.3897/rio.5.e36178
111. B. D. Ondov, T. J. Treangen, P. Melsted, A. B. Mallonee, N. H. Bergman, S. Koren,
A. M. Phillippy, Mash: Fast genome and metagenome distance estimation using
MinHash. Genome Biol. 17, 132 (2016). doi:10.1186/s13059-016-0997-x Medline
112. V. Lefort, R. Desper, O. Gascuel, FastME 2.0: A comprehensive, accurate, and fast
distance-based phylogeny inference program. Mol. Biol. Evol. 32, 2798–2800
(2015). doi:10.1093/molbev/msv150 Medline
First release: 25 March 2021 www.sciencemag.org
113. I. Letunic, P. Bork, Interactive Tree Of Life (iTOL) v4: Recent updates and new
developments. Nucleic Acids Res. 47, W256–W259 (2019).
doi:10.1093/nar/gkz239 Medline
114. D. Rothschild, S. Leviatan, A. Hanemann, Y. Cohen, O. Weissbrod, E. Segal, An
atlas of robust microbiome associations with phenotypic traits based on large-
scale cohorts from two continents. bioRxiv 2020.05.28.122325 [Preprint]. 18
October 2020; https://doi.org/10.1101/2020.05.28.122325.
115. S. C. Watts, S. C. Ritchie, M. Inouye, K. E. Holt, FastSpar: Rapid and scalable
correlation estimation for compositional data. Bioinformatics 35, 1064–1066
(2019). doi:10.1093/bioinformatics/bty734 Medline
116. J. Friedman, E. J. Alm, Inferring correlation networks from genomic survey data.
PLOS Comput. Biol. 8, e1002687 (2012). doi:10.1371/journal.pcbi.1002687
Medline
117. J. Revell, phytools: An R package for phylogenetic comparative biology (and other
things). Methods Ecol. Evol. 3, 217–223 (2012). doi:10.1111/j.2041-
210X.2011.00169.x
118. C. M. Mlewa, J. M. Green, Biology of the marbled lungfish, Protopterus aethiopicus
Heckel, in Lake Baringo, Kenya. Afr. J. Ecol. 42, 338–345 (2004).
Downloaded from http://science.sciencemag.org/ on March 26, 2021
doi:10.1111/j.1365-2028.2004.00536.x
119. J. H. Steele, S. A. Thorpe, K. K. Turekian, Elements of Physical Oceanography: A
Derivative of the Encyclopedia of Ocean Sciences (Academic Press, 2009).
120. A. M. Martín-Platero, E. Valdivia, M. Ruíz-Rodríguez, J. J. Soler, M. Martín-Vivaldi,
M. Maqueda, M. Martínez-Bueno, Characterization of antimicrobial substances
produced by Enterococcus faecalis MRR 10-3, isolated from the uropygial gland
of the hoopoe (Upupa epops). Appl. Environ. Microbiol. 72, 4245–4249 (2006).
doi:10.1128/AEM.02940-05 Medline
121. W. Christidis, W. E. Boles, Systematics and Taxonomy of Australian Birds (Csiro
Publishing, 2008).
122. J. del Hoyo, A. Elliott, J. Sargatal, Eds., Ostrich to Ducks, vol. 1 of Handbook of the
Birds of the World (Lynx Edicions, 1992).
123. J. del Hoyo, A. Elliott, J. Sargatal, Eds., Hoatzin to Auks, vol. 3 of Handbook of the
Birds of the World (Lynx Edicions, 1996).
124. H. Shirihai, L. Svensson, Handbook of Western Palearctic Birds: Passerines (Helm,
2018).
125. J. Ferguson-Lees, D. A. Christie, Raptors of the World (Houghton Mifflin Harcourt,
2001).
126. D. E. Wilson, R. A. Mittermeier, Eds., Monotremes and Marsupials, vol. 5 of
Handbook of the Mammals of the World (Lynx Edicions, 2015).
