In eriqande/CKMRpop: Forward-in-Time Simulation and Tallying of Pairwise Relationships

Trying to get some stuff for a grant proposal.

library(tidyverse)
library(CKMRpop)

# get the data generated on the cluster in 006
crel <- read_rds("../cluster-run-one-big-kelp-rf-pop/crel.rds")

samples <- read_rds("../cluster-run-one-big-kelp-rf-pop/samples.rds")

relat_counts <- read_rds("../cluster-run-one-big-kelp-rf-pop/relat_counts.rds")

Now we look those over. First, the highly summarized relationships:

relat_counts$highly_summarised

s2 <- samples %>%
  mutate(
    samp_year = unlist(samp_years_list),
    age = samp_year - born_year
  )

Look at distribution of ages of the sampled fish:

ggplot(s2, aes(x = age)) + 
  geom_histogram(binwidth = 1)

Cool, that is what it should look like. So, get the number of juveniles and the number of adults sampled:

s2 %>%
  mutate(adult = age > 2) %>%
  count(adult)

OK, in Diana's study there were 1847 adults and 4244 juveniles.

So, the number of adult-juvie pairs in Diana's study would be

1847 * 4244

Roughly 7.8 million. And, if she found 8 PO pairs, that is suggesting a pool of about 1 million parents.

The number of adult-juvie pairs in the simulation study is:

6679 * 2925

About 19.5 million.

And we found 61 PO pairs in the simulation. If we had the same number of pairs as diana we would have expected.

61 * (7.8 / 19.5)

About 24. And so, this suggests that we actually overpredicting the number of pairs that we might see by a factor of a little less than 3.

It is hard to know how to deal with this. On the one hand, self-recruitment might be quite low in Monterey and Carmel Bay, in which case the possible candidate parents contributing might be considerably higher than the number just in Monterey Bay, and it might be that most of the parent sampled in Monterey Bay don't have much of a chance of created offspring that stay there. I suspect that is likely. On the other hand, if self-recruitment is quite high, then each little segment of the coast is separate, in which case, from Diana's result, that would mean that the number of adults within Carmel Bay was actually very high (like 1.5 million). That just seems way beyond what Dan and Mark think. So, clearly there is a lot of dispersal I would say.

So, to first approximation, let's say that it is all well mixed. That means that our expected numbers of different kin categories might be closer to:

counts <- relat_counts$highly_summarised %>%
  mutate(adjusted = n * 24 / 61)
counts

Let's put that in (roughly) decreasing kinship order:

newc <- counts %>%
  mutate(relat = factor(dom_relat, levels = relationship_zone_names)) %>%
  arrange(relat, desc(max_hit)) %>%
  select(relat, max_hit, adjusted) %>%
  mutate(adjusted = floor(adjusted))

newc

eriqande/CKMRpop documentation built on Jan. 25, 2024, 2:10 p.m.