Long-term species-level observations indicate the critical role of soil moisture in regulating China's grassland productivity relative to phenological and climatic factors.

As a sensitive indicator of climate change and a key variable in ecosystem surface-atmosphere interaction, vegetation phenology, and the growing season length, as well as climatic factors (i.e., temperature, precipitation, and sunshine duration) are widely recognized as key factors influencing vegetation productivity. Recent studies have highlighted the importance of soil moisture in regulating grassland productivity. However, the relative importance of phenology, climatic factors, and soil moisture to plant species-level productivity across China's grasslands remains poorly understood. Here, we use nearly four decades (1981 to 2018) of in situ species-level observations from 17 stations distributed across grasslands in China to examine the key mechanisms that control grassland productivity. The results reveal that soil moisture is the strongest determinant of the interannual variability in grassland productivity. In contrast, the spring/autumn phenology, the length of vegetation growing season, and climate factors have relatively minor impacts. Generally, annual aboveground biomass increases by 3.9 to 25.3 g∙m2 (dry weight) with a 1 % increase in growing season mean soil moisture across the stations. Specifically, the sensitivity of productivity to moisture in wetter and colder environments (e.g., alpine meadows) is significantly higher than that in drier and warmer environments (e.g., temperate desert steppes). In contrast, the sensitivity to the precipitation of the latter is greater than the former. The effect of soil moisture is the most pronounced during summer. Dominant herb productivity is more sensitive to soil moisture than the others. Moreover, multivariate regression analyses show that the primary climatic factors and their attributions to variations in soil moisture differ among the stations, indicating the interaction between climate and soil moisture is very complex. Our study highlights the interspecific difference in the soil moisture dependence of grassland productivity and provides guidance to climate change impact assessments in grassland ecosystems.

Evolutionary trends and analysis of the driving factors of Ulva prolifera green tides: A study based on the random forest algorithm and multisource remote sensing images.

Understanding the prolonged spatiotemporal evolution and identifying the underlying causes of Ulva prolifera green tides play pivotal roles in managing such occurrences, restoring water ecology, and fostering sustainable development in marine ecosystems. Satellite remote sensing represents the primary choice for monitoring Ulva prolifera green tides due to its capability for extensive, long-term ocean monitoring. Based on multi-source remote sensing images, ecological and environmental datasets, and machine learning algorithms, therefore, this study focused on "remote sensing modelling - evolution history - change trends - mechanism analysis" to elucidate both the remote sensing monitoring models and the underlying driving factors governing the spatiotemporal evolution of Ulva prolifera green tides in the highly impacted South Yellow Sea of China. With the use of GOCI Ⅰ/Ⅱ images, an hybrid remote sensing extraction model merging the robustness of the random forest (RF) model and the optical algae cloud index (ACI) was established to map Ulva prolifera distribution patterns. The ACI-RF method exhibited exceptional performance, with an F1 score surpassing 0.95, outperforming alternative methods such as the support vector machine (SVM) and K-nearest neighbour (KNN) methods. On the basis, we analysed the evolutionary trends and the driving factors determining these distribution patterns using meteorological data, runoff data, and data on various water quality parameters (SST, ocean current speed, wind speed, precipitation, DO, PAR, Si, NO3-, PO43-and N/P). Over the period from 2011 to 2022, excluding 2021, there was a notable decline in the area of Ulva prolifera green tides, varying between 397 and 2689.9 km2, with an average annual reduction rate of 3%. The maximum annual biomass varied between 0.12 and 15.9 kt. Notably, more than 75% of the area of Ulva prolifera green tides exhibited northward drift, which was significantly influenced by northern currents and wind fields. The analysis of driving factors indicates that factors such as average sea surface temperature, eastward wind speed, northward wind speed, precipitation, PO43- and N/P/Si significantly influence the biological growth rate of Ulva prolifera. Furthermore, coastal land use change and surface runoff, particularly surface runoff in June, significantly impacted the growth rate of Ulva prolifera, with Pearson correlation coefficients of 0.74 and 0.67, respectively. Against the background of global warming and severe deterioration in the marine environment, Ulva prolifera blooms persist. Consequently, two distinct management strategies were proposed based on the distribution patterns and cause analysis results for addressing Ulva prolifera green tides: establishing a continuous protection framework for rivers, lakes, and nearshore areas to mitigate pollutant inputs and implementing precise environmental monitoring measures in urban expansion areas and farmlands to combat overgrowth-induced green tides. This methodology could be applied in other regions affected by marine ecological disasters, and the criteria for selecting influencing factors offer a valuable reference for designing tailored and proactive measures aimed at controlling Ulva prolifera green tides.

