Search

Upscaling flux tower measurements based on machine learning (ML) algorithms is an essential approach for large‐scale net ecosystem CO2 exchange (NEE) estimation, but existing ML upscaling methods face some challenges, particularly in capturing NEE interannual variations (IAVs) that may relate to lagged effects. With the capacity to characterize temporal memory effects, the Long Short‐Term Memory (LSTM) networks are expected to help solve this problem. Here we explored the potential of LSTM for predicting NEE across various ecosystems using flux tower data over 82 sites in North America. The LSTM model with differentiated plant function types (PFTs) demonstrates the capability to explain 79.19% (R2 = 0.79) of the monthly variations in NEE within the testing set, with RMSE and Mean Absolute Error values of 0.89 and 0.57 g C m−2 d−1 respectively (r = 0.89, p < 0.001). Moreover, the LSTM model performed robustly in predicting cross‐site variability, with 67.19% of the sites that can be predicted by both LSTM models with and without distinguished PFTs showing improved predictive ability. Most importantly, the IAV of predicted NEE highly correlated with that in flux observations (r = 0.81, p < 0.001), clearly outperforming that by the random forest model (r = −0.21, p = 0.011). Among all nine PFTs, solar‐induced chlorophyll fluorescence, downward shortwave radiation, and leaf area index are the most important variables for explaining NEE variations, collectively accounting for approximately 54.01% in total. This study highlights the great potential of LSTM for improving carbon flux upscaling with multi‐source remote sensing data. Plain Language Summary: Net ecosystem exchange (NEE) of CO2 is a crucial process that regulates carbon exchange between terrestrial ecosystems and the atmosphere. Currently, the growing availability of flux tower measurements and the rapid development of artificial intelligence technology, have made the machine learning‐based data‐driven flux upscaling approach a popular way for estimating carbon fluxes over large scales. Several upscaling NEE data sets have been derived with different machine learning methods; however, the lack of representing memory effects of climate and environmental factors in the modeling remains an important source of uncertainty for NEE estimates. To address this issue, we constructed site‐level Long Short‐Term Memory (LSTM) training models by plant function types in North America to improve the simulation of monthly scale NEE and its interannual variations. The established LSTM model improves the prediction of the temporal variability of NEE, showing great potential for improving carbon flux upscaling. Key Points: The Long Short‐Term Memory (LSTM) model with differentiated plant function types (PFTs) demonstrates the capability to explain 79.19% of the monthly variations in net ecosystem exchange (NEE)The LSTM model exhibited clear advantages over the RF model in capturing the interannual variations of NEEThe relative importance of feature variables for predicting monthly NEE dynamics across different PFTs in North America was quantified [ABSTRACT FROM AUTHOR]

Fagus L. is a key component in temperate deciduous broadleaf forests of the Northern Hemisphere. However, its biogeographic history has not been examined under the framework of a fully resolved and reasonably time‐calibrated phylogeny. In this study, we sequenced 28 nuclear single/low‐copy loci (18 555 bp in total) of 11 Fagus species/segregates and seven outgroups. Phylogenetic trees were reconstructed using both concatenation‐based (maximum parsimony, maximum likelihood, and Bayesian inference) and coalescent‐based methods (StarBEAST2, ASTRAL). The monophyly of two subgenera (Fagus and Engleriana) and most sections was well supported, except for sect. Lucida, which was paraphyletic with respect to sect. Longipetiolata. We also found a major phylogenetic conflict among North American, East Asian, and West Eurasian lineages of subgen. Fagus. Three segregates that have isolated distribution (F. mexicana, F. multinervis, and F. orientalis) were independent evolutionary units. Biogeographic analysis with fossils suggested that Fagus could have originated in the North Pacific region in late early Eocene. Major diversifications coincided with a climate aberration at the Eocene/Oligocene boundary and the global cooling since mid‐Miocene. The late Miocene accelerated global cooling and the Pleistocene glaciations would have driven beeches into East Asia, North America, and West Eurasia. Meanwhile, range reduction and extinction in high latitudes, central Asia, and western North America converged to form the beech modern distribution pattern. This study provides a first attempt to disentangle the biogeographic history of beeches in the context of a nearly resolved and time‐calibrated phylogeny, which could shed new insights into the formation of the temperate biome in the Northern Hemisphere. [ABSTRACT FROM AUTHOR]

Copyright of Functional Ecology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Increased resource supply commonly facilitates invasion by exotic plants, raising concerns over atmospheric nitrogen (N) deposition; fast-growing annual invaders may have exceptional abilities to outperform native perennials in response to N pulses. However, it remains unclear whether this advantage is due to growth differences or to shifts in competitive outcomes, and whether annual invaders are favored by N deposition in their introduced range over native range. We conducted an experiment to compare the growth and competitive ability of Bromus tectorum and its native perennial grasses either at three different N regimes or between China and North America. The soil used in this experiment was from mountain grasslands as a neutral growth medium. The total biomass of three natives from China and North America did not increase along the N deposition gradient. Nitrogen addition enhanced the growth of North American B. tectorum instead of Chinese B. tectorum. Nitrogen addition increased the competitive ability of B. tectorum, but had no effect on that of natives. North American B. tectorum was bigger and had greater competitive ability and root weight ratio than Chinese B. tectorum. In contrast, North American natives were less competitive than Chinese natives. There was a significantly positive correlation between the growth of B. tectorum grown alone and its competitive ability. These findings suggest that N deposition may enhance the B. tectorum invasion through disproportionally increasing the growth and maintaining inherent competitive advantages of North American B. tectorum, further increasing threats to introduced ranges. There were differences in the growth and competitive ability of B. tectorum and natives between China and North America, which explains why B. tectorum is a minor component at home and becomes a successful invader abroad. [ABSTRACT FROM AUTHOR]

