快猫短视频

MORE than a century ago, Charles Darwin wrote to his friend Thomas Henry
Huxley: 鈥淭he time will come, I believe, though I shall not live to see it, when
we shall have fairly true genealogical trees of each great kingdom of Nature.鈥�
Darwin鈥檚 prediction was born of his conviction that the secrets of evolution are
to be found in the bodies of all living things. He was confident that the
relationships between species could be established by comparing their anatomical
features.

What Darwin could not have predicted was that a century after the publication
of his On the Origin of Species, biologists would begin to use
molecules such as proteins and DNA to uncover the shape of the tree of life.
This approach has taken the biological stage by storm, and has produced many
surprising results.

Consider some of the announcements made during 1997 alone. The elephant
shrew, consigned by traditional analysis to the order insectivores (which
includes moles and hedgehogs), is in fact more closely related to its behemoth
namesake, the true elephant. Cows are more closely related to dolphins than they
are to horses. The duck-billed platypus, an egg-laying mammal from Australia,
does not represent the most primitive form of mammal after all, but is on an
equal evolutionary footing with those marsupial mammals from Australia,
kangaroos and koalas. And forget examining the shape of seedlings in flowering
plants to seek out their evolutionary history: molecular evidence shows that the
microscopic form of pollen grains gives the best clues.

But who cares about the true relatives of elephant shrews? Let cows take a
dive in the ocean, if that鈥檚 where their roots are. And platypuses are just
cute, no matter who their closest cousin is. Flowers, well, their place is in a
vase, isn鈥檛 it, not at the centre of a debate over the tree of life. Maybe. But,
if these results and countless others like them are correct, it means that what
Darwin had in mind, and what biologists have been doing for more than a century,
was misguided at best and at worst a waste of time.

Has morphological analysis had its day when it comes to understanding the
evolutionary relationships鈥攖he phylogeny鈥攐f life on Earth? Some
proponents of molecular phylogenetics believe it has. Blair Hedges, a biologist
at Pennsylvania State University, puts it this way: 鈥淚鈥檓 not saying that
morphological data are useless鈥攖hey are extremely valuable for many
things, but not for phylogeny. If you want to know about phylogeny, using
molecular data is the only way to go.鈥� However, behind this bravado鈥攁nd
behind the eye-catching headlines鈥攍ies another story.

Understanding the shape of the tree of life and the details of its branches
is more than a quaint sideline of biology, even though the science of this
quest鈥攌nown as systematics鈥攈as come to be regarded by many modern
biologists as dowdy and old fashioned, little more than stamp collecting. But
such an understanding is probably the best foundation for a larger appreciation
of life, including evolution, ecology and behaviour. As Colin Patterson, a
palaeontologist at the Natural History Museum in London, says: 鈥淭o retrieve the
history of life, to reconstruct the evolutionary tree, is still the central aim
of evolutionary biology.鈥� Getting it right is therefore important.

Family resemblance?

Getting it right, however, is much harder than might be imagined. Inferring
an evolutionary relationship from morphology rests on identifying anatomical
features, or characters, that are shared by two species because of their common
descent. Such features might include the shape of teeth, the form of a
particular nerve canal, the number of certain flower parts, and so on.
Ironically, the thing most likely to confound the well-intentioned systematist
in identifying such characters is the power of natural selection itself. Many
shared characteristics do not reflect a common ancestry, but instead are the
result of distantly related species independently adapting their bodies to meet
the demands of similar lifestyles.

This is known as convergent evolution. The eerie resemblance between European
wolves (placental mammals) and Tasmanian wolves (marsupial mammals), is a good
example. 鈥淚dentifying convergent evolution is the key to phylogenetics,鈥� says
Patterson. 鈥淚t is also often extremely difficult.鈥� In fact, so tricky is the
problem that by the middle of this century systematics was in the doldrums.
Traditional methods of analysis (which was not much more than eyeballing
combined with experience) simply weren鈥檛 up to the task.