Remote sensing retrieval and driving analysis of phytoplankton density in the large storage freshwater lake: A study based on random forest and Landsat-8 OLI.

Remote sensing monitoring of seasonal changes in phytoplankton density and analyses of the driving factors of phytoplankton densities are necessary for assessing the health of aquatic ecosystems, controlling lake eutrophication, and formulating ecological restoration policies. Building upon the satellite-ground synchronization experiment that involves the in situ aquatic ecological monitoring conducted in Nansi Lake, which is the largest storage lake situated along the eastern route of the South-to-North Water Diversion Project, we developed a phytoplankton density retrieval model utilizing the random forest (RF) method and Landsat-8 OLI data. On this basis, we mapped the seasonal fluctuations and spatial disparities in the phytoplankton densities from 2013 to 2023. Subsequently, we conducted a detailed analysis of the driving factors and considered both the natural and anthropogenic aspects. The results indicate that (1) the RF model, when utilizing three band combinations, yielded favorable results with R2, RMSE and MAE values of 0.67, 1.31 × 106 cells/L and 1.18 × 106 cells/L, respectively. (2) The phytoplankton densities exhibited both seasonal and spatial variations, with higher concentrations in summer and autumn than in spring and winter. Significantly, the northwestern region of Zhaoyang Lake and the southeastern region of Weishan Lake had substantially greater phytoplankton densities than did the other areas. Furthermore, overarching upward trends were observed from 2013 to 2023, reflecting an annual rate of increase of 3.32%. (3) An analysis of the causal factors indicated that temperatures and gross agricultural production levels are the primary drivers influencing the seasonal variations and distributions of phytoplankton densities. In the future, we will delve into the potential of deep learning and utilize various satellite sensors to explore the intricacies of phytoplankton monitoring, as well as the complex mechanisms that influence aquatic ecological health.

A novel multivariate time series prediction of crucial water quality parameters with Long Short-Term Memory (LSTM) networks.

Intelligent prediction of water quality plays a pivotal role in water pollution control, water resource protection, emergency decision-making for sudden water pollution incidents, tracking and evaluation of water quality changes in river basins, and is crucial to ensuring water security. The primary methodology employed in this paper for water quality prediction is as follows: (1) utilizing the comprehensive pollution index method and Mann-Kendall (MK) trend analysis method, an assessment is made of the pollution status and change trend within the basin, while simultaneously extracting the principal water quality parameters based on their respective pollution share rates; (2) employing the spearman method, an analysis is conducted to identify the influential factors impacting each key parameter; (3) subsequently, a water quality parameter prediction model, based on Long Short-Term Memory (LSTM) analysis, is constructed using the aforementioned driving factor analysis outcomes. The developed LSTM model in this study showed good prediction performance. The average coefficient of determination (R2) of the prediction of crucial water quality parameters such as total nitrogen (TN) and dissolved oxygen (DO) reached 0.82 and 0.86 respectively. Additionally, the error analysis of WQI prediction results showed that >75% of the prediction errors were in the range of 0-0.15. The comparative analysis revealed that the LSTM model outperforms both the random forest (RF) model in time series prediction and demonstrates superior robustness and applicability compared to the AutoRegressive Moving Average with eXogenous inputs model (ARMAX). Hence, the model developed in this study offers valuable technical assistance for water quality prediction and early warning systems, particularly in economically disadvantaged regions with limited monitoring capabilities. This contribution facilitates resource optimization and promotes sustainable development.

A machine learning-based study on the impact of COVID-19 on three kinds of pollution in Beijing-Tianjin-Hebei region.

