Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium t...Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium to long distances over different cameras.However,beef cattle can tend to frequently move and change their feeding position during feeding.Furthermore,the great variations in their head direction and complex environments(light,occlusion,and background)can also lead to some difficulties in the recognition,particularly for the bio-similarities among individual cattle.Among them,AlignedReID++model is characterized by both global and local information for image matching.In particular,the dynamically matching local information(DMLI)algorithm has been introduced into the local branch to automatically align the horizontal local information.In this research,the AlignedReID++model was utilized and improved to achieve the better performance in cattle re-identification(ReID).Initially,triplet attention(TA)modules were integrated into the BottleNecks of ResNet50 Backbone.The feature extraction was then enhanced through cross-dimensional interactions with the minimal computational overhead.Since the TA modules in AlignedReID++baseline model increased the model size and floating point operations(FLOPs)by 0.005 M and 0.05 G,the rank-1 accuracy and mean average precision(mAP)were improved by 1.0 percentage points and 2.94 percentage points,respectively.Specifically,the rank-1 accuracies were outperformed by 0.86 percentage points and 0.12 percentage points,respectively,compared with the convolution block attention module(CBAM)and efficient channel attention(ECA)modules,although 0.94 percentage points were lower than that of squeeze-and-excitation(SE)modules.The mAP metric values were exceeded by 0.22,0.86 and 0.12 percentage points,respectively,compared with the SE,CBAM,and ECA modules.Additionally,the Cross-Entropy Loss function was replaced with the CosFace Loss function in the global branch of baseline model.CosFace Loss and Hard Triplet Loss were jointly employed to train the baseline model for the better identification on the similar individuals.AlignedReID++with CosFace Loss was outperformed the baseline model by 0.24 and 0.92 percentage points in the rank-1 accuracy and mAP,respectively,whereas,AlignedReID++with ArcFace Loss was exceeded by 0.36 and 0.56 percentage points,respectively.The improved model with the TA modules and CosFace Loss was achieved in a rank-1 accuracy of 94.42%,rank-5 accuracy of 98.78%,rank-10 accuracy of 99.34%,mAP of 63.90%,FLOPs of 5.45 G,frames per second(FPS)of 5.64,and model size of 23.78 M.The rank-1 accuracies were exceeded by 1.84,4.72,0.76 and 5.36 percentage points,respectively,compared with the baseline model,part-based convolutional baseline(PCB),multiple granularity network(MGN),and relation-aware global attention(RGA),while the mAP metrics were surpassed 6.42,5.86,4.30 and 7.38 percentage points,respectively.Meanwhile,the rank-1 accuracy was 0.98 percentage points lower than TransReID,but the mAP metric was exceeded by 3.90 percentage points.Moreover,the FLOPs of improved model were only 0.05 G larger than that of baseline model,while smaller than those of PCB,MGN,RGA,and TransReID by 0.68,6.51,25.4,and 16.55 G,respectively.The model size of improved model was 23.78 M,which was smaller than those of the baseline model,PCB,MGN,RGA,and TransReID by 0.03,2.33,45.06,14.53 and 62.85 M,respectively.The inference speed of improved model on a CPU was lower than those of PCB,MGN,and baseline model,but higher than TransReID and RGA.The t-SNE feature embedding visualization demonstrated that the global and local features were achieve in the better intra-class compactness and inter-class variability.Therefore,the improved model can be expected to effectively re-identify the beef cattle in natural environments of breeding farm,in order to monitor the individual feed intake and feeding time.展开更多
The Qinling Mountains, known for their rich vegetation and diverse pollinating insects, have seen a significant decline in bee species richness and abundance over recent decades, largely due to the introduction and sp...The Qinling Mountains, known for their rich vegetation and diverse pollinating insects, have seen a significant decline in bee species richness and abundance over recent decades, largely due to the introduction and spread of Apis mellifera. This decline has caused cascading effects on the region's community structure and ecosystem stability. To improve the protection of native bees in the natural and agricultural landscape of the Qinling Mountains and its surrounding areas, we investigated 33 sampling sites within three habitats: forest, forest-agriculture ecotones, and farmland. Using a generalized linear mixing model, t-test, and other data analysis methods, we explored the impact of Apis mellifera on local pollinator bee richness, abundance, and the pollination network in different habitats in these regional areas. The results show that(1)Apis mellifera significantly negatively affects the abundance and richness of wild pollinator bees,while Apis cerana abundance is also affected by beekeeping conditions.