Rough draft - Last revision
Reviewers - if any! - are asked to read this disclaimer carefully
By Cesare Brizio
Address: Via Fornace Tanari 900/c- San Benedetto –
40018 San Pietro in Casale BO -
- Introduction
- The
small fossil connection: where are they? I don't know, but the small
animals were there!
- The
easy way: average body mass categories in the vertebrate faunas of our
time
- The
reference signal: populations as raw material and skeletons as refined
products
- The
noise: organic matter degradation and taphonomic factors
- Geologic
Diagenesis
- A
review of the main steps involved in the determination of the actual
signal
- The
actual signal and the fossil record
- What
does a bone stand for?
In
recent years many paleontologists throughout the world have devoted their best
energies to the assessment of the completeness of the fossil record, as well as
to the research of patterns of extinction and speciation based on fossil data.
At present, there seems to be a general consensus about the fact that
preservational, taphonomic, diagenetic and collecting biases - along with
geological factors - have not compromised the readability of fossil data in
this respect, and that statistical analyses about temporal diversity can be
performed on fossil data sets with a reasonable degree of accuracy.
Unfortunately,
some of the methods used in literature - when examined from a more general
point of view - seem error-prone or self-justifying. In the first category fall
all the attempts of validating any speciation/extinction pattern by the
comparison with computer-based simulations using random number generation as a
simulation of natural selection or diversity. There has never been a serious
attempt to illustrate the drawbacks of the random number seed methods and to
elucidate the techniques for compensating the distinctive statistical pattern
intrinsic to any computer-based randomizer, a software device not substantially
better than "heads or tails" in generating random sequences. In other
words, correspondence in observed patterns and patterns based on any level of
computer randomization should be regarded as suspect, because random number
generation - whichever its level in the simulation process - is a strong
pattern generator which tends to obliterate the expression of the mathematical
model being explored. It is not a case that many studies about temporal
diversity have come to similar conclusions.
Since
the concept of "stratigraphic debt" was proposed in literature, many
studies have centered on "stratigraphic congruence" of phylogenies
and on the adequateness of the fossil record in representing the evolutionary
history. With a rough semplification, from a logic point of view, estimating
the quality of fossil record by comparison with phylogenies, themselves based
on that very same fossil record and evaluated according to their congruency
with First Appearance Data (or better Lowest Stratigraphic Data), is some sort
of vicious circle.
I do
not contend the practice of inferring the possibility of finding a given
species in rocks dating to a well defined time interval, but rather I contend
the exhaustiveness of evaluations of the completeness of the fossil records
based on its respondence to our phylogenetic expectations - which unavoidably
are often misplaced.
In
this paper I will concentrate on terrestrial vertebrates, and i will try to
outline a different, more straightforward method to obtain information about
this fascinating subject, a method drawn from ecological considerations based
on our present biosphere, an entity that can reasonably be supposed to
represent mechanisms active also in the past biospheres, and surely more easily
and completely accessible.
My
aim is making an histograms of number of species and number of individuals in
discrete body mass categories representing our global non-marine vertebrate
fauna, and then try to derive from these data an histogram representing the
number of skeletons entering the sediments and thus being available to
fossilization processes (what I call the "signal" our non-marine
vertebrate fauna is sending).
I
propose the concept that the relation between the living animal distributions
and the number of skeletons available should be constant through the geologic
ages, and that what could be considered like a "complete fossil
record", could reproduce, or at least well compare with, our
"signal".
Furthermore,
I would like to elucidate the relations among fossil remains and living faunas,
trying to quantify the small animals deficit in the fossil record, and showing
how even now a great part of our terrestrial or non-marine vertebrate diversity
is obliterated from accessing the depositional environments well before the
fossilization itself could begin.
The small fossil connection: where are they? I don't know, but the small
animals were there!
Small
animals - at least in subaerial environments - fossilize less often than big
ones. Even though this fundamental concept has been common sense in modern
paleontology since its beginning, until now nobody seems to have fully accepted
its implications. To my knowledge, nobody has tried to quantify the small
animal gap in fossil remains, and, as long as there is no equation or
statistical provision to compensate for this lack of small weight vertebrates,
their absence is perfectly known and completely, plainly ignored in wide scope phylogentic
reconstructions.
There
are some groups - namely, the Mammals - quite well documented also in the small
individual body mass range, thanks to the screening for teeth of the
fossiliferous sediments, a practice completely blind to edentulous small
animals, like Birds, who presently constitute the most diverse non-marine
vertebrate group.
