摘要
探讨了在人为控制不同因子变量的条件下各景观格局指标对由中性随机模型产生的不同景观格局系列的反应 ,以评价一些常用指标的实用性和局限性。研究结果表明 ,大部分指标所指示的格局特征往往是不全面的 ,即它们只对格局系列中个别因子的变化敏感 ,而对另一些因子的变化反应迟钝。比较值得推荐的指标有 :总斑块数目 ,平均斑块大小 ,总边界密度 ,分维数 ,蔓延度 ,聚集度。但即使是这些指标 ,也各有其局限性 ,且存在冗余。由此提醒读者在运用景观格局指标时应在了解其实际意义的基础上 ,结合生态学过程慎重选择和解释 ,避免陷于数字游戏。
Landscape metrics have been widely used to describe landscape patterns quantitatively in the last two decades. With the quick development of geographical information system, more and more metrics are designed and calculated with ever increasing speed. Packages which can provide numerous varieties of landscape metrics, such as FRAGSTATS and APACK, make it more convenient for users to calculate those metrics without querying about their ecological meaning, or even without knowing their arithmetic formulas. Therefore more and more users are using and interpreting landscape metrics arbitrarily, especially in China. It is high time to clarify if the metrics we are using so often can really indicate the desired pattern or not.This paper tested the behavior of some landscape pattern metrics against six pattern scenarios generated by neutral landscape models, by changing one spatial parameter while keeping all the others stable. The scenarios include: (1) Number of Classes, with map size 1000×1000, and 2～100 classes randomly distributed at equal area percentages; (2) Scale - Map Extent, with 10 equal classes randomly distributed at map size 8×8, 16×16,…, 500×500, and 1000×1000; (3) Scale - Resolution, with a 3-class predefined 10×10 map resampled by cell size 1, 0.5,…, and 0.05; (4) Proportion of One Class, with the area of one class changing from 1%, 10%, …, to 99% consequently in binary maps; (5) Aggregation Level - Rule, with 4-equal-classes in 1024×1024 cells maps at different aggregation levels according to the neutral landscape model RULE; and (6) Aggregation Level - SimMap, with 4-equal-classes in 500×500 cells maps at different aggregation levels according to the neutral landscape model SimMap. Landscape metrics were calculated and compared for different pattern cases in each scenario.Results demonstrated that most of the metrics response to some of the pattern scenarios only, for example, Average Patch Perimeter/Area Ratio and Shonnon Diversity, while they are not sensitive to the others. Therefore none of them can indicate all aspects of a landscape pattern. However, in spite of those limitations, still some of the metrics are recommended for future use: Total number of patches, Average patch size, Total edge density, Fractal double-logged, Contagion (Li & Reynolds), and Aggregation index. But attention must be paid on the limitation, redundancy and real meaning of the metrics. The relationship between metric values and ecological processes are more important than the values themselves. For instance, suitable habitat area, number of patches and lacunarity metrics might be well related to population size and growth, while pattern metrics of erosion-sensitive land use types could be related to hydrological indicators such as modulus of runoff, or modulus of erosion. However, this paper could not answer all questions regarding to the relationship between metrics and landscape patterns. For example, why there is a peak at class level when N=5 for Number of Patches in the Number of Classes scenario? Why the landscape level Average Patch Size is the lowest when one class reaches 10%～20% in a binary map? Similar questions also arise for more complicated indicators such as Fractal and Contagion. Further study on mathematical analysis of the research results is expected for future consideration.
出处
《生态学报》
CAS
CSCD
北大核心
2004年第1期123-134,共12页
Acta Ecologica Sinica
基金
中国科学院"引进国外杰出人才"资助项目
国家重点基础研究专项经费资助项目 (2 0 0 2 CB1115 0 6)
国家自然科学基金资助项目(4 0 3 3 10 0 8
40 0 0 10 0 2
3 0 2 70 2 2 5
40 1710 3 7)
中国科学院知识创新资助项目 (SCXZY0 10 2 )~~
关键词
景观格局
格局指标
反应
中性模型
landscape pattern
pattern metrics (indices)
response
neutral model
