Get access

Evaluating the relative importance of patch quality and connectivity in a damselfly metapopulation from a one-season survey


  • Takehiko Yamanaka,

  • Koichi Tanaka,

  • Kenji Hamasaki,

  • Yukinobu Nakatani,

  • Nobusuke Iwasaki,

  • David S. Sprague,

  • Ottar N. Bjørnstad

T. Yamanaka (, K. Tanaka and K. Hamasaki, Biodiversity Division, National Inst. for Agro-Environmental Sciences, 3-1-3 Kannondai, Tsukuba, JP–305-8604 Ibaraki, Japan. – Y. Nakatani, Natural Resources Inventory Center, National Inst. for Agro-Environmental Sciences, 3-1-3 Kannondai, Tsukuba, JP–305-8604 Ibaraki, Japan. – N. Iwasaki and D. S. Sprague, Ecosystem Informatics Division, National Institute for Agro-Environmental Sciences, 3-1-3 Kannondai, Tsukuba, JP–305-8604 Ibaraki, Japan. – O. N. Bjørnstad, Depts of Entomology and Biology, 501 Ag Sciences and Industries Building, The Pennsylvania State Univ., University Park, PA 16802, USA.


The area-and-isolation paradigm, which has been the primary focus of metapopulation research, may not hold in some animal metapopulations if within-patch preference is more important than patch area or connectivity. Recently, regression analyses have been used to evaluate the effect of patch connectivity and various patch qualities including area. However, their relative importance is not easy to determine, because patch qualities and connectivity are often spatially autocorrelated. In this paper, we try to evaluate the relative importance of within-patch quality, patch connectivity and spatial autocorrelation using variation partitioning methods from community ecology. We constructed three regression models: within-patch quality, PCNM (principal coordinates of neighbor matrices) and patch connectivity based on a one-season survey of a damselfly Copera annulata metapopulation. The contribution of within-patch quality was larger than that of connectivity. There was no prominent effect of patch area. We conclude that the area-and-isolation paradigm is not applicable to this C. annulata metapopulation. The spatial autocorrelation extracted by PCNM had the largest contribution; it contained almost all of the variation of connectivity and overlapped with variation explained by within-patch quality. Connectivity corresponded most closely to medium-scale spatial structure captured by PCNM (ca 640 m). The mean effective dispersal scale was estimated to be 53 m. Within-patch quality, debris accumulation and vegetation cover in the pond corresponded with the medium and small (ca 201 m) spatial scales from PCNM, though we could not clearly explain the cause of this correspondence. We believe that our method will contribute to quick and effective evaluation of spatial and non-spatial aspects of metapopulation.