Intro to spind

Sam Levin, Gudrun Carl, Ingolf Kuehn

2017-04-07

spind is a package dedicated to removing the spectre of spatial autocorrelation in your species distribution models (hereafter referred to as SDMs). It contains many of the tools you need to calculate probabilities of occurence and/or abundances, assess model performance, and conduct multimodel inference for 2-D gridded datasets using methods that are robust to spatial autocorrelation.

The theory underlying the use of GEEs, WRMs, and many of the other tools in this package is covered elsewhere in the literature, and for the purposes of this vignette, we assume that you have already read those papers. If you haven’t, citations are included in the footnotes of this vignette as well as in the documentation of each function. We also assume that you have a working knowledge of how to use R. This vignette will focus on demonstrating how to utilize this package to create an SDM and assess its accuracy. Along the way, we will use a couple different data sets to examine how these functions work and investigate how one might use them to create a robust SDM.

Generalized Estimating Equations (GEEs) for species distribution modeling 1

This package utilizes the functions already written for GEEs from the packages gee2 and geepack3 and adapts them for easy use in the context of an SDM. Let’s start with a fairly simple GEE using the simulated musdata data set included in the package.

data(musdata)
data(carlinadata)

# Examine the structure to familiarize yourself with the data
?musdata
head(musdata)

?carlinadata
head(carlinadata)
# Next, fit a simple GEE and view the output
coords<-musdata[ ,4:5]

mgee<-GEE(musculus ~ pollution + exposure, family="poisson", data=musdata,
      coord=coords, corstr="fixed", plot=TRUE, scale.fix=FALSE)


summary(mgee, printAutoCorPars=TRUE)
#> 
#>  Call: 
#> GEE(formula = musculus ~ pollution + exposure, family = "poisson", 
#>     data = musdata, coord = coords, corstr = "fixed", plot = TRUE, 
#>     scale.fix = FALSE)
#> --- 
#>  Coefficients: 
#>             Estimate  Std.Err z value  Pr(>|z|)    
#> (Intercept) -1.90475  1.31091 -1.4530 0.1462252    
#> pollution    3.36216  0.91416  3.6779 0.0002352 ***
#> exposure    -1.46348  0.88010 -1.6629 0.0963410 .  
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> --- 
#>  QIC:  1139.159 
#> --- 
#> Autocorrelation of GLM residuals 
#>  [1]  0.685338504  0.509680590  0.363021118  0.247398654  0.144726020
#>  [6]  0.084220961  0.050228656  0.022369044 -0.001985639 -0.027296083
#> 
#>  Autocorrelation of GEE residuals 
#>  [1] -0.001277974 -0.004261554  0.045280260  0.022738750  0.005821352
#>  [6]  0.004289166  0.008311357  0.003437398  0.001030847 -0.010359040
#> --- 
#>  Autocorrelation parameters from  fixed  model 
#> [1] "a=alpha^(d^v)  , alpha=0.685 , v=1.093"

predictions<-predict(mgee, newdata=musdata)

As you can see, this package includes S3 methods for summary and predict. These are useful in evaluating model fit and autocorrelation of residuals compared to a non-spatial model (in this case, a GLM with the same family as the GEE). Additionally, the plot argument in GEE can be used to visually inspect the autocorrelation of the residuals from each regression. Note that a QIC (Quasi-information criterion) score is reported as opposed to AIC. This is calculated based on the method described in Hardin & Hilbe4,5 and is implemented using the function qic.calc.

Note that trying to fit GEEs with corstr="fixed" to large data sets (i.e. number of observations is approximately sqrt(.Machine$integer.max)) will result in errors, as the resulting variance-covariance matrices will be too large to be handled in R (you may well run into problems before this point due to memory allocation issues). This is where fitting clustered models can come in handy, as they work with smaller, more manageable matrices. These can be specified by changing the corstr to either "quadratic" or "exchangeable".

