Temporal Dynamics In The CO₂ Flux And Canopy Characteristics Of A Temperate Deciduous Forest At The Maoershan Site In Northeast China

The temperate forest in the East Asia is one of the three major temperate forests in the world,but the long-term interannual variations(IAV)in the ecosystem exchange of CO2(NEE)and their driving mechanism are rarely explored,which limited our understanding of the carbon sink of the temperate secondary forest(deciduous broadleaved forest)and the controlling mechanisms of the IAV of CO2 fluxes.Based on the continuous 11-year(2008-2018)measurements of the CO2 fluxes of a temperate secondary forest at the Maoershan site in northeastern China,supplemented by the canopy(e.g.leaf area index and phenology),woody growth,and environmental fators based on the ground measuremets and remote sensing products,we explored the responses of the NEE to the biotic and environmental fators and their underlying drivers.The main results were as follows:(1)The IAV in the leaf area index(LAI)and canopy phenology were monitored by the litterfall collection and remote sensing vegetation indices.To balance the measuring time/labor consumption and the accuracy of LAI estimation,we recommend focusing on the temporal variations and ignoring spatial variations in specific leaf area(SLA)for monitoring the temporal dynamics,and correcting for the leaf area shrinkage relative to the green leaf.The MODIS LAI underestimated the LAI by 9%,but well tracked the IAV in LAI.The broadband Normalized Difference Vegetation Index(NDVIB)best tracked the IAV in the end of growing season(EOS)among the six metrics of the NDVIB investigated,while the MODIS Enhanced Vegetation Index(EVIM)did so among the six metrics of the MODIS vegetation index,with the determination coefficient(R2)of 0.66 and 0.44 with the EOS estimated by the leaf-litterfall.Radiometer and MODIS could effectively track the autumn phenology in the forest,with the leaf-litterfall collection being a complementary approach.(2)The NEE was continuously measured by the eddy covariance technique across 2008-2018 for exploring the IAV of NEE and its environmental drivers.The mean values of NEE,gross primary production(GPP),and ecosystem respiration(Re)of the stand over the 11 years were-157 ± 64,1356± 148,and 1200 ± 138 g C m-2 yr-1,respectively.The environmental factors had a weak impact on the IAV of NEE largely because of the offset between the positive responses of the annual GPP and Re to the spring and autumn soil water content,respectively.Spring precipitation and autumn photosynthetically active radiation were the main environmental drivers of the spring and autumn NEE,respectively.Temperature did not have significant influence on the IAV of NEE,while precipitation and radiation did by changing the end of net CO2 uptake in autumn.(3)The LAI estimated by the litterfall collection,phenology and physiology were used to investigated the biotic driving mechanism of the IAV of NEE.The IAV of NEE was jointly controlled by the length of net CO2 uptake period and the summer peak of the net CO2 uptake,whereas those of GPP and Re were dominantly controlled by the summer peaks(GPPmax and Remax).The increasing trend of annual GPP was dominantly controlled by the GPPmax due to the leaf-level photosynthetic capacity rather than the maximum LAI.Phenology was the dominant biotic driver of the spring and autumn CO2 fluxes.Taking the influences of the ecosystem-and leaf-level physiology,canopy structure,and phenology,together with environmental interactions,on CO2 fluxes into account will improve the understanding and prediction of the temporal dynamics in the forest carbon budget.(4)The phenology and summer peaks of GPP and those of four vegetation indices from the near-surface remote sensing and MODIS were compared during 2008-2018,and the performances of four growing-season integral vegetation indices for characterizing the annual GPP were evaluated.The start of growing season defined by 25%-35%of EVIB amplitude well tracked the IAV of spring photosynthetic phenology(R2:0.56-0.60,bias<4 d);the EOS defined by 45%or 50%amplitudes of NDVIB and NDVIM were effective proxies of the autumn photosynthetic phenology(R2:0.58-0.67,bias<3 d).However,the summer peaks of the four vegetation indices all failed to character the IAV of GPPmax.The growing season integral NDVIB and EVIM were robust surrogates for the IAV of annual GPP(R2:0.44-0.63),regardless of the phenology definition.The definition threshold of vegetation index phenology is important for characterizing the photosynthetic phenology,but not for explaining the IAV in GPP.(5)The tree(diameter)growth was continuously measured in the nine permenant plots around the eddy-flux tower.Relationships between tree growth and carbon input and environmental factors were explored,and the relationships between woody growth and different vegetation indices were compared.The woody biomass increment was positively correlated with the GPP in the current and next year(R2=0.48 and 0.63,respectively),suggesting a strong correlation between the canopy photosynthesis and the vegetation growth.Spring soil water content was an important environmental factor controlling the tree growth.NDVIM showed the closest correlation with the tree growth among the four vegetation indices tested(R=0.682),and was a reasonable proxy of the status of the vegetation growth status.Overall,we provided a standard protocol of LAI measurement in temperate deciduous forests by the litterfall collection.LAI had little influence on the NEE,while phenology controlled the spring and autumn CO2 fluxes,highlighting the importance of vegetation physiology in the IAV in NEE.The integral vegetation indices generally reflected the IAV of GPP,but there was a hysteric relationship between vegetation growth and GPP.These findings improved our understanding of the temporal dynamics in the canopy and CO2 fluxes in temperate deciduous forests.