You are on page 1of 10

Landscape Ecology

Author(s): Dean L. Urban, Robert V. O'Neill and Herman H. Shugart, Jr.


Reviewed work(s):
Source: BioScience, Vol. 37, No. 2 (Feb., 1987), pp. 119-127
Published by: University of California Press on behalf of the American Institute of Biological Sciences
Stable URL: http://www.jstor.org/stable/1310366 .
Accessed: 23/07/2012 04:57

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

University of California Press and American Institute of Biological Sciences are collaborating with JSTOR to
digitize, preserve and extend access to BioScience.

http://www.jstor.org
Landscape
Ecology
A hierarchicalperspectivecan help scientists
understandspatialpatterns

Dean L. Urban, Robert V. O'Neill, and Herman H. Shugart, Jr.

A landscapeis a mo-
terrestrial death to the large-scale effects of
saic of heterogeneous land A landscape is a wildfires, drought, and epidemic dis-
forms, vegetation types, and ease. Biotic, or regenerative, processes
land uses. The study of landscape-its mosaic of patches, also vary in scale from the regrowth
spatial patterns and how they devel- the components of an individual to the reorganization
op-is presently emerging as a new of species assemblages. Environmen-
discipline in the field of ecology (For- of pattern tal constraints include microclimatic
man 1981, 1983, Forman and Go- and fine-scale soil conditions govern-
dron 1981, 1986, Naveh and Lieber- ing seed germination, and also sub-
man 1984, Noss 1983, Risser et al. phasis is on forested landscapes,we continental climatic regimes that de-
1984). Landscape ecology is motivat- can generalizea theory of landscape lineate biomes, such as the Eastern
ed by a need to understand the devel- ecology. Deciduous Forest.
opment and dynamics of pattern in The agents of pattern formation are
ecological phenomena (Clark et al. interwoven in landscape develop-
1978, Levin 1976a,b, 1978, Whitta- Landscape pattern and process ment. This interaction allows some
ker and Levin 1977, Wiens 1976), the We will first focus on the wide range sites to be especially prone to, or
role of disturbance in ecosystems of phenomenain a naturalterrestrial sheltered from, disturbances. For ex-
(Mooney and Godron 1983, Pickett landscapeby consideringthe appar- ample, topographic position interacts
and White 1985, Sousa 1984, White ent complexityof landscapedynamics with fire frequency; dry ridges burn
1979), and characteristic spatial and and illustrating how a hierarchical more frequently than moist (mesic)
temporal scales of ecological events paradigm lends itself to simplifying coves. Regenerative processes are in-
(Allen and Starr 1982, O'Neill et al. suchcomplexity.Ourperspectivealso fluenced by site quality, and also vary
1986). affordsinsightsinto the management with the age and life-history attri-
Pattern, generated by processes at of man-dominatedlandscapes. butes of the regenerating individuals
various scales, is the hallmark of a (Odum 1969, Shugart and Hett
landscape. In this paper we outline an Developmentof landscapepattern.A 1973). Moreover, both disturbances
approach to landscape study that em- landscapeis a mosaic of patches, the and regeneration may be constrained
ploys a hierarchical paradigm of pat- componentsof pattern.The agentsof by the existing spatial pattern (Curtis
tern and behavior. Although our em- pattern formation on natural land- 1956, Forman 1981, Watt 1947). Fi-
scapes can be categorizedas distur- nally, new patches are continually
bances, biotic processes (especially superimposed on existing patches
Dean L. Urbanis a researchassociateand the demographicprocesses of birth, (Reiners and Lang 1979). The emer-
HermanH. Shugart,Jr. is the Corcoran death, and dispersal), and environ- gent scenario is a mosaic of patches of
Professor of EnvironmentalSciences in mental constraints (Levin 1978). various size, of various origins, in
the Department of EnvironmentalSci- Each of these agents can be consid- various stages of regeneration, ap-
ences, Universityof Virginia,Charlottes- ered across a spectrumof spatial and proaching microenvironmental equi-
ville, VA 22903. Robert V. O'Neill is a libria at various rates. Such complex-
senior researchecologist in the Environ- temporalscales. For example, distur-
mentalSciencesDivision, Oak RidgeNa- bances that affect terrestrial land- ity would seem overwhelming at first
tionalLaboratory,Oak Ridge,TN 37831. scapes vary in spatial extent, recur- and any attempt to fully understand
They study the developmentand dynam- rence interval, and intensity (Pickett landscapes would appear futile.
ics of scaled pattern in ecosystems. ? and White 1985, Sousa 1984, White
1987 American Institute of Biological 1979). Disturbancesrange from the Organization of landscape pattern.
Sciences. localized effects of an individual But importantly, the complexity of

February 1987 119


landscape pattern is organized in a ters Canoe Area of Minnesota. The still larger scales and over longer
special way: the component events concept of disturbance is itself implic- spans of time, genetic fluxes define
and patches occur at characteristic itly scaled; if disturbances are events the domainsof adaptationand evolu-
scalesthat are positivelycorrelatedin that kill tress prematurely, then "dis- tionaryprocesses.
time and space. Disturbances,for ex- turbance" is confined to a relatively Environmental constraints are
ample, occur over a wide range of narrow window of time correspond- scaled by the manner in which we
space and time scales, but those af- ing to the lifespan of trees (Figure la). witness them. Atmospheric condi-
fectinga particularlandscapetypical- Events outside this domain, while tions can be observedat virtuallyany
ly can be divided into events occur- they may affect trees, are not consid- scale, but "microclimate,""weath-
ring on characteristic scales. For ered disturbances. Small-scale distur- er," and "climate"connotephenome-
example, Romme and Knight (1982) bances tend to occur more frequently na witnessed over increasinglylarger
were able to break the fire regimein while larger events tend to be less areasand longertimespans.Likewise,
Yellowstone Park into two compo- frequent. we could also measuresoil processes
nents:frequent,small fires (everyfew Demographic processes are also at virtuallyany scale, but we would
decades, affecting areas of less than scaled. Individuals have characteristic not measurethe chemicalweathering
100 ha); and larger,less-frequentfires sizes and lifespans that dictate corre- of mineralsover the entire Canadian
(everyfew centuries,affectingareasof spondingly scaled patterns (Watt Shield (we would use a small, repre-
severalhundredha). Similarly,Hein- 1947). Species dispersal distances and sentativesurface),nor would we ob-
selman (1973) documented a two- rates define the scale of larger and servethe evolutionof landformson a
scale fire regimein the BoundaryWa- longer-term patterns (Figure lb). At single hillslope (where we would

