Abstract.We compare assimilation and respiration rates, and water use strategies in four divergent ecosystems located in cold-continental central Siberia and in semi-arid southern Africa. These seemingly unrelated systems have in common a harsh and highly seasonal environment with a very sharp transition between the dormant and the active season, with vegetation facing dry air and soil conditions for at least part of the year. Moreover, the northern high latitudes and the semi-arid tropics will likely experience changes in key environmental parameters (e.g., air temperature and precipitation) in the future; indeed, in some regions marked climate trends have already been observed over the last decade or so.The magnitude of instantaneous or daily assimilation and respiration rates, derived from one to two years of eddy covariance measurements in each of the four ecosystems, was not related to the growth environment. For instance, respiration rates were clearly highest in the two deciduous systems included in the analysis (a Mopane woodland in northern Botswana and a Downy birch forest in Siberia; >300 mmol m −2 d −1 ), while assimilation rates in the Mopane woodland were relatively similar to a Siberian Scots pine canopy for a large part of the active season (ca. 420 mmol m −2 d −1 ). Acknowledging the limited number of ecosystems compared here, these data nevertheless demonstrate that factors like vegetation type, canopy phenology or ecosystem age can override larger-scale climate differences in terms of their effects on carbon assimilation and respiration rates.By far the highest rates of assimilation were observed in Downy birch, an early successional species. These were achieved at a rather conservative water use, as indicated by relatively low levels of λ, the marginal water cost of plant Correspondence to: A. Arneth (almut.arneth@nateko.lu.se) carbon gain. Surprisingly, the Mopane woodland growing in the semi-arid environment had significantly higher values of λ. However, its water use strategy included a very plastic response to intermittently dry periods, and values of λ were much more conservative overall during a rainy season with low precipitation and high air saturation deficits. Our comparison demonstrates that forest ecosystems can respond very dynamically in terms of water use strategy, both on interannual and much shorter time scales. But it remains to be evaluated whether and in which ecosystems this plasticity is mainly due to a short-term stomatal response, or alternatively goes hand in hand with changes in canopy photosynthetic capacity.