Steinmann, Linder, and Zimmermann. 2009. Modelling plant species richness using functional groups. Ecological Modelling
In the ever-elusive goal of better plant richness modelling, these folk take on the challenge by using various higher order divisions in their models, to see what shakes out best for modelling plant species richness across Switzerland. They mainly attempted a middle-path sort of route- not a bottom-up, model the individual plant responses to environmental gradients and then overlay all those maps together (a common approach which completely disregards community interactions), and not a top-down, community model which doesn't model individual species responses (an obvious deficiency).
They found little benefit to their approach, which was sort of a let-down- but, in another way, refreshing. Non-results are rarely published, which is a shame. There was an interesting facet in that some functional groups were modeled far better than others, indicating some interesting differences (but, one of the best were trees, which usually model well because of their longevity and, therefore, sensitivity to environmental factors/relative immunity from stochastic events).
Rocchini, Chiarucci, and Loiselle. 2004. Testing the spectral variability hypothesis by using satellite multispectral imagery. Acta Oecologica
My interest in the spectral variability hypotheses (also described in an earlier post about Palmer, et al 2002) stems mainly from my dislike of classified imagery as such. It eliminates so much information from the image, and locks you into whatever errors are present at the time of the classification. The SVH is a way around it, but relatively unexplored. Rocchini and others seem to be the only group working on the idea, and they always use high spatial resolution stuff (they also have a kicking dataset, like Steinmann et al, which apparently they get mostly from other people. Lucky.). They've published a few more papers on the idea, but this 2004 is the first, so... start at the beginning.
They explain ~50% of the species richness at a 1 ha scale (over landscapes, of course), which is pretty impressive, since it's all done without looking at the individual spectral responses of the plants. The image is ground-truthed, of course, but only by count; i.e. anybody can do that, without any special tools. Plus, this is with Quickbird, so it's only four bands, and I'd bet good money you can improve on that 50% really easily with some higher spectral resolution stuff. At 30 m Landsat? That's a good question, and I'm sure it really depends on the scale of vegetiation present. People have used NDVI for this sort of thing via Landsat (Gould 2000), but I'm unaware of any attempts at full-on variability style methodologies.
Monday, June 15, 2009
Friday, June 12, 2009
VP free
Variance partitioning
In an attempt to re-understand VP, I’m going to attempt to explain it to myself in word-form…
Variance, in regressions, is always present. R2 values are given to illustrate the amount of variance the model (regression) explains. Multiple regressions use several variables (x1, x2, etc) in an attempt to explain the response variable (y); each x contributes some “explanatory power” to the model. That’s stuff we know already…
How much does each x contribute to the explanatory power of the model? Well, first note that the order of insertion of independent variables into the model doesn’t change the final result (x1b + x2b + x3b…). Intermediate results will differ, but the final result is the same. Secondly, the contribution (in terms of increasing r2, if that’s our scorecard) of the second variable entered will change depending on what was entered first. Not only that, some variables which don't add anything to r2 might still be important, as they could mediate the effect of another variable in some important way. Overlapping disturbances would be a candidate.
Basically, it appears that you look at the change in r2 when each independent variable is added last, and then call that difference the “unique variance.”
Nothing super new so far, it’s the commonality analysis which seems like it’ll be useful- is it possible to look at the influence of A on B even if A adds nothing to an explanatory model of B, but mediates the effect of variable C on B?
Perhaps it works this way: A -> B <- C
Or this: C -> A -> B <- C
In the second case, A would appear, in a regression, to add nothing to a predictive model of B, but perhaps it’s mediating the effect of C in some way. Since they are correlated, it would be difficult (impossible?) to separate the two influences specifically, but perhaps it would be possible to get a measure of the importance of C via subtraction…
VARIATION PARTITIONING OF SPECIES DATA MATRICES: ESTIMATION AND COMPARISON OF FRACTIONS
Pedro R. Peres-Neto1, Pierre Legendre, Stéphane Dray, and Daniel Borcard
Ecology 87(10)
In an attempt to re-understand VP, I’m going to attempt to explain it to myself in word-form…
Variance, in regressions, is always present. R2 values are given to illustrate the amount of variance the model (regression) explains. Multiple regressions use several variables (x1, x2, etc) in an attempt to explain the response variable (y); each x contributes some “explanatory power” to the model. That’s stuff we know already…
How much does each x contribute to the explanatory power of the model? Well, first note that the order of insertion of independent variables into the model doesn’t change the final result (x1b + x2b + x3b…). Intermediate results will differ, but the final result is the same. Secondly, the contribution (in terms of increasing r2, if that’s our scorecard) of the second variable entered will change depending on what was entered first. Not only that, some variables which don't add anything to r2 might still be important, as they could mediate the effect of another variable in some important way. Overlapping disturbances would be a candidate.
