Issues with formatLong (df, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)

126 views
Skip to first unread message

Mark Thompson

unread,
Jul 30, 2019, 8:45:00 PM7/30/19
to unmarked
Hello,

I'm new to unmarked - just learning. My data has numerous NA's and I'm trying to make some decisions to retain or drop years / sites. This is the terrestrial subset of my data that also includes wetland surveys. However, in the meantime I'm trying to get my data into an unmarkedMultFrame from long into wide format using unmarked:

formatWideLong

I sampled twice from from 2014:2018. I am attempting to follow on the example by  Loading and finalizing Carolina Sandhills occupancy data, Adam Smith (https://rpubs.com/adamsmith_fws/CSNWR_load_occup_data). However, I get stuck on the example provided:

BHNU <- newformatLong(BHNU, type = "unmarkedFrameOccu", timename = "visit")

I thought I should be able to use:

amma <- formatLong(amma, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)
anbo
<- formatLong(anbo, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)
ralu
<- formatLong(ralu, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)
lisy
<- formatLong(lisy, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)
amphib
<- formatLong(amphib, type="unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)

However, this does not seem to be organizing my data correctly. 

> str(amma)
Formal class 'unmarkedMultFrame' [package "unmarked"] with 7 slots
 
..@ numPrimary    : num 5
 
..@ yearlySiteCovs:'data.frame': 5150 obs. of  1 variable:
 
.. ..$ FWCPdata.visit: num [1:5150] 1 1 1 1 1 2 2 2 2 2 ...
 
..@ y             : int [1:103, 1:10] 0 0 0 0 0 1 0 0 0 0 ...
 
..@ obsCovs       :'data.frame': 1030 obs. of  6 variables:
 
.. ..$ y.count   : int [1:1030] 0 NA NA NA NA NA NA NA NA NA ...
 
.. ..$ doy       : num [1:1030] 16954 NA NA NA NA ...
 
.. ..$ wetdist.3 : num [1:1030] 0.96 0.574 -0.74 0.96 0.574 ...
 
.. ..$ cutdist.4 : num [1:1030] -0.74 0.96 0.574 -0.74 0.96 ...
 
.. ..$ rddist.5  : num [1:1030] 0.574 -0.74 0.96 0.574 -0.74 ...
 
.. ..$ JulianDate: num [1:1030] 1 1 1 1 1 2 2 2 2 2 ...
 
..@ siteCovs      :'data.frame': 103 obs. of  2 variables:
 
.. ..$ place.1: Factor w/ 2 levels "m","p": 1 1 1 1 1 1 1 1 1 1 ...
 
.. ..$ vri.2  : Factor w/ 9 levels "AT","LA","PL",..: 9 7 9 3 3 9 9 9 9 9 ...
 
..@ mapInfo       : NULL
 
..@ obsToY        : num [1:10, 1:10] 1 0 0 0 0 0 0 0 0 0 ...

I don't understand where JulianDate came from (doy is date in Julian day) and I get an odd output when I try to put this into a colext model (:number of rows of matrices must match). I'm hoping that someone might be able to provide some assistance / guidance. I have Kery and Royle (2016) - took a course with Daryl Mackenzie on occupancy modelling with WinBugs and I have other applicable literature at hand that I've reviewed. I'm familiar, but still learning. I apologize for my inexperience here. 

Thank you,

Mark Thompson

FWCP OccupancyMult.R
FWCPDataLongTsp.csv

Mark Thompson

unread,
Jul 31, 2019, 1:07:58 PM7/31/19
to unmarked
The amphib variable is the detection results from all amphibian species combined amma + anbo + lisy + ralu. There were very few wood frogs (lisy) or spotted frogs (ralu) detected in my terrestrial plots. For now, I'm not including the wetland data and just focused on my terrestrial plots. I'm not even sure it's all that important at this time to differential the different species; amma = long-toed salamanders and anbo = western toads, however, I thought I should retain this information for now. The first year (2014) was more of a pilot year and very few of the terrestrial grids were sampled - I survey plots in grids for presence/absence. There is more site specific data I collected (e.g., soil depth, canopy cover %, coarse woody debris volumes, etc...), but I need to submit an interim report and will leave that information out for now. Below is what I I think I need to achieve after I get the data transformed from long to wide using the amphib data - or perhaps I should be using the four species as covariates?  

