ArticlePDF Available

Functional Convergence of Ecosystems: Evidence from Body Mass Distributions of North American Late Miocene Mammal Faunas

February 2006
Ecosystems 9(1):97-118

February 2006
9(1):97-118

Authors:

All Saints Episcopal School

There is disagreement among ecologists as to whether ecosystem system behavior in general is the net result of all of the complex internal system interactions (bottom-driven) or if the behavior is driven by a limited number of key processes (top-driven). If ecosystems are primarily bottom-driven in nature, then it is unlikely that any two complex ecosystems will ever behaviorally converge as a simple matter of probability, or that their behaviors will ever be predictable. Conversely, evidence of ecosystem convergence would suggest that the systems are top-driven, with the corollary that their behaviors can be understood (and therefore in principle predicted) without a complete understanding of their internal workings. Research has demonstrated that body mass distributions of terrestrial animals broadly reflect ecosystem function. Thus, comparable but causally disconnected terrestrial ecosystems that demonstrate similar body mass distributions would suggest ecosystem convergence. To look for this possible convergence, I generated body mass distributions in a time series for late Miocene North American mammal faunas from the Gulf Coastal Plains, Great Plains, and Pacific coastal region, and compared them with data from the modern Serengeti savanna region. The data show that during the early late Miocene Gulf Coastal Plain faunas resembled each other but were distinctly different from that of the Serengeti, the Great Plains fauna resembled the Serengeti, while the Pacific Coast fauna showed no resemblance to any of the others. However, during the latest Miocene the Gulf Coastal Plain faunas were transformed so as to strongly resemble the Serengeti fauna. The resemblance with the Serengeti was maintained by the Great Plains faunas until at least the end of the Miocene, while the Pacific Coast faunas remained distinctly different from the others. These findings suggest that the late Miocene ecosystems of the Gulf Coastal Plain and Great Plains regions (but not the Pacific Coast region) converged with that of the Serengeti savanna fauna and thus that these ecosystems were/are top-driven rather than bottom-driven in nature.

. Families represented in Late Miocene Mammal Faunas of the Gulf Coastal Plain and the Modern Seregenti Mammal Fauna, Listed by Order

…

. List of Species and their Body Masses in Genera Shared between Contemporaneous (either Middle Hemphillian or Latest Hemphillian) Gulf Coastal Plain and Great Plains faunas

…

Figures - uploaded by W. David Lambert

Content may be subject to copyright.

Content uploaded by W. David Lambert

Content may be subject to copyright.

Functional Convergence of

Ecosystems: Evidence from

Body Mass Distributions of

North American Late Miocene

Mammal Faunas

W. David Lambert*

The Louisiana School for Mathematics, Science, and the Arts, 715 College Ave., Natchitoches, Louisiana 71457, USA

ABSTRACT

There is disagreement among ecologists as to

whether ecosystem system behavior in general is

the net result of all of the complex internal system

interactions (bottom-driven) or if the behavior is

driven by a limited number of key processes (top-

driven). If ecosystems are primarily bottom-driven

in nature, then it is unlikely that any two complex

ecosystems will ever behaviorally converge as a

simple matter of probability, or that their behaviors

will ever be predictable. Conversely, evidence of

ecosystem convergence would suggest that the

systems are top-driven, with the corollary that their

behaviors can be understood (and therefore in

principle predicted) without a complete under-

standing of their internal workings. Research has

demonstrated that body mass distributions of ter-

restrial animals broadly reﬂect ecosystem function.

Thus, comparable but causally disconnected ter-

restrial ecosystems that demonstrate similar body

mass distributions would suggest ecosystem con-

vergence. To look for this possible convergence, I

generated body mass distributions in a time series

for late Miocene North American mammal faunas

from the Gulf Coastal Plains, Great Plains, and Pa-

ciﬁc coastal region, and compared them with data

from the modern Serengeti savanna region. The

data show that during the early late Miocene Gulf

Coastal Plain faunas resembled each other but were

distinctly different from that of the Serengeti, the

Great Plains fauna resembled the Serengeti, while

the Paciﬁc Coast fauna showed no resemblance to

any of the others. However, during the latest

Miocene the Gulf Coastal Plain faunas were trans-

formed so as to strongly resemble the Serengeti

fauna. The resemblance with the Serengeti was

maintained by the Great Plains faunas until at least

the end of the Miocene, while the Paciﬁc Coast

faunas remained distinctly different from the oth-

ers. These ﬁndings suggest that the late Miocene

ecosystems of the Gulf Coastal Plain and Great

Plains regions (but not the Paciﬁc Coast region)

converged with that of the Serengeti savanna fauna

and thus that these ecosystems were/are top-driven

rather than bottom-driven in nature.

Key words: thermodynamics; convergence; body

size; mammals; landscape; Miocene.

NTRODUCTION

Nonlinear, non-equilibrium systems (henceforth

called complex systems) are divisible into two

Received 9 June 2004; accepted 11 November 2004; published online

*Corresponding author; e-mail: dlambert@lsmsa.edu

Ecosystems (2006) 9: 1–28

DOI: 10.1007/s10021-004-0076-0

major categories on the basis of the processes that

generate and control their behavior. Some systems,

although characterized by a large number of com-

plex internal interactions, have their basic behavior

governed by relatively few overriding factors. Such

systems are here called top-driven systems, a ref-

erence to the hierarchical position of the driving

factors. In contrast, the behavior of other complex

systems is the result of the sum total of the

numerous interactions among components of those

systems, each interaction contributing signiﬁcantly

(although not necessarily equally). I call such sys-

tems bottom-driven due to the critical importance

of the lower levels in the process hierarchy (Perry

1994; Jo

˜rgensen 1992).

Knowing whether ecosystems are top-driven or

bottom-driven in nature is crucial for understand-

ing, and ultimately predicting their behavior. If

ecosystems are primarily bottom-driven, then

understanding and predicting their behavior re-

quires in principle that the majority if not all of the

interactions within that system be understood.

Considering the number and complexity of the

interactions present in all but the simplest ecosys-

tems, this is a daunting if not outright impossible

task. On the other hand if ecosystems are primarily

top-driven, then there is no need to pinpoint all of

the internal interactions within a given system to

understand its basic behavior. Rather, one need

only identify and understand the relatively few key

processes that drive the behavior of that system (by

no means necessarily an easy task, but clearly more

practical than unraveling all or most of the internal

interactions). Such processes could include ﬁre,

ﬂood, damaging wind, and large herbivore activity

among many others (Holling 1992). Considering

the current importance of conserving and/or

managing numerous ecosystems endangered by

human disturbance, such information about eco-

systems is valuable if not outright critical.

The above discussion leads to a crucial question:

are ecosystems predominantly top-driven, bottom-

driven, or are both conditions widespread? One

possible way to address this question relates to

ecosystem functional convergence. Imagine a

hypothetical scenario in which two bottom-driven

ecosystems with similar starting conditions are al-

lowed to evolve completely independently. Be-

cause the behavior of these systems is heavily

inﬂuenced by their numerous complex internal

interactions, the odds of the two of them inde-

pendently evolving similar behaviors in parallel is

extremely low. If however the systems are top-

driven in nature, then the base probability of them

evolving convergently is greatly increased because

many fewer factors must independently reach a

parallel state. In fact, the second law of thermo-

dynamics may in principle actually favor (though

not guarantee) convergence under these condi-

tions. A corollary of the second law of thermody-

namics as applied to complex systems is that, all

other factors being equal, systems that generate

large amounts of entropy are more likely to spon-

taneously develop than comparable systems gen-

erating less entropy. Thus, the second law of

thermodynamics has the potential to directionally

shift the evolution of complex system behaviors

(Chaisson 2001). This entropy-driven directional

shift in system behavior may be related to the

attractor phenomenon commonly invoked in dis-

cussions of complex, dynamic systems (for exam-

ple, Glass and Mackey 1988).

Even if it is accepted that behavioral convergence

is a plausible if indirect indicator of top-driven

ecosystems, actually demonstrating its existence

convincingly is problematic. One could in principle

examine and compare the behaviors of apparently

similar but spatially separate modern ecosystems in

great detail, looking for congruencies in underlying

function. However, even where such similarities

are noted their value as evidence for ecosystem

convergence is questionable, because the possibility

that the compared ecosystems are actually causally

linked components of a larger system cannot be

ruled out. For example, an increase in North

American snow goose populations due to rice

agriculture in the Gulf Coastal Plain has wreaked

havoc on prairie landscape around Hudson Bay in

Canada, showing a signiﬁcant connection between

these two ecosystems despite their seeming eco-

logical and spatial disparity (Zolkewich 1997). A

more convincing case could be made if behavioral

convergence was demonstrated between ecosys-

tems that were clearly causally disconnected, for

example widely separated in space, time, and even

biotic composition (for example, comparable mod-

ern and ancient ecosystems in different geographic

regions). An obvious and seemingly fatal method-

ological drawback to this approach presents itself,

namely that information about paleoecosystem

behavior is extremely difﬁcult to extract from the

fossil record because of the destruction of crucial

information. However, a new methodology has

been developed that can potentially reveal at least

broad aspects of the behavior of certain paleoeco-

systems despite this information loss, and this

methodology can in turn be used to make com-

parisons between these and modern ecosystems.

Research by Holling and others (1996) has

demonstrated that animal body sizes in terrestrial

2W. D. Lambert

communities are distinctly clumped rather than

continuous. This pattern is explained by the Tex-

tural Discontinuity Hypothesis, which can be

summarized as follows. Ecosystems are structured

by relatively few key processes operating at speciﬁc

temporal and spatial scales. The distinct temporal

frequencies and spatial scales characterizing these

processes create discrete, hierarchical landscape

structures with a scale-speciﬁc signature. The scale-

speciﬁc effect of these processes leads to a discon-

tinuous distribution of ecological structure and

pattern, which in turn constrains attributes of

animals residing on the resulting landscape,

including body size. These constraints force animals

to adopt body sizes that accommodate the speciﬁc

discontinuities in the landscape on which they live,

with the result that animal body size distributions

are discontinuous as well as the landscape (Holling

1992, 1995; Holling and others 1996; Lambert and

Holling 1998; Allen and others 1999; Holling and

Allen 2002). A corollary of the Textural Disconti-

nuity Hypothesis is the Bioassay Corollary, which

states that faunal body mass distributions (speciﬁ-

cally the pattern of clumps and gaps, forming a

kind of ‘‘spectrum’’) indirectly contain information

about both the landscape structures in an ecosys-

tem, and of particular importance for this study, the

ecological processes that produced them (Lambert

and Holling 1998). Reliable body size (mass) dis-

tributions can be generated for paleofaunas that

satisfy key conditions (see below). Thus, following

the Bioassay Corollary, it is apparent that by com-

paring the body mass distributions of modern fau-

nas and paleofaunas, one can indirectly compare

the ecological processes that structured their

respective ecosystems, and thus at least some aspect

of their behaviors.

It is important to recognize that the Textural

Discontinuity Hypothesis and the Bioassay Corol-

lary implicitly assume that the terrestrial ecosys-

tems in question are fundamentally top-driven.

Thus, using the Textural Discontinuity Hypothesis

to determine ultimately whether ecosystems are

top-driven suggests the possibility of circular rea-

soning. However, as shown in the above references

considerable evidence prior to this study has ac-

crued supporting the Textural Discontinuity

Hypothesis, and therefore by implication wide-

spread top-driven terrestrial ecosystems. Thus, its

viability is not strictly dependent upon the results

of this study, a fact that to at least some degree

mitigates the possible problem of circularity. One

can look upon an investigation into ecosystem

convergence as both an effort to determine

whether ecosystems are top-driven in nature, and

secondarily a test of whether the Textural Discon-

tinuity Hypothesis is a viable explanation for the

discontinuous body mass distributions observed in

many terrestrial faunas. Plausible observation of

ecosystem convergence would provide support for

the Textural Discontinuity Hypothesis, although a

failure to observe this phenomenon would not

necessarily disprove this hypothesis.

It has long been recognized that the middle-late

Miocene (15–5 million years ago [mya]) mammal

faunas of North America superﬁcially resembled

those of modern African savannas such as the

Serengeti region in terms of the adaptations of their

respective animals, and these North American

faunas have been called savanna faunas accord-

ingly. (Here, savanna is deﬁned as a moderately to

very open landscape containing both grass and

trees, with the tree density never becoming great

enough to form a deﬁned canopy [Webb 1977,

1983, 1984; Webb and others 1995]). However, no

one has investigated whether these similarities are

the result of similar structuring system processes, or

merely chance superﬁcial resemblances unrelated

to fundamental system function, probably at least

in part because of logistical difﬁculties. Thus, the

important question of whether these causally dis-

connected ecosystems actually converged func-

tionally remains unanswered.