Wavelet Revised Models (WRMs) 6

Next, we’ll examine the other main model that is introduced in this package - the Wavelet Revised Model. These are implemented using wavelet transforms from the waveslim package.7 Let’s start with a fairly simple WRM using the same musdata data set as above.



mwrm<-WRM(musculus ~ pollution + exposure, "poisson", musdata,
coord=coords, level=1, plot=TRUE)


summary(mwrm)
#> 
#>  Call: 
#> WRM(formula = musculus ~ pollution + exposure, family = "poisson", 
#>     data = musdata, coord = coords, level = 1, plot = TRUE)
#> 
#>  Pearson Residuals: 
#>       Min.    1st Qu.     Median       Mean    3rd Qu.       Max. 
#> -1.6140000 -0.3057000  0.0047510 -0.0003873  0.3039000  3.0620000 
#> --- 
#>  Coefficients: 
#>             Estimate Std.Err z value Pr(>|z|)   
#> (Intercept)  -1.9360  1.9177 -1.0095 0.312717   
#> pollution     3.1841  1.2251  2.5991 0.009348 **
#> exposure     -1.2286  1.5063 -0.8156 0.414723   
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> --- 
#>  Number of observations n:  400 ,  n.eff:  300 ,  AIC:  1110.845 
#> 
#>  Number of iterations:  7 
#> --- 
#> Autocorrelation of glm.residuals 
#>  [1]  0.685338504  0.509680590  0.363021118  0.247398654  0.144726020
#>  [6]  0.084220961  0.050228656  0.022369044 -0.001985639 -0.027296083
#> Autocorrelation of wavelet.residuals 
#>  [1]  0.024855393 -0.086311686  0.007820356  0.024501828 -0.016578686
#>  [6]  0.002798656 -0.002977017 -0.004611334  0.018150352 -0.008727321

predictions<-predict(mwrm, newdata=musdata)

Let’s try padding with mean values.


# Padding with mean values

padded.mwrm<-WRM(musculus ~ pollution + exposure, "poisson", musdata,
                 coord=coords, level=1, pad=list(padform=1), plot=TRUE)


summary(padded.mwrm)
#> 
#>  Call: 
#> WRM(formula = musculus ~ pollution + exposure, family = "poisson", 
#>     data = musdata, coord = coords, level = 1, pad = list(padform = 1), 
#>     plot = TRUE)
#> 
#>  Pearson Residuals: 
#>       Min.    1st Qu.     Median       Mean    3rd Qu.       Max. 
#> -1.6140000 -0.3057000  0.0047510 -0.0003873  0.3039000  3.0620000 
#> --- 
#>  Coefficients: 
#>             Estimate Std.Err z value Pr(>|z|)   
#> (Intercept)  -1.9360  1.9177 -1.0095 0.312717   
#> pollution     3.1841  1.2251  2.5991 0.009348 **
#> exposure     -1.2286  1.5063 -0.8156 0.414723   
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> --- 
#>  Number of observations n:  400 ,  n.eff:  300 ,  AIC:  1110.845 
#> 
#>  Number of iterations:  7 
#> --- 
#> Autocorrelation of glm.residuals 
#>  [1]  0.685338504  0.509680590  0.363021118  0.247398654  0.144726020
#>  [6]  0.084220961  0.050228656  0.022369044 -0.001985639 -0.027296083
#> Autocorrelation of wavelet.residuals 
#>  [1]  0.024855393 -0.086311686  0.007820356  0.024501828 -0.016578686
#>  [6]  0.002798656 -0.002977017 -0.004611334  0.018150352 -0.008727321

padded.predictions<-predict(padded.mwrm, newdata=musdata)

WRM has many of the same features as GEE. Setting plot=TRUE allows you to visually examine the autocorrelation of residuals from a GLM of the same error family as your WRM. S3 methods for predict and summary allow you to examine outputs from the model using the same code as you might use for a GLM. However, note that this reports an AIC score, rather than a QIC score as in the GEE.

Other features specific to WRMs

WRM has a number of other model-specific functions that you may find useful in diagnosing model fit and understanding your results. For example, you might want to plot the variance or covariance of each of your variables as a function of level. The covar.plot function allows you to visually examine the wavelet relationships from your model. However, we are going to switch to the carlinadata data set now.


coords<-carlinadata[ ,4:5]

covar.plot(carlina.horrida ~ aridity + land.use - 1,
           data=carlinadata, coord=coords, wavelet="d4",
           wtrafo='modwt', plot='covar')

