Download Agents of Pattern Formation: Biotic Processes

Agents of Pattern Formation: Biotic Processes
Instructor: K. McGarigal
Assigned Reading: Turner et al. 2001 (Chapter 4); Watt (1947)
Objective: Provide an overview of biotic processes operating within the physical template at
various scales as agents of pattern formation. Highlight importance of biotic interactions and
feedback.
Topics covered:
1. Watt's unit pattern
2. Demographic processes as agents of pattern
3. Interactions and Feedbacks
4. Implications for conservation planning
5. Implications for climate change
Comments: Some material taken from Dean Urban’s Landscape Ecology course notes, Duke
University.
6.1
1. Foundations of the Pattern and Process Paradigm
Much of the way we approach pattern in ecology can be traced to seminal papers by Alex Watt
(1925, 1947), who in many ways gave us the "pattern-process paradigm" that we use today.
Watt based much of this notion on gap dynamics as observed in forests, in which a tree-scale
area undergoes a more-or-less predictable trajectory from gap, to old field, to young forest, and
so on, until a large tree dies and the cycle repeats. His insight was as follows: if one were to
look at a large area of forest, various gap-scale elements would be undergoing similar cycles of
gap dynamics, but independently. Averaged over a large area, there would be some elements in
all stages of succession represented (somewhere) at any given time, and further, the relative
abundance of these stages would reflect the duration of each phase.
6.2
Watt (1947) postulated the notion of a unit pattern:
The "unit pattern" is the full representation of the pattern in all its phases (the `community in
harmony with itself and its environment').
Moreover, Watt suggested that the phases of the community (seral stages, in his cases) should be
present in relative abundances corresponding to the duration of each seral stage: a strict
translation of temporal dynamics into pattern on the ground.
6.3
He further observed:
“departures from this phasic equilibrium either in space or in time could then be measured
and correlated with the changed factors of the environment”.
This is often heralded as one of the earliest and most lucid translations of temporal dynamics into
spatial pattern. This insight into the value of the unit pattern as a benchmark or reference state
against which to compare real patterns also represents one of the earlier neutral models in
ecology.
6.4
There have been several other variations on this theme:
1. The "climax pattern" (Whittaker 1953) is a steady-state pattern of a variety of community
types across a landscape, reconciling Clementsian and Gleasonian notions of vegetation
pattern.
2. The "shifting mosaic steady-state" (Borman and Likens 1979) is a stationary distribution of
successional stages, wherein fine-scale elements undergo gap dynamics while the aggregate
forest appears constant.
6.5
Each of these models shares two main heuristic points
1. First, demographic processes, acting in and of themselves, should generate a pattern with
predictable characteristics.
2. Secondly, this expectation might be useful as an interpretive model. That is, the expectation
could serve as a reference standard and departures from this, observed for real systems, could
be interpreted as interesting deviations.
6.6
2. Demographic Processes as Agents of Pattern
Any and all of the basic demographic processes (dispersal, birth or establishment, growth, and
mortality) can contribute to vegetation or ecosystem pattern. Some issues that may be relevant in
some cases:
1. Establishment.--Small plants see a different environment than big plants. For example,
seedlings and herbs experience light as filtered through the canopy, and soil moisture in
the topsoil, in contrast to these variables as experienced by a large tree. Thus, while the
constraints may have the same names they are not the same variables; moreover, they
may vary on quite different spatial and temporal scales.
2. Growth.--Growth is the focus, often only implicitly, of much of gradient analysis and
interspecific competition studies (which thus might ignore other demographic processes).
Response to direct physical factors (temperature, pH) may differ from that for resource
gradients (light, water, nutrients) (Austin and Smith 1989). For direct physical factors, a
Shelfordian response model may be appropriate (i.e., ranging from "not enough" through
"just right" to "too much"), while for resource gradients a more one-sided response may
be the rule (i.e., from "plenty" to "not enough"). Moreover, most (all?) species would be
expected to show qualitatively similar responses to resource gradients, which thus
6.7
invokes competition as the primary mechanism for sorting species along resource
gradients (Smith and Huston 1989 and see below).
3. Mortality.--Mortality often reflects multiple factors (the `death spiral'). For example,
trees suppressed by drought or shading might be more likely to succumb to insect attack.
Mortality patterns may be in response to chronic stress or episodic events, in which case
mortality may be explained by the mean environmental conditions or by the variance in
environmental conditions.
