Biodiversity Data Journal :
Research Article
|
Corresponding author: Shoko Nakamura (nakamuras1220@affrc.go.jp)
Academic editor: Paolo Biella
Received: 24 Jan 2023 | Accepted: 26 Jun 2023 | Published: 08 Sep 2023
© 2023 Shoko Nakamura, Hisatomo Taki, Tomonori Arai, Ken Funayama, Shunsuke Furihata, Yuki Furui, Takamasa Ikeda, Hiromitsu Inoue, Kiyohiko Kagawa, Hidenari Kishimoto, Mitsuko Kohyama, Michiyo Komatsu, Akihiro Konuma, Ken Nakada, Suguru Nakamura, Nobuo Sawamura, Shoji Sonoda, Masahiro Sueyoshi, Seishi Toda, Katsuhiko Yaginuma, Shunsuke Yamamoto, Koki Yoshida, Tomoyuki Yokoi, Masatoshi Toyama
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Nakamura S, Taki H, Arai T, Funayama K, Furihata S, Furui Y, Ikeda T, Inoue H, Kagawa K, Kishimoto H, Kohyama M, Komatsu M, Konuma A, Nakada K, Nakamura S, Sawamura N, Sonoda S, Sueyoshi M, Toda S, Yaginuma K, Yamamoto S, Yoshida K, Yokoi T, Toyama M (2023) Diversity and composition of flower-visiting insects and related factors in three fruit tree species. Biodiversity Data Journal 11: e100955. https://doi.org/10.3897/BDJ.11.e100955
|
|
Animal-mediated pollination is an essential ecosystem service for the production of many fruit trees. To reveal the community composition of flower-visiting wild insects which potentially contribute to fruit production and to examine the effects of geographic location, local meteorological conditions and locally introduced domesticated pollinators on them, we investigated the community composition of insects visiting the flowers (hereafter, “visitors”) of apple, Japanese pear and Oriental persimmon for 1‒3 years at 20 sites around Japan. While most of the variation (82%) of the community composition was explained by tree species with a slight contribution by geographic distance (2%), maximum temperature and tree species contributed 62% and 41% of the variation in total abundance of the visitors, respectively. Though the dominant families of the visitors varied spatiotemporally, the community composition of the visitors of apple and Japanese pear clearly differed from that of Oriental persimmon. While Andrenidae and Syrphidae together accounted for 46%‒64% of the visitors of apple and Japanese pear, Apidae represented 57% of the visitors of Oriental persimmon. The taxonomic richness, diversity and evenness of the visitors were best predicted by locally introduced domesticated pollinators and local meteorological conditions of wind speed and maximum temperature. Amongst these selected factors, locally introduced domesticated pollinators could have the largest impact. It seemed to be strongly related to the reduction of taxonomic richness, diversity and evenness of the visitors, accounting for 41‒89% of the variation. Results suggested that the community composition and total abundance of potential pollinators were predominantly determined by tree species and temperature, but locally introduced domesticated pollinators could have a determinantal pressure on the taxonomic diversity of the community.
insect pollination, agriculture, Apis mellifera, meteorological conditions, bees, Diptera, Coleoptera
Pollination is an important ecosystem service provided by flower-visiting animals (
Recent studies have emphasised the importance of complementary pollination by diverse insect pollinators (
The introduction of domesticated pollinators, Apis mellifera or Osmia cornifrons began in apple orchards in Japan in the 1950s, following decreased fruiting success likely from the shortage of wild insect pollinators (
We aimed to answer the following questions to understand potential pollinators of major entomophilous fruit trees in Japan, namely, apple, Japanese pear and Oriental persimmon: 1) How does the community composition of flower-visiting insects (hereafter, “visitors”) differ amongst the fruit tree species? 2) How do local meteorological conditions and the introduction of domesticated pollinators influence the composition and abundance of the communities of visitors and how large are these effects? 3) How does each taxonomic group of visitors respond to these factors?