umf_AMPHIB <- unmarkedMultFrame(y=y1,
                                siteCovs
= data.frame(place.1, vri.2, wetdist.3, cutdist.4, rddist.5),
                                obsCovs
=list(visit, date),
                                yearlySiteCovs
=list(year),
                                numPrimary
=T)

I'm having trouble conceptualizing the matrix dimensions as I've tried to organize the data several other times. The method I posted seems to be approaching what I am after, but yearlySiteCovs looks more like my obsCovs, while obsCovs look like my siteCovs, and I don't know why JulianDate is showing up. I hope this gives the information needed to solve my problem. I understand that y needs to be a MxJT matrix and tried to organize the data using another rscript (attached) that I was using to learn / play around, but I think I made some errors on the date. I'm hoping that the new approach will help me to better understand the data structure.

Thank you!
FWCPDataLongT.csv
FWCP Occupancy.R

Adam Smith

unread,
Jul 31, 2019, 1:08:45 PM7/31/19
to unmarked
Hi Mark,

That `newformatLong` function was a custom function that's not included in the example and likely doesn't work any more even if you had it. The current `formatLong` function is to be preferred, and I don't suspect `visit` is a yearlySiteCov.

You'll need to modify your `visit` variable to reflect your primary/secondary sampling scheme. You have 10 possible visits to a site (5 primary periods, each with 2 secondary samples), so `visit` would need to take values in 1:10 across all sites. That is, primary period 1 (2014) with be associated with visits 1 & 2, primary period 2 (2015) will be associated with visits 3 & 4, and so on.

I've not intensively checked this, but the following should set up your data to be ready for `formatLong`:

library(unmarked)
library
(dplyr)
library
(lubridate)

FWCPdata <- read.csv("~/FWS_Projects/sandbox/Data/FWCPDataLongTsp.csv", header = TRUE, stringsAsFactors = FALSE) %>%
 
# Drop records with no response; we'll complete data set below
  filter
(!is.na(y))

# Explicitly define the sampling scheme so we can complete the data set with missing values later
sampling_scheme
<- expand.grid(spp.6 = unique(FWCPdata$spp.6),
                               visit
= 1:10,
                               site
= unique(FWCPdata$site),
                               stringsAsFactors
= FALSE)

FWCPdata <- FWCPdata %>%
  arrange
(spp.6, site, date) %>%
  mutate
(date = dmy(date), # Convert string to date
         doy
= yday(date), # Day of year variable; I think this is what you want instead of the 'Julian' function
         year
= lubridate::year(date),
         visit
= visit + ((year - min(year)) * 2 )) %>% # Number visits assuming 5 primary (yearly), 2 secondary periods
 
# Use sampling scheme to fill the data set with NAs, keeping the visit sequence correct
  left_join
(sampling_scheme, .) %>%
 
# Rearrange columns for unmarked
 
select(site, visit, y, spp.6, place.1:doy) %>%
 
# Rearrange rows for human eyes
  arrange
(spp.6, site, visit)

head
(FWCPdata, 20)
#>    site visit  y spp.6 place.1 vri.2 wetdist.3 cutdist.4 rddist.5 doy
#> 1   m.1     1 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 2   m.1     2 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 3   m.1     3 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 4   m.1     4 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 5   m.1     5  0  amma       m    TM   3116.42    245.48   254.55 154
#> 6   m.1     6 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 7   m.1     7 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 8   m.1     8 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 9   m.1     9 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 10  m.1    10 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 11 m.10     1 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 12 m.10     2 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 13 m.10     3 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 14 m.10     4 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 15 m.10     5  0  amma       m    SX    112.00    388.70   582.56 201
#> 16 m.10     6 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 17 m.10     7 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 18 m.10     8 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 19 m.10     9 NA  amma    <NA>  <NA>        NA        NA       NA  NA
#> 20 m.10    10 NA  amma    <NA>  <NA>        NA        NA       NA  NA

