\documentclass[11pt]{article}

\usepackage[margin=1in]{geometry}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{tikz}
\usepackage{pgfplots}
\pgfplotsset{compat=1.18}
\usepackage[colorlinks=true, linkcolor=black, citecolor=black, urlcolor=blue]{hyperref}

% --- skin palette: calculus teal, old-ivory amber, well-provenance rust -----
\definecolor{dpteal}{HTML}{2F6B63}
\definecolor{dptealdeep}{HTML}{1F4A44}
\definecolor{dpamber}{HTML}{93672B}
\definecolor{dprust}{HTML}{8A3B2B}
\definecolor{fone}{HTML}{2A6F8F}
\definecolor{ftwo}{HTML}{B5651D}
\definecolor{fthree}{HTML}{3F8F6F}
\definecolor{ffour}{HTML}{7C4FA0}
\definecolor{ffive}{HTML}{A68B1F}
\definecolor{fsix}{HTML}{B1495F}

\title{The Denisovans Finally Have a Face}
\author{The Science Journaling Club}
\date{Field Notes --- Paper Analysis --- Paleogenomics}

\begin{document}
\maketitle

\begin{abstract}
For 92 years the most complete archaic human skull ever found in Asia had no
scientific name, because a laborer in occupied Manchuria hid it down a well in
1933 rather than hand it to the wrong army, and his family kept the secret for
three generations. By the time it reached a laboratory in 2018 and was
described in 2021 as a new species, \textit{Homo longi}, nobody could say what
it actually \emph{was} in genetic terms, because its bone held no usable DNA.
In July 2025, Qiaomei Fu's team found the answer somewhere stranger: the
hardened dental plaque on its teeth. Fu, Cao, Dai, Bennett, Feng, Yang, Ping,
P\"a\"abo \& Ji reported mitochondrial DNA and roughly 95 endogenous proteins
recovered from that calculus --- the first time host DNA has ever been pulled
from dental plaque on a Pleistocene human fossil --- and both lines of evidence
say the same thing: this is a Denisovan \cite{fu2025}. The mtDNA carries 27
diagnostic positions matching all seven previously sequenced Denisovan genomes
and none of them modern human, which is exactly backwards from what museum
contamination would produce. It is also the first time anyone has had a face to
put on this population, previously known only from a finger bone, a few teeth,
and two isolated jaws. We rebuild why dental calculus, of all things, turns out
to protect biomolecules that bone loses, and we do our own arithmetic --- not
the paper's --- on how deep the Denisovan mitochondrial branch really runs:
about 1,000,000 years back to its common ancestor with ours, roughly
$1.5\times$ as old as the 660,000-year split between Neanderthal and modern
human mtDNA, which is why 27 positions out of roughly 16,569 were all it took.
\end{abstract}

%============================================================================
\section{Eighty-Five Years in a Well}

Start with the part that has nothing to do with genetics.

In 1933, during the construction of a bridge over the Songhua River in
Japanese-occupied Harbin, a labourer found a nearly complete human skull in the
riverbank. The Peking Man discoveries of the previous few years had made
fossil skulls newsworthy even to people with no training in anthropology, and
rather than hand his find to the Manchukuo authorities, he wrapped it and
lowered it down an abandoned well. Then came 1945 and the Soviet invasion of
Manchuria, and the man still could not come forward, because doing so meant
explaining how he had spent the war years under Japanese occupation. So he
told no one outside his family, and kept the secret until near the end of his
life, when he finally told his grandchildren where to look. In 2018 the third
generation of that family retrieved the skull from the well and, persuaded by
the palaeoanthropologist Ji Qiang, donated it to Hebei GEO University.

\begin{table}[htbp]
\centering
\small
\begin{tabular}{@{}p{2.1cm}p{11.5cm}@{}}
\toprule
\textbf{Year} & \textbf{Event} \\
\midrule
1933 & Found in riverbank sediment during bridge construction in
  Japanese-occupied Harbin; hidden in a disused well by the finder, to keep it
  from the occupying authorities. \\
1933--2018 & Eighty-five years underground and undocumented. No institution,
  no catalogue number, no scientific record of any kind. \\
2018 & Retrieved by the finder's grandson on his deathbed instructions;
  donated to Hebei GEO University at Ji Qiang's urging. \\
2021 & Described in three simultaneous papers as the holotype of a new
  species, \textit{Homo longi} (``Dragon Man''); direct dating places it
  between roughly 146,000 and 309,000 years old \cite{ji2021species,
  ni2021lineage, shao2021dating}. No usable DNA recoverable from the bone
  itself. \\
2025 & Mitochondrial DNA and about 95 endogenous proteins recovered from
  hardened dental calculus on the molars identify the individual as Denisovan
  \cite{fu2025}. \\
\bottomrule
\end{tabular}
\caption{Chain of custody, reconstructed by the club from the 2021 description
papers and the 2025 genetic paper. Every date is one the original sources
report.}
\end{table}