Large-scale restrictions on anthropogenic activities in China in 2020 due to the Corona Virus Disease 2019 (COVID-19) indirectly led to improvements in air quality. Previous studies have paid little attention to the changes in nitrogen dioxide (NO2), fine particulate matter (PM2.5) and ozone (O3) concentrations at different levels of anthropogenic activity limitation and their interactions. In this study, machine learning models were used to simulate the concentrations of three pollutants during periods of different levels of lockdown, and compare them with observations during the same period. The results show that the difference between the simulated and observed values of NO2 concentrations varies at different stages of the lockdown. Variation between simulated and observed O3 and PM2.5 concentrations were less distinct at different stages of lockdowns. During the most severe period of the lockdowns, NO2 concentrations decreased significantly with a maximum decrease of 65.28 %, and O3 concentrations increased with a maximum increase of 75.69 %. During the first two weeks of the lockdown, the titration reaction in the atmosphere was disrupted due to the rapid decrease in NO2 concentrations, leading to the redistribution of Ox (NO2 + O3) in the atmosphere and eventually to the production of O3 and secondary PM2.5. The effect of traffic restrictions on the reduction of NO2 concentrations is significant. However, it is also important to consider the increase in O3 due to the constant volatile organic compounds (VOCs) and the decrease in NOx (NO+NO2). Traffic restrictions had a limited effect on improving PM2.5 pollution, so other beneficial measures were needed to sustainably reduce particulate matter pollution. Research on COVID-19 could provide new insights into future clean air action.

Rare soil bacteria are more responsive in desertification restoration than abundant bacteria.

Soil microbes play key roles in ecosystem functions, especially in the recovery of ecosystems from disturbance, and exploring community assembly under changing environments has long been a central theme in microbial ecology. The response of abundant and rare bacteria in desertified land to restoration is still unclear. Here, we investigated the effects of vegetation restoration on the assemblage patterns of abundant and rare bacteria in soil across the four sandy lands (Hulunbeir, Horqin, Otindag, and Mu Us) in northern China. Our results revealed that abundant bacteria maintained a relatively stable state under restoration, whereas rare taxa were more responsive, indicating the higher resilience of the rare community to change. Our network analysis also showed that restoration promoted destabilizing properties in rare, but not in abundant, bacterial co-occurrence networks in soil. Environmental selection played a key role in abundant and rare community assembly under restoration. Of the two, the rare subcommunity was mainly affected by environmental filtering. The variations in the abundant and rare communities at the sampling sites under restoration were controlled mainly by plant species richness, and stronger effects were observed in the rare taxa. Overall, these results provide new insight into the mechanisms controlling bacterial community assembly in response to vegetation restoration.

The response of net primary productivity to climate change and its impact on hydrology in a water-limited agricultural basin.

Climate change has remarkably altered growing-season vegetation growth, but the impacts of vegetation variability on the regional hydrological cycle remain poorly understood. Exploring the relationships between climate change, vegetation dynamics, and hydrologic factors would contribute to the sustainable management of ecosystems. Here, we investigated the response of vegetation dynamics to climate change and its impact on hydrologic factors in a traditional agricultural basin with limited water resources in China, Nansi Lake Basin (NLB). To this end, CASA (Carnegie-Ames-Stanford Approach) model and the SWAT (Soil and Water Assessment Tool) model were applied to simulate the net primary productivity (NPP), evapotranspiration (ET), and soil water in the growing season (April-October) from 2000 to 2016. Results showed that the mean growing-season NPP (NPPGS) exhibited an ascending trend at a rate of 2.93 g C/m2/year during the 17-year period. The intra-annual variation of NPPGS displayed two peaks in May and July, respectively. The first peak in May was accompanied by relative deficits in soil water, which might inhibit vegetation productivity. Precipitation was the principal climatic factor controlling NPPGS dynamics in the water-limited NLB. The positive influence of temperature on NPPGS was relatively weak, and even future warming could negatively affect ecosystem productivity in the south-central regions of the NLB. Furthermore, a strongly positive relationship between NPPGS and ET was detected, suggesting that increasing NPP in the future might stimulate the rise in ET and then exacerbate drought at the watershed scale. This study provides an integrated model for a comprehensive understanding of the interaction between vegetation, climate, and hydrological cycle, and highlights the importance of water-saving agriculture for future food security.

