In this study, we set up a wood decomposition experiment to i) quantify the percent of mass remaining, decay constant and performance strength of aspen stakes (Populus tremuloides) in dry and moist boreal (Alaska and Minnesota, USA), temperate (Washington and Idaho, USA), and tropical (Puerto Rico) forest types, and ii) determine the effects of fragmentation on wood decomposition rates as related to fragment size, forest age (and/or structure) and climate at the macro- and meso-scales. Fragment sizes represented the landscape variability within a climatic region. Overall, the mean small fragments area ranged from 10-14 ha, medium-sized fragments 33 to 60 ha, and large fragments 100-240 ha. We found that: i) aspen stakes decayed fastest in the tropical sites, and the slowest in the temperate forest fragments, ii) the percent of mass remaining was significantly greater in dry than in moist forests in boreal and temperate fragments, while the opposite was true for the tropical forest fragments, iii) no effect of fragment size on the percent of mass remaining of aspen stakes in the boreal sites, temperate dry, and tropical moist forests, and iv) no significant differences of aspen wood decay between forest edges and interior forest in boreal, temperate and tropical fragments. We conclude that: i) moisture condition is an important control over wood decomposition over broad climate gradients; and that such relationship can be non linear, and ii) the presence of a particular group of organism (termites) can significantly alter the decay rates of wood more than what might be predicted based on climatic factors alone. Biotic controls on wood decay might be more important predictors of wood decay in tropical regions, while abiotic constraints seems to be important determinants of decay in cold forested fragments.