A specimen with no accession history sits under more suspicion than one with a
clean find-spot, and every methods section in the 2025 paper is partly a
response to that suspicion. The well is not colour; it is the reason the
contamination argument in Section~\ref{sec:tells} had to be so careful.

%============================================================================
\section{A Skull Everyone Could See but No One Could Test}

Once the skull surfaced, its anatomy alone made a strong claim. The braincase
is long and low, the brow ridge heavy, the face unusually wide, and the
cranial capacity large --- comparable to modern humans and larger than most
Neanderthals. Two of the 2021 papers used that anatomy to argue \textit{Homo
longi} might sit closer to us on the family tree than Neanderthals do, a
striking claim to make from bone shape alone, and it drew real pushback from
other palaeoanthropologists who read the same skull differently
\cite{ji2021species, ni2021lineage}.

What nobody could do in 2021 was settle the argument with genetics, because
the bone had none to offer. Ancient DNA needs specific chemical conditions to
survive across a hundred thousand years, and this particular hundred thousand
years included a stretch of unmonitored handling nobody can fully
characterise. A dating study nailed the age bracket using material associated
with the skull rather than the skull itself \cite{shao2021dating}, but a
species-level identity resting entirely on cranial shape is exactly the kind
of claim molecular evidence either confirms or quietly dismantles. For four
years it stood on shape alone.

%============================================================================
\section{The Strange Place They Looked Instead}

Dental calculus is what a dentist scrapes off your teeth: hardened,
mineralised plaque. Plaque itself is a living biofilm, a community of oral
bacteria embedded in a matrix of their own making, feeding on whatever passes
through the mouth. Left alone, minerals dissolved in saliva --- mostly calcium
and phosphate --- slowly precipitate into that biofilm and turn a soft film
into rock-hard scale, layer by layer, a little more added every few days a
person goes without cleaning their teeth. A person accumulates this their
whole life, and it fossilises while they are still alive.

That is the strange part. Calculus is not a preservation environment bone
falls into after death; it is a mineral that forms around living tissue on
purpose, trapping whatever the mouth's bacteria, food debris, and the person's
own shed cells and proteins happened to be doing at the time. A study of
ancient calculus from human skeletal remains found it holds pathogen DNA,
dietary microparticles, and host immune proteins well enough to reconstruct
disease history and oral microbiome composition across archaeological
populations \cite{warinner2014}. Weyrich and colleagues went further and
pulled a working oral microbiome, dietary evidence, and clues to specific
health complaints out of calculus from two Neanderthals in Spain and Belgium
--- the same category of source material, on a close evolutionary relative, at
comparable antiquity \cite{weyrich2017}.

Bone, by contrast, is porous and alive right up until death, after which
groundwater, soil microbes, and enzymatic breakdown all get to work on it for
as long as it sits buried. Dental calculus is already a sealed mineral block
by the time its owner dies. It was never as exposed to begin with.

%============================================================================
\section{Bone Said Nothing. The Plaque Said Everything.}

The 2025 paper's cleanest result is a negative and a positive from the same
skull. Standard ancient-DNA extraction from the Harbin bone produced nothing
usable. The team then sampled the calculus still adhering to the teeth and got
both an authenticatable mitochondrial genome and a set of ancient proteins ---
described as the first recovery of host DNA from dental calculus in any
Pleistocene human fossil \cite{fu2025}. Nobody had shown before that calculus
could outperform the skeleton sitting directly underneath it.

