$$nll(x) = \sum{-log(\frac{1}{{\sigma \sqrt {2\pi } }}e^{{{ -
\frac{\left( {C - \hat{C} } \right)^2}{2 \sigma^2}
}}}}) $$
With the right indices to make catches by week make sense.
Chapter 1 Results
Are there seasonal patterns in weekly exploitation rates?
Is there evidence, in the depletion series, for the
commonly held belief that more than \( 90 \% \) of the legal
crabs are harvested?
$$cv=.05$$
Chapter 2 Vulnerability Regime
Adding Size Composition
Able to estimate a scalar natural mortality
Likelihood profiles of various vulnerability regimes
Time-Dependent Vulnerability Hypothesis
Crabs feed more intensely just prior to and
after hardening up after a molt and make themselves more vulnerable
to capture during these shell conditions
There are simply seasonal
changes in vulnerability, these changes do not depend
on shell condition or number of traps in the water.
Crabs aggregate due to the presence of baited
traps, increasing catch rate
Shell condition vulnerability
The Catch equation becomes
$$C=qv_{a,t}E_tA_t$$
\begin{eqnarray}
v_{a,t}&=&1 \mbox{ if } SC_a=1 \\
v_{a,t}&=&\alpha \mbox{ if } SC_a\ne 1
\end{eqnarray}
Seasonal Vulnerability
The Catch equation becomes
$$C=qv_{a,t}E_tA_t$$
Crabs are fully vulnerable from May until September.
These dates are arbitrary.
\begin{eqnarray}
v_{a,t}&=&1 \mbox{ if } 18\leq t \leq36 \\
v_{a,t}&=&\alpha \mbox{ otherwise }
\end{eqnarray}
Shortened period of full vulnerability
\begin{eqnarray}
v_{a,t}&=&1 \mbox{ if } 22\leq t \leq26 \\
v_{a,t}&=&\alpha \mbox{ otherwise }
\end{eqnarray}
Effort Vulnerability
\begin{eqnarray}
\label{eq:effortvul}
VW_{t+1}&=&.4\frac{E_t}{40000}+(1-.4)VW_t\\
v_{a,t}&=&1 \mbox{ if } \alpha\cdot VW_t >=1 \mbox{ or } SC_a=1 \mbox{ and } weeks<=2\\
v_{a,t}&=&\alpha VW_t
\end{eqnarray}
Time-Independent Vulnerability
Vulnerability varies randomly between crabs
Vulnerability is constant between all crabs.
Variable vulnerability
$$V_{a}\sim U(0,1)$$
Negative Log-Likelihood
$$nll(x) = \sum{-log(\frac{1}{{\sigma \sqrt {2\pi } }}e^{{{ -
\frac{\left( {C - \hat{C} } \right)^2}{2 \sigma^2}
}}}})-\sum{S \cdot log(\hat{S}) W_{sample} }$$
With the right indices to make proportions at size for each year and
catches by week make sense.
Chapter 2 Results
How to figure out quantitative vulnerability scenarios?
What drives effort?
What Management Scenarios and indicators am I missing?
Seasonal NLL
Seasonal Short NLL
Shell NLL
Effort NLL
Three different rules for setting opening and closing
dates
Four different vulnerability hypothesis: constant,
variable, shell condition and effort.
Two different models of effort: a linear
function of cpue in the previous week and a
linear function of cpue in the previous week and
average windspeed in present week.
\begin{eqnarray}
Effort_{t+1}&=& \beta (catch_t)^\alpha\\
\alpha&=&0.54\\
\beta&=&1.59\\
\end{eqnarray}
Effort and CPUE