This
is just one of the aspects of the multifaceted problem of "sampling
bias": small remains are very often overlooked, because finding them
requires a specific attention, as they don't stick out like big fossils do.
This factor cumulates with disuniform exposition of the fossiliferous sediments
in compromising our possibilities of a completely exhaustive reading of the
fossil record. In other words, even a perfectly complete fossil record would
unavoidably seem spotty when seen through small, sparse windows (the rocky
outcrops being studied) with dirty panes. The factors influencing these
sampling problems are not directly related with the biological factors I would
like to concentrate upon, and become evident during fossil collection. In this
paper I would rather like to concentrate on fossil generation, leaving the task
of overcoming our well known sampling biases to the continuously evolving
paleontological techniques.
A
most fascinating theory about avian evolution, namely the Core Group theory by
George Olshevsky, whose implications seem to be increasingly accepted also by
other paleontologists, centers on the concept of a "core group", a
bundle of strictly intersecting phyletic lineages of small, unspecialized
animals in an arboreal ecological niche, evolving through time an increasing
"birdhood" from early archosaurian (small animals with arboreal
preadaptations) to modern birds.
This
theory overcomes many limitations of the current view of bird as descendants of
a specific dinosaurian lineage: in violation of Dollo's and Cope's Laws the
current view expects the birds' phyletic lineage to descend from the trees,
increase in size, specialize as cursorial predator, undergo proportional
reduction of forelimbs, then again to get smaller, despecialize, redevelope
forelimbs, climb the trees.
The
current view is at odds with the recent findings of very birdlike dinosaurs
belonging in different, not strictly related families, and with the widespread
presence of birdlike characters in dinosaurs, while the Core Group Theory sees
many dinosaurian lineages as byproducts of bird evolution, caught in Cope's
drift towards specialization as bigger and cursorial, terrestrial animals - the
same fate of recent and contemporary Ratites with one main difference: at the
time of the origination of the dinosaurian lineages, the ancestral quasi-avian
forelimb still retained grasping functions.
The
problem is that the fossil remains of the Core Group species are virtually
impossible to find, both for their dimensions and for the niches that they
occupied (let's think of rain forest - presently our main source of vertebrate
diversity - with its immediate degradation of organic matter), so this is not a
bone-based theory.
In an
example of hard-ground reasoning, professionals throughout the world do not
compromise themselves and, rather than trying to take into account the obvious
gap in fossil remains, stick to the fossils they have, even though they
perfectly know that there were much more animals than those that will ever be
found.
Which
is good science? Not compensating this gap at all or trying to compensate?
Let's help the guys, and attempt an evaluation of the status of our present
time non-marine vertebrate biosphere, to come out with some statistical data
they can deal with.
The easy way: average body mass categories in the vertebrate faunas of
our time
One
rather simple way of describing how many "small" animals were there,
is an histogram of non-marine vertebrate species in discrete average adult body
mass categories. How many in the 0-10 g range? How many in the 10_100 g range?
And so on. Even though I devoted some of my time to the research of such a
diagram, I didn't succeed in finding one, and I am thinking that maybe nobody
ever tried to produce it.
Even
more stringent could be a similar representation in terms of number of
individuals in any given discrete average adult body mass category.
This
paper could greatly benefit from those elaborates, if available. Anyway, I
think that the kind of data therein represented could be judged as too
"static" and simplistic. Apart from that, the species and individual
distribution histograms in body mass categories do not directly relate with
bone available for fossilization, even though they dramatically show the
prevalence of small organisms both in number of species and number of
individuals.
The
aim of this paper is to stress the even more dramatic prevalence of
"big" animals in the non-marine vertebrate fossil record,
irrespective of the undisputed numeric dominance of small animals, so it's
interesting to put a stress on the fate of the bones and treat the whole thing
in dynamic, descriptive terms of signal/noise ratio.
The reference signal: populations as raw material and skeletons as
refined products
I
will write about bones, including in this very approximate definition all the
fossilizable skeletal remains, including teeth, tendons, etcetera.
I do
not specifically address the problem linked with the definition of a reference
time lapse. My impression is that age distribution in a supposedly stable
population of organisms can be expected to be constant through time: I am
aiming at a statistic profile of fossilizable skeletons distribution in
different body mass categories at the level of subphylum Vertebrata. This kind
of "fingerprint" shouldn't be strictly time-related, but should
rather be the expression of a general trend that may be supposed as constant
through time. What we should consider as the original, "clean" signal
- in terms of fossil generation - is this distribution of bones available to
diagenetic processes in different dimensional / original body mass categories.
