Ectotherms generally shrink under experimental warming, but whether this pattern extends to wild populations is uncertain. We analysed ten million visual survey records, spanning the Australian continent and multiple decades and comprising the most common coastal reef fishes (335 species). We found that temperature indeed drives spatial and temporal changes in fish body size, but not consistently in the negative fashion expected. Around 55% of species were smaller in warmer waters (especially among small-bodied species), while 45% were bigger. The direction of a species’ response to temperature through space was generally consistent with its response to temperature increase through time at any given location, suggesting that spatial trends could help forecast fish responses to long-term warming. However, temporal changes were about ten times faster than spatial trends (~4% versus ~40% body size change per 1 °C change through space and time, respectively). The rapid and variable responses of fish size to warming may herald unexpected impacts on ecosystem restructuring, with potentially greater consequences than if all species were shrinking.

通常在实验性变暖下，温升会缩小，但是这种模式是否扩展到野生种群尚不确定。我们分析了横跨澳大利亚大陆和几十年的一千万种视觉调查记录，其中包括最常见的沿海珊瑚礁鱼类（335种）。我们发现温度确实驱动着鱼体大小的时空变化，但并非始终如预期那样负面。在温暖的水域中（尤其是小物种）约有55％的物种变小，而更大的物种中约有45％。在任何给定位置，物种对空间温度响应的方向通常与其对温度随时间升高的响应一致，这表明空间趋势可以帮助预测鱼类对长期变暖的响应。然而，时间变化比空间趋势快十倍（在空间和时间上每1°C的变化，身体大小的变化分别为4％和40％）。鱼的大小对变暖的快速而多变的反应可能预示着对生态系统重组的意想不到的影响，其后果可能比所有物种都在萎缩的情况更大。