• Peak growing season GPP (GPP PGS) and climate extremes impacts on GPP were studied. • Satellite models reasonably reproduced GPP PGS spatial pattern over North America. • Process models generally failed to reveal the typical spatial pattern of GPP PGS. • Satellite models performed better in capturing climate extremes impacts on GPP. • Satellite data are recommended to be used for improving GPP simulations. Representations of the seasonal peak uptake of CO 2 and climate extremes effects have important implications for accurately estimating annual magnitude and inter-annual variations of terrestrial carbon fluxes, however the consistency of such representations among different satellite models and process-based (PB) models remain poorly known. Here we investigated these issues over North America based on a large ensemble of state-of-the-art gross primary production (GPP) models, including two solar-induced chlorophyll fluorescence (SIF)-based models (WECANN and GOPT), three remote sensing driven light-use efficiency (LUE) models, and 10 PB models. We found that the two SIF-based GPP estimates were bilaterally consistent in spatial patterns of peak growing season GPP (GPP PGS ; with the largest uptake at the Corn-Belt area in the United States) and climate extremes-driven responses. The simulations from three LUE models showed relatively consistent spatial patterns of GPP PGS and climate extremes-driven responses, which agreed well with SIF-based estimates and satellite based metrics. Obviously differed from SIF and LUE based estimates, the simulations from PB models exhibited noticeable divergences and mostly failed to reasonably replicate the spatial pattern of GPP PGS. In addition, satellite models and PB models were comparably able to capture the effects of climate extremes on GPP, but showing obvious divergences in the magnitude of impacts among different models, and the former outperformed the latter in locating GPP changes caused by climate extremes. We discussed the possible origins of such discrepancies in state-of-the-art models with focus on PB models. Improving the parameterizations of critical variables (e.g. leaf area index) and better characterizing environmental stresses could lead to more robust estimates of large-scale terrestrial GPP with PB models, thus serving for accurately assessing global carbon budget and better understanding the impacts of climate change on the terrestrial carbon cycle. Our study offers a baseline for improving large-scale estimation of terrestrial GPP. [ABSTRACT FROM AUTHOR]

• Five LUE models were evaluated against data-driven GPP products and ancillary data. • Data-driven models are able to reasonably well represent GPP dynamics at site levels. • The simulated GPP by LUE models reasonably responded to droughts at regional scales. • Considerable differences existed in the amplitude and phase of GPP seasonal anomalies. • LUE models consistently exhibited a likely misrepresentation of drought legacy effects. Remote sensing driven light-use efficiency (LUE) models are an important tool for estimating vegetation gross primary productivity (GPP) and assessing climate change impacts on the terrestrial carbon cycle, but their capacity in representing drought impact on GPP over large-scales is not fully explored. Here, we evaluated a suite of LUE models, i.e., VPM, TL-LUE, CASA, EC-LUE and MuSyQ, against two data-driven flux products (GOSIF-GPP and FluxSat), with the aids of environmental variables and eddy covariance flux measurements over North America. The two data-driven GPP products were firstly validated against with eddy covariance flux measurements, suggesting that they are able to reasonably indicate drought impacts at site scales. They were then used as a reference to evaluate these LUE models for indicating the drought impacts in 2011 and 2012 over North America at regional scales. In general, the simulated GPP by these models reasonably responded to droughts, but showed considerable differences in the seasonal anomalies for both amplitude and phase. Especially, these models consistently showed a clearly earlier trough (about 1 ∼ 2 months) of flux reduction than those in data-driven models over both entire drought-impacted regions and individual ecosystems (mostly pronounced for grasslands), indicating a likely misrepresentation of drought legacy effects, which needs to be considered in developing new-generation LUE models. Among the grass-dominated drought areas, TL-LUE and VPM performed best in indicating the impacts of the 2011 and 2012 droughts on regional GPP, respectively, while EC-LUE showed a significant underestimation of the impacts in both events, which may be attributed to the different representations of water stress factors in models. This study could provide some useful hints toward improving LUE models in characterizing drought impacts on regional GPP. [ABSTRACT FROM AUTHOR]

Climate warming may drive invasions by exotic plants, thereby raising concerns over the risks of invasive plants. However, little is known about how climate warming influences the growth and competitive ability of exotic plants from their home and introduced ranges. We conducted a common garden experiment with an invasive plant Centaurea maculosa and a native plant Poa pratensis, in which a mixture of sand and vermiculite was used as a neutral medium, and contrasted the total biomass, competitive effects, and competitive responses of C. maculosa populations from Europe (home range) and North America (introduced range) under two different temperatures. The warming-induced inhibitory effects on the growth of C. maculosa alone were stronger in Europe than in North America. The competitive ability of C. maculosa plants from North America was greater than that of plants from Europe under the ambient condition whereas this competitive ability followed the opposite direction under the warming condition, suggesting that warming may enable European C. maculosa to be more invasive. Across two continents, warming treatment increased the competitive advantage instead of the growth advantage of C. maculosa, suggesting that climate warming may facilitate C. maculosa invasions through altering competitive outcomes between C. maculosa and its neighbors. Additionally, the growth response of C. maculosa to warming could predict its ability to avoid being suppressed by its neighbors.