[Bug]: error in time delay between spawning and settlement when settlement age is not equal to 0

[Rick-Methot-NOAA]

Comparison of 3 situations discloses a problem:
The Jan run had recruitment on Jan 1 at age 1, a year after spawning.
The Dec run shifted settlement to ~Dec 31 at age 0, so at end of the year of spawning. such that they would become age 1 the following day (not implying that SS3 works in days, only in decimal years).

The two runs produced nearly identical logL, but the sequence of recdevs is shifted by a year. A big recdev in 1988 for the Dec run becomes a big recdev in 1989 for the Jan run. So the labeling of recdevs is referring to the calendar year in which the settlement happens.

The settlement mechanism allows for the total recruitment to be divided among multiple settlements, which would seem to confuse that recdev naming protocol. Another test in which the recruits are evenly split between dec 31 age 0 and Jan 1 age 1 confirms that the code is not handling it as intended because the recdevs are now high for both 1988 and 1989.

Need to check the components that go into this assignment:
natage(t + Settle_seas_offset(settle), p, g, Settle_age(settle)) +=
Recruits * recr_dist(y, GP(g), settle, p) * platoon_distr(GP2(g)) *
mfexp(natM(t, p, GP3(g), Settle_age(settle)) * Settle_timing_seas(settle));

[Rick-Methot-NOAA]

After the spawning event produces the total number of recruits, they are placed forward in time into the N-at-age matrix.
The error was in the calculation of how far forward in time when age at settlement was > 0.

This line in read_control:
Settle_seas_offset(settle_time) = Settle_seas(settle_time) - spawn_seas + j * nseas; // number of seasons between spawning and the season in which settlement occurs

needed to be:
Settle_seas_offset(settle_time) = Settle_seas(settle_time) - spawn_seas + j * nseas + Settle_age(settle_time) * nseas; // number of seasons between spawning and the season in which settlement occurs

This change also verifies that the year referred to in recdev(y) is the year of spawning.

With this change, the results for the Dec settlement, the Jan settlement at age 1, and the run with recruitment split between those two settlement times are now very close to identical and all place the large recdevs in the same year.

• @e-perl-NOAA For the manual: Recruitment year is the year in which spawning occurred, and which also is the year in which the recruits are age 0. Recdevs are labelled with this year. Settlement events the following year begin at age 1.

Note that assessments that previously had recruitment at age 1 in the year after spawning will see that the recdev time series will shift by a year. This could cause a small change in results because you will lose one year of devs at the beginning.