9 9
I I I I I I I I I I
8 a- 8

7 7

a6 I6
0
U
-.5 . 5
0O
03 4 <
(4
-J -I 4
O
0 3 0.3

2 -2 2

1 - 1

U
I I I I I I I I I 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
SPATIAL SCALE (log m2) SPATIAL SCALE (log m2)

9 i I I
I I I I I I I 9

8 - 8

7 7

0) 6
2 GLACIAL 6
CYCLES
w
Lu 5 PEDOGENESIS,
-J
4 CLIMATE FLUX
0 4 -i 4
o4
(1) TOPOGRAPHY
3
- _ L 3
IL
1
2 22

1 MICRO- 1
ENVIRONS
C u
0 2 4 6 8 10 12 0 2 4 6 8 10 12
SPATIAL SCALE (log m2) SPATIAL SCALE (log m2)

Figure1. (a) Disturbanceregimes, (b) forest processes,(c) environmentalconstraints,and (d) vegetationpatterns,viewed in the
context of space-timedomains.Modifiedfrom Delcourtet al. (1983).

120 BioScience Vol. 37 No. 2


study erosional processes).The phe- Hierarchical structure. Components scramble competition ensues, until
nomenawe study tend to be positive- of a hierarchical system are organized another tree grows large enough to
ly correlatedin time and space (Figure into levels according to functional become dominant. When this tree
Ic). scale (Figure 2). Events at a given dies, the cycle repeats. The spatial
Vegetationpatternscan be resolved level have a characteristic natural fre- unit of gap dynamics is equal to the
on differentscales (Figureld). Whit- quency and, typically, a correspond- area affected by the death of a cano-
taker's (1953) notion of climax pat- ing spatial scale. In general, low-level py-dominant tree; its natural frequen-
tern was a two-level description of events are comparatively small and cy reflects the lifespan of the domi-
vegetationpatternthat reconciledthe fast; higher-level behaviors are larger nant species (Shugart 1984, Shugart
broad-scale idea of a single stable and slower. More strictly, compo- et al. 1981).
communityof plants dictated by cli- nents of a hierarchical system may be Trees within a forest gap interact
mate (Clements1916) with fine-scale ordered into levels according to a much more among themselves, by vir-
observationsof individualplant com- number of criteria (Allen et al. 1984). tue of their shared regime of available
munities (Gleason 1939). Moreover, Higher levels may be larger than, resources (especially light), than they
the large-scalegeneralizationis only slower than, constrain (control), or do with trees beyond the gap. These
loosely coupled to the fine-scalede- contain lower levels. In many of the interactions define the boundaries of
tails of a phenomenon. The succes- hierarchies we will consider, all crite- a gap. By extension, a larger forest
sional schemesdescribedby Clements ria will apply. The rules structuring area can be decomposed into a mosa-
may apply in general to large areas, hierarchies can be conveniently illus- ic of gap-sized patches, in which each
but they do not predictwith certainty trated through an extended example. gap undergoes its own dynamics. But
what one will actuallyfind on a par- Let us develop a hierarchy to study the gaps are neither identical nor
ticularsite at a giventime. This loose- the species-composition dynamics of completely independent. The gaps
ly coupled, multileveledorganization a deciduous forest system in the east- comprising a mesic cove share similar
of landscapesrequiresa new concep- ern United States. The forest gap has species under similar growing condi-
tual model. We suggest a paradigm long been recognized as a functional tions, and they exchange seeds and
that comes fromhierarchytheory(Al- unit in forest systems (Watt 1925, nutrients more often within the cove
len and Starr1982, Allen et al. 1984, 1947; see also Bormann and Likens than with gaps on a nearby ridge.
O'Neill et al. 1986, Pattee 1973, 1979, Shugart 1984). When a large Again, these similarities allow us to
Whyte et al. 1969). tree dies, it creates a gap where subor- delineate an area of characteristic size
dinate trees may thrive under a re- within which gaps interact at a char-
The paradigm gime of greater resources now avail- acteristic frequency, and allow us to
able to them. A transitional stage of define stands as higher-level compo-
Hierarchytheory is concerned with
systems that have a certain type of
organizedcomplexity. Hierarchically
organizedsystems can be divided,or
decomposed,into discretefunctional
components operating at different LEVEL3
scales (Simon 1962). As applied to
landscape ecology, the hierarchical
paradigmprovides guidelinesfor de-
finingthe functionalcomponentsof a
system,and definesways components
at differentscales are related to one
another (e.g., lower-levelunits inter-
act to generatehigher-levelbehaviors
and higher-levelunits controlthose at
lower levels). This paradigmcan aid
the design of studies in landscape LEVEL 2 ---- -- C-
ecology and the prediction of how
external factors will alter an
ecosystem.
Natural phenomena often are not
perfectly decomposable: spatial
boundariesmay be difficultto define
0-0
preciselyand componentsmay inter- C-
act. Yet many complex, naturalphe- C
nomenaare nearlydecomposable(Al- LEVEL1 ..'< /\ ?
len and Starr 1982, O'Neill et al. . . . --- o -
o
1986, Simon 1973) and thus can be
conceptualized usefully as hierarchi- Figure2. A generalizedhierarchicalsystem.Thick arrowsindicatestronginteractions;
cal systems. brokenarrows,weak interactions.

