Forgot your password?
Document Actions
test file for sharing
Day 3 ideas and discussion issues:
From CW Osenberg (I took these for my own use, but figured they might be useful; this file also contains the notes from the “traits” group).
Start off morning discussion:
EcoPath parameterization for a well studied system; generate predictions and compare to experimental
Bring in someone with a hideously complex model already (e.g., everglades) and use this as the “reality” model from which we can then explore simplications and rules of thumb, and generate “experimental” data.
existing models or foodweb descriptions: bascompte, roughgarden and goldwasser (Caribbean), Jennifer dunne, Ythan Estuary (emmerson is working with this).
I see two projects:
How well do simple models to in predicting responses to foodweb perturbations? How much complexity is needed? What's the importance of particular features…?
What are rules of them for identifying “important” species? Take a model and then perturb it. What rules identify the key taxa? Compare rules of thumb with 1) dn/dt or dn/ndt (initial transient); 2) dN* and do these both for 3) small perturbation dNj vs. 4) larger perturbations (e.g., removal of Nj).
(Others expressed interest in a system with mobile predators to look at this connecvity, patch choice, resource issues.)
We need to differentiate b/w magnitudes of change (“big effect”) vs. the specific pattern of change (i.e., removing X will cause a huge change relative to removing Y, but we don't know the final state: vs. we can say how the system will change). It's possible that we can say something more concrete about magnitudes, but that the direction is really indeterminate.
If we take the “who are the key players” question and define it in terms of change to the system, then we could direct managers to focus on a couple of key players. Thus, ES management could be focused on “an old style population based management approach”: i.e., define ES impact but then manage via focus on key species.
How do you anticipate surprises…or gather data to best anticipate surprises? (Jim Estes). Or use data to predict effects of future changes.
Andy - wants us to focus more on ecosystem function/services and their response.
So, what data do you need to do this exploration? Approach?...
REVIEW PAPERS or META-ANALYSES we could do/write:
case studies. E.g., kelp forests and observed changes.
what are the key traits of strong interactors and phylogenetic distributions of traits? Or what are the features of systems that allow for strong effects.
are surprises common in well studied system? Include disease effects… e.g., if you manage well, but then disease outbreaks….and messes up the system…where are you? This changes with historical context…old surprises are now part of the expectation.
predator impacts on ecosystem services (not novel response parameters).
…Take TWO:
Ecopath vs. Loop analysis : can a detailed simulation vs. a quality approach predict responses of Kelp forest or Intertidal to perturbations? But some questioned the +/- approach to loop analysis - is that really helpful? You could use a very qualitative loop analysis approach to pinpoint the key interactions that you might have to quantify to get a more specific answer (e.g., big change vs. small change).
allometric scaling and community matrix approach - mark and tim have been discussing this.
spp-level indicators - I think this is essentially the “what are traits/features of key species”: e.g., mobility, size, trophic level, omnivory, connectedness, mean aij, etc. see other notes above.
other possible people to involve: rich Williams (neo martinez). Marrisa basket (student from levin's lab; interviewed UCDavis). Carl Walters (ecospace).
other systems: kluane national park hare (krebs system), gulf of maine, carpenter lakes.
Brose et al. 12-13 coauthors who provided data on body size. In Ecology 2005 (eco archives data paper)
Specialization and its effects:
lag times in switches to prey as densities change
covariance in handling efficiency and searching effort; effect of averaging - how do you characterize the population (a “super” otter vs. an “average” otter) if there are trade-offs or costs of generalization?
how does specialization change the intensity of intraspecific competition? It might actually increase it, if foraging is more effective for specialists. Contrast with bolnick paper?
could you compare the degree of feeding trade-offs and how it affects competition if you assume certain resource dynamics?
BREAK OUT GROUP:
TRAITS of key species that determine if they are strong interactors….
Key papers: Power (keystone spp); Borer (trphic cascade traits); Menge 1994 (keysteons);
Topics of relevance:
direct vs. indirect effects
mutualisms.
Can we examine the traits of keys species in different environments and therefore determine if the importance of the trait is context dependent? Does this just mean we've identified the wrong traits (if it was the right trait, then it would'nt be context dependent) or we've got an Importance*Trait interaction?