Possible
sources of information are essentially field studies, to be integrated in a
statistical frame. I am almost sure that something interesting can be found in
literature.
Even
though it is well known that teeth and long bones are more common in the fossil
record than, say, flat bones, and that the bones of big animals are much more
easy to find - this being the single reason for this paper, I will equanimously
consider all bones as equal, and give a most general look at this subject.
Among
the innumerable factors influencing this bone availability, the most important
relate with population dynamics, and particularly with ontogenetic and
ecological factors.
At
the individual level, the bone is generated: the degree of ossification
can be expected to grow during ontogeny, with a plateau in the adult life and
some degradation towards the extreme age range. In this respect, any member of
the population can be viewed as a bone producer - a process taking some time.
The individual fate, which obviously influences the survival of skeletal
elements, is best examined statistically at the population level. What we
may expect to obtain at the individual level is a statistic estimate of
"overall individual bone mass".
At
the population and specific level, the bone is distributed in a peculiar
age/mass distribution: attention must be paid at age distribution in
the population, with healthy population comprising an adequate surplus of
young, poorly ossified individuals. Provisions should be made to compensate for
the incomplete ossification of subadults.
In
other words, if we plot this kind of age distribution (number of individuals
for each age), and use age as a direct correlate of body mass (number of
individual per individual body mass range), the histogram describing the
population (either based on absolute quantities - e.g. 1 to 10 g, 10 to 100 g,
100 to 1000 g, or - maybe better - based on percentages of average adult body
mass) will show a peak in the early life stages and another more prominent peak
around the average age for that particular population, fading to zero at the
absolute maximum age.
If we
assume that the population is reasonably stable through time, the maintenance
of that particular age/mass distribution profile is requiring an exactly
similar pattern in the overall mortality by natural causes, that when examined
in terms of "age at death" presumably will also show a peak in the
early life stages and another more prominent peak around the average age for
that particular population, with just a few individuals lucky enough to die at
very old ages. Thus, for the aims of our study, what we may expect to obtain
at the population level is some sort of histogram showing the
"distribution of dead body mass available in any given moment",
subdivided in discrete individual body mass categories. The specific level
is nothing more than the integration of population data in a geographic
framework.
At
the ecosytem level, bone is distributed in peculiar environmental contexts, and
predator/ prey interactions become apparent, that is, intraspecific
differences come into play. The two main factors influencing the future destiny
of the bones that will hopefully be available for fossilization are grossly
size-related and also depend on the position of the species in the food chain.
Factors as the species' average adult body mass and feeding strategies exert
their influence at the ecosystem level. The ecosystem tests each population
(each species) against all the others. Which percentage of the potential
prey is effectively caught and eaten by predators? Which percentage of their
bone mass survives predation? A size-related pattern should become
apparent, with degradation roughly inversely proportional to body mass (small
animals who can be swallowed whole opposed to big herbivores in which just the
flesh is eaten). The overall probability of falling victim to a predator (in
terms of what percentage of deaths is caused by predation) shouldn't
necessarily be size related, even though there are many more predators that can
deal with small animals than those who can engage a big one. What could
hopefully be obtained from the integration of population-level bone
distributions in the ecosystem framework is a statistical representation of the
distribution of dead bone mass surviving after predators have taken their toll
in terms of mechanical and chemical degradation.
As an
alternative approach, the predation-associated bone degradation could be
treated as one of the "noise" factors, but in my opinion being eaten
is just one way of dying, and we should photograph the situation immediately
after the death cause, destructive or not, has completed its action.
So
here is our clean, unspoiled signal: the overall statistical distribution of unscavenged,
undigested bone from dead individuals. But how should we express this
statistical distribution?
I
would suggest an histogram with base 10 logarithm x-axis, and the discrete
average body mass categories yet mentioned above: how many individual skeletons
are available in each body mass range (1-10 g, 10-102 g, 102-103 g, 103-104 g, 104-105 g, more than 105 g)? With
reference to the living animal distribution, I expect the difference between
small and big animals to be less accentuated as small animals are mole likely
to be eaten whole.
Our
signal should be evaluated for the Subphylum Vertebrata as a whole, but should
also be calculated for all the main kind of vertebrate faunas, because - as we
are going to demonstrate - the next step, noise evaluation, will lead to
geographically-related data.
