[R-sig-ME] glmm question

Tue Sep 1 22:58:44 CEST 2015

Hello, I´m trying to model de abundance of a grassland shorebird
(Buff-breasted Sandpiper, BBSA) as a function of the abundance of American
Golden Plover (AMGP) abundance, grass height and distance to the lagoon.
I´m using area as an offset.

I´ve tried a glmm with poisson errors and wanted you to see if residuals
are fine (they don´t look great, but wanted to know if they area minimally
acceptable). Also, I´m trying to understand two way interactions but I´m no
sure how to proceed, and also would like to plot the interactions and
unique variables effects (can you please provide suggestions?).

I also have problems with the predict function. In the script are the
notes. I´m attaching the data as well as the script with commentaries.

Let me know if there´s some format problem with the files.

Thank you very much in advanced.

Joaquín.

-- 
*Joaquín Aldabe*

*Grupo Biodiversidad, Ambiente y Sociedad*
Centro Universitario de la Región Este, Universidad de la República
Ruta 15 (y Ruta 9), Km 28.500, Departamento de Rocha

*Departamento de Conservación*
Aves Uruguay
BirdLife International
Canelones 1164, Montevideo

https://sites.google.com/site/joaquin.aldabe
<https://sites.google.com/site/perfilprofesionaljoaquinaldabe>
-------------- next part --------------
Field_name	Year	Month	Grassland_type	Prop_numero_potr_pastalto	Grass_height	Has_pastalt	Prop_sup_pastoalt	Field_area	Distance_to_lagoon	Flood	Field_enclosure_700m	Field_enclosure_350	Grazing_regime	BBSA	AMGP
Casas	2006	December	Natural	0,375	13,0	264,0	0,1	47	4699	No	61,0	42,8	Rotational	0	0
Pradera_B	2006	December	Improved	0,375	37,0	264,0	0,1	50	4426	No	66,8	46,4	Rotational	0	0
Pradera_A	2006	December	Improved	0,375	4,5	264,0	0,1	32	4073	No	76,9	68,9	Rotational	0	0
Ba�o	2006	December	Natural	0,375	7,0	264,0	0,1	142	2611	Yes	9,2	2,7	Continuous	65	100
Puesto	2006	December	Natural	0,375	6,5	264,0	0,1	261	1864	Yes	3,1	0,0	Continuous	1	39
Tres_Porteras	2006	December	Natural	0,375	6,0	264,0	0,1	92	2480	No	43,9	24,3	Continuous	0	0
Del_Medio	2006	December	Natural	0,375	4,5	264,0	0,1	82	2962	No	39,6	18,5	Continuous	0	0
Sancho	2006	December	Natural	0,375	3,0	264,0	0,1	244	874	Yes	26,8	6,8	Continuous	57	30
Laguna	2006	December	Natural	0,375	4,5	264,0	0,1	345	459	Yes	6,5	0,0	Continuous	94	201
Bolsa_Chica	2006	December	Natural	0,375	8,0	264,0	0,1	122	562	Yes	18,3	4,3	Continuous	0	1
Estero	2006	December	Improved	0,375	8,0	264,0	0,1	179	2600	No	25,8	10,6	Rotational	0	1
Coronillas	2006	December	Natural	0,375	3,5	264,0	0,1	150	342	Yes	7,5	0,5	Continuous	14	2
Pradera_Balneario	2006	December	Improved	0,375	35,0	264,0	0,1	36	4037	No	63,9	48,1	Rotational	0	0
Primer_Potrero	2006	December	Improved	0,375	35,0	264,0	0,1	43	3584	No	62,6	35,9	Rotational	0	0
Pileta	2006	December	Improved	0,375	40,0	264,0	0,1	46	3046	No	43,2	28,8	Rotational	0	0
Aljibe	2006	December	Improved	0,375	40,0	264,0	0,1	42	2361	No	47,4	38,3	Rotational	0	0
Casas	2007	December	Natural	0,1875	4,0	347,0	18,0	47	4699	No	61,0	42,9	Rotational	0	0
Pradera_B	2007	December	Improved	0,1875	30,0	347,0	18,0	50	4426	No	66,8	46,4	Rotational	0	0
Pradera_Balneario	2007	December	Improved	0,1875	6,0	347,0	18,0	36	4037	No	63,9	48,1	Rotational	0	0
Primer_Potrero	2007	December	Improved	0,1875	7,0	347,0	18,0	43	3584	No	62,6	35,9	Rotational	0	30
Pradera_A	2007	December	Improved	0,1875	6,0	347,0	18,0	32	4073	No	76,9	68,9	Rotational	0	0
Ba�o	2007	December	Natural	0,1875	3,5	347,0	18,0	142	2611	Yes	9,2	2,7	Continuous	7	64
Laguna	2007	December	Natural	0,1875	4,0	347,0	18,0	345	459	Yes	6,5	0,0	Continuous	4	91
