Transcript Slide 1

Assembly Rules
Jeff Ott and Jackie White
Early history of assembly rules
Gotelli (2000)
• Diamond (1975)
– Argued that interspecific competition among bird species in the
Bismark Archipelago resulted in observable community
assembly rules, e.g.
• forbidden species combinations
• checkerboard distributions
• incidence functions
• Conner and Simberloff (1979)
– Used null model analysis to show that many of Diamond’s
patterns could also arise in competition-free communities
assembled by random colonization
• Triggered a debate that has lasted 25+ years
Null Models
Sp. 1 Sp. 2 Sp. 3 Sp. 4 Sp. 5 Sp. 6 Richness
Plot 1
1
0
1
0
1
0
3
Plot 2
0
1
1
0
0
0
2
Plot 3
1
0
1
1
1
0
4
Plot 4
1
0
1
1
0
1
4
Frequency
3
1
4
2
2
1
• Tokeshi principle: “the null model must include all features of
the observed data except the one it is intended to test
• Questions:
– How important are null models in ecology?
– Do all patterns need to be tested against a null model to be “proven”?
Broadening of assembly rule definitions
• Filtering of the regional species pool to produce
a local community (Belyea and Lancaster 1999)
– Geographic filters
– Environmental filters
– Biotic filters
• Filtering of species traits due to environmental
constraints (Keddy and Weiher 1999)
• The dynamic process by which communities
develop (Booth and Swanton 2002)
Assembly rules:
Wilson and Agnew’s definition
• “Restrictions on the observed patterns of species
presence or abundance that are based on the presence
or abundance of one or other species or groups of
species (not simply the response of individual species to
the environment)”
–
–
–
–
Emphasizes pattern, not process
Most likely to apply to equilibrium communities
Ignores within-species genetic variation and phenotypic plasticity
Null models required
• Questions:
– Can species interactions and environmental responses really be
treated as isolated topics?
– Is environmental characterization of species distributions really
the “easy task of community ecology” and therefore
uninteresting?
Ruling out environmental variation
• Is variance in richness different from a null model?
•
Example 1
– Yes, if you look across the entire sample
– No, if you look within specific environments
Alternative question:
Is richness correlated with environment?
•
Example 2
– Yes, whether across entire sample
or within environments
– However, quadrats within environments
are pseudoreplicated: only one quadrat
allowed per environment… True???
Alternative question:
Is composition correlated with environment?
Ruling out environmental variation
• Is variance in richness different from a null model?
•
•
Example 1
– Yes, if you look across the entire sample
– No, if you look within specific environments
Alternative question:
Is richness correlated with environment?
Example 2
– Yes, whether across entire sample
or within environments
– However, quadrats within environments
are pseudoreplicated: only one quadrat
allowed per environment
Alternative question:
Is composition correlated with environment?
(Pseudoreplication: lack of independence
among sample units, e.g. in space or time)
Patch model
Questions:
Is Wilson and Agnew’s approach for ruling out environmental variation adequate?
Are there other ways besides the “patch model” for mitigating pseudoreplication?
Ruling out environmental variation
• Is variance in richness different from a null model?
•
Example 1
– Yes, if you look across the entire sample
– No, if you look within specific environments
Alternative question:
Is richness correlated with environment?
•
Example 2
– Yes, whether across entire sample
or within environments
– However, quadrats within environments
are pseudoreplicated: only one quadrat
allowed per environment… True???
Alternative question:
Is composition correlated with environment?
Zonation on gradients
The work of Whittaker and other ecologists has
suggested that species boundaries are randomly-spaced
rather than clustered along environmental gradients.
Are there theoretical reasons to continue searching for
clustered boundaries on gradients?
Species sorting
• Hypothesis 1: Species associations will become stronger over the
course of succession as random processes are replaced by
deterministic ones.