The purpose of this study is to investigate whe-

ther functional convergence occurred between a

modern African savanna ecosystem and apparently

similar ecosystems from the Gulf Coastal Plain and

Great Plains of late Miocene North America, and

therefore try to determine whether these ecosys-

tems were/are top-driven in nature. To answer this

question, body mass distributions were obtained for

the modern Serengeti, late Miocene Gulf Coastal

Plain, and late Miocene Great Plains mammal

faunas, which were compared for congruency.

Body mass distributions from late Miocene mam-

mal faunas in the Paciﬁc Coast region were also

obtained. The Textural Discontinuity Hypothesis

and Bioassay Corollary imply that patterns in body

mass distributions indirectly reﬂect modes of eco-

system function, and thus congruence between

modern African mammal faunas and those of the

late Miocene Gulf Coastal Plain and Great Plains

would suggest that functional convergence be-

tween these respective ecosystems occurred. This

ﬁnding would be further supported if the body

mass distributions of the Paciﬁc Coast faunas dif-

fered signiﬁcantly from those of the more eastern

regions and modern Serengeti, since evidence

indicates that this region possessed a landscape

structure different from the rest of North America

Functional Convergence of Ecosystems 3

during this time (Webb 1977; Wing 1998). A

plausible ﬁnding of functional convergence be-

tween these causally disconnected ecosystems

would have important implications for the broader

issue of how ecosystems function, suggesting that

these ecosystems (and probably many others)

were/are primarily top-driven in nature and,

therefore that their behaviors are at least in prin-

ciple amenable to human understanding and pre-

diction. If no convergence is observed however,

then the question of whether these ecosystems

were/are top- or bottom-driven would remain

unresolved.

ERMS

Chronological subdivisions of the Miocene are dis-

cussed in terms of North American land mammal

ages. The Barstovian refers to the middle to late

middle Miocene, and ranges from 15 to 12.5 mya.

The Clarendonian refers to the late Miocene, and

ranges from 12.5 to 8.9 mya in age. The Hemphil-

lian refers to the latest Miocene to the very earliest

Pliocene, and ranges in age from 8.9 to 4.5 mya

(Prothero 1998). In this study the late Clarendonian

is deﬁned as 10–8.9 mya, early Hemphillian 8.9–7.5

mya, middle Hemphillian 7.5–6.5 mya, late Hem-

phillian 6.5–5.5 mya, and latest Hemphillian 5.5–

4.5 mya.

ETHODS

Specimens and Measurements

Linear measurements of specimens were made

with dial calipers to the nearest millimeter. The

circumferences of humeri and femora were mea-

sured with a tape measure to the nearest millime-

ter. The following specimen collections were

utilized in this study:

The Florida Museum of Natural History (Gaines-

ville, Florida)

The Nebraska State Museum of Natural History

(Lincoln, Nebraska)

The Los Angeles County Museum of Natural His-

tory (Los Angeles, California)

A list of specimens examined and measurement

data from those specimens can be obtained from

the author upon request.

Selection of Faunas

As explained in detail in Lambert and Holling

(1998), to be utilized in this type of analysis a pa-

leofauna must satisfy at least two requirements. (1)

It must be large enough to be reasonably repre-

sentative of the original mammal fauna (at least

over the size range of interest). And (2) it must

have a reasonably wide size range, although not

necessarily mouse to elephant. Because of prob-

lems with making adequate body mass estimates

(see below) and rarity, both very small (less than 3

kg) and large (greater than 2,000 kg) taxa were

excluded from the data sets. Generally, 20 was

considered the minimum acceptable faunal size for

inclusion in this study. However, for some faunas

where the distribution was robust in the central

region (a section observed to be particularly

important for characterizing the body mass distri-

butions of these faunas), sizes as low as 18 were

utilized. There were some cases in which no one

fauna in a critical geographical region and/or time

period was sufﬁciently diverse for inclusion in this

study. In this situation, faunal lists from localities

that were reasonably close in geography and age

were combined to form regional faunas that were

subject to analysis as if they represented a single

locality (locality faunal lists used in this study can

be obtained from the author upon request). The

modern African mammal faunas used for compar-

ison consisted of the Sierra Leone National Forest

(faunal data from Davies 1987) and the Serengeti

National Park (faunal data from Swynnerton

1963).

Body Mass Estimates

Body masses were estimated using regression

equations from Alexander (1989) and various pa-

pers in Damuth and MacFadden (1990) (body mass

estimates used in this study and associated regres-

sion equations can be obtained from the author

upon request). Average predictor error for these

regression equations (where given) ranged from

15–30%, with equations with values greater than

30% avoided if at all possible. Because of these

signiﬁcant and unavoidable regression predictor

errors, any body mass estimate for a fossil mammal

must be considered approximate at best. For the

modern African mammals, body mass values for

individual species were taken from Macdonald

(1984) and Nowak and Paradiso (1983).

Clump Assignment in Body Mass

Distributions

In an analysis of Pleistocene faunas utilizing the

Textural Discontinuity Hypothesis and Bioassay

Corollary, Lambert and Holling (1998) utilized an

4W. D. Lambert

index called the split moving window ratio to as-

sign species in a fauna to clumps; these indices

were plotted against rank order, with peaks in the

plot representing gaps between clumps. This

methodology is useful for detecting broad patterns

within body mass distributions (which commonly

can be easily discerned through a simple visual

examination of the data), but it is problematic be-

cause it lacks statistical rigor (Holling and Allen

2002). To minimize the subjective aspect of the

clump assignment process, each body mass distri-

bution in the data set was subjected to univariate

cluster analysis (CLUSTAN; Wishart 2003). In this

analysis, the body mass values in each faunal data

set were log transformed and unweighted, with

increasing sum of squares used as the clustering

algorithm. Similarity trees generated for the mod-

ern and Miocene mammal faunas are presented in

Figures 1, 2, 3, and 4. The body mass data for the

modern and Miocene faunas utilized in this study,

including taxa, rank orders, and clump assign-

ments, are presented in Appendices 1–13 (all

appendices referred to in this paper are available

online at www.springerlink.com).

ESULTS

Modern African Forest and Savanna

Faunas

Superimposed body mass plots from the modern

Sierra Leone National Forest and Serengeti faunas

are presented in Figure 5, with the body mass data

presented in Appendices 1 and 2 (http://www.

springerlink.com). Because the landscape structures

of these habitats are very different, the Textural

Discontinuity Hypothesis predicts that both the

number and/or location of the clumps in these dis-

tributions should be distinctly different, which is

indeed observed.

Clarendonian and Early Hemphillian

Faunas of the Great and Gulf Coastal

Plains

The body mass distributions of the late Clarendo-

nian and early Hemphillian Gulf Coastal Plain

faunas are presented in Figures 6 and 7, whereas

that of a late Clarendonian Great Plains fauna is

presented in Figure 8 (no adequate early

Hemphillian Great Plains fauna was available for

analysis). The body mass data are presented in

Appendices 3–5 (http://www.springerlink.com).

There is a complication involving the clump

assignments for the late Clarendonian Love fauna

shown in Figure 6. Examination of the similarity

tree generated for this body mass distribution

(Figure 2A) shows that the cluster analysis has split

clump 2, assigning its first seven members into a

distinct group that includes clump 1. Examination

of the Love fauna body mass data (Appendix 3;

http://www.springerlink.com) reveals that in fact

this splitting is based on a relatively small change in

body masses (Prosthenops sp. at 88 kg, Calippus

cerasinus at 102 kg) considering the error levels

underlying these estimates. When different masses

Figure 1. Similarity

trees generated for

modern African faunas

using cluster analysis.

(A) Sierra Leone Forest;

(B) Serengeti savanna.

Functional Convergence of Ecosystems 5

for these taxa were generated on the basis of this

uncertainty (20% was used as a reasonably con-

servative and representative error factor) and

plugged into the cluster analysis, more often than

not this splitting of clump 2 disappeared. When a

different clustering algorithm, average linkage, was

applied to the data set the taxa in clump 2 were

clearly separated from clump 1 in the resulting

similarity tree. Thus, in this instance I have chosen

to regard the splitting of clump 2 as artifacts of both

body mass estimation error and the increasing sum

of squares clustering algorithm.

Examination of the clump distributions for both

Gulf Coastal Plain faunas reveals that they are both

quite similar to each other, but signiﬁcantly dif-

ferent from that of the modern Serengeti fauna. In

Figure 2. Similarity trees generated for late Clarendonian through latest Hemphillian faunas from the Gulf Coastal Plain.

(A) Love fauna; (B) Mixon’s/McGehee Farm regional fauna; (C) Moss Acres Racetrack/Withlacoochee 4A River regional

fauna; (D) Palmetto fauna.

6W. D. Lambert

particular, both Gulf Coastal Plain distributions

have a wide, diverse clump centered near 100 kg

that bridges the third and fourth modern Serengeti

clumps. In striking contrast to these two Gulf

Coastal Plain faunas however, the body mass

clump distribution of the late Clarendonian Great

Plains fauna corresponds almost exactly with that

of the modern Serengeti fauna, although there are

differences in the clump widths that are probably at

least partly the result of errors in the data such as

incomplete faunal preservation. This ﬁnding begs

the question of when this congruence between the

late Miocene Great Plains and modern Serengeti

faunas ﬁrst arose. Unfortunately, no suitable Great

Plains or Gulf Coastal Plain faunas from the early or

middle Clarendonian could be found for analysis.

However, a suitable late Barstovian fauna from

Nebraska called the Valentine Railroad Quarry was

available, with its body mass clump distribution

presented in Figure 9 and the body mass data

presented in Appendix 6 (http://www.springer-

link.com). The late Barstovian faunal distribution

Figure 3. Similarity trees generated for late Barstovian through latest Hemphillian faunas from the Great Plains.

(A) Valentine Railroad Quarry fauna; (B) North Shore/Blue Jay regional fauna; (C) Cambridge fauna; (D) Santee/Devil’s

Nest regional fauna.

Functional Convergence of Ecosystems 7

shows a pattern that is clearly very different from

that of both the late Clarendonian Blue Jay/North

Shore and the modern Serengeti, as well as those

from the Gulf Coastal Plain. Thus, it appears that

the congruence in body mass distributions between

the Great Plains and Serengeti faunas occurred

sometime during the early/middle Clarendonian

time window (roughly 12–10 mya). It is unfortu-

nate that no early Hemphillian Great Plains fauna

was available for comparison with that from the

Gulf Coastal Plain. However, as discussed below the

data from the Great Plains faunas of the middle and

latest Hemphillian show a dramatic similarity with

the data from the modern Serengeti, and it is

therefore likely that this similarity extended into

the early Hemphillian as well.

Middle Hemphillian Faunas of the Great

and Gulf Coastal Plains

Body mass data for these faunas are presented in

Appendices 7 and 8 (http://www.springerlink.com).

Although the Gulf Coastal Plain and Great Plains

faunal body mass distributions were clearly different

during the late Clarendonian and presumably the

early Hemphillian, by the middle Hemphillian these

differences had essentially vanished (Figures 10,

11), with both late Miocene faunas showing clump

correspondences with each other and the modern

Serengeti. The most significant discrepancy is a small

third clump in the Gulf Coastal Plain Moss Acres

Racetrack/Withlacoochee River 4A body mass dis-

tribution that is contained by the larger third clump

Figure 4. Similarity trees generated for late Clarendonian through latest Hemphillian Pacific Coast faunas. (A) Ricardo

fauna; (B) Merhten/Mount Eden/Peace Valley regional fauna; (C) Schutler/Rome regional fauna.

8W. D. Lambert

Figure 5. Superimposed

body mass plots (number

of species against log

body mass, grams) for

the modern Sierra Leone

National Forest and

Serengeti faunas,

showing the location of

both clumps and gaps in

each distribution.

Figure 6. Superimposed

body mass plots (number

of species against log

body mass, grams) for

the modern Serengeti

fauna and the late

Clarendonian Bluejay/

Northshore faunas

(western Nebraska, 9

million years), showing

the location of both

clumps and gaps in each

distribution.

Functional Convergence of Ecosystems 9

Figure 7. Superimposed

body mass plots (number

of species against log

body mass, grams) for

the modern Serengeti

fauna and the late

Barstovian Valentine

Railroad Quarry fauna

(western Nebraska, 13

million years), showing

the location of both

clumps and gaps in each

distribution.

Figure 8. Superimposed

body mass plots (number

of species against log

body mass, grams) for

the modern Serengeti

fauna and the late

Clarendonian Love Bone

Bed fauna (North

Florida, 9 million years),

showing the location of

both clumps and gaps in

each distribution.

10 W. D. Lambert

in the Great Plains Cambridge distribution. In addi-

tion, the third clump in the Cambridge body mass

distribution (apparently corresponding to the third

clump in the Serengeti and the fourth clump in the

Moss Acres Racetrack-Withlacoochee River 4A dis-

tributions) is anomalously wide, with the fourth

clump distinctly displaced to the right. In evaluating

the significance of these differences (including dif-

ferences noted between late Clarendonian and early

Hemphillian faunas noted above), it is important to

remember that the data are subject to unavoidable

and essentially unquantifiable error from such

sources as body mass estimations and incomplete

faunal preservation. In addition, if indeed the dis-

tribution patterns reflect ecosystem function, then it

is unreasonable to expect spatially and temporally

highly disparate ecosystems to exactly resemble each

other even if they have converged. Thus, perfect

congruence between compared faunas is an unre-

alistic expectation, and small pattern irregularities

are ignored in this study unless there is a compelling

reason to do otherwise.