\begin{figure}[htbp]
\centering
\begin{tikzpicture}[line width=1pt, font=\small]
  \draw (0,0) rectangle (5.4,3.4);
  \node[anchor=north] at (2.7,3.3) {\textbf{BONE}};
  \draw (0.3,2.75) -- (5.1,2.75);
  \node[anchor=west] at (0.3,2.1) {mtDNA:};
  \node[fsix, anchor=west] at (2.3,2.1) {$\times$\none usable};
  \node[anchor=west] at (0.3,1.2) {proteins:};
  \node[fsix, anchor=west] at (2.3,1.2) {$\times$\none reported};
  \node[anchor=south, font=\scriptsize] at (2.7,0.15) {standard ancient-DNA extraction};

  \begin{scope}[xshift=6.2cm]
  \draw[dpteal, line width=1.4pt] (0,0) rectangle (5.4,3.4);
  \node[dpteal, anchor=north] at (2.7,3.3) {\textbf{DENTAL CALCULUS}};
  \draw[dpteal] (0.3,2.75) -- (5.1,2.75);
  \node[anchor=west] at (0.3,2.1) {mtDNA:};
  \node[fthree, anchor=west] at (2.3,2.1) {\checkmark\ Denisovan match};
  \node[anchor=west] at (0.3,1.2) {proteins:};
  \node[fthree, anchor=west] at (2.3,1.2) {\checkmark\ $\sim$95 identified};
  \node[anchor=south, font=\scriptsize] at (2.7,0.15) {same protocol, same skull};
  \end{scope}
\end{tikzpicture}
\caption{The two outcomes reported for the same individual \cite{fu2025}. The
skeleton, out of controlled storage for most of a century, gave the team
nothing to authenticate. The mineral scale on its teeth, sealed since long
before the well, gave them a mitochondrial genome and dozens of proteins.}
\end{figure}

%============================================================================
\section{A Third, Much Older Branch}
\label{sec:branch}

Mitochondrial DNA is the easy entry point into any ancient-genetics problem:
hundreds of copies per cell instead of two, so it survives in samples where
the nuclear genome is long gone, and it changes at a fairly steady rate, so a
raw count of differences from a reference sequence converts into a rough date.
What it does not give you is the whole ancestry of a person --- only the
single unbroken maternal line back through their mother, and her mother, and
so on. Still, for a first identification, it is usually enough to place a
sample on the right branch of the tree.

The reason it worked so cleanly here goes back fifteen years. In 2010, a team
that included Qiaomei Fu herself --- the same scientist who led the 2025
Harbin study --- sequenced the complete mitochondrial genome from a finger
bone in Denisova Cave, in the Altai Mountains of southern Siberia, and found
something nobody expected: an mtDNA lineage sharing a common ancestor with
both modern human and Neanderthal mtDNA roughly 1,000,000 years ago
\cite{krause2010}. That is a strikingly deep split. Two years earlier, a
separate team had sequenced a complete Neanderthal mitochondrial genome and
dated its own divergence from the modern human line to about $660{,}000 \pm
140{,}000$ years ago \cite{green2008}. Denisovans, in other words, were not a
variant of anything already known. Their mitochondrial lineage was an outlier
branch old enough to be unmistakable the moment any of it turned up.

\paragraph{Club arithmetic, not the paper's.}
$1{,}000{,}000 \div 660{,}000 \approx 1.5\times$. The Denisovan mtDNA branch
point is about half again as old as the Neanderthal one. That gap is exactly
what makes mtDNA such a blunt but reliable tool for a first call: you are not
hunting for a subtle statistical signal, you are asking whether a sequence
sits on a branch that split off a million years ago instead of 660,000, and
the two are not easily confused.