The noise: organic matter degradation, and taphonomic factors
Immediately
after death factors have acted on the organism, a percentage of his bone mass
ranging from 100% (death by natural causes not involving predation) to 0% (a
small animal being completely eaten). Then this "signal" begins its
degradation. The main degradation factors are:
Scavenging
of carcasses: it can affect any dead animal, regardless to the
death agent. Its degrading effects are size-related, as small carcasses can be
completely obliterated while dead elephant bones are supposed to survive scavenging
almost unaltered. Any kind of organic matter degradation by any scavenging
organism should be taken into account.
Physical
degradation from atmospheric factors, including fluitation: although evenly
acting on big and small remains, purely physical factors will unavoidably be
more sensible on the small, more fragile bones.
The
action of these factors requires bones to be exposed. Any taphonomic factor
influencing the exposition of bones to the above mentioned agents obviously
directly affects their preservation.
From
this point of view, we can state that the single most important factor in the
degradation of our signal is the environment itself, in terms of climate and
geographical setting of the scene of death. In first approximation, destructive
forces are roughly related with relief energy and with thermal excursion, with
humidity influencing the speed of degradation.
Looking
at relief energy, extremes can be found in the following situations:
- Low-energy, anoxic
conditions coupled with a rapid burial, epitomized by tar pits like Rancho
La Brea's, where the entire body - not excluding internal organs - can be
preserved. In general, any lake, marsh, river meander or lagoon is a
potential setting for rapid burial with very short transport of the carcass.
- High-energy conditions,
like a torrent: even the body of a big animal can be dismembered and
fragmented in unrecognizable portions
Looking
at temperature, constant temperatures - even when extreme - concur in favouring
preservation, especially when coupled with low humidity.
- Desert sands can
rapidly bury and mummify a dead body
- Permafrost has
demonstrated its ability to preserve exceptional fossil remains
I propose to define as TAPHOPLASTIC those
environments that favor the preservation of articulated and recognizable
biologic remains, particularly bones. The continental, taphoplastic
depositional environments can be easily identified in today's geography.
The opposite adjective of TAPHOSCLERIC applies
to those environments in which the fossilization is exceptional or impossible.
Apart
from this very preliminary and approximate descriptions, the integration of
signal with noise is a very complex, or even impossible, task to accomplish
accurately. In subaerial environments, we have very rare, almost unique taphoplastic
settings that can preserve our signal with exceptional fidelity, while the
great majority of geographical settings completely damps the signal.
When
trying to integrate the "signal" data with these powerful
"noise" generator, we can identify in first approximation the most
common kinds of continental environments, and can define their degree of taphoplasticity,
that is to say, fitness as fossil generators. This should express as a taphoplasticity
coefficient that will be maximum for the above mentioned low-energy
depositional settings, and equal to zero for the taphoscleric more energetic,
humid, highly variable temperature environments.
Taphoplasticity
itself could be expressed as a series of coefficients, one for each mass
category, or by a single coefficient that will be applied to all the columns of
the histogram.
To
integrate signal and noise, peculiar signal profiles for the main vertebrate
faunas should be available from the previous step of "signal
evaluation", and for any of the reference fauna/environment couples the
signal has to be multiplied for the preservational coefficient.
This
is the rough estimate of the final signal, "skeletons expected to have
entered the sediments".
After
their burial, carcasses enter the geological domain and are subject of
diagenetic processes including permineralization and deformation, that can
finally lead to the complete destruction of the bodily remains.
The
amount of time needed for a rigorous evaluation of the incidence of these
factors is enormous, and requires a complex process of integration of
geological and geographical information. All that can be said is that the
quality of the fossil record roughly degrades with age, and that this
degradation more sensibly affects small, fragile bones than big ones.
For
the sake of our study, I would suggest to ignore this factors, stating that as
soon as the bones enter the geological domain and begin their way towards
complete fossilization, they exit the ecological domain and our visibility.
A review of the main steps involved in the determination of the actual
signal
- Definition of the
general composition of the Subphylum Vertebrata for non-marine animals.
Species count (in terms of discrete ranges of average adult body mass) and
individual count (in terms of discrete ranges of body mass) for each
category (HISTOGRAMS I and II).
- Definition of the main
non-marine vertebrate faunas. For each reference fauna (e.g., "the
rain forest fauna of
- The percent incidence
of each reference fauna (e.g., "the rain forest fauna of
- Definition of
"signal" for each non-marine vertebrate "kind of
fauna" as illustrated above: a "signal" histogram (dead
bone availability) is generated for each worldwide fauna.