#> $result
#>                            [,1]   [,2]   [,3]   [,4]   [,5]
#> carlina.horrida-aridity  0.0368 0.0450 0.0623 0.0780 0.0466
#> carlina.horrida-land.use 0.4782 0.1191 0.0332 0.0126 0.0055

covar.plot(carlina.horrida ~ aridity + land.use - 1,
           carlinadata, coord=coords, wavelet="d4",
           wtrafo='modwt', plot='var')

#> $result
#>                   [,1]   [,2]   [,3]   [,4]   [,5]
#> carlina.horrida 0.7235 0.1792 0.0628 0.0242 0.0093
#> aridity         0.0691 0.1025 0.2028 0.3588 0.2657
#> land.use        0.7556 0.1851 0.0420 0.0119 0.0044

You may also want to view the smooth components of your wavelets at different scale levels. For this, we offer the upscale function, which allows you to visually examine your matrices for a number of different levels of scale, which controls the resolution of the grid cells in your data set.8 It also offers the option to adjust padding settings so you can see how that influences your smooth components as well. The default is mean values of your input vector, but it can be easily switched using the pad argument, which works the same way as in the other WRM functions. A quick example below using carlinadata data set.


upscale(carlinadata$land.use, carlinadata$x, carlinadata$y)

Multi-model inference with GEEs and WRMs

spind provides a couple of frameworks for conducting multi-model inference analyses and some helper functions that we hope will make your life easier when examining the results. The first that we’ll examine here is the step.spind function, which implements step-wise model selection. The process is loosely based on MASS::stepAIC and stats::step, but is specific to classes GEE and WRM. For GEEs, step.spind uses models with the lowest QIC scores to determine what the next step will be. For WRMs, you have the option of using AIC or AICc (AIC corrected for small sample sizes) using the logical AICc argument.

Currently, the function only supports backwards model selection. In other words, you have to start with all of the variables in your model formula and remove them in a stepwise fashion. We hope to add forward model selection methods shortly. Additionally, step.spind is written to always respect the hierarchy of variables in the model and the user cannot override this currently. For example, step.spind would not remove race while retaining I(race^2). We may change that in the future, but it will remain like this at least until the next major release. Currently, it recognizes polynomial variables by matching variable names located inside of I(var^some_power) and interaction terms by searching for var1:var2 in the model terms. If you want to use a higher order polynomial variable and are not worried about the variable hierarchy, you can create a separate variable (i.e. race_2) and use that in the model.

We’ll go through an example of step.spind using a GEE on the birthwt data set in the MASS package below. The data in birthwt aren’t at all related to SDMs and are not spatially structured, but we hope that in using this data set, we will demonstrate how this function can work with many types of data sets.


# For demonstration only. We are artificially imposing a grid structure
# on data that is not actually spatial data

library(MASS)
data(birthwt)


x<-rep(1:14,14)
y<-as.integer(gl(14,14))
coords<-cbind(x[-(190:196)],y[-(190:196)])

formula<-formula(low ~ age + lwt + race + smoke + ftv +  bwt + I(race^2))

mgee<-GEE(formula, family="gaussian", data=birthwt,
          coord=coords, corstr="fixed",scale.fix=TRUE)

mwrm<-WRM(formula, family="gaussian", data=birthwt,
          coord=coords, level=1)