\begin{figure}[htbp]
\centering
\begin{tikzpicture}[line width=1.1pt, font=\footnotesize,
    x=6.2cm, y=1cm]
  % time axis: x in units of 1 = 1,000,000 years, 0 = present at x=1 (right)
  \draw[->] (1.0,-0.3) -- (-0.05,-0.3) node[left, font=\scriptsize]
    {years before present};
  \foreach \t/\lab in {1.0/0, 0.8/200k, 0.6/400k, 0.4/600k, 0.2/800k, 0.0/1.0M} {
    \draw (\t,-0.35) -- (\t,-0.25);
    \node[below, font=\scriptsize] at (\t,-0.35) {\lab};
  }
  % ancestor lineage
  \draw[ffive] (0.0,1.4) -- (0.34,1.4);
  \draw (0.34,1.0) -- (0.34,1.8);
  \draw[ffive] (0.34,1.8) -- (1.0,1.8) node[right] {modern human mtDNA};
  \draw[ffour] (0.34,1.0) -- (0.96,1.0) node[right] {Neanderthal mtDNA};
  \node[fill=white, inner sep=1pt] at (0.34,-0.05) {\scriptsize split $\sim$660,000 ya};
  \node[fill=white, inner sep=1pt] at (0.02,1.55) {\scriptsize split $\sim$1.0M ya};
  \draw (0.0,1.4) -- (0.0,0.5);
  % Denisovan lineage, with Harbin age band
  \draw[fthree, line width=1.4pt] (0.0,0.5) -- (1.0,0.5) node[right] {Denisovan mtDNA};
  \fill[fsix, opacity=0.3] (0.691,0.42) rectangle (0.854,0.58);
  \node[fsix, font=\scriptsize] at (0.77,0.25) {Harbin skull};
  \node[fsix, font=\scriptsize] at (0.77,0.1) {146,000--309,000 ya};
\end{tikzpicture}
\caption{Mitochondrial branch points from the primary sources
\cite{krause2010,green2008}, with the Harbin skull's independently dated age
bracket \cite{shao2021dating} marked on the Denisovan branch. The individual
lived a few hundred thousand years after its lineage split off, and long
before anyone would try to identify it from its teeth.}
\end{figure}

%============================================================================
\section{Twenty-Seven Tells}
\label{sec:tells}

Knowing which branch a sequence belongs to is one thing. Proving it is not
contamination is another, and this is where the Harbin result earns its
confidence rather than assuming it. The calculus mtDNA fell within the known
range of Denisovan variation and grouped with a branch carried by early
Denisovan individuals from southern Siberia \cite{fu2025}. Specifically, the
sequence carried 27 positions that match all seven previously sequenced
Denisovan mitochondrial genomes and do not match modern human mtDNA at all.

That direction of mismatch matters enormously. If the calculus sample had
picked up modern contamination during a century of uncontrolled handling --- a
real risk for any specimen with this provenance --- the contaminating
sequences would look like ordinary modern human mtDNA. They could not
manufacture 27 positions that happen to match seven Siberian fossils nobody
involved in the discovery had ever touched. The direction of error the authors
were most worried about is the one direction the data cannot produce.

\paragraph{Club arithmetic, not the paper's.} A modern human mitochondrial
genome runs about 16,569 base pairs. $27 \div 16{,}569 \approx 0.16\%$. Well
under a fifth of one percent of the sequence was needed to make the call, once
you already know which branch of a million-year-old split you are checking
against.

\begin{figure}[htbp]
\centering
\begin{tikzpicture}[x=0.19cm, y=0.19cm, font=\scriptsize]
  \foreach \i in {0,...,26} {
    \fill[fthree] (\i*1.3,10) rectangle +(1,1);
  }
  \node[anchor=east, font=\small] at (-1,10.5) {7 known Denisovans};
  \node[fthree, anchor=west, font=\small] at (36.5,10.5) {27 / 27};

  \foreach \i in {0,...,26} {
    \fill[fthree] (\i*1.3,6) rectangle +(1,1);
  }
  \node[anchor=east, font=\small] at (-1,6.5) {Harbin calculus mtDNA};
  \node[fthree, anchor=west, font=\small] at (36.5,6.5) {27 / 27};

  \foreach \i in {0,...,26} {
    \draw[ffive] (\i*1.3,2) rectangle +(1,1);
  }
  \node[anchor=east, font=\small] at (-1,2.5) {modern human reference};
  \node[ffive, anchor=west, font=\small] at (36.5,2.5) {0 / 27};
\end{tikzpicture}
\caption{The diagnostic logic in one picture, following the paper's own
description of the comparison \cite{fu2025}. Every previously sequenced
Denisovan genome agrees at these 27 positions by definition --- that is how
the set was chosen. The Harbin calculus sequence matches all 27. A modern
human reference matches none of them, which is the pattern contamination
cannot fake.}
\end{figure}

%============================================================================
\section{The Proteins Told the Same Story}

DNA is not the only molecule that survives in old teeth. Proteins are
chemically tougher than DNA and can persist far longer, which is why
palaeoproteomics has become its own branch of ancient-biomolecule work over
the last decade, and it is exactly the method that identified two other
Denisovan specimens before Harbin even entered the genetic record: a mandible
from Baishiya Karst Cave on the Tibetan Plateau, dated to more than 160,000
years old, whose collagen sequence variants matched the Denisovan lineage when
no DNA could be recovered from the bone at all \cite{chen2019}, and a mandible
dredged from the seabed of the Taiwan Strait, identified the same way in 2025
\cite{tsutaya2025}.