Spp vs. functional groups
Species traits |
Community Traits |
Ecosystem Traits |
Metabolic efficiency: assimilation efficiency, production; endo/ecotherm. |
Diversity |
Physical stress (as perceived by the focal sp or the system?). What is it? “Wave exposure”, tidal height? |
Mobility |
Diversity at the same trophic level of focal species |
Spatial and heterogeneity |
Abundance |
Food web structure, connectivity (e.g., number of two links; some weighting by aij): some of these are more like “traits of species”. How do you weight different connections (e.g., two steps away) - what do people report and is this info comparable? |
Temporal heterogeneity (perhaps scaled to generation time of the focal species) |
Reproductive output/recruitment, or open vs. closed |
Functional redundancy/dominance: can this be quantified by comparing the vector of ai1 vs. ai2 (i.e., effect of 1 on all other species) - how similar is sp 1 compared to other specie or the most similar species in the system. How can this be obtained? Expert option, but possible bias from the results of the expt. |
Habitat (e.g., kelp, intertidal, mudflat, oceanic) |
Spatial distribution pattern (i.e., - cov in space with responses of interest) : natural experiments? - perhaps this goes with “traits of ecosystem (spatial heterogeneity) |
|
Temporal variability in abiotic factors - could link this to habitat (e.g., intertidal undergoes more variation in abiotic conditions than subtidal) |
Body size |
|
Temperature and other abiotic variables |
Diet breadth (number of links to prey) |
|
Large scale geomorphology, tectonics |
Ecosystem engineer/foundation spp |
|
|
Vert/invert |
|
|
Sensory system (visual, olfactory) |
|
|
Novelty: invaders, specialists, mutualists, etc. (related to functional redundancy) |
|
|
Low R* |
|
|
Defensive traits. Could this be quantified by mean aij connecting to the next higher trophic level; but aren't some key species not defended (i.e., they transmit effects rather than function as dead ends?): i.e. flow through species vs. dead ends. |
|
|
Longevity/ storage/ diapause |
|
|
Diet Plasticity |
|
|
Reproductive mode: semel/itero; fraction of lifetime that it's reproductive. |
|
|
Abundance, biomass |
|
|
Approach that could be taken:
Look for experiments that have manipulated a species and quantified response of community. Compare similarity of system +/- spp.
But can we get the above information for systems in which these expts have been done?
VS. taking a modeling approach (i.e., construct foodwebs, parameterize it, and similulate perturbations), then ask what the relationship is b/w traits and effects. Problems: existing models might only include consumptive links; dynamics may not well match “real” system; …
I'm a bit concerned about the dual approach here - although they are complimentary. The MA focuses on species that WERE manipulated (in DIFFERENT systems) and we are therefore exploring effects for spp that were thought to be important. In contrast, if we use a set of models to ask the question, then we can perturb all species and investigate their role (but with little link to real responses). I suppose there's also an issue of comparison within a system (how do you identify the strong interactor WITHIN a system) vs. between a system (i.e., how do you identify the SYTEMS with strong effects and only secondarily the traits of the SECIES within those systems). This seems very important to clarify.
Day 4: Traits
Take an example system:
Tim w: birds in intertidal: cages to exclude birds (but there are 3 species…); responses: snails increase, algae respond, other sessile species change,
RESPONSE VARIABLES:
Dissimilarity in community composition (of a guild, trophic level, total community)
Change in biomass or abundance (or specific taxa, prey, “prey of prey”, or total community)
Variance (spatial or temporal - can this be extracted?: e.g., transformations, scaling with mean, no data on individual plots through time, etc.)
APROACH:
Meta-analysis using existing experiments (manip and natural) looking at how effect sizes vary with “traits”. We'll need to deal with the inference limitations since many spp within a com were not manipulated (see caveats above). Perhaps the issue here is more about what we can say about the role of system traits rather than the spp traits?
Modeling effort in which we perturb each species, quantify response of community and relate this response to “traits”. Do this for different food web structures. The challenges are:
if we care about biological traits (e.g., body size, metabolic rate), then how well does the model really match “real” responses? If not, then the traits we “find” may not have much to do with the actual traits that we can use to identify key species. We can overcome this by ignoring bio traits and instead focusing on food web traits: e.g., connectivity, mean direct effect, mean consumptive rate, mean vulnerability to preds, two-link species, etc.
the models are “simple”. How do we incorporate reasonable complexities, like prey switching, size structure, nonlinearities, etc.
MA: start with Menge's review of indirect effects (1997; Am Nat). Ben, Craig, Fio start the effort. We need to find a student/tech to do the extractions. Ben will check on the logistics of doing this via nceas.
Model effort: will depend on the kelp/intertidal project. I also will try to explore other models that might exist that could be used.