The 1960s brought hope to depressed systematists in the shape of two new, but
very different, analytical techniques, known as phenetics and cladistics. Both
techniques claimed a degree of scientific objectivity that had been absent in
earlier times, because they involve collecting quantitative data that are
analysed by computer programs, rather than qualitative data assessed
subjectively. Phenetics focused on comparison of overall morphology between
species to create a picture of the consequences of evolutionary change without
regard to evolutionary relationships. Cladistics, on the other hand, explicitly
aimed at tracking the course of evolution. Battles raged between the two
philosophies, each camp claiming that theirs was the True Way. Cladistics won,
partly because it is a more natural system.

One serendipitous outcome of that war was that comparative anatomy (as a
science) was lifted out of the doldrums, and its techniques honed, just in time
to withstand attack from a totally unexpected direction: genetics. In the early
1960s, Linus Pauling and Emile Zuckerkandl, then of the California Institute of
Technology, showed that genetic relationships among species can be inferred by
comparing certain properties of blood proteins. Molecular phylogenetics was
born.

Tree of life

The approach blossomed over the next two decades, with the development of new
techniques including immune reactions with proteins, comparison of protein
sequences, matching the entire DNA complement of two species and, ultimately,
reading the sequence of segments of DNA, or even of whole genomes. Molecular
biologists were soon clambering around in the tree of life, examining its
branches with their new tools and declaring that their approach was inherently
superior to the traditional methods.

With good reason, as participants heard at the Third International Congress
on Systematics and Evolution, held in Brighton in 1985. The conference was
unambiguously titled Molecules versus Morphology, and was convened by
Patterson.

On parade were numerous success stories. Morris Goodman, of Wayne State
University, showed that humans are more closely related to African apes than to
Asian apes鈥攃ontrary to the prevailing classification. Alan Wilson and
Vincent Sarich claimed victory in their epic battle with anthropologists, by
using molecular evidence to show that the human family originated close to 5
million years ago, not the 30 million years that anthropologists had deduced
from the comparative anatomy of fossils.

Even more convincing were the findings of Walter Fitch, then of the
University of Southern California, and William Atchley, then of the University
of Wisconsin. They chose a phylogeny that was well-recorded鈥攖he 70-year
evolutionary history of mice reared in the laboratory and inbred鈥攁nd
successfully reconstructed it by comparing protein variants from 97 gene sites.
By contrast, morphological data from 10 measurements of the lower jaw produced
the wrong answer. 鈥淢olecular data appear to be superior to morphological data,鈥�
said Fitch and Atchley.

There were theoretical reasons supporting this apparent superiority, too.
Mutation of DNA sequences was regarded as quite steady in rate, unlike
morphological evolution which often proceeds erratically. Molecular evolution
tends largely to be isolated from natural selection, so the problem of
convergent evolution was considered minimal. Moreover, the chances of the same
site in the same gene in two distantly related species mutating in the same way
seemed very low. At least as important in giving molecular phylogenetics an edge
over traditional methods was the imminent prospect of a flood of molecular data,
as DNA sequencing slipped into top gear. The consensus among delegates at the
Brighton conference was that if you want to know phylogenetic history, go with
the molecules.

Stephen Jay Gould, a palaeontologist and therefore not a natural ally of
molecular biology, was even more forthright. Impressed by the revolutionary
reclassification of Australian birds by Charles Sibley and Jon Ahlquist, then of
Yale University, Gould wrote: 鈥淭he problem of phylogeny has been solved . . . I
do not fully understand why we are not proclaiming the message from the
rooftops.鈥� The problem of phylogeny, of course, was the snare of convergent
evolution. Molecular analysis had helped Sibley and Ahlquist to show that
despite anatomical similarities with European species, Australian birds evolved
locally. Gould鈥檚 exuberance in support of the molecular approach provoked
indignation among his fellow comparative anatomists, who saw their profession
being shunted into obscurity.