We selected apple (Malus pumila Mill.), Japanese pear (Pyrus pyrifolia (Burm.f.) Nakai) and Oriental persimmon (Diospyros kaki Thunb.) as targets for monitoring visitors. The apple and Japanese pear produce hermaphrodite, but self-incompatible flowers and the Oriental persimmon produces monoecious entomophilous flowers. The fruits are relatively familiar to Japanese people, with the top three production volumes of 763,300 (33% of the total production of fruit trees in Japan), 170,500 (7%) and 193,200 (8%) tonnes for apple, Japanese pear and Oriental persimmon, respectively, amongst the entomophilous fruits (
The surveys were conducted in orchards at 20 sites distributed in nine prefectures in Japan. The sites were within the Japanese agricultural landscape, with apples being grown at six sites (MOC, YKYEX1, YKYEX2, MN1, SN1, HKC), Japanese pear being grown at nine sites (UTC, MKC, TKC, TEO, TED, TEY, TES, KKU, KKA) and Oriental persimmon being grown at five sites (IZC, IZE, ODC, HDC, AKC) (Table S1: Suppl. material
Monitoring surveys were conducted on the days of full bloom (late March to late May) without rain at each site for one to three years, 2017–2019, covering morning to afternoon (Table S1: Suppl. material
We preserved the visitors captured during the surveys in 99.9% ethanol or 98% propylene glycol. The major flower visitors, namely Hymenoptera, Diptera and Coleoptera, were morphologically identified to family level and the others were identified to order level (hereafter, “taxon” for these identification units) using stereoscopic binoculars (Nikon SMZ800, Nikon Corporation, Tokyo, Japan). The top three taxa (bees: Andrenidae and Apidae; hoverflies: Syrphidae) and the most abundant coleopteran family, Scarabaeidae, were identified to the genus level. We did not identify visitors to the species level because functional diversity often explains ecosystem functioning including pollination better than species diversity (
We extracted daily-local meteorological data of maximum and minimum temperature, precipitation, sun duration and average wind speed from the nearest weather station (Table S3: Suppl. material
We conducted the following analyses to explore the factors related to the composition and abundance of the visitor communities and the abundance of each taxon. All statistical analyses were conducted in R version 4.1.0 (
The abundance records, based on family (Hymenoptera, Diptera and Coleoptera) or order (other taxa) identification of the visitors, were pooled for each site–year combination to construct the visitor communities. After excluding small communities (n < 50) and sites where the survey was conducted on only one day, we removed the taxa with relatively rare occurrences. These were taxa represented by only one individual across all the sites and years sampled. These reductions were to remove possible sampling bias and to compare communities, based on the taxa that potentially contribute to pollination. Rare taxa would have quite less contribution to pollination due to their low abundances. We then performed abundance-based rarefaction, rarefying each community to 51 individuals and used the data for the subsequent community analysis. According to the rarefaction analysis, these communities encompassed 83%‒99% of the taxa richness expected in these communities (Fig. S1: Suppl. material
We calculated the Bray–Curtis distances, based on the log-transformed community data and performed non-metric multi-dimensional scaling (nMDS) to visualise the differences amongst the communities. Log-transformation allows us to reduce the overappreciated large effect by dominant taxa (
The contribution of each variable to the overall variation of the response variable explained by the best model, which was interpreted as the proportion reflecting the relative importance of the variable in the model, was calculated using deviance partitioning. The exclusive contributions from each explanatory variable in the best model were calculated as:
{(deviance of the model in which the explanatory variable of interest was removed from the best model) - (deviance of the best model)} / {(deviance of the null model) - (deviance of the best model)}.
