Temperature variation and distribution of living cells within tree stems: implications for stem respiration modeling and scale-up

Few studies have examined variation in respiration rates within trees, and even fewer studies have focused on variation caused by within-stem temperature differences. In this study, stem temperatures at 40 positions in the stem of one 30-year-old Norway spruce (Picea abies (L.) Karst.) were measured during 40 days between July 1994 and June 1995. The temperature data were used to simulate variations in respiration rate within the stem. The simulations assumed that the temperature-respiration relationship was constant (Q10 = 2) for all days and all stem positions. Total respiration for the whole stem was calculated by interpolating the temperature between the thermocouples and integrating the respiration rates in three dimensions. Total respiration rate of the stem was then compared to respiration rate scaled up from horizontal planes at the thermocouple heights (40, 140, 240 and 340 cm) on a surface area and on a sapwood volume basis. Simulations were made for three distributions of living cells in the stems: one with a constant 5% fraction of living cells, disregarding depth into the stem; one with a living cell fraction decreasing linearly with depth into the stem; and one with an exponentially decreasing fraction of living cells. Mean temperature variation within the stem was 3.7 degrees C, and was more than 10 degrees C for 8% of the time. The maximum measured temperature difference was 21.5 degrees C. The corresponding mean variation in respiration was 35% and was more than 50% for 24% of the time. Scaling up respiration rates from different heights between 40 and 240 cm to the whole stem produced an error of 2 to 58% for the whole year. For a single sunny day, the error was between 2 and 72%. Thus, within-stem variations in temperature may significantly affect the accuracy of scaling respiration data obtained from small samples to whole trees. A careful choice of chamber position and basis for scaling is necessary to minimize errors from variation in temperature.