Basically, it appears that you look at the change in r2 when each independent variable is added last, and then call that difference the “unique variance.”
Nothing super new so far, it’s the commonality analysis which seems like it’ll be useful- is it possible to look at the influence of A on B even if A adds nothing to an explanatory model of B, but mediates the effect of variable C on B?
Perhaps it works this way: A -> B <- C
Or this: C -> A -> B <- C
In the second case, A would appear, in a regression, to add nothing to a predictive model of B, but perhaps it’s mediating the effect of C in some way. Since they are correlated, it would be difficult (impossible?) to separate the two influences specifically, but perhaps it would be possible to get a measure of the importance of C via subtraction…
VARIATION PARTITIONING OF SPECIES DATA MATRICES: ESTIMATION AND COMPARISON OF FRACTIONS
Pedro R. Peres-Neto1, Pierre Legendre, Stéphane Dray, and Daniel Borcard
Ecology 87(10)
Haven't been posting, but have been reading...
This morning-
Palmer, M., et al. 2002. Quantitative tools for perfecting species lists. Environmetrics. 13:121-137
Zimmermann, N., et al. 2007. Remote sensing based predictors improve distribution models of rare, early successional and broadleaf tree species in Utah. Journal of Applied Ecology. 44:1057-1067
Palmer, M., et al. 2002. Quantitative tools for perfecting species lists. Environmetrics. 13:121-137
Zimmermann, N., et al. 2007. Remote sensing based predictors improve distribution models of rare, early successional and broadleaf tree species in Utah. Journal of Applied Ecology. 44:1057-1067
Wednesday, January 21, 2009
Ecosystem Values
Social Goals and the Valuation of Ecosystem Services. Costanza 2000
A good read, although I disagreed with most things. There are a few interesting statements which I think are worth considering.
First, the paper makes the arguement that valuation of ecosystem services is inevitable, and all choices can be evaluated via projection of relative value (i.e. we choose x over y, therefore x is more valuable than y). I disagree. That is the dominant economic paradigm of our society (both now and historical). But it's not impossible to think of other thought paradigms where this inevitable tradeoff doesn't occur. It's not that choices are not made, but they are not seen in the same light. Therefore, the thought of "inevitable valuation" doesn't even occur. For example, if you don't care between two outcomes, than there's no assumption of value necessary. Or, if you are at state A and are choosing to either transform A to B or leave A alone, you could just consider it a transformation, not a tradeoff.
Building Lego's (should I build an airplane or a boat?) doesn't require a choice between the two based on value, because the inherent thing isn't changed, just altered. The assignment of value is a Western paradigm, and sure, it can be applied to every situation. But so could a paradigm based on choosing whatever option was mentioned first, or options whose title comes first in the alphabet. Those are silly, but they are still options.
I am annoyed with this "let's take this paradigm for granted," because once you start down the valuation path, ecosystems lose. Maybe not today, maybe not tomorrow, but soon... they lose. Today, wetlands purify our water cheaper then we can. But tomorrow, we invent a new technology which purifies water cheaper. All of a sudden, those wetlands have lost value. And putting astetic value (in terms of money) on ecosystems, or even moral value, will only fail in the long run- conservation is a luxury of the rich, as many have said, and people will always choose food over furry friends (and let's be honest, bringing the whole world to our level of luxury won't happen. even if it does, we'll want more, and the problem will begin all over again).
Later, the author says "valuation ultimately refers to the contribution of the item to meeting a specific goal or objective," and "one cannot state a value without stating the goal being served." So without a goal, something has no value? This is a very user-centered view of the world, which is what got us into this mess in the first place.
He argues "the decisions we make, as a society, about ecosystems imply valuations." Sure, according to the paradigm of "everything is a tradeoff between two values." But I think that thought process is what gets us into trouble in the first place. The philosophy of assigning values moves easily into saying "well, this place is nice, but I can make money building condos. I'll just donate to the Sierra club and conserve through them." Thinking in terms of valuation leads to only thinking in the present tense (and one time step ahead). It's all now- you choose between two options now.
