search

brianstock / MixSIAR

A framework for Bayesian mixing models in R:
http://brianstock.github.io/MixSIAR/
87 stars 73 forks source link

How does uneven number of replicates affects the model? #120

Open EmiliaIOPAN opened 6 years ago

EmiliaIOPAN commented 6 years ago

Hi, I am working on a community-wide approach. My aim is to describe the main food sources of all species collected at two station (field study). To do that, I run MixSIAR on several consumers that represent macrobenthos species. Each species is treated as a separate consumer and then I am comparing the diet among the species. I have a different number of replicates for each of the species. I am wondering how does the uneven number of replicates affects the model output. Is it possible to compare diet of different species with uneven number of replicates from the statistical point of view? Thank you in advance for help. Emilia

brianstock commented 6 years ago

Hi Emilia,

Yes, you can compare the different species with different sample sizes. Two considerations come to mind:

  1. Species with lower sample sizes will have wider credible intervals.
  2. Species with lower sample sizes will be affected more by whatever prior you use. The "uninformative" prior is not really uninformative - it pulls the proportion estimates towards the generalist diet (e.g. 33%, 33%, 33% for a 3 source model). This effect is greater for low sample sizes (for both mixture and source data) and mixing polygons with less area. Because several factors are involved (mix sample size, source sample size, mix variance, source variance, and mixing polygon geometry), it's hard to say what a reasonable sample size is. Brett (2014) looked at how the influence of the "uninformative" prior changes when you change the source variance and mixing polygon geometry, keeping the mix sample size = 1.
Johnme24 commented 6 years ago

hi，Emilia and Brian， It's very lucky to see the dialogue above.I am a aftergraduate student ,majoring in marine ecology.The present job I am working on are analysing trophic relationship between diffferent species and identifing diet composition of main consumers in artifical reefs via stable isotopes,(just a little similar to yours,Emilia) .On the Mixsiar, everything seems to go well after a short tough journey ,except for several question basiclly(forgive my poor understanding).The problems are as follows: a,I wonder how to build a discrimination. csv, I did't find the approach when I read the manual,can you help me？ b，I don't quite understand the meaning of concentration dependence,in what circumstance do I select Yes？ Thank You very much.

AndrewLJackson commented 6 years ago

hi

the discrimation csv file is built manually by entering the names of your sources, names of your isotopes and then values representing the discrimination factor between sources and consumers for your specific species and tissue. Often these are not available, and so people often pick a value from a similar species and same tissue type and use that value with an amount of error included to reflect your uncertainty in this estimate (which is included as a mean +/- standard deviation). This open access paper presents a statistical method to derive discrimination factors if you have a mammal or a bird, and it also includes many relevant references for more information and approaches https://onlinelibrary.wiley.com/doi/abs/10.1111/ecog.03371

Concentration dependence is explained in this open access paper under Point 8, which also refers to the original paper that introduced the method in Don Phillips IsoConc paper http://www.nrcresearchpress.com/doi/full/10.1139/cjz-2014-0127#.W0XMuS01TUI

best wishes

andrew

--

Dr Andrew Jackson, PhD, FTCD Associate Professor Irish Research Council Laureate School of Natural Sciences, Department of Zoology Trinity College Dublin, the University of Dublin Dublin 2, Ireland.

+353 1 896 2728 | a.jackson@tcd.iemailto:a.jackson@tcd.ie Twitter: @yodacomplexhttps://twitter.com/yodacomplex http://www.tcd.ie/Zoology/research/http://www.tcd.ie/Zoology/research/groups/jackson/groups/jackson/

On 11 Jul 2018, at 10:14, Johnme24 notifications@github.com<mailto:notifications@github.com> wrote:

hi，Emilia and Brian， It's very lucky to see the dialogue above.I am a aftergraduate student ,majoring in marine ecology.The present job I am working on are analysing trophic relationship between diffferent species and identifing diet composition of main consumers in artifical reefs via stable isotopes,(just a little similar to yours,Emilia) .On the Mixsiar, everything seems to go well after a short tough journey ,except for several question basiclly(forgive my poor understanding).The problems are as follows: a,I wonder how to build a discrimination. csv, I did't find the approach when I read the manual,can you help me？ b，I don't quite understand the meaning of concentration dependence,in what circumstance do I select Yes？ Thank You very much.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/brianstock/MixSIAR/issues/120#issuecomment-404101385, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ADK1sCv1z4SpPZdFq1uyvmw2tjlYBeUXks5uFcHpgaJpZM4Q7ChE.

Johnme24 commented 6 years ago

much thanks，Andrew！