Latest Hemphillian Faunas of the Great

and Gulf Coastal Plains

The faunal body mass data for these faunas are

presented in Appendices 9 and 10 (http://

www.springerlink.com), whereas plots of these

body mass data are presented in Figures 12 and 13.

As in the middle Hemphillian, a striking congruence

exists among the Gulf Coastal Plain, Great Plains,

and the modern Serengeti faunas.

Paciﬁc Coast Faunas

Body mass distributions from late Clarendonian,

late Hemphillian, and latest Hemphillian Paciﬁc

Coast faunas are presented in Figures 14, 15, and

16, with the body mass data presented in Appen-

dices 11–13 (http://www.springerlink.com). The

late Clarendonian Pacific Coast faunal distribution

shows a rough similarity to the Serengeti savanna

fauna. However, unlike any of the other faunas

examined the clumps at the large end of the dis-

tribution are distinctly fragmented, while one

clump lies completely in a gap between two clumps

in the Serengeti distribution. The late Hemphillian

Pacific Coast fauna, in contrast, bears virtually no

resemblance to any of the other modern or Mio-

cene faunal distributions observed, with at least

two and possibly three clumps filling gaps between

clumps in the Serengeti savanna distribution. The

latest Hemphillian Pacific Coast fauna, like that of

the late Clarendonian Pacific Coast fauna, bears a

partial resemblance to the Serengeti savanna fauna.

Figure 9. Superimposed

body mass plots (number

of species against log

body mass, grams) for

the modern Serengeti

fauna and the early

Hemphillian Mixon’s

Farm/McGehee faunas

(North Florida, 8 million

years), showing the

location of both clumps

and gaps in each

distribution.

Functional Convergence of Ecosystems 11

However, the broadest clump in the distribution

strongly overlaps two clumps from the Serengeti

savanna. Thus, although there are some partial

similarities, none of the Pacific Coast faunas shows

a compelling congruence with any of the more

eastern late Miocene faunas or the Serengeti sa-

vanna fauna.

ISCUSSION

Demonstration of Ecosystem Functional

Convergence

A fundamental issue addressed in this study is

ecosystem functional convergence, an issue that

has important implications for ecosystem properties

(speciﬁcally, whether they tend to be top- or bot-

tom-driven in nature). Thus, a crucial question

arises: do the faunal body mass data from the late

Miocene North American and modern Serengeti

mammal faunas plausibly support ecosystem

functional convergence?

As discussed above, according to the Textural

Discontinuity Hypothesis and Bioassay Corollary

faunal body mass distributions indirectly reﬂect

ecosystem function. Thus, causally disconnected

ecosystems that show congruent faunal body mass

distributions strongly suggest that functional con-

vergence has occurred. The Gulf Coastal Plain late

Clarendonian and early Hemphillian faunas show

body mass distributions that are consistent with

each other, but very different from that of the

modern Serengeti fauna. Thus, despite the fact that

this region is commonly described as having both a

savanna landscape and climate (for example, Webb

1977; see below), the evidence suggests that these

ecosystems functioned in a manner signiﬁcantly

different from the modern Serengeti. In contrast

however, the body mass distribution of the late

Clarendonian Blue Jay/North Shore fauna from the

Great Plains shows a deﬁnite similarity to that of

the Serengeti. If indeed the Textural Discontinuity

Hypothesis is correct, then it would appear that the

late Clarendonian savanna of the Great Plains re-

gion both resembled and in some important way

functioned like the modern Serengeti savanna, and

therefore that these two ecosystems are conver-

gent. Moreover, because the late Barstovian Val-

Figure 10. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the middle Hemphillian Cambridge fauna (western Nebraska, 7 million years), showing the location of both

clumps and gaps in each distribution.

12 W. D. Lambert

entine Quarry fauna from the Great Plains shows a

faunal body mass distribution that is very different

from its late Clarendonian counterpart (as well as

those from the late Clarendonian and early Hem-

phillian of the Gulf Coastal Plain), we can deduce

that the convergence occurred sometime during

the early or middle Clarendonian. Thus, there is

evidence not just of convergence between discon-

nected ecosystems, but regional ecosystem evolu-

tion. Signiﬁcantly, once established this new

pattern continued with little change in the Great

Plains until the very end of the Miocene.

The Gulf Coastal Plain experienced regional

evolution as well, but not until the middle Hem-

phillian. The body mass distribution for the Moss

Acres Racetrack/Withlacoochee River 4A fauna

resembles those from the late Miocene Great Plains

and the modern Serengeti rather than its older

counterparts from the late Clarendonian and early

Hemphillian. Thus, the ecosystem of the Gulf

Coastal Plain apparently experienced both regional

evolution and convergence at the same time, like

the Great Plains did sometime during the early or

middle Clarendonian (but see discussion below).

Like the Great Plains, the Gulf Coastal Plain

maintained this pattern until the very end of the

Miocene.

A signiﬁcant aspect of the body mass distributions

of the Gulf Coastal Plain and Great Plains faunas is

their relative consistency through time. For exam-

ple, the Gulf Coastal Plain faunas maintained a

uniform body mass distribution from the late

Clarendonian into the early Hemphillian before

making a transition to a pattern congruent with the

modern Serengeti and Great Plains faunas in the

middle Hemphillian, which in turn remained un-

changed for the rest of the Miocene. Similarly, the

Great Plains faunas maintained consistent body

mass distributions from the late Clarendonian to the

end of the Miocene. This regional coherence sug-

gests that these patterns have a shared, underlying

cause as opposed to being caused by simple random

chance, which I here propose is the result of shared

landscape-structuring ecological processes.

If this hypothesis is correct, than in principle one

would predict that different patterns would emerge

in comparable data sets from regions where the

landscape was signiﬁcantly different from that of

Figure 11. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the middle Hemphillian Moss Acres Racetrack/Withlacoochee River 4A faunas (north Florida, 7 million years),

showing the location of both clumps and gaps in each distribution.

Functional Convergence of Ecosystems 13

the late Miocene Gulf Coastal Plain and Great

Plains. One such region in late Miocene North

America is the Paciﬁc Coast region. Signiﬁcant

faunal and ﬂoral evidence indicate that the Paciﬁc

Coast experienced a precocious increase in aridity

compared to the rest of North America during the

late Miocene, developing extensive desert and

chaparral habitats by the Clarendonian (Webb

1977; Webb and others 1995; Wing 1998). If this

interpretation of the evidence is correct, then

according to the Textural Discontinuity Hypothesis

and Bioassay Corollary the body mass distributions

of the late Miocene mammal faunas from this re-

gion should differ signiﬁcantly from those of the

Gulf Coastal Plains and Great Plains. A contrary

ﬁnding, although not necessarily supporting ran-

dom chance as a causal factor, would suggest that a

different explanation underlies the patterns seen in

the late Miocene and modern Serengeti faunas.

Consistent with expectations of the Textural Dis-

continuity Hypothesis, none of the late Miocene

faunas from the Paciﬁc Coast shows more than a

modest resemblance to any of their eastern coun-

terparts. Indeed, unlike the Gulf Coastal Plain and

Great Plains there is no regional coherence among

these faunas. This lack of coherence may be partly

explainable by geographic disparity, the late

Clarendonian Ricardo fauna having come from

central Southern California, the Merhten/Peace

Valley/Mount Eden fauna from central California

within the Central Valley, and the Schutler/Rome

fauna from western Oregon. Despite this disparity

however, the Paciﬁc Coast data can still be regarded

as validly testing the Textural Discontinuity

Hypothesis because substantial paleoﬂoral evidence

indicates that the late Miocene drying trend

encompassed the entire Paciﬁc Coast region south

of Washington (Wing 1998).

In conclusion, following the precepts of the

Textural Discontinuity Hypothesis and the Bioassay

Corollary, there is substantial evidence that some

level of functional convergence occurred between

the modern Serengeti savanna ecosystem and those

of the late Miocene Gulf Coastal Plain and Great

Plains of North America. Based on the principles of

probability and complex system thermodynamics,

this ﬁnding implies that these ecosystems were/are

top driven in nature, their behavior driven pri-

Figure 12. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the latest Hemphillian Devil’s Nest/Santee faunas (western Nebraska, 5 million years), showing the location of

both clumps and gaps in each distribution.

14 W. D. Lambert

marily by a relatively small number of key factors.

Since the late Miocene ecosystems have vanished

(essentially changed beyond any recognition

through time), this information has no practical

relevance to them. However, this is not true for the

modern Serengeti savanna ecosystem, which is

indeed subject to and most likely threatened by a

host of human disturbances (see papers in Sinclair

and Arcese 1995). By focusing study on the key

factors that drive its behavior, one could in prin-

ciple discover whether this ecosystem is being

fundamentally affected by current levels of human

disturbance, how or if it is going to fundamentally

change in response, and what steps should be taken

to prevent this change should it prove necessary or

desirable. Similar studies could also be done for

other threatened ecosystems, assuming that they

are top-driven as well, although clearly this would

be more practical for some than others.

Alternative Explanations to Ecosystem

Functional Convergence

It is important to remember that the conclusions

about ecosystem behavior drawn from this study

are completely inferential in nature. For obvious

reasons none of the Miocene ecosystems examined

in this study has been directly observed by ecolo-

gists, nor will they ever be barring the unlikely

invention of a time machine. The body mass dis-

tribution patterns and associated paleontological

evidence are broadly consistent with the Textural

Discontinuity Hypothesis and the Bioassay Corol-

lary of Holling and colleagues, with implications for

ecosystems as discussed above. However, other

explanations only partially related or completely

unrelated to the Textural Discontinuity Hypothesis

could conceivably account for or contribute to

these patterns as well.

One such possible alternative explanation is

taxonomic bias. In this scenario, one could argue

that the similarities observed in the body mass

distributions are the result of comparisons between

mammal faunas with similar faunal compositions

(and thus not truly independent) rather than eco-

system functional convergence. The taxonomic bias

hypothesis does a poor job of explaining the simi-

larities between the modern Serengeti and the

Miocene mammal faunas. Only two genera are

shared between all of the combined Miocene

Figure 13. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the latest Hemphillian Palmetto fauna (central Florida, 5 million years), showing the location of both clumps

and gaps in each distribution.

Functional Convergence of Ecosystems 15

mammal faunas and that of the Serengeti, Felis and

Canis (both only found in latest Hemphillian Mio-

cene faunas) (see Appendices 2–13). Not surpris-

ingly, the similarity increases when comparisons

are made at the family level (Table 1). However,

only five families are shared at this higher taxo-

nomic level, representing 28% of the Late Miocene

collective fauna and 31% of the Serengeti fauna

respectively. Thus, there is no reason to question

the faunal independence of these temporally and

spatially disparate faunas.

The case for taxonomic bias helping to produce

similarities between the contemporaneous late

Miocene mammal faunas has considerably more

plausibility. For example, consider the Cambridge

(Great Plains, middle Hemphillian) and Moss Acres

Racetrack/Withlacoochee River 4A (Gulf Coastal

Plain, middle Hemphillian) faunas, which are the

ﬁrst late Miocene faunas to show interregional

convergence in their body mass distributions. These

two faunas share at least four well identiﬁed species

(17% of the Cambridge fauna, 19% of the Moss

Acres Racetrack/Withlacoochee River 4A fauna),

with the actual number possibly being higher be-

cause accurate species level identiﬁcation of fossil

specimens can be difﬁcult, and 12 genera (52% of

the Cambridge fauna, 63% of the Moss Acres

Racetrack/Withlacoochee River 4A fauna). Such an

overlap is not surprising when one considers that

many if not most of these mammals are large and

highly mobile (for example, horses, camels,

pronghorn ‘‘antelopes’’, and rhinoceroses among

others), and that no major geographical barriers,

such as mountain ranges, separate the Gulf Coastal

Plain and the Great Plains. A similar faunal overlap

exists between the Palmetto (Gulf Coastal Plain,

latest Hemphillian) and Devil’s Nest/Santee faunas

(Great Plains, latest Hemphillian), which share

eight species (26% of the Palmetto fauna, 36% of

the Devil’s Nest/Santee fauna) and 13 genera (41%

of the Palmetto fauna, 59% of the Devil’s Nest/

Santee fauna) respectively.

The species shared between contemporaneous

faunas create a clear overlap in the compared data

sets, but it is unlikely that these overlaps by

themselves are sufﬁcient to create observed pattern

similarities. However, the overlap at the generic

level is considerably larger, greater than 50% in

Figure 14. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the late Clarendonian Ricardo fauna (southern California, 9 million years), showing the location of both clumps

and gaps in each distribution.

