Aims:The increasing adoption of machine learning in ecological research has enabled the modeling of complex,nonlinear ecological patterns.However,the"black-box"nature of many machine learning models limits their interpretability,hindering the extraction of ecological insights.This review aims to introduce the core concepts,methods,and practical tools of interpretable machine learning(IML),and to demonstrate how these techniques can enhance ecological understanding from predictive models. Methods:We first clarify key distinctions among white-box model and black-box model,global interpretability and local interpretability,and intrinsic interpretability versus post-hoc interpretability models.Using a simulated dataset representing plant diversity and environmental variables(e.g.,elevation,temperature,soil moisture),we apply both white-box models(e.g.,linear regression,decision trees)and black-box models(e.g.,random forest)to illustrate major interpretability techniques,including regression coefficients,permutation importance,partial dependence plots(PDP),accumulated local effects(ALE),Shapley additive explanations(SHAP),and local interpretable model-agnostic explanations(LIME). Results:White-box models offer direct and transparent interpretability through their model structure,while black-box models require additional tools to derive explanations.Our case study shows that both model types can yield consistent insights about variable importance and ecological relationships.Furthermore,methods such as ALE and SHAP effectively address common limitations in conventional approaches like PDP by accounting for feature interactions and dependencies. Conclusion:IML provides a valuable toolkit for improving model transparency and interpretability in ecological research.It serves as a crucial complement to traditional statistical modeling,enabling researchers to extract meaningful ecological interpretations from complex models.As ecological data and modeling complexity continue to grow,the integration of IML techniques will become increasingly important for hypothesis generation and ecological decision-making.
Aims:Pteropods(Mollusca:Gastropoda)play a key role in marine ecosystems,contributing to critical processes such as the marine food web and carbon cycle.However,their diversity and distribution in the South China Sea remain poorly understood.To address these gaps and enhance the understanding of their regional biogeography,this study aimed to systematically investigate the species diversity and spatial distribution patterns of pteropods(Mollusca:Gastropoda)in the South China Sea. Methods:Zooplankton samples were collected during four cruises across various ecological regions,including estuarine,shelf,basin,and island/reef areas of the South China Sea between 2022 and 2023.Species were identified based on morphological examination using stereomicroscopy and scanning electron microscopy.Ecological indices,including abundance,Shannon-Wiener diversity index,and Pielou evenness index,were calculated to analyze community structure. Results:A total of 26 pteropod species belonging to 13 families and 18 genera were identified,including four new records for the South China Sea:Diacavolinia elegans,Desmopterus gardineri,Notobranchaea tetrabranchiata,and Cliopsis krohnii.Notable regional heterogeneity was observed:the Pearl River Estuary-western South China Sea shelf showed the highest species richness(22 species),while the Northern South China Sea islands and reefs had the lowest(6 species).Horizontally,coastal areas exhibited higher abundance but lower diversity,whereas open-sea areas showed the reverse pattern.Vertically,pteropods were predominantly concentrated in the 0-200 m layer,with abundance and diversity declining sharply with depth.The composition of dominant species varied markedly among different ecological regions. Conclusion:This study updates the species checklist of pteropods in the South China Sea and reveals clear spatial partitioning in their community structure,reflecting adaptations to distinct habitats.The findings provide essential baseline data for understanding regional marine biodiversity and monitoring ecosystem changes in this marginal sea.
Aims:Pollinator-mediated variation in floral morphology is mostly observed in specialized pollination systems.For widely distributed plants that rely on generalized pollination,it remains unclear how their morphological traits and pollinator assemblages vary along geographic gradients,and their correlations with latitude,longitude and altitude.Our preliminary study has revealed that Dipsacus asper(Dipsacaceae)is widely distributed in southwestern China and exhibits a typical generalized pollination system,making it an ideal material for investigating this question. Methods:In this study,we took D.asper as the experimental material and conducted field investigations across 13 wild populations in Hubei,Sichuan,Yunnan,and Guizhou provinces.We measured the morphological traits of the plants,compared the morphological characteristics among populations,and calculated the coefficients of variation(CV)within and between populations.Additionally,we observed and recorded the pollinator types in each population and calculated their visitation frequencies.We also measured the body characteristics of the main pollinator bumblebees.Finally,we quantified natural seed set of D.asper across the different populations. Results:The results indicated significant variation in the morphological traits of D.asper across different populations.Specifically,the CV were high for leaf and stem traits,whereas inflorescence size showed low variation.Dipsacus asper plants from Xiaojin,Jianshi,Yanyuan,and Xishui populations exhibited larger inflorescences compared to other populations.The primary pollinators of D.asper included Bombus,Apis,Eristalis,and Lepidoptera,with bees serving as key pollinators.Both pollinator composition and visitation frequency varied markedly among populations.The larger-bodied bumblebees was main pollinator for plant in Xiaojin,Jianshi,Yanyuan,and Xishui populations.There was no significant correlation between the floral tube length of D.asper and the proboscis length of widely distributed Bombus friseanus.Plant height,stem,leaf,and inflorescence size of D.asper,as well as the morphological traits of B.friseanus,decreased significantly with increasing elevation and they also varied with longitude and latitude.Plant height and the proboscis length of B.friseanus increased significantly with increasing longitude,while the sizes of stems,leaves,inflorescences,and the abdomen of B.friseanus decreased significantly with increasing longitude.Leaf size,inflorescence size,the thickness of the head and thorax,and the abdomen of B.friseanus increased significantly with increasing latitude,whereas the proboscis length of B.friseanus decreased significantly with increasing latitude.Additionally,the natural seed set of D.asper with different geographic distributions showed an increasing trend with the increase of longitude,a significant decreasing trend with the increase of latitude,and a significant increasing trend with the increase of altitude. Conclusion:The morphological traits and pollinator diversity of widely distributed plant species exhibit significant variation along geographic gradients.This variation may result from the combined effects of plant-pollinator interactions and abiotic environmental factors.Across different geographic environments,plants ensure their reproductive success by adapting floral morphology and utilizing diverse pollinators.