# Pull amma data for `formatLong` try
amma
<- filter(FWCPdata, spp.6 == "amma")
ammaMF
<- formatLong(amma, type = "unmarkedMultFrame", numPrimary = 5)
head
(ammaMF)
#> Data frame representation of unmarkedFrame object.
#>    y.1 y.2 y.3 y.4 y.5 y.6 y.7 y.8 y.9 y.10 spp.6 place.1 vri.2 wetdist.3
#> 1   NA  NA  NA  NA   0  NA  NA  NA  NA   NA  amma       m    TM   3116.42
#> 2   NA  NA  NA  NA   0  NA  NA  NA  NA   NA  amma       m    SX    112.00
#> 3   NA  NA  NA  NA   0  NA  NA  NA  NA   NA  amma       m    TM    814.55
#> 4   NA  NA  NA  NA   0  NA  NA  NA  NA   NA  amma       m    PL    515.90
#> 5   NA  NA  NA  NA   0  NA  NA  NA  NA   NA  amma       m    PL    624.13
#> 6   NA  NA  NA  NA   1  NA  NA  NA  NA   NA  amma       m    TM    161.50
#> 7   NA  NA  NA  NA  NA  NA   0  NA   0    0  amma       m    TM    239.89
#> 8   NA  NA  NA  NA  NA  NA  NA   0   0    0  amma       m    TM    307.97
#> 9   NA  NA  NA  NA  NA  NA  NA   0   0    0  amma       m    TM    244.50
#> 10  NA  NA  NA  NA  NA  NA  NA   0   0    0  amma       m    TM    364.40
#>    rddist.5 cutdist.4.1 cutdist.4.2 cutdist.4.3 cutdist.4.4 cutdist.4.5
#> 1    254.55          NA          NA          NA          NA      245.48
#> 2    582.56          NA          NA          NA          NA      388.70
#> 3    135.83          NA          NA          NA          NA      452.42
#> 4    619.85          NA          NA          NA          NA      264.33
#> 5    105.13          NA          NA          NA          NA      272.44
#> 6    454.77          NA          NA          NA          NA      638.66
#> 7    379.10          NA          NA          NA          NA          NA
#> 8    425.74          NA          NA          NA          NA          NA
#> 9    490.66          NA          NA          NA          NA          NA
#> 10   488.16          NA          NA          NA          NA          NA
#>    cutdist.4.6 cutdist.4.7 cutdist.4.8 cutdist.4.9 cutdist.4.10 doy.1
#> 1           NA          NA          NA          NA           NA    NA
#> 2           NA          NA          NA          NA           NA    NA
#> 3           NA          NA          NA          NA           NA    NA
#> 4           NA          NA          NA          NA           NA    NA
#> 5           NA          NA          NA          NA           NA    NA
#> 6           NA          NA          NA          NA           NA    NA
#> 7           NA      709.97          NA      709.97       709.97    NA
#> 8           NA          NA      786.75      786.75       786.75    NA
#> 9           NA          NA      720.56      720.56       720.56    NA
#> 10          NA          NA      840.29      840.29       840.29    NA
#>    doy.2 doy.3 doy.4 doy.5 doy.6 doy.7 doy.8 doy.9 doy.10 JulianDate.1
#> 1     NA    NA    NA   154    NA    NA    NA    NA     NA            1
#> 2     NA    NA    NA   201    NA    NA    NA    NA     NA            1
#> 3     NA    NA    NA   155    NA    NA    NA    NA     NA            1
#> 4     NA    NA    NA   153    NA    NA    NA    NA     NA            1
#> 5     NA    NA    NA   153    NA    NA    NA    NA     NA            1
#> 6     NA    NA    NA   153    NA    NA    NA    NA     NA            1
#> 7     NA    NA    NA    NA    NA   207    NA   156    211            1
#> 8     NA    NA    NA    NA    NA    NA   208   156    211            1
#> 9     NA    NA    NA    NA    NA    NA   208   156    211            1
#> 10    NA    NA    NA    NA    NA    NA   208   156    211            1
#>    JulianDate.2 JulianDate.3 JulianDate.4 JulianDate.5 JulianDate.6
#> 1             2            3            4            5            6
#> 2             2            3            4            5            6
#> 3             2            3            4            5            6
#> 4             2            3            4            5            6
#> 5             2            3            4            5            6
#> 6             2            3            4            5            6
#> 7             2            3            4            5            6
#> 8             2            3            4            5            6
#> 9             2            3            4            5            6
#> 10            2            3            4            5            6
#>    JulianDate.7 JulianDate.8 JulianDate.9 JulianDate.10
#> 1             7            8            9            10
#> 2             7            8            9            10
#> 3             7            8            9            10
#> 4             7            8            9            10
#> 5             7            8            9            10
#> 6             7            8            9            10
#> 7             7            8            9            10
#> 8             7            8            9            10
#> 9             7            8            9            10
#> 10            7            8            9            10