Enhanced spatiotemporal heterogeneity and the climatic and biotic controls of autumn phenology in northern grasslands.

Knowledge of the response of grassland phenology towards climatic factors is essential to improve our understanding of ecological processes under global warming. To date, however, it remains unclear how climate change and associated changes in vegetation dynamics might affect autumn phenology of grasslands at the global scale. In this study, the trends in start of growing season (SOS) and end of growing season (EOS) dates were explored using remote sensing data (1981-2014). The responses of EOS to preseason temperature, rainfall, SOS, and net primary productivity (NPP) were then investigated for the mid-latitude (30°N ~ 55°N) grasslands of the Northern Hemisphere. The remotely-sensed SOS/EOS and PhenoCam-based SOS/EOS were first compared and a good correlation was observed. Trend analysis revealed that the time span of SOS/EOS (from the earliest SOS/EOS to the last SOS/EOS) and the growing season length (from SOS to EOS) have extended for the entire study region. Furthermore, a forward shift in all SOS pixels was observed in Central-West Asian grasslands, whereas no such significant trend was observed for North American grasslands and East Asian grasslands. The duration of EOS completion had shortened within North American grasslands but lengthened in Asian grasslands. Next, correlation analysis uncovered a stronger relationship between EOS and previous rainfall than between EOS and temperature, indicating the key role of water availability in controlling autumn phenology. The sensitivity of EOS to both temperature and rainfall was higher in drier and warmer locations. Moreover, a significant negative correlation between EOS and SOS was observed in part of the study region, but no significant relationship between NPP and EOS was observed. Overall, this study highlights the spatially intensified heterogeneity of spring and autumn phenology in northern grasslands and that climatic changes in precipitation might act as key drivers for modifying autumn phenology of grassland vegetation in the Northern Hemisphere.

Diverse responses of spring phenology to preseason drought and warming under different biomes in the North China Plain.

Global-change-type drought, a combination of drought and warmer temperatures, is projected to have severe effects on vegetation growth and ecosystem functions. Spring phenology is an important biological indicator to understand the response of vegetation growth to climate change. However, the differences in the response of spring phenology to global-change-type drought among various vegetation types remain unclear. Here, we extracted the start of growing season (SOS) from NDVI (Normalized Difference Vegetation Index) data using Spline-midpoint, HANTS-Maximum, and Timesat-SG methods in the North China Plain over the period 1982-2015. Then, we investigated the effects of preseason drought on SOS (based on the Standardized Precipitation Evapotranspiration Index, SPEI), and compared responses of SOS to the minimum temperature (Tmin), maximum temperature (Tmax), and mean temperature (Tmean) in different biomes. Results showed a trend of advanced SOS in 81.7% of pixels in the North China Plain, with an average rate of -0.5 days/yr. Negative correlations were found between preseason SPEI and SOS in 72.1% of the study region, and the SOS of grassland showed the least resistance to drought. Interannual variations of SOS were triggered by Tmin more than by Tmax in the North China Plain. Multiple regression analysis exhibited that a 1 °C increase in Tmin would advance SOS by 10.5, 7.6, 2.9, 2.1 days for wheat, other crops, forests, and grasslands, indicating warming displayed greater effects on advancing the SOS of wheat. Considering the coupled effects of preseason drought and warming on spring phenology, future warming would trigger earlier spring green-up, while drought might slow the trend. Besides, nonlinear responses of SOS to preseason SPEI and Tmin along the humidity gradient were discovered. This research provides a new reference for the biome-specific and nonlinear responses in phenology models to promote the understanding of phenology changes, contributing to ecosystem management under future global-change-type drought.

Contrasting wheat phenological responses to climate change in global scale.