127. S. H. Ridgway, R. Harrison, Eds., The Second Book of Dolphins and the Porpoises,
vol. 6 of Handbook of Marine Mammals (Elsevier, 1981).
128. R. A. Mittermeier, D. E. Wilson, A. B. Rylands, Eds., Primates, vol. 3 of Handbook
of the Mammals of the World (Lynx Edicions, 2013).
129. D. E. Wilson, R. A. Mittermeier, T. E. Lacher, Rodents II, vol. 7 of Handbook of the
Mammals of the World (Lynx Edicions, 2017).
130. D. E. Wilson, R. A. Mittermeier, Handbook of the Mammals of the World (Lynx
Edicions, 2009).
131. D. E. Wilson, R. A. Mittermeier, Carnivores, vol. 1 of Handbook of the Mammals of
the World: Carnivores (2009).
132. L. Baskin, K. Danell, Ecology of Ungulates: A Handbook of Species in Eastern
Europe and Northern and Central Asia (Springer, 2003).
133. W. D. Emmerson, A Guide to, and Checklist for, the Decapoda of Namibia, South
Africa and Mozambique (Volume 3) (Cambridge Scholars Publishing, 2017).
134. Clarias gariepinus summary page; www.fishbase.us/summary/Clarias-
gariepinus.html.
135. R. Burton, International Wildlife Encyclopedia (Marshall Cavendish, 2002).
136. R. C. Stebbins, A Field Guide to Western Reptiles and Amphibians (Houghton
Mifflin Harcourt, 2003).
137. S. P. Mackessy, Handbook of Venoms and Toxins of Reptiles (CRC Press, 2016).
138. T. Seemann, Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30,
2068–2069 (2014). doi:10.1093/bioinformatics/btu153 Medline
139. S. Kumar, G. Stecher, M. Suleski, S. B. Hedges, TimeTree: A resource for
timelines, timetrees, and divergence times. Mol. Biol. Evol. 34, 1812–1819 (2017).
doi:10.1093/molbev/msx116 Medline
140. D. Orme, R. Freckleton, G. Thomas, T. Petzoldt, S. Fritz, N. Isaac, W. Pearse, The
caper package: comparative analysis of phylogenetics and evolution in R. R
(Page numbers not final at time of first release) 11
 package version. 5, 1–36 (2013). Published online 25 March 2021
141. D. Melo, G. Garcia, A. Hubbe, A. P. Assis, G. Marroig, EvolQG – An R package for 10.1126/science.abb5352
evolutionary quantitative genetics. F1000 Res. 4, 925 (2015).
doi:10.12688/f1000research.7082.2 Medline
142. K. Afiahayati, K. Sato, Y. Sakakibara, MetaVelvet-SL: An extension of the Velvet
assembler to a de novo metagenomic assembler utilizing supervised learning.
DNA Res. 22, 69–77 (2015). doi:10.1093/dnares/dsu041 Medline
143. H. Noguchi, T. Taniguchi, T. Itoh, MetaGeneAnnotator: Detecting species-specific
patterns of ribosomal binding site for precise gene prediction in anonymous
prokaryotic and phage genomes. DNA Res. 15, 387–396 (2008).
doi:10.1093/dnares/dsn027 Medline
144. B. Buchfink, C. Xie, D. H. Huson, Fast and sensitive protein alignment using
DIAMOND. Nat. Methods 12, 59–60 (2015). doi:10.1038/nmeth.3176 Medline
ACKNOWLEDGMENTS
The authors wish to thank the entire team of Wild Biotech and the Segal lab at the
Weizmann Institute of Science for valuable assistance with this study; Dr. M.
Kopel and Dr. I. Horowitz for providing valuable assistance with sampling; Ms. K.

