Content uploaded by Clive Gareth Jones
Author content
All content in this area was uploaded by Clive Gareth Jones
Content may be subject to copyright.
http://www.jstor.org
Positive and Negative Effects of Organisms as Physical Ecosystem Engineers
Author(s): Clive G. Jones, John H. Lawton, Moshe Shachak
Source:
Ecology,
Vol. 78, No. 7, (Oct., 1997), pp. 1946-1957
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/2265935
Accessed: 02/06/2008 11:23
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=esa.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We enable the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
1946 SPECIAL FEATURE Ecology
Vol. 78, No. 7
Ecology, 78(7), 1997, pp. 1946-1957
? 1997 by the Ecological Society of America
POSITIVE AND NEGATIVE EFFECTS OF ORGANISMS AS
PHYSICAL ECOSYSTEM ENGINEERS
CLIVE G. JONES,' JOHN H. LAWTON,
12 AND MOSHE SHACHAK"3
'Institute of Ecosystem Studies, Box AB, Millbrook, New York 12545 USA
2Centre for Population Biology, Imperial College, Silwood Park, SL5 7PY, UK
3Mitrani Center for Desert Ecology, Blaustein Institute for Desert Research,
Ben-Gurion University of the Negev, Sede Boqer 84900, Israel
Abstract. Physical ecosystem engineers are organisms that directly or indirectly control
the availability of resources to other organisms by causing physical state changes in biotic
or abiotic materials. Physical ecosystem engineering by organisms is the physical modi-
fication, maintenance, or creation of habitats. Ecological effects of engineers on many other
species occur in virtually all ecosystems because the physical state changes directly create
nonfood resources such as living space, directly control abiotic resources, and indirectly
modulate abiotic forces that, in turn, affect resource use by other organisms. Trophic in-
teractions and resource competition do not constitute engineering. Engineering can have
significant or trivial effects on other species, may involve the physical structure of an
organism (like a tree) or structures made by an organism (like a beaver dam), and can, but
does not invariably, have feedback effects on the engineer. We argue that engineering has
both negative and positive effects on species richness and abundances at small scales, but
the net effects are probably positive at larger scales encompassing engineered and nonen-
gineered environments in ecological and evolutionary space and time. Models of the pop-
ulation dynamics of engineers suggest that the engineer/habitat equilibrium is often, but
not always, locally stable and may show long-term cycles, with potential ramifications for
community and ecosystem stability. As yet, data adequate to parameterize such a model
do not exist for any engineer species. Because engineers control flows of energy and
materials but do not have to participate in these flows, energy, mass, and stoichiometry do
not appear to be useful in predicting which engineers will have big effects. Empirical
observations suggest some potential generalizations about which species will be important
engineers in which ecosystems. We point out some of the obvious, and not so obvious,
ways in which engineering and trophic relations interact, and we call for greater research
on physical ecosystem engineers, their impacts, and their interface with trophic relations.
Key words: cascades, coupled trophic and engineering; community stability; ecosystem engineers;
ecosystem function; feedbacks; habitat formation and destruction; population dynamics; positive and
negative effects; species diversity.
WHAT Is THE ECOLOGICAL ROLE OF A
TREE IN A FOREST?
What does a tree do in a forest? Of course the living
and dead tissues are eaten by many animals and mi-
croorganisms, and the tree competes with other plants
for light, water, and nutrients. But a tree does much
more than provide food and directly compete for re-
sources. The branch, bark, root, and living and dead
leaf surfaces make shelter, resting locations, and living
space. Small ponds full of organisms form where
throughfall gets channelled into crotches (Kitching
1971, 1983), and the soil cavities that form as roots
grow provide animals with places to live and cache
food (Foster 1988, Vander Wall 1990). The leaves and
Manuscript received 1 December
1995; accepted
3 January
1997. For reprints of this Special Feature, see footnote 1,
page 1945.
branches cast shade, reduce the impact of rain and
wind, moderate temperature extremes, and increase hu-
midity for organisms in the understory and the soil
(Holling 1992, Callaway and Walker 1997). Root
growth aerates the soil, alters its texture, and affects
the infiltration rate of water (Bouma and Anderson
1973, Tisdall and Oades 1982, Smiles 1988, Juma
1993). Dead leaves fall to the forest floor altering rain-
drop impact, drainage, and heat and gas exchange in
the soil habitat, and make barriers or protection for
seeds, seedlings, animals, and microbes (Facelli and
Pickett 1991, Callaway and Walker 1997). The trunk,
branches, and leaves can fall into forest streams cre-
ating debris dams and ponds for species to live in (Lik-
ens and Bilby 1982, Hedin et al. 1988). The roots can
bind around rocks, stabilizing the substrate and ame-
liorating hurricane impacts on other species (Basnet et
al. 1992). If the tree falls, the downed trunk, branches,
1946
October
1997 POSITIVE
INTERACTIONS
IN COMMUNITIES 1947
and resulting tip-up soil pit and mound create habitats
for numerous organisms (Collins and Pickett 1982, Pe-
terson and Pickett 1990, Peterson et al. 1990).
WHAT SHOULD WE CALL THESE EFFECTS?
It is probable that more species are affected by these
things that a tree does than directly use the tree for
food or compete with it for light, water, and nutrient
resources. And yet, these diverse ecological effects are
not trophic. Nor is the tree in the forest unique. Most
plants have similar effects, and many animals and mi-
croorganisms cause ecologically significant physical
changes in their environments (Jones et al. 1994). A
woodpecker or rot fungus may make holes in the tree
that are then used by other species (Kitching 1971,
Bradshaw and Holzapfel 1985, 1992, Daily et al. 1993),
and a beaver may come along and cut down the tree,
making a dam and pond in which hundreds of species
live (Pollock et al. 1995).
If these effects are not trophic, what are they? De-
spite their diversity, all of them involve changes from
one physical state or condition to another in the tree
(e.g., a tree without crotch ponds to a tree with them)
or its local environment (e.g., a stream without a debris
dam to a dam and pond). These physical state changes
are caused by the tree itself or by an organism, like
the woodpecker, fungus, or beaver, that changes the
physical state of the tree (e.g., a tree without holes to
a tree with holes) or the local environment (e.g., a
stream without a beaver dam to a tree dam and a pond).
In all of these cases the physical state changes can have
both positive and negative ecological consequences for
other species that live in the old or the new environment
that is created. This is because these other species de-
pend on resources whose availability is directly or in-
directly controlled by these physical state changes.
Examples of species that benefit from these changes
include mosquito species that breed exclusively in wa-
ter-filled tree holes (Kitching 1971, Bradshaw and Hol-
zapfel 1985), stream organisms that live in ponds be-
hind debris or beaver dams (Likens and Bilby 1982,
Pollock et al. 1995), and plant species that are suscep-
tible to the effects of hurricanes in the forest, and that
in part, are dependent upon the protection conferred by
the stabilizing effects of tree roots binding around rocks
(Basnet et al. 1992). But not all species benefit from
such physical state changes. Negative effects are also
common, for example, terrestrial organisms flooded out
behind a debris or beaver dam, or grassland plants and
animals excluded by succession to forest (a similar
balance between positive and negative effects is dis-
cussed by Callaway and Walker [1997]).