The Harbin team ran the same style of analysis on the calculus and pulled out
roughly 95 endogenous proteins --- proteins belonging to the individual, not
to the oral bacteria living in the plaque around them --- and those proteins
independently pointed to the same classification as the mtDNA \cite{fu2025}.
Two different molecule classes, extracted from the same small sample by two
different methods, agreeing with each other, is a considerably stronger
position than either result standing alone.

%============================================================================
\section{A Population With Fragments Instead of a Portrait}

Before 2025, everything anyone knew about Denisovans came from pieces small
enough to fit in an envelope. The type specimen is a finger bone from Denisova
Cave, sequenced in 2010 \cite{krause2010, reich2010}. The rest of the direct
evidence is scattered across a handful of teeth from the same cave, a
fragment of arm or leg bone belonging to a Neanderthal-Denisovan hybrid
nicknamed Denny, a mandible from the Tibetan Plateau identified by protein
sequencing because no DNA remained in it \cite{chen2019}, sediment at that
same Tibetan cave carrying Denisovan mitochondrial DNA with no attached
skeleton at all \cite{zhang2020}, a single child's molar from a cave in Laos
whose protein signal was suggestive but not conclusive \cite{demeter2022}, and
the Taiwan Strait mandible, identified in April 2025, three months before the
Harbin result \cite{tsutaya2025}.

\begin{table}[htbp]
\centering
\small
\begin{tabular}{@{}lllll@{}}
\toprule
\textbf{Site} & \textbf{Year} & \textbf{Body part} & \textbf{Evidence} & \textbf{Ref.} \\
\midrule
Denisova Cave, Siberia & 2010 & finger bone, teeth & DNA & \cite{krause2010,reich2010} \\
Baishiya Cave, Tibet & 2019 & mandible & protein & \cite{chen2019} \\
Tam Ngu Hao 2, Laos & 2022 & molar & protein (tentative) & \cite{demeter2022} \\
Penghu, Taiwan Strait & 2025 & mandible & protein & \cite{tsutaya2025} \\
\textbf{Harbin, China} & \textbf{2025} & \textbf{complete cranium} & \textbf{DNA + protein} & \cite{fu2025} \\
\bottomrule
\end{tabular}
\caption{Every Denisovan specimen identified by 2025. Four of five results are
isolated teeth, jaws, or bone chips. Harbin is the only complete skull, and
the only specimen confirmed by both DNA and protein evidence at once.}
\end{table}

That scatter of fragments is not a filing accident. It is the whole reason
Denisovans have spent fifteen years as a genetically well-characterised
population with no agreed face. Geneticists could describe their relationship
to Neanderthals and to us in exacting statistical detail while
palaeoanthropologists had almost nothing to hold up next to a Neanderthal or
\textit{Homo erectus} skull for comparison. Harbin changes that asymmetry for
the first time: a nearly complete cranium, already fully described
anatomically in 2021, now has a secure genetic identity attached to it.

%============================================================================
\section{How Far They Got}

The 2025 paper's own framing is that the Harbin mtDNA relates to a branch
carried by early Denisovan individuals in southern Siberia, and that this
establishes Denisovans inhabiting a large geographical range across Middle
Pleistocene Asia \cite{fu2025}. Put the site names on a map and the range is
not subtle. Denisova Cave sits in the Altai Mountains of southern Siberia.
Baishiya Karst Cave sits on the Tibetan Plateau, thousands of metres above sea
level. Harbin sits in the temperate lowlands of Heilongjiang, in China's far
northeast. Add the Taiwan Strait and Laos specimens, both identified only by
protein because their DNA did not survive, and the known Denisovan footprint
runs from Siberian steppe to tropical Southeast Asia to a Chinese river valley
--- the kind of ecological range that, among living hominins, only our own
species matches.

%============================================================================
\section{What Mitochondrial DNA Cannot Tell You}

None of this is a nuclear genome, and it is worth saying plainly what that
limits. The Harbin identification rests on mitochondrial DNA, which records
only the maternal line, plus a set of proteins that support the same lineage
but do not carry the population-genetic resolution a full nuclear genome
would. That means the paper cannot yet say how Harbin relates to Denisova 11,
the individual with one Neanderthal parent and one Denisovan parent, or to the
deeper population structure geneticists have inferred from nuclear data at
Denisova Cave \cite{reich2010}. A single mitochondrial sequence tells you
which branch someone's mother's mother's line sits on. It does not tell you
the whole of who they were.