Downloaded from http://science.sciencemag.org/ on March 26, 2021

Hanna Bajic and Mr. O. Nachman Bajic for valuable discussions; Ms. H. Jessel for
assistance with graphical design and figure preparation; Mr. Y. Shoshan and Dr.
A. Abu-Horowitz for valuable administrative assistance; and Mr. Gal Zanir for
hosting sampling videos on his YouTube channel. Special Acknowledgments for
critical assistance with sample collection are given to Prof. L. Kei-Po Leung and
Ms. S. Minnis from Queensland University, Australia; the South Atlantic
Environment Research Institute (SAERI), Falkland Islands; the Budapest Zoo,
Hungary; Mr. Yosef Kiat and the Israeli Nature and Parks Authority, Israel; and
the Institute for the Conservation of Tropical Environments and Ms. A.
Andriatiavina, Madagascar. Funding: This work was funded by Wild Biotech, Ltd.
Author contributions: DL, NR, IB, and ES conceived the research. DL, NR, IB,
and YP performed experiments, analyzed data, and wrote the manuscript. GZ,
NM, and DG collected samples and analyzed data. DR, DF, AG, SL, and DZ
performed experiments and analyzed data. IB and ES provided valuable assets,
analyzed data, wrote the manuscript, and supervised research. Competing
interests: DL, NR, IB, YP, DF, GZ are shareholders and/or employees in Wild
Biotech Ltd, a synthetic metagenomics and discovery company from Rehovot,
Israel. ES is a paid consultant to Wild Biotech Ltd. The rest of the authors declare
no competing interests. Data and materials availability: Dataset is available for
academic use at http://fx9.cloud. Academic users will receive access to the
data, without intervention from the authors, once academic credentials have
been provided. They will be able to perform unlimited cloud-based computation
and bioinformatic analyses (including implementing their own software on the
cloud) and use the output freely. Commercial users will be directed to the
authors in order to gain access. The data are also available at Zenodo (102)
under accession number 4419908. Access must be requested and will be
granted to all academic users. Data collected using the app is available at
http://shorturl.at/fmzOS. Sampling videos are available at G. Zanir’s YouTube
channel (www.youtube.com/channel/UCnvJjBrio2ACkyIZqO6EVMw). Sampling
videos include emu (D. novaehollandiae, https://youtu.be/OYEMLs6dVb0);
black swan (C. atratus, https://youtu.be/Z4hcfbc5SxQ); red kangaroo (O. rufus,
https://youtu.be/kcCizhvdHYI) tasmanian devil (S. harrisii,
https://youtu.be/t3zNAbXr_-M); southern elephant seal (M. leonine,
https://youtu.be/5E5wviq1AtQ); common wombat (V. ursinus,
https://youtu.be/IZIkVIR7jxA). Requests for materials and collaborations with
Wild Biotech should be addressed to neta@wildbio.tech (N.R.).
SUPPLEMENTARY MATERIALS
science.sciencemag.org/cgi/content/full/science.abb5352/DC1
Materials and Methods
Figs. S1 to S4
Tables S1 to S7
References (103-144)
MDAR Reproducibility Checklist

2 March 2020; resubmitted 17 July 2020

Accepted 9 March 2021

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 12
 Downloaded from http://science.sciencemag.org/ on March 26, 2021
Fig. 1. Study cohort and design. (A) A collage of representative wild animal species profiled in the
present study. Animals were photographed in their natural habitats by the collection teams. Columns
cluster species based on sampling regions: (from left to right) Australia (Queensland), Falkland
Islands, Madagascar, Uganda, Israel. Animal names are color-coded based on class (see B for color
legend). (B) A pruned NCBI taxonomy tree of all species included in the cohort. Arc colors denote
animal class, dots outside arc denote sampling region. Branches represent taxonomic relations and
are not scaled to evolutionary distance. (C) World map showing sampling regions with sample size
from each region. (D) Study design.

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 13
 Fig. 2. Identification of unknown bacterial genomes
in the microbiota of wild animals. (A) Adding
genomes from our metagenomic constructed
bacterial genome database of wild animals
significantly improves the mapping rate of the
samples. A reference database was used [41,904
known bacterial genomes, collected from various
large-scale human microbiome studies and NCBI,
representing all known bacterial genomes (X-axis),
and using an expanded genome database that also
includes the 1,209 genomes that we assembled from
our cohort (Y-axis)]. Mapping rate median (across
samples) significantly increased from 7.2% to 21.0%
(inset) (P<4x10−8, Wilcoxon signed-rank test). (B) Wild
genomes introduce microbial members to the tree of
life. For each medium or higher-quality bin
(completeness >50%, contamination <5%), maximal

Downloaded from http://science.sciencemag.org/ on March 26, 2021

ANI score with the reference database is shown, per
animal class. As a control, the same is shown for bins
constructed from 240 human gut microbiome
samples from 240 individuals whose data was not
used to construct the human reference database.
Genomes with maximum ANI lower than 83% were
considered novel. (C to F) Accumulation of genomes
that represent species (by randomly ordering samples
and counting only genomes that represent species
that were not yet counted) per host class (C) and for
wild and captive hosts (E). Accumulation of novel
species per host class (D) and for wild and captive
hosts (F) is shown.