4. Dispersal.--Dispersal can limit response to physical gradients, by preventing a species
from occupying potentially usable resources. Dispersal might also act as a pattern
amplifier, reinforcing existing (or initial) patterns through the positive feedback that
species already present will contribute to the seed rain in the future. Which effect occurs
might depend on the scale of dispersal versus the grain of the physical environment (see
Interactions, below). That is, if dispersal distances are long relative to the grain of the
physical environment, then dispersal should distribute species across a variety of
microsites, allowing chance establishment events and founder effects to mediate the
effects of competition. Conversely, if dispersal distances are comparatively short then
local seed rain might amplify species response to the environment. (This is somewhat
idle speculation, given our rather limited understanding of how dispersal interacts with
environmental heterogeneity. In fact, most studies of dispersal have made the simplifying
assumption that the environment is homogeneous.)
6.8
3. Interactions and Feedbacks
3.1. How do biotic processes themselves interact?
Watt's notion of the unit pattern coupled demographic processes to generate spatial pattern.
Others have done this more explicitly. For example, Shugart (1984, 1987) categorized tree
species in terms of their mode of mortality (does/does not create a gap on dying, a function
mostly of size) and regeneration mode (requires/does not require a gap to regenerate). This
pair of dichotomies defines four functional roles of trees:
Role 1: needs gap and creates gap;
Role 2: does not need gap but creates gap;
Role 3: needs gap but does not create gap; and
Role 4: does not need gap and does not create gap.
Shugart noted that the coupling of mortality and regeneration could lead to some interesting
feedbacks that would generate characteristic patterns in abundance of the various functional
roles:
1. Roles 1 and 4 are self-reinforcing strategies, creating conditions on dying that favor
6.9
their own regeneration.
2. Role 2 creates conditions that favor other species; i.e., it "gives its advantage away."
3. Role 3 cannot create the conditions it needs to regenerate, and so must depend on
other agents for persistence.
Shugart suggested that environmental conditions that affected modes of mortality or
regeneration would lead to characteristic combinations of roles of trees. For example,
frequent disturbance would favor species that require gaps (especially role 3). Conversely,
low sun angles at high latitudes would not favor gap-phase regeneration but rather, would
favor roles 2 and 4. These are patterns due to interacting biotic processes overlaid onto the
physical template.
One of the most useful implications of these ideas is the expectation that biotic processes
should generate characteristic patterns in species abundance and distribution. Reciprocally,
we would like to be able to infer generating processes by observing spatial patterns. But this
goal is difficult to achieve.
6.10
3.2. How do biotic processes interact with the physical template?
A number of classic efforts in plant ecology have addressed the way that demographic
processes interact with the physical template to generate pattern. Among them:
1. Tilman (1982, 1985) posed the resource-ratio hypothesis:
<
<
<
Species vary in their relative growth rates in response to levels of resource
abundance;
The species best adapted to local resource levels "wins" in competition;
Spatial heterogeneity in these resources can generate patterns in species diversity.
6.11
2. Grime (1977, 1979) proposed that there are three primary plant strategies: competitor,
stress-tolerant, and ruderal (ephemeral), reflecting basic trade-offs in life-history traits.
These strategies play out to yield characteristic patterns of abundance:
<
<
<
Competitor: continuously abundant, but depressed under low resource levels (in time
or space);
Stress-tolerant: continuously rare, displaced spatially to resource-poor sites;
Ruderal: temporarily abundant, displaced in time by more competitive species.
6.12
3. Smith and Huston (1989) used a simulation model to explore the interactions between
demographic processes and the physical template. Their basic premise: life-history
strategies entail compromises that dictate competitive abilities under varying resource
levels. Generally, a plant can be competitive under high levels of resource availability or
tolerant of low levels (stress), but these are exclusive adaptations (a plant cannot do both
at once). Specifically, a plant can:
<
<
<
be competitive for light, in which case it will not do well in the shade;
be tolerant to drought (or low nutrients), in which case it won't be competitive under
more favorable (higher) resource levels;
be competitive under high light and high moisture (nutrients), but not the converse
(tolerant of drought and shade).
So there are three "pure" strategies, which form the corners of a triangle of possible
strategies. They used hypothetical species of varying strategies to illustrate the
implications of these strategies as they play out along environmental gradients in
available water (in space) and light (in successional time). Their simulation results
suggested that:
6.13
<
<
<
The species that grew best under high levels of light and/or moisture "won" on
favorable sites.
More tolerant species "lost" (were out-competed) on good sites but won on poorer
(drier, shady) sites--not because they achieved their own best growth there, but
because they persisted there when more competitive species could not.
Because life-history traits are correlated, the same explanation that accounts for
gradient response also accounts for succession (a gradient in time): species that are
shade-intolerant may also be drought-tolerant, and so may be early successional but
may persist on dry sites (because shade-tolerant species tend also to be
drought-sensitive, and so do not come to dominate dry sites).