ssgee<-step.spind(mgee, birthwt)
#> Iteration:  1 
#>  Single term deletions
#>  Deleted Term:  age 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 112.4177  -52.68206
#> 2          age 111.9314  -52.67027
#> 3          lwt 112.3100  -52.74725
#> 4         race 112.1267  -52.68782
#> 5        smoke 112.1349  -52.68652
#> 6          ftv 112.1632  -52.70359
#> 7          bwt 299.1973 -121.62394
#> 8    I(race^2) 112.1329  -52.69176
#> 
#> Iteration:  2 
#>  Single term deletions
#>  Deleted Term:  race 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 111.9314  -52.67027
#> 2          lwt 111.8146  -52.72965
#> 3         race 111.6474  -52.67038
#> 4        smoke 111.6566  -52.67100
#> 5          ftv 111.7364  -52.72503
#> 6          bwt 299.0924 -121.56174
#> 7    I(race^2) 111.6567  -52.67607
#> 
#> -----
#> Model hierarchy violated by last removal
#> New deleted term:  smoke 
#> Previously deleted term added back into model
#>  -----
#> Iteration:  3 
#>  Single term deletions
#>  Deleted Term:  race 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 111.6566  -52.67100
#> 2          lwt 111.5311  -52.72809
#> 3         race 111.3802  -52.67366
#> 4          ftv 111.4711  -52.72790
#> 5          bwt 300.0954 -123.16568
#> 6    I(race^2) 111.3877  -52.67890
#> 
#> -----
#> Model hierarchy violated by last removal
#> New deleted term:  I(race^2) 
#> Previously deleted term added back into model
#>  -----
#> Iteration:  4 
#>  Single term deletions
#>  Deleted Term:  ftv 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 111.3877  -52.67890
#> 2          lwt 111.2494  -52.72805
#> 3         race 111.3122  -52.76311
#> 4          ftv 111.2088  -52.73630
#> 5          bwt 298.6147 -123.33017
#> 
#> Iteration:  5 
#>  Single term deletions
#>  Deleted Term:  lwt 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 111.2088  -52.73630
#> 2          lwt 111.0717  -52.78793
#> 3         race 111.1415  -52.82335
#> 4          bwt 298.6351 -123.35038
#> 
#> Iteration:  6 
#>  Single term deletions
#>  Deleted Term:  race 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 111.0717  -52.78793
#> 2         race 110.9656  -52.86072
#> 3          bwt 295.8817 -123.12477
#> 
#> Iteration:  7 
#>  Single term deletions
#>  Deleted Term:  <none> 
#>  -------------------- 
#>   Deleted.Vars      QIC  Quasi.Lik
#> 1       <none> 110.9656  -52.86072
#> 2          bwt 296.2879 -123.98743
#> 
#> 
#> ---------------
#> Best model found:
#> low ~ bwt
sswrm<-step.spind(mwrm, birthwt, AICc=TRUE)
#> Iteration:  1 
#>  Single term deletions
#>  Deleted Term:  race 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -36.27411  90.54822  91.55381
#> 2          age -36.27965  88.55930  89.35930
#> 3          lwt -36.30071  88.60143  89.40143
#> 4         race -36.03376  88.06752  88.86752
#> 5        smoke -36.13543  88.27087  89.07087
#> 6          ftv -36.25766  88.51532  89.31532
#> 7          bwt -88.97789 193.95579 194.75579
#> 8    I(race^2) -36.03879  88.07758  88.87758
#> 
#> -----
#> Model hierarchy violated by last removal
#> New deleted term:  I(race^2) 
#> Previously deleted term added back into model
#>  -----
#> Iteration:  2 
#>  Single term deletions
#>  Deleted Term:  smoke 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -36.03879  88.07758  88.87758
#> 2          age -36.04158  86.08317  86.70195
#> 3          lwt -36.09072  86.18144  86.80022
#> 4         race -36.11776  86.23552  86.85430
#> 5        smoke -35.90087  85.80174  86.42052
#> 6          ftv -36.02003  86.04007  86.65885
#> 7          bwt -88.71971 191.43941 192.05820
#> 
#> Iteration:  3 
#>  Single term deletions
#>  Deleted Term:  ftv 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -35.90087  85.80174  86.42052
#> 2          age -35.90797  83.81593  84.27747
#> 3          lwt -35.95683  83.91366  84.37520
#> 4         race -35.97398  83.94797  84.40950
#> 5          ftv -35.88997  83.77995  84.24149
#> 6          bwt -90.79731 193.59462 194.05616
#> 
#> Iteration:  4 
#>  Single term deletions
#>  Deleted Term:  age 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -35.88997  83.77995  84.24149
#> 2          age -35.89084  81.78169  82.10956
#> 3          lwt -35.94335  81.88670  82.21457
#> 4         race -35.96301  81.92602  82.25389
#> 5          bwt -90.81649 191.63298 191.96085
#> 
#> Iteration:  5 
#>  Single term deletions
#>  Deleted Term:  lwt 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -35.89084  81.78169  82.10956
#> 2          lwt -35.94877  79.89754  80.11494
#> 3         race -35.95486  79.90972  80.12711
#> 4          bwt -91.14883 190.29766 190.51505
#> 
#> Iteration:  6 
#>  Single term deletions
#>  Deleted Term:  race 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -35.94877  79.89754  80.11494
#> 2         race -36.00167  78.00334  78.13307
#> 3          bwt -91.88479 189.76958 189.89931
#> 
#> Iteration:  7 
#>  Single term deletions
#>  Deleted Term:  <none> 
#>  -------------------- 
#>   Deleted.Vars    LogLik       AIC      AICc
#> 1       <none> -36.00167  78.00334  78.13307
#> 2          bwt -92.26134 188.52268 188.58719
#> 
#> 
#> ---------------
#> Best model found:
#> low ~ bwt