Hopefully that's doing what you want it to.

Best,
Adam

Mark Thompson

unread,
Jul 31, 2019, 6:03:33 PM7/31/19
to unmarked
Wow! Thank you!

This almost accomplishes what I am after. A few things:

1. I think you have to specify the yearlySiteCovs:

ammaMF <- formatLong(amma, type = "unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)

Otherwise, they do not appear in ammaMF

2. The covariates need to be transformed. I note in your teaching post that you had done the following:

covars <- select(occ2012, -date, -site, -observer, -spp, -count, -year, -visit, -wind)
covar_meds <- apply(covars, 2, median)
covar_mads <- apply(covars, 2, mad)
# Now scale them in the original data set
which_cols <- which(names(occ2012) %in% names(covar_meds)) # Identifies columns to standardize
occ2012[, which_cols] <- median_scale(occ2012[, which_cols])

I took the following approach (changed here from my original script:

# Standardizing covariates Mean-Xi/SD (see Wilson and Schmidt. (2015).Ecosphere 6(9):168. )
# Obtain covariates
covars <- select(FWCPdata, wetdist.3, cutdist.4, rddist.5)

# Now scale them in the original data set
which_cols <- which(names(FWCPdata) %in% names(covars)) # Identifies columns to standardize
covar_mean <- apply(covars, 2, mean, na.rm = TRUE)
covar_sd <- apply(covars, 2, sd, na.rm = TRUE)
covar_wet <- (covar_mean[1] - (apply(covars[1], 2, function(x) x)))/covar_sd[1]
covar_cut <- (covar_mean[2] - (apply(covars[2], 2, function(x) x)))/covar_sd[2]
covar_rd <- (covar_mean[3] - (apply(covars[3], 2, function(x) x)))/covar_sd[3]
FWCPdata[, which_cols] <- cbind(covar_wet,covar_cut,covar_rd)


It seems necessary to scale the covariates. However, ammaMF output looks as follows:

> str(ammaMF)

Formal class 'unmarkedMultFrame' [package "unmarked"] with 7 slots
 
..@ numPrimary    : num 5

 
..@ yearlySiteCovs:'data.frame': 1030 obs. of  1 variable:
 
.. ..$ amma.visit: num [1:1030] 1 2 3 4 5 6 7 8 9 10 ...
 
..@ y             : int [1:103, 1:10] NA NA NA NA NA NA NA NA NA NA ...
 
..@ obsCovs       :'data.frame': 1030 obs. of  3 variables:
 
.. ..$ cutdist.4 : num [1:1030] NA NA NA NA 2.21 ...
 
.. ..$ doy       : num [1:1030] NA NA NA NA 154 NA NA NA NA NA ...
 
.. ..$ JulianDate: num [1:1030] 1 2 3 4 5 6 7 8 9 10 ...
 
