To be presented, with the permission of the Faculty of Science and Forestry of the University of Eastern Finland, for public examination in auditorium BOR 100 in borealis building on Joensuu campus of the University of Eastern Finland, Yliopistokatu 7, Joensuu on 23 May 2017 at 12 o' clock noon.
ABSTRACTThe main aim of this study was to investigate the dynamics and biophysical controls of carbon, water and energy exchange over a semiarid shrub ecosystem in the Mu Us desert, northern China, using continuous eddy-covariance (EC) measurements. Specific objectives were as follows: (1) To examine intra-annual variations in net ecosystem CO 2 exchange (NEE) and its biophysical controls (Paper I); (2) To quantify the diurnal and seasonal variations in surface energy-balance components, and to examine the partitioning of net radiation (R n ) among different energy components at diurnal and seasonal timescales (Paper II); and (3) To examine how ecosystem production and water use efficiency (WUE) vary inter-annually with contrasting precipitation (PPT) and soil moisture patterns (Paper III).The results showed that, soil water content (i.e. at 30 cm depth, SWC_30), or water deficit, imposed a major control on the seasonal dynamics of carbon assimilation and energy partitioning. Water deficit (i.e. SWC_30 < 0.10 m 3 m -3 ) was a major constraint over daytime NEE, and also interacted with other stresses, e.g. heat stress and photoinhibition (Paper I). Low soil moisture reduced the temperature sensitivity (Q 10 ) of total ecosystem respiration (TER). Rain events triggered immediate pulses of carbon release from the ecosystem, followed by peaks of CO 2 uptake 1-2 days later. Leaf area index (LAI) accounted for 45 and 65% of the seasonal variation in NEE and gross ecosystem production (GEP), respectively. On the other hand, sensible heat flux (H) exceeded latent heat flux (λE) during most time of the year (Paper II). The evaporative fraction (EF, i.e. λE/R n ), Priestley-Taylor coefficient (α), surface conductance (g s ) and decoupling coefficient (Ω) all correlated positively with SWC_30 and LAI. The direct enhancement of λE by high vapor pressure deficit (VPD) was buffered by a concurrent suppression of g s , which controlled EF and α by mediating the effects of LAI, SWC_30 and VPD.At the annual scale, net ecosystem production (NEP, here defined as −NEE) indicated a rapid shift from an annual sink of carbon in 2012 (NEP = 77 ± 10 g C m -2 yr -1 ) to a source of carbon in 2014 (NEP = -22 ± 5 g C m -2 yr -1 ), with the year 2013 being close to carbon neutral (NEP = -4 ± 10 g C m -2 yr -1 ) (Paper III). GEP, TER and evapotranspiration (ET) also declined over the three years. Suppressed annual carbon and water fluxes were observed in years with low spring soil moisture. GEP declined more than TER and ET, leading to reduced carbon sequestration and WUE (i.e. GEP/ET). Neither annual nor growing-season PPT amount could explain the year-to-year variation in carbon fluxes. ET was a better proxy for water available to ecosystem carbon...