best.mgee<-GEE(ssgee$model, family = "gaussian", data=birthwt,
           coord=coords, corstr="fixed",scale.fix=TRUE)

best.wrm<-WRM(sswrm$model, family="gaussian", data=birthwt,
              coord=coords, level = 1)

summary(best.mgee, printAutoCorPars=FALSE)
#> 
#>  Call: 
#> GEE(formula = ssgee$model, family = "gaussian", data = birthwt, 
#>     coord = coords, corstr = "fixed", scale.fix = TRUE)
#> --- 
#>  Coefficients: 
#>                Estimate     Std.Err t value Pr(>|t|)    
#> (Intercept)  1.2492e+00  4.9121e-01  2.5430  0.01099 *  
#> bwt         -3.0919e-04  6.5913e-05 -4.6909 2.72e-06 ***
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> --- 
#>  QIC:  110.9656 
#> --- 
#> Autocorrelation of GLM residuals 
#>  [1]  0.837748633  0.724407532  0.602588671  0.500754270  0.387294592
#>  [6]  0.275433941  0.147728669  0.008716423 -0.130798183 -0.268641655
#> 
#>  Autocorrelation of GEE residuals 
#>  [1]  0.43453709  0.35186795  0.27457621  0.21231229  0.10255028
#>  [6]  0.08028419  0.07174312  0.04070057  0.02919975 -0.06904364
summary(best.wrm)
#> 
#>  Call: 
#> WRM(formula = sswrm$model, family = "gaussian", data = birthwt, 
#>     coord = coords, level = 1)
#> 
#>  Pearson Residuals: 
#>     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
#> -0.32380 -0.03533 -0.01311  0.00000  0.03328  0.42380 
#> --- 
#>  Coefficients: 
#>                Estimate     Std.Err     t value  Pr(>|t|)    
#> (Intercept)  1.2809e+00  2.3089e-09  5.5477e+08 < 2.2e-16 ***
#> bwt         -3.4305e-04  7.4898e-06 -4.5802e+01 < 2.2e-16 ***
#> ---
#> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
#> --- 
#>  Number of observations n:  189 ,  n.eff:  142 ,  AIC:  78.00334 
#> 
#>  Number of iterations:  2 
#> --- 
#> Autocorrelation of glm.residuals 
#>  [1]  0.837748633  0.724407532  0.602588671  0.500754270  0.387294592
#>  [6]  0.275433941  0.147728669  0.008716423 -0.130798183 -0.268641655
#> Autocorrelation of wavelet.residuals 
#>  [1] 0.427865614 0.201246818 0.139896894 0.103437367 0.046485859
#>  [6] 0.045779123 0.025277466 0.034175107 0.005430915 0.055360212

Additionally, we offer multimodel inference tools for GEEs and WRMs which are loosely based on the MuMIn package. These are implemented in mmiWMRR and mmiGEE. They enable you to examine the effect that the grid resolution and variable selection have on the resulting regressions, and then select the appropriate model for subsequent analyses. Note that mmiWMRR has two more arguments than mmiGEE that must be specified.


# Example for WRMs
data(carlinadata)
coords<- carlinadata[,4:5]

wrm<- WRM(carlina.horrida ~ aridity + land.use, "poisson",
          carlinadata, coords, level=1, wavelet="d4")

ms1<-scaleWMRR(carlina.horrida ~ aridity + land.use,"poisson",
                carlinadata,coords,scale=1,wavelet='d4',plot=F)

mmi<- mmiWMRR(wrm, data=carlinadata, scale=1, detail=TRUE)
#> --- 
#>  Level =  1 
#>      (Int) aridity land.use df    logLik    AIC delta weight
#> 3  3.63243         -3.78957  2 -1185.487 2375.0  0.00  0.995
#> 4  1.94780 2.11872 -3.42550  3 -1189.834 2385.7 10.69  0.005
#> 2 -0.82469 1.13710           2 -1200.512 2405.0 30.05  0.000
#> 1 -0.13692                   1 -1221.456 2444.9 69.94  0.000

