High‐frequency temperature monitoring with time‐resolved wavelet analysis reveals unrecognized complexities across depth on a coral reef ecosystem

Abstract Temperature variation at fine scales of minutes to hours and ~ 1–100 m is ubiquitous in marine ecosystems and often exceeds variability at larger scales. However, we know little of how well temperatures at one location or depth can serve as proxies to represent other locations, which is important in predicting thermal patterns relevant to organismal thermal stress. We used small data loggers to record fine‐scale temporal variability in water temperature across a depth gradient, over multiple years, on the barrier reef system in Belize. Using pairwise correlation and wavelet coherence, which quantifies the time‐ and frequency‐resolved correlation between pairs of time series, we examine how water temperature varies across reef zones (backreef, forereef inner, and forereef outer) and depths, and estimate the coherence of this variability across time scales. Although spectral analysis across the entire study period suggested a high degree of similarity in temperature patterns, time‐resolved wavelet analyses at daily, weekly, and monthly periodicities revealed significant differences across reef‐zone‐by‐depth combinations. We also detected a signal consistent with the Madden–Julian Oscillation. Coherence and phase locking at biologically relevant periodicities diminished with increasing depth difference between sites, suggesting that patterns at depth were not predictable from those near the surface. Overall, our findings show that wavelet analysis can reveal fine‐scale thermal variation on coral reefs that coarser sampling misses, and they make clear that accurately predicting how organismal performance will shift in space and time under global change depends on collecting and analyzing depth‐resolved, high‐frequency, in situ data wherever possible.