The value is not affected by the collinearity amongst other variables in the model (
Given that tree species was selected in the best model, Dunnett's multiple comparisons were carried out to test for the differences between tree species as a post hoc test. The distances between communities of different tree species were compared against those within the same tree species using the "glht" function in the R package "multcomp" (
We examined the effects of geographic and local meteorological factors, tree species, introduction of domesticated pollinators and annual fluctuations on the following diversity indices for each community: taxonomic richness, taxonomic diversity (Shannon diversity) and taxonomic evenness (Pielou’s evenness). We fitted generalised linear models (GLMs) for these response variables for each community and conducted model selection using AIC. The full model included the explanatory variables of mean maximum temperature, squared mean maximum temperature, average wind speed and squared average wind speed as local meteorological factors. Latitude was included as a geographical factor and tree species, the status of introduction of the domesticated pollinators and year were included as categorical variables. We did not include longitude as an explanatory variable in the full model to avoid strong multicollinearity with latitude (Fig. S2: Suppl. material
To explore the factors affecting the abundance of all visitors and each taxonomic group, we constructed generalised linear mixed effect models (GLMMs) and conducted model selection based on AIC. For these models, we assumed negative binomial error distribution and the response variables were the abundance of all the visitors (total abundance) or those of each taxonomic group recorded each day. To account for the varying sampling efforts amongst sites and days, we used the offset term of effort (min) in each day. The explanatory variables were daily maximum temperature, squared daily maximum temperature, daily average wind speed, squared daily average wind speed, latitude, tree species, year and the status of introduction of domesticated pollinators. Site was included as a random effect. For the analyses of each taxonomic group, the zero data were limited to the possible pairs of tree-insect combinations that had been detected at least once throughout the three-year survey. This was to examine the effects of local meteorological conditions and status of introduction of domesticated pollinators on the abundance of visitors, while eliminating the strong effect of tree species. The numeric explanatory variables were scaled to zero mean and unit variance. In the analyses for the abundance of each taxon, families that were sufficiently abundant to appear within the accumulated proportion of 80% in either of the three tree species (hereafter, “major” families) were analysed separately. The remaining less abundant 65 taxa (hereafter, “minor” families or taxa) were analysed in one model with the taxonomic group being an additional random variable. The contribution by each explanatory variable selected was calculated using a similar procedure used for the community composition.
The data underpinning the analysis reported in this paper are deposited at GBIF, the Global Biodiversity Information Facility, https://doi.org/10.15468/he2hm9 (
In total, we identified 5,411 individuals of wild visitors from 37 communities ("community" as assemblages per site per year) across 20 sites (Table S5: Suppl. material
The dominant families of wild visitors differed amongst communities, creating spatiotemporal variation in the community composition (Table S5: Suppl. material
The most dominant genus within the major family also varied amongst sites (Table S6: Suppl. material
The rarefied data for the community and diversity analyses consisted of 26 communities, including 10 for Japanese pear, nine for apple and seven for Oriental persimmon (Table S7: Suppl. material
Numbers and proportions of visitors pooled by tree species based on the rarefied communities. Families with asterisks represent major families, in which the accumulated proportion > 80% in each tree species.
Order | Family | Japanese pear | Apple | Oriental persimmon | |||||||||
n | % | Order Total n | Order Total % | n | % | Order Total n | Order Total % | n | % | Order Total n | Order Total % | ||
Hymenoptera | Andrenidae | 124 | 24.3 * | 159 | 31.2 | 159 | 34.6 * | 222 | 48.4 | 11 | 3.1 | 248 | 69.5 |
Apidae | 0 | 0.0 | 27 | 5.9 * | 205 | 57.4 * | |||||||
Halictidae | 16 | 3.1 | 29 | 6.3 * | 26 | 7.3 * | |||||||
Tenthredinidae | 18 | 3.5 * | 0 | 0.0 | 0 | 0.0 | |||||||
Icheumonidae | 0 | 0.0 | 4 | 0.9 | 1 | 0.3 | |||||||
Megachilidae | 0 | 0.0 | 0 | 0.0 | 3 | 0.8 | |||||||
Bethylidae | 1 | 0.2 | 1 | 0.2 | 0 | 0.0 | |||||||
Blasticotomidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Colletidae | 0 | 0.0 | 0 | 0.0 | 1 | 0.3 | |||||||
Figitidae | 0 | 0.0 | 0 | 0.0 | 1 | 0.3 | |||||||
Scoliidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Diptera | Syrphidae | 109 | 21.4 * | 309 | 60.6 | 135 | 29.4 * | 198 | 43.1 | 14 | 3.9 | 26 | 7.3 |
Empididae | 95 | 18.6 * | 23 | 5.0 * | 0 | 0.0 | |||||||
Anthomyiidae | 52 | 10.2 * | 3 | 0.7 | 1 | 0.3 | |||||||
Chironomidae | 12 | 2.4 | 16 | 3.5 | 1 | 0.3 | |||||||
Ephydridae | 12 | 2.4 | 1 | 0.2 | 2 | 0.6 | |||||||
Bibionidae | 7 | 1.4 | 7 | 1.5 | 0 | 0.0 | |||||||
Calliphoridae | 6 | 1.2 | 1 | 0.2 | 0 | 0.0 | |||||||
Muscidae | 3 | 0.6 | 2 | 0.4 | 2 | 0.6 | |||||||
Sciaridae | 4 | 0.8 | 1 | 0.2 | 1 | 0.3 | |||||||
Bombyliidae | 0 | 0.0 | 4 | 0.9 | 0 | 0.0 | |||||||
Conopidae | 1 | 0.2 | 2 | 0.4 | 0 | 0.0 | |||||||
Tachinidae | 3 | 0.6 | 0 | 0.0 | 0 | 0.0 | |||||||
Tipulidae | 1 | 0.2 | 0 | 0.0 | 2 | 0.6 | |||||||
Scatopsidae | 0 | 0.0 | 0 | 0.0 | 2 | 0.6 | |||||||
Agromyzidae | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | |||||||
Asilidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Cylindrotomidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Fanniidae | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | |||||||
Sphaeroceridae | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | |||||||
Tabanidae | 0 | 0.0 | 0 | 0.0 | 1 | 0.3 | |||||||
Therevidae | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | |||||||
Xylophagidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Coleoptera | Scarabaeidae | 17 | 3.3 * | 41 | 8.0 | 16 | 3.5 | 37 | 8.1 | 50 | 14.0 * | 79 | 22.1 |
Elateridae | 1 | 0.2 | 0 | 0.0 | 21 | 5.9 * | |||||||
Chrysomelidae | 8 | 1.6 | 4 | 0.9 | 3 | 0.8 | |||||||
Cerambycidae | 6 | 1.2 | 0 | 0.0 | 1 | 0.3 | |||||||
Coccinellidae | 1 | 0.2 | 3 | 0.7 | 2 | 0.6 | |||||||
Melandryidae | 0 | 0.0 | 6 | 1.3 | 0 | 0.0 | |||||||
Nitidulidae | 3 | 0.6 | 3 | 0.7 | 0 | 0.0 | |||||||
Cantharidae | 2 | 0.4 | 1 | 0.2 | 2 | 0.6 | |||||||
Melyridae | 2 | 0.4 | 0 | 0.0 | 0 | 0.0 | |||||||
Pyrochroidae | 0 | 0.0 | 2 | 0.4 | 0 | 0.0 | |||||||
Curculionidae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Oedemeridae | 0 | 0.0 | 1 | 0.2 | 0 | 0.0 | |||||||
Staphylinidae | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | |||||||
Hemiptera | 0 | 0.0 | 0 | 0.0 | 2 | 0.4 | 2 | 0.4 | 4 | 1.1 | 4 | 1.1 | |
Lepidoptera | 1 | 0.2 | 1 | 0.2 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | |
Total | 510 | 100.0 | 459 | 100.0 | 357 | 100.0 |
The difference in the community composition was best predicted by the geographic distance between sites and tree species (Table
Results of model selection for the fitted generalised least squares (GLS) model for the distances between communities. The best model selected by AIC and the null model with AIC value are shown. The result of Dunnett's multiple comparison for testing differences in the community composition between different tree species and within the same tree species is also shown. Contributions of selected variables were calculated, based on the deviance partitioning. The full GLS model included Bray–Curtis distances, based on the log-transformed community data as a response variable, combination of tree species between the communities, distances separately derived from local meteorological conditions (maximum temperature and average wind speed), introduction of domesticated pollinators, geographic location and year were included as explanatory variables, with a set of Maximum-likelihood population effects covariance structure amongst sites (MLPE; (
Estimate | Std. Error | t-value | p-value | AIC | Contributions | ||
Null model | -286.23 | ||||||
Intercept |
0.580 |
0.030 | 22.10 | < 0.01 | |||
Best model | -547.47 | ||||||
Intercept (Tree species combination = Within) | 0.368 | 0.035 | 13.754 | < 0.01 | |||
log (Geographic distance) | 0.008 | 0.002 | 3.551 | < 0.01 | 2% | ||
Tree species combination | 82% | ||||||
Apple–Pear | 0.028 | 0.014 | 1.949 | 0.052 | |||
Apple–Persimmon | 0.187 | 0.018 | 10.277 | < 0.01 | |||
Pear–Persimmon | 0.309 | 0.016 | 18.894 | < 0.01 | |||
Post-hoc (Dunnett's multiple comparison) | z-value | p-value | |||||
Apple–Pear vs Within | 1.949 | 0.135 | |||||
Apple–Persimmon vs Within | 10.277 | < 0.01 | |||||
Pear–Persimmon vs Within | 18.894 | < 0.01 |
nMDS plot (a) with the names and years of the communities (information for each community is provided in Table S1: Suppl. material
Blue: Japanese pear; red: apple; green: Oriental persimmon.