# Example for GEEs
library(MASS)
data(birthwt)

# impose an artificial (not fully appropriate) grid structure
x<-rep(1:14,14)
y<-as.integer(gl(14,14))
coords<-cbind(x[-(190:196)],y[-(190:196)])

formula<-formula(low ~ race + smoke +  bwt)

mgee<-GEE(formula, family = "gaussian", data = birthwt,
         coord=coords, corstr="fixed", scale.fix=TRUE)

mmi<-mmiGEE(mgee, birthwt)
#> 
#>  Model selection table: 
#>  
#>     (Int)     race    smoke      bwt df     QLik   QIC  delta weight
#> 5 1.24916                   -0.00031  3  -52.861 111.0   0.00  0.275
#> 6 1.25578 -0.00240          -0.00031  4  -52.788 111.1   0.11  0.261
#> 7 1.25003          -0.00118 -0.00031  4  -52.874 111.3   0.31  0.236
#> 8 1.25964 -0.00288 -0.00293 -0.00031  5  -52.787 111.3   0.38  0.228
#> 4 0.29818  0.02756  0.04083           4 -121.597 294.6 183.65  0.000
#> 3 0.35269           0.02366           3 -123.187 295.2 184.28  0.000
#> 2 0.32531  0.02216                    3 -123.125 295.9 184.92  0.000
#> 1 0.36431                             2 -123.987 296.3 185.32  0.000
#> 
#>  --- 
#>  Relative variable importance: 
#>  
#>  race smoke   bwt 
#> 0.489 0.464 1.000

Finally, we offer one further model selection procedure specific to WRMs. rvi.plot uses mmiWMRR and creates a plot of the relative importance of each explanatory variable as a function of the resolution of the grid (in other words, as a function of the scale argument in mmiWMRR). It will also print the resulting model selection tables to the console.

data(carlinadata)
coords<- carlinadata[,4:5]

rvi.plot(carlina.horrida ~ aridity + land.use,"poisson",
         data=carlinadata,coord=coords,maxlevel=4,detail=TRUE,wavelet="d4")
#> 
#>  Model selection tables: 
#>  
#> --- 
#>  Level =  1 
#>      (Int) aridity land.use df    logLik    AIC delta weight
#> 3  3.63243         -3.78957  2 -1185.487 2375.0  0.00  0.995
#> 4  1.94780 2.11872 -3.42550  3 -1189.834 2385.7 10.69  0.005
#> 2 -0.82469 1.13710           2 -1200.512 2405.0 30.05  0.000
#> 1 -0.13692                   1 -1221.456 2444.9 69.94  0.000
#> --- 
#>  Level =  2 
#>      (Int) aridity land.use df    logLik    AIC  delta weight
#> 4  2.23516 0.59909 -2.93096  3 -1184.169 2374.3   0.00      1
#> 2 -0.73426 0.75117           2 -1209.233 2422.5  48.13      0
#> 3  2.72922         -3.16193  2 -1228.572 2461.1  86.81      0
#> 1 -0.40235                   1 -1262.854 2527.7 153.37      0
#> --- 
#>  Level =  3 
#>      (Int) aridity land.use df    logLik    AIC  delta weight
#> 4  2.93438 0.49042 -3.50797  3 -1178.072 2362.1   0.00      1
#> 3  3.11997         -3.40857  2 -1198.412 2400.8  38.68      0
#> 2 -0.55307 0.46073           2 -1217.184 2438.4  76.22      0
#> 1 -0.28444                   1 -1238.104 2478.2 116.06      0
#> --- 
#>  Level =  4 
#>      (Int) aridity land.use df    logLik    AIC  delta weight
#> 2 -1.36217 1.87641           2 -1205.696 2415.4   0.00      1
#> 1 -0.02245                   1 -1220.984 2444.0  28.58      0
#> 3  8.29497         -8.35099  2 -1272.480 2549.0 133.57      0
#> 4  7.65184 1.81274 -9.01193  3 -1292.419 2590.8 175.45      0
#> 
#>  --- 
#>  Relative variable importance: 
#> 