(2)There are significant negative effects of Apis mellifera on the community structure of pollinator bees in the Qinling Mountains and its surrounding areas: the Shannon-Wiener diversity index, Pielou evenness index, and Margalef richness index of bee communities at sites with Apis mellifera influence were significantly lower than those at sites without Apis mellifera influence.(3)The underlying driver of this effect is the monopolization of flowering resources by Apis mellifera. This species tends to visit flowering plants with large nectar sources, which constitute a significant portion of the local plant community. By maintaining a dominant role in the bee-plant pollination network, Apis mellifera competitively displaces native pollinator bees, reducing their access to floral resources. This ultimately leads to a reduction in local bee-plant interactions, decreasing the complexity and stability of the pollination network. These findings highlight the need for targeted conservation efforts to protect native pollinator species and maintain the ecological balance in the Qinling Mountains.展开更多
The cicada genus Vietanna is reviewed based on the descriptions of two new species,V.perparva sp.nov.and V.longiloba sp.nov.,from China.The relationship of this genus to related taxa is discussed based on the phylogen...The cicada genus Vietanna is reviewed based on the descriptions of two new species,V.perparva sp.nov.and V.longiloba sp.nov.,from China.The relationship of this genus to related taxa is discussed based on the phylogeny of Vietanna and representative species from subtribes Puranina,Leptopsaltriina,Euterpnosiina and Leptosemiina based on the mitochondrial gene COI and nuclear genes EF-1αand ARD1.展开更多
Two new species of Mukariini,Tiaobeinia coarseata sp.nov.from Shaanxi and Tiaobeinia yuani sp.nov.from Gansu,are described.Detailed morphological descriptions and illustrations of these new species are given.A checkli...Two new species of Mukariini,Tiaobeinia coarseata sp.nov.from Shaanxi and Tiaobeinia yuani sp.nov.from Gansu,are described.Detailed morphological descriptions and illustrations of these new species are given.A checklist of all known species in this tribe from China is provided and a key is proposed for all species of Tiaobeinia.展开更多
Background,aim,and scope Soil saturated hydraulic conductivity(K_(s))is a key parameter in the hydrological cycle of soil;however,we have very limited understanding of K_(s) characteristics and the factors that inf lu...Background,aim,and scope Soil saturated hydraulic conductivity(K_(s))is a key parameter in the hydrological cycle of soil;however,we have very limited understanding of K_(s) characteristics and the factors that inf luence this key parameter in the Mu Us sandy land(MUSL).Quantifying the impact of changes in land use in the Mu Us sandy land on K_(s) will provide a key foundation for understanding the regional water cycle,but will also provide a scientific basis for the governance of the MUSL.Materials and methods In this study,we determined K_(s) and the basic physical and chemical properties of soil(i.e.,organic matter,bulk density,and soil particle composition)within the first 100 cm layer of four different land use patterns(farmland,tree,shrub,and grassland)in the MUSL.The vertical variation of K_(s) and the factors that influence this key parameter were analyzed and a transfer function for estimating K_(s) was established based on a multiple stepwise regression model.Results The K_(s) of farmland,tree,and shrub increased gradually with soil depth while that of grassland remained unchanged.The K_(s) of the four patterns of land use were moderately variable;mean K_(s)values were ranked as follows:grassland(1.38 mm·min^(-1))<tree(1.76 mm·min^(-1))<farmland(1.82 mm·min^(-1))<shrub(3.30 mm·min^(-1)).The correlation between K_(s) and organic matter,bulk density,and soil particle composition,varied across different land use patterns.A multiple stepwise regression model showed that silt,coarse sand,bulk density,and organic matter,were key predictive factors for the K_(s) of farmland,tree,shrub,and grassland,in the MUSL.Discussion The vertical distribution trend for K_(s) in farmland is known to be predominantly influenced by cultivation,fertilization,and other factors.The general aim