We have called these processes "physical ecosystem
engineering" and the organisms responsible "physical
ecosystem engineers" (Jones et al. 1994, Lawton and
Jones 1995). A tremendous diversity of examples can
be found in the literature (see Viles 1988, Meadows
and Meadows 1991, Wilson and Agnew 1992, Jones et
al. 1994, Butler 1995, Jones and Lawton 1995, Flecker
1996), including many effects caused by Homo sapiens,
a physical ecosystem engineer par excellence (see
Jones et al. 1994). We have argued that physical eco-
system engineering by organisms plays a major role in
determining the structure and functioning of most eco-
systems, and we have yet to find an ecosystem in which
physical ecosystem engineering by organisms does not
play some role, even in such hostile environments as
Arctic ice (Buynitskiy 1968, Arrigo et al. 1991). Here,
we refine the concept of engineering, distinguishing it
from trophic relations and competitive interactions. We
explore the probable net effects of physical ecosystem
engineers on species diversity and abundances, and
upon population, community, and ecosystem stability.
We also ask whether there are ways to predict which
species will be important physical engineers, and which
ecosystems will be the most affected by them. Lastly,
we explore ways in which physical engineering and
trophic interactions can be integrated.
WHAT Is ECOSYSTEM ENGINEERING?
We first need to formally define what we mean by
engineering (the definition that follows is modified
slightly from Jones et al. [1994], further clarifying the
concept and should be read in consultation with Fig.
1). "Physical ecosystem engineers are organisms that
directly or indirectly control the availability of re-
sources to other organisms by causing physical state
changes in biotic or abiotic materials. Physical eco-
system engineering by organisms is the physical mod-
ification, maintenance, or creation of habitats. The eco-
logical effects of engineering on other species occur
because the physical state changes directly or indirectly
control resources used by these other species."
Living space and engineering
It is debatable whether or not the direct provision of
living space by the structure of an organism (e.g., leaf,
bark, root surfaces of a tree), and quantitative changes
in the amount of living space as an organism grows
(e.g., a bigger tree) should be considered physical en-
gineering. In our original paper (Jones et al. 1994), we
said that inclusion or exclusion was a matter of choice,
and elected to exclude these processes in our subse-
quent discussion. We now include these processes in
physical engineering. The changes in living space via
branch growth of a tree has more in common with the
creation of living space in soil cavities caused by root
growth, which is engineering, than with the consump-
tion of tree tissues, which is not engineering.
Engineering vs. other processes
The definition of engineering can be sharpened by
contrasting it with other important ecological process-
es. Clearly, the utilization of the living or dead tissues
1948 SPECIAL FEATURE Vol. 78, No 7
PHYSICAL PHYSICAL EXAMPLES
OF CONTROLS ON RESOURCE
AVAILABILITY
STATE
I STATE 2 TO
OTHER
ORGANISMS
CREATION OF CONTROL MODULATION
RESOURCES OVER OF ABIOTIC
ABIOTIC FORCES
RESOURCES
a. FF
TREE I TREE WITHI
I TREE | | TCROTCH i Living Space Water Capture
Growth IPOND Sediment Capture
autogenic Nutrient Capture
b. TREE ~~~~TREE
WITH == I> LvnSpc TREE HOES
LLving space
Caching
Space
WOODPECKER
OR
allogenic ROT FUNGUS
feeding
SOL SOIL WITHS Living Space Water Infiltration
Caching Space Soil Texture
TREE
allogenic fact growth Aeration
d. PLANT
COVER UNDERSTORY
AND SOIL AND
SOIL C> Shaded Habitat Water Input Rain impact
Nutrient Input Wind Impact
TREE Relative Humidity
allogenic growth and eaf Temperature
production
e. SOLWT
SOILLEAF LITTER Living
Space Physical
Barrier Rainsplash Impact
Water
Drainage Heat
Exchange
autogenic TREE Soil Erosion
and Gas Exchange
allogenic leaf absclasson
AND
SOIL SUBSTRATE Soil Erosion Hurricane Impact
autogenic TREE
and rotgotah
allogenic binding
STREAM DEBRIS DAM Living Space Water Retention
Sediment Retention
autogenic TREE Nutrient Retention
and log, branch Oxygenation
allogenic abaclalon
October 1997 POSITIVE
INTERACTIONS
IN COMMUNITIES 1949
of one organism as food by a consumer or decomposed,
or the direct uptake and utilization of an abiotic re-
source (light, water, nutrients) by an organism is not
engineering. While trophic or competitive interactions
can lead to the physical engineering of habitats, they
do not inevitably do so; nor are they necessary for
engineering to occur. When a beaver cuts down the tree
to make the dam it does not have to eat any part of the
tree in order for the dam and resulting pond to have
an effect. And while disturbance (Pickett and White
1985) and engineering can often have similar effects,
not all engineering is disturbance (e.g., tree growth)
and not all disturbance is engineering (e.g., a hurri-
cane). Many "keystone species" (Mills et al. 1993,
Menge et al. 1994) are engineers (e.g., beavers), but
others (e.g., sea otters) are not; some engineering has
big effects on other species (e.g., beaver dams), while
other impacts may be relatively trivial (e.g., a cow
hoofprint).
We suspect that one reason ecology has placed little
emphasis on physical ecosystem engineering as an im-
portant general phenomenon is because of the bewil-
dering apparent variety of ways in which organisms
engineer habitats. By comparison, trophic or compet-
itive effects seem relatively straightforward. While the
devil is in the, as yet, poorly understood details, en-
gineering is a very simple concept with relatively few
key features.
Who does the engineering?
Understanding who the engineer is and what is en-
gineered is critical for predicting how other ecological
processes will influence the impact of an engineer. Eco-
system engineers bring about physical state changes in
two basic ways. Autogenic physical engineers directly
transform the environment via endogenous processes
(e.g., tree growth, development) that alter the structure
of the engineer, and the engineer remains as part of the
engineered environment. Good examples of autogenic
engineering by plants are summarized elsewhere in this
Special Feature (Bertness and Leonard 1997, Callaway
and Walker 1997, Hacker and Gaines 1997). In con-
trast, allogenic engineers change the environment by
transforming living or nonliving materials from one
physical state to another, and the engineer is not nec-
essarily part of the permanent physical ecosystem
structure (e.g., beavers). Both animals and plants can
be both autogenic and allogenic engineers (Jones et al.
1994). There are many examples of animals acting as
autogenic engineers (corals, for instance), and plants
as allogenic engineers (e.g., tree canopies affect the
understory that does not contain the tree engineer).
Trees often have mixed autogenic and allogenic engi-
neering impacts (see Fig. 1).
Feedbacks to engineers
Feedbacks occur when the physical state change di-
rectly affects the engineer either positively or nega-
tively. Jones et al. (1994) referred to positive feedbacks
as "extended phenotype engineering," because the en-
gineered habitat has direct consequences for the fitness
of the engineer (Dawkins 1982). For example, beavers
build dams and then use ponds as a place to live, avoid
predators, and cache food (positive effects). In the lon-
ger term, the level of the beaver pond may extend into
habitats that can no longer be effectively engineered,
and beavers will lose a habitat, abandoning the area
(negative effects) (Johnston and Naiman 1990, Hart-
man 1994).
However, physical engineering does not necessarily
have a feedback effect on the engineer. For example,
the formation of a debris dam by tree branches might
have no significant beneficial or adverse consequences
for the tree. Such "accidental engineering" (see Jones
et al. 1994) may have profound effects on other taxa,
but has no effect on the organism responsible for the
physical transformation of the habitat. There is an im-
portant difference here between the variable fitness
consequences of physical ecosystem engineering, and
eating, being eaten, or competing for resources, which
always have direct consequences for the fitness of the
participants.
