Non-linear fitness gradients with maxima between extremes are expected for any environmental variable to which free-living populations are exposed. For exceedingly toxic agents, including ionizing radiation, such deviations from linearity are close to zero exposure and are conventionally called hormesis. Accordingly, hormesis is an extreme version of the non-linear fitness gradients for general environmental stresses such as temperature fluctuations, for which maximum fitness occurs at the moderate temperature fluctuations to which free-living populations are most commonly exposed. Some metabolic reserves should occur under moderate temperature stresses because of the need for pre-adaptation enabling survival during exposure to occasional periods of more extreme stress, especially at species borders where selection for stress resistance is likely to be most intense. Because heat shock proteins are induced by all stresses, adaptation to extreme temperatures should translate into adaptation to other stresses. Consequently, metabolic reserves from adaptation to extreme temperatures in the past should translate into protection from correlated abiotic stresses, especially in human populations where modern cultural processes are now ameliorating exposure to extreme stresses. Ambient and man-made radiations of non-catastrophic dimensions should therefore lead to stress-derived radiation hormesis. Other stresses can, in principle, be incorporated into this model.
Accordingly, evolutionary and ecological considerations suggest two components of hormesis in relation to ionizing radiation: background radiation hormesis based upon the background exposure to which all organisms on earth are subjected; and stress-derived radiation hormesis. Exposure under stress-derived radiation hormesis is considerably larger than under background radiation hormesis, so significant deleterious effects from non-catastrophic radiation normally may be impossible to detect. Suggestions are provided for testing such postulated deviations from the commonly assumed linear no-threshold (LNT) hypothesis for the biological consequences of exposure to radiation.