#>          level=1 level=2 level=3 level=4
#> aridity    0.005       1       1       1
#> land.use   1.000       1       1       0

Goodness of fit and model performance

You may also find that a model not implemented by this package works best for your data. We still have some spatially corrected accuracy measures that you can use to assess goodness of model fit. The first two of these are categorized according to whether or not their outputs are dependent on the chosen threshold and first appeared in the spind 1.0 9. th.dep (threshold dependent) and th.indep (threshold independent) are designed to work on any number of model types, all you need is a set of actual values, predictions, and their associated coordinates. We’ll use the hook data set to see how these work.

data(hook)

# Familiarize yourself with the data
?hook
head(hook)

df<-hook[,1:2]
coords<-hook[,3:4]

# Threshold dependent metrics
th.dep.indices<-th.dep(data=df, coord=coords, spatial=TRUE)

# Confusion Matrix
th.dep.indices$cm
#>      [,1] [,2] [,3] [,4]
#> [1,]    5    2    0    0
#> [2,]    3    1    1    3
#> [3,]    2    0    0    8
#> [4,]    2    3    0   70

# Kappa statistic
th.dep.indices$kappa
#> [1] 0.628529

# Threshold independent metrics
th.indep.indices<-th.indep(data=df, coord=coords, spatial=TRUE, plot.ROC=TRUE)


# AUC
th.indep.indices$AUC
#> [1] 0.9424119

# TSS
th.indep.indices$TSS
#> [1] 0.7425474

Additionally, we include the function acfft to calculate spatial autocorrelation of model residuals using Moran’s I statistic. A quick example below using a GLM and the musdata data set.


coords<- musdata[,4:5]
mglm <- glm(musculus ~ pollution + exposure, "poisson", musdata)

ac<-acfft(coords[ ,1], coords[ ,2], resid(mglm, type="pearson"),
          lim1=0, lim2=1, dmax=10)
ac
#>  [1]  0.685338504  0.509680590  0.363021118  0.247398654  0.144726020
#>  [6]  0.084220961  0.050228656  0.022369044 -0.001985639 -0.027296083

Note that you can adjust the number of distance bins to examine in acfft using the dmax argument. The default is 10.

Wrapping up

Hopefully, you are now ready to utilize GEEs and WRMs to conquer the world of species distribution modeling. However, if this vignette has not served its purpose and you still have questions about how to use these tools (or how to improve this vignette), please let us know. Of course, no package is complete without bugs and we are always trying to improve our code. If you find any bugs that need squashing, have suggestions for additional functionality or improvements to existing functionality, please don’t hesitate to contact us10.

  1. Carl G & Kuehn I, 2007. Analyzing Spatial Autocorrelation in Species Distributions using Gaussian and Logit Models, Ecol. Model. 207, 159 - 170]↩

  2. Carey, V. J., 2006. Ported to R by Thomas Lumley (versions 3.13, 4.4, version 4.13)., B. R. gee: Generalized Estimation Equation solver. R package version 4.13-11.↩

  3. Yan, J., 2004. geepack: Generalized Estimating Equation Package. R package version 0.2.10.↩

  4. Hardin, J.W. & Hilbe, J.M. (2003) Generalized Estimating Equations. Chapman and Hall, New York.↩

  5. Barnett et al. Methods in Ecology & Evolution 2010, 1, 15-24.↩

  6. Carl, G., Kuehn, I. (2010): A wavelet-based extension of generalized linear models to remove the effect of spatial autocorrelation. Geographical Analysis 42 (3), 323 - 337↩

  7. Whitcher, B. (2005) Waveslim: basic wavelet routines for one-, two- and three-dimensional signal processing. R package version 1.5.↩

  8. Carl G, Doktor D, Schweiger O, Kuehn I (2016) Assessing relative variable importance across different spatial scales: a two-dimensional wavelet analysis. Journal of Biogeography 43: 2502-2512.↩

  9. Carl G, Kuehn I (2016) Spind: a package for computing spatially corrected accuracy measures. Ecography. DOI: 10.1111/ecog.02593↩

  10. Contact email - levisc8@gmail.com or visit the Github repo and create an issue at http://github.com/levisc8/spind/issues.↩