is to improve the water-holding capacity of shallow soil on farmland(0-30 cm in depth)to conserve water and nutrients;research has shown that the K_(s) of farmland increases with soil depth.The root growth of tree and shrub in sandy land exerts mechanical force on the soil due to biophysical processes involving rhizospheres,thus leading to a significant change in K_(s).We found that shallow high-density fine roots increased the volume of soil pores and eliminated large pores,thus resulting in a reduction in shallow K_(s).Therefore,the K_(s) of tree and shrub increased with soil depth.Analysis also showed that the K_(s) of grassland did not change significantly and exhibited the lowest mean value when compared to other land use patterns.This finding was predominantly due to the shallow root system of grasslands and because this land use pattern is not subject to human activities such as cultivation and fertilization;consequently,there was no significant change in K_(s) with depth;grassland also had the lowest mean K_(s).We also established a transfer function for K_(s) for different land use patterns in the MUSL.However,the predictive factors for K_(s) in different land use patterns are known to be affected by soil cultivation methods,vegetation restoration modes,the distribution of soil moisture,and other factors,thus resulting in key differences.Therefore,when using the transfer function to predict K_(s) in other areas,it will be necessary to perform parameter calibration and further verification.Conclusions In the MUSL,the K_(s) of farmland,tree,and shrub gradually increased with soil depth;however,the K_(s) of grassland showed no significant variation in terms of vertical distribution.The mean K_(s) values of different land use patterns were ranked as follows:shrub>farmland>tree>grassland;all land use patterns showed moderate levels of variability.The K_(s) for different land use patterns exhibited differing degrees of correlation with soil physical and chemical properties;of these,clay,silt,sand,bulk density,and organic matter,were identified as important variables for predicting K_(s) in farmland,tree,shrub,and grassland,respectively.Recommendations and perspectives In this study,we used a stepwise multiple regression model to establish a transfer function prediction model for K_(s) for different land use patterns;this model possessed high estimation accuracy.The ability to predict K_(s) in the MUSL is very important in terms of the conservation of water and nutrients.展开更多
基金National Key Research and Development Program(2023YFD1301801)National Natural Science Foundation of China(32272931)+1 种基金Shaanxi Province Agricultural Key Core Technology Project(2024NYGG005)Shaanxi Province Key R&D Program(2024NC-ZDCYL-05-12)。
文摘Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium to long distances over different cameras.However,beef cattle can tend to frequently move and change their feeding position during feeding.Furthermore,the great variations in their head direction and complex environments(light,occlusion,and background)can also lead to some difficulties in the recognition,particularly for the bio-similarities among individual cattle.Among them,AlignedReID++model is characterized by both global and local information for image matching.In particular,the dynamically matching local information(DMLI)algorithm has been introduced into the local branch to automatically align the horizontal local information.In this research,the AlignedReID++model was utilized and improved to achieve the better performance in cattle re-identification(ReID).Initially,triplet attention(TA)modules were integrated into the BottleNecks of ResNet50 Backbone.The feature extraction was then enhanced through cross-dimensional interactions with the minimal computational overhead.Since the TA modules in AlignedReID++baseline model increased the model size and floating point operations(FLOPs)by 0.005 M and 0.05 G,the rank-1 accuracy and mean average precision(mAP)were improved by 1.0 percentage points and 2.94 percentage points,respectively.Specifically,the rank-1 accuracies were outperformed by 0.86 percentage points and 0.12 percentage points,respectively,compared with the convolution block attention module(CBAM)and efficient channel attention(ECA)modules,although 0.94 percentage points were lower than that of squeeze-and-excitation(SE)modules.The mAP metric values were exceeded by 0.22,0.86 and 0.12 percentage points,respectively,compared with the SE,CBAM,and ECA modules.Additionally,the Cross-Entropy Loss function was replaced with the CosFace Loss function in the global branch of baseline model.CosFace Loss and Hard Triplet Loss were jointly employed to train the baseline