Seasonality and inter-annual forest—atmosphere carbon and water exchanges in a Portuguese Eucalyptus plantation (Mediterranean climate)

Eddy covariance flux data from the Portuguese CarboEurope forest site ‘Herdade da Espirra' obtained from 2003 to 2005 is analysed. A 32 meter tower contains at its top the eddy covariance unit, comprised by an ultrasonic Gill anemometer R2 and an open path analyzer IRGA LI-7500, as well as an automatic weather station. The site (38º38'N, 8º36'W) is an Eucalyptus Globulus 300 hectare plantation, 11 years aged forest with 20 meter mean canopy height and trees spacing 3x3 meter and an estimated mean leaf area index (LAI) of 2.41. The cover is located in flat terrain, and it extends for distances from 700m to 1800m in the several directions, from the measurement tower. Footprint analyses shows that the average measured flux contribution from the target forest is about 87 percent, showing that the site allows a good perspective on the impact that this type of plantations have in terms of carbon assimilation in Mediterranean climates. These climates are characterized by discrete periods of intense rain events occurring mainly from October to April. Periodically these events are very short or less intense, resulting in dry years, with the frequency of this occurrence increasing in the last decades of the 20th century. The linkage between water and carbon balances in Mediterranean ecosystems is therefore a major issue. The partitioning of the carbon flux, expressed as Net Ecosystem Exchange (NEE), in Gross Primary Production (GPP) and Ecosystem Respiration (Reco) allows a better understating of the phenomena influencing carbon exchanges between forest and the atmosphere. This year-long type assimilation forest acts as a strong carbon sink reaching an uptake of 938 gC.m^-2.yr^-1 in 2003, year with 11% less precipitation than the 30 years mean value for the location (709mm). A depression of carbon uptake occurred during summer, due to increased stomatal resistance associated with atmospheric water vapour deficit and a peak in soil respiration. In 2004 the precipitation was 48% under the mean value, and although the annual carbon balance pattern was quite similar when compared with the previous year, the carbon uptake was reduced in 10%, due to a reduced GPP and an increased Reco. The second consecutive dry year (2005), suffered 67% rain and 47% NEE reduction in the period between January and September relatively to the same period in 2003. The impact in the NEE was caused by a high stomatal control of the evaporatory demand, affecting negatively the carbon assimilation. The McNaughton and Jarvis decoupling factor (Ω) evaluated along the three years shows a pattern that confirms the prevalence of the imposed evapotranspiration, with very low values (<0.2) during 2003 and 2004 summertime and the whole 2005 (until Sept.), linked with the NEE evolution. Water availability is clearly the key factor determining the forest-atmosphere carbon exchanges variability, either in daily, monthly and yearly time scales. GPP positively correlates with evapotranspiration (r² = 0.55), considering the monthly values. The correlation between NEE and evapotranspiration at the same time scale is less evident due to deviations produced by respiration. A simple linear regression between annual net carbon uptake and precipitation, although with only three years (Jan. to Sept.), shows a high positive correlation (r²= 0.95). The Radiation use Efficiency (RUE) defined as the ratio between GPP and Photosynthetic Active Radiation (PAR), and Water use Efficiency (WUE), consisting in the ratio between GPP and evapotranspiration reach their minimum values at the end of the dry season (Sept.). After the first rains the recovery of both parameter is out of fase, with RUE increasing immediately and WUE presenting a slower response for the first two months. Both parameters are highly correlated (r² = 0.71) from January until September.