R help - Aug 2006 - Bootstraping for groups and subgroups and joing with other table

Dear R-experts,
   
  I have a table with following collumns: State, SamplePlot, Species and
BodySize. I sampled bird species at 34 SamplePlots and 5 States (regions)
monthly during two years. On each bird record I measured bodysize and identified
the species. So I have many records of each species (about 150 species) at each
SamplePlot and each Region (State).
   
  Now I would like bootstrap these data, selecting 50 records for each
State/SamplePlot combinations and count how many species (richness) were sampled
at bootstrap. I need to do this 1.000 times.
   
  After that and need join the number of species [obtained at each bootstrap and
for each State/SamplePlot combination] with a dataframe that have other
attributes for SamplePlot (like Area, Perimeter etc).  Bellow follow a sample of
these two tables (dataframes).
   
  Could someone help me on this task.
   
  Kind regards,
   
  Miltinho
   
  ========  Table 1 - Bird records
                    State  SamplePlot  Species  Bodysize    SaoPaulo  Site1 
Spp01  4.35    SaoPaulo  Site1  Spp04  0.80    SaoPaulo  Site1  Spp01  4.45   
SaoPaulo  Site1  Spp02  6.46    SaoPaulo  Site1  Spp02  3.98    SaoPaulo  Site1 
Spp02  9.82    SaoPaulo  Site2  Spp02  9.71    SaoPaulo  Site2  Spp01  7.38   
SaoPaulo  Site2  Spp03  9.30    SaoPaulo  Site2  Spp01  4.32    SaoPaulo  Site2 
Spp01  6.80    SaoPaulo  Site2  Spp02  5.80    SaoPaulo  Site2  Spp01  1.53   
RioJaneiro  Site1  Spp02  1.37    RioJaneiro  Site1  Spp02  4.43    RioJaneiro 
Site1  Spp01  3.31    RioJaneiro  Site1  Spp03  6.25    RioJaneiro  Site1  Spp01
3.40    RioJaneiro  Site2  Spp01  9.12    RioJaneiro  Site2  Spp01  0.80   
RioJaneiro  Site2  Spp01  6.37    RioJaneiro  Site2  Spp02  1.66    RioJaneiro 
Site2  Spp01  3.49    RioJaneiro  Site3  Spp02  8.98    RioJaneiro  Site3  Spp02
5.96    RioJaneiro  Site3  Spp01  8.28    RioJaneiro  Site3  Spp03  1.51   
Bahia  Site1  Spp01  6.76    Bahia
 Site1  Spp01  9.99    Bahia  Site2  Spp01  0.72    Bahia  Site2  Spp02  6.63   
Bahia  Site2  Spp02  3.41    Bahia  Site2  Spp01  8.85    Bahia  Site2  Spp01 
8.13    Bahia  Site3  Spp02  4.41    Bahia  Site3  Spp01  9.49    Bahia  Site3 
Spp02  6.77    Bahia  Site3  Spp02  0.64    Bahia  Site4  Spp01  6.97    Bahia 
Site4  Spp03  8.34    Bahia  Site4  Spp01  5.46    Bahia  Site4  Spp01  5.52   
Bahia  Site5  Spp01  5.37    Bahia  Site5  Spp02  8.66
   
  =====  Table 2 - Attributes of SamplePlot
                  State  SamplePlot  Area    Bahia  Site1  10    Bahia  Site2 
25    Bahia  Site3  70    Bahia  Site4  15    Bahia  Site5  5    RioJaneiro 
Site1  32    RioJaneiro  Site2  45    RioJaneiro  Site3  10    SaoPaulo  Site1 
23    SaoPaulo  Site2  45
   

 		
---------------------------------

 Música para ver e ouvir: You're Beautiful, do James Blunt
	[[alternative HTML version deleted]]

>   I have a table with following collumns: State, SamplePlot, Species and
BodySize. I sampled bird species at
> 34 SamplePlots and 5 States (regions) monthly during two years. On each
bird
record I measured bodysize
> and identified the species. So I have many records of each species (about
150
species) at each SamplePlot
> and each Region (State). 
> 
>   Now I would like bootstrap these data, selecting 50 records for each
State/SamplePlot combinations and
> count how many species (richness) were sampled at bootstrap. I need to do
this
1.000 times. 
> 
>   After that and need join the number of species [obtained at each
bootstrap
and for each State/SamplePlot
> combination] with a dataframe that have other attributes for SamplePlot
(like
Area, Perimeter etc). 



I asked a similar question earlier... IF you have data frame birds and
bird.plots, maybe something like this.

#initialize empty variable 
boot<-NULL

for(i in 1:100)
{
   ## split on state and site and create list a
   a <-split(birds, paste(birds$State,birds$SampleSite), drop=T)

   # sample 50 rows each OR by number of observations (better?)
   # b<-lapply( a, function(x) x[sample(nrow(x), 50, replace=T),])
     b<-lapply( a, function(x) x[sample(nrow(x), replace=T),])

   ## count number of unique species or other statistic? 
   ###  and add row to boot matrix
   boot<-rbind(boot, unlist( lapply(b, function (x) length(unique(x$Species))
) ))
}

## mean

y<-apply(boot, 2, mean)


## convert to data frame for merge

y<-data.frame(y)

names(y)<-"boot.count"


## add row names to bird.plots for easy join
rownames(bird.plots)<-paste(bird.plots$State,bird.plots$SampleSite)

merge(bird.plots,y, by=0)

          Row.names      State SampleSite Area boot.count
1       Bahia Site1      Bahia      Site1   10       1.00
2       Bahia Site2      Bahia      Site2   25       1.96
3       Bahia Site3      Bahia      Site3   70       1.72
4       Bahia Site4      Bahia      Site4   15       1.73
5       Bahia Site5      Bahia      Site5    5       1.42
6  RioJaneiro Site1 RioJaneiro      Site1   32       2.49
7  RioJaneiro Site2 RioJaneiro      Site2   45       1.63
8  RioJaneiro Site3 RioJaneiro      Site3   10       2.37
9    SaoPaulo Site1   SaoPaulo      Site1   23       2.41
10   SaoPaulo Site2   SaoPaulo      Site2   45       2.57


Chris Stubben