model
{
	Hyper.gamma <- 1.0E-2
	Tau.error <- 1.0E+1
	Tau.noninformative <- 1.0E-4

	# observation
	for (i in 1:N.sample) {
		Flc[i] ~ dnorm(density.flc[i], Tau.error)
		density.flc[i] <- exp(log.max) / (1 + exp(-logit.flc[T.flc[i]]))
	}
	log.max ~ dnorm(4.0, 0.2)

	# time change of density
	for (j in 1:N.id) {
		#logit.flc[Time.start[j]] ~ dnorm(0, Tau.noninformative)
		logit.flc[Time.start[j]] ~ dnorm(logit.flc[Time.start[j] + 1], tau)
		for (t in Time.start[j]:(Time.end[j] - 1)) {
			# flc
			logit.flc[t + 1] ~ dnorm(l.flc[t], tau)
			l.flc[t] <- log(p.flc[t]) - log(1 - p.flc[t])
			p.flc[t] <- (
				q.flc[t] * (1 - r[t, 1])
				+ (1 - q.flc[t]) * r[t, 2]
			)
			logit(q.flc[t]) <- logit.flc[t]
			# rates
			logit(r[t, 1]) <- b[1] + b[2] * Temp[t]
			logit(r[t, 2]) <- b[3] + b[4] * Temp[t]
		}
	}

	# parameters
	for (k in 1:N.b) {
		b[k] ~ dnorm(0.0, Tau.noninformative)
	}
	tau ~ dgamma(Hyper.gamma, Hyper.gamma)
}