Aim:The chloroplast genome has been widely used for plant species identification,delimitation,and phylogenetic reconstruction due to its structural conservation,maternal uniparental inheritance,and high copy number.However,the chloroplast genome cannot effectively reflect pollen-mediated gene flow,is sensitive to genetic drift,and is susceptible to incomplete lineage sorting and hybridization events.Therefore,in plant identification and population evolutionary history studies,relying solely on chloroplast genome evidence may lead to biased or incomplete conclusions. Methods:This study performed whole-chloroplast genome sequencing,assembly,and annotation for seven Pterocarya individuals,and conducted comparative genomic and phylogenetic analyses incorporating published data. Results:The chloroplast genome ranged in length from 160,176 to 160,318 bp and exhibited a typical quadripartite structure,encoding 131 genes(86 protein-coding genes,8 rRNAs,and 37 tRNAs).The number of simple sequence repeats(SSRs)loci ranged from 87 to 96,dominated by mononucleotide repeats.Comparative genomic analysis indicated that chloroplast genomes among Pterocarya species were highly conserved overall.The sequence conservation of the inverted repeat regions and coding regions was higher than that of the single-copy regions and non-coding regions.Notably,within the coding regions,the ndhF gene exhibited significantly higher sequence variation than other protein-coding genes.The results of phylogenetic analysis showed that the phylogenetic trees constructed based on different data partitions(whole genome,protein-coding sequence,hypervariable regions,and sliding windows)exhibited various topological structures with two to five major clades.Individuals of species such as P.stenoptera,P.macroptera var.macroptera,and P.insignis consistently showed lineage admixture and cross-clade distribution across different analyses. Conclusion:The chloroplast genome exhibits significant limitations in resolving species identification and biogeographic history within the genus Pterocarya,and relying solely on its data may lead to systematic bias.Therefore,future studies on plant phylogenetics should integrate nuclear genomic data and employ multispecies coalescent methods to achieve more accurate species identification and evolutionary relationship reconstruction.
Background & Aims:Global biodiversity is declining at an unprecedented rate.As a widespread and diverse taxonomic group,birds are crucial indicators,and understanding drivers and consequences of their population dynamics is vital for biodiversity conservation.Passive acoustic monitoring(PAM),a non-invasive technique with high spatiotemporal resolution,offers a revolutionary tool for large-scale,long-term avian diversity surveys.Despite the rapid development of PAM globally,its application in China is currently hampered by scattered projects,heterogeneous equipment,and inconsistent methodologies.This has resulted in a"data island"dilemma,significantly hindering data integration and comparability and limiting the utility of these data for biodiversity policy and conservation.This study aims to provide a scientific blueprint and implementation guide for establishing a standardized,networked,and intelligent avian acoustic monitoring system in China. Method & Results:We systematically reviewed the corresponding urgent needs of PAM in light of the global loss of avian biodiversity.We then compared and analyzed the technical development,network construction,and data-sharing experiences of avian PAM both internationally and domestically,identifying the core challenges China faces due to the lack of unified standards.Drawing on successful international PAM practices,we propose a comprehensive framework of standards suitable for China.This framework covers the entire workflow,from monitoring design and core technical parameters to metadata standards and data processing and analysis. Perspectives:Finally,we present a series of actionable recommendations and a strategic roadmap for future implementation.These proposals address multiple facets,including top-level design,national network construction,technological innovation,data sharing mechanisms,and essential policy support,charting a clear course for the future of avian acoustic monitoring in the nation.