- Definition of an
eco-geographical context for each of the above mentioned worldwide faunas.
- Assignment of a "”taphoplasticity
index" to each eco-geographical context.
- Evaluation of the actual
signal by the application to each worldwide fauna of the
"preservational coefficient" of its respective eco-geographical
context
- Determination of the
final results by sum:
- How many living
species in the different adult average body mass categories
- How many living
vertebrates in the different body mass categories (sum of all the
"kinds of fauna")
- What signal in terms
of "post-predation" skeletons available to degradation, in the
different body mass categories (sum of all the "kinds of
fauna")
- What signal in terms
of "skeletons that entered the sediments" and became available
to diagenetic, geologic process, in the different body mass categories
(sum of all the "kinds of fauna") (HISTOGRAM NUMBER III)
The aim of this job is showing that the pattern
represented in HISTOGRAM NUMBER
IIIis the expression of a fauna like
the one represented in HISTOGRAM NUMBER I and HISTOGRAM NUMBER II.
- Maybe an algorithmic
link between HISTOGRAM III and HISTOGRAM II (and vice versa) could be
observed. This equation - if any - could be considered as the single most
important result of this research, and should be proposed for application
in environmental reconstructions.
The actual signal and the fossil record
This
is how we could roughly evaluate the amount of skeletons that, in discrete
individual body mass categories, enter the sediment. But now that we have
painstakingly obtained the signal and the noise, what should be said about
fossil record quality? How good is our non-marine vertebrate fossil record?
This question has two faces.
1)
Have we found all the fossils available?
We
could make both an overall comparison, or a comparison limited to the fossils
of a particular geologic era. If we make an histogram like the ones mentioned
above with all the non-marine vertebrate fossils of the desired era (I know
that it's almost impossible to count each and every fossil individual ever
found, but someone should try!) and compare it with HISTOGRAM III (roughly representing
"fossils that in theory could be found"), we will probably notice
differences for the past existence of very big terrestrial animals, but we will
also probably notice that in particular body mass categories there is still
much to be found. I imagine that small bodied animals are still to be found.
Even a very approximate numeric evaluation of all the fossils theoretically
available could be possible:
- As a first step, the
area of all the present continents has to be calculated (good
approximations exist in any geographic atlas)
- The second step is
calculating the area of all the exposed geological formations referrable
to continental environments, and assign a specific taphoplasticity index
to those environments
- The third step is a
simple division based on the values of HISTOGRAM III: a very rough
evaluation of the number of species and number of individuals that should
reasonably be expected to be found in each body size category, regardless
of diagenetic obliteration.
2)
Even if we find all the fossils, what is the relation between the diversity read
from the fossils, and the diversity in the original biosphere from which they
were generated ?
This
is the very final question: how well do fossils diversity represent the
original vertebrate diversity? Even without direct comparison with fossil data,
the answer comes from the comparison of HISTOGRAM I and II with HISTOGRAM III.
Taking
for good this relations (and maybe with the help of some algorithm proposed in
this paper, from an histogram of fossil individuals (or species) another
histogram of "then living species" could be derived, and the deficit
in small species could be more correctly evaluated.
We
should also briefly mention some seldom discussed implications about inferences
made from the bone alone.
Looking
at the vertical, chronological plane it's obvious that a single bone stands for
a successful story of reproduction in the evolutionary line departing from the
"common ancestors" of living organisms to the parents of that
particular animal whose bone we have found. One must be much more prudent about
what happened next: from the most reductive point of view, the only thing that
can be inferred from a well-recognizable bone is the existence of the very
single animal who owned that bone. If the animal is a subadult (not sexually
mature), it's absolutely impossible that it could have been engaged in
reproduction, so - as far as we can see - it could well be the last of its
kind. Even if the animal is a fully fledged adult, the possibility for its
phyletic lineage to extend in time beyond this particular specimen remains just
a possibility. In this respect, any phylogenetic reconstruction based on
isolated specimens could be criticized on a purely logical plan, while more
extensive monospecific assemblages are much more reliable indicators of a
breeding population, as well as the remains of eggs and nests with hatchlings.
What
this paper deals with, is something like an horizontal, isocronic view of that
particular bone. From the remains of a big predator it's easy to imagine a
complex food chain, or at least the existence of a guild of vegetarian animals
as preys. Extending this concept, a whole ecosystem must be taken into account,
with that particular bone as one of the pixels of the big picture.