Direct and indirect control
Physical engineering controls the availability of re-
sources to other species either directly or indirectly.
Tree crotch ponds and beaver dams directly control
water availability. On the other hand, tree roots that
bind around rocks and ameliorate the impact of hur-
ricanes on other species exert indirect control. The re-
sources used by other species that are controlled or
modulated by the engineer can be energy, materials,
space, food organisms, or combinations of these re-
sources.
FIG. 1. Examples of physical ecosystem engineering involving a tree in a forest. Physical ecosystem engineering requires
a change in physical state (physical state 1 to physical state 2) that then controls the availability of resources to other
organisms. Control may be direct, via the creation of habitat and control over the supply of abiotic resources, and/or indirect,
by modulation of abiotic forces that, in turn, affect resource use by other organisms. With autogenic engineering (a), the
engineer is a part of the new physical state (via tree growth in this example). With allogenic engineering (b-d) the new
physical state is caused by the engineer (X), but the engineer is not part of the new physical state. In examples e-g, both
autogenic and allogenic engineering are involved because the tree not only creates a new physical state for parts of the
environment in which it does not occur, but also becomes a part of the new environment. The processes involved in engineering
(e.g., growth, feeding, etc.) are shown in the boxes using lowercase type.
1950 SPECIAL FEATURE Ecology
V Nnd'R r '7
NET EFFECTS OF ENGINEERS
Impacts of engineers on species diversity and
abundances
At first sight it might appear that engineering will
have mostly positive effects on other species. After all,
if engineers make habitat, other species will now have
a place to live (see Hacker and Gaines 1997, for an
excellent example). In practice, the impacts that en-
gineers have on species' abundances and richness vary
from trivial to enormous, and they are not necessarily
positive. It is certainly true that beavers make habitats
for a very large number of species. But, it is also true
that the conversion of a stream to a beaver pond must
also have negative effects on many organisms, includ-
ing aquatic species. First, a section of stream has been
eliminated. Second, the upstream dam and pond may
decrease the availability of downstream resources, via
reduced water, oxygen, or nutrients, adversely affecting
many organisms living downstream. Third, the changes
in the riparian habitat that result from the felling of
trees will have negative effects on the species that live
there, even though it will create new habitat for other
species. It is probable that in many cases the effects of
the transformation of the habitat will be sufficient to
entirely eliminate some species from the environs of
the beaver pond, and to make others much rarer. Only
some species will benefit from the changes.
From the perspective of the locale in' which engi-
neering takes place, we see no a priori rationale for
assuming that the total number of other species that
can now live in the new habitat should be more, less,
or the same as the number of species that disappear
when the old habitat is eliminated. Nor is there any
reason to suppose that species able to live in both en-
gineered and unmodified habitat will necessarily be
commoner or rarer as a result of the engineer's activ-
ities; some will benefit, others will not. The answer
will depend on the magnitude and types of changes that
occur, the resources that are controlled, the number of
species in the habitat that depend on these resources,
and the extent to which these resources are adequate
to support persistence in the new habitat. Whether or
not it will be possible to predict such impacts in the
future remains to be seen.
On the other hand, if our temporal and spatial scale
encompasses more than the time and place that is en-
gineered, we see a different picture. At the landscape
level, beaver dams result in a net increase in the number
of habitat types, space, and resources for other species,
because what was once a valley with a stream and a
riparian zone is now a valley with unmodified stream
and riparian habitat, a pond, new type of riparian hab-
itat, modified downstream habitats, and sites with aban-
doned and collapsed beaver dams supporting their own
distinctive flora and fauna. If we watch the valley over
time, as beavers come and go, we will certainly see a
more dynamic landscape with greater rates of change
in habitats, resources, and organisms compared to the
adjacent valley where beavers never came. Hence, at
sufficiently large scales, encompassing unmodified
habitats, engineered habitats, and areas abandoned by
engineers, the net effect of engineering will almost in-
evitably be to enhance regional species richness via a
net increase in habitat diversity.
A similar sort of scale dependency of positive and
negative effects must have existed as engineers
evolved. In a given habitat, the evolution of an engineer
may have destroyed some niches, but at the same time
created new ecological opportunities to be filled by new
species. For example, the evolution and increasing im-
portance in marine environments of mobile, benthic
bioturbators from the Devonian onwards appear to have
created a highly unfavorable environment for sessile
suspension feeders, which are now largely confined to
hard substrates (Thayer 1979); on the other hand, taxa
able to survive the disturbance created by these "bi-
ological bulldozers," swimming particle feeders for ex-
ample, may have benefitted. In general, we suggest that
the net effect of the evolution of physical ecosystem
engineers, across a mosaic of habitats, will be to in-
crease species richness. Today we have sediment en-
vironments that contain bioturbators and the organisms
that thrive because of them, as well as sediments and
hard substrates without bioturbators, containing dif-
ferent organisms. No doubt many species have been
dragged along through evolution in the wake of en-
gineers, while others were lost by the wayside. What
happened to the species that depended on the putatively
massive soil, sediment, and rock engineering effects of
dinosaurs?
Our tentative suggestion that physical ecosystem en-
gineering by organisms on a global scale has a net
positive effect on total species richness would benefit
from more critical evaluation. Estimating how big the
effects are, where they are the greatest, how the net
positive effects arise from the balance of habitat ad-
dition and elimination, and how these have changed in
evolutionary time are major research challenges of the
future.
Impacts of engineers on population,
community, and ecosystem stability
Gurney and Lawton (1996) have recently formulated
variations on a simple, very general differential equa-
tion model in Lotka-Volterra form for the population
dynamics of ecosystem engineers. In this family of
models, the engineers (allogenic or autogenic) must
physically modify virgin habitat in order to survive,
that is, they are extended phenotype engineers. In the
simplest case, individual engineers work alone; in a
second version of the model they must collaborate to
successfully physically modify the habitat. A key fea-
ture of both versions of the model is that engineered
October 1997 POSITIVE
INTERACTIONS
IN COMMUNITIES 1951
habitat decays and eventually becomes unsuitable for
occupation by the engineering population. A recovery
period is necessary before degraded habitat returns to
the virgin state, and again becomes suitable for reco-
lonization and reuse by engineers. A third version of
the model explores the population dynamic conse-
quences of the distribution of residence times of habitat
in a degraded state. One interesting result to emerge
from this exercise is that no single population of eco-
system engineers has been sufficiently well studied to
completely parameterize any version of the model. Ex-
amples of the essential features (the necessity of habitat
modification for survival, cooperative and noncoop-
erative engineering, habitat decay, and the recovery of
degraded habitat) are all well documented in the lit-
erature (see Gurney and Lawton [1996] for a review),
but have not been adequately quantified for any single
species of engineer.
The model seeks to predict the dynamics and stability
of the population of engineers, and of the three habitat
states (virgin, engineered, and degraded). It makes no
attempt to predict the dynamics of other species that
are either dependent upon virgin habitat (and therefore
suffer from the impacts of the engineer), or require
engineered or degraded habitat (and therefore benefit
from the presence of the engineer). The broad conse-
quences for other organisms are, however, implicit in
the dynamics of these three habitat states, although time
delays and trophic and interspecific competitive inter-
actions could greatly complicate the general picture,
and have not yet been studied in detail. (Wilson and
Nisbet's [1997] model in this Special Feature is for-
mulated in the same spirit as Gurney and Lawton's
model, but does not require the creation of engineered
habitat for the survival of the engineer. Rather, it fo-
cuses on the role of engineering in creating sheltered
settlement or germination sites for young individuals
of the engineer or other species.)
