Re: [R-sig-eco] logistic regression and spatial autocorrelation

Hi, to weigh in on this: @Aitor, Harrell's rules of thumb are assuming independent predictors without any fancy covariance function. To model the covariance of the residuals you are now estimating extra 2nd order parameters from the data, so even more data is needed to stabilize the parameter esti

Re: [R-sig-eco] Making dynamite plots

Hi, The best route is not to do it. the following web sites explain it better than I can in a short email, and have example code http://biostat.mc.vanderbilt.edu/wiki/Main/DynamitePlots http://emdbolker.wikidot.com/blog:dynamite Nicholas > From: ZAID M MCKIE-KRISBERG > To: r-sig-ecology@r-proje

[R-sig-eco] question about appropriate model for abundance studies

Hi Sacha, A few things: 1) If counts are large, a normal error structure may be fine as an approximation, large counts means conditional on the covariates, the counts in each cell are relatively 'large'. if there are not many 0's a log transformation might be appropriate. This will be a lot e

[R-sig-eco] MCMCglmm and power analysis

Hi Krista, You might look at posterior predictive p-values and Bayes factors, However, you are mixing your metaphors so to speak. You can't really talk about power in a Bayesian context, It really is a frequentest concept dealing with classical error rates. You can think about how much information

[R-sig-eco] post-treatment measurements as covariable?

Hi Melissa, I think the problem is confounding. If you do not have pre and post measurements of soil moisture, you cannot really fully model the effect of warming on on growth vs the effect of drying, without making some assumptions. Though it would seem to me, rusty though my botany is, that inc

[R-sig-eco] (no subject)

Hi Manuel, I am guessing the problem is that because you have categorical predictors, you are getting empty cells in your cross validation sets, and hence infinite coefficients. Unfortunately, you are now in a very tricky situation, to get at the generalization error of your model you need to have

Re: [R-sig-eco] meaning of extreme outliers in bootstrapped statistic

Hi Tom, You have 45 observations, and remember the bootstrap samples with replacement. If you have a few extreme data points it is entirely likely that some of your parameters may have some extreme cases that arise from rare bootstrap samples where the extreme data appears more than once, especial

Re: [R-sig-eco] Manova repeated measures

Hi, Yep, you seem to have more behaviors than observations. "Error in anova.mlm(object) : residuals have rank 70 < 72" is the clue. On top of that be aware that MANOVA has very strong assumptions about the covariance structure. You may want to look into a mixed modeling approach. A solution is

[R-sig-eco] (no subject)

Hi, Just to add a note to the good advice you have gotten so far. Redundancy Aanalysis (RA) is a linear method, as Jari explained. If you apply transformations until your data conform to the assumptions of RA and then do RA, you are no longer applying a linear method. You will get back a configura

Re: [R-sig-eco] Proper Contrasts for an Ordered ANOVA

Hi Joseph, I think you are making things a bit more complicated than they need to be. You have 4 levels of instar as treatment and 2 presumably (-)correlated responses algae and zooplankton. You can assume you know something about the spacing of levels and try and fit a linear or quadratic contrast

Re: [R-sig-eco] lme post-hoc help

our replicates in each of those categories. The metric I used > was the hourly % loss of a tethered set of algae. Basically, I set the > tether out, came back an hour later, and quantified the percent loss > in terms of mass. > > > > On Mar 4, 2010, at 10:32 AM, Nicholas L

Re: [R-sig-eco] lme post-hoc help

d set of algae. Basically, I set the > tether out, came back an hour later, and quantified the percent loss > in terms of mass. > > > > On Mar 4, 2010, at 10:32 AM, Nicholas Lewin-Koh wrote: > > > Hi Nathan, > > I don't use SPSS, so I can't commen

[R-sig-eco] lme post-hoc help

Hi Nathan, I don't use SPSS, so I can't comment on what it is doing, but if you look at the bottom of the output from multcomp it says:(Adjusted p values reported -- single-step method) What that means is that multcomp is adjusting for the fact that you are doing six comparisons. So a quick and di

[R-sig-eco] clumping vs. random

Hi Kristen, Are you are just interested in "clumpiness" in time, or "clumpiness" in space and time? Are your sites far enough apart that the environmental process you are looking at can be assumed independent? If you have the spatial coordinates, even approximate, you could calculate Moran's I for

[R-sig-eco] Low counts

Hi Tore, What your seeing in the residuals may just be due to the "discreetness" of count data. Gordon Smyth has a nice paper on this topic (and code in the statmod package): Dunn, P. K., and Smyth, G. K. (1996). Randomized quantile residuals. Journal of Computational and Graphical Statistics 5, 23

Re: [R-sig-eco] Time series and GLS

Hi Lisa, The point of including time in the model is that your data are ordered in time. If there is reason to believe that food availability at year t has no influence on food availability in year t + 1 than all you need is success~food. But alas ecology is rarely that simple and I would guess t

Re: [R-sig-eco] R-sig-ecology Digest, Vol 22, Issue 3

Hi, To add to all the good advice you have gotten, I would second Scott's advice about using a GAM. Two issues that come to mind are that the interpretation might be tricky as population, I assume geographical unit, is your "Experimental unit" and individual nests are within the populations so flu

Re: [R-sig-eco] low predicted vales in GAMs

Hi Anna, A couple of thoughts: Did you try fitting a straight Poisson model? The quasi Poisson model is assuming the variance is not a strict function of the mean, and that may be interacting with your weighting function. Also how exactly are the weights defined? is it (# squares counted)/(total po

Re: [R-sig-eco] creating cluster with between points arbitrary distances

Hi Alex, Well in the example you show, almost any distance based clustering method would identify the relationship, from your example dd<-dist(cbind(x,y)) cuttree(hclust(dd),2) However, what I think you are after is simply thresholding. So if x are your bird coordinates, then, library(igraph) x<-m