February 1987 121


FOUR LEVELS OF A FOREST HIERARCHY relationto a specifiedphenomenonof
interest. Different phenomena may
LEVEL BOUNDARYDEFINITION SCALE call for differences in the hierarchies
we use to study them. While the hier-
LANDSCAPE PHYSIOGRAPHIC PROVINCES; 10000s ha archy, gap-stand-watershed-land-
CHANGES IN LAND USE OR scape, might be appropriate for a
DISTURBANCE REGIME study of either species composition or
nutrient cycling in forests, these levels
WATERSHED LOCAL DRAINAGE BASINS; 100s-1000s ha need not be relevant for other pur-
TOPOGRAPHIC DIVIDES poses. For a landscape dominated by
recurrent fires, hierarchical levels cor-
STAND TOPOGRAPHIC POSITIONS; 1s-10s ha
responding to the scales of individual
DISTURBANCES PATCHES fires and to the larger fire-mosaic
GAP LARGE TREE'S INFLUENCE 0.01-0.1 ha might be a more appropriate concep-
tualization. For a study of forest
birds, an obvious focal scale would be
Figure3. A forestedlandscapeas a hierarchyof gaps, stands,and watersheds. territories. The higher levels, for ex-
ample, the forest stand, may or may
not be suitable to studies of fires and
nents of a forest hierarchy. of a higher-level component (Koestler birds. Thus, the species-composition
Moving upscale once again, we 1967). hierarchy for a forest system is not the
might define watershedsat the next Landscapes have a special kind of only one possible.
higherlevel, because stands within a vertical structure: they are nested spa-
watershed share a similar resource tially. Each level of the hierarchy con- Mechanistic explanation. The behav-
base and interactmore among them- tains the levels below it. This property ior of a forest gap through time de-
selves than they do with stands in of containment provides for a number pends on the individual trees in the
other watersheds. At a still higher of special features (Allen and Hoek- gap. Their growth rates, lifespans,
level, we might define landscapesas stra 1984). Levels in a nested hierar- shade tolerance, response to moisture
units of similar, interacting water- chy may coincide when defined by a and nutrient levels, and initial sizes
sheds.At the landscapescale, bound- variety of ordering criteria or mea- collectively determine which trees
aries might be coincidentwith large- sured attributes. Forest gaps might be come into dominance and which trees
scale physiographic features (e.g., recognizable in terms of species com- are suppressed. Gaps, in turn, interact
mountainranges)that governweath- position, biomass, ambient sunlight to generate stand dynamics. Whether
er patterns and limit frequenciesof (insolation), or other attributes. They sugar maple (Acer saccharum) will
species movement. Of course, such might also be recognizable in terms of become more important than white
landscapes interact as well, giving the limited basic resources that con- oak (Quercus alba) in a particular
definition to still higher levels, for strain tree growth. Thus, the hierar- stand may depend on the interaction
example, regional forest provinces, chy is also ordered on the criterion of of nearby gaps. If oak-dominated
such as a spruce-fir forest. constraint. gaps tend to become maple-dominat-
We have constructed a four-level Indeed, it seems that constraint and ed gaps as a result of seed rain from
hierarchy to represent a forest (Figure interaction can be mutually reinforc- nearby maples, stand composition
3). At each level, similar and interact- ing as ordering criteria: patches delin- will shift toward maple. This general
ing components become the function- eated by a spatially distributed con- rule carries upscale: stands, then wa-
al aggregates at the next higher level. straint (e.g., topographic moisture) tersheds, then landscapes interact to
This is a rate-structured hierarchy, may interact to generate higher-level generate successively higher-level
because components of one aggregate aggregates. Forest stands defined on behaviors.
interact more frequently and inten- topographic moisture may be joined We can go a long way toward un-
sively among themselves than with by seed dispersal to generate an inter- derstanding a complex phenomenon
components of other aggregates. This acting landscape mosaic. As a further when we explain its behavior in terms
rule defines the horizontal structure consequence, note that because land- of interactions among its parts. But
(within levels) as well as the vertical scapes contain stands, higher levels of understanding a hierarchical phe-
structure (between levels) of a hierar- this hierarchy are larger than lower nomenon requires more than mecha-
chical system. Interactions among levels; also, because stands interact nisms. Understanding requires that
components at one level generate the more within themselves than among the mechanisms be considered in
behaviors of a component at the next themselves, higher levels are slower context.
higher level. A gap has its own inter- than lower levels. This special rela-
nal dynamics, but it also contributes tionship among ordering criteria in a Constraint and higher-level context.
to the behavior of a stand. In turn, a landscape hierarchy makes its struc- Specific conditions within the gap in-
stand's behaviors are not only its ture very robust. The hierarchical lev- fluence the processes, such as tree
own, but also are a part of watershed els are often evident as patches in growth and longevity, that generate
function. Each patch, at any level, is nature. gap dynamics. The pattern of avail-
at once an integral whole and a part Each hierarchy is constructed in able light in the gap, for example,

