Relative resource abundance explains butterfly biodiversity in island communities
Naoaki Yamamoto, Jun Yokoyama, and Masakado Kawata*
Department of Ecology and Evolutionary Biology, Graduate School of Life Sciences, Tohoku University, Aoba-ku Sendai 980-8578, Japan
Edited by James H. Brown, University of New Mexico, Albuquerque, NM, and approved May 1, 2007 (received for review February 20, 2007) Ecologists have long been intrigued by the factors that control the diversity of species is determined by interspecific competition pattern of biodiversity, i.e., the distribution and abundance of and the diversity of resources. Many studies have demonstrated species. Previous studies have demonstrated that coexisting spe- that coexisting species partition their resources (15–17), which cies partition their resources and/or that the compositional simi- lends support to the niche-assembly model. However, recent larity between communities is determined by environmental fac- studies have shown that the neutral theory can explain the tors, lending support to the niche-assembly model. However, no biodiversity of several plant communities (1, 18), although other attempt has been made to test whether the relative amount of studies have refuted the theory (17, 19–21). An appropriate test resources that reflects relative niche space controls relative species that differentiates between the niche and neutral models is to abundance in communities. Here, we demonstrate that the relative assess the different predictions from the competing theories (17, abundance of butterfly species in island communities is signifi- 19). The neutral model predicts that the compositional similarity cantly related to the relative biomasses of their host plants but not between communities will decrease as the distance between two to the geographic distance between communities. In the studied points increases (17). In contrast, the niche-apportionment communities, the biomass of particular host plant species posi- model predicts that the similarity of relative species abundance tively affected the abundance of the butterfly species that used between communities will increase with an increase in the them, and consequently, influenced the relative abundance of the similarity of relative resource abundance. The relative contri- butterfly communities. This indicated that the niche space of bution of distance and resource abundance to the similarity butterflies (i.e., the amount of resources) strongly influences but- should be evaluated (22). terfly biodiversity patterns. We present this field evidence of the Previous studies supporting the niche theory have demon- niche-apportionment model that propose that the relative amount strated that coexisting species partition their resources or that of niche space explains the pattern of the relative abundance of the community composition is related to environmental conditions. species in communities. However, these studies did not evaluate the relative abundance of available resources; consequently, the relationships between neutral theory ͉ niche theory ͉ relative species abundance niche space and species abundance were not directly tested. Thus, to date, there has been no attempt to test whether the iodiversity is often considered to be synonymous with spe- relative abundance of available resources, i.e., available niche Bcies richness and relative species abundance (1). For more space, affects the relative abundance of species. In this study, than half a century, ecologists have paid attention to the factors butterfly communities were examined because both the larval that determine the relative species abundance in ecological and adult stages of butterflies depend almost entirely on specific communities; i.e., the commonness and rarity of species (2). Host plants for their dietary requirements, and information as to plants and their herbivore communities are good systems for which host plant a butterfly uses is generally available. In examining how the diversity of herbivores is influenced by their addition, we could estimate the host plant abundance using a resources; indeed, insect ecologists have focused considerable high-resolution aerial photograph and field surveys. Therefore, attention on the question of how host plant communities affect using the host plant–butterfly system, we were able to directly the species richness and composition of insects. Although a test the relationship between fundamental niche space and recent review (3) pointed out that little progress on this subject relative species abundance. has been made since the 1980s, a number of important studies have demonstrated that a large number of plant species is Results frequently correlated with a large number of insect species We examined butterfly species diversity on five of the Urato (4–6). Experimental studies have also demonstrated a positive Islands, Japan (Fig. 1). Throughout the sampling period, a total relationship between the diversity of plant species and the of 39 species were identified and used in the study (see Materials diversity of consumers (7–9). The abundance of resources also and Methods: Mahanashi, 30; Hoh, 25; Katsura, 31; Nono, 34; represents an important factor in the structuring of insect Sabusawa, 35); these butterflies potentially use 58 host plant communities (10). A number of studies have demonstrated that species [supporting information (SI) Fig. 3]. Of the butterfly resource abundance explains the variation in the abundance and species, nine are restricted to the use of a single host plant species, species richness of herbivorous insects (11–14). However, these and, for eight species, the host plants are used by no other studies on host plant and herbivore communities have treated species richness and species composition as components of the diversity, and no studies have examined the relative abundance Author contributions: N.Y. and M.K. designed research; N.Y., J.Y., and M.K. performed patterns of both host plants and herbivores. research; N.Y., J.Y., and M.K. analyzed data; and M.K. wrote the paper. On the other hand, recent theoretical arguments have focused The authors declare no conflict of interest. on whether the neutral or niche theory better explains the This article is a PNAS Direct Submission. patterns of relative species abundance. The neutral theory (1) of Freely available online through the PNAS open access option. biodiversity based on a dispersal-assembly perspective assumes Abbreviation: AIC, Akaike information criterion. that the relative abundance of species in communities is deter- *To whom correspondence should be addressed. E-mail: [email protected]. mined by random dispersal and stochastic local extinction. This article contains supporting information online at www.pnas.org/cgi/content/full/ Conversely, the niche-assembly view proposes that coexisting 0701583104/DC1. species should have different niches so that the abundance and © 2007 by The National Academy of Sciences of the USA
10524–10529 ͉ PNAS ͉ June 19, 2007 ͉ vol. 104 ͉ no. 25 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0701583104 Downloaded by guest on September 25, 2021 2 km 1 2 3 4
5 6 7 8
9 10 11 12 Matsushima Bay
Hoh
Urato Islands
Pacific Ocean
Nono
Mahanashi Katsura
2 km Sabusawa
Fig. 1. Maps of the Urato Islands and vegetation classification. Eight communities were considered when the Katsura and Sabusawa Islands were divided into two and three areas, respectively, as shown by the black lines. 1, residential areas, roads, and other man-made structures; 2, rice fields; 3, other crop fields; 4, grasslands with vegetation Ͻ50-cm high; 5, grasslands dominated by Miscanthus sinensis and other tall grasses; 6, wetland vegetation mainly dominated by Phragmites communis; 7, bamboo or dwarf bamboo thickets; 8, Cryptomeria japonica plantations; 9, Pinus densiflora (P. thunbergii in part) forests; 10, Machilus thunbergii forests; 11, deciduous forests mainly dominated by Quercus serrata; 12, mixed secondary forest dominated by Juglans mandshurica var. sachalinensis, Celtis sinensis var. japonica, and Neolitsea sericea.
butterfly species. In this study, plant biomass was estimated as source, but there was no significant correlation between the the weight of the edible parts of the plants. An analysis of the matrices of Resource and Distance (r ϭ 0.3945, P ϭ 0.103). The diversity and abundance of host plants and butterfly species was multiple regression of the Resource and Distance matrices on conducted for eight communities, of which two communities Butterfly matrix indicated that the relative abundance pattern of were classified on Katsura Island and three were classified on butterflies was highly significantly influenced by the relative Sabusawa Island (the analysis of five communities, where each abundance pattern of their host plants, but not by geographic community corresponded to a single island, was also conducted, distance (Table 1 and Fig. 2). It is possible that communities and the results are provided in SI Table 3 and SI Fig. 5). This within islands were similar in the relative abundance of butter- community division was based on the observation that Katsura flies. When within- and between-islands factors were used as Island and Sabusawa Island support different vegetation types independent variables instead of the Distance matrix for the (Fig. 1; see also Materials and Methods). multiple regression, the Butterfly matrix was not significantly The number of butterfly species did not differ greatly among influenced by within-island effects (P ϭ 0.3996). the communities (26–30 species), and there was no significant The effects of the biomass densities of host plant species on the effect of host plant species richness on butterfly species richness density of each butterfly species that used them were examined. 2 (regression analysis, b ϭ 0.0884, F1,6 ϭ 1.151, P ϭ 0.3247, R ϭ 0.1615, SI Fig. 4a). Simple regression analysis indicated that the total host biomass densities significantly affect butterfly biomass Table 1. Results of a multiple regression of the Resource and 2 densities (b ϭ 0.01513, F1,6 ϭ 31.95, P ϭ 0.0013, R ϭ 0.8419, SI Distance matrices on the Butterfly matrix by using an extension Fig. 4b), but butterfly biomass densities were not observed to of the Mantel test ECOLOGY have any effect on butterfly species richness (b ϭϪ0.491, F ϭ 2 1,6 AB.C AC.B R 1.142, P ϭ 0.3264, R2 ϭ 0.1599, SI Fig. 4c). By using an extension of the Mantel test (23, 24), we examined Distance 0.8052 0.1529 0.7654 whether the similarities of the relative abundance pattern of P Ͻ 0.0002 P ϭ 0.2258 P Ͻ 0.0002 butterfly species between the communities (Butterfly matrix) Log distance 0.8121 0.1603 0.7685 are determined by the similarities of the relative abundance P Ͻ 0.0002 P ϭ 0.224 P Ͻ 0.0002
pattern of their host plant species (Resource matrix), by geo- AB.C, regression coefficient of the Resource matrix on the Butterfly matrix. graphic distances (Distance matrix), or by both. There was a AC. B, regression coefficient of the Distance matrix on the Butterfly matrix. The clear correlation between the matrices of Butterfly and Re- results for five-community analysis are shown in SI Table 3.
Yamamoto et al. PNAS ͉ June 19, 2007 ͉ vol. 104 ͉ no. 25 ͉ 10525 Downloaded by guest on September 25, 2021 Table 2. Effects of host plant biomass on the butterfly densities a 0.8 of each species Butterfly Host plant 0.6 Graphium sarpedon nipponum 5* (0.907) Hestina persimilis japonica — 0.4 Papilio xuthus 2¶ (0.429) Papilio macilentus 3†, 2 (0.895) Papilio bianor dehaanii 2¶ (0.393) 0.2 Papilio machaon hippocrates 6‡ (0.657) r = 0.895 Papilio protenor 2 (0.294) Dissimilarity of Butterfly P = 0.003 Atrophaneura alcinous 1* (0.997) 0 Eurema hecabe 12‡, 16, 17 §, 18 † (0.429) 0 0.2 0.4 0.6 0.8 1 Artogeia melete 11† (0.830) Dissimilarity of Resource Colias erate poliographys 10‡, 12‡, 15‡, 20§ (0.982) Artogeia rapae crucivora 11¶ (0.368) Everes argiades 31‡, 16, 29, 30 (0.967) b 0.8 Rapala arata 25‡, 16, 21, 30 (0.978) Lycaena phlaeas daimio — 0.6 Pseudozizeeria maha — Celastrina argiolus ladonides 25, 22, 28¶, 30¶ (0.745) Vanessa indica 40† (0.894) 0.4 Limenitis camilla japonica — Sasakia charonda — Polygonia c-aureum — 0.2 r = 0.476 Hestina persimilis japonica — Dissimilarity of Butterfly P = 0.150 Neptis sappho intermedia 14 (0.307) Cynthia cardui 38§ (0.509) 0 Kaniska canace — 024 6 Geographic Distance (km) Libythea celtis — Lethe diana 48 (0.267) c 1 Mycalesis francisca perdiccas — Minois dryas bipunctata 51* (0.999) Lethe sicelis 54¶, 48, 53, 55 (0.706) 0.8 Ypthima argus — Mycalesis gotama fulginia — 0.6 Neope niphonica 54 (0.476) Parnara guttata 55†, 46, 56 (0.840) 0.4 Polytremis pellucida 53§, 48, 43, 55 (0.938) Potanthus flavus — 0.2 r = 0.394 Thoressa varia 53¶, 48 (0.578) Dissimilarity of Resource P = 0.294 Daimio tethys — ¶ 0 Isoteinon lamprospilus 44 , 47 (0.503) 0246 Pelopidas jansonis — Geographic Distance (km) The numbers below indicate host plant species and are also shown in SI Fig. 3. The host plant species that were included in the model that yielded the Fig. 2. Relationship between the dissimilarities between butterfly commu- lowest AIC are shown. The symbol–indicates that no host plant species was nities and host plant communities (a), dissimilarities between butterfly com- selected in the model. Italicized numbers indicate host plants with a negative munities and geographic distance (b), and dissimilarities between host plant regression coefficient. Values in parentheses indicate the R2 of the regression. communities and geographic distance (c). The r and P values shown in the Significant values (P) were adjusted for controlling multiple comparisons by figures were obtained by the Mantel test. Dissimilarities in relative species using the false discovery rate (FDR) procedure. 1, Aristolochia kaempferi;2, abundance of butterfly and relative resource abundance were calculated by Zanthoxylum ailanthoides;3,Orixa japonica;5,Neolitsea sericea;6,Oenanthe using Odum’s percentage difference method. The results for five-community javanica;10, Trifolium pratense; 11, Brassica rapa; 12, Medicago polymorpha; analysis are shown in SI Fig. 5. 14, Lespedeza buergeri; 15, Trifolium repens; 16, Lespedeza homoloba;17, Albizia julibrissin; 18, Lespedeza juncea var. serpens; 20, Lotus corniculatus Ϸ var. japonicus; 21, Deutzia crenata; 22, Vicia angstifolia; 25, Pueraria lobata; The results showed that in 23 ( 60%) of 39 butterfly species, the 28, Rosa wichuraiana; 29, Lathyrus japonicus; 31, Kummerowia stipulacea; 38, regression model that included the biomass of one or more host Cirsium nipponicum; 40, Boehmeria longispica; 43, Pleioblastus chino; 44, plant species as explanatory variables showed a lower Akaike Spodiopogon sibiricus; 46, Carex doniana; 47, Miscanthus sinensis; 48, information criterion (AIC) value than that obtained without the Sasamorpha borealis; 51, Calamagrostis arundinacea var. brachytricha; 53, biomass, and the densities positively increased with increasing Phyllostachys bambusoides; 54, Sasa nipponica; 55, Phyllostachys heterocycla; host plant biomass (Table 2). After controlling the error of 56, Oryza sativa. *, P Ͻ 0.001; †, P Ͻ 0.01; ‡, P Ͻ 0.05; §, P Ͻ 0.1; ¶, P Ͻ 0.2. multiple comparisons, one or more host plant species weakly (8 butterfly species, P Ͻ 0.2) or significantly (12 butterfly species, P Ͻ 0.05) affected the butterfly densities. abundance of consumer species. The results indicate that the niche space of butterflies (i.e., the amount of their resources) Discussion strongly influences the abundance of butterflies and, conse- The present results demonstrate that the relative abundance of quently, butterfly biodiversity patterns. This was also supported resource biomass has a highly significant effect on the relative by the results that the total biomass densities significantly affect
10526 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0701583104 Yamamoto et al. Downloaded by guest on September 25, 2021 the total biomass densities of all of the butterfly species and that negative relations were observed between the densities of Papilio the biomass of certain particular host plant species explains the protenor and the host plant Zanthoxylum ailanthoides. This might abundance of the butterfly species that use them. The positive be because the increasing biomass of Z. ailanthoides increased relationships between the total biomass and total butterfly the densities of the competitors Papilio macilentus and Papilio biomass suggest that resource availability actually regulates bianor dehaanii, resulting in Papilio protenor changing its host butterfly abundance. However, butterfly species richness did not plants. Second, unknown ecological or environmental factors increase with increasing butterfly biomass densities. Thus, the associated with the relative abundance of host plant species increasing total biomass density of butterflies was attributable might influence the relative abundance of butterflies. To exam- not to an increase in species richness but to an increase in the ine the relative contribution of these factors, additional studies densities of certain species that was caused by large increases in are required. For instance, to confirm the potential for host plant the biomass of their host plants. For instance, several species switches by butterflies, the actual resource selection by butter- (e.g., Graphium sarpedon nipponum, Papilio machaon hip- flies should be observed; this would be an interesting topic for pocrates, Artogeia melete, and Colias erate poliographys) could be future research. found at high densities in one or more communities, and the Previous studies on butterflies and moths have demonstrated densities of these species were significantly affected by the the relationship between lepidopteran species richness and plant biomass densities of their host plants (Table 2). In the present species richness (26, 27). However, the local movements of adult study, the species richness of butterflies was not related to that butterflies might be influenced by nectar resources. Indeed, a of their host plants, and this might be because the number of study on butterfly communities in open successional fields butterfly species did not differ greatly among the communities. demonstrated that butterfly species richness was explained by Therefore, in the present communities, species richness was not flower abundance (28). However, butterflies oviposit on specific useful for comparing communities; however, the relative abun- host plants and their larvae depend on these plants. Hence, if dance of species was an important aspect characterizing the adult butterflies do not frequently disperse from the natal island biodiversity of butterfly communities. The relative amount of or community, the abundance of adult butterflies will be related resources was a major factor that explained the biodiversity to that of the juveniles that depend on the host plant abundance pattern of the butterfly communities. on that island. In this study, the locations of the sampling points The neutral theory predicts that the compositional similarity were selected at random with regard to the locations of host between butterfly communities will decrease as the distance plants and nectar plants. Thus, the estimated densities of but- between two communities increases because dispersal is a pre- terflies are considered to reflect the butterflies indigenous to dominant factor determining community composition. Butter- each island. flies may be able to disperse to distant locations; however, the Most studies on plant–herbivore communities have focused distances between the studied communities were not very large. on species richness and species composition as representative A recent study showed that the similarity between communities components of biodiversity; however, few studies have treated begins to decrease at a distance of several hundred kilometers the effect and consequences of the relative abundance of species. (25). Hence, there may have been an adequate number of Although species richness and species composition are signifi- migrants between communities regardless of distance. However, cantly correlated to ecosystem functioning (29), differences in at present, there is no available data that demonstrates the relative species abundance can affect ecosystem functioning. For relationship between the number of migrants and the distance instance, the portfolio effects of ecosystem stability could be between the studied communities. The distance effects should be considerably influenced by relative species abundance. Even if examined to confirm that there might be no distance effects in species richness does not change, large increase in particular the present communities. As expected, our studies demonstrated species might decrease stability by the portfolio effects (29). In that geographic distance did not have a significant effect on the present study, increasing the biomass of host plants pro- butterfly diversity. In addition to the distance effect, the simi- moted an increase in the densities of certain butterfly species, larity between communities within islands might be larger than thus causing a change in relative species abundance. Thus, the that between communities on different islands if migrations present studies indicated that vegetation changes caused by occur more frequently within islands. However, there was no differences in land use would largely affect butterfly community evidence to suggest that communities within islands were more stability by changing the abundance of particular species. similar than those on different islands. Therefore, the present There are several hypotheses explaining the relative abun- results do not support the neutral model at least for the studied dance of species and diversity [reviewed by Tokeshi (16)]. spatial scale (Ͻ6 km). Previous studies designed to test the neutral and niche theories In this study, one of the major factors that might have have examined the relationships between community composi- contributed to the significant effect of the relative abundance of tions and environmental gradients, and have provided no direct host plant species on the relative abundance of butterfly species evidence for the effects of relative resource abundance on is the positive relationship between the abundance of each relative species abundance patterns. These results are field butterfly species and the biomass of its host plant species. For evidence of the niche-apportionment model that proposes that Ϸ60% of 39 butterfly species, the biomass densities of the host the relative amount of niche space explains the pattern of plants were observed to have positive effects on the densities of relative species abundance at the consumer level. The results the butterfly species that used them, and for Ϸ30% of the clearly demonstrate that butterfly communities depend on avail- species, the effects were significant. These positive effects of the able resource abundance. This implies that the conservation of ECOLOGY host plant biomass found in many (but not all) of the butterfly butterfly diversity and that of plant diversity should consider not species indicate a significant relationship between the relative only the maintenance of species richness but also prevent the biomass of host plants and relative abundance of butterflies. Two prevalence of a few dominant species. additional factors may have contributed to the results, particu- larly in the case of those species that did not exhibit high positive Materials and Methods relations between butterfly densities and their host plant bio- Study Region and Butterfly Sampling. The study was conducted on mass. First, some species alter their host plant preference the Urato Islands in Matsushima Bay, Japan (Fig. 1). The depending on the relative abundance of potential host plant diversity of butterfly species and their host plants was investi- species and the abundance of other (competitor) butterfly gated on five islands: Mahanashi, Katsura, Sabusawa, Nono, and species that use the same host plants. For instance, weakly Hoh. The respective areas of these islands are 0.15, 0.76, 1.45,
Yamamoto et al. PNAS ͉ June 19, 2007 ͉ vol. 104 ͉ no. 25 ͉ 10527 Downloaded by guest on September 25, 2021 0.56, and 0.15 km2. The locations of 4, 12, 22, 9, and 4 sampling host plants recorded in the literature were not identified to the points were randomly determined on Mahanashi, Katsura, Sabu- species level. In the case of Parnara guttata, known as a pest of rice, sawa, Nono, and Hoh, respectively, such that the number of all of the plants belonging to the Poaceae and Cyperaceae were sampling points per area was approximately the same (0.04–0.06 considered as host plants. This is because a relatively wide range of km2). A census of butterflies at all of the sampling locations was host plants has been recorded for this butterfly, although food plant conducted three times, in August 2004 and in July and August preference at the species level remains undetermined (30). 2005. To minimize the effect of the seasonal change in butterfly Vegetation surveys were carried out by using the following appearance, the census was conducted only in August and July. procedures. We divided the vegetation and land use into the The number of individuals of each species observed and cap- following 12 types according to the results of preliminary field tured by sweep nets within 5 m either side of a 10-m line transect surveys and an aerial digital photograph (one pixel ϭ 10 cm ϫ 10 was recorded by two investigators during 12 min when weather cm) taken in July 2003: (1) residential areas, roads, and other conditions were suitable for butterfly activity. man-made structures; (2) rice paddy fields; (3) other crop fields; (4) The analysis of the diversity and abundance of host plants and grasslands with vegetation Ͻ50-cm high, usually dominated by butterfly species was conducted mainly at eight communities annual grasses such as Trifolium spp.; (5) grasslands dominated by because these islands support different vegetation types (Fig. 1). Miscanthus sinensis and other perennial tall grasses; (6) wetland At the five-community scale, each community corresponded to vegetation mainly dominated by Phragmites communis; (7) bamboo a single island, and the results for the 5 communities were shown or dwarf bamboo thickets; (8) Cryptomeria japonica afforested in SI Table 3 and Fig. 5. At the eight-community scale, two and areas; (9) Pinus densiflora (P. thunbergii, in part) forests; (10) three communities were classified on Katsura Island and Sabu- sawa Island, respectively (Fig. 1). Sampling points 6, and 7, and Machilus thunbergii evergreen forests; (11) deciduous forests mainly 8, were included on Katura Island and Sabusawa Island, respec- dominated by Quercus serrata; and (12) mixed secondary forest tively. In the east of Katsura Island, a large area was occupied by dominated by Juglans mandshurica var. sachalinensis, Celtis sinensis deciduous forests and this was unique compared with the other var. japonica, and Neolitsea sericea (Fig. 1). We then established ϫ areas on the island. On Nono Island, the northeast area includes 8–26 quadrants of 10 m 10 m for each vegetation or land-use type. vegetation types that are somewhat different than those found in Measurements were taken of the number of individuals of each other areas of the island; however, this area is a narrow peninsula plant species in a quadrant, the height of each tree individual, and and only one sampling point was included. Owing to the position the degree of cover by herbaceous plants. From these data, the of the randomly chosen sampling points, it was difficult to average number of individuals per unit area for each size class of adequately divide Nono Island into two communities. Thus, each tree species and the average degree of cover per unit area for Nono Island was not divided for the eight-community analysis. each herbaceous plant were calculated for each vegetation or To remove the effects of different sampling effort on the land-use type. In this study, we assumed that the leaves of each plant different islands, 16 samplings were conducted on all of the species were the edible parts for the larvae of butterflies and islands, with the exception of Sabusawa where sampling was estimated the wet weights of leaves per individual for tree species conducted once at all 22 sampling points. For instance, on and those per unit area for herbaceous plants. For each plant Katsura, the sampling was conducted at all 12 sampling points species, we measured the wet weights of leaves and counted the and 4 random sampling points selected from the original 12. For number of leaves per individual or per unit area. We then estimated the 8-community analysis, sampling was conducted twice on the unit biomass of the edible parts based on the average weights Mahanashi and Hoh islands in each census. Six to nine samplings per leaves and the average number of leaves per individual or unit were conducted in the communities. The numbers of individuals area. The available resource biomass of a given plant species in each of each species per sampling point (100 m2) per sampling time local community was estimated from the unit biomass and the area were used as the butterfly densities for the analysis. of vegetation and land-use types in which the plant species was Throughout the sampling period, a total of 43 butterfly species distributed. Hence, in the analysis, the biomass of a given plant were identified. Parantica sita niphonica and Dichorragia nesi- species was used as biomass density (unit biomass per unit area). machus were omitted from the analysis because P. sita niphonica does not reproduce on these islands, and the host plant of D. Analysis of Butterfly and Host Plant Diversity. The relationships nesimachus could not be found. In addition, in the field, it was among the species richness of butterflies and host plants, the difficult to distinguish Limentis glorifira and Neope gosch- total biomass densities of butterflies and host plants, and the kevitschii from Limenitis camilla japonica and Neope niphonica, total biomass densities of butterfly species and species richness L. glorifira N. gosch- respectively. Therefore, the data for and of butterflies were examined by using a linear regression anal- kevitschii were combined with that of L. camilla japonica and N. ysis. The total biomass densities of butterflies were calculated as niphonica, respectively. Thus, in total, 39 butterfly species were ⌺ d w , where d is the densities of species i, and w is the average used in the study. i i i i weights of adult butterflies of species i. The total biomass densities of the host plants were calculated as the sum of the Estimation of Host Plant Biomass. Investigations of the distribution of resources for butterflies on the Urato Islands were conducted from biomass densities of all of the 58 host plant species. 2003 to 2005. In this study, we defined resources for butterflies as We tested the effects of resource and geographic distance on host plants for the juvenile stages; resource distribution was esti- butterfly biodiversity by applying an extension of the Mantel test for mated by using the following procedure. The host plants used by estimating the regression coefficients of independent distance each species of butterfly were selected by comparing the floral matrices (23, 24). We considered a multiple regression equation of ϭ ϩ ϩ composition based on preliminary field investigations and the host the form: aij 0 AB.Cbij AC.Bcij, where bij is the dissimilarity plant records listed in Fukuda et al. (30). Species belonging to the matrix of the relative resource abundance between communities i Satyridae and Hesperiidae (with the exception of Daimio tethys) and j (Resource matrix), cij is the geographic distance (Distance are, however, largely polyphagous, and few accurate records of host matrix, both linear and log-transformed were considered), and aij is plant species are available. In these cases, only the plant species on the dissimilarity matrix of butterfly species diversity (Butterfly which the larvae of a given butterfly were frequently recorded were matrix). AB.C and AC.B are the regression coefficients, and the classified as host plants. For butterflies that subsist on bambusoid significances were tested by randomization. The geographic dis- plants, their host plants were defined as all of the bambusoid plant tances were measured between the centers of gravity of two species distributed in the study area. This was necessary because the communities. To measure the similarity of butterfly relative abun-
10528 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0701583104 Yamamoto et al. Downloaded by guest on September 25, 2021 dance and host plant resource distribution, we used Odum’s per- Effects of Host Plant Biomass on Butterfly Density. The effects of the centage difference method (31). The dissimilarity between com- biomass of the host plant on the density of each butterfly species munities 1 and 2 is represented as follows: were examined in eight communities. We used a linear regression in which the density of a butterfly species was a dependent variable, p and the biomass density of its potential host plant species (SI Fig. ͉x Ϫ x ͉ 3) was used as an independent variable. If there was a high 1i 2i correlation coefficient between the biomass densities of the host iϭ1 D ϭ , plants, one of either host plant species was omitted in the analysis. p We explored the set of predictors of the host plant species with a ͑ ϩ ͒ x1i x2i stepwise AIC procedure; independent variables were removed and iϭ1 added to the models to determine the set of predictors that yielded the lowest AIC by using the R software. Given the model with the where x1i and x2i are the numbers of individuals of species i in lowest AIC, we obtained the significance of the regression coeffi- community 1 and community 2, respectively. For butterfly cients. We conducted an analysis of 39 butterfly species, and thus, species, the average numbers of individuals per sampling time to control the errors of multiple comparison, we used the false per sampling point were used as x. When resource similarity was discovery rate (FDR) procedure (33). considered, x1i and x2i are the biomasses of host plant species i in community 1 and community 2, respectively. The biomass was We thank I. Hanski, R. Butlin, T. Fukami, J. Urabe, A. Mizuno, T. Yoshida, J. Bridle, M. Tokeshi, A. Satake, and S. Chiba for valuable D divided by the area of the community. values increase with comments on the manuscript and K. Saito, Y. Nukatsuka, and other decreasing similarity. The number of individuals and the biomass member of the laboratory of Evolutionary Biology for field assistance. of host plant species were log transformed because the evalua- M.K. was supported by Grant-in-Aid for Scientific Research 15370007 tion of dominant and rare species were equivalent (32). from the Japan Society for the Promotion of Science.
1. Hubbell SP (2001) The Unified Neutral Theory of Biodiversity and Biogeography 17. Gilbert B, Lechowicz MJ (2004) Proc Natl Acad Sci USA 101:7651–7656. (Princeton Univ Press, Princeton). 18. Bell G (2005) Ecology 86:1757–1770. 2. Preston FW (1948) Ecology 29:254–283. 19. Dornelas M, Connolly SR, Hughes TP (2006) Nature 440:80–82. 3. Lewinsohn TM, Novotny V, Basset Y (2005) Annu Rev Ecol System 36:597–620. 20. Graves GR, Rahbek C (2005) Proc Natl Acad Sci USA 102:7871–7876. 4. Andow DA (1991) Annu Rev Entomol 36:561–586. 21. McGill BJ (2003) Nature 422:881–885. 5. Murdoch MW, Evans FC, Peterson CH (1972) Ecology 53:819–829. 22. Tuomisto H, Ruokolainen K, Yli-Halla M (2003) Science 299:241–244. 6. Novotny V, Drozd P, Miller SE, Kulfan M, Janda M, Basset Y, Weiblen GD 23. Manly BFJ (1991) Randomization and Monte Carlo Methods in Biology (Chap- (2006) Science 313:1115–1118. man & Hall, London). 7. Armbrecht I, Perfecto I, Vandermeer J (2004) Science 304:284–286. 24. Manly BFJ (1986) Res Pop Ecol 28:201–218. 8. Haddad NM, Tilman D, Haarstad J, Ritchie M, Knops JMH (2001) Am Nat 25. Soininen J, McDonald R, Hillebrand H (2007) Ecography 30:3–12. 158:17–35. 26. Kitahara M, Watanabe M (2003) Ecological Research 18:503–522. 9. Siemann E, Tilman D, Haarstad J, Ritchie M (1998) Am Nat 152:738–750. 27. Usher MB, Keiller SW (1998) Biodiversity Conservation 7:725–748. 10. Hunter MD (1992) in Effects of Resource Distribution on Animal–Plant Interactions, 28. Steffan-Dewenter I, Tscharntke T (1997) Oecologia 109:294–302. eds Humter MD, Ohgushi T, Price PW (Academic, New York), pp 287–325. 29. Hooper DU, Chapin I, Ewel JJ, Hector A, Inchausti P, Lavorel S, Lawton JH, 11. Kelly CK, Southwood TRE (1999) Proc Natl Acad Sci USA 96:8013–8016. Lodge DM, Loreau M, Naeem S, et al. (2005) Ecol Monogr 75:3–35. 12. Marques ESDA, Price PW, Cobb NS (2000) Environ Entomol 29:696–703. 30. Fukuda H, Haba E, Kuzuya K, Takahasi A, Takahasi M, Tanaka B, Tanaka H, 13. Ohgushi T (1992) in Effects of Resource Distribution on Animal–Plant Interactions, Wakabayashi M, Watanabe Y (1982) The Life Histories of Butterflies in Japan eds Hunter D, Ohgushi T, Price PW (Academic, New York.), pp 287–325. (Hoikusha, Osaka, Japan), Vol 1–4 (in Japanese). 14. Potts SG, Vulliamy B, Dafni A, Neåfeman G, Willmer P (2003) Ecology 31. Quinn GP, Keough MJ (2002) Experimental Design and Data Analysis for 84:2628–2642. Biologists (Cambridge Univ Press, Cambridge). 15. Schoener TW (1974) Science 185:27–39. 32. Legendre P, Legendre L (1998) Numerical Ecology (Elsevier Scientific Pubish- 16. Tokeshi M (1999) Species Coexistence: Ecological and Evolutionary Perspectives ing, Amsterdam). (Blackwell, Oxford). 33. Benjamini Y, Hochberg Y (1995) J R Stat Soc Ser B 57:289–300. ECOLOGY
Yamamoto et al. PNAS ͉ June 19, 2007 ͉ vol. 104 ͉ no. 25 ͉ 10529 Downloaded by guest on September 25, 2021