[R-sig-eco] RE: Plot a polynomial regression

Hi, Doing that with xyplot is a bit of a PITA, but try: fits<-fitted(model) xyplot((GPP)~(Discharge), groups=Site, data=nighttimeds1.dat, fits=fits, panel=function(x,y,fits,...){ panel.xyplot(x,y,...) panel.lines(x,fits) },col=c("red","blue","green","yellow"

Re: [R-sig-eco] R-sig-ecology Digest, Vol 19, Issue 2

No No No No No! The log likelihood of the Poisson and the Gaussian are not comparable. One is a discrete distribution and the other continuous, you can get into all sorts of trouble there and not just pathological cases. They are on totally different scales. You need to make a decision if you wan

Re: [R-sig-eco] Removing the effect from a qualitative variable into a GAM

Hi Clement, I am not sure exactly what you are proposing. Is there any data or is this all simulated? My first question is why is years ordinal? It would seem that richness would vary smoothly from year to year. The second question is why is spatial location lumped in with gradient? wouldn't a "bet

[R-sig-eco] Subject: Re: missing data in PCA

Hi, It's not quite so simple, at what stage do you recombine your samples. Recombining the covariances into an "average" covariance from the imputed samples has different implications than recombining the projected data. I think that in terms of pca uncertainty the projection directions is the issu

[R-sig-eco] correlated parameters, absolute model fit, and MCMC

Hi Dan, Since you are fitting a fully Bayesian model, the output from your MCMC sample is a sample from the joint posterior distribution. So given that your chain is mixing properly and converged (a whole nother kettle of fish) I don't see why the correlation is a problem. You can assess the model

[R-sig-eco] About data distribution model

- > > Message: 1 > Date: Thu, 18 Dec 2008 16:52:46 +0530 Hi, If your counts are relatively high, you might start with a log-normal or gamma distribution. What you are talking about here are species abundance distributions on which the

[R-sig-eco] Classical statistics in R

Hi, I have yet to see a book that was a good introduction to statistics. I have always felt that before going into anything about inference, discussion should center on what is an experimental unit, and what is an observational unit. Then incorporate notions of likelihood, and probability. S&R and

[R-sig-eco] Analyzing hemispheric photos in R? (opheliaw...@mail.utexas.edu)

Hi Ophelia, You are working in an area where R is pretty weak. You will have to write your own code to do this, however there are a few packages that support image analysis. They are EBImage on bioconductor biOps rimage Some of the Spatial packages like RGDAL and RGrass might help as well. You c

[R-sig-eco] Interpreting large clusters

Hi Phil, Can you plot the dendorgam in pieces? if you look at example(hclust) the show an example where they plot the dendrogram showing 10 aggregated groups from the centroids of a previous dendrogram cut at 10 clusters. You could also do the opposite using as.dendrogram example(hclust) hcd<-as.d

[R-sig-eco] (no subject)

Hi, Mark Taper, Subhash Lele, and I organized an ESA invited session around this topic at the 98? ESA meeting in Baltimore MD. Which resulted in Mark L. Taper and Subhash R. Lele (eds): The Nature of Scientific Evidence: Statistical, Philosophical, and Empirical Considerations University of Chic

[R-sig-eco] polar plot mappng

Hi Fraser, I believe the package plotrix has a polar plot function. For circular statistics there are the packages CircStats, and circular on CRAN. Those should have what you need. Nicholas > -- > > Message: 3 > Date: Tue, 8 Jul 2008 14:35:16 -0400 > From: Fraser Smith

Re: [R-sig-eco] Calculating EC50s

Hi, Short answer: fit a 4 parameter logistic model using nls in base or use the package drm. The parameter names differ, but the third parameter is the ec50. Longer Answer: Need more information from you to give you a longer answer. Nicholas PS: I changed the title of this thread. Something info

[R-sig-eco] Why use Rpy?

Hi Phil, The main reason would be for accessing R functionality in broader applications. Python does use a more efficient memory model than R, however when using R functions through rpy, R will still make copies with assignment within called R functions, so I am not clear that there is any gain on

[R-sig-eco] zip models in R

What about the packages COZIGAM and ZIGP and the orphan package in the archives, zicounts? Nicholas > > Message: 2 > Date: Tue, 10 Jun 2008 11:45:44 +0200 > From: "Dr. Cornelia B. Krug" <[EMAIL PROTECTED]> > Subject: [R-sig-eco] zip models in R > To: r-sig-ecology@r-project.org > Message-ID: <[

Re: [R-sig-eco] Publication quality graphics in R

Hi, following this thread I have seen several misunderstandings that I think should be cleared up. Firstly, we should be careful what is meant by "publication quality", on interpretation is for a particular journal, a good resolution graphic in the format they require. In general, the meaning ref

Re: [R-sig-eco] nlme model specification

Hi, Last month, or so, Doug was talking about comparing models using likelihood ratio tests, anova(m1,m2) and pointed out that the way things are calculated in lmer the ML and REML estimates are equivalent. I am not sure if this is because the bias in the REML estimates cancels out or if the estim

Re: [R-sig-eco] beam forming

Hi Stephen, Can you be a bit more specific, are the peaks already marked and you want to find all pairs of peaks distance k appart? or is it that the peaks are unknown and the signal/noise needs to be considered as well? I know Tibshirani and Hastie have a package for peak probability contrasts but

Re: [R-sig-eco] R-sig-ecology Digest, Vol 2, Issue 4

Hi Melanie, It has been a long time since i looked at gravity models but if I remember correctly, they are formulated as log linear models with a particular weight matrix. I would guess that that could be written as a generalized linear mixed model, and fit with lme, or lmer, though if a paramet