122 BioScience Vol. 37 No. 2


provides a context that constrains in- at the next lower level, and its signifi- derstanding the event of interest. For
dividual tree growth, usually limiting cance in the context of higher-level example, in describing gap dynamics,
replacement of a dominant tree to constraints(O'Neill et al. 1986). we know that safe sites for germina-
shade-tolerant species. This knowl- tion are important to seedling estab-
edge may not, however, allow us to Hierarchyas an analytictool. Simul- lishment. However, we need not con-
predict with certainty which shade- taneouslyconsideringa complex eco- sider these details at the level of the
tolerant species will come into domi- logical system on many scales is an gap, because within gaps safe sites
nance: it might be sugar maple, or intimidating prospect that is made occur with predictable frequency. We
beech (Fagus grandifolia), or bass- simplerby the mannerin which pat- also know that the specific forest ex-
wood (Tilia americana), or some oth- ternstranslateacrosshierarchicallev- ists because the post-Pleistocene cli-
er equally tolerant species. The great- els. Hierarchyisolates the phenome- mate favors trees, but this knowledge
er the functional redundancy among non of interest. In a forest, a tree has little bearing on the event at hand.
trees, the less predictable is case-spe- integrates fine-scale variation in its To recap, a hierarchical perspective
cific behavior. The constraint in effect physical environment, experiencing would emphasize three strategic con-
judges all trees by a single standard. only a blend. For example, a tree's cerns in an analysis of landscape pat-
Some other constraints on tree growthringis an integrationof grow- tern: (1) to detect pattern and define
growth are soil moisture levels, nutri- ing conditionsduringthe year, and it its spatial and temporal scale, which
ent availability, and disturbance re- is an instant cross-section of any long- is to define functional patches at a
gimes such as fires or wind. All these er-term trends. In addition, a tree specific level; (2) to infer which fac-
constraints affect the trees within the experiences large patterns as a rela- tors generate the pattern, whether
gap, favoring some over others. The tively constant context. Each tree they be demographic processes, envi-
constraints sort individually since within a gap is not subject to the full ronmental constraints, disturbances,
trees differ individually with respect range of soil patterns in a watershed, or a combination of these; and (3) to
to such characteristics as fire toler- but only to the small sample of such relate this pattern to adjacent levels.
ance, drought tolerance, and ability patterns that constitute the local We have emphasized that the notions
to withstand wind. condition. of mechanistic interaction and con-
The factors producing these con- A component at a given level of a straining context explicitly involve
straints are themselves patterned on hierarchy experiences as variable only multiple levels of reference. In the
some spatial scale. Topographic pat- those patterns that are similarly next section we pursue a variation on
tern governs soil factors, insolation, scaled in rate, as well as in size. By this theme.
and moisture and defines the charac- comparison lower-level dynamics are
teristic spatial scale of these con- so fast that they are experienced as
straints. Each of these constraints average values; higher-level dynamics Consequences of landscape
provides a context for the behaviors are too slow to be experienced as
pattern
of the lower levels of the hierarchy. variable. Thus, the complexity of We have seen that the apparent com-
Nested hierarchical organization dynamics and spatial patterns at sev- plexity of landscapes can be partially
further specifies the context for low- eral scales is resolved into a few varia- resolved by decomposing them into a
er-level behaviors. Returning to the bles and a set of constants, defined hierarchical framework. It is when
example of tree species replacement relative to the reference level. one considers landscape phenomena
in gap dynamics, we might not be A powerful consequence of a hier- at different levels that the conse-
able to specify precisely which tree archical paradigm is that it allows one quences of pattern at characteristic
will come into dominance, but we can to focus on an event at a particular scales emerge. These consequences
predict that it will be a species that is scale, while recognizing that there are are far reaching; reviews by Levin
well-represented in the local seed pool other scales relevant to that event. (1976a) and Wiens (1976) illustrate
and that has not been excluded by the When describing an event, its charac- the scope of the subject. One funda-
other constraints acting on that par- teristic scale dictates an appropriate mental consequence of hierarchical
ticular site. Collectively, the various sampling scale and frequency. The structure is that events causing pat-
constraints provide a context that al- scale of an event defines the observa- tern on one scale can be incorporated
lows us to make sense of what we tional level through which the rest of into higher-level behavior (O'Neill et
observe in the forest at a given time, a the hierarchy is accessed. The next al. 1986). Through this incorpo-
contextual explanation. In general, lower level provides the components ration, effectively nonequilibrium
the more constraints we consider at of the event and its mechanistic expla- dynamics or spatial heterogeneity at
one level that are relevant to one nation. Higher levels provide the con- one scale can be translated to equilib-
criterion, the greater our predictive text that gives the event greater signif- rium or constancy at a higher level.
power. icance. These higher-level factors can There are two aspects of incorpo-
Thus, to understand a complex, be treated as constants when viewed ration that are especially pertinent to
hierarchically organized system we from the reference level, though they landscape ecology. The first concern
must consider multiple levels. The may be quite variable at larger scales. is whether the biological mechanisms
reference level is the scale on which Levels of the hierarchy further re- necessary to a pattern disturbance at
the phenomenon is witnessed as an moved from the reference level are a given scale are available in the sys-
interesting event. Once specified, the not of immediate concern, in that tem; that is, do the mechanisms exist?
event has its mechanistic explanation they contribute very little toward un- Obviously a logical consequence of