Although Gurney and Lawton's models are a mini-
malist caricature of the dynamics of a population of
physical ecosystem engineers, they make some simple
and interesting predictions amenable to field testing.
For instance, where the engineers do not co-operate to
any significant degree, and where the distribution of
residence times of habitat in the degraded state is wide
(very variable rates of recovery at different points in
the landscape), the engineer/habitat equilibrium is al-
ways locally stable, and simulations suggest that it is
also globally stable. We think this will be a very general
result, implying that in the absence of severe abiotic
environmental disturbance, many engineers create very
stable and predictable conditions for those species that
are dependent upon them for habitat, and presumably,
a concomitant degree of stability in ecosystem pro-
cesses.
The necessary conditions for unstable limit cycles
in the abundances of engineers and habitat are either:
(1) highly cooperative engineers exploiting short-lived,
slowly recovering habitat; or (2) habitat where the dis-
tribution of recovery times from a degraded state varies
very little (i.e., the variance in recovery time is small
relative to the mean recovery time). Possible examples
of populations of engineers that cycle are beavers (in
the absence of human persecution) (see, for example,
Hartman 1994), and Dendroctonus bark beetles that
periodically erupt and kill pine-tree hosts over huge
areas (Raffa and Berryman 1987), with massive knock-
-on effects for many other species, both via the gen-
eration of large quantities of dead timber, and the mod-
ulation of many other ecosystem resources.
Obvious extensions of these simple models are to
couple them explicitly to the dynamics of nonengi-
neering species that either require unmodified habitat
to survive, or require engineered or degraded habitat.
If these nonengineering species interact among them-
selves (as competitors or predators and prey), and/or
interact with the engineer as competitors, predators, or
via shared enemies, it is not at all clear to us whether
the net effects will be positive or negative, and how
the outcome might vary over plausible parameter
space. We will return to the issue of coupling the pop-
ulation dynamics of engineers and nonengineers later.
A second, major variant will be to develop models of
engineering with explicit, spatial structure (Gurney and
Lawton 1996), with obvious implications both for the
stability of the interactions, and for local and regional
species richness.
WHICH SPECIES WILL BE IMPORTANT
PHYSICAL ENGINEERS?
As other contributors to this Special Feature point
out (e.g., Callaway and Walker 1997), when pairwise
trophic interactions are embedded in a complex food
web, counterintuitive positive outcomes, such as in-
creases in the density of prey or an inferior competitor,
can occur for many reasons (see Yodzis 1988, Pimm
1991, Miller 1994, Menge 1995). These indirect effects
and higher order interactions are now reasonably well
understood, at least in principle. In many ways, trying
to predict a priori which species will be important phys-
ical engineers is at least as, if not more, difficult as
trying to predict the outcomes of these types of pair-
wise trophic interactions, or which species is a "key-
stone" predator (Mills et al. 1993, Menge et al. 1994)
or a "strong" trophic interactor (Power 1995).
While many species cause physical state changes in
the environment, not all of the changes have important
(positive or negative) ecological consequences; some
engineers have trivial effects, just as some trophic in-
teractions have trivial population dynamic conse-
quences and some interspecific competitive interac-
tions are feeble. For example, a beaver pond has a much
bigger and longer lasting effect than the small pond
formed in an animal hoofprint. It would be very useful
1952 SPECIAL FEATURE Vol. 78, No. 7
to be able to predict which species will be important
ecosystem engineers.
Key factors scaling engineering impact
We have previously identified six factors that scale
the impact of engineers (Jones et al. 1994) and they
are worth reiterating here: (1) lifetime per capita ac-
tivity of the individual engineering organisms; (2) pop-
ulation density; (3) the local and regional spatial dis-
tribution of the population; (4) the length of time the
population has been at a site; (5) the type and formation
rate of the constructs, artifacts, or impacts, and their
durability in the absence of the engineers; and (6) the
number and types of resources that are directly or in-
directly controlled, the ways these resources are con-
trolled, and the number of other organisms that depend
on these resources. Factors 1-5 could be readily mea-
sured for many physical engineer species. Getting a
handle on factor 6 is much more difficult, but lies at
the heart of understanding the impact. Removing or
adding the engineer species, comparing naturally oc-
curring sites with and without the engineer, and arti-
ficially manipulating the environment to mimic the ef-
fects in the absence of the engineer could be useful in
many, but not all situations. Field manipulation ex-
periments to understand the effects of ecosystem en-
gineers are still uncommon (Bertness and Leonard
1997); good examples are provided by Bertness
(1984a, b, 1985), Flecker (1996), Callaway and Walker
(1997), and Hacker and Gaines (1997).
Engineering, mass flow, and the
conservation of energy
In trophic interactions, being a part of the direct flow
of energy and materials is at least a necessary precon-
dition for membership in a food web. Since the laws
of thermodynamics prevent energy or matter from be-
ing in more than one place or organism at the same
time, one organism must gain benefits (positive) from
trophic relations at the expense of another (negative).
With physical engineering, however, it is hard to see
what preconditions define membership of an "engi-
neering web" (in the sense of Martinez [1995]). We
can also say that the sorts of principles used in under-
standing trophic dynamics in food web and ecosystem
theory do not appear to be of much value in under-
standing engineering. Trophic relations must conform
to the principles of mass flow and conservation of en-
ergy. The mass consumed minus the wastes produced
times the growth efficiency equals the mass gained by
the consumer. Engineering does not conform to this
principle. The amount of mass or energy put into a
beaver (minus its wastes and the energy it uses to build
the dam) does not equal the mass of the dam or the
water it holds, nor the magnitude of the many and
varied ecosystem effects that flow from the construc-
tion of the dam. Trophic relationships must also con-
form to stoichiometric requirements. A predator has
the elemental composition of its prey minus the ele-
mental composition of its wastes. The elemental ratios
of the materials in a beaver dam, or of the organisms
in the pond, bear no relationship to the elemental stoi-
chiometry of the beaver. Perhaps the fundamental rea-
son why energy, mass, and stoichiometry appear to be
of little value in understanding engineering is that en-
gineers do not have to be a part of the energy and
material flows among the trophically connected organ-
isms they affect. They are controllers of these flows,
not participants in the flows.
Engineering and the idiosyncrasies of species
We also see another stumbling block to prediction
that pervades much of ecology, not just engineering.
The propensity of beavers to build dams is a peculiarity
of this species and does not seem to us to be predictable
a priori. Nevertheless, it is the key design feature of
this engineer. Once it is recognized, existing knowledge
about hydrology, sedimentology, and stream and pond
ecology can go a long way toward telling us what will
happen if the beaver starts to build. And we could
probably predict many of the effects if we discovered
a very different taxon that also built dams (e.g., hu-
mans). For some species, as with the beaver, there may
be no substitute for starting with natural history and
behavior in order to discover the key design feature
and thereby understand the potential engineering im-
pact. While it is likely that every engineer species will
have at least some unique attributes and impacts, many
will share common features. Trees are a good case in
point. It should be feasible to measure many of the key
structural attributes that determine the engineering im-
pact of trees in forests. Comparison of these attributes
among tree species in different ecosystems could lead
to valuable insights and generalizations. The same may
be true for the common features found among burrow-
ing animals (Meadows and Meadows 1991, Hansell
1993, Butler 1995), and so on.
WHERE WILL PHYSICAL ENGINEERING BE THE
MOST IMPORTANT?
