Estimated groundfish density for key SSL prey species in BS/AI region

[tristan-sebens]

Data product requested:
Estimated density across space, by species

Output type:
CSV or data.frame of predicted point density

Species:
Pollock, PCod, Atka Mackerel, all rockfish (aggregated) EDIT: shortraker rockfish (I was mistaken), all skates (aggregated)

Region :
EBS Shelf and Aleutian Islands

Research team making the request:
Tristan Sebens (@tristan-sebens) [email protected]
CADQ branch of Sustainable Fisheries, AKRO

I've been asked to recreate a series of figured that Cecilia O'Leary (@coleary-noaa) created for the GOA Biological Assessment report that the SF recently finished. Here's an example of one of these figures:

These images indicate the spatial distribution of estimated biomass and commercial catch relative to the locations of Stellar Sea Lion rookeries/haulouts. When I took over the figures for the GOA report, Cecilia had already generated the estimated spatial distributions for each of the species. Now the division has moved on to writing the same report for the BS/AI.

I spoke with Cecilia, and she told me that she estimated the spatial distribution using GAP's standard process, since she was a part of the team. I'm wondering if it is possible to have those same distributions estimated for the above species in the BS/AI region.

I'm also about to sound REALLY high maintenance, which I apologize for, but wondering if it would be possible to have this in the next couple of weeks? I'm supposed to have these figures finished by the end of February (2025). I know I should have requested this sooner, I just didn't realize the kind of data I was going to need to put them together. Thanks in advance for your help and your patience :)

The GOA data came in .csv format, and included the following fields:

• year
• density - the estimated density
• geometry - Point vector indicating the estimate location

I've attached one of the datasets as an example. Please let me know if this is something you can help me with, and/or if there's anything else I can provide. Thanks very much!

pcod_average_density.csv

[zoyafuso-NOAA]

Hi @tristan-sebens,

Were mean densities summarized directly from observed survey densities or via some model-derived product like VAST/sdmTMB? What was the report that was published that Cecilia contributed (this will help us with methods)?

The AI and Bering Sea surveys are conducted with different trawls so plotting densities on the same map might be challenging as a result. Currently we don't have a standard way to correcting for potentially different catchabilities between the gears. Is the idea to plot these regions separately or as one BSAI region?

[tristan-sebens]

Hi Zack,

The densities were model-derived products. Cecilia's code extracted the densities from a fitted VAST model. The report was the GOA Biological Opinion report published Dec 23 2024.

In terms of BS vs AI, it's fine for them to be separate sets of estimates. It's not critical that they be plotted in a single figure.

[tristan-sebens]

I was also wrong about the Rockfish. I thought we needed all Rockfish aggregated, but turns out we only need Shortraker rockfish.

[zoyafuso-NOAA]

Hi Tristan,

Given the very short amount of time we have here, I don’t think we have the capacity to fulfil model-based spatial distributions for the species and regions listed in the same way it was done in the past. But, I do have some alternatives that could be provided in the time frame:

Aleutian Islands: We currently don't have any standard process to conduct model-based indices for the AI, nor do we have an internal plan to prioritize going into the direction of extending our model-based efforts towards the AI. Most of the model-based spatiotemporal distribution effort used for is used to support the GOA and EBS/NBS species requests and there’s a formal yearly process that goes into this. This is to make sure we’re not overburdening our staff, these models are gigantic and take a while to run and maintain.

What we usually produce is something like this below (from the AI 2022 data report). It’s just the observed densities as a bubble map plot, we can easily get the lat/lon, observed density, and years for you. Would something like this suffice?

Bering Sea: pcod and pollock have model-based output that could be of use, we would need to ask Lewis Barnett if he thinks it would be appropriate to co-opt the pcod and pollock modsquad production products from 2024. Atka mackerel occur in very low frequencies (fewer than a handful of stations) in the Bering Sea and shortraker RF have not been observed. We do not currently have a skate model.

What we provide in the data reports are inverse distance weighted densities maps, so an interpolated observed density map (see below). We can provide something like this to you pretty quickly.

Unfortunately, this is all we can provide to you. Thank you for your understanding and please let me how to proceed.

[tristan-sebens]

Hi Zack,

I understand, and thanks for your work on this. I'll run this by my people and see what they want to do.