February 1987 123


patchinesswould be the evolution of over the landsscape, although not at changed the level of resolution in the
mechanisms to deal with it. Both specific local sites. Cherry wins the data, and accessed a hierarchical sys-
plants and animals have evolved di- war, thoughit loses every battle. Thus tem at two different levels (Allen et al.
versetacticsto utilizepatchilydistrib- the dynamics of regional patchiness 1984).
uted resources.One strategyfor deal- are incorporatled at a higher level. Incorporation can be passive; a dis-
ing with patchiness is to employ That species;assemblages in nature turbance is incorporated simply by
superiorcompetitiveabilityto persist can be catego*rized into types (con- increasing the scale of reference. In
within a given patch, maybe even stant assembl; ages) at a variety of the example of cherry abundance, the
exerting some control over the envi- scales (e.g., Braun 1950, Holdridge cherry does not affect the regional
ronment.An alternativestrategyis to 1967, von Humboldt 1807) suggests disturbance regime. In other cases,
concede competitive advantage by that many natlural patterns are incor- adaptive mechanisms may evolve
playingthe role of fugitiveon a larger poratedby eco)logical systems. In par- such that the disturbance regime is
scale. ticular,a disturbance regime that can actually modified. Many fire-adapted
The cherry (Prunus spp.), which be incorporate :d is not disturbing at systems evolve such that the member
requiresopen sites for regeneration,is all (Allenand .Starr1982). At a higher species produce volatile substances
a fugitive in both time and space level, disturba.nce frequency can be and serotinous seeds, which require
(Auclair and Cottam 1971, Marks viewedas a coinstraint that governs an the heat of a fire to germinate. These
1974). Cherryseedsmay be dispersed equilibrium slpecies assemblage. An species both exert some control over
great distances by birds, and at the analysisof prairie plant species com- fire frequency and capitalize on fires
site of deposition, the seeds may re- position in response to fire illustrates for episodic regeneration (Mutch
main viable for several decades. Thus, this two-level approach. Allen and 1970). This incorporation with evolu-
cherry can take advantage of an ap- Wyleto (1983:)used the "time since tionary integration has repercussions
propriate regeneration site elsewhere, fire" variable to emphasize the suc- that we will discuss later.
or it can await the recurrence of a cessional dyno amics of plant species A second aspect of incorporation
disturbance that brings an appropri- composition f(ollowing a burn. Using pertinent to landscape ecology con-
ate site to it. At the level of gap the "fire frequency" as a variable, cerns whether incorporation can be
dynamics, cherry is a loser: it is even- they demonstr;ated that different equi- realized within a particular bounded
tually replaced by longer-lived, shade- librium species assemblages were system. That is, given a geographical-
tolerant species. But the cherry main- maintained uinder different fire re- ly defined region (e.g., a park) and the
tains a nearly constant abundance gimes. In effec:t, data transformation perturbation affecting it, is the region
of sufficient scale to incorporate the
disturbance?Shugart and West (1981)
8 have addressed this idea, using a for-
I l l r est simulation model. In their simula-
V>\
\?i6s tions, individual trees suffered sto-
7 - chastic, age-related mortality, and
G>r/
each model plot (0.08-ha gap) was
(-J
c?
E 6 (\0/S> \ _ independent of other plots. They
" found that 50 model plots were neces-
Q.).x,
0
-1
/<;GS \
'/~O ~ sary to stabilize the statistical vari-
- 5
- ance in biomass associated with gap
4
LU
dynamics, that is, to incorporate gap
III dynamics using biomass as a standard.
s4 Shugart and West then compared
LU
O several bounded landscapes to the
z scale of their disturbance regimes.
43
m
cc
They defined a quasi-equilibrating
landscape as one in which the area
n 2 ratio of bounded landscape to distur-
bance regime was at least 50:1. Small-
er landscapes were called effectively
1 4- nonequilibrating(Figure4).
HED PARKS A bounded landscape that is large
0 I I I enough to incorporate the factors that
7~8 9 10 disturb its component patches has a
0 1 2 3 4 5 6
constant frequency distribution of
FOCAL AREA [LOG(m2)]
patches of all types at all times, and is
Figure 4. Classification of landscapes relative to the spatial scale of their disturbance considered to be a strictly equilibrat-
regimes,accordingto a 50:1 ratio as calculatedfor forestbiiomassdynamics(Shugart ing landscape. A smaller landscape
and West 1981). A forest stand may incorporate single treefalls but not larger that is unable to incorporate a distur-
windthrowsor fires;a smallwatershedcan incorporatetreef;allsbut perhapsnot larger bance has a transient frequency distri-
windthrows,and not wildfire;a largerparkmay incorporateall of these disturbances. bution of patch types, which changes