We have argued that engineers are found in all eco-
systems (Jones et al. 1994). While this contention has
yet to be tested, it is nevertheless very likely that en-
gineering is more important in some ecosystems than
others. Being able to predict the types of ecosystems
in which engineers play the most critical roles is both
fundamentally interesting and of considerable prag-
matic value in conservation and management. While
we certainly do not have anything approaching a de-
finitive answer, our surveys of the literature and some
of the points raised earlier provide some postulated
nonexclusive generalizations that are amenable to test-
ing via surveys and comparative or experimental stud-
October
1997 POSITIVE
INTERACTIONS
IN COMMUNITIES 1953
ies. Some generalizations are obvious, while others are
not.
The dominant organisms are relatively
massive and persistent structures
The mere presence of such physical structures, their
continual growth and replacement, and their persis-
tence over long periods of time (including evolutionary
time) should lead to systems in which many other spe-
cies are dependent upon both the autogenic creation of
surface area for living space and the autogenic and
allogenic modulation of resources controlled by these
structures. Forests, Sphagnum bogs, and coral reefs are
obvious examples. It seems likely that most forests will
have many qualitatively similar engineering effects.
However, it also seems reasonable to expect that en-
gineering effects would be bigger in a Redwood or
Pacific old-growth Douglas-fir forest than in an early-
successional or second-growth forest, because of the
size and persistence, ecological and perhaps evolu-
tionary, of these huge trees.
Plant cover is extensive
As with the forest examples above, the presence of
plants affects physical structure and, hence, ecosystem
functioning. An early-successional habitat with shrubs
and young trees, a grassland, a kelp forest and a sea-
grass prairie are all plant-engineered environments. We
might expect that early-successional forest, with shrub
and saplings, has more engineering than the grassland
it invaded, because the plants are bigger. It is not true,
however, that size alone will determine the importance
of plant engineering in an ecosystem. Many deserts are
sparse in cover by higher plants. Instead, the soil is
extensively covered by dominant microphytic com-
munities of blue-green algae, cyanobacteria, and fungi
that are barely visible to the naked eye (West 1990,
Zaady and Shachak 1994). These organisms certainly
cannot be construed as massive structures. Neverthe-
less, these communities have potent engineering effects
because they secrete polysaccharides that bind the soil.
This controls stability, erosion, runoff, and site avail-
ability for germination by higher plants (West 1990,
Zaady and Shachak 1994). The same situation exists
with diatom carbohydrate secretions that bind sandy
sediments (Daborn et al. 1993).
Animals build or destroy massive, persistent,
abiotic structures
Beavers, gophers, pack rats, mole rats, alligators,
some termite species, tilefish, and corals all build large
structures above or below ground that can have eco-
system-level effects and that may last for long periods
of time, even centuries or more (Jones et al. 1994,
Butler 1995). Animals can also destroy abiotic struc-
tures on a massive scale. Puffin burrowing on the island
of Grassholm (UK) between 1898 and 1928 was so
great that the entire soil surface of the island more or
less completely eroded into the sea (Furness 1991).
Many of these species have restricted geographical
ranges, and their effects on the ecosystem are often
species specific (e.g., beavers). In this sense their en-
gineering effects on ecosystems are somewhat idio-
syncratic and unpredictable.
Large animals are abundant
Large animals such as elephants, bison, other un-
gulates, whales, etc. tend to have large per capita en-
gineering effects on the ecosystems they occupy (e.g.,
Naiman 1988, see Butler 1995). While many of these
effects may be more species specific than others (e.g.,
elephants tend to knock down more trees than wilde-
beest), many of the effects of trampling, tearing, paw-
ing, etc., may be quite similar across ecosystems. Most
of these animals occurs in herds at high densities, which
must lead to additive impacts on their ecosystems, at
the very least.
Abiotic substrates are amenable to
biogeomorphic action
There are many habitats that may or may not fit any
of the above categories, but nevertheless are exten-
sively engineered. A tremendous diversity of small and
large animals dig, burrow, or otherwise disturb soils
and sediments (Meadows and Meadows 1991, Jones et
al. 1994, Butler 1995). In one sense, the ecosystem
effects of these animals arise because these substrates
are soft enough for organisms to act on. We do not
expect much animal digging in hard granite rocks! On
the other hand, there are many rock environments
where the substrate is not too hard to prevent animal
erosion (e.g., Bloom 1978, Krumbein and Dyer 1985,
Shachak et al. 1987). In fact, animals are so adept at
digging, scraping, burrowing, boring, and even chem-
ically eroding these substrates (Butler 1995), that we
doubt whether there are many ecosystems in which
these types of activities do not play a key role. Plant
roots and chemical exudates from lichens likewise have
marked engineering effects on soils, sediments, and
rocks (Bloom 1978, Krumbein and Dyer 1985, Jones
et al. 1994).
The ecosystem has persistent structure
The magnitude of the effects of physical ecosystem
engineering depends on the persistence of physical
structures (a tree, a dam, a mound, etc.) created by
organisms. So ecosystems that do not have a high de-
gree of physical structuring (either the organisms or
what they make), such as the pelagic zones of the wa-
ters of the earth, seem unlikely to be dominated by
engineering effects. Living planktonic organisms do
physically modify the environment, however, for in-
stance by contributing to the formation of a thermocline
(Mazumder et al. 1990, Townsend et al. 1992). Fur-
Ecology
1954 SPECIAL FEATURE Vol. 78, No. 7
thermore, the effluvia of pelagic organisms, their car-
apaces and feces (marine and lake "snow" [Silver et
al. 1995]), can have surprisingly large effects on the
functioning of the entire ecosystem, and some of these
effects constitute physical engineering. Nevertheless,
we should expect the relative contribution of engi-
neering to ecosystem functioning to be less in these
relatively unstructured environments than in many oth-
er ecosystems.
Many abiotic resources are integrated
Organisms that engineer rivers, streams, soils, and
sediments tend to have large ecosystem-level effects
(e.g., beavers, earthworms, benthic bioturbators; Nai-
man et al. 1988, Lal 1991, Thompson et al. 1993, Lev-
inton 1995, Butler 1995, Pollock et al. 1995). The most
likely reason is that water, soil, and sediments integrate
many resources (living space, nutrients, prey, etc.)
within one locale, thus modifying them has big effects.
Water, in particular, tends to move readily from place
to place carrying nutrients, sediments, oxygen, etc., all
of which are key resources for many species.
Environments are extreme
It might be expected that strong abiotic forces in
extreme environments (e.g., hurricanes, wave action,
heat, and drought) would diminish the importance of
engineering, or at least make it more difficult to detect.
This does not seem to be the case; indeed, as pointed
out elsewhere in this Special Feature, positive inter-
actions in general are often important in harsh envi-
ronments (Bertness and Leonard 1997, Hacker and
Gaines 1997). Examples involving physical ecosystem
engineers include Dacryodes excelsa trees in Puerto
Rican forests, which resist hurricanes because their
roots bind around rocks (Basnet et al. 1992), and crus-
tose and coralline algae on the outer margins of coral
reefs, which resist tropical storms (Anderson 1992).
The Negev Desert, Israel, contains engineers in the
form of rock-eating snails (Shachak et al. 1987, Jones
and Shachak 1990), burrowing desert isopods (Shachak
and Jones 1995), and digging porcupines (Yair and Ru-
tin 1981, Gutterman 1982), as well as the microphytic
crust communities previously mentioned; the engi-
neering of soil and rock by each of these species has
very large effects on desert productivity and species
diversity. In fact, one could argue that natural selection
might particularly favor the evolution of extended phe-
notype engineers in extreme environments, as a means
of enhancing survival, with obvious consequences for
cohabiting, but nonengineering taxa.