16 W. D. Lambert

each case. Thus, if many or most of these genera

contain species in a relatively narrow size range

then these overlaps could deﬁnitely create a bias in

favor of similarity. To check for this possibility, the

body masses of species in genera shared between

middle and latest Hemphillian Gulf Coastal Plain

and Great Plains faunas were compared (Table 2).

The intrageneric size similarity was considerable for

the middle Hemphillian faunas, particularly if a

20% error factor is incorporated to compensate for

body mass estimation errors. With the error factor

included, the following middle Hemphillian genera

contain species that overlap in body size: Thinoba-

distes (T. segnis =Thinobadistes sp.), Nannippus

(N. minor =Nannippus sp.), Calippus (C. mccar-

tyi =Calippus sp.), Hipparion (Hipparion cf.

forcei =Hipparion cf. tehonense), and Pseudoceras (cf.

Pseudoceras =Pseudoceras sp.) (five additional species

in all). Similarly, the following latest Hemphillian

genera contain species that overlap in body size:

Machairodus (Machairodus sp. = Machairodus sp.),

Nannippus (N. aztecus =N. lenticularis), Teleoceras (T.

fossiger =Teleoceras sp.), and Platygonus (Platygonus

sp. = Platygonus sp.) (four additional species in all).

Based upon the fundamental premise of the Tex-

tural Discontinuity Hypothesis, congeneric species

that are similar in size can be considered functional

equivalents, and thus nine species can be consid-

ered shared between the middle Hemphillian

faunas (42% of the Moss Acres Racetrack/Withla-

coochee River 4a fauna, 38% of the Cambridge

fauna) and 12 species shared between the latest

Hemphillian faunas (39% of the Palmetto fauna,

55% of the Santee/Devil’s Nest fauna). The revised

number of species (including functional species)

shared between these faunas is considerable. Thus,

taxonomic bias cannot be ruled out as a significant

contributor to the interregional similarities ob-

served between these late Miocene faunas, as op-

posed to ecosystem convergence alone.

A plausible alternative explanation for this in-

crease in interregional similarity is ecosystem

expansion. During the late Clarendonian and pre-

sumably early Hemphillian, the Gulf Coastal Plains

and Great Plains appear to have constituted eco-

logically distinct regions, with their body mass

distributions differing accordingly. However, from

the middle Hemphillian through the end of the

Figure 15. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the late Hemphillian Merhten/Mount Eden/Peace Valley faunas (central California, 6 million years), showing

the location of both clumps and gaps in each distribution.

Functional Convergence of Ecosystems 17

Miocene ecological conditions in the eastern and

central regions of North America may have chan-

ged by becoming more broadly uniform, with the

Great Plains expanding so as to encompass the Gulf

Coastal Plain and form a single broad, multi-re-

gional ecosystem that was convergent with the

modern Serengeti savanna of Africa (see discussion

below). This proposed absorption of the Gulf

Coastal Plain by the Great Plains ecosystem, if it

indeed occurred, is interesting because it runs

counter to a widespread belief that the Gulf Coastal

Plain served as a relatively mesic savanna refuge for

certain mammal taxa (for example, three-toed

horses and browsing ungulates) that vanished from

Great Plains during the late Miocene, particularly at

the end when a ﬁnal conversion of the Great Plains

to a steppe landscape is thought to have occurred

(for example, Webb 1977, 1983; Webb and others

1995).

Taxonomic bias can, at least in part, be consid-

ered a methodological artifact. However certain

macroscale (as opposed to mesoscale) geographical

factors, such as temperature, rainfall, and eleva-

tion, have been observed to correlate with body

size in mammals and thus in principle could have

helped produce the faunal similarities in question

in addition to or instead of ecosystem functional

convergence (Currie 1991). For example, a number

of researchers (for example, Currie 1991; Badgely

and Fox 2000; Cannon 2004) have observed a po-

sitive relationship between medium and large

mammal diversity and potential evapotranspiration

[a measure of solar energy received in a given area

rather than an actual measure of the amount of

water transpired by plants (Perry 1994)] in both

modern and Pleistocene North American habitats.

Based on this ﬁnding, they have proposed that

mammal diversity in these size categories is dis-

tinctly limited by total available environmental

energy. Unfortunately, the exact causal basis for

this limitation has not been clearly explained, and

thus it is difﬁcult to evaluate the implications of

these ﬁndings for the Textural Discontinuity

Hypothesis and the hypothesis of ecosystem func-

tional convergence. Indeed, because the exact

nature of the ecological processes that generate

discontinuous habitat structure is not well under-

stood, it is conceivable that potential evapotrans-

Figure 16. Superimposed body mass plots (number of species against log body mass, grams) for the modern Serengeti

fauna and the latest Hemphillian Schutler/Rome faunas (western Oregon, 5 million years), showing the location of both

clumps and gaps in each distribution.

18 W. D. Lambert

piration is actually incorporated in the Textural

Discontinuity Hypothesis. Nevertheless, pending

further investigation, this and perhaps other geo-

graphical factors cannot be ruled out as possible

alternative causes for the body mass patterns ob-

served in this study, although as of now there is no

strong evidence to this effect.

Possible Causes for Ecosystem

Convergence Between Modern Savanna

Africa and Late Miocene North America

Assuming that ecological convergence did indeed

occur between the Serengeti ecosystem of modern

Africa and some ecosystems of late Miocene North

America, an obvious question presents itself: what

changes occurred in the Gulf Coastal Plain and

Great Plains during the late Miocene that caused

these two regional ecosystems (or perhaps com-

bined super-regional ecosystem) to transform and

begin to function at least approximately like the

modern Serengeti ecosystem? Unfortunately body

mass distributions, although providing evidence for

broad behavior, reveal little about the actual pro-

cesses producing that behavior. However, impor-

tant ecological changes known to have occurred in

North America during the late Miocene may pro-

vide intriguing circumstantial clues.

Through most of the Miocene, North America as

a whole experienced a deﬁnite trend in landscape

change, with forests being steadily replaced by

more open savanna. By the middle Miocene

(Barstovian), much of the continent including the

Great Plains had been converted into woodland

savanna, although the Gulf Coastal Plain is thought

to have been a forest refuge at this time, holding

onto animals that had already vanished from other

regions (particularly archaic browsing ungulates).

This trend continued through the rest of the Mio-

cene, incorporating the Gulf Coastal Plain in addi-

tion to the Great Plains. Woodland savannas

became steadily more open to form grassland sav-

annas, and the diversity of browsing ungulates

dropped accordingly. In the Great Plains, this trend

reached its peak with the formation of a grass

steppe in the late and latest Hemphillian, although

the Gulf Coastal Plains is believed to have acted as a

savanna refuge until the end of the Miocene (Webb

1977, 1983, 1984; Webb and others 1995).

The usual explanation given for this landscape

trend is increasing aridity, both in terms of reduced

annual rainfall and more intensely seasonal rainfall

(for example, Webb 1977), which is presumed to

have steadily concentrated trees into areas with

permanently high moisture. Similarly, seasonal

rainfall is a critical factor helping to maintain the

savanna landscape of the modern African Serengeti,

inducing grazing ungulates to migrate and helping

to inﬂuence tree and grass distribution (see

numerous papers in Sinclair and Norton-Grifﬁths

1979). In what is considered to have been an analog

to this situation in the late Miocene of North

America, Hulbert (1982) observed evidence of sea-

sonal occupation of the late Clarendonian Love site,

a permanent water source in the form of a sinkhole

pond, by the three-toed horse Neohipparion.

If however both the Gulf Coastal Plain and Great

Plains were dominated by a savanna landscape and

seasonal rainfall during the late Miocene as sug-

gested by the available (if limited) circumstantial

evidence, then we are faced with a conundrum.

The body mass data suggest that the Great Plains

ecosystem functioned in some basic respect like the

modern Serengeti from at least the late Clarendo-

nian through to the very end of the Miocene,

consistent with the traditional view that this region

resembled the savanna region of modern Africa

Table 1. Families represented in Late Miocene

Mammal Faunas of the Gulf Coastal Plain and the

Modern Seregenti Mammal Fauna, Listed by Order

Late Miocene Gulf Coastal Plain Serengeti

Order Edentata Order Rodentia

Megalonychidae (EH, MH, LH) Pedetidae

Mylodontidae (EH, MH, LH) Order Primates

Order Carnivora Cercopithecidae

Mustelidae (LC, EH, MH, LH) Order Carnivora

Canidae (LC, EH, MH, LH) Canidae

Procyonidae (LC, EH, LH) Mustelidae

Ursidae (MH, LH) Viverridae

Felidae (LC, EH, MH, LH) Hyaenidae

Nimravidae (LC) Felidae

Order Artiodactyla Order Tubulidendata

Tayassuidae (LC, EH, MH, LH) Orycteropodidae

Protoceratidae (EH, LH) Order Pholidota

Camelidae (LC, EH, MH, LH) Manidae

Gelocidae (LC, MH) Order Hyracoidea

Antilocapridae (LC, MH, LH) Procaviidae

Cervidae (LH) Order Artiodactyla

Dromomerycidae (LC, EH, MH) Suidae

Order Perrisodactyla Hippopotamidae

Equidae (LC, EH, MH, LH) Giraffidae

Tapiridae (LC, EH, MH, LH) Bovidae

Rhinoceratidae Order Perissodactyla

(LC, EH, MH, LH) Equidae

Rhinoceratidae

Note that the late Miocene list reﬂects family presence in the studied faunas only.

(Abbreviations: LC, late Clarendonian Love fauna; EH, early Hemphillian Mi-

xon/McGehee Farm fauna; MH, middle Hemphilian Moss Acres Racetrack/

Withlacoochee River 4A fauna; LH, latest Hemphillian Palmetto fauna).

Functional Convergence of Ecosystems 19

during this time. However, contradicting the tra-

ditional view, the body mass data from the Gulf

Coastal Plain faunas suggests that this ecosystem

functioned signiﬁcantly differently from those of

both the modern Serengeti and Great Plains during

the late Clarendonian and early Hemphillian.

In trying to resolve this apparent contradiction,

some facts must be considered. First, it can be very

difﬁcult to tease details of landscape structure from

the fossil record. For example, a paleoﬂora can

provide information on what kinds of plants were

present at a locality and their relative abundance,

but not their exact spatial distribution (Lambert

1994). Thus, two paleoﬂoras could closely resemble

each other in most respects, yet still represent

localities with distinctly different landscapes. Sec-

ondly, a variety of ecological factors beyond

seasonal rainfall can help produce an open,

Table 2. List of Species and their Body Masses in Genera Shared between Contemporaneous (either Middle

Hemphillian or Latest Hemphillian) Gulf Coastal Plain and Great Plains faunas

Species Mass (grams) Fauna

Middle Hemphillian Faunas

Thinobadistes

T. segnis 948,000 Moss Acres Racetrack/With. River 4A

Thinobadistes sp. 1,066,000 Cambridge

Nannippus

N. morgani 36,000 Moss Acres Racetrack/With. River 4A

N. minor 79,000 Moss Acres Racetrack/With. River 4A

Nannipus sp. 97,000 Cambridge

Calippus

Calippus sp. 77,000 Cambridge

C. theristes 132,000 Cambridge

C. mccartyi 73,000 Moss Acres Racetrack/With. River 4A

Hipparion

Hipparion cf. tehonense 151,000 Moss Acres Racetrack/With. River 4A

Hipparion cf. forcei 174,000 Cambridge

Dinohippus

Dinohippus sp. 330,000 Moss Acres Racetrack/With. River 4A

Dinohippus sp. 259,000 Cambridge

Aphelops

Aphelops mutilis 764,000 Moss Acres Racetrack/With. River 4A

Aphelops kimballensis 1,792,000 Cambridge

Pseudoceras

Pseudoceras sp. 11,100 Cambridge

cf. Pseudoceras 9,050 Moss Acres Racetrack/With. River 4A

Pediomeryx

Pediomeryx hemphillensis 108,000 Moss Acres Racetrack/With. River 4A

Pediomeryx figginsi 414,000 Cambridge

Latest Hemphillian Faunas

Machairodus

Machairodus sp. 150,000 Palmetto

Machairodus sp. 134,000 Santee/Devil’s Nest

Nannippus

N. aztecus 79,000 Palmetto

N. lenticularis 98,000 Santee/Devils Nest

Teleoceras

Teleoceras sp. 963,000 Palmetto

T. fossiger 1,139,000 Santee/Devil’s Nest

Platygonus

Platygonus sp. 240,000 Palmetto

Platygonus sp. 240,000 Santee/Devil’s Nest

Hemiauchenia

Hemiauchenia sp. 204,000 Palmetto

H. vera 159,000 Santee/Devil’s Nest

20 W. D. Lambert

savanna-like landscape, including ﬁre, physio-

graphic ﬂatness, and animal activity. For example,

cattle grazing has been observed to create savannas

in regions of South America that would otherwise

have been forest (Retallack 1982; Sarmiento 1984).

Thus, apparently similar savanna landscapes can in

principle be generated by different underlying

processes.

With the above facts in mind, the following is a

plausible hypothetical scenario that is consistent

with the available evidence and also resolves the

aforementioned apparent contradiction. During

the late Clarendonian and early Hemphillian, both

the Gulf Coastal Plains and Great Plains had rel-

atively open landscapes. The landscape of the

Great Plains was very much like the modern

Serengeti (that is, local pockets of open woodlands

around water sources separated by open grass-

lands), and it was generated by similar ecological

factors (for example, strongly seasonal rainfall,

facilitative grazing by ungulates, grass ﬁre, tree

predation by proboscideans, and so on). Thus, it

would have been truly functionally convergent

with the Serengeti. In contrast, the landscape of

the Gulf Coastal Plain was open, but in a fashion

that differed in detail from the Great Plains (for

example, the trees could have been more widely

dispersed throughout the terrain, ﬁre could have

played a lesser role in maintaining the structure,

and so on), with this particular structure main-

tained by different combinations of ecological

factors. Given the proximity of the Gulf of Mexico,

it is plausible that the rainfall may not have been

strongly seasonal at this time. However, as North

America continued its pan-Miocene trend towards

increasing aridity, by the middle Hemphillian the

conditions governing the Great Plains spread to

the Gulf Coastal Plain. As a result, these land-

scapes became similar and the Gulf Coastal Plain

developed faunal body mass distributions like

those found in the Great Plains.

Unfortunately, at this time it is not possible to

adequately test this scenario. Faunal body mass

distributions unfortunately represent at best diffuse

shadows of the processes governing their respective

ecosystems, and currently provide little or no de-

tailed information. Hopefully, future research will

allow ecologists to better interpret the ecological

meaning of patterns within body mass distribu-

tions, or perhaps completely new methodologies

will be invented that provide a clearer picture into

how different paleoecosystems functioned, many

almost certainly without modern analogs and thus

not amenable to analysis by the methodology used

in this study.

ACKNOWLEDGEMENTS

The Florida Museum of Natural History, Nebraska

State Museum of Natural History, and Los Angeles

County Museum of Natural kindly gave me access

to their collections. S. David Webb, Buzz Holling,

Paul Marples, Craig Allen, and Jan Zenzimir pro-

vided guidance and valuable suggestions during the

conduction of this study. Craig Allen and three

anonymous reviewers provided invaluable criticism

of this manuscript. This project received ﬁnancial

support from NASA Earth Observation Systems and

Terrestrial Ecosystems grants, and a generous grant

from the Richard Brown Foundation of the Louisi-

ana School for Math, Science, and the Arts.

REFERENCES

Alexander RM. 1989. Dynamics of dinosaurs and other extinct

giants. New York: Columbia University Press.

Allen CR, Forys EA, Holling CS. 1999. Body mass patterns pre-

dict invasions and extinctions in transforming landscapes.

Ecosystems 2:114–21.

Badgely C, Fox DL. 2000. Ecological biogeography of North

American mammals: species density and ecological structure in

relation to environmental gradients. J Biogeogr 27:1437–67.

Cannon MD. 2004. Geographic variability in North American

mammal community richness during the terminal Pleisto-

cene. Quatern Sci Rev 23:1099–123.

Chaisson EJ. 2001. Cosmic evolution: the rise of complexity in

nature. Cambridge (MA): Harvard University Press.

Currie DJ. 1991. Energy and large-scale patterns of animal- and

plant-species richness. Am Nat 137:27–49.

Damuth J, MacFadden B. 1990. Body size in mammalian pa-

leobiology. Cambridge (UK): Cambridge University Press.

Davies AG. 1987. The Gola Forest Reserves, Sierra Leone. Gland,

Switerzland: IUCN.

Glass L, Mackey M. 1988. From clocks to chaos: the rhythm of

life. Princeton (NJ): Princeton University Press.

Holling CS. 1992. Cross-scale morphology, geometry, and

dynamics of ecosystems. Ecol Monogr 62:447–502.

Holling CS. 1995. What barriers? What bridges? In: Gunderson

LH, Holling CS, Light SS, Eds. Barriers and bridges to the re-

newal of ecosystems and institutions. New York: Columbia

University Press.

Holling CS, Peterson G, Marples P, Sendsimir J, Redford K,

Gunderson L, Lambert W. 1996. Self-organization in ecosys-

tems: lumpy geometries, periodicities, and Morphologies. In:

Self W, Walker B, Eds. Global change and terrestrial ecosys-

tems. Oxford (UK): Oxford University Press.

Holling CS, Allen CR. 2002. Adaptive inference for distinguish-

ing credible from incredible patterns in nature. Ecosystems

5:319–28.

Hulbert RC. 1982. Population dynamics of the three-toed horse

Neohipparion from the late Miocene of Florida. Paleobiology

8(2):159–67.

˜rgensen SE. 1992. Integration of ecosystem theories: a pattern.

Dordrecht: Kluwer.

Lambert WD. 1994. The fauna and paleoecology of the late

miocene moss acres racetrack site, Marion County,

Functional Convergence of Ecosystems 21

Florida. PhD dissertation. Gainesville, FL: University of

Florida.

Lambert WD, Holling CS. 1998. Causes of ecosystem transfor-

mation at the end of the Pleistocene: evidence from mammal

body mass distributions. Ecosystems 1:157–75.

Macdonald D. Ed. 1984. The encyclopedia of mammals. New

York: Facts on File Publications.

Nowak RM, Paradiso JL. 1983. Walker’s Mammals of the World.

Baltimore: Johns Hopkins University Press.

Perry DA. 1994. Forest ecosystems. Baltimore: Johns Hopkins

University Press.

Prothero DR. 1998. The chronological, climatic, and paleogeo-

graphic background to North American evolution. In: Janis CM,

Scott KM, and Jacobs LL, Eds. Evolution of Tertiary mammabls

of North America. Oxford (UK): Oxford University Press.

Retallack GJ. 1982. Paleopedological perspectives on the devel-

opment of grasslands during the Tertiary. Third North Am

Paleontol Conv Proc 2:417–21.

Sarmiento G. 1984. The ecology of neotropical savannas. Cam-

bridge (MA): Harvard University Press.

Sinclair ARE, Norton-Griffiths M. 1979. Serengeti. Chicago:

University of Chicago Press.

Sinclair ARE, Arcese P. 1995. Serengeti II: dynamics, manage-

ment, and conservation of an ecosystem. Chicago: University

of Chicago Press.

Swynnerton GH. 1963. Fauna of the Serengeti National Park.

Mammalia 22:435–50.

Webb SD. 1977. A history of savanna vertebrates in the New

World. Part 1:North–America. Annu Rev Ecol Syst 8:355–

80.

Webb SD. 1983. The rise and fall of the late Miocene ungulate

fauna in North America. In: Nitecki MH, Ed. Coevolution.

Chicago: University of Chicago Press.

Webb SD. 1984. Ten million years of mammal extinctions in

North America. In: Martin PS, Klein RG, Eds. Quaternary

extinctions: a prehistoric revolution. Tucson: University of

Arizona Press.

Webb SD, Hulbert RC, Lambert WD. 1995. Climatic implica-

tions of large herbivore distributions in the Miocene of

North America. In: Vrba ES, Denton GH, Partridge TC,

Burckle LH, Eds. Paleoclimate and evolution: with

emphasis on human origins. New Haven (CT): Yale Uni-

versity Press.

Wing SL. 1998. Tertiary vegetation of North America as a con-

text for mammal evolution. In: Janis CM, Scott K, Jacobs L,

Eds. Evolution of tertiary mammals of North America. Cam-

bridge (UK): Cambridge University Press.

Wishart D. 2003. CLUSTAN. Edinburgh (UK): Clustan Limited.

Zolkowich S. 1997. Tundra trouble: snow geese populations

wreak havoc on arctic habitat. Conservator 18(4):4.

22 W. D. Lambert

Appendix 1. Body Mass Data (grams) and Clump Assignments for the Modern Sierra Leone Forest Mammal

Fauna

Species Rank Mass Log mass Clump no

Cercopithecus campbelli 1 3,250 3.51 1

Herpestes ichneumon 2 3,265 3.51 1

Atilax paludinosus 3 3,400 3.53 1

Cercopithecus aethiops 4 3,900 3.59 1

Procolobus verus 5 4,300 3.63 1

Cephalophus monticola 6 4,350 3.64 1

Cercopithecus diana 7 5,000 3.70 1

Cercopithecus petaurista 8 5,500 3.74 1

Procolobus badius 9 7,750 3.89 1

Colobus polykomos 10 8,900 3.95 1

Cercocebus torquatus 11 10,000 4.00 1

Civettictus civetta 12 11,250 4.05 1

Cephalophus zebra 13 12,500 4.10 1

Aonyx capensis 14 13,275 4.12 1

Felis aurata 15 15,750 4.20 1

Cephalophus dorsalis 16 22,000 4.34 1

Tragelaphus scriptus 17 35,000 4.54 2

Panthera pardus 18 45,600 4.66 2

Potamochoerus porcus 19 60,450 4.78 2

Pan troglodytes 20 62,250 4.79 2

Cephalophus sylvicultor 21 62,500 4.80 2

Cephalophus jentinki 22 70,000 4.85 2

Choeropsis libericus 23 210,500 5.32 3

Tragelaphus euryceros 24 277,000 5.44 3

Functional Convergence of Ecosystems 23

Appendix 2. Body Mass Data (grams) and Clump

Assignments for the Modern Serengeti Savanna

Mammal Fauna

Species Rank Mass

Log

mass

Clump

no.

Heterohyrax brucei 1 3,000 3.48 1

Procavia capensis 2 3,300 3.52 1

Herpestes ichneumon 3 3,300 3.52 1

Atilax paludinosus 4 3,400 3.53 1

Pedetes capensis 5 3,500 3.54 1

Cercopithecus aethiops 6 3,900 3.59 1

Ichneumia albicauda 7 4,000 3.60 1

Otocyon megalotis 8 4,200 3.62 1

Cephalophus monticola 9 4,350 3.64 1

Felis lybica 10 4,600 3.66 1

Madoqua kirkii 11 5,100 3.71 1

Cercopithecus mitis 12 5,800 3.76 1

Manis temnickii 13 7,200 3.86 1

Canis mesomelas 14 7,200 3.86 1

Erythrocebus patas 15 8,500 3.93 1

Canis adustus 16 8,800 3.94 1

Proteles cristatus 17 10,000 4.00 1

Felis serval 18 10,400 4.02 1

Raphicerus campestris 19 11,100 4.05 1

Mellivora capensis 20 11,700 4.07 1

Oreotragus oreotragus 21 11,900 4.08 1

Felis caracal 22 12,900 4.11 1

Civettictis civetta 23 13,000 4.11 1

Aonyx capensis 24 13,300 4.12 1

Sylvicapra grimmea 25 14,000 4.15 1

Ourebia ourebi 26 14,100 4.15 1

Hystrix cristata 27 20,000 4.30 2

Gazella thomsoni 28 23,000 4.36 2

Lycaon pictus 29 23,000 4.36 2

Papio ursinus 30 24,000 4.38 2

Redunca fulvorufula 31 29,400 4.47 2

Hyaena hyaena 32 35,000 4.54 2

Redunca redunca 33 45,000 4.65 2

Acinonyx jubatus 34 48,000 4.68 2

Panthera pardus 35 50,000 4.70 2

Tragelaphus scriptus 36 52,000 4.72 2

Orycteropus afer 37 52,000 4.72 2

Gazella granti 38 60,000 4.78 2

Potamochoerus porcus 39 60,000 4.78 2

Crocuta crocuta 40 61,000 4.79 2

Phacochoerus aethiopicus 41 68,000 4.83 2

Damaliscus lunatus 42 133,000 5.12 3

Alcephalus buselaphus 43 136,000 5.13 3

Connochaetes taurinus 44 215,000 5.33 3

Panthera leo 45 238,000 5.38 3

Kobus ellipsiprymnus 46 260,000 5.41 3

Hippotragus equinus 47 270,000 5.43 3

Equus burchelli 48 307,000 5.49 3

Taurotragus oryx 49 372,000 5.57 3

Syncerus caffer 50 775,000 5.89 4

Giraffa camelopardalis 51 1,000,000 6.00 4

Diceros bicornis 52 1,100,000 6.04 4

Hippotamus ampbibius 53 1,400,000 6.15 4

Appendix 3. Body Mass Data (grams) and Clump

Assignments for the Love Mammal Fauna (North

Florida, 9 mya)

Species Rank Mass

Log

(mass)

Clump

no.

Leptarctus cuspidatus 1 7,500 3.88 1

Pseudoceras sp. 2 10,000 4.00 1

Arctonasua floridana 3 26,500 4.42 2

Aelurodon cf. saevus 4 36,000 4.56 2

Aelurodon cf. haydeni 5 40,000 4.60 2

Calippus elachistus 6 49,000 4.69 2

Prosthennops sp. (small) 7 50,000 4.70 2

Antilocapridae

(genus indet.)

8 55,000 4.74 2

Pseudhipparion skinneri 9 61,000 4.79 2

Prosthennops sp. (large) 10 88,000 4.94 2

Calippus cerasinus 11 102,000 5.01 2

‘‘Hemiauchenia’’ minima 12 103,000 5.01 2

Nimravides galiani 13 126,000 5.10 2

Barbourofelis lovei 14 129,000 5.11 2

Cormohipparion ingenuum 15 139,000 5.14 2

Neohipparion trampasense 16 142,000 5.15 2

Hipparion cf. tehonense 17 151,000 5.18 2

Hemiauchenia sp. 18 157,000 5.20 2

Cormohipparion plicatile 19 198,000 5.30 2

Protohippus gidleyi 20 204,000 5.31 2

Pediomeryx hamiltoni 21 211,000 5.32 2

Tapirus simpsoni 22 318,000 5.50 3

Teloceras proterum 23 475,000 5.68 3

Procamelus grandis 24 491,000 5.69 3

Aphelops malacorhinus 25 935,000 5.97 4

Aepycamelus major 26 950,000 5.98 4

24 W. D. Lambert

Appendix 4. Body Mass Data (grams) and Clump

Assignments for the Combined McGehee and Mi-

xon’s Mammal Faunas (North Florida, 8 mya)

Species Rank Mass

Log

(mass)

Clump

no.

Procyonidae (genus indet.) 1 32,000 4.51 1

Osteoborus cf. galushai 2 34,000 4.53 1

Calippus regulus 3 42,000 4.62 1

Pseudhipparion sp. 4 60,000 4.78 1

Astrohippus martini 5 73,000 4.86 1

Nannippus aztecus 6 79,000 4.90 1

Nannippus westoni 7 93,000 4.97 1

‘‘Hemiauchenia’’ minima 8 103,000 5.01 1

Nimravides galiani 9 126,000 5.10 1

Cormohipparion ingenuum 10 139,000 5.14 1

Pediomeryx hemphillensis 11 150,000 5.18 1

Hemiauchenia sp. 12 156,000 5.19 1

Pliohippus cf. supremus 13 186,000 5.27 1

Synthetoceras tricornatus 14 188,000 5.27 1

Cormohipparion plicatile 15 198,000 5.30 1

Tapirus simpsoni 16 318,000 5.50 1

Teleoceras sp. 17 710,000 5.85 2

Pliometanastes protistus 18 851,000 5.93 2

Aphelops malacorhinus 19 935,000 5.97 2

Thinobadistes segnis 20 943,000 5.97 2

Aepycamelus major 21 950,000 5.98 2

Appendix 5. Body Mass Data (grams) and Clump

Assignments for the Combined Blue Jay and North

Shore Mammal Faunas (Nebraska, 9 mya)

Species Rank Mass

Log

mass

Clump

no.

Leptocyon sp. 1 6,000 3.78 1

Longirostromeryx cf. merriami 2 20,000 4.30 2

Aelurodon saevus 3 36,000 4.56 2

Aelurodon haydeni 4 40,000 4.60 2

Barbourofelis morrisi 5 87,000 4.94 3

Pseudhipparion gratum 6 105,000 5.02 3

Parahippus cognatus 7 113,000 5.05 3

Nimravides galiani 8 126,000 5.10 3

Hemiauchenia sp. 9 133,000 5.12 3

Protohippus sp. 10 172,000 5.24 3

Cranioceras cf. unicornis 11 179,000 5.25 3

Hypohippus cf. affinis 12 213,000 5.33 3

Ustatochoerus major 13 223,000 5.35 3

Neohipparion occidentale 14 230,000 5.36 3

Pliauchenia magnifontis 15 237,000 5.37 3

Miolabis sp. 16 247,000 5.39 3

Dinohippus sp. 17 419,000 5.62 4

Teleoceras sp. 18 423,000 5.63 4

Procamelus grandis 19 491,000 5.69 4

Megatylopus primaevus 20 653,000 5.81 5

Aphelops sp. 21 935,000 5.97 5

Appendix 6. Body Mass Data (grams) and Clump

Assignments for the Late Barstovian Valentine

Railroad Quarry Mammal Fauna (Nebraska, 12

mya)

Species Rank Mass

Log

mass

Clump

no.

ruminant, genus indet. 1 3,600 3.56 1

Leptocyon vafer 2 4,800 3.68 1

Pseudoparablastomeryx

francesita

3 5,400 3.73 1

Submeryceros minor 4 7,500 3.88 1

Pseudaelurus marshi 5 8,000 3.90 1

Tomarctos cf. euthos 6 10,000 4.00 1

Cynarctos saxatilis 7 11,000 4.04 1

Tomarctos temerius 8 11,700 4.07 1

Ursavus sp. 9 13,000 4.11 1

Longirostromeryx blicki 10 14,000 4.15 1

Arctonasua sp. 11 15,500 4.19 1

Carpocyon cuspidatus 12 16,000 4.20 1

Mionictis sp. 13 18,000 4.26 1

Merycodus warreni 14 18,000 4.26 1

Parablastomeryx sp. 15 21,000 4.32 1

Aelurodon ferox 16 25,000 4.40 1

Pseudaelurus intrepidus 17 34,000 4.53 1

Dyseohyus xiphodonticus 18 58,000 4.76 1

Cranioceras skinneri 19 105,000 5.02 2

Parahippus cognatus 20 113,000 5.05 2

Merychippus cf. insignus 21 120,000 5.08 2

Calippus placidus 22 130,000 5.11 2

Ustatochoerus medius 23 180,000 5.26 2

Neohipparion republicanus 24 193,000 5.29 2

Pliohippus mirabilis 25 202,000 5.31 2

Tapiravis cf. polkensis 26 214,000 5.33 2

Anchitherium sp. 27 228,000 5.36 2

Protolabis cf. gracilis 28 268,000 5.43 2

Megahippus mckennai 29 305,000 5.48 2

Aepycamelus robustus 30 347,000 5.54 2

Procamelus robustus 31 372,000 5.57 2

Teleoceras cf. mediacornatus 32 732,000 5.86 3

Peraceras cf. crassus 33 1,951,000 6.29 4

Functional Convergence of Ecosystems 25

Appendix 7. Body Mass Data (grams) and Clump

Assignments for the Combined Moss Acres Race-

track and Withlacoochee River 4A Mammal Faunas

(North Florida, 7 mya)

Species Rank Mass

Log

mass

Clump

no.

Vulpes stenognathus 1 6,000 3.78 1

cf. Pseudoceras 2 9,050 3.96 1

Osteoborus orc 3 18,130 4.26 2

Enhydritherium terraenovae 4 21,060 4.32 2

Nannippus morgani 5 36,000 4.56 2

Calippus mccartyi 6 73,000 4.86 3

Nannippus minor 7 79,000 4.90 3

Pediomeryx hemphillensis 8 108,000 5.03 3

Hipparion cf. tehonense 9 151,000 5.18 4

Machairodus sp. 10 165,000 5.22 4

Neohipparion eurystyle 11 166,000 5.22 4

Cormohipparion emslei 12 186,000 5.27 4

Cormohipparion plicatile 13 198,000 5.30 4

Astrohippus 14 225,000 5.35 4

Tapirus simpsoni 15 318,000 5.50 4

Dinohippus sp. 16 330,000 5.52 4

Indarctos sp. 17 363,000 5.56 4

Aepycamelus sp. 18 618,000 5.79 5

Aphelops mutilis 19 764,000 5.88 5

Pliometanastes protistus 20 851,000 5.93 5

Thinobadistes segnis 21 948,000 5.98 5

Appendix 8. Body Mass Data (grams) and Clump

Assignments for the Cambridge Mammal Fauna

(Nebraska, 7 mya)

Species Rank Mass

Log

mass

Clump

no.

Vulpes stenognathus 1 6,000 3.78 1

Sthenictis sp. 2 10,600 4.03 1

Pseudoceras sp. 3 11,100 4.05 1

Texoceros cf. guymonensis 4 37,000 4.57 2

Epicyon validus 5 39,000 4.59 2

Prosthenops graffamhami 6 43,000 4.63 2

Calippus sp. 7 77,000 4.89 3

Nannippus sp. 8 97,000 4.99 3

Calippus theristes 9 132,000 5.12 3

Hemiauchenia vera 10 159,000 5.20 3

Neohipparion eurystyle 11 166,000 5.22 3

Hipparion cf. forcei 12 174,000 5.24 3

Pliohippus nobilis 13 201,000 5.30 3

Barbourofelis fricki 14 242,000 5.38 3

Dinohippus sp. 15 259,000 5.41 3

Tapirus simpsoni 16 318,000 5.50 3

Pediomeryx figginsi 17 414,000 5.62 3

Alforjas taylori 18 524,000 5.72 3

Pliometanastes protistus 19 851,000 5.93 4

Agriotherium schneideri 20 986,000 5.99 4

Thinobadistes sp. 21 1,066,000 6.03 4

Teleoceras fossiger 22 1,139,000 6.06 4

Megatylopus sp. 23 1,259,000 6.10 4

Aphelops kimballensis 24 1,792,000 6.25 4

26 W. D. Lambert

Appendix 9. Body Mass Data (grams) and Clump

Assignments for the Palmetto Mammal Fauna

(Central Florida, 5 mya)

Species Rank Mass

Log

mass

Clump

no.

Mephitine (genus indet.) 1 1,590 3.20 1

cf. Bassariscus 2 3,200 3.51 2

Vulpes stenognathus 3 6,000 3.78 2

Euocyon davisi 4 9,000 3.95 2

cf. Procyon 5 10,000 4.00 2

Subantilocapra garciae 6 12,000 4.08 2

Leptarctus cf. progressus 7 12,200 4.09 2

Enhydritherium terraenovae 8 21,000 4.32 3

Felis rexroadensis 9 32,000 4.51 3

Eocoileus gentryorum 10 36,000 4.56 3

Borophagus dudleyi 11 37,000 4.57 3

Arctonasua eurybates 12 37,000 4.57 3

Megantereon hesperus 13 45,000 4.65 3

Hexameryx simpsoni 14 48,000 4.68 3

Pseudhipparion simpsoni 15 53,000 4.72 3

lamine, small 16 76,000 4.88 4

Nannippus aztecus 17 79,000 4.90 4

Astrohippus stocki 18 85,000 4.93 4

Machairodus sp. 19 150,000 5.18 5

Neohipparion eurystyle 20 166,000 5.22 5

Cormohipparion emslei 21 186,000 5.27 5

Mylohyus elmorei 22 194,000 5.29 5

Hemiauchenia sp. 23 204,000 5.31 5

Platygonus sp. 24 240,000 5.38 5

Dinohippus mexicanus 25 250,000 5.40 5

Tapirus simpsoni 26 318,000 5.50 5

Megalonyx curvidens 27 319,000 5.50 5

Kyptoceras amatorum 28 448,000 5.65 5

Teleoceras sp. 29 963,000 5.98 6

Agriotherium schneideri 30 985,000 5.99 6

Megatylopus sp. 31 1,336,000 6.13 6

Appendix 10. Body Mass Data (grams) and Clump

Assignments for the Combined Devil’s Nest and

Santee Mammal Faunas (Nebraska, 5 mya)

Species Rank Mass

Log

mass

Clump

no.

Trigonictis sp. 1 3,200 3.51 1

Vulpes stenognathus 2 6,000 3.78 2

Pliotaxidea garberi 3 6,600 3.82 2

Euocyon davisi 4 9,000 3.95 2

Lutra sp. 5 14,000 4.15 2

Canis cf. condoni 6 15,000 4.18 2

Felis rexroadensis 7 30,000 4.48 3

Eocoileus gentryorum 8 36,000 4.56 3

Antilocaprid, genus indet. 9 48,000 4.68 3

Nannippus lenticularis 10 98,000 4.99 4

Machairodus sp. 11 134,000 5.13 4

Pediomeryx hemphillensis 12 150,000 5.18 4

Hemiauchenia vera 13 159,000 5.20 4

Neohipparion eurystyle 14 166,000 5.22 4

Dinohippus sp. 15 239,000 5.38 4

Platygonus sp. 16 240,000 5.38 4

Tapirus simpsoni 17 318,000 5.50 4

Megalonyx curvidens 18 319,000 5.50 4

Aphelops mutilis 19 764,000 5.88 5

Agriotherium schneideri 20 986,000 5.99 5

Teleoceras fossiger 21 1,139,000 6.06 5

Megacamelus cf. merriami 22 1,479,000 6.17 5

Functional Convergence of Ecosystems 27

Appendix 11. Body Mass Data (grams) and Clump

Assignments for the Ricardo Mammal Fauna

(Southern California, 9 mya)

Species Rank Mass

Log

mass

Clump

no.

Leptocyon sp. 1 7,000 3.85 1

Borophagus diabolensis 2 9,000 3.95 1

Cosoryx sp. 3 16,000 4.20 1

Merycodus sp. 4 21,000 4.32 1

Sphenophalos sp. 5 28,000 4.45 1

Aelurodon saevus 6 36,000 4.56 1

Aelurodon haydeni 7 40,000 4.60 1

Hipparion tehonense 8 151,000 5.18 2

felid, genus indet. 9 160,000 5.20 2

Pliauchenia sp. 10 202,000 5.31 2

Hadracyon mohavense 11 208,000 5.32 2

Cormohipparion occidentale 12 230,000 5.36 2

Hipparion mohavense 13 233,000 5.37 2

Ticholeptus undescribed sp. 14 243,000 5.39 2

Hipparion forcei 15 265,000 5.42 2

Procamelus coartatus 16 270,000 5.43 2

Pliohippus tantalus 17 468,000 5.67 3

Ticholeptus major 18 511,000 5.71 3

Neohipparion sp. 19 556,000 5.75 3

Paracamelus sp. 20 608,000 5.78 3

Ustatochoerus californicus 21 968,000 5.99 4

Megatylopus sp. 22 1,034,000 6.01 4

Appendix 12. Body Mass Data (grams) and Clump

Assignments for the Combined Merhten, Peace

Valley, and Mount Eden Mammal Faunas (Central

California, 6 mya)

Species Rank Mass

Log

mass

Clump

no.

Martes sp. 1 3,000 3.48 1

Euocyon davisi 2 8,000 3.90 2

Ottoceras peacevalleyensis 3 15,000 4.18 2

Borophagus parvus 4 20,000 4.30 2

Sphenophalos sp. 5 23,000 4.36 2

Plesiogulo marshalli 6 61,000 4.79 3

Prosthenops sp. 7 77,000 4.89 3

Pediomeryx hemphillensis 8 108,000 5.03 3

Hemiauchenia vera 9 159,000 5.20 3

Neohipparion sp. B 10 246,000 5.39 4

Neohipparion cf. gidleyi 11 306,000 5.49 4

Dinohippus interpolatus 12 320,000 5.51 4

Neohipparion sp. A 13 401,000 5.60 4

Pliohippus osborni 14 535,000 5.73 4

Megatylopus sp. 15 710,000 5.85 4

Aphelops sp. 17 1,951,000 6.29 5

Titanotylopus nebraskensis 18 1,964,000 6.29 5

Appendix 13. Body Mass Data (grams) and Clump

Assignments for the Combined Schutler and Rome

Mammal Faunas (Western Oregon, 5 mya)

Species Rank Mass

Log

mass

Clump

no.

Martes sp. 1 3,000 3.48 1

Bassariscus sp. 2 3,000 3.48 1

Euocyon davisi 3 8,000 3.90 1

Aelurodon sp. 4 12,000 4.08 1

Borophagus pugnator 5 18,000 4.26 1

Satherium sp. 6 24,000 4.38 1

Sphenophalos sp. 7 34,000 4.53 1

Eocoileus gentryorum 8 36,000 4.56 1

Prosthenops sp. 9 82,000 4.91 2

Machairodus sp. 10 150,000 5.18 3

Nannippus sp. 11 154,000 5.19 3

Hipparion sp. 12 167,000 5.22 3

camelid, genus indet

(small)

13 189,000 5.28 3

Neohipparion sp. 14 240,000 5.38 3

Megalonyx cf. curvidens 15 319,000 5.50 3

camelid, genus indet

(large)

16 529,000 5.72 4

equid, genus indet

(large)

17 554,000 5.74 4

Camelid, genus indet

(large)

18 578,000 5.76 4

Agriotherium schneideri 19 985,000 5.99 5

Teleoceras sp. (small) 20 1,078,000 6.03 5

28 W. D. Lambert

Isotopic evidence for mammalian diets and environment in Early Pliocene Yepómera, Mexico

Article

Jul 2021
PALAEOGEOGR PALAEOCL

Concurrent expansion of grassland habitat in the Americas and first appearances of immigrant taxa to Mexico via the Great American Biotic Interchange (GABI) during the Late Miocene to Early Pliocene suggest a possible link between the timing of migrations and changes in the environmental conditions. However, the causes and environmental context of GABI migrations are not well understood. This study examined the stable isotope compositions of tooth enamel samples of the Yepómera fauna from western Chihuahua, Mexico, to explore environmental conditions of this region between 4.99 and 5.23 Ma, just after the arrival of C4 plants in North and South America and before the first major migration of GABI. At Yepómera, there was distinct niche partitioning into C3 diets, mixed diets, and C4 diets. Despite expectations, no niche partitioning between equid species (Dinohippus mexicanus, Nannippus aztecus, Astrohippus stockii, and Neohipparion eurystyle) can be determined from carbon isotope ratios. The enamel carbon and oxygen isotope data suggest a relatively dry, open habitat dominated by either savanna or grassland, with a substantial C4 vegetation component and a warmer and somewhat wetter climate than today. These reconstructions are consistent with a rise in C4 biomass before 5.23 Ma and suggest that the conditions needed for growth of C4 vegetation were prevalent in this region of Mexico. Future work along the GABI migration route will lead to a more complete understanding of the ecologic responses to changing climate and faunal interchange events.

MEGAHERBIVORE FOSSIL RECORD FROM THE NEOGENE OF JALISCO, MEXICO, AND ITS IMPLICATIONS FOR PALEOENVIRONMENTAL INTERPRETATION

Article

Full-text available

Aug 2022

Pliocen megaherbivores from Jalisco, Mexico

Functional convergence of biosphere–atmosphere interactions in response to meteorological conditions

Article

Full-text available

Apr 2021
BG

Understanding the dependencies of the terrestrial carbon and water cycle with meteorological conditions is a prerequisite to anticipate their behaviour under climate change conditions. However, terrestrial ecosystems and the atmosphere interact via a multitude of variables across temporal and spatial scales. Additionally these interactions might differ among vegetation types or climatic regions. Today, novel algorithms aim to disentangle the causal structure behind such interactions from empirical data. The estimated causal structures can be interpreted as networks, where nodes represent relevant meteorological variables or land-surface fluxes and the links represent the dependencies among them (possibly including time lags and link strength). Here we derived causal networks for different seasons at 119 eddy covariance flux tower observations in the FLUXNET network. We show that the networks of biosphere–atmosphere interactions are strongly shaped by meteorological conditions. For example, we find that temperate and high-latitude ecosystems during peak productivity exhibit biosphere–atmosphere interaction networks very similar to tropical forests. In times of anomalous conditions like droughts though, both ecosystems behave more like typical Mediterranean ecosystems during their dry season. Our results demonstrate that ecosystems from different climate zones or vegetation types have similar biosphere–atmosphere interactions if their meteorological conditions are similar. We anticipate our analysis to foster the use of network approaches, as they allow for a more comprehensive understanding of the state of ecosystem functioning. Long-term or even irreversible changes in network structure are rare and thus can be indicators of fundamental functional ecosystem shifts.

Phytoplankton size- and abundance-based resilience assessments reveal nutrient rather than water level effects

Article

Jul 2020
SCI TOTAL ENVIRON

The use of discontinuity analysis to assess resilience and alternative regimes of ecosystems has mostly been based on animal size. We so far lack systematic comparisons of size-based and abundance-based approaches necessary for assessing the performance and suitability of the discontinuity analysis across a broader range of organism groups. We used an outdoor mesocosm setup to mimic shallow lake ecosystems with different depths (1.2 m deep, “shallow”; 2.2 m deep, “deep”) and trophic status (i.e. low and high nutrient status characteristic of mesotrophic and hypertrophic lakes, respectively). We compared resilience assessments, based on four indicators (cross-scale structure, within-scale structure, aggregation length and gap size) inferred from the size and abundance (biovolume) structure of phytoplankton communities. Our results indicate that resilience assessments based on size and biovolume were largely comparable, which is likely related to similar variability in the size and abundance of phytoplankton as a function of nutrient concentrations. Also, nutrient enrichment rather than water depth influenced resilience, manifested in decreased cross-scale structure and increased aggregation lengths and gap sizes in the high-nutrient treatment. These resilience patterns coupled with decreased phytoplankton diversity and dominance of cyanobacteria in the high nutrient treatment support the use of discontinuity analysis for testing alternative regimes theory. Concordance of size-based and abundance-based results highlights the approach as being potentially robust to infer resilience in organism groups lacking discrete size structures.

Resilience of natural lake phytoplankton communities to pulse disturbances from micropollutant exposure and vertical mixing

Article

Full-text available

Jul 2019

Freshwaters are increasingly exposed to complex mixtures of pharmaceutical and personal care products (PPCPs) from municipal wastewater. PPCPs are known to alter freshwater communities’ structure and functioning, however, their interaction with other disturbances and whether their combined effects can impact ecological resilience (i.e. the ability of a system to tolerate disturbances without altering the system's original structure and processes) remain unexplored. Using in situ mesocosms in two lakes with different nutrient levels (mesotrophic and eutrophic), we assessed whether a pulse exposure to sub‐lethal concentrations of 12 PPCPs affects the ecological resilience of natural phytoplankton communities that experienced an abrupt environmental change involving the destabilization of the water column through mixing. Such mixing events are predicted to increase as the effects’ climate change unfold, leading to more frequent storms, which disrupt stratification in lakes and force communities to restructure. We assessed their combined effects on community metrics (biomass, species richness and composition) and their relative resilience using four indicators (cross‐scale, within scale, aggregation length and gap length), inferred from phytoplankton communities by discontinuity analysis. The mixing disturbance alone had negligible effects on the community metrics, but when combined with chemical contaminants significant changes were measured: reducing total biomass, species richness and altered community composition of phytoplankton. Once these changes occurred, they persisted until the end of the experiment (day 20) where the communities’ structures from the two highest exposure levels diverged from the controls. The resilience indicators were not affected by PPCPs, but differed significantly between lakes, with lower resilience found in the eutrophic lake. PPCPs can significantly alter community structures and reinforce mechanisms that maintain ecosystems in a “degraded state”. This article is protected by copyright. All rights reserved.

Correction: Body-size structure of Central Iberian mammal fauna reveals semidesertic conditions during the middle Miocene Global Cooling Event

Article

Full-text available

Aug 2018
PLOS ONE

[This corrects the article DOI: 10.1371/journal.pone.0186762.].

Observations on graphing paleozoological data: Suggestions for better graphs

Article

Aug 2018

Paleozoologists have long used graphs of diverse styles to describe, analyze, and summarize their data. Some of these graphs provide excellent visual representations of complex data and are readily deciphered. Other graph styles require close study to be interpreted. Ease of visual decoding of information contained in a graph – graph perception – varies from graph style to graph style. Historical instances of graphing paleozoological data indicate some difficult to decipher graph styles have been used for at least a century. Graphs with three-dimensions, moiré effects, superimposed lines, or segmented bars, or which demand simultaneous decipherment of position and magnitude, are ill-advised. Temporal trends in data are best graphed following the principle of superposition such that data from old material is graphed at the bottom and data from younger material is graphed at the top of the diagram.

Complex Adaptive Systems: Cross-Scale Structure and Resilience

Thesis

Jan 2018

Shana M Sundstrom

This dissertation is focused on scaling and resilience of complex adaptive systems, including ecological and economic systems. In particular it is concerned with the textural discontinuity hypothesis (hereafter called the discontinuity hypothesis), which describes how the distinct spatial and temporal scales of processes that shape systems in turn generates distinct spatial and temporal scales in system structure and entities interacting with that structure; the cross-scale resilience model, which uses the discontinuity hypothesis as the foundation of a theory about specific system features that drive ecological resilience; panarchy and adaptive cycles, which articulate how system dynamics at the above-mentioned scales change over time and how feedbacks across those scales informs system behavior; and the notion of spatial regimes in ecological structure. I both expand existing frameworks to accommodate non-ecological complex systems, and test my hypotheses in a variety of economic and ecological systems. ^ Some general findings of my analyses are that the objective identification of scale domains in many types of complex systems can be useful for understanding how pattern and process shape structure and impact system-level resilience. Economic systems, for example, as expressed by Gross Domestic Product, fall into distinct, non-random size classes that suggest there are scale-specific processes generating basins of attraction. I expand the cross-scale resilience model to incorporate abundance, a species and community attribute that is mechanistically related to the provision of function and resilience. The coral reef fish communities of the Hawaiian archipelago were analysed to see if their cross-scale resilience differed amongst coral dominated and macroalgal and turf dominated reefs, with the surprising result that the macroalgal-turf communities were more resilient. In a twist on classic regime shift theory, which typically focuses on temporal shifts within a single ecosystem, I used a novel information theory method to successfully detect spatial boundaries and transition zones between types of ecological systems by using animal community data. Finally, I argue why the adaptive cycle may be a result of endogenous processes in complex adaptive systems, and is not just a convenient metaphor for cycling behavior and dynamics.

Middle Miocene paleoenvironmental reconstruction in the central Great Plains, USA, from stable carbon isotopes in ungulates

Article

Mar 2022
PALAEOGEOGR PALAEOCL

Middle Miocene (~18–12 Ma) mammalian faunas of the North American Great Plains contained a higher diversity of presumed browsers than any modern biome. This has been attributed to greater primary productivity, which commonly corresponds with densely vegetated, forested habitats today. However, several lines of evidence suggest that open biomes, which support low browser diversity today, dominated the middle Miocene. To gain a better understanding of middle Miocene habitats on the northern Great Plains, we generated a large dataset of stable carbon isotope values (δ¹³C) from a wide variety of fossil ungulates. Tooth enamel was sampled from four local faunas in Nebraska of late Barstovian age (14.8–12.5 Ma). Paleoenvironmental interpretations were made using a predictive model based on δ¹³C values in vegetation and mammalian herbivores in modern biomes. Resulting mean δ¹³C fossil values for local faunas plot in the upper range expected for C3 vegetation, indicating open biomes. Most mean values for genera plot in the range expected for open canopy habitats with only two genera in the closed canopy range. Of all the taxa, only the grazing horse, Neohipparion, plots in the range for likely C4 consumption. Neohipparion has the highest mean δ¹³C value of any taxon, with one individual plotting in the range for unequivocal C4 consumption and others in the mixed C4/C3 or C3 water-stressed range. Neohipparion appears to be the earliest known taxon that included C4 vegetation in its dietary niche. The range in carbon isotope values suggests browsers, mixed-feeders, and grazers all consumed vegetation in mostly open areas. Our model implies a fairly dry, open biome with limited riparian habitat. This interpretation is generally consistent with previous interpretations of an open savanna-like biome. A high diversity of middle Miocene browsers and other taxa appear to have lived in an open biome with no modern analog.

Paleozoology and Paleoenvironments: Fundamentals, Assumptions, Techniques

Book

Feb 2019

Paleozoology and Paleoenvironments outlines the reconstruction of ancient climates, floras, and habitats on the basis of animal fossil remains recovered from archaeological and paleontological sites. In addition to outlining the ecological fundamentals and analytical assumptions attending such analyzes, J. Tyler Faith and R. Lee Lyman describe and critically evaluate many of the varied analytical techniques that have been applied to paleozoological remains for the purpose of paleoenvironmental reconstruction. These techniques range from analyses based on the presence or abundance of species in a fossil assemblage to those based on taxon-free ecological characterizations. All techniques are illustrated using faunal data from archaeological or paleontological contexts. Aimed at students and professionals, this volume will serve as fundamental resource for courses in zooarchaeology, paleontology, and paleoecology.

Walker's Mammals of The World

Article

Full-text available

Feb 1984

Cosmic Evolution: The Rise of Complexity in Nature

Book

Full-text available

Feb 2001

Eric Chaisson

Using astronomical telescopes and biological microscopes, among a virtual arsenal of other tools of high technology, modern scientists are weaving a thread of understanding spanning the origin, existence, and destiny of all things. Now emerging is a unified scenario of the cosmos, including ourselves as sentient beings, based on the time-honored concept of change. From galaxies to snowflakes, from stars and planets to life itself, we are beginning to identify an underlying, ubiquitous pattern penetrating the fabric of all the natural sciences—a sweepingly encompassing view of the order and structure of every known class of object in our richly endowed Universe. We call this subject "cosmic evolution." Recent advances throughout the sciences suggest that all organized systems share generic phenomena characterizing their emergence, development and evolution. Whether they are physical, biological or cultural systems, certain similarities and homologies pervade evolving entities throughout an amazingly diverse Universe. How strong are the apparent continuities among Nature's historical epochs and how realistic is the quest for unification? To what extent might we broaden conventional evolutionary thinking, into both the pre-biological and post-biological domains? Is such an extension valid, merely metaphorical, or just plain confusing? For many years at Harvard University, starting in the 1970s and continuing to the present, I have taught, initially with George B. Field, an introductory course on cosmic evolution that sought to identify common denominators bridging a wide variety of specialized science subjects—physics, astronomy, geology, chemistry, biology, and anthropology, among others. The principal aim of this interdisciplinary course explored a universal framework against which to address some of the most basic issues ever contemplated: the origin of matter and the origin of life, as well as how radiation, matter, and life interact and change with time. Our intention was to help sketch a grand evolutionary synthesis that would better enable us to understand who we are, whence we came, and how we fit into the overall scheme of things. In doing so, my students and I gained a broader, integrated knowledge of stars and galaxies, plants and animals, air, land, and sea. Of paramount import, we learned how the evident order and increasing complexity of the many varied, localized structures within the Universe in no way violate the principles of modern physics, which, prima facie, maintain that the Universe itself, globally and necessarily, becomes irreversibly and increasingly disordered. Beginning in the late 1980s while on sabbatical leave at MIT, and continuing for several years thereafter while on the faculty of the Space Telescope Science Institute at Johns Hopkins University, I occasionally offered an advanced version of the introductory course. This senior seminar attempted to raise substantially the quantitative aspects of the earlier course, to develop even deeper insights into the nature and role of change in Nature, and thus to elevate the subject of cosmic evolution to a level that colleague scientists and intelligent lay persons alike might better appreciate. This brief and broadly brushed monograph—written mostly in the late 1990s during a stint as Phi Beta Kappa National Lecturer, and polished while resuming the teaching at Harvard of my original course on cosmic evolution--is an intentionally lean synopsis of the salient features of that more advanced effort. Some will see this work as reductionistic, with its analytical approach to the understanding of all material things. Others will regard it as holistic, with its overarching theme of the whole exceeding the sum of Nature's many fragmented parts. In the spirit of complementarity, I offer this work as an evolutionary synthesis of both these methodologies, integrating the deconstructionism of the former and the constructivist tendencies of the latter. Openly admitted, my inspiration for writing this book has been Erwin Schroedinger's seminal little tract of a half-century ago, What is Life?, yet herein to straighten and extend the analysis to include all known manifestations of order and complexity in the Universe. No attempt is made to be comprehensive in so far as details are concerned; much meat has been left off the bones. Nor is this work meant to be technically rigorous; that will be addressed in a forthcoming opus. Rather, the intent here is to articulate a skeletal précis—a lengthy essay, really—of a truly voluminous subject in a distilled and readable manner. To bend a hackneyed cliché, although the individual trees are most assuredly an integral part of the forest, in this particular work the forest is of greater import. My aim is to avoid diverting the reader from the main lines of argument, to stay focused on target regarding the grand sweep of change from big bang to humankind. Of special note, this is not a New Age book with mystical overtones however embraced or vulgarized by past scholars, nor one about the history and philosophy of antiquated views of Nature. It grants no speculation on the pseudo-science fringe about morphic fields or quantum vitalism or interfering dieties all mysteriously affecting the ways and means of evolution; nor do we entertain epistemological discussions about the limits of human knowledge or post-modernist opinions about the sociological implications of science writ large. This is a book about mainstream science, pure and simple, outlining the essence of an ongoing research program admittedly multidisciplinary in character and colored by the modern scientific method's unavoidable mix of short-term subjectivity and long-term objectivity. In writing this book, I have assumed an undergraduate knowledge of natural science, especially statistical and deterministic physics, since as we shall see, much as for classical biological evolution, both chance and necessity have roles to play in all evolving systems. The mathematical level includes that of integral calculus and differential equations, with a smattering of symbolism throughout; the units are those of the centimeter-gram-second (cgs) system, those most widely used by practitioners in the field, editorial conventions notwithstanding. And although a degree of pedagogy has been included when these prerequisites are exceeded, some scientific language has been assumed. "The book of Nature is written in the language of mathematics," said one of my two intellectual heroes, Galileo Galilei, and so are parts of this one. Readers with unalterable math phobia will benefit from the unorthodox design of this work, wherein the "bookends" of Prologue-Introduction and Discussion-Epilogue, comprising more than half of the book, can be mastered without encountering much mathematics at all. What is presented here, then, is merely a sketch of a developing research agenda, itself evolving, ordering and complexifying—an abstract of scholarship-in-progress incorporating much data and many ideas from the entire spectrum of natural science, yet which attempts to surpass scientific popularizations (including some of my own) that avoid technical lingo, most numbers, and all mathematics. As such, this book should be of interest to most thinking people—active researchers receptive to an uncommonly broad view of science, sagacious students of many disciplines within and beyond science, the erudite public in search of themselves and a credible worldview—in short, anyone having a panoramic, persistent curiosity about the nature of the Universe and of our existence in it. -- Summary Abstract of This Work -- The essence of this book outlines the grand scenario of cosmic evolution by qualitatively and quantitatively examining the natural changes among radiation, matter, and life within the context of big-bang cosmology. The early Universe is shown to have been flooded with pure energy whose radiation energy density was initially so high as to preclude the existence of any appreciable structure. As the Universe cooled and thinned, a preeminent phase change occurred a few hundred centuries after the origin of all things, at which time matter's energy density overthrew the earlier primacy of radiation. Only with the onset of technologically manipulative beings (on Earth and perhaps elsewhere) has the energy density contained within matter become, in turn, locally dominated by the rate of free energy density flowing through open organic structures. Using non-equilibrium thermodynamics at the crux, especially energy flow considerations, we argue that it is the contrasting temporal behavior of various energy densities that have given rise to the environments needed for the emergence of galaxies, stars, planets, and life forms. We furthermore maintain that a necessary (though perhaps not sufficient) condition—a veritable prime mover—for the emergence of such ordered structures of rising complexity is the expansion of the Universe itself. Neither demonstrably new science nor appeals to non-science are needed to explain the impressive hierarchy of the cosmic-evolutionary scenario, from quark to quasar, from microbe to mind.

From Clocks to Chaos: The Rhythms of Life

Book

Jan 1988

Integration of Ecosystem Theories: A Pattern

Article

Jan 1992

Sven Erik Jørgensen

Ecosystems are still a puzzle for mankind. We would like to be able to know their reactions and control them, but repeatedly we have been surprised by their unexpected reactions to our somewhat hasty actions. We unfortunately have to admit that our present knowledge about ecosystems and their true nature is rather limited. Many excellent contributions to a more profound understanding of ecosystems have been launched during the last two decades, but if you do not know the field, it looks as if all the presented ecosystem theories are in complete discord with each other. However, ecosystems are extremely complex and only a pluralistic view will be able to reveal their basic properties. The different approaches therefore have much in common, when you go deeper into the core material, than the first superficial more glance will be able to tell and there is therefore a natural need for a unification of the various approaches to ecosystem theories. It has for many years been my desire to attempt to make a unification of the many excellent thoughts, ideas and observations about ecosystems, that scientists have contributed. These thoughts, ideas and hypotheses have not been made in vain.

Serengeti II: Dynamics, Management and Conservation of an Ecosystem.

Article

Aug 1996

The aim of this text is to provide an up-to-date understanding of the Serengeti-Mara ecosystem in East Africa, home to one of the largest and most diverse populations of animals in the world. Building on the groundwork laid by "Serengeti: Dynamics of an Ecosystem", published in 1979 by the University of Chicago Press, this work integrates studies of the ecosystem at every level, from the plants at the bottom of the visible food chain to the many species of herbivores and predators, as well as the system as a whole. Drawing on data from long-term studies, the contributors examine the processes that have produced the Serengeti's biological diversity, with its species-species and species-environment interactions. The book also discusses computer modelling as a tool for exploring these interactions, employing it to test and anticipate the effects of social, political and economic changes on the entire ecosystem and on particular species, with the aim of assisting the development of future conservation and management strategies.

Forest Ecosystems.

Article

Jun 1996

Fauna of the Serengeti National Park

Article

Jan 1958
MAMMALIA

G. SWYNNERTON

Ten million years of mammal extinctions in North America.

Article

Jan 1984

S.D. Webb

Examines Pleistocene extinctions in the light of a longer perspective gained by examining the last 10 Myr. During this period six major extinction episodes have devastated the land mammal fauna of North America. Some evidence shows that these extinction episodes were correlated with terminations of glacial cycles, when climatic extremes and instability are thought to have reached their maxima. -after Author

Walker ´ s Mammals of the World

Article

Jan 1999

Ronald M Nowak

Barriers & Bridges to the Renewal of Ecosystems and Institutions

Article

Oct 1997

Functional Convergence of Ecosystems: Evidence from Body Mass Distributions of North American Late Miocene Mammal Faunas

Abstract and Figures

Recommended publications

Do Ecosystems Ever Converge? Evidence From Faunal Size Distributions Of Late Miocene North American...

Terrestrial mammal remains from the Miocene Chesapeake Group of Calvert Cliffs, Maryland, and compar...

Original Articles: Causes of Ecosystem Transformation at the End of the Pleistocene: Evidence from M...

Implications of discoveries of the shovel-tusked gomphothere Konobelodon (Proboscidea, Gomphotheriid...