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 14
 Downloaded from http://science.sciencemag.org/ on March 26, 2021

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 15
 Fig. 3. Microbial genomes from wild animals expand the known phylogenetic diversity. (A) A
maximum-likelihood alignment based phylogenetic tree of the 1,209 rSGBs assembled in this study.
Inner color strip, bacterial phylum; Outer color strip, host class. Clades of novel genomes (novel
defined as having ANI lower than 83% compared to the reference database) are colored dark-red.
(B) Distribution of maximal ANI distance to the reference database is shown for the bacterial
genomes constructed from each phylum. Phyla are sorted by percentage of novel genomes within
the phyla. (C) Genomes that were taxonomically annotated as Verrucomicrobia were used to
construct a separate distance-based phylogenetic tree (left). A clade, consisting of only novel
species, was identified and denoted as a novel clade (star, top). Normalized counts of significantly
enriched (or depleted) KEGG pathways in the novel clade relative to the whole phylum are
represented by color intensity (bottom). (D) Genomes that were taxonomically annotated as
Spirochaete were used to construct a separate distance-based phylogenetic tree (left). The
Treponema genus is marked with a star. Normalized counts of significantly enriched (or depleted)
KEGG pathways of the Treponema genus relative to the whole phylum are represented by color
intensity (right).

Downloaded from http://science.sciencemag.org/ on March 26, 2021

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 16
 Downloaded from http://science.sciencemag.org/ on March 26, 2021

Fig. 4. Microbiota composition is associated with host class and exhibits co-existing bacterial clusters.
(A) Pairwise Bray-Curtis dissimilarity between pairs of samples within a host taxonomic rank shows that
bacterial similarity increases with host similarity (*P<0.05; ****P<0.0001, Mann-Whitney U test). (B) A
heatmap representing relative abundance of each rSGB (columns, colored by bacterial phylum) in each sample
(rows, colored by the host class). The heatmap was hierarchically clustered using average linkage based on
Euclidean distance, showing co-existence of groups of rSGBs within particular host groups. (C and D)
Correlation matrix calculated between vectors of abundances of rSGBs across all samples. rSGBs were
hierarchically clustered with average linkage based on Euclidean distance. Four clusters of co-existing bacteria
were visually selected and their abundance in the corresponding hosts is represented in (D). The colors in (D)
share the color key from (B).

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 17
 Downloaded from http://science.sciencemag.org/ on March 26, 2021
Fig. 5. Microbiota composition and richness are associated with host class and phenotypes.
(A) Alpha diversity of the microbiota of all samples from the wild, calculated using Shannon’s
diversity index, showing that diversity is significantly associated with dietary adaptations, while
considering the phylogenetic relationships between the hosts (***P<0.001, phyloANOVA with a
post-hoc test). (B) Differential abundance analysis of the rSGBs between groups of host classes.
Each panel represents a volcano plot of a pair of classes. The difference in average abundance per
animal class is shown on the X-axis and FDR corrected P-value (Mann-Whitney U test) is shown on
the Y-axis. (C) A heatmap representing the effect size for each rSGB (columns) per trait or taxonomy
(rows) for all samples from the wild. For each rSGB and trait/taxonomy pair (a heatmap cell), the
effect size was measured using Mann-Whitney U test. (D) Heatmap representing the effect size for
each microbial phylum (rows) per trait or taxonomy (columns) for all samples from the wild [similar
to (C)]. (E and F) Association between microbial abundance and average body mass (lower row) and
lifespan (upper row) in Mammalia (E) and Aves (F), while accounting for phylogenetic relationships
between the hosts using phylogenetic generalized least squares (PGLS).