model for the better identification on the similar individuals.AlignedReID++with CosFace Loss was outperformed the baseline model by 0.24 and 0.92 percentage points in the rank-1 accuracy and mAP,respectively,whereas,AlignedReID++with ArcFace Loss was exceeded by 0.36 and 0.56 percentage points,respectively.The improved model with the TA modules and CosFace Loss was achieved in a rank-1 accuracy of 94.42%,rank-5 accuracy of 98.78%,rank-10 accuracy of 99.34%,mAP of 63.90%,FLOPs of 5.45 G,frames per second(FPS)of 5.64,and model size of 23.78 M.The rank-1 accuracies were exceeded by 1.84,4.72,0.76 and 5.36 percentage points,respectively,compared with the baseline model,part-based convolutional baseline(PCB),multiple granularity network(MGN),and relation-aware global attention(RGA),while the mAP metrics were surpassed 6.42,5.86,4.30 and 7.38 percentage points,respectively.Meanwhile,the rank-1 accuracy was 0.98 percentage points lower than TransReID,but the mAP metric was exceeded by 3.90 percentage points.Moreover,the FLOPs of improved model were only 0.05 G larger than that of baseline model,while smaller than those of PCB,MGN,RGA,and TransReID by 0.68,6.51,25.4,and 16.55 G,respectively.The model size of improved model was 23.78 M,which was smaller than those of the baseline model,PCB,MGN,RGA,and TransReID by 0.03,2.33,45.06,14.53 and 62.85 M,respectively.The inference speed of improved model on a CPU was lower than those of PCB,MGN,and baseline model,but higher than TransReID and RGA.The t-SNE feature embedding visualization demonstrated that the global and local features were achieve in the better intra-class compactness and inter-class variability.Therefore,the improved model can be expected to effectively re-identify the beef cattle in natural environments of breeding farm,in order to monitor the individual feed intake and feeding time.
基金funded by the National Key R&D Program of China (2022YFE0115200)the Biodiversity Survey and the Assessment Project of the Ministry of Ecology and Environment, China (2019HJ2096001006)the National Animal Collection Resource Center, China。
文摘The Qinling Mountains, known for their rich vegetation and diverse pollinating insects, have seen a significant decline in bee species richness and abundance over recent decades, largely due to the introduction and spread of Apis mellifera. This decline has caused cascading effects on the region's community structure and ecosystem stability. To improve the protection of native bees in the natural and agricultural landscape of the Qinling Mountains and its surrounding areas, we investigated 33 sampling sites within three habitats: forest, forest-agriculture ecotones, and farmland. Using a generalized linear mixing model, t-test, and other data analysis methods, we explored the impact of Apis mellifera on local pollinator bee richness, abundance, and the pollination network in different habitats in these regional areas. The results show that(1)Apis mellifera significantly negatively affects the abundance and richness of wild pollinator bees,while Apis cerana abundance is also affected by beekeeping conditions.(2)There are significant negative effects of Apis mellifera on the community structure of pollinator bees in the Qinling Mountains and its surrounding areas: the Shannon-Wiener diversity index, Pielou evenness index, and Margalef richness index of bee communities at sites with Apis mellifera influence were significantly lower than those at sites without Apis mellifera influence.(3)The underlying driver of this effect is the monopolization of flowering resources by Apis mellifera. This species tends to visit flowering plants with large nectar sources, which constitute a significant portion of the local plant community. By maintaining a dominant role in the bee-plant pollination network, Apis mellifera competitively displaces native pollinator bees, reducing their access to floral resources. This ultimately leads to a reduction in local bee-plant interactions, decreasing the complexity and stability of the pollination network. These findings highlight the need for targeted conservation efforts to protect native pollinator species and maintain the ecological balance in the Qinling Mountains.
基金supported by the National Natural Science Foundation of China(32070476,32270496)。
文摘The cicada genus Vietanna is reviewed based on the descriptions of two new species,V.perparva sp.nov.and V.longiloba sp.nov.,from China.The relationship of this genus to related taxa is discussed based on the phylogeny of Vietanna and representative species from subtribes Puranina,Leptopsaltriina,Euterpnosiina and Leptosemiina based on the mitochondrial gene COI and nuclear genes EF-1αand ARD1.
