This document records what SPoRC development is working
towards, what is saved for later to be revisited, and what is likely not
in the development plan (out of scope) .
Three tracks are under active development: structural equation models linking the deviation processes to each other and to environmental covariates, process error on the remaining parameter blocks, and electronic tagging. Close-kin mark-recapture is planned, but has not been started.
Work to date has involved the following released versions:
Released versions
| Version | Date | What it added |
|---|---|---|
| 1.0.0 | 2025-11-24 | First public release. Age, sex, season and region structured dynamics, closed loop simulation and management strategy evaluation. |
| 1.1.0 | 2026-03-31 | Movement estimated as a continuous time Markov chain with preference functions. |
| 1.2.0.9000 | in development | Population specific (natal homing) and seasonal dynamics, discarding and retention, time varying and semi-parametric growth, conditional age at length, at age data sources, estimated index observation error, per fleet ageing error, numbers at age as a state space process, and catchability process error. |
In progress
Structural equation models and environmental covariates
Recruitment, the state space numbers at age, time varying and semi-parametric growth, continuous movement and catchability each estimate annual deviations, and each takes its penalty from its own module. Putting those deviations on one grid alongside environmental covariates, with estimated paths between them, would let a covariate enter any of these processes and let one process inform another. Structural equation models are under consideration for this.
Process error on the remaining parameter blocks
Selectivity and growth already take a deviation surface with a choice of covariance, fishing mortality takes independent, random walk or first order autoregressive deviations, and catchability now takes those same three forms. Natural mortality is what is left: it is constant within a population, region and time block. Giving it annual deviations is also what a temperature or body condition covariate on natural mortality would attach to.
Electronic tagging
Conventional tags inform movement through recaptures. Recaptures are sparse, they happen only where the fishery is, and they arrive filtered through a reporting rate. Archival and satellite tags record position through the fish’s life independently of the fishery, so they inform the movement rates themselves rather than the movement rates confounded with fishing processes. An experimental implementation was written and taken back out of the package while the movement parameterization changed, and it is being reworked onto the current one.
Implemented
Numbers at age as a state space process
Recruitment used to be the only stochastic element of the population state. From age two onwards, numbers at age were a deterministic consequence of recruitment, mortality and movement, so any misspecification in natural mortality, in selectivity or in the reported catch was absorbed by the recruitment deviations. Process error on the whole numbers at age surface, as in SAM and WHAM, lets that misspecification enter at the age it occurs instead. The state runs at season boundaries as well as year boundaries. It is not advanced by the deterministic projection, so a forecast from a state space fit omits that process error, but the closed loop operating model does advance it.
To be revisited
Close-kin mark-recapture
Close-kin mark-recapture estimates absolute abundance from the rate at which pairs of sampled fish turn out to be related. The quantities it needs are already in the report file: numbers at age by population, region, year and sex, maturity at age, and weight at age, from which relative reproductive output follows. What is missing is the pairwise comparison data and a likelihood over it. It is planned, and no work has started on it.
Outside current scope
| Capability | Rationale |
|---|---|
| Multispecies models and predation mortality |
SPoRC is designed for single-species population
dynamics, potentially with multiple populations or regions. Multispecies
interactions and predation mortality are therefore outside the current
scope of the framework. |
| Fully length-structured dynamics |
SPoRC uses age as the primary population-structure
dimension. Length-based observations and selectivity can be represented
through age–length relationships without requiring the population
dynamics themselves to be fully length structured. A fully
length-structured implementation would substantially expand the model
structure and computational burden without being necessary for the
primary applications of SPoRC. |
| Hermaphroditism and sex change |
SPoRC assumes sex is fixed after recruitment. Given
that hermaphroditism and sex change are relevant to relatively few
species and are not a primary focus of the intended applications, they
are outside the current scope of the framework. |