..@ siteCovs      :'data.frame': 103 obs. of  5 variables:
 
.. ..$ spp.6    : Factor w/ 1 level "amma": 1 1 1 1 1 1 1 1 1 1 ...

 
.. ..$ place.1  : Factor w/ 2 levels "m","p": 1 1 1 1 1 1 1 1 1 1 ...
 
.. ..$ vri.2    : Factor w/ 9 levels "AT","LA","PL",..: 9 7 9 3 3 9 9 9 9 9 ...

 
.. ..$ wetdist.3: num [1:103] -7.782 0.539 -1.407 -0.58 -0.88 ...
 
.. ..$ rddist.5 : num [1:103] 0.787 -1.478 1.607 -1.735 1.819 ...

 
..@ mapInfo       : NULL
 
..@ obsToY        : num [1:10, 1:10] 1 0 0 0 0 0 0 0 0 0 ...


Why is cutdist.4 in my obsCov and not in siteCovs? JulianDate is still being introduced.

Thanks!

Mark Thompson

unread,
Jul 31, 2019, 6:50:23 PM7/31/19
to unmarked
I have learned that there is something particular about the data in cutdist.4 that puts it into the obsCovs. If I run the following:

FWCPdata$cutdist.4 <- FWCPdata$wetdist.3


cutdist.4 stays with siteCovs. The creation of JulianDate is still a problem. I've tried a different transformation on cutdist.4, but this doesn't seem to fix the problem.

Adam Smith

unread,
Aug 1, 2019, 10:04:22 AM8/1/19
to unmarked
Hi again Mark,


1. I think you have to specify the yearlySiteCovs:

ammaMF <- formatLong(amma, type = "unmarkedMultFrame", numPrimary = 5, yearlySiteCovs = visit)

Please see Dan Linden's nice description of these types of covariates. A yearlySiteCov is a variable the varies among sites in a given year, and within a site(s) over multiple years.

None of your covariates fits this description and, as such, you do not have any yearlySiteCovs.

FWCPdata <- read.csv("~/FWS_Projects/sandbox/Data/FWCPDataLongTsp.csv", header = TRUE, stringsAsFactors = FALSE) %>%
  # Drop records with no response; we'll complete data set below
  filter(!is.na(y))

# Explore covariate temporal variation within sites
cov_temp <- group_by(FWCPdata, site) %>%
  summarise_at(vars(place.1, vri.2, wetdist.3, cutdist.4, rddist.5), n_distinct)
summary(cov_temp)
#>      site              place.1      vri.2     wetdist.3   cutdist.4    
#>  Length:103         Min.   :1   Min.   :1   Min.   :1   Min.   :1.000  
#>  Class :character   1st Qu.:1   1st Qu.:1   1st Qu.:1   1st Qu.:1.000  
#>  Mode  :character   Median :1   Median :1   Median :1   Median :1.000  
#>                     Mean   :1   Mean   :1   Mean   :1   Mean   :1.049  
#>                     3rd Qu.:1   3rd Qu.:1   3rd Qu.:1   3rd Qu.:1.000  
#>                     Max.   :1   Max.   :1   Max.   :1   Max.   :2.000  
#>     rddist.5
#>  Min.   :1  
#>  1st Qu.:1  
#>  Median :1  
#>  Mean   :1  
#>  3rd Qu.:1  
#>  Max.   :1

This summary does highlight why `cutdist.4` is being treated as an obsCov rather than a siteCov. There must be a few sites that have different values within a site within a year, likely some kind of data entry error based on your surprise at it being treated different than you other siteCovs.