124 BioScience Vol. 37 No. 2


in response to each disturbance event. ANTHROPOGENIC EFFECTS ON LANDSCAPES
These are called nonequilibrating.
The implication is that the relative HUMANACTIVITY CONSEQUENCES
abundance of each patch type can be
predicted readily at the higher level, if RESCALE PATCH DYNAMICS RENDER ADAPTIVE MECHANISMS
the lower-level dynamics can be in-
LESS EFFECTIVE
corporated. In a national park that is
CHANGE CONSTRAINING RULES
large enough to incorporate wildfire,
a constant and predictable proportion ALTER PATCH INTERACTIONS
of fire-successional vegetation (per- RESCALE BOUNDED REGIONS REDEFINE FROM EQILIBRATING
haps critical habitat for some species) TO NONEQUILIBRATING STATE
is maintained. Conversely, a park that
is too small to incorporate a natural INTRODUCE NOVEL PATCHES RENDER ADAPTIVE MECHANISMS
wildfire regime does not lend itself to AND DYNAMICS LESS EFFECTIVE
straightforward predictions of the rel- REDUCE POTENTIAL FOR SPECIES
ative abundance of patch types. TO EVOLVE ADAPTIVE
The phenomenon of incorporation MECHANISMS
is a natural extension of pattern at
characteristic scales (O'Neill et al. HOMOGENIZE PATTERNS REDUCE TREE SPECIES DIVERSITY
1986). This perspective is especially THROUGH LAND USE REDUCE HABITAT DIVERSITY
illuminating when applied to man- FOR FOREST WILDLIFE
dominated landscapes.
Figure 5. Summaryof the effects of anthropogenicrescalingof natural landscape
patternsand processes.
Man-dominated landscapes
There is a tendency to view man-
dominated landscapes as being differ- change and that role may no longer bations that might differ in spatial or
ent from natural landscapes. Excel- be advantageous. Finally, small burns temporal scale from natural regimes.
lent discussions of human impacts on in a fire mosaic might depend on an For example, the spatial scale and
landscapes are available (Burgess and immediate seed rain from adjacent dynamics of human land use may be
Sharpe 1981, Forman 1981, Forman unburned forest in order to regener- different from any natural forest
and Godron 1981 and 1986, Mooney ate. Rescaling the burns to cover larg- process. Many changes in land use
and Godron 1983). We focus here on er areas might decrease the influx of cover large areas but are frequent or
man's influence on the characteristic seeds, slowing the regenerative chronic, contrary to the natural rule
scales of landscape phenomena. dynamics of the mosaic. Though of large/slow or small/fast. Man-dom-
these effects are not independent, the inated landscapes may change ac-
Effects of anthropogenic scaling. A specific actions of rescaling are (1) to cording to such nonecological factors
primary influence of man is to rescale render natural incorporating mecha- as price of commodities or transfers
patterns in time and space (Figure 5). nisms less effective; (2) to change the of land ownership. One would expect
Human control of forest fires illus- set of constraints (including distur- that such anthropogenic regimes
trates several ramifications of this re- bance frequencies) governing lower- would disrupt the natural system,
scaling. Fire suppression retards the level biotic processes; and (3) to leaving only behaviorally plastic spe-
natural frequency of burns in systems change the degree of interaction cies. This is consistent with the fre-
that have incorporated fire. When among patches, thus altering behav- quent association of generalists and
wildfires do occur as a result of fuel iors that influence higher levels. weeds with man-dominated regions.
buildup, they may escape to burn Man also rescales natural regions Man's activities at some scales may
over a larger area and at a greater by establishing new boundaries. Pipe- homogenize a forest stand's fine-scale
intensity than they would otherwise. lines, drainage canals, and roads all patterns that result from gap dynam-
A thick-barked tree that could survive set new bounds if they are effective ics. Chronic use of woodlots for graz-
a low-intensity fire might succumb to barriers to patch interactions, espe- ing or as a fuelwood source can oblit-
a hotter fire. A species with seeds that cially species dispersal. This is critical erate natural patterns in regeneration,
require episodic fires in order to re- when the scale is redefined relative to so that the entire woodlot assumes a
generate might decrease in regional the scale at which disturbances can be high degree of similarity. Such an
abundance because fire suppression incorporated. In such cases, an equili- effect can be indirectly imposed by
removes opportunities for germina- brating system may be rescaled to a natural edge effects in very small
tion. In each case, an incorporating nonequilibrating state. Forest frag- woodlots. In small woodlots, in-
mechanism has been short-circuited. ments in the eastern United States creased insolation and convection can
Under the constraint of periodic illustrate this effect. Some fragments alter the physical environment to such
burns, the role of a fugitive, like the are large enough to incorporate dis- an extent that natural gap-phase re-
cherry, might have a winning strategy turbances; most are not (Pickett and placement mediated by shade and
at the landscape scale. With fire sup- Thompson 1978). moisture is not expressed. Very small
pression, the constraining rules Man also introduces novel pertur- woodlots do not develop an interior

February 1987 125


of mesic species; they remain essen- thoughtfully cutting or planting just a as a conceptualand analyticmodel to
tially all edge (Levenson 1981). This few trees. be exercised as long as it is useful.
has an obvious effect on local and A third mode of prescriptive scaling That hierarchytheoryis conceptually
regional forest diversity (Noss 1983) proceeds from our discussion of appropriate to landscape ecology
but may have further consequences. incorporation. We suggest that an in- should be apparentfrom our discus-
The regional abundance of many for- herently nonequilibrium landscape sion. We have statedvery little that is
est birds and small mammals may that cannot incorporate its internal actually novel: we have merely re-
depend on a continual availability of dynamics can be equilibrated by re- phrasedfamiliarnotions in terms of
specific forest microhabitats (Seagle scaling its internal dynamics to effect patterns at characteristicscales. In-
et al. 1984, Whitcomb et al. 1981). smaller patches. Thus, prescribed deed, severalearlyclassics(e.g., Tans-
Man's homogenizing effect may thus burning of small patches could rein- ley 1935, Watt 1947, Whittaker
contribute to a regional decline in state fire into a park that could other- 1953) and a host of more recent
forest microhabitat specialists. wise not incorporate wildfire. In efforts (Allen and Starr 1982) em-
Man's various effects on landscape rangelands, controlled rotation of braceideas that are implicitlyhierar-
pattern are neither exclusive nor inde- grazing pressure in small paddocks chical. Hierarchy theory takes this
pendent but are typically interactive mimicks but rescales the natural re- conceptuallygood fit and pushes it
and cumulative. A forest fragment gime of far-ranging herbivores, effec- deductivelytoward new insightsinto
has an imposed size. It may have its tively creating rangeland microcosms. complex, natural phenomena. As a
component events rescaled and its It seems fruitful to attempt to general- nascent interdisciplinary endeavor,
internal patterns altered. It may be ize these familiar examples to other landscapeecology can benefitfrom a
operating under a new or rescaled set systems in managed landscapes. hierarchytheory as a unifying con-
of higher-level constraints. Each of ceptualand analyticframework.
these factors contributes to the con- Experimental landscapes. Man-domi-
founded and confounding behavior of nated landscapes can provide natural
woodlots (Burgess and Sharpe 1981, experiments from which we can learn
Acknowledgments
Curtis 1956, 1959). a great deal about ecological scaling This researchwas sponsored by the
in natural systems. Human land-use National Science Foundation's Eco-
Prescriptive scaling in land manage- patterns may be more variable than system Study Program,under Inter-
ment. A knowledge of the characteris- many natural environmental patterns, agency Agreement No. BSR-
tic spatial scale and natural frequency because human land use reflects not 8103181-A02 with the US
of patch dynamics on a landscape only natural constraints (Bowen and Department Energy, under Con-
of
lends itself to prescriptive applica- Burgess 1981) but also the financial tractNo. DE-AC05-840R21400with
tions in natural resource manage- resources and personal whims of pri- MartinMariettaEnergySystems,Inc.
ment. In general, resource manage- vate landowners. Thus, these land- PublicationNo. 2807, Environmental
ment should be scaled to mimic scapes often provide a spectrum of SciencesDivision, ORNL. We appre-
natural patch dynamics, so as to take anthropogenic patches of various ciate the helpfulcommentsof T. E H.
advantage of preselected adaptive sizes within the same area (e.g., Am- Allen, B. Milne, D. Miller, P. Del-
mechanisms in the local species pool. buel and Temple 1983, Burgess and court, D. DeAngelis,and A. King.
Foresters use prescriptive scaling Sharpe 1981, Forman et al. 1976).
when they mimic natural distur- Such landscapes can provide the nec- References cited
bances with clearcuts, a practice that essary empirical observations from
Allen, T. E H., and T. W. Hoekstra. 1984.
represents the collective wisdom of which to infer critical thresholds of Nested and non-nested hierarchies: a signifi-
generations of foresters, who have interaction among components of hi- cant distinction for ecological systems. Pages
found a successful clearcutting strate- erarchical phenomena. Specifically, 175-180 in A. W. Smith, ed. Proceedings of
the Society for General Systems Research. I.
gy through trial and error. A hierar- studies of man-modified landscapes
chical approach dictates the same may indicate how inter-patch dis- Systems Methodologies and Isomorphies. In-
tersystems Publ., Coutts Lib. Serv., Lewis-
strategy deductively. It seems likely tance, connectivity, and spatial con- ton, NY.
that the mimicry approach to re- figuration modify patch interactions Allen, T. F. H., T. W. Hoekstra, and R. V.
source management should prove to generate higher-level behaviors in O'Neill. 1984. Interlevel relations in ecologi-
useful in other land use applications. mosaics; what the minimum differ- cal research and management: some working
Another method of prescriptive ence in scale might be for a perturba- principles from hierarchy theory. USDA For-
est Service Gen. Tech. Rep. RM-110. Rocky
scaling might capitalize on the notion tion to be incorporated at the next Mountain Forest and Range Experiment Sta-
that lower-level interactions are prop- higher level; and whether these rules tion, Fort Collins, CO.
agated upward in a hierarchy to gen- are constant or vary systematically Allen, T. F. H., and T. B. Starr. 1982. Hierar-
erate higher-level behaviors. This sug- for different kinds of phenomena. We chy: Perspectives for Ecological Complexity.
University of Chicago Press, Chicago.
gests that management could be suggest that general answers to such Allen, T. F H., and E. P. Wyleto. 1983. A hier-
tailored such that a minimal amount questions will form the basis of a archical model for the complexity of plant
of management, at the proper time theory of landscape ecology. communities. J. Theor. Biol. 101: 529-540.
and place, could be amplified at high- Ambuel, B., and S. A. Temple. 1983. Area-
er levels to have maximal effect. One Conclusion dependent changes in the bird communities
and vegetation of southern Wisconsin wood-
might envision steering long-term, lots. Ecology 64: 1057-1068.
large-scale forest dynamics by The purpose of a paradigm is to serve Auclair, A. N., and G. Cottam. 1971. Dynam-