Perhaps thinking in terms of place would work better. Thinking in terms of place doesn't assign value (although the valuation paradigm can be imposed on it, of course). Thinking in terms of place assigns consequences. It's not inherently a tradeoff between a human affected environment and a "pristine natural environment," it's just the environment and how it's changed. This leads to longer term thinking- if I do x, and want to do y later, is that possible? How should I structure x so that y is possible in 10 years?
I (being a good Western thinker in the tradition of the liberal arts) assign value unconsciously, as do most people around me. But our society has REALLY screwed up the planet we're supposed to steward- so perhaps that paradigm needs rethinking. I assign great moral value to "unspoiled lands," although even that term implies valuation. But somebody else sees unlimited development, or great beauty in a golf course. The vaguries of human perceptions of beauty are such that eventually, all land will be in control of somebody that thinks it will look better as a community rec center, and our development is far easier carried out then undone. If we rely on valuation to save landscapes, we're going to lose.
A good read, although I disagreed with most things. There are a few interesting statements which I think are worth considering.
First, the paper makes the arguement that valuation of ecosystem services is inevitable, and all choices can be evaluated via projection of relative value (i.e. we choose x over y, therefore x is more valuable than y). I disagree. That is the dominant economic paradigm of our society (both now and historical). But it's not impossible to think of other thought paradigms where this inevitable tradeoff doesn't occur. It's not that choices are not made, but they are not seen in the same light. Therefore, the thought of "inevitable valuation" doesn't even occur. For example, if you don't care between two outcomes, than there's no assumption of value necessary. Or, if you are at state A and are choosing to either transform A to B or leave A alone, you could just consider it a transformation, not a tradeoff.
Building Lego's (should I build an airplane or a boat?) doesn't require a choice between the two based on value, because the inherent thing isn't changed, just altered. The assignment of value is a Western paradigm, and sure, it can be applied to every situation. But so could a paradigm based on choosing whatever option was mentioned first, or options whose title comes first in the alphabet. Those are silly, but they are still options.
I am annoyed with this "let's take this paradigm for granted," because once you start down the valuation path, ecosystems lose. Maybe not today, maybe not tomorrow, but soon... they lose. Today, wetlands purify our water cheaper then we can. But tomorrow, we invent a new technology which purifies water cheaper. All of a sudden, those wetlands have lost value. And putting astetic value (in terms of money) on ecosystems, or even moral value, will only fail in the long run- conservation is a luxury of the rich, as many have said, and people will always choose food over furry friends (and let's be honest, bringing the whole world to our level of luxury won't happen. even if it does, we'll want more, and the problem will begin all over again).
Later, the author says "valuation ultimately refers to the contribution of the item to meeting a specific goal or objective," and "one cannot state a value without stating the goal being served." So without a goal, something has no value? This is a very user-centered view of the world, which is what got us into this mess in the first place.
He argues "the decisions we make, as a society, about ecosystems imply valuations." Sure, according to the paradigm of "everything is a tradeoff between two values." But I think that thought process is what gets us into trouble in the first place. The philosophy of assigning values moves easily into saying "well, this place is nice, but I can make money building condos. I'll just donate to the Sierra club and conserve through them." Thinking in terms of valuation leads to only thinking in the present tense (and one time step ahead). It's all now- you choose between two options now.
Perhaps thinking in terms of place would work better. Thinking in terms of place doesn't assign value (although the valuation paradigm can be imposed on it, of course). Thinking in terms of place assigns consequences. It's not inherently a tradeoff between a human affected environment and a "pristine natural environment," it's just the environment and how it's changed. This leads to longer term thinking- if I do x, and want to do y later, is that possible? How should I structure x so that y is possible in 10 years?
I (being a good Western thinker in the tradition of the liberal arts) assign value unconsciously, as do most people around me. But our society has REALLY screwed up the planet we're supposed to steward- so perhaps that paradigm needs rethinking. I assign great moral value to "unspoiled lands," although even that term implies valuation. But somebody else sees unlimited development, or great beauty in a golf course. The vaguries of human perceptions of beauty are such that eventually, all land will be in control of somebody that thinks it will look better as a community rec center, and our development is far easier carried out then undone. If we rely on valuation to save landscapes, we're going to lose.
Climate change and range shifts
Predicting climate-induced range shifts: Model differences and model reliability. Lawler et al, 2006
Modeling! Modeling in R, which is more impressive. This is a great paper- not in terms of being particularly interesting, or revolutionary, but in terms of usefulness. There needs to be more papers like this, a straightforward comparison between modeling approaches using the same data, so relative errors, types of errors, etc can all be directly seen. The researchers take 100 different mammal species present in and around Brazil, and model their current distribution based on nine climate variables and one land cover variable- basically the inputs for most climate change scenarios. Then, they compare the results to actual distributions. Fun.
