what to do with bird spp not in all BCRs?

[achubaty]

@ilaneshaw @SteveCumming for the moment, I'm skipping processing of species that don't have >1 bootstrap replicate for all 14 BCRs.

This means species that are A) not found in all BCRs, or B) haven't had all their maps uploaded are being skipped.

A priori I can't tell which case is which. I'd need to dig into the bird biology to determine if it's A vs. B.

So my question for you: is it sensible to omit these (for now or otherwise)?

My guess is not, for case B. So we will need to double check the mosaic code to handle situations where the resulting mosaic only covers a portion of all the BCR areas, and downstream, the user needs to be made aware that their study area may have species missing in portions of it because of ths.

[achubaty]

@tati-micheletti can you weigh in on this too please?

[tati-micheletti]

So, I think that this is highly likely case A (I know Diana automates the code, so she probably wouldn't manually suppress some species for some BCR's). What I would do (although I don't know exactly how the code was built, so this might be a tricky thing to do), is to forecast all species in all BCR's that they do exist, and then make a zeroed raster for the BCRs where these species don't exist. This is what I did for the backcasting. The rasters's borders "should" be very close in terms of abundance if the reason why these are being omitted is lack of data.

[SteveCumming]

I mean ideally we could account for variation in the number of bootstrap replicates among species and BCRs, so that we don't have to depend on access to completed models.....Isolde's code had attempted to do that, simply by counting the number of replicated that were present at each level. Variance estimates for n=1 would be NA

[achubaty]

@SteveCumming Yes, we track the number of replicates so that can be included downstream. My question here is what to do with species for which we don't have layers in some BCRs.

Tati's suggestion of filling in the missing regions with zeroes (or NA?) would work, provided the reason we don't have those rasters is because the species isn't found in that region.

Probably something @ilaneshaw should check on and confirm before I proceed with implementing Tati's suggestion.

[ilaneshaw]

Sorry for the slow reply- I was out getting a covid test, annoyingly with only one week to go before I am due to get the vaccine... 😐

Yes, I was thinking that perhaps the most sensible thing is for me to send an email to Diana to clarify if it is situation A or B is for most of these cases.

Steve- are there any particular species that you know Sam will definitely want to be present? Or does the data just have to be available for all species that Diana has models for?

Thanks, Isolde

[SteveCumming]

I think we should process all species, BCRs and replicates as provided by BAM

On Thu, May 20, 2021 at 12:18 PM ilaneshaw @.***> wrote:

Sorry for the slow reply- I was out getting a covid test, annoyingly with only one week to go before I am due to get the vaccine... 😐

Yes, I was thinking that perhaps the most sensible thing is for me to send an email to Diana to clarify if it is situation A or B is for most of these cases.

Steve- are there any particular species that you know Sam will definitely want to be present? Or does the data just have to be available for all species that Diana has models for?

Thanks,

Isolde

Sent from Mail https://go.microsoft.com/fwlink/?LinkId=550986 for Windows 10

From: Alex Chubaty @.> Sent: Thursday, May 20, 2021 14:51 To: ilaneshaw/bootRasterCombine @.> Cc: ilaneshaw @.>; Mention @.> Subject: Re: [ilaneshaw/bootRasterCombine] what to do with bird spp not in all BCRs? (#1)

@SteveCumming https://github.com/SteveCumming Yes, we track the number of replicates so that can be included downstream. My question here is what to do with species for which we don't have layers in some BCRs.

Tati's suggestion of filling in the missing regions with zeroes (or NA?) would work, provided the reason we don't have those rasters is because the species isn't found in that region.

Probably something @ilaneshaw https://github.com/ilaneshaw should check on and confirm before I proceed with implementing Tati's suggestion.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub < https://github.com/ilaneshaw/bootRasterCombine/issues/1#issuecomment-845384966

, or unsubscribe < https://github.com/notifications/unsubscribe-auth/AIPWPPIXB7RJ4IG2MXFIETDTOVK2FANCNFSM45F7TWSA

.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ilaneshaw/bootRasterCombine/issues/1#issuecomment-845409151, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABXH6O4WLGC2CMDKWJZPO53TOVOA7ANCNFSM45F7TWSA .

-- Steve Cumming Professeur agrégé Département des sciences du bois et de la forêt Université Laval (418) 656-2131 poste 2593

[achubaty]

any update on this? @ilaneshaw did Diana confirm with you?

I'll be returning to this in July, so I'd like this sorted out before them.

[ilaneshaw]

Diana said that she has cycled through all species and BCRs that she is currently planning to do. If there are simply only one or a few replicates of a particular BCR, it might be that she just has not created all the bootstraps for it yet. However, for the cases where there is a BCR missing entirely she said it will be because the models could not be built for that BCR. She has not confirmed why this would be yet (she is away on fieldwork at the moment I think), but I assume lack of data?