126 BioScience Vol. 37 No. 2


ics of black cherry (Prunus serotina Erhr.) in graphic islands in southeastern Wisconsin. Shugart, H. H., Jr. 1984. A Theory of Forest
southern Wisconsin oak forests. Ecol. Mon- Pages 13-39 in R. L. Burgess and D. M. Dynamics. Springer-Verlag, New York.
ogr. 41: 153-177. Sharpe, eds. Forest Island Dynamics in Man- Shugart, H. H., Jr., and J. M. Hett. 1973.
Bormann, F H., and G. E. Likens. 1979. Pat- dominated Landscapes. Springer-Verlag, Succession: similarities of species turnover
tern and Process in a Forested Ecosystem. New York. rates. Science 180: 1379-1381.
Springer-Verlag,New York. Levin, S. A. 1976a. Population dynamic models Shugart, H. H., Jr., and D. C. West. 1977.
Bowen, G. W., and R. L. Burgess. 1981. A in heterogeneous environments. Ann. Rev. Development of an Appalachian deciduous
quantitative analysis of forest island pattern Ecol. Syst. 7: 287-310. forest succession model and its application to
in selected Ohio landscapes. ORNL/TM- 1976b. Spatial patterning and the struc- the assessment of the impact of the chestnut
7759. Oak Ridge National Laboratory, Oak ture of ecological communities. Pages 1-36 blight. Environ. Manage. 5: 161-179.
Ridge, TN. in S. A. Levin, ed. Some Mathematical Ques- 1981. Long-term dynamics of forest
Braun, E. L. 1950. Deciduous Forests of East- tions in Biology. VII. Lectures on Mathemat- ecosystems. Am. Sci. 69: 647-652.
ern North America (1974 facsimile edition). ics in the Life Sciences, vol. 8. American Shugart, H. H., Jr., D. C. West, and W. R.
Hafner Press, New York. Mathematical Society, Providence, RI. Emanuel. 1981. Patterns and dynamics of
Burgess, R. L., and D. M. Sharpe, eds. 1981. 1978. Pattern formation in ecological forests: an application of simulation models.
Forest Island Dynamics in Man-dominated communities. Pages 433-465 in J. S. Steele, Pages 74-94 in D. C. West, H. H. Shugart,
Landscapes. Springer-Verlag, New York. ed. Spatial Pattern in Plankton Communi- Jr., and D. B. Botkin, eds. Forest Succession:
Clark, W. C., D. D. Jones, and C. S. Holling. ties. Plenum Press, New York. Concepts and Application. Springer-Verlag,
1978. Patches, movements, and population Marks, P. L. 1974. The role of pin cherry New York.
dynamics in ecological systems: a terrestrial (Prunus pensylvanica) in the maintenance of Simon, H. A. 1962. The architecture of com-
perspective. Pages 385-432 in J. S. Steele, ed. stability in northern hardwood ecosystems. plexity. Proc. Am. Philos. Soc. 106: 467-
Spatial Pattern in Plankton Communities. Ecol. Monogr. 44: 73-88. 482.
Plenum Press, New York. Mooney, H. A., and M. Godron. 1983. Distur- 1973. The organization of complex
Clements, F E. 1916. Plant succession: an anal- bance and Ecosystems. Springer-Verlag,New systems. Pages 3-27 in H. H. Pattee, ed.
ysis of the development of vegetation. Carne- York. Hierarchy Theory. George Braziller, New
gie Institute Publ. 242, Washington, DC. Mutch, R. W. 1970. Wildland fires and ecosys- York.
Curtis, J. T. 1956. The modification of mid- tems: a hypothesis. Ecology 51: 1046- Sousa, W. P. 1984. The role of disturbance in
latitude grasslands and forests by man. Pages 1051. natural communities. Annu. Rev. Ecol. Syst.
721-736 in W. L. Thomas ed. Man's Role in Naveh, Z., and A. S. Lieberman. 1984. Land- 15: 353-391.
Changing the Face of the Earth. University of scape Ecology: Theory and Application. Tansley, A. G. 1935. The use and abuse of
Chicago Press, Chicago. Springer-Verlag, New York. vegetational concepts and terms. Ecology
1959. The Vegetation of Wisconsin: Noss, R. F. 1983. A regional landscape ap- 16: 284-307.
An Ordination of Plant Communities. Uni- proach to maintain diversity. BioScience 33: von Humboldt, A. 1807. Ideen zu einer Geo-
versity of Wisconsin Press, Madison. 700-706. graphic der Pflanzen nebst einem Naturge-
Delcourt, H. R., P. A. Delcourt, and T. Webb Odum, E. P. 1969. The strategy of ecosystem milde der Tropenlander. Bey F. G. Cotta,
III. 1983. Dynamic plant ecology: the spec- development. Science 164: 262-270. Tiibingen, FRG.
trum of vegetation change in space and time. O'Neill, R. V., D. L. DeAngelis, J. B. Waide, Watt, A. S. 1925. On the ecology of British
Quat. Sci. Rev. 1: 153-175. and T. F. H. Allen. 1986. A Hierarchical beech woods with special reference to their
Forman, R. T. T. 1981. Interactions among Concept of the Ecosystem. Princeton Univer- regeneration. II. The development and struc-
landscape elements: a core of landscape ecol- sity Press, Princeton, NJ. ture of beech communities on the Sussex
ogy. Pages 35-48 in S. P. Tjallingii and A. A. Pattee, H. 1973. Hierarchy Theory. George Downs. J. Ecol. 13: 27-73.
deVeer, eds. Perspectives in Landscape Ecol- Braziller, New York. 1947. Pattern and process in the plant
ogy. Pudoc, Wageningen, the Netherlands. Pickett, S. T. A., and J. N. Thompson. 1978. community. J. Ecol. 35: 1-22.
1983. An ecology of the landscape. Patch dynamics and the design of nature Whitcomb, R. E, C. S. Robbins, V. . Lynch,
BioScience 33: 535. reserves. Biol. Conserv. 13: 27-37. B. L. Whitcomb, M. K. Klimkiewicz, and D.
Forman, R. T. T., A. E. Galli, and C. F Leck. Pickett, S. T. A., and P. S. White, eds. 1985. Bystrak. 1981. Effects of forest fragmenta-
1976. Forest size and avian diversity in New The Ecology of Natural Disturbance and tion on avifauna of the eastern deciduous
Jersey woodlots with some land use implica- Patch Dynamics. Academic Press, Orlando, forest. Pages 125-205 in R. L. Burgess and
tions. Oecologia 26: 1-8. FL. D. M. Sharpe, eds. Forest Island Dynamics
Forman, R. T. T., and M. Godron. 1981. Reiners, W. A., and G. E. Lang. 1979. Vegeta- in Man-dominated Landscapes. Springer-
Patches and structural components for a tion patterns and processes in the balsam fir Verlag, New York.
landscape ecology. BioScience 31: 733-740. zone, White Mountains, New Hampshire. White, P. S. 1979. Pattern, process, and natural
1986. Landscape Ecology. John Wiley Ecology 60: 403-417. disturbance in vegetation. Bot. Rev. 45:
& Sons, New York. Risser, P. G., J. R. Karr, and R. T. T. Forman. 230-299.
Gleason, H. A. 1939. The individualistic con- 1984. Landscape ecology: directions and ap- Whittaker, R. H. 1953. A consideration of
cept of the plant association. Am. Midi. Nat. proaches. Illinois Natural History Survey climax theory: the climax as a population
21: 92-110. Special Publ. No. 2. Illinois Natural History and pattern. Ecol. Monogr. 23: 41-78.
Heinselman, M. L. 1973. Fire in the virgin Survey, Champaign. Whittaker, R. H., and S. A. Levin. 1977. The
forests of the Boundary Waters Canoe Area, Romme, W. H., and D. H. Knight. 1982. Land- role of mosaic phenomena in natural com-
Minnesota. Quat. Res. 3: 329-382. scape diversity: the concept applied to Yel- munities. Theor. Popul. Biol. 12: 117-139.
Holdridge, L. R. 1967. Life Zone Ecology. lowstone Park. BioScience 32: 664-670. Whyte, L. L., A. G. Wilson, and D. Wilson,
Tropical Science Center, San Jose, Costa Seagle, S. W., H. H. Shugart, and D. C. West. eds. 1969. Hierarchical Structures. Elsevier
Rica. 1984. Habitat availability and animal com- Science Publ., New York.
Koestler, A. 1967. The Ghost in the Machine. munity characteristics. ORNL/TM-8864, Wiens, J. A. 1976. Population responses to
Macmillan Publ., New York. Oak Ridge National Laboratory, Oak Ridge, patchy environments. Annu. Rev. Ecol. Syst.
Levenson, J. B. 1981. Woodlots as biogeo- TN. 7: 81-120.

February 1987 127

You might also like