Puesto	2007	December	Natural	0,1875	12,0	347,0	18,0	261	1864	Yes	3,1	0,0	Continuous	54	131
Sancho	2007	December	Natural	0,1875	4,5	347,0	18,0	244	874	Yes	26,8	6,8	Continuous	66	65
Bolsa_Chica	2007	December	Natural	0,1875	4,0	347,0	18,0	122	562	Yes	18,3	4,3	Continuous	52	92
Tres_Porteras	2007	December	Natural	0,1875	4,0	347,0	18,0	92	2480	No	43,9	24,3	Continuous	0	0
Del_Medio	2007	December	Natural	0,1875	6,5	347,0	18,0	82	2962	No	39,6	18,5	Continuous	0	0
Pileta	2007	December	Improved	0,1875	14,0	347,0	18,0	46	3046	No	43,2	28,8	Rotational	0	0
Aljibe	2007	December	Improved	0,1875	6,0	347,0	18,0	42	2361	No	47,4	38,3	Rotational	2	54
Coronillas	2007	December	Natural	0,1875	5,0	347,0	18,0	150	342	Yes	7,5	0,5	Continuous	55	157
Estero	2007	December	Improved	0,1875	7,0	347,0	18,0	179	2600	No	25,8	10,6	Rotational	0	0
Estero	2008	December	Improved	0,125	5,5	82,0	4,3	179	2600	No	25,8	10,6	Rotational	22	155
Del_Medio	2008	December	Natural	0,125	6,5	82,0	4,3	82	2962	No	39,6	18,5	Continuous	0	29
Tres_Porteras	2008	December	Natural	0,125	5,0	82,0	4,3	92	2480	No	43,9	24,3	Continuous	3	25
Primer_Potrero	2008	December	Improved	0,125	8,5	82,0	4,3	43	3584	No	62,6	35,9	Rotational	0	0
Pradera_Balneario	2008	December	Improved	0,125	8,0	82,0	4,3	36	4037	No	63,9	48,1	Rotational	0	0
Casas	2008	December	Natural	0,125	6,5	82,0	4,3	47	4699	No	61,0	42,9	Rotational	0	0
Laguna	2008	December	Natural	0,125	5,0	82,0	4,3	345	459	Yes	6,5	0,0	Continuous	14	27
Aljibe	2008	December	Improved	0,125	7,5	82,0	4,3	42	2361	No	47,4	38,3	Rotational	0	0
Bolsa_Chica	2008	December	Natural	0,125	4,5	82,0	4,3	122	562	Yes	18,3	4,3	Continuous	7	22
Pileta	2008	December	Improved	0,125	8,0	82,0	4,3	46	3046	No	43,2	28,8	Rotational	0	0
Coronillas	2008	December	Natural	0,125	3,0	82,0	4,3	150	342	Yes	7,5	0,5	Continuous	60	118
Sancho	2008	December	Natural	0,125	4,5	82,0	4,3	244	874	Yes	26,8	6,8	Continuous	59	198
Pradera_A	2008	December	Improved	0,125	12,0	82,0	4,3	32	4073	No	76,9	68,9	Rotational	0	0
Pradera_B	2008	December	Improved	0,125	20,0	82,0	4,3	50	4426	No	66,8	46,4	Rotational	0	0
Ba�o	2008	December	Natural	0,125	5,5	82,0	4,3	142	2611	Yes	9,2	2,7	Continuous	18	152
Puesto	2008	December	Natural	0,125	9,0	82,0	4,3	261	1864	Yes	3,1	0,0	Continuous	0	46
Puesto	2012	December	Natural	0,75	24,6	1512,0	79,0	261	1864	Yes	3,1	0,0	Continuous	0	0
Coronillas	2012	December	Natural	0,75	13,9	1512,0	79,0	150	342	Yes	7,5	0,5	Continuous	0	0
Sancho	2012	December	Natural	0,75	15,4	1512,0	79,0	244	874	Yes	26,8	6,8	Continuous	0	0
Primer_Potrero	2012	December	Improved	0,75	25,4	1512,0	79,0	43	3584	No	62,6	35,9	Rotational	0	0
Pradera_A	2012	December	Improved	0,75	16,0	1512,0	79,0	32	4073	No	76,9	68,9	Rotational	0	0
Pradera_B	2012	December	Improved	0,75	12,8	1512,0	79,0	50	4426	No	66,8	46,4	Rotational	0	0
Estero	2012	December	Improved	0,75	8,5	1512,0	79,0	179	2600	No	25,8	10,6	Rotational	7	143
Tres_Porteras	2012	December	Natural	0,75	9,7	1512,0	79,0	92	2480	No	43,9	24,3	Continuous	0	0
Del_Medio	2012	December	Natural	0,75	3,4	1512,0	79,0	82	2962	No	39,6	18,5	Continuous	57	273
Ba�o	2012	December	Natural	0,75	11,4	1512,0	79,0	142	2611	Yes	9,2	2,7	Continuous	0	0
Laguna	2012	December	Natural	0,75	14,8	1512,0	79,0	345	459	Yes	6,5	0,0	Continuous	0	0
Casas	2012	December	Natural	0,75	14,8	1512,0	79,0	47	4699	No	61,0	42,9	Rotational	0	0
Pradera_Balneario	2012	December	Improved	0,75	13,3	1512,0	79,0	30	4037	No	63,9	48,1	Rotational	0	557
Pileta	2012	December	Improved	0,75	19,0	1512,0	79,0	46	3046	No	43,2	28,8	Rotational	0	0
Aljibe	2012	December	Improved	0,75	3,4	1512,0	79,0	42	2361	No	47,4	38,3	Rotational	66	253
Bolsa_Chica	2012	December	Natural	0,75	35,0	1512,0	79,0	122	562	Yes	18,3	4,3	Continuous	0	0