Comprehensive analysis of how wheat phenology responds to environmental factors in global scale is helpful for tackling the possible adverse effects of ongoing climate change on wheat production. In this study, six phenological parameters of global wheat, i.e., the growing season start (SGS), peak (PGS), end (EGS), length (LGS), as well as the vegetative period length (LVP) and reproductive period length (LRP), were retrieved from remote sensing data (1981-2014) by threshold-, logistic-, and shape-based methods. And then, we analyzed the effects of temperature, precipitation, short-wave (SW) radiation, and frost on spatiotemporal patterns of wheat phenology. In addition, haze impacts on wheat phenology were investigated in China and India where haze weather appears frequently in winter-spring seasons. Results showed that the occurrence time of SGS/PGS/EGS is gradually advanced from the pole to the equator and annual mean air temperature can explain >70% of their spatial variations. A dominant advanced SGS/PGS/EGS and a shortened LGS/LVP/LRP were detected in the study region due to the significant increase in temperature and SW radiation, as well as the decrease in frost days. Interannual fluctuations of SGS/PGS/EGS are primarily controlled by air temperature, while precipitation and frost only exerted some obvious impacts in some locations. Higher preseason temperature would induce an earlier wheat phenology and a shorter growing season, while adequate precipitation and frequent frost in preseason could delay the occurrence timing of wheat phenology and lead to a longer growing season. Besides, the decreased temperature resulted from severe haze weather may have partly counteracted the global-warming-induced advancing trend of wheat phenology in China, but further advanced the occurrence timing of wheat phenology through prompting vernalization in India. Overall, though wheat growth is largely constrained by human management, we still highlight the strong impacts of global climate change on wheat phenology.

Temporal coherence of phenological and climatic rhythmicity in Beijing.

Using woody plant phenological data in the Beijing Botanical Garden from 1979 to 2013, we revealed three levels of phenology rhythms and examined their coherence with temperature rhythms. First, the sequential and correlative rhythm shows that occurrence dates of various phenological events obey a certain time sequence within a year and synchronously advance or postpone among years. The positive correlation between spring phenophase dates is much stronger than that between autumn phenophase dates and attenuates as the time interval between two spring phenophases increases. This phenological rhythm can be explained by positive correlation between above 0 °C mean temperatures corresponding to different phenophase dates. Second, the circannual rhythm indicates that recurrence interval of a phenophase in the same species in two adjacent years is about 365 days, which can be explained by the 365-day recurrence interval in the first and last dates of threshold temperatures. Moreover, an earlier phenophase date in the current year may lead to a later phenophase date in the next year through extending recurrence interval. Thus, the plant phenology sequential and correlative rhythm and circannual rhythm are interacted, which mirrors the interaction between seasonal variation and annual periodicity of temperature. Finally, the multi-year rhythm implies that phenophase dates display quasi-periodicity more than 1 year. The same 12-year periodicity in phenophase and threshold temperature dates confirmed temperature controls of the phenology multi-year rhythm. Our findings provide new perspectives for examining phenological response to climate change and developing comprehensive phenology models considering temporal coherence of phenological and climatic rhythmicity.

Assessing plant senescence reflectance index-retrieved vegetation phenology and its spatiotemporal response to climate change in the Inner Mongolian Grassland.

Plant phenology is a key link for controlling interactions between climate change and biogeochemical cycles. Satellite-derived normalized difference vegetation index (NDVI) has been extensively used to detect plant phenology at regional scales. Here, we introduced a new vegetation index, plant senescence reflectance index (PSRI), and determined PSRI-derived start (SOS) and end (EOS) dates of the growing season using Moderate Resolution Imaging Spectroradiometer data from 2000 to 2011 in the Inner Mongolian Grassland. Then, we validated the reliability of PSRI-derived SOS and EOS dates using NDVI-derived SOS and EOS dates. Moreover, we conducted temporal and spatial correlation analyses between PSRI-derived SOS/EOS date and climatic factors and revealed spatiotemporal patterns of PSRI-derived SOS and EOS dates across the entire research region at pixel scales. Results show that PSRI has similar performance with NDVI in extracting SOS and EOS dates in the Inner Mongolian Grassland. Precipitation regime is the key climate driver of interannual variation of grassland phenology, while temperature and precipitation regimes are the crucial controlling factors of spatial differentiation of grassland phenology. Thus, PSRI-derived vegetation phenology can effectively reflect land surface vegetation dynamics and its response to climate change. Moreover, a significant linear trend of PSRI-derived SOS and EOS dates was detected only at small portions of pixels, which is consistent with that of greenup and brownoff dates of herbaceous plant species in the Inner Mongolian Grassland. Overall, PSRI is a useful and robust metric in addition to NDVI for monitoring land surface grassland phenology.