Results
We recorded 99 pollinator species (33 bees, 14 flies, 13 lepidopterans; 18 wasps and 22 diverse small insects) and 125 plant species in our field surveys, resulting in 6615 interactions and 492 links (Fig. 2). In total, we collected 4772 floral visitors, where 4607 showed legitimate visiting behaviour and, among them, 2008 had pollen on their body surface. Therefore, the plant-centred network from Pantanal comprised 4607 interactions among 99 pollinators and 106 plant species, while the animal-centred network comprised 2008 interactions and a total of 48 pollinators and 95 plants species. Among plants, 78 species were recovered by both sampling methods, 31 species only with the plant- and 19 species only in the animal-centred networks. In relation to pollinators, 44 species occurred in both networks, 50 species occurred only in the plant-centred network and three species were unique to the animal-centred network (Centris sp. 2, Formicidae sp. 2 and Formicidae sp. 3), as these were recorded visiting flowers illegitimately, but later inspection revealed pollen on their bodies.
Beta diversity of interactions and beta diversity indices
We recorded a high beta diversity of interactions across both sampling methods and in the combined network. Regardless of the method used, network beta diversity of interactions (βWN) was mostly due to turnover (βST) in comparison to rewiring (βOS) (Fig. 3a, in the combined network βST 0.72±0.18 and βOS 0.13±0.09; T = 23.45; P <0.0001). Plants showed higher dissimilarity than pollinators between networks (β-plants: 0.83±0.14, β-pollinators: 0.67±0.12; P <0.0001, Fig. 3b). Considering the different sampling methods, the plant-centred method presented the highest values of β-plants (χ2 = 127.48, P <0.0001), while β-pollinators did not differ (χ2 = 1.89, P = 0.38; Fig. 3b; Table S3). Following the same trend, the plant-centred method returned the highest values of different interaction beta diversity measures, with the exception of interaction rewiring which was lower in comparison to other methods (Fig. 3a). In relation to species turnover (βST), plant-centred networks had the highest values, and the animal-centred network did not differ from the combined network (χ2 = 88.22, P <0.0001; Fig. 3a; Table S3). Finally, when calculating the true components of beta diversity excluding the influence of differences on richness between the networks, the results returned were similar (βOS: χ2 = 58.01, P <0.0001; βWN: χ2 = 7.46, P <0.03 Fig. 3a; Table S3).
Considering the habitats, while rewiring (βOS) did not differ within and between habitats (χ2 = 1.32, P = 0.07, Fig. 4a), turnover (βST) was significantly higher between than within habitats (χ2 = 15.68, P <0.001, Fig. 4b). The beta diversity (βWN) in the landscape considering the combined network was generally high but did not differ between and within habitats (βWN within habitats: 0.92 ± 0.08; βWN between habitats: 0.94 ± 0.05; χ2 = 58.49, P = 0.07; Fig. 4c; Table S4). When considered separately, β-plants were significantly higher between than within habitats (χ2 = 340.53, P <0.001, Fig. 4d), but this was not the case for β-pollinators (χ2 = 1.21, P = 0.27; Fig. 4e; Table S4).
Network structure, species level indices and their relationship with floral traits
The 36 animal-centred networks showed higher levels of interaction partitioning than the plant-centred ones considering specialization H2' (animal-centred: 0.69±0.14, plant-centred: 0.58±0.16; χ2 = 20.47, P <0.001), z H2' (animal-centred: 23.03±8.99, plant-centred: 8.31±7.18; χ2 = 48.63, P <0.0001), modularity Qw (animal-centred: 0.73±0.09, plant-centred: 0 35±0.18; χ2 = 87.90, P <0.0001) and z Qw (animal-centred: 30.53±13.32, plant-centred: 6.09±7.02; χ2 = 65.17, P <0.0001; Fig. 5; Table S3). When considering the combined networks, network metrics returned intermediate values between those found for each method separately (Fig. 5).
Considering species level indices, degree (animal-centred: 3.94±5.29, plant-centred: 3.78±5.80; χ2 = 75.14, P <0.001) and specialization (animal-centred: 0.47±0.22, plant-centred: 0.41±0.26; χ2 = 49.45, P <0.0001) were higher in animal-than plant-centred networks (Table S3). Aggregating animal- and plant-centred networks resulted in higher degree than in each of the two methods separately ( 4.38±6.37), while specialization ( 0.45±0.24) resulted in intermediate values.
Plants with distinct pollination systems differed in species degree (
F= 4.29;
P <0.01; Fig. 6a) and specialization (
F = 6.60;
P <0.0001; Fig. 6b; Table S5) in the combined network. Specifically, plant species with oil-bee pollination had a higher degree and specialization than plants pollinated by large bees and dsi (diverse small insects). This result was consistent even after removing from the analysis the outlier mono-dominant
Byrsonima cydoniifolia (Malpighiaceae), a highly connected oil-producing species. For specialization, plants pollinated by oil-bees showed higher values than plants pollinated by small bees and dsi (Fig. 6b; Tukey test
P<0.05).
None of the evaluated factors explained ∆degree estimated between the plant-centred and the combined network (
F = 1.44;
df = 4;
P = 0.18). On the other hand, oil-bee pollinated plants showed higher differences on specialization between sampling methods than the observed for small-bee pollinated plants (Fig. 6c). In addition, there was a large difference on the specialization of flag and gullet flowers between sampling methods compared to the less restrictive dish flowers (Fig. 6d).
Discussion
We show, for the first time, that the sampling method significantly affects how interaction dissimilarity across space is characterized in a heterogeneous tropical landscape. This finding is likely caused by differences in the mobility of organisms embedded in the ecological web, since across all analyses, pollinators showed lower dissimilarity than plants. Our results also reflect changes in the composition of plant species between different habitats in the landscape. Importantly, some differences between methods may be linked to plant’s floral traits, especially the type of flowers and the pollination systems of plant species. These outcomes have important consequences on how plant-pollinator networks are described and that these groups are affected differently by the complementary sampling methods.
Plant-centred sampling indicated higher interaction and species beta diversity in space. Beta diversity of interactions is driven by both changes in species composition, which depends largely on the environmental tolerances, phenology, and dispersal ability of interacting partners (
Trøjelsgaard et al., 2015; Graham & Weinstein, 2018; Noreika
et al., 2019), and rewiring, when co-occurring species interact differently over gradients owed to changes in the costs and benefits of interactions (
CaraDonna et al., 2017). While plant-centred sampling is limited in space by the chosen sampling unit, animal-centred networks vary in spatial coverage according to the mobility of pollinators. Reflecting that, with the plant-centred method around half of plants (54%) and pollinator (51%) species were found in more than three sampled networks, while in the animal-centred sampling most plants (85%) and pollinator (96%) species occurred in more than three networks. Therefore, the higher mobility of pollinators compared to plants results in lower beta diversity of interactions, indicating that although we studied three habitat types, many plant and animal