-------------- next part --------------
my4<-read.table(file.choose(), header=T, dec=",")

#scale variables
my4s<-as.data.frame(scale(my4[,c(6,10,12,13,16)], center=T, scale=T))
my4S<-cbind(BBSA=my4$BBSA, Field_name=my4$Field_name,Grassland_type=my4$Grassland_type, Flood=my4$Flood, Year=my4$Year,my4s)
my4S$log.AMGP=scale(log(my4$AMGP+1), scale=T, center=T)

# Add Field_area to use as offset
my4S<-cbind(my4S, Field_area.o=my4$Field_area)

require(lme4)
m3.1=glmer(BBSA~log.AMGP*Distance_to_lagoon+Grass_height*Distance_to_lagoon+log.AMGP*Grass_height+Grassland_type+(1|Field_name),family="poisson", offset=log(Field_area.o),data=my4S[-61,])
summary(m3.1)

plot(fitted(m3.1),residuals(m3.1))
qqnorm(residuals(m3.1)); qqline(residuals(m3.1))

m3.1.1=update(m3.1,~.-Grassland_type)
summary(m3.1.1)

#Prediction with m3.1.1. I�m leaving as fix the variables amgp, distance to lagoon (dist), to see the effect of grass height (grass) when distance to lagoon is lowest. But I could not make prediction function work.  

#Distance to lagoon 

amgp=rep(-0.79,500)#0.47
dist=rep(-1.59,500)#0.52
grass=seq(-0.85,2.83, length.out=500)
area1=seq(30,345,500)#52.5
newdata=data.frame(log.AMGP=amgp, Distance_to_lagoon=dist,Grass_height=grass, Field_area.o=area1)
predglmm=predict(m3.1.1,newdata, Reform=~(1|Field_name))