– Pioneer stage: random or weak negative associations
– Building stage: some positive and negative associations
– Mature stage: strong associations, mostly negative
• Evidence for:
– Tussock grasses after burning
• Evidence against:
– Sand quarries (negative associations less common over time)
– Canyon walls (greater dissimilarity in frequently-disturbed patches than
undisturbed—relevant?)
• Uncertain
– Tropical rain forest (non-random throughout succession?)
– Pastures in which negative associations increased, but the total number
of significant associations decreased over time
Species sorting
• Hypothesis 2: Species composition will converge over
time in similar conditions (“It would be fascinating to see
how similar species assembly was in similar conditions.”)
Table 5.x. Dry mass (g m-2) of selected species in six replicate
plots in the experiment of Crawley et al. (1999).
– If conditions were truly identical, would the outcome of
succession (species sorting) also be identical?
– What factors in natural communities seem to be most
responsible for preventing successional convergence?
Limiting Similarity
•
Hypothesis: If two species are similar in resource use patterns then one will
be excluded unless character displacement occurs.
•
Hubbell (2005): “The empirical evidence, in general, has not borne out
these predictions… particularly in plant communities… does limiting niche
similarity for functional groups exist?... I believe that the answer is no (at
least in plants)
•
Wilson and Agnew: Hubbell (2005) was too dismissive. Limiting similarity in
plant communities can be demonstrated.
•
Examples:
–
–
–
–
Rooting depth in desert and fybnos (Cody 1986)
Leaf shape/length in fybnos (Cody 1986)
Pollinators in tropics (Armbruster 1986, 1994)
Multiple characters in herbaceous riverside vegetation (Weiher et al. 1998)
and sand dunes (Stubbs and Wilson 2004)
– Flowering phenology (Stiles 1977 etc.)
– Vegetative phenology (Parrish and Bazzaz 1976 etc.)
•
Conclusion: greater support for limiting similarity due to vegetative
characters (e.g. rooting patterns, leaf water control) than reproductive
characters (e.g. pollinators, flowering phenology)
Guild Proportionality
•
Pianka (1990) – guilds are groups of species that compete more with each
other than with species in other guilds.
– same α niche
•
•
Result: relative constancy in the proportion of species from each guild
Within guild differences (“Micro-niches”) are assumed
– Not simply the abundance of individuals within a guild
•
Why species and not abundance? Why would we expect to see more than
one spp per guild?
Guild Proportionality Tests
• Ratio of variance in guild proportionality observed with
that expected under the null model
• Patch and site null models
• Test statistic is relative variance in guild proportionality
(RVgp)
Evidence: constancy in space
RV >1 is likely attributed to environmental heterogeneity
Evidence?
YES
• Salt Marsh
• Inter-tussock
vegetation
• Subalpine meadow
• Lawn
NO
• Beech forest
• Riparian habitat
MAYBE
• Dune slack
– In forb layer
Is the scale relevant?
• Scale (4 x 4 cm, 10 x 10 cm, 13 x 13 mm)
• “We deliberately started from the smallest scale that
seemed ecologically sensible: 1 x 1 m. Smaller scales
would not be conceptually realistic in a forest, since
extrapolation of the quadrat into the canopy would be
meaningless.” Bycroft et al. 1993
• If significant guild proportionality is observed at
these scales, is this evidence of a relevant
assembly rule in plant communities? What scale
is relevant? At what scale do communities
occur?
Evidence: removal experiments
• If species from one guild are removed then, in order to
maintain proportionality, any increase in species will
likely be from the same guild
• Is this a logical argument? What other outcomes might
we expect? Will the outcome depend on the type of
species removed or the time since removal?