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 18
 Downloaded from http://science.sciencemag.org/ on March 26, 2021
Fig. 6. Functional bacterial gene content is associated with host phenotypes. (A) Gene calling and
annotation were performed using eggNOG. Genes that mapped to a known COG category were
counted. The categorized functional annotation of 1,880,067 genes within medium or higher quality
bins (completeness>50%, contamination<5%) predicted in the cohort. Outer ring: COG category.
Inner ring: existence or absence of KEGG pathway annotation. (B to F) PCoA (Euclidean metric) of
sample-level KEGG pathway vectors (features are pathway-associated ortholog counts per sample).
Mammals from the wild colored by diet type (B) (R2=0.35; P<1.3x10−3), activity hours (C) (R2=0.31;
P<1.3x10−3), social structure (D) (R2=0.24; P<8x10−3), body mass (E) (R2=0.36; P<1.3x10−3) and
lifespan (F) (R2=0.41; P<1.3x10−3) (phyloMantel, FDR corrected). (G) A heatmap of log odds ratio of
KEGG orthologs that are enriched in diet groups. Rows represent orthologs, columns are different
diet groups compared. Only orthologs that are enriched (by means of Fisher exact test) with P<0.05
after FDR correction and phylogenetic P<0.05 are shown. (H and I) Existence map of KEGG orthologs
that are associated with ABC transporters (H) and phosphotransferase system (I), both enriched in
carnivores relative to herbivores. Rows represent orthologs, columns represent samples, colored by
dietary adaptations.

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 19
 Downloaded from http://science.sciencemag.org/ on March 26, 2021
Fig. 7. Discovery and experimental validation of unknown toxin-metabolizing genes in the
microbiota of carrion eaters. (A to D) Characterization of protease genes from the gut microbiota
of carrion eaters. Length distribution of the 693 proteases that were initially selected for screening
is shown in (A). Classification of the proteases by type (up to the third hierarchy of EC numbers) is
shown in (B). DIAMOND similarity score distribution (aligned to KEGG protease database) of a
deduplicated list of 531 proteases is shown in (C) and (D). (D) is a magnification of (C), with Y-axis
spanning to 60 instead of 300; X-axis remains unchanged. (E) Structures of 12 bacterial toxins from
which 120 peptides were extracted as targets for the screening, with isolated peptide substrate
positions marked in red. Structures are semi-transparent so peptides from all sides are visible.
Structures were drawn from PDB (a full list of PDB IDs can be found in table S7). (F) Protease
screening results. Protease genes were synthetized and tested on a printed microarray of
biotinylated peptides in order to identify cleavage events. Microarray staining was done post
cleavage by FITC-labeled streptavidin, such that cleaved peptides were not stained. Significant hits
(FDR of 0.01 or less) are marked by black squares. Each row in the heatmap is a specific protease
(enzyme) gene; each column is a specific peptide. (G) Validation of a selected cleavage event, a
protease termed WBT001, identified by the screening. Validation was done in solution, using the
specific peptide synthesized as a FRET substrate, modified with 6-carobyfluorescein (6-FAM) and
4-(dimethylaminoazo)benzene-4-carboxylic acid (DABCYL), and measured by fluorometry.

First release: 25 March 2021 www.sciencemag.org (Page numbers not final at time of first release) 20
Diversity and functional landscapes in the microbiota of animals in the wild
Doron Levin, Neta Raab, Yishay Pinto, Daphna Rothschild, Gal Zanir, Anastasia Godneva, Nadav Mellul, David Futorian, Doran
Gal, Sigal Leviatan, David Zeevi, Ido Bachelet and Eran Segal

published online March 25, 2021

Downloaded from http://science.sciencemag.org/ on March 26, 2021

ARTICLE TOOLS http://science.sciencemag.org/content/early/2021/03/24/science.abb5352

SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2021/03/24/science.abb5352.DC1
MATERIALS

REFERENCES This article cites 121 articles, 15 of which you can access for free
http://science.sciencemag.org/content/early/2021/03/24/science.abb5352#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2021, American Association for the Advancement of Science