Environmental variation in time and space affects biological processes such as extinction risk and speed of adaptation to environmental change. The spatial structure of environmental variation may vary among ecosystems, for instance due to differences in the flow of nutrients, genes and individuals. However, inferences about ecosystem spatial scale should also include spatial autocorrelation in environmental stochasticity, such as fluctuations in weather or climate. We used spatially structured time series (19-36 yr) on temperature from 4 different ecosystems (terrestrial, limnic, coastal sea and open ocean) to assess the spatiotemporal patterns of environmental variation over large geographical scales (up to 1900 km) during summer and winter. The distance of positive spatial autocorrelation in mean temperature was greatest for the terrestrial system (range: 592-622 km), and shorter for the open ocean (range: 472-414 km), coastal sea (range: 155-814 km) and the limnic systems (range: 51-324 km), suggesting a stronger spatial structure in environmental variation in the terrestrial system. The terrestrial system had high spatial synchrony in temperature (mean correlation: winter = 0.82, summer = 0.66) with a great spatial scaling (>650 km). Consequently, populations of terrestrial species experience similar environmental fluctuations even at distances up to 1000 km, compared to species in the aquatic systems (<500 km). There were clear seasonal differences in environmental synchrony in the terrestrial and limnic systems, but less so in the other systems. Our results suggest that biological processes affected by environmental stochasticity occur at the largest spatial scale in terrestrial systems, but their magnitude depends on whether the process is affected by winter or summer conditions.