ENGINEERING
MEETS TROPHIC ECOLOGY
So far we have emphasized the distinction between
engineering and trophic interactions. There is certainly
heuristic value in this, since it serves to highlight en-
gineering as an ecological phenomenon worthy of
study. But if we are ever going to understand nature
in all its complexity, we will need to integrate engi-
neering and trophic ecology (Jones and Lawton 1995),
as touched on in our concluding remarks on modeling
engineering. At its most fundamental and simplest, the
connection between engineering and trophic ecology
lies in the recognition that the creation of physical
structure by organisms controls the distribution and
abundance of resources for other species. There are a
number of obvious, and not so obvious, ways in which
engineering and trophic relations interact that can be
deduced from real-world examples.
Direct consumption of engineers
Herbivores that graze grasses, eat algae, or defoliate
trees, and pathogens or bark beetles that kill plants
obviously have impacts that go well beyond the direct
impact of consumption on the resource. The same is
true for predators of animal engineers. In essence, the
interaction is fairly straightforward. If you know what
the engineer does to other species in the habitat, the
effects of removing or reducing the density of the en-
gineer should be relatively easy to figure out, at least
in comparison to some of the other examples we discuss
below.
Competition between engineers
Plant engineers can compete with each other for abi-
otic resources, with the success of one species affecting
the performance of the other. The consequences of
these competitive outcomes for other species in the
ecosystem that do not feed upon the two competing
engineers will therefore depend on the species-specific
and nonspecific effects that the two engineers have on
the environment. Knowing exactly which species en-
gineers what features with what consequences may not
be easy. On the other hand, once this is known, working
out the consequences of changes in the relative abun-
dance of these two engineers ought to be relatively
straightforward.
Coupled engineering and trophic cascades
Sandy shorelines in the Bay of Fundy are subject to
wave action that constantly changes the physical struc-
ture. Diatoms that dominate in certain areas produce
carbohydrate exudates. These chemical secretions
cause a physical state change in the environment by
binding the sand, stabilizing its movement. The dia-
toms are auto- and allogenic engineers, and this pre-
sumably has important effects on both the diatoms and
all the other organisms that live in this habitat. Am-
phipods are the dominant grazers of diatoms in these
environments. Where amphipods are abundant, stabi-
lization is reduced. Sandpipers, the dominant predators
on the amphipods, reduce amphipod grazing and hence
promote restabilization of the habitat by diatoms (Da-
born et al. 1993). We will call this a "coupled engi-
October
1997 POSITIVE
INTERACTIONS IN COMMUNITIES 1955
neering and trophic cascade." The reason why sand-
piper distribution and abundance have such a large ef-
fect is not simply because they eat many amphipods.
Rather, it is because the engineer has big effects, and
the engineer is part of a food web. Some might call
the sandpiper a keystone species or a keystone predator,
but it is important to recognize that the effects of sand-
pipers only occur because the diatoms are engineers.
As we have pointed out previously (Jones et al.
1994), a very similar coupling of engineering and tro-
phic cascades occurs with sea otters, urchins, and kelp
forests on the Pacific coast of the United States (Estes
and Palmisano 1974, Estes 1995). Otters eat urchins,
urchins eat kelp, and kelp are auto- and allogenic en-
gineers. Kelp reduce impacts of waves and currents,
maintain water clarity, and prevent sediment move-
ment, providing a habitat for numerous species that do
not feed on kelp. The otter can be considered a keystone
predator, but only because it eats urchins, which destroy
the kelp engineers. We suspect that coupled engineer-
ing and trophic cascades will be very common. Rec-
ognition that the effects arise because engineers often
belong to food webs that exist in the engineered habitat
is crucial to predicting the impacts on species that occur
in the habitat but do not necessarily belong in the food
web.
Multiple engineers and coupled and uncoupled
trophic interactions
Coupling between trophic and engineering interac-
tions does not require all engineers to be a part of the
same food web, however. The Negev Desert example
of autogenic and allogenic engineering by the micro-
phytic crust (West 1990, Zaady and Shachak 1994)
discussed earlier illustrates this point particularly clear-
ly. The physical state changes caused by the crust have
a major influence on, among other things, the avail-
ability of germination sites for the seeds of annuals,
which are the major food source for both native grazing
species and livestock (Boeken and Shachak 1994).
These large herbivores do not feed on the microphytic
crust. Most of these animals are hooved, and when they
graze on the annuals they disturb the adjacent crust.
The small-scale pits and mounds caused by hooves
(accidental allogenic engineering) trap seeds of annuals
and runoff water, creating an ideal environment for seed
germination and growth. Here the animal engineers de-
stroy the effects of the microbial physical engineers,
increasing the production of annuals, a positive feed-
back from their engineering activities.
However, the small soil pits and mounds that are
occupied by the annuals are highly dependent on the
runoff water that comes as overland flow. The amount
of runoff is controlled by the microphytic crust, be-
cause the polysaccharide secretions of these organisms
form a surface that is more or less impermeable to water
infiltration, generating runoff into the pits and mounds.
As a consequence, the productivity of annuals, and
hence food for large grazers in this desert, depends on
there being enough crusted soil to generate sufficient
runoff. If grazing becomes too intense, the destruction
of the crust-engineered environment has a negative im-
pact on productivity. This is not just a short-term effect
of overgrazing the annuals, but a much longer term
effect on ecosystem productivity. After overgrazing,
productivity only recovers once the slow-growing crust
organisms re-engineer the habitat.
In this example, we have one physical engineer (the
microphytic crust community) that is not part of the
annual plant-ungulate food web, but that nevertheless
prevents annuals from getting established (few cracks
in the soil in which to germinate, little moisture infil-
trating). Paradoxically, this first engineer also facili-
tates the growth of annual plants (via runoff) once a
second physical engineer (hooved mammals) creates
suitable habitat (pits and mounds). This second engi-
neer, in a different food web from the microphytic crust,
also negatively impacts the crust (via disturbance from
hooves). While this example is very complex, we sus-
pect that nature will be full of similar, or even more
complex examples. It is our job to figure them out.
CONCLUDING REMARKS
Compared with the huge efforts that ecologists have
devoted to the study of trophic interactions among spe-
cies, interspecific competition, species diversity, and
ecosystem fluxes, engineering is a very poor relative.
To start to reverse this imbalance, we need empirical
data from comparative and experimental studies, mod-
els, and conceptual integration of the phenomenon we
call physical ecosystem engineering. We need to know
which attributes of species are relevant to engineering,
and which are not relevant, and which attributes are
general, and which are not. We need to understand
which species or functional types exert what types of
controls on what type of resource flows, in which eco-
systems. Until we have as good a conceptual and em-
pirical understanding of engineering as we do of trophic
relations, integrating these two aspects of ecology, or
even distinguishing engineering from trophic effects in
real ecosystems, will be difficult.
One thing, at least, should by now be clear. Ecolo-
gists interested in the significance of positive interac-
tions in ecosystems cannot ignore engineering. Not all
effects of engineers are positive-within the ambit of
the physically altered locality there may be as many
species that suffer from the resulting changes in re-
source flows and habitat structure as benefit. But at
regional or landscape scales, among a mosaic of en-
gineered and nonengineered habitat, the overall im-
pacts of engineering are most likely to enhance species
richness. Similar remarks probably apply to the effects
of engineers over evolutionary time.