It is also, on its own terms, about as strong a case as mitochondrial evidence
gets. The 27 diagnostic positions were chosen precisely because they are
shared across every previously sequenced Denisovan and absent from modern
humans, and the independent protein result removes the single-method risk
that any one ancient-DNA claim carries. The honest description is: a secure
mitochondrial and proteomic identification, on a specimen that badly needed
one, that leaves the nuclear-genome question for whoever tries next.

%============================================================================
\section{Notes From the Club Table}

We spent more of one meeting than we expected on a single question: does the
well story matter to the science, or is it just a good story attached to good
science? We did not fully agree. The position that won out, barely, is that
provenance and molecular authentication are not separate topics here --- a
specimen that spent 85 years outside any institution is exactly the kind of
specimen where a modern-contamination false positive is most worth worrying
about, and the 27-position argument in Section~\ref{sec:tells} is precisely
the kind of evidence built to survive that worry.

A second disagreement, still open: how much weight a mitochondrial-only
identification should carry against a full nuclear genome, in a field that
increasingly treats nuclear data as the gold standard. Our answer, for now, is
that 27 independently corroborating positions plus 95 independently
corroborating proteins is a great deal more than a typical single-locus claim,
even if it is less than everyone eventually wants.

%============================================================================
\section{What We Would Tell Someone Who Read Only the Headline}

Four things. The skull did not sit in a museum drawer for eighty-five years;
it sat down a well, hidden from an occupying army, and nobody outside one
family knew it existed. The species question from 2021, \textit{Homo longi}
versus something else, is now a lineage question with a molecular answer:
Denisovan. The breakthrough sample was not bone, which had already failed,
but the hardened plaque on the teeth, which is exactly the kind of source the
field is now going to look at harder on every other archaic specimen with
disappointing bone DNA. And the identification itself rests on a small,
specific, well-chosen piece of evidence --- 27 positions out of more than
sixteen thousand --- not a sweeping genomic pronouncement. Small, sharp
evidence, aimed at the one place it could not be faked, is what actually
closed a fifteen-year gap between a genetically well-known population and a
face nobody had for it.

%============================================================================
\begin{thebibliography}{99}

\bibitem{fu2025}
Fu, Q., Cao, P., Dai, Q., Bennett, E.~A., Feng, X., Yang, M.~A., Ping, W.,
P\"a\"abo, S. \& Ji, Q. (2025). Denisovan mitochondrial DNA from dental
calculus of the $>$146,000-year-old Harbin cranium. \textit{Cell} 188(15),
3919--3926.e9. doi:10.1016/j.cell.2025.05.040

\bibitem{ji2021species}
Ji, Q., Wu, W., Ji, Y., Li, Q. \& Ni, X. (2021). Late Middle Pleistocene
Harbin cranium represents a new \textit{Homo} species. \textit{The
Innovation} 2(3), 100132. doi:10.1016/j.xinn.2021.100132

\bibitem{ni2021lineage}
Ni, X., Ji, Q., Wu, W., Shao, Q., Ji, Y., Zhang, C., Liang, L., Ge, J., Guo,
Z., Li, J., Li, Q., Gr\"un, R. \& Stringer, C. (2021). Massive cranium from
Harbin in northeastern China establishes a new Middle Pleistocene human
lineage. \textit{The Innovation} 2(3), 100130. doi:10.1016/j.xinn.2021.100130

\bibitem{shao2021dating}
Shao, Q., Ge, J., Ji, Q., Li, J., Wu, W., Ji, Y., Zhan, T., Zhang, C., Li, Q.,
Gr\"un, R., Stringer, C. \& Ni, X. (2021). Geochemical provenancing and direct
dating of the Harbin archaic human cranium. \textit{The Innovation} 2(3),
100131. doi:10.1016/j.xinn.2021.100131

\bibitem{krause2010}
Krause, J., Fu, Q., Good, J.~M., Viola, B., Shunkov, M.~V., Derevianko, A.~P.
\& P\"a\"abo, S. (2010). The complete mitochondrial DNA genome of an unknown
hominin from southern Siberia. \textit{Nature} 464(7290), 894--897.
doi:10.1038/nature08976