The average taxonomic richness in the communities, based mostly on the family identification, was 10.0 (5‒12 taxa) in Japanese pear, 9.1 (3‒15 taxa) in apple and 6.9 (3‒14 taxa) in Oriental persimmon. The average Shannon diversity in the communities was 1.7 (0.7‒2.0) in Japanese pear, 1.5 (0.8‒2.4) in apple and 1.2 (0.4‒2.1) in Oriental persimmon (Fig.
The responses of taxonomic richness, Shannon diversity and Pielou's evenness to the introduction of domesticated pollinators, average wind speed, maximum temperature and year. Panels of responses to the non-selected variables are shaded in grey. The predicted lines are drawn, based on the predictions for the year 2018. Richness, diversity and evenness calculations were based basically on family identification (see Materials and Methods for details).
**: < 0.01, *: < 0.05, †: < 0.1
The taxonomic richness in the communities was best predicted by introduction of domesticated pollinators and linear and quadratic terms of average wind speed (Table
The results of model selection for the generalised linear models (GLMs) for the response variables of taxonomic richness, Shannon diversity and Pielou's evenness. The best models selected by AIC, with the respective null models with AIC values are shown. The contributions of the variables selected were calculated, based on deviance partitioning. The full model included one of the three diversity indices derived from the rarefied community as a response variable and mean maximum temperature, squared mean maximum temperature, average wind speed, squared average wind speed, latitude, tree species, year and the status of introduction of domesticated pollinators as explanatory variables. We assumed negative binomial error distribution for the model of taxonomic richness and gamma error distribution with log-link was assumed for diversity and evenness.
Estimate | Std. Error | t-value | p-value | AIC | Contributions | ||
Richness | Null model | 144.7 | |||||
Intercept | 2.180 | 0.081 | 26.80 | < 0.01 | |||
Best model | 138.6 | ||||||
Intercept (Domesticated bee = Unintroduced) | 2.517 | 0.127 | 19.88 | < 0.01 | |||
Domesticated bee | -0.578 | 0.167 | -3.45 | < 0.01 | 88% | ||
Wind | 0.183 | 0.085 | 2.15 | 0.032 | 44% | ||
Wind^2 | -0.153 | 0.079 | -1.94 | 0.053 | |||
Shannon diversity | Null model | 48.3 | |||||
Intercept | 0.410 | 0.070 | 5.830 | < 0.01 | |||
Best model | 40.0 | ||||||
Intercept (Domesticated bee = Unintroduced, Year = 2017) | 0.683 | 0.146 | 4.675 | < 0.01 | |||
Domesticated bee | -0.613 | 0.179 | -3.430 | < 0.01 | 50% | ||
Wind | 0.198 | 0.081 | 2.453 | 0.024 | 23% | ||
Wind^2 | -0.214 | 0.100 | -2.161 | 0.044 | |||
Temp^2 | 0.163 | 0.079 | 2.061 | 0.053 | 18% | ||
Year (2018) | 0.070 | 0.142 | 0.495 | 0.626 | 22% | ||
Year (2019) | -0.444 | 0.225 | -1.978 | 0.063 | |||
Pielou's evenness | Null model | -22.0 | |||||
Intercept | -0.348 | 0.038 | -9.146 | < 0.01 | |||
Best model | -32.7 | ||||||
Intercept (Domesticated bee = Unintroduced, Year = 2017) | -0.190 | 0.078 | -2.449 | 0.024 | |||
Domesticated bee | -0.312 | 0.095 | -3.272 | < 0.01 | 41% | ||
Wind | 0.092 | 0.043 | 2.149 | 0.045 | 23% | ||
Wind^2 | -0.122 | 0.053 | -2.307 | 0.032 | |||
Temp^2 | 0.111 | 0.042 | 2.627 | 0.017 | 27% | ||
Year (2018) | 0.015 | 0.075 | 0.202 | 0.842 | 32% | ||
Year (2019) | -0.310 | 0.120 | -2.587 | 0.018 |
