Remote sensing of subtropical tree diversity:The underappreciated roles of the practical definition of forest canopy and phenological variation

Tree species diversity is vital for maintaining ecosystem functions,yet our ability to map the distribution of tree diversity is limited due to difficulties in traditional field-based approaches.Recent developments in spaceborne remote sensing provide unprecedented opportunities to map and monitor tree diversity more efficiently.Here we built partial least squares regression models using the multispectral surface reflectance acquired by Sentinel-2 satellites and the inventory data from 74 subtropical forest plots to predict canopy tree diversity in a national natural reserve in eastern China.In particular,we evaluated the underappreciated roles of the practical definition of forest canopy and phenological variation in predicting tree diversity by testing three different definitions of canopy trees and comparing models built using satellite imagery of different seasons.Our best models explained 42%–63%variations in observed diversities in cross-validation tests,with higher explanation power for diversity indices that are more sensitive to abundant species.The models built using imageries from early spring and late autumn showed consistently better fits than those built using data from other seasons,highlighting the significant role of transitional phenology in remotely sensing plant diversity.Our results suggested that the cumulative diameter(60%–80%)of the biggest trees is a better way to define the canopy layer than using the subjective fixeddiameter-threshold(5–12 cm)or the cumulative basal area(90%–95%)of the biggest trees.Remarkably,these approaches resulted in contrasting diversity maps that call attention to canopy structure in remote sensing of tree diversity.This study demonstrates the potential of mapping and monitoring tree diversity using the Sentinal-2 data in species-rich forests.

Research on the Key Techniques of TCP Protocol Normalization for Mimic Defense Architecture

The Mimic Defense(MD)is an endogenous security technology with the core technique of Dynamic Heterogeneous Redundancy(DHR)architecture.It can effectively resist unknown vulnerabilities,backdoors,and other security threats by schedule strategy,negative feedback control,and other mechanisms.To solve the problem that Cyber Mimic Defense devices difficulty of supporting the TCP protocol.This paper proposes a TCP protocol normalization scheme for DHR architecture.Theoretical analysis and experimental results show that this scheme can realize the support of DHR-based network devices to TCP protocol without affecting the security of mimicry defense architecture.

鼎湖山南亚热带常绿阔叶林叶功能性状沿群落垂直层次的种内变异

探究功能性状沿着环境梯度如何变化一直以来是基于性状的群落生态学的核心问题之一。尽管功能性状存在种内和种间变异,但种内变异沿环境梯度如何变化仍有待探究。本文以鼎湖山南亚热带常绿阔叶林1.44 ha塔吊样地内16个树种的2,820个个体为研究对象,探究4种叶功能性状(比叶面积、叶干物质含量、叶厚度和叶面积)沿群落垂直层次的种内变异。首先,利用随机效应线性模型量化塔吊样地内的种内变异和种间变异;其次,利用Kmeans函数将森林的垂直层次划分为灌木层、亚冠层和林冠层,并通过构建回归模型探究叶功能性状在群落垂直层次中的种内变异格局。最后,应用混合线性模型和单因素方差分析的方法探究叶功能性状沿垂直层次的种内变异是否具有物种依赖性。结果表明:在局域群落中,并非所有叶功能性状的种内变异都低于种间变异;叶功能性状在不同垂直层次的种内变异格局存在显著差异,且种内变异与垂直范围呈正相关;叶功能性状的种内变异具有较强的物种依赖性,因此树种差异相对于小环境解释了更多的性状变异;此外,不同叶功能性状的种内变异沿垂直层次的变化趋势并不一致。本研究发现种内变异对于物种共存具有重要作用。

基于枝条木材密度分级的鼎湖山南亚热带常绿阔叶林树高曲线模型

如何便捷准确地测量树高一直是林学及群落生态学所关心的问题。由于木材密度与树木生长密切相关,因此基于木材密度建立树高曲线模型能够为测量树高提供新的方法。本文以鼎湖山南亚热带常绿阔叶林1.44 ha塔吊样地内119个物种的4,032个个体为研究对象,利用树高、胸径和木材密度数据来探究基于枝条木材密度分级的树高曲线模型。首先,对个体进行随机抽样,将其划分为建模样本(占总样本量的70%)和检验样本(占总样本量的30%),并通过聚类分析将所有个体的木材密度划分为4级。其次,基于建模样本利用常见的5种理论生长方程(Richards、Korf、Logistic、Gompertz和Weibull方程)对不同分级建立树高–胸径模型;基于检验样本检验模型精度,并确定各分级的最适模型。最后,构建基于物种分类的树高曲线模型,并比较其与木材密度分级模型的差异。结果表明:基于木材密度分级的模型,各分级小组检验样本的平均绝对误差(MAE)和均方根误差(RMSE)最小值所对应的模型类型与建模样本结果一致,确定Gompertz模型和Weibull模型为鼎湖山南亚热带常绿阔叶林最适树高模型类型。比较基于木材密度分级的模型与基于物种分类的模型,发现二者的MAE、RMSE指数差异小。综上,基于木材密度分级的树高曲线模型对树高估测精度高,使用方便,为树高预测提供了新方法,可以更好服务森林调查等生产实践。