There are some interesting results. Random forest models were far and away the best. Genetic algorithms were second place, although they had problems with errors of commisson. Other models, like classification trees, GAMs, etc did relatively worse. I'd hypothesize that random forests did better specifically because they sampled from the 10 inputs with replacement, meaning they randomlly added weights to the various factors, so there was an added feature that neural networks and the other machine learning algorithms didn't have.
But, like all models, they have their problems, which follow.....
Prediction of plant species distributions across six millenia. Pearman et al, 2008.
This is a more applied study, where Pearman and friends attempt to model tree species in Europe. For a twist, they model from 6000 BP to now, 6000 BP only, and now only, in an attempt validate future predictions (using past climate records and pollen finds). Their main finding was that some species are modeled quite well, and others not so well- the ones which came out winners are dominant competitors for light. This makes perfect sense, and highlights the main limitation of this study, Lawler's, and most other species distribution models I've seen (and they readily admit to this limitation as well). Light competition is the only way in which most plants experience biotic constraints on their distribution. Most plants don't interact directly, especially these trees, beyond the race for good spots in the canopy. Thus, species which are competitively dominant won't really experience biotic controls on their distribution, because wherever they live, they win. Species which are not competitively dominant will experience this biotic limit to their range, and therefore models which don't take biotic factors into account won't be as accurate.
A second limition is both assume that species are at current equilibria with respect to their distribution. But the Johnstone article (reviewed earlier) indicates that some species are still expanding their range, invalidating this assumption.
Pearman also mentionsthe possibility of rapid niche shifts, which I don't know much about, but which would also invalidate predictions into the future.
Modeling! Modeling in R, which is more impressive. This is a great paper- not in terms of being particularly interesting, or revolutionary, but in terms of usefulness. There needs to be more papers like this, a straightforward comparison between modeling approaches using the same data, so relative errors, types of errors, etc can all be directly seen. The researchers take 100 different mammal species present in and around Brazil, and model their current distribution based on nine climate variables and one land cover variable- basically the inputs for most climate change scenarios. Then, they compare the results to actual distributions. Fun.
There are some interesting results. Random forest models were far and away the best. Genetic algorithms were second place, although they had problems with errors of commisson. Other models, like classification trees, GAMs, etc did relatively worse. I'd hypothesize that random forests did better specifically because they sampled from the 10 inputs with replacement, meaning they randomlly added weights to the various factors, so there was an added feature that neural networks and the other machine learning algorithms didn't have.
But, like all models, they have their problems, which follow.....
Prediction of plant species distributions across six millenia. Pearman et al, 2008.
This is a more applied study, where Pearman and friends attempt to model tree species in Europe. For a twist, they model from 6000 BP to now, 6000 BP only, and now only, in an attempt validate future predictions (using past climate records and pollen finds). Their main finding was that some species are modeled quite well, and others not so well- the ones which came out winners are dominant competitors for light. This makes perfect sense, and highlights the main limitation of this study, Lawler's, and most other species distribution models I've seen (and they readily admit to this limitation as well). Light competition is the only way in which most plants experience biotic constraints on their distribution. Most plants don't interact directly, especially these trees, beyond the race for good spots in the canopy. Thus, species which are competitively dominant won't really experience biotic controls on their distribution, because wherever they live, they win. Species which are not competitively dominant will experience this biotic limit to their range, and therefore models which don't take biotic factors into account won't be as accurate.
A second limition is both assume that species are at current equilibria with respect to their distribution. But the Johnstone article (reviewed earlier) indicates that some species are still expanding their range, invalidating this assumption.
Pearman also mentionsthe possibility of rapid niche shifts, which I don't know much about, but which would also invalidate predictions into the future.
Thursday, January 15, 2009
RS and N (no weeds)
Canopy nitrogen, carbon assimilation, and albedo in
temperate and boreal forests: Functional relations
and potential climate feedbacks
S. V. Ollinger, et al.
Another remote sensing paper, and it continues on with my recent readings in foliar nitrogen. The researchers use AVIRIS (and a little Hyperion, where necessary) to estimate foliar nitrogen levels over a few landscapes in temperate/boreal forests scattered over Canada. I received this paper after emailing S. Ollinger himself about some AVIRIS work he's doing, and he graciously sent me this along with a few thoughts... but back to the paper. They found a strong relationship between measured foliar N and reflectance across most of the NIR spectrum- cool, because that means you can get foliar N estimates from other (read: free) sensors. To quote directly:
"Collectively, these results suggest that we already have a basis
for detecting variation in %N and CAmax of forest canopies at
continental scales by using scaled relationships with albedo
and/or simple measures of NIR reflectance obtained during the
peak period of the growing season."