Sure enough, looks like `cutdist.4` values change within a handful of sites:

# What's up with cutdist.4?
weird_cutdist <- filter(cov_temp, cutdist.4 > 1) %>% pull(site)
filter(FWCPdata, site %in% weird_cutdist, spp.6 == "amma") %>%
  mutate(date = dmy(date)) %>% # Convert string to date)
  arrange(site, spp.6, date)
#>    site       date y y.count place.1 vri.2 wetdist.3 cutdist.4 rddist.5
#> 1  p.44 2014-05-13 0       0       p    TC     42.96    779.07   295.73
#> 2  p.44 2014-06-17 0       0       p    TC     42.96    779.07   295.73
#> 3  p.44 2015-06-04 1       1       p    TC     42.96    779.07   295.73
#> 4  p.44 2015-08-09 0       0       p    TC     42.96    779.07   295.73
#> 5  p.44 2016-05-16 0       0       p    TC     42.96    779.07   295.73
#> 6  p.44 2016-07-27 0       0       p    TC     42.96    832.21   295.73
#> 7  p.47 2014-05-13 0       0       p    TC    133.11    692.67   264.37
#> 8  p.47 2014-06-17 0       0       p    TC    133.11    692.67   264.37
#> 9  p.47 2015-06-04 0       0       p    TC    133.11    692.67   264.37
#> 10 p.47 2015-08-09 0       0       p    TC    133.11    880.01   264.37
#> 11 p.47 2016-05-16 0       0       p    TC    133.11    880.01   264.37
#> 12 p.47 2016-07-27 0       0       p    TC    133.11    880.01   264.37
#> 13 p.48 2015-06-04 0       0       p    TM     75.77    880.01   452.11
#> 14 p.48 2015-08-09 0       0       p    TM     75.77    783.21   452.11
#> 15 p.50 2015-06-04 0       0       p    TC    345.67    806.51   328.85
#> 16 p.50 2015-08-09 0       0       p    TC    345.67    817.46   328.85
#> 17 p.50 2016-05-15 0       0       p    TC    345.67    817.46   328.85
#> 18 p.50 2016-07-28 0       0       p    TC    345.67    817.46   328.85
#> 19 p.51 2014-06-16 0       0       p    TC    753.59    817.46   187.64
#> 20 p.51 2015-06-04 0       0       p    TC    753.59    535.95   187.64
#> 21 p.51 2015-08-10 0       0       p    TC    753.59    535.95   187.64
<clip>

If these are data entry errors, fixing them to match the other `cutdist.4` values within a site will move `cutdist.4` back into the siteCovs group.

2. The covariates need to be transformed. I note in your teaching post that you had done the following:

The covariates don't necessarily need to be transformed, but it can improve computation efficiency/speed in some case. I applied an slightly unusual transformation in that example for various reasons. You can use whatever approach makes most sense---no transformation, more typical approaches (see `?scale`), etc...
 
Why is cutdist.4 in my obsCov and not in siteCovs? JulianDate is still being introduced.

`JulianDate` is simply what `formatLong` calls the variable you denoted as the "observation number" variable. In your case, this was `visit`. Thus, `JulianDate` is simply the same thing as what was originally called `visit`. 
 
Best,
Adam

Mark Thompson

unread,
Aug 1, 2019, 12:37:53 PM8/1/19
to unmarked
Thank you very much Adam!

I had put yearlySiteCovs as visit based on the information in Fiske et al. (2019) Package 'unmarked' where they put 'years' in as 'yearlySiteCovs' under 'unmarkedMultFrame' (pg. 111). I previously had a layer called 'visitYr' and got this confused with the renaming of my seasonal revisit time to 'visit'. I'm thinking I should create a variable out of date that marks years:

years: data.frame(year=c(t(years)))


Data entry error with cutdist.4 was next on my list to check just before you submitted your response - so thank you.

I believe everything else is sorted out now.

You have been a tremendous help.

Adam Smith

unread,
Aug 1, 2019, 3:06:10 PM8/1/19
to unma...@googlegroups.com
Very good. Yes, you could create a "year" yearly site covariate a la the unmarked documentation example to model a linear annual trend in occupancy (or separate annual estimates if you made it categorical)... 

Good luck,
Adam

--
You received this message because you are subscribed to a topic in the Google Groups "unmarked" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/unmarked/jaqo2hK94es/unsubscribe.
To unsubscribe from this group and all its topics, send an email to unmarked+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/unmarked/e470ac98-ae25-42de-9ef1-1a58d5c801fb%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages