Need help - data has too many zero?

78 views
Skip to first unread message

tamari yu

unread,
Jan 4, 2018, 4:13:07 AM1/4/18
to distance-sampling
Dear All, 

I conducted line transect distance sampling for mammalian species in a nature reserve. Since most species were in low density, I only managed to have one species that seems to have large enough sampling size for analysis in the program. 
I am a newbie in R and Distance, and i would like to have some advice for you 

(1) My data have quite a lot of Zero Sighting distance, as most of the animals were seen on the transect itself (i think it is the nature of the animals that they like to use the road system to travel within the reserve), i wonder if this will affect the analysis and if there is any transformation needed before the analysis
I try to run the data according to the example in "montrave robin example" but it seems that the data just doesnt fit in the model, (most have highly significant P-value and data points were far off from the goodness of fit ...) 

(2)Did i miss anything or the sample size is still to small to run the analysis? Or it is just that this data just couldnt be analysed using Distance??? If so , do you have any suggestions what program i could use? 

I am stuck at the moment and unsure what to do next. I have attached the data file and the steps that i did for the analysis. 

Any advice would be useful !! Much thanks!!

Best, 
Tamari

testing_porcupine10122017.csv
R_all_porcupine_distance_sampling_.docx

Eric Rexstad

unread,
Jan 4, 2018, 9:03:53 AM1/4/18
to tamari yu, distance-sampling

Tamari

The document you provided describing your analysis is very clear.  You have drawn the correct conclusion that because > half of your detections are recorded at distance 0 from the transect, you cannot fit a reasonable detection function to the data:

> tamari <- read.csv(file.choose())
> sum(!is.na(tamari$distance))
[1] 65
> sum(tamari$distance==0, na.rm=TRUE)
[1] 33

As you note your study organism likes to use the same roads that you used for your sampling.  This violates the design assumption of distance sampling.  Transects should be placed randomly in the landscape and roads are not placed randomly.  Equally important, there should be as many animals in the strip, say between 10 and 11m from the  transect as are located between 0 and 1m from the transect.  Another way of saying this is animals should not know where the transects are located (and either avoid or gravitate toward the transects).

There really is not a remedy to apply during the analysis.  The better solution is a re-design of your survey so as to produce random placement of transects throughout the study area.
--
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 post to this group, send email to distance...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/d17ba052-5604-4646-a9a9-a751673120d1%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

-- 
Eric Rexstad
Research Unit for Wildlife Population Assessment
Centre for Research into Ecological and Environmental Modelling
University of St. Andrews
St. Andrews Scotland KY16 9LZ
+44 (0)1334 461833
The University of St Andrews is a charity registered in Scotland : No SC013532

tamari yu

unread,
Jan 7, 2018, 3:17:17 AM1/7/18
to distance-sampling
Thank you for your reply Eric! Very well explained :)

Best, 
Tamari
Reply all
Reply to author
Forward
0 new messages