Alpha diversity and invasion success in kīpuka centers, edges, and continuous forest habitat
We collected a total of 38,289 individual arthropod specimens overall (Stainback: 11,610, Kona Hema: 12,130; kīpuka center: 7,887; kīpuka edge: 5,768; lava: 894). Different numbers of specimens were found for the different size categories (0 – 2: 11,653 total, 224 average; 2 – 4: 5,532 total, 106 average; 4 – 7: 1,917 total, 37 average; >7: 1,713 total, 33 average; Collembola: 17,474 total, 336 average). These specimens represented a total of 923 arthropod OTUs and 3,022 zOTUs. The zOTUs belonged to 23 orders, of which the most diverse were Hemiptera (675), Araneae (423), Diptera (337), Lepidoptera (299) and Coleoptera (249).
Different areas showed relatively comparable order compositions. This held true for relative zOTU read abundances (Suppl. Fig. 1A) as well as zOTU richness (Suppl. Fig. 1B). However, a clear difference was evident when comparing lava and kīpuka edges with the remaining areas, specifically within the orders Coleoptera and Diptera: lava and edges were distinguished by significantly fewer reads and zOTUs.
The continuous forests of Kona Hema and Stainbeck hosted similar arthropod richnesses, but a significantly higher arthropod richness than kīpuka sites (Fig. 2A). Moreover, we found a significantly higher richness in kīpuka cores than in kīpuka edge habitats. An even lower richness was found in barren lava (average zOTU|OTU richness Stainback: 332|145; Kona Hema: 331|152; kīpuka center: 246|120; kīpuka edges: 207|101; lava: 116|56). These general trends also largely held up when separate orders were analyzed (Suppl. Fig. 2). However, for Diptera, kīpuka centers showed comparable richness to continuous forest sites; for Psocoptera, all sites showed comparable richness; and, for Hemiptera, only lava sites were less species rich.
We found a significant positive correlation of logarithmic kīpuka size and the recovered zOTU and 3 % radius OTU richness in kīpuka centers (Fig. 2B; r2zOTU = 0.54, r2OTU = 0.47, LM p < 0.05). Larger kīpuka harbored a significantly more diverse community in their center than smaller ones. The increase of richness with area was steeper for zOTU, increasing by 79 % from 165 to 295 zOTUs over kīpuka sampled, than it was for 3 % radius OTUs, where it increased only 31 % from 86 to 125. Small kīpuka showed a considerably lower 3 % radius OTU and zOTU richness than continuous forests. However, large kīpuka approached richness values comparable to those documented for continuous forest, with richness approaching saturation in the cores of the seven largest kīpuka (7,388 m2 – 100,081 m2), particularly for OTUs. In contrast to kīpuka centers, we did not find an association of zOTU or 3 % radius OTU richness with kīpuka area for edge habitats (Fig. 2B; r2zOTU = 0.07, r2OTU = 0.12, LM p > 0.05). The richness-area relationship for kīpuka centers could also be replicated when separate orders were analyzed (Suppl. Fig. 3). It was particularly strong for Coleoptera (r2 = 0.48) and Diptera (r2 = 0.73). Araneae was the only order for which richness and kīpuka area were significantly associated in edge habitat (r2 = 0.38; LM p < 0.05).
While we found an overall higher spider species richness in continuous forest habitats than in kīpuka (Suppl. Fig. 2A), an inverse pattern was found for non-native spider species (Fig. 3A). The highest proportion of reads of non-native spiders was found in barren lava and kīpuka edge habitat. Significantly lower proportions of non-native spider reads were found in continuous forest and in kīpuka centers, with no significant differences between kīpuka centers and continuous forests. Kīpuka centers harbored a significantly lower proportion of non-native spiders than edges (Fig. 3A).
A strong species-area effect was also found for the proportion of non-native spider reads. The proportion of non-native spider OTUs had a highly significant, negative association with logarithmic kīpuka size (Fig. 3B). Large kīpuka harbored a considerably lower proportion of non-native spider reads and OTUs than small ones. A particularly strong drop of invasion success was visible between the four smallest (555 – 1,619 m2) and the remaining nine kīpuka (4,212 – 100,081 m2). This effect was visible in kīpuka centers and edges, but was particularly pronounced for edge habitat, which generally showed a higher presence of non-natives.