\bibitem{green2008}
Green, R.~E., Malaspinas, A.-S., Krause, J., Briggs, A.~W., Johnson, P.~L.~F.,
Uhler, C., Meyer, M., Good, J.~M., Maricic, T., Stenzel, U., Pr\"ufer, K.,
Siebauer, M., Burbano, H.~A., Ronan, M., Rothberg, J.~M., Egholm, M., et al.\
\& P\"a\"abo, S. (2008). A complete Neandertal mitochondrial genome sequence
determined by high-throughput sequencing. \textit{Cell} 134(3), 416--426.
doi:10.1016/j.cell.2008.06.021

\bibitem{chen2019}
Chen, F., Welker, F., Shen, C.-C., Bailey, S.~E., Bergmann, I., Davis, S.,
Xia, H., Wang, H., Fischer, R., Freidline, S.~E., Yu, T.-L., Skinner, M.~M.,
Stelzer, S., Dong, G., Fu, Q., Dong, G., Wang, J., Zhang, D. \& Hublin, J.-J.
(2019). A late Middle Pleistocene Denisovan mandible from the Tibetan
Plateau. \textit{Nature} 569(7756), 409--412. doi:10.1038/s41586-019-1139-x

\bibitem{warinner2014}
Warinner, C., Rodrigues, J.~F.~M., Vyas, R., Trachsel, C., Shved, N.,
Grossmann, J., Radini, A., Hancock, Y., Tito, R.~Y., Fiddyment, S., Speller,
C., Hendy, J., Charlton, S., Luder, H.~U., Salazar-Garc\'ia, D.~C., et al.\ \&
Cappellini, E. (2014). Pathogens and host immunity in the ancient human oral
cavity. \textit{Nature Genetics} 46(4), 336--344. doi:10.1038/ng.2906

\bibitem{weyrich2017}
Weyrich, L.~S., Duch\^ene, S., Soubrier, J., Arriola, L., Llamas, B., Breen,
J., Morris, A.~G., Alt, K.~W., Caramelli, D., Dresely, V., Farrell, M.,
Farrer, A.~G., Francken, M., Gully, N., Haak, W., Hardy, K., Harvati, K., et
al.\ \& Cooper, A. (2017). Neanderthal behaviour, diet, and disease inferred
from ancient DNA in dental calculus. \textit{Nature} 544(7650), 357--361.
doi:10.1038/nature21674

\bibitem{tsutaya2025}
Tsutaya, T., Sawafuji, R., Taurozzi, A.~J., Fagern\"as, Z., Patramanis, I.,
Troch\'e, G., Mackie, M., Gakuhari, T., Oota, H., Tsai, C.-H., Olsen, J.~V.,
Kaifu, Y., Chang, C.-H., Cappellini, E. \& Welker, F. (2025). A male
Denisovan mandible from Pleistocene Taiwan. \textit{Science} 388(6743),
176--180. doi:10.1126/science.ads3888

\bibitem{reich2010}
Reich, D., Green, R.~E., Kircher, M., Krause, J., Patterson, N., Durand,
E.~Y., Viola, B., Briggs, A.~W., Stenzel, U., Johnson, P.~L.~F., et al.\ \&
P\"a\"abo, S. (2010). Genetic history of an archaic hominin group from
Denisova Cave in Siberia. \textit{Nature} 468(7327), 1053--1060.
doi:10.1038/nature09710

\bibitem{zhang2020}
Zhang, D., Xia, H., Chen, F., Li, B., Slon, V., Cheng, T., Yang, R., Jacobs,
Z., Dai, Q., Massilani, D., et al.\ \& Fu, Q. (2020). Denisovan DNA in Late
Pleistocene sediments from Baishiya Karst Cave on the Tibetan Plateau.
\textit{Science} 370(6516), 584--587. doi:10.1126/science.abb6320

\bibitem{demeter2022}
Demeter, F., Zanolli, C., Westaway, K.~E., Joannes-Boyau, R., Duringer, P.,
Morley, M.~W., Welker, F., R\"uther, P.~L., Skinner, M.~M., McColl, H., et
al. (2022). A Middle Pleistocene Denisovan molar from the Annamite Chain of
northern Laos. \textit{Nature Communications} 13, 2557.
doi:10.1038/s41467-022-29923-z

\end{thebibliography}
\end{document}