[Alex Nankivell]

Hi Eric,

 

I am doing a multi species analysis using the following formula and I can’t for life of me figure out what might be causing the warning message below?

 

head(mult.spec_2020)

 

Region.Label Area       date  transects Sample.Label repeats Effort species size distance max.temp min.temp month

1   Witchelina 4219 13/07/2020 TS-W-T1-P1            1       1      1     roo    1      129     18.1      2.5  July

2   Witchelina 4219 13/07/2020 TS-W-T1-P1            2       1      1  rabbit    1       54     18.1      2.5  July

3   Witchelina 4219 13/07/2020 TS-W-T1-P3            3       1      1     roo    1      228     18.1      2.5  July

4   Witchelina 4219 13/07/2020 TS-W-T1-P4            4       1      1     roo    1      196     18.1      2.5  July

5   Witchelina 4219 13/07/2020 TS-W-T1-P4            5       1      1     roo    1      203     18.1      2.5  July

6   Witchelina 4219 13/07/2020 TS-W-T1-P5            6       1      1     roo    1      380     18.1      2.5  July

 

dht.spec.ests_2020 <- dht2(ddf = mult.spec.est_2020,

                           flatfile = mult.spec_2020,

                           strat_formula = ~ species,

                           convert_units = 0.001,

                           stratification = "object")

 

Warning messages:

1: In `[<-.data.frame`(`*tmp*`, flatfile$Sample.Label %in% sl_diff,  :

  provided 14 variables to replace 13 variables

2: In `[<-.data.frame`(`*tmp*`, flatfile$Sample.Label %in% sl_diff,  :

  provided 14 variables to replace 13 variables

 

Help greatly appreciated.

 

Cheers

 

Alex

 

Sent from Mail for Windows 10

 

[Eric Rexstad]

Alex

Thanks for sending along your code and data.  From your survey you are trying to produce annual estimates of kangaroo abundance from surveys stretching over three years.

In this instance, you need not adopt the approach that uses `dht2` as described in the example at https://examples.distancesampling.org, designed for producing species-specific estimates.  If you wish to have annual estimates, treat year as strata (Region.Label) and use Region.Label as a covariate.  This will provide year-specific detection functions (from the covariate) and will report year-specific estimates, without the need for `dht2`.

Here's a snippet of the code:

#====================
roo_all$Region.Label <- as.factor(roo_all$year)
#====================

mult.spec.est_year <-ds(roo_all,
                        key = "hn",
                        adjustment = NULL,
                        transect = "point",
                        convert.units = 0.001,
                        formula = ~Region.Label,
                        truncation = "5%")

summary(mult.spec.est_year)
plot(mult.spec.est_year, pdf= TRUE)
gof_ds(mult.spec.est_year)

and a snippet of the output from `ds`

Summary for individuals

Summary statistics:
  Region  Area CoveredArea Effort   n        ER      se.ER      cv.ER mean.size    se.mean
1   2019  4219   44.849710    104 127 1.2211538 0.07094362 0.05809556  1.270000 0.06942040
2   2020  4219    4.743719     11   9 0.8181818 0.12196734 0.14907120  1.000000 0.00000000
3   2021  4219   10.781180     25  48 1.9200000 0.25768197 0.13420936  2.000000 0.25537696
4  Total 12657   60.374609    140 184 1.3142857 0.07455639 0.05672769  1.383459 0.07373279

Abundance:
  Label  Estimate       se        cv      lcl       ucl       df
1  2019  53290.37  6791.47 0.1274427 41487.33  68451.34 190.5947
2  2020  44820.02 22380.80 0.4993482 17633.24 113923.14 139.3424
3  2021  32670.74 12502.74 0.3826893 15738.55  67819.31 152.7468
4 Total 130781.13 26519.06 0.2027743 87963.44 194441.06 151.3944

Density:
  Label  Estimate       se        cv      lcl      ucl       df
1  2019 12.631043 1.609735 0.1274427 9.833451 16.22454 190.5947
2  2020 10.623375 5.304763 0.4993482 4.179484 27.00240 139.3424
3  2021  7.743717 2.963438 0.3826893 3.730398 16.07474 152.7468
4 Total 10.332712 2.095209 0.2027743 6.949786 15.36233 151.3944

You will note the quite large CVs on the 2020 and 2021 estimates, to be expected from the small number of detections in those years.

--
You received this message because you are subscribed to the Google Groups "distance-sampling" group.
To unsubscribe from this group and stop receiving emails from it, send an email to distance-sampl...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/527B8D0D-CA3B-48FB-AE1A-D9B2F090F9CC%40hxcore.ol.

-- 
Eric Rexstad
Centre for Ecological and Environmental Modelling
University of St Andrews
St Andrews is a charity registered in Scotland SC013532

[Isha Bopardikar]

Hello,

I had a question based on this analysis. I have surveys from 2020-2023 and wanted to get density estimates by year. I don't have enough detections across all years to build separate detection functions, so I thought to follow the method outlined here to have years as strata, but I ran into a few questions about how best to allocate effort. 
Some of the transects were not surveyed all year due to weather constraints, so effort across the years is a bit uneven. When I ran the analysis to get overall densities, I multiplied the effort by the number of times the transect was conducted; I then used this same data frame and used year as strata. My question is, when I do it this way, am I overestimating effort through the years? The estimated numbers do add up to what I get when I run it with the region, though. When I corrected for effort, the yearly estimates seemed slightly off from the main model.
Apologies if my text here is a bit confusing, I have attached the dataframe and model summaries from the analysis I ran (hope that helps!) 

Thanks so much!!