It is also important to realize that engineers and
1956 SPECIAL FEATURE Ecology
"keystone species" are not synonymous. Many engi-
neers have small, difficult-to-detect effects; only some
have dramatic effects, but where they do, understand-
ing how the engineers modify and modulate resource
flows for other species, and create and maintain entire
habitats, are among the most significant and poorly
researched questions in ecology.
ACKNOWLEDGMENTS
We are indebted to the engineers of the planet earth, humans
included, for providing insights. The Mary Flagler Charitable
Trust, The John Simon Guggenheim Memorial Foundation
(C. G. Jones), NSF (DEB-9311600), and the Core Grant to
the NERC Centre for Population Biology provided financial
support. Contribution to the program of the Institute of Eco-
system Studies, and paper number 214 from The Mitrani Cen-
ter for Desert Ecology.
LITERATURE CITED
Anderson, R. A. 1992. Diversity of eukaryotic algae. Bio-
diversity and Conservation 1:267-292.
Arrigo, K. R., C. W. Sullivan, and J. N. Kremer. 1991. A
bio-optical model of Antarctic sea ice. Journal of Geo-
physical Research 96 C6:10581-10592.
Basnet, K., G. E. Likens, F N. Scatena, and A. E. Lugo.
1992. Hurricane Hugo: damage to a tropical rain forest in
Puerto Rico. Journal of Tropical Ecology 8:47-55.
Bertness, M. D. 1984a. Habitat and community modification
by an introduced herbivorous snail. Ecology 65:370-381.
1984b. Ribbed mussels and Spartina alternifiora
production in a New England salt marsh. Ecology 65:1794-
1807. 1985. Fiddler crab regulation of Spartina alternifiora
production on a New England salt marsh. Ecology 66:
1042-1055.
Bertness, M. D., and G. H. Leonard. 1997. The role of pos-
itive interactions in communities: lessons from intertidal
habitats. Ecology 78:1976-1989.
Bloom, A. L. 1978. Geomorphology. Prentice Hall, Engle-
wood Cliffs, New Jersey, USA.
Boeken, B., and M. Shachak. 1994. Desert plant commu-
nities in human made patches-implications for manage-
ment. Ecological Applications 4:702-716.
Bouma, J., and J. L. Anderson. 1973. Relationships between
soil structure characteristics and hydraulic conductivity.
Pages 77-105 in M. Steely, R. C. Dinauer, and J. M. Hach,
editors. Field soil water regime. Soil Science Society of
America Special Publication Number 5. Soil Science So-
ciety of America, Madison, Wisconsin, USA.
Bradshaw, W. E., and C. M. Holzapfel. 1985. The distri-
bution and abundance of treehole mosquitoes in eastern
North America: perspectives from north Florida. Pages 3-
23 in L. P. Lounibos, J. R. Rey, and J. H. Frank, editors.
Ecology of mosquitoes: proceedings of a workshop. Florida
Medical Entomology Laboratory, Vero Beach, Florida,
USA.
Bradshaw, W. E., and C. M. Holzapfel. 1992. Resource lim-
itation, habitat segregation, and species interactions of Brit-
ish tree-hole mosquitoes in nature. Oecologia 90:227-237.
Butler, D. R. 1995. Zoogeomorphology: animals as geo-
morphic agents. Cambridge University Press, New York,
New York, USA.
Buynitskiy, V. K. 1968. The influence of microalgae on the
structure and strength of Antarctic sea ice. Oceanology 8:
771-776.
Callaway, R. M., and L. R. Walker. 1997. Competition and
facilitation: a synthetic approach to interactions in plant
communities. Ecology 78:1958-1965.
Collins, B. S., and S. T. A. Pickett. 1982. Vegetation com-
position and relation to environment in an Allegheny hard-
woods forest. American Midland Naturalist 108:117-123.
Daborn, G. R., C. L. Amos, M. Brylinsky, H. Christian, G.
Drapeau, R. W. Faas, J. Grant, B. Long, D. M. Paterson,
G. M. E. Perillo, and M. C. Piccolo. 1993. An ecological
cascade effect: migratory birds affect stability of intertidal
sediments. Limnology and Oceanography 38:225-231.
Daily, G. C., P. R. Ehrlich, and N. M. Haddad. 1993. Double
keystone bird in a keystone species complex. Proceedings
of the National Academy of Sciences (USA) 90:592-594.
Dawkins, R. 1982. The extended phenotype. Oxford Uni-
versity Press, Oxford, England.
Estes, J. A. 1995. Top-level carnivores and ecosystem ef-
fects: questions and approaches. Pages 151-158 in C. G.
Jones and J. H. Lawton, editors. Linking species and eco-
systems. Chapman and Hall, New York, New York, USA.
Estes, J. A., and J. F Palmisano. 1974. Sea-otters: their role
in structuring nearshore communities. Science 185:1058-
1060.
Facelli, J. M., and S. T. A. Pickett. 1991. Plant litter: its
dynamics and effects on plant community structure. Bo-
tanical Review 57:1-32.
Flecker, A. S. 1996. Ecosystem engineering by a dominant
detritivore in a diverse tropical stream. Ecology 77:1845-
1854.
Foster, R. C. 1988. Microenvironments of soil microorgan-
isms. Biology and Fertility of Soils 6:189-203.
Furness, R. W. 1991. The occurrence of burrow-nesting
among birds and its influence on soil fertility and stability.
Symposium of the Zoological Society of London 63:53-
67.
Gurney, W. S. C., and J. H. Lawton. 1996. The population
dynamics of ecosystem engineers. Oikos 76:273-283.
Gutterman, Y. 1982. Observations on the feeding habits of
the Indian crested porcupine (Hystrix indica) and the dis-
tribution of some hemicryptophytes and geophytes in the
Negev desert highlands. Journal of Arid Environments 5:
261-268.
Hacker, S. D., and S. D. Gaines. 1997. Some implications
of direct positive interactions for community species di-
versity. Ecology 78:1990-2003.
Hansell, M. H. 1993. The ecological impact of animal nests
and burrows. Functional Ecology 7:5-12.
Hartman, G. 1994. Long-term population development of a
reintroduced beaver (Castor fiber) population in Sweden.
Conservation Biology 8:713-717.
Hedin, L. O., M. S. Mayer, and G. E. Likens. 1988. The
effect of deforestation on organic debris dams. Proceedings
of the International Association for Theoretical and Ap-
plied Limnology 23:1135-1141.
Holling, C. S. 1992. Cross-scale morphology, geometry, and
dynamics of ecosystems. Ecological Monographs 62:447-
502.
Johnston, C. A., and R. J. Naiman. 1990. Aquatic patch
creation in relation to beaver population trends. Ecology
71:1617-1621.
Jones, C. G., and J. H. Lawton, editors. 1995. Linking spe-
cies and ecosystems. Chapman and Hall, New York, New
York, USA.
Jones, C. G., J. H. Lawton, and M. Shachak. 1994. Organ-
isms as ecosystem engineers. Oikos 69:373-386.
Jones, C. G., and M. Shachak. 1990. Fertilization of the
desert soil by rock-eating snails. Nature 346:839-841.
Juma, N. G. 1993. Interrelationships between soil structure/
texture, soil biota/soil organic matter and crop production.
Geoderma 57:3-30.
Kitching, R. L. 1971. An ecological study of water-filled
tree-holes and their position in the woodland ecosystem.
Journal of Animal Ecology 40:281-302.
October
1997 POSITIVE
INTERACTIONS IN COMMUNITIES 1957
. 1983. Community structure in water-filled treeholes
in Europe and Australia-comparisons and speculations.
Pages 205-222 in J. K. Frank and L. P. Lounibos, editors.
Phytotelmata: terrestrial plants as hosts for aquatic insect
communities. Plexus, Medford, New Jersey, USA.
Krumbein, W. E., and B. D. Dyer. 1985. This planet is alive-
weathering and biology, a multi-faceted problem. Pages
143-160 in J. I. Drever, editor. The chemistry of weath-
ering. D. Reidel, Dordrecht, The Netherlands.
Lal, R. 1991. Soil conservation and biodiversity. Pages 89-
103 in D. L. Hawksworth, editor. The biodiversity of mi-
croorganisms and invertebrates: its role in sustainable ag-
riculture. CAB International, Wallingford, England.
Lawton, J. H., and C. G. Jones. 1995. Linking species and
ecosystems: organisms as ecosystem engineers. Pages 141-
150 in C. G. Jones and J. H. Lawton, editors. Linking
species and ecosystems. Chapman and Hall, New York,
New York, USA.
Levinton, J. 1995. Bioturbators as ecosystem engineers: con-
trol of the sediment fabric, inter-individual interactions, and
material fluxes. Pages 29-36 in C. G. Jones and J. H. Law-
ton, editors. Linking species and ecosystems. Chapman and
Hall, New York, New York, USA.
Likens, G. E., and R. E. Bilby. 1982. Development, main-
tenance, and role of organic-debris dams in New England
streams. Pages 122-128 in E J. Swanson, R. J. Janda, T.
Dunne, and D. N. Swanston, editors. Sediment budgets and
routing in forest drainage basins. USDA Forest Service
General Technical Report PNW-141.
Martinez, N. D. 1995. Unifying ecological subdisciplines
with ecosystem food webs. Pages 166-175 in C. G. Jones
and J. H. Lawton, editors. Linking species and ecosystems.
Chapman and Hall, New York, New York, USA.
Mazumder, A., W. D. Taylor, D. J. McQueen, and D. R. S.
Lean. 1990. Effects of fish and plankton on lake temper-
ature and mixing depth. Science 247:312-315.
Meadows, P. S., and A. Meadows, editors. 1991. The en-
vironmental impact of burrowing animals and animal bur-
rows. Clarendon, Oxford, England.
Menge, B. A. 1995. Indirect effects in marine rocky intertidal
interaction webs: patterns and importance. Ecological
Monographs 65:21-74.
Menge, B. A., E. L. Berlow, C. A. Blanchette, S. A. Navarrete,
and S. B. Yamada. 1994. The keystone species concept:
variation in interaction strength in a rocky intertidal habitat.
Ecological Monographs 64:249-286.
Miller, T. E. 1994. Direct and indirect species interactions
in an early old-field plant community. American Naturalist
143:1007-1025.
Mills, L. S. M., M. E. Soul, and D. R Doak. 1993. The
keystone-species concept in ecology and conservation.
BioScience 43:219-224.
Naiman, R. J. 1988. Animal influences on ecosystem dy-
namics. BioScience 38:750-752.
Naiman, R. J., C. A. Johnston, and J. C. Kelley. 1988. Al-
teration of North American streams by beaver. BioScience
38:753-762.
Peterson, C. J., W. P. Carson, B. C. McCarthy, and S. T. A.
Pickett. 1990. Microsite variation and soil dynamics with-
in newly created treefall pits and mounds. Oikos 58:39-
46.
Peterson, C. J., and S. T. A. Pickett. 1990. Microsite and
elevational influences on early forest regeneration after cat-
astrophic windthrow. Journal of Vegetation Science 1:657-
662.
Pickett, S. T. A., and R S. White, editors. 1985. The ecology
of natural disturbance and patch dynamics. Academic
Press, Orlando, Florida, USA.
Pimm, S. L. 1991. The balance of nature? University of
Chicago Press, Chicago, Illinois, USA.
Pollock, M. M., R. J. Naiman, H. E. Erickson, C. A. Johnston,
J. Pastor, and G. Pinay. 1995. Beaver as engineers: influ-
ences on biotic and abiotic characteristics of drainage ba-
sins. Pages 117-126 in C. G. Jones and J. H. Lawton,
editors. Linking species and ecosystems. Chapman and
Hall, New York, New York, USA.
Power, M. E. 1995. Floods, food chains, and ecosystem pro-
cesses in rivers. Pages 52-60 in C. G. Jones and J. H.
Lawton, editors. Linking species and ecosystems. Chapman
and Hall, New York, New York, USA.
Raffa, K. F, and A. A. Berryman. 1987. Interacting selective
pressures in conifer-bark beetle systems: a basis for recip-
rocal adaptations? American Naturalist 129:234-262.
Shachak, M., and C. G. Jones. 1995. Ecological flow chains
and ecological systems: concepts for linking species and
ecosystem perspectives. Pages 280-294 in C. G. Jones and
J. H. Lawton, editors. Linking species and ecosystems.
Chapman and Hall, New York, New York, USA.
Shachak, M., C. G. Jones, and Y. Granot. 1987. Herbivory
in rocks and the weathering of a desert. Science 236:1098-
1099.
Silver, M. W., S. L. Coale, D. K. Steinberg, and C. H. Pilskaln.
1995. Marine snow: what it is and how it affects ecosystem
function. Pages 45-51 in C. G. Jones and J. H. Lawton,
editors. Linking species and ecosystems. Chapman and
Hall, New York, New York, USA.
Smiles, D. E. 1988. Aspects of the physical environment of
soil organisms. Biology and Fertility of Soils 6:204-215.
Thayer, C. W. 1979. Biological bulldozers and the evolution
of marine benthic communities. Science 203:458-461.
Thompson, L., C. D. Thomas, J. M. A. Radley, S. Williamson,
and J. H. Lawton. 1993. The effect of earthworms and
snails in a simple plant community. Oecologia 95:171-178.
Tisdall, J. M., and J. M. Oades. 1982. Organic matter and
water-stable aggregates in soils. Journal of Soil Science 33:
141-163.
Townsend, D. W., M. D. Keller, M. E. Sieracki, and S. G.
Ackleson. 1992. Spring phytoplankton blooms in the ab-
sence of vertical water column stratification. Nature 360:
59-62.
Vander Wall, S. B. 1990. Food hoarding in animals. Uni-
versity of Chicago Press, Chicago, Illinois, USA.
Viles, H. A., editor. 1988. Biogeomorphology. Basil Black-
well, Oxford, UK.
West, N. E. 1990. Structure and function of microphytic soil
crusts in wildland ecosystems of arid to semi-arid regions.
Advances in Ecological Research 20:180-223.
Wilson, J. B., and A. D. Q. Agnew. 1992. Positive-feedback
switches in plant communities. Advances in Ecological Re-
search 23:263-336.
Wilson, W. G., and R. M. Nisbet. 1997. Cooperation and
competition along smooth environmental gradients. Ecol-
ogy 78:2004-2017.
Yair, A., and J. Rutin. 1981. Some aspects of the regional
variation in the amount of available sediment produced by
isopods and porcupines, northern Negev, Israel. Earth Sur-
face Processes and Landforms 6:221-234.
Yodzis, P. 1988. The indeterminacy of ecological interactions
as perceived through perturbation experiments. Ecology
69:508-515.
Zaady, E., and M. Shachak. 1994. Microphytic soil crust and
ecosystem leakage in the Negev Desert. American Journal
of Botany 81:109.
