An environmental scientist measures stream temperature daily for 365 days to estimate the annual mean. Consecutive daily temperatures are highly autocorrelated: r = 0.96 (tomorrow closely resembles today). If assuming independence, the standard error would be $\text{SE}_{\text{naive}} = \frac{\sigma}{\sqrt{n}} = \frac{5°C}{\sqrt{365}} ≈ 0.26°C$. What is the actual standard error accounting for autocorrelation?