So which data are indeed superior, molecular or morphological? 鈥淚f you鈥檇
asked me that question back in 1985, I would have said I have no idea,鈥� says
Patterson. Despite organising the Brighton conference, he was one of the few who
had not been bowled over by the apparent superiority of molecular phylogenetics.
But the approach generated even more successes in the late 1980s. 鈥淏y 1990, I
would have had no hesitation鈥攊f you can get the money, get the sequencing
done; that鈥檚 the way to go,鈥� says Patterson. 鈥淣ow I鈥檇 say that if the organisms
you are working with have a good set of characters, stick with the
morphology鈥攜ou get into much deeper trouble with molecular data.鈥�

Patterson鈥檚 shifting opinion reflects an increasing understanding of just how
complex evolutionary change at the molecular level really is. Molecular
biologists once thought of genes in higher organisms as strings of beads that
mutated in a straightforward, random way. As research accumulated, it became
apparent that this was far too simplistic. For instance, biologists discovered
that genes are split into pieces鈥攅xons, the protein-coding sections of DNA
are interrupted by introns, non-coding sections鈥攚hich can be shuffled
during evolution to form new genes.

Genes may also exist as mini-families made up of multiple repeats of a single
ancestral gene. Members of these families sometimes evolve in unison, sometimes
diverge from each other, and occasionally become inactivated. The rate of
mutation of a gene may vary through time, and different segments of a gene
mutate at different rates, too.

The overall effect is that molecular phylogenetics is by no means as
straightforward as its pioneers believed. For a start, convergent evolution is
common. In 1995, Michael Donoghue and Michael Sanderson of Harvard University
combed through data from 42 studies of morphological phylogenetics and 18
studies of molecular phylogenetics. They found that the levels of convergent
evolution were similar. One reason is because some segments of a gene are highly
susceptible to mutation, so the chances of identical mutations in two separate
species is much higher than if mutations were equally distributed over the whole
gene.

The Byzantine dynamics of genome change has many other consequences for
molecular phylogenetics, including the fact that different genes tell different
stories. Even the tale of Fitch and Atchley鈥檚 inbred mice, which did so much to
convince observers of the power of molecular phylogenetics back in 1985, is now
seen as more complicated and perplexing. The two researchers have now analysed
data from more than 200 genetic sites in 24 strains of mice. Some of these sites
are from general protein-coding genes, and here the molecular phylogeny remains
consistent with the known phylogeny. Others are in immune system genes and genes
from viruses innate to the mice. Neither of these gives the correct evolutionary
history. 鈥淭his tells us that not all genetic data have equivalent information
content for phylogenetic reconstruction,鈥� warns Atchley. 鈥淭here鈥檚 a lot we still
don鈥檛 know.鈥�

One final practical problem for molecular phylogenetics is coping with
so-called 鈥渕ultiple hits鈥�, which is what happens when a single site, having
mutated once, mutates again. As time passes, the probability of multiple hits
increases. Molecular phylogeneticists have come up with some ingenious
statistical tricks to take account of the problem, but it has by no means
disappeared.

Patterson鈥檚 pessimism for the future of molecular phylogeny is born out in
practice as well as theory. 鈥淟ook at the Paris meeting,鈥� he says, referring to a
conference titled Molecules and Morphology, held last March. 鈥淚t was a replay of
the Brighton meeting, but there weren鈥檛 many surprises, and there was Gavin
Naylor鈥檚 presentation on how to be deceived.鈥�

Until recently, Naylor was working with Wesley Brown, a pioneer of molecular
phylogeny at the University of Michigan, Ann Arbor. The lab had sequenced the
entire protein-coding region (some 12 234 nucleotides) of the mitochondrial DNA
of amphioxus. This small marine creature, often called the lancelet because of
its blade-like shape, is generally assumed to be descended from a sister group
to the early vertebrates. Naylor decided to compare this mitochondrial sequence
with the equivalent sequence from 18 other species from across the animal
kingdom to produce a phylogenetic tree. 鈥淲e came up with an unequivocal answer,鈥�
recalls Naylor. 鈥淭he wrong one.鈥�

Naylor and Brown鈥檚 analysis showed that sea
urchins, not amphioxus, are the sister group of vertebrates. 鈥淭hat鈥檚
ridiculous,鈥� says Naylor. So where had the mistake been made? To pinpoint the
source of error, Naylor decided to turn the usual way of working upside down and
discover which genetic sites gave the 鈥渞ight鈥� answer and which did not. He found
that while many gave the wrong answer, some DNA sequences, such as those for
amino acids involved in maintaining the three-dimensional structure of protein,
did reflect the known phylogeny. 鈥淭hese are obviously important to the function
of the protein, and will therefore be conserved over long periods of time,鈥� says
Naylor.

Digging for roots

The phylogenetic tree on which Naylor was working has roots back to the
Cambrian explosion, some 510 million years ago. But the very conservative,
slow-changing DNA sequences that can resolve such a deeply rooted tree will
yield little or no information about recent events, such as the evolution of Old
World monkeys during the past 50 million years.

Similarly, fast-changing sequences that can capture recent events will be
swamped by multiple hits over 500 million years. But over short time periods,
where morphological change is likely to be minimal, fast-changing molecular
sequences may be the only reliable source of phylogenetic information. 鈥淭he
bottom line message is that context is everything, both in terms of timescale
and biological function,鈥� says Naylor.

Morphological analysis comes into its own for resolving short-term bursts of
evolution that happened long ago. Here molecular phylogenetics faces a quandary
because sequences that change fast enough to capture such events will be swamped
by multiple hits in the subsequent millions of years. And sequences that change
slowly enough to avoid being swamped will be too slow to record ancient but
rapid changes such as the one that seems to have given rise to all 18 modern
mammalian orders in a relatively short period around the time of the demise of
the dinosaurs, 65 million years ago. Morphological data are the best bet here,
limited though they might be.

Both molecules and morphology have their strengths. Which technique is best
may not be the right question to ask. 鈥淭hey are both important, but they answer
different questions,鈥� says Robert Foley, a biological anthropologist at
Cambridge University. He argues that genetic analysis is best for assessing the
relationships between living organisms. 鈥淚f you want to know about an organism鈥檚
evolution, its behavioural ecology, natural selection and adaptation, then
molecules won鈥檛 help much,鈥� he adds. 鈥淵ou need morphological data to complete a
full picture.鈥� It is this fuller picture that increasingly concerns evolutionary
biologists.

Phylogenies based on molecular analysis may be grabbing the headlines and
creating the impression that they are constantly overturning the status quo. In
fact, for the most part the answers produced by molecular techniques are the
same as phylogenies inferred from morphology. But as Foley points out: 鈥淵ou
don鈥檛 get a paper in Nature by saying, `Hey, my molecular analysis
confirms what these morphological types have been saying all along鈥�.鈥� In fact, a
recent study by Patterson and two colleagues shows that new studies using
molecules are no less likely to be at odds with each other than they are to
challenge conclusions based on morphology.

Still, analysing molecules seems more like 鈥渞eal science鈥� than measuring
bumps on bones, which may go some way to explain why biologists including Foley
and Brown remain confident that molecules hold the best route to phylogeny.
鈥淒eep in my soul, I believe looking at genes is better than looking at form,鈥�
admits Brown. Then, with a modern echo of Darwin鈥檚 sentiment of a century ago,
he adds: 鈥淚t鈥檚 a lot more complicated than I thought it was when I started this
trip thirty years ago. I used to think that we鈥檇 get the outline of the tree of
life worked out in my lifetime. But I don鈥檛 think so any more.鈥