Then, they tie the N measurements to carbon fixation, using flux towers (through BOREAS sites I assume; I believe that's where the imagery is). All well and good.
My thoughts, to tie into the previous post, are about using this foliar N to potentially address biodiversity levels, potential for invasion, and perhaps even invasion monitoring (I don't know a lot of biogeochemistry, so hopefully I don't say something dumb at this point, but I'll try anyway). Would higher biodiversity sites (in the understory: say vascular plants, or invertebrates, or whatever) also have higher foliar N in the overstory? I would imagine so, given simple things like nutrient availability, more potential niches, etc. Would invasive species be reflected in the overstory N (either via nitrogen fixation themselves, like Scotch broom or sea buckthorn, or by exclusion of other nitrogen fixing plants, see previous post)? To speculate further (what fun!), what would the edge of higher nitrogen zones look like? If you look at the map in the paper, there's a lot of heterogeneity in the foliar nitrogen levels across forests- so... could any sort of meta-population/meta-community dynamics be happening across that edge?
temperate and boreal forests: Functional relations
and potential climate feedbacks
S. V. Ollinger, et al.
Another remote sensing paper, and it continues on with my recent readings in foliar nitrogen. The researchers use AVIRIS (and a little Hyperion, where necessary) to estimate foliar nitrogen levels over a few landscapes in temperate/boreal forests scattered over Canada. I received this paper after emailing S. Ollinger himself about some AVIRIS work he's doing, and he graciously sent me this along with a few thoughts... but back to the paper. They found a strong relationship between measured foliar N and reflectance across most of the NIR spectrum- cool, because that means you can get foliar N estimates from other (read: free) sensors. To quote directly:
"Collectively, these results suggest that we already have a basis
for detecting variation in %N and CAmax of forest canopies at
continental scales by using scaled relationships with albedo
and/or simple measures of NIR reflectance obtained during the
peak period of the growing season."
Then, they tie the N measurements to carbon fixation, using flux towers (through BOREAS sites I assume; I believe that's where the imagery is). All well and good.
My thoughts, to tie into the previous post, are about using this foliar N to potentially address biodiversity levels, potential for invasion, and perhaps even invasion monitoring (I don't know a lot of biogeochemistry, so hopefully I don't say something dumb at this point, but I'll try anyway). Would higher biodiversity sites (in the understory: say vascular plants, or invertebrates, or whatever) also have higher foliar N in the overstory? I would imagine so, given simple things like nutrient availability, more potential niches, etc. Would invasive species be reflected in the overstory N (either via nitrogen fixation themselves, like Scotch broom or sea buckthorn, or by exclusion of other nitrogen fixing plants, see previous post)? To speculate further (what fun!), what would the edge of higher nitrogen zones look like? If you look at the map in the paper, there's a lot of heterogeneity in the foliar nitrogen levels across forests- so... could any sort of meta-population/meta-community dynamics be happening across that edge?
Remote Sensing and Weeds
Remote analysis of biological invasion and biogeochemical change
Gregory P. Asner and Peter M. VitousekMore weeds, this time over to Hawaii for my favorite tree, Myrica faya (I believe it was recently renamed, because I learned it as M. faya and then was told there was a new name, but can't remember it). One of my favorites because the wood isn't super dense, and it's hackable with a machete when doing invasive species control.
Regardless, it's a nitrogen fixing invasive, and the Asner/Vitousek study looks at using foliar N as measured via AVIRIS to detect M. faya prior to it's domination of the canopy. It works, quite well. And for another note, they also found they could detect another invasive, a ginger (and not a fun thing to attempt to control), via water content measurements from AVIRIS- and it's an understory plant. This is a slight vindication for me, since I heard that something like this was possible from Greg in Hawaii, but when taking remote sensing back in Washington later, was told that sensing anything below the canopy wasn't possible. Awsome, it's in print.
This has me thinking: Scotch broom is another invasive species, nitrogen fixing, present in the west and up into BC and Alaska. While I don't think it makes it to the canopy, I wonder if you can detect it's presence via overstory N levels, especially since foliar N seems to be pretty low in most of our forests. The next review has a paper which also inspires that thought....
Subscribe to:
Posts (Atom)