Shannon diversity in the communities was best predicted by introduction of domesticated pollinators, linear and quadratic terms of average wind speed, quadratic term of maximum temperature and year (Table
Pielou's evenness in the communities was best predicted by similar explanatory variables to those in the best model for Shannon diversity (Table
Total abundance of visitors was best predicted by tree species and the quadratic term of maximum temperature (Table
The results for the generalised linear mixed effect models (GLMMs) for total abundance of visitors, abundances of major families and other minor taxa. The best models selected by AIC, with the respective AIC value of null models are shown. The figures are the estimates in the models and those in parentheses represent p-values. The full model included the response variable of total abundance of visitors or the abundance of each taxon for each day, with the offset term of sampling effort. The explanatory variables were the daily maximum temperature, squared daily maximum temperature, daily average wind speed, squared daily average wind speed, latitude, tree species, year and the status of introduction of domesticated pollinators and the random variable was site. We assumed a negative binomial error distribution for these models. For the model on the minor taxa, the random variable of taxonomic group was also included.
Intercept | Plant | Domestic bee | Max temp | Max temp^2 | Wind | Wind^2 | Year | AIC | n | ||
Total | Null | -2.211 | 1068.7 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -1.584 | Pear: -1.081 | -0.241 | 1063.7 | 111 | ||||||
(< 0.01) | (0.560) | (< 0.01) | |||||||||
Persimmon: -0.262 | |||||||||||
(0.026) | |||||||||||
Contributions | 41% | 62% | |||||||||
Andrenidae | Null | -4.544 | 648.4 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -2.944 | Pear: -0.453 | 0.502 | -0.333 | 623.7 | 111 | |||||
(< 0.01) | (0.463) | (< 0.01) | (0.024) | ||||||||
Persimmon: -4.262 | |||||||||||
(< 0.01) | |||||||||||
Contributions | 62% | 50% | |||||||||
Syrphidae | Null | -4.573 | 632.7 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -3.463 | Pear: 0.043 | 599.8 | 111 | |||||||
(< 0.01) | (0.850) | ||||||||||
Persimmon: -3.39 | |||||||||||
(< 0.01) | |||||||||||
Contributions | 100% | ||||||||||
Apidae | Null | -6.186 | 415.6 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -4.755 | Pear: -3.094 | 0.880 | -0.812 | -0.357 | 388.6 | 111 | ||||
(< 0.01) | (< 0.01) | (< 0.01) | (< 0.01) | (0.090) | |||||||
Persimmon: 0.202 | |||||||||||
(0.849) | |||||||||||
Contributions | 26% | 66% | 8% | ||||||||
Empididae | Null | -6.23 | 341.3 | 83 | |||||||
(< 0.01) | |||||||||||
Best | -4.661 | -16.862 | 2018: 0.812 | 316.9 | 83 | ||||||
(< 0.01) | (0.933) | (< 0.01) | |||||||||
2019: -0.162 | |||||||||||
(0.806) | |||||||||||
Contributions | 66% | 34% | |||||||||
Scarabaeidae | Null | -5.913 | 394.9 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -5.214 | 0.673 | -0.675 | -0.804 | -0.851 | 373.6 | 111 | ||||
(< 0.01) | (0.045) | (0.020) | (0.021) | (< 0.01) | |||||||
Contributions | 37% | 50% | |||||||||
Halictidae | Null | -5.335 | 446.5 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -3.814 | Pear: -1.079 | -1.338 | 0.568 | -0.385 | -0.311 | 429.9 | 111 | |||
(< 0.01) | (0.118) | (0.038) | (<0.01) | (0.013) | (0.050) | ||||||
Persimmon: -1.746 | |||||||||||
(< 0.01) | |||||||||||
Contributions | 21% | 12% | 64% | 14% | |||||||
Anthomyiidae | Null | -5.985 | 255.1 | 83 | |||||||
(< 0.01) | |||||||||||
Best | -3.987 | -3.137 | -0.500 | 2018: -1.265 | 246.1 | 83 | |||||
(< 0.01) | (< 0.01) | (0.134) | (0.060) | ||||||||
2019: 0.368 | |||||||||||
(0.680) | |||||||||||
Contributions | 42% | 15% | 29% | ||||||||
Elateridae | Null | -8.95 | 127.9 | 111 | |||||||
(< 0.01) | |||||||||||
Best | -8.765 | Pear: -0.597 | -2.435 | 0.371 | 0.559 | 2018: 0.560 | 119.5 | 111 | |||
(< 0.01) | (0.679) | (0.036) | (0.031) | (0.144) | (0.397) | ||||||
Persimmon: 3.148 | 2019: -2.057 | ||||||||||
(0.022) | (0.641) | ||||||||||
Contributions | 40% | 18% | 15% | 9% | 43% | ||||||
Tenthredinidae | Null | -6.55 | 187.0 | 83 | |||||||
(< 0.01) | |||||||||||
Best | -5.588 | -1.822 | -0.478 | 183.2 | 83 | ||||||
(< 0.01) | (0.039) | (0.084) | |||||||||
Contributions | 50% | 47% | |||||||||
Minor taxa | Null | -8.99 | 3171.7 | 4323 | |||||||
(< 0.01) | |||||||||||
Best | -8.077 | -1.51 | -0.188 | -0.304 | -0.158 | 3148.7 | 4323 | ||||
(< 0.01) | (0.040) | (0.090) | (< 0.01) | (< 0.01) | |||||||
Contributions | 12% | 57% | 27% |
Based on the accumulated abundance (>80%), nine families, namely Andrenidae, Halictidae, Apidae, Tenthredinidae, Syrphidae, Empididae, Anthomyiidae, Scarabaeidae and Elateridae were determined to be major families and separately analysed. Each taxonomic group responded differently to the factors of tree species, local meteorological conditions and year (Table
Either of the factors related to local meteorological conditions, namely maximum temperature and average wind speed were related to the abundance, except for Syrphidae and Empididae (Table
Introduction of domesticated pollinators had a negative or no clear effect, but never had a positive effect on the abundance of each taxon (Fig. S3: Suppl. material
According to the deviance partitioning, the most contributing factor to the variation in the abundance of each taxon differed between taxonomic groups. Tree species contributed best to Andrenidae and Syrphidae; introduction of domesticated pollinators contributed best to Empididae, Anthomyiiade and Tenthredinidae; maximum temperature contributed best to Apidae, Halictidae and the minor taxa; average wind speed did best for Scarabaeidae; and year contributed best for Elatridae (Table
We are the first to describe and compare the composition and abundance of visitor communities for three fruit tree species around Japan. The community composition, total abundance and taxonomic diversity of visitors varied amongst communities. This would have resulted from the varying responses of the taxonomic groups of the visitors to tree species, local meteorological conditions, local introduction of domesticated pollinators and/or other annually fluctuating factors.
The community composition and total abundance were best explained by tree species and geographic distance between sites or daily maximum temperature. Though we must note that our sampling sites were geographically clustered by tree species, this result might imply that these factors could have considerable effects on the characteristics of the visitor communities, compared with the other variables examined including introduction of domesticated pollinators.
The visitor communities significantly differed between the Rosaceae species (apple and Japanese pear) and Oriental persimmon (Tables
This clear difference between the visitor communities of the two Rosaceae species and Oriental persimmon may have partially resulted from the responses of insects to different floral traits, such as shape and floral orientation (Rosaceae species: disc-shaped and multi-directional depending on the positions within inflorescences; Oriental persimmon: bell-shaped and downward), floral colour (Rosaceae species: white to pale pink; Oriental persimmon: greenish-pale yellow) or reward (Rosaceae species: hermaphrodite flowers provide both nectar and pollen; Oriental persimmon: monoecious flowers either of nectar or pollen). Phenological compatibility between the flowering and the active season of the visitors might also have played a key role (
Geographic distance explained a relatively small proportion (2%) of the variation in the model for community composition, compared to tree species (82%; Table
Taxonomic richness, diversity and evenness in the communities were not related to tree species, but to introduction of domesticated pollinators, average wind speed, maximum temperature and/or year (Table
Introduction of domesticated pollinators (mainly A. mellifera) was strongly related to the decrease in the taxonomic richness, diversity and evenness in the communities (41%‒88% of the variations explained; Table
Our analyses have shown that the composition of potential pollinator communities is largely determined by fruit tree species. This implies that preliminary conservation measures may be constructed, based on the knowledge of visitors to focal fruit tree species. However, given that our results have also suggested a spatiotemporal turnover of dominant taxa, conservation measures that are too specific, based on temporally and spatially limited observations, may possess a high risk of failure in the context of crop production. We have also highlighted the different levels of vulnerability to increasing temperature in terms of pollination services provided by wild visitors amongst crop species and potential harm from domesticated pollinators to the diversity of visitors which may be related to the stability of pollination services.
We are grateful to the local farmers who kindly allowed us to visit and conduct insect surveys in their fields. We thank Masayoshi K. Hiraiwa, Tsunashi Kamo, Shigeki Kishi, Taro Maeda, Koji Mishiro, Aoi Nikkeshi, Satoshi Toda, Yoshinori Tokuoka and Nami Uechi for their support in the field study. This work was supported by the Japanese Ministry of Agriculture, Forestry and Fisheries through the research project “Monitoring and enhancement of pollinators for crop production (JPJ006239)” and JSPS KAKENHI Grant Number JP21K14868.
Monitoring and enhancement of pollinators for crop production (JPJ006239)
JSPS KAKENHI Grant Number JP21K14868
Locations, sampling methods, sampling efforts and status of introduction of domesticated pollinators at each site and year. Sampling efforts were converted to the minutes of the survey per person in the column "Effort".
Normal weather conditions at the nearest weather station as the 30-year (1991–2020) average for every 10 days during antheses of Japanese pear, apple and Oriental persimmon from March to May. Figures with @ represent statistics that are not based on data for the 30 years, due to termination of meteorological observations.
Locations of the nearest weather stations and the distances to the sites.
Meteorological conditions of the survey days at each site. Data were extracted from the nearest weather station (see Table S3).
The numbers of visitors in each taxa collected from each site and year. Light grey cells represent undetected taxa and orange-shaded cells represent the most dominant taxa within the community. Dark grey-shaded lines were communities with less than 50 individuals. Apis mellifera and Osmia cornifrons were only collected in the first year (2017).
Genera detected in the families Andrenidae, Apidae, Syrphidae and Scarabaeidae and their proportions in the site-pooled communities. Grey cells represent undetected genera and orange, yellow, green or blue cells represent the most abundant genus in the families of Andrenidae, Apidae, Syrphidae or Scarabaeidae, respectively.
Rarefied community composition, based on family or order identification. This data were used for the community analysis. Grey cells represent undetected taxa and orange-shaded cells represent the most dominant taxa within the community.
Rarefaction curves for the communities of visitors. The curves are drawn after excluding small communities (n < 50) and sites where the survey was conducted on only one day and removing the taxa with relatively rare occurrences. The vertical line represents 51 individuals, to which we rarefied these communities for the community analyses. The rarefied communities encompassed 83%‒99% of the taxa richness expected in these communities.
Correlation plots for the candidates of explanatory variables for meteorological factors.
The responses of visitor abundance to tree species, maximum temperature, wind speed, introduction of domesticated pollinators and year. Figures in parentheses above the boxes of the boxplots represent sample sizes. Panels drawn for non-selected explanatory variables by the model selection are shaded in grey. The yellow shades in the panels for the responses to maximum temperature correspond to the maximum temperature range for normal years (16.4℃‒22.0℃ in apple and pear sites, 23.3℃‒24.8℃ in persimmon sites).