基金supported by the National Natural Science Foundation of China(3207047932270497)National key Research and Development Program"Intergovernmental Cooperation on International Science and Technology Innovation"Special Project(2022YFE0115200)。
文摘Two new species of Mukariini,Tiaobeinia coarseata sp.nov.from Shaanxi and Tiaobeinia yuani sp.nov.from Gansu,are described.Detailed morphological descriptions and illustrations of these new species are given.A checklist of all known species in this tribe from China is provided and a key is proposed for all species of Tiaobeinia.
文摘Background,aim,and scope Soil saturated hydraulic conductivity(K_(s))is a key parameter in the hydrological cycle of soil;however,we have very limited understanding of K_(s) characteristics and the factors that inf luence this key parameter in the Mu Us sandy land(MUSL).Quantifying the impact of changes in land use in the Mu Us sandy land on K_(s) will provide a key foundation for understanding the regional water cycle,but will also provide a scientific basis for the governance of the MUSL.Materials and methods In this study,we determined K_(s) and the basic physical and chemical properties of soil(i.e.,organic matter,bulk density,and soil particle composition)within the first 100 cm layer of four different land use patterns(farmland,tree,shrub,and grassland)in the MUSL.The vertical variation of K_(s) and the factors that influence this key parameter were analyzed and a transfer function for estimating K_(s) was established based on a multiple stepwise regression model.Results The K_(s) of farmland,tree,and shrub increased gradually with soil depth while that of grassland remained unchanged.The K_(s) of the four patterns of land use were moderately variable;mean K_(s)values were ranked as follows:grassland(1.38 mm·min^(-1))<tree(1.76 mm·min^(-1))<farmland(1.82 mm·min^(-1))<shrub(3.30 mm·min^(-1)).The correlation between K_(s) and organic matter,bulk density,and soil particle composition,varied across different land use patterns.A multiple stepwise regression model showed that silt,coarse sand,bulk density,and organic matter,were key predictive factors for the K_(s) of farmland,tree,shrub,and grassland,in the MUSL.Discussion The vertical distribution trend for K_(s) in farmland is known to be predominantly influenced by cultivation,fertilization,and other factors.The general aim is to improve the water-holding capacity of shallow soil on farmland(0-30 cm in depth)to conserve water and nutrients;research has shown that the K_(s) of farmland increases with soil depth.The root growth of tree and shrub in sandy land exerts mechanical force on the soil due to biophysical processes involving rhizospheres,thus leading to a significant change in K_(s).We found that shallow high-density fine roots increased the volume of soil pores and eliminated large pores,thus resulting in a reduction in shallow K_(s).Therefore,the K_(s) of tree and shrub increased with soil depth.Analysis also showed that the K_(s) of grassland did not change significantly and exhibited the lowest mean value when compared to other land use patterns.This finding was predominantly due to the shallow root system of grasslands and because this land use pattern is not subject to human activities such as cultivation and fertilization;consequently,there was no significant change in K_(s) with depth;grassland also had the lowest mean K_(s).We also established a transfer function for K_(s) for different land use patterns in the MUSL.However,the predictive factors for K_(s) in different land use patterns are known to be affected by soil cultivation methods,vegetation restoration modes,the distribution of soil moisture,and other factors,thus resulting in key differences.Therefore,when using the transfer function to predict K_(s) in other areas,it will be necessary to perform parameter calibration and further verification.Conclusions In the MUSL,the K_(s) of farmland,tree,and shrub gradually increased with soil depth;however,the K_(s) of grassland showed no significant variation in terms of vertical distribution.The mean K_(s) values of different land use patterns were ranked as follows:shrub>farmland>tree>grassland;all land use patterns showed moderate levels of variability.The K_(s) for different land use patterns exhibited differing degrees of correlation with soil physical and chemical properties;of these,clay,silt,sand,bulk density,and organic matter,were identified as important variables for predicting K_(s) in farmland,tree,shrub,and grassland,respectively.Recommendations and perspectives In this study,we used a stepwise multiple regression model to establish a transfer function prediction model for K_(s) for different land use patterns;this model possessed high estimation accuracy.The ability to predict K_(s) in the MUSL is very important in terms of the conservation of water and nutrients.
