Reef fish temperature index followed sea temperature across 2,781 Australian reef sites, but changed only 0.15 °C per °C year to year at 333 sites in 1992–2025, with lags and persistence

Question: does the community temperature index (CTI) of reef fish track sea surface temperature (SST) on Australian reefs, in space and in time? Approach: we used Reef Life Survey, ATRC and Parks Victoria Method 1 fish counts (31,187 surveys, 2,790 sites, 1992-2025). We calculated the species thermal index (STI) of 1,579 species from OBIS presence cells, without the RLS records, and NOAA OISST v2.1. CTI is the log-abundance-weighted mean STI. Survey SST is the OISST mean of the 12 months to the survey. We wrote the tests before we calculated CTI: spatial slope, year-to-year first differences, site trends, lag models and an out-of-sample test. Null models shuffle STI among species. Results: across sites, CTI followed SST with a slope of 0.94 (R2 0.94). At 333 sites with 8 or more survey years (mostly NSW, WA, Tasmania, Victoria and SA), CTI changed 0.15 C per C of SST change between surveyed years (95% CI 0.11-0.19; permutation p = 0.019). SST from 1-3 years before the survey predicted CTI as well as or better than current SST. The summed lag 0-5 coefficient was 0.51. Small fish responded in the same year; larger fish responded later. Mean CTI rose 0.21 C per decade and mean SST rose 0.20 C per decade, but sites that warmed more did not gain more CTI (across-site slope -0.43). In 2008-2025 at 179 sites, the CTI trend was 0.54 of the SST trend. At 42 Jurien Bay sites, CTI was 1.07 C (0.95-1.18) higher in 2015-2021 than in 1999-2009, after the 2011 marine heatwave, while SST at survey time was 0.31 C lower. A model trained on years to 2012 predicted CTI at 243 sites in 2013-2025 only 1.7% better than no change. Space-for-time and full-tracking predictions were 45-52% worse. Exploratory: cooling steps lowered CTI more (0.28) than warming steps raised it (0.05). Conclusion: reef fish composition matches SST in space, but its response in time is weak, delayed and path dependent. A match of mean long-term trends is not evidence of site-level tracking. Limitations: STI from OBIS has sampling bias and some wrong identifications (sensitivity versions agree). Divers are not identified in the public data. OISST does not resolve nearshore temperature. The panel is temperate and subtropical, not the Great Barrier Reef.

community temperature index · reef fish · Reef Life Survey · sea surface temperature · climate tracking · marine heatwave · Western Australia · Jurien Bay · tropicalization · species thermal index · OBIS · OISST · lagged response · hysteresis · out-of-sample prediction