6.14
3.3. Interactions: An Illustration
Simple spatial models can be used to illustrate the complex interactions among competition and
gradient response as modified by local dispersal. The examples shown are generated with a
cellular automaton called MetaFor (Urban; see Green 1989 for a similar model). The model
includes a physical template and the biotic processes of competition and dispersal, and is
configured so that each of these factors can be switched on or off. The model simulates forest
pattern as a raster map of tree-sized cells, in which each cell may be occupied by a single
species. The examples shown here are based on forest pattern in the Sierra Nevada, but have
been abstracted to emphasize general agents of pattern formation.
The physical template..--The abiotic environment is simulated as an elevation gradient in
which higher elevations are cooler and wetter and the lower elevations are warmer and drier;
alternatively, the gradient can be set to constant intermediate (mid-elevation) conditions.
6.15
Competitive relationships.--The model includes three species:
<
White fir (Abies concolor, or ABCO) dominates the middle elevations and is the most
competitive of the three species;
< Lodgepole pine (Pinus contorta, PICO) is less competitive but more cold-tolerant
than the other two species;
< Ponderosa pine (Pinus ponderosa, PIPO) is less competitive than white fir but more
drought tolerant, and is characteristic of the warmer and drier lower elevations.
6.16
Seed dispersal.--Effects of local dispersal are implemented as a simple neighborhood
function, in which the probability of a given species establishing in an empty cell is biased
by the abundance of that species in an 8-cell window around the empty cell.
Each time step during a simulation, a cell may become empty as a result of mortality (a
probability based on species longevity). Empty cells may be occupied by a species, in which
event the actual establishment probability for each species is the product of the neighborhood
effect and the environmental response of that species. Cells that do not "die" become one time
step older. Simulations were conducted for a 2x2 factorial of Gradient/No Gradient crossed with
Neighborhood Effects/No Neighborhood Effects. Each simulation was for a 100x100-cell grid,
and ran 500 years. Cover maps for the final year illustrate the spatial patterns generated by each
combination of agents.
6.17
The clumping effect of the neighborhood dispersal is visually obvious in these maps. What is
less obvious is the effect on the relative abundance of the three species. The interaction between
dispersal and competition along the gradient can be illustrated by ‘sampling’ the grid and
graphing the relative abundance of each species in each row of the map. Note that in the ‘no
gradient’ simulations, ‘elevation’ is merely map position.
In the simplest case of no gradient and no local dispersal, the relative abundances of each species
simply reflect their relative competitive abilities, i.e., ABCO > PIPO >> PICO. Competition
effectively excludes PICO from the map in the case with no gradient and with local dispersal.
The gradient in the absence of neighborhood effects allows some sorting of species along the
gradient, but there is considerable overlap in species distributions. Adding neighborhood effects
to the gradient allows PICO to not only persist but to dominate the higher elevations as PIPO
dominates the lower elevations; the environmental sorting of species is greatly amplified.
These simulations illustrate that simple biotic processes, interacting among themselves and with
the physical template, have the potential to generate rather complex patterns across landscapes.
This rich complexity generated from extremely simple processes should temper inferences made
of patterns observed on real landscapes.
6.18
4. Implications for Conservation Practice
One implication of Watt's unit pattern is the expectation that a landscape at equilibrium with its
environment and with its internal biotic processes should exhibit a stationary distribution of
patch area in various types; i.e., if the unit pattern is realized, the relative abundances of patch
types should remain constant through time. This stability is very much central to practical issues
in conservation biology related to the 'minimum size of ecosystems' that could be self-sustaining.
Because disturbances--natural or anthropogenic--are often responsible for perturbing systems
away from equilibrium or steady-state, we will return to this notion later in the course.
The simulation study of Smith and Huston has some interesting implications for conservation
practice. If their results are general, then the most effective conservation strategy might be
different on physically homogeneous, mostly mesic landscapes as compared to physically
heterogeneous landscapes with some less favorable sites:
•
•
On homogeneous, mesic sites the strategy would be to manage in time, maintaining a
variety of stand ages, because succession would lead to less diversity through
competitive exclusion.
On heterogeneous sites, the strategy would be to manage in space because a variety of
physically contrasting sites will support (and maintain) more diversity.
6.19
5. Implications for Climatic Change
The results of Tilman, Grime, and Smith and Huston all argue that, because competition
displaces species from their optimal sites (fundamental niche) to sites where they can persist
under competitive pressures (realized niche), we have a limited ability to infer the actual
environmental tolerances of species by observing their realized distributions. This undermines
naive efforts to predict species responses to climatic change simply by translating species
distributions according to a shifting environmental template: we must also consider competitive
relationships and how these might change in a changing environment.
6.20