• Herben et al 2003:
– Removed Festuca rubra
– Grass biomass increased more than dicots
• Symstad 2000
– Removed 3 guilds
– Additions after 3 years showed only weak evidence of resident
guilds excluding similar species
Removal experiments
• Fargione et al 2003
– Addition experiments at Cedar Creek
– Resident guilds had an inhibitory effect on invasion by
its own guild
• Von Holle and Simberloff 2004
– Removal of species from riparian plots
– Additions
– No effect
Conclusion: removal experiments give little
evidence for guild based assembly rules
Evidence: Successional
convergence
Fukami et al. 2005 – convergence in trait
groups
Guilds -- Conclusions
• Is there support for guild proportionality?
• What do Wilson and Agnew conclude
regarding the existence of guilds and guild
proportionality?
• Why is guild proportionality generally only
observed in the herb layer?
Intrinsic guilds
• A priori guild selection is arbitrary
– Subjective selection
– Classification by trait groups
• Solution: ‘interview the plants’
– Maximize RVgp -- Distributional data
• Assume there is guild proportionality
– Maximize mean RYM – Interference experiments and
the “Jack Spratt effect”
– Response to removal
• Evidence of stochastic assembly
• Benefit is that this technique can fail to find
significant guilds
Intrinsic guilds
• What do you think about the method for
determining intrinsic guilds? Can all
species realistically be assigned to two or
three guilds?
Texture Convergence
• Functional characteristics
• “in comparable habitats in different areas, whilst
the actual species present may be different, the
texture may be the same.”
• Mean texture or distribution of functional traits
Texture
• Wilson et al 1994
– Wooded fen communities
– Texture – traits related to light capture
– Convergence in PSU width and area was observed
when species were weighted by photosynthetic
biomass
– Each community has representation from the range of
functional characteristics
– ‘Strong evidence that species are being sorted by
their characters…for their entry into the community’
Time
• No-analogue communities
• Combinations of climatic variates differ
• Community misfits occur in same place/time as climate
misfits
• What is the relevance of this section?
Enviro
factor
B
Time 1
Existing
environments
Enviro
factor
B
Time 2
Existing
environments
Niche of
Species X
Species Y
Environmental factor A
Environmental factor A
Abundance
• Biomass constancy per unit area
– What about biomass constancy in guilds?
• Relative abundance distribution (RAD)
– Niche pre-emption model (Whittaker 1965) –
competition
– Broken stick (MacArthur 1957) – random assignment
of resources
– “fits to an model based on an ecological theory would
be most interesting, and usually ambiguous…’
Abundance
• Sparse species
– Are there special rare niches being filled?
– Zobel et al 1994 – no difference in the number
of invasions
– Do rare species have a distinct effect?
– Lyons and Schwartz 2001 – invasion of exotic
species was greater after removal of rare
species
Keystone species
• Vital for maintenance of the community
• Greatest effect on others relative to its
biomass
Aspen, Saguaro, and Figs
Exotic species
• Is invasion surprising?
– Native species should be more adapted to the
local environment
– Species should not be intrinsically different
• All spp are native somewhere
• Drivers or passengers of change?
• 5 explanations for successful invasion
Alternative hypotheses
• Depauperate flora
– Implies missing guilds
• Weak competitors
– Sub-optimal guild members
• Invaders are r species
– More disturbed habitats
• Escape from natural enemies
• Co-evolution
• Why is invasion limited in tropical forests?
Exotic establishment and
community assembly
• How exotic species assemble when they reach a new territory
• Wilson et al. 2000 – Roadside communities
– Two distinct groups in UK and NZ
• Community assembly by pre-adaptation
• Only strong environmental filtering is able to reassemble
communities (Calcareous grasslands)
• Can this provide insight into assembly
rules?
• Are there different assembly rules for
different communities?
Conclusions
• There are no real conclusions
• Plants interact and species differ
– Must be limitations to coexistence
Fig. 5.16: Profile through a part of the Botany Lawn.
• SUMMARY QUESTIONS:
Did any sections of this chapter seem
irrelevant or only tangentially related to the
topic?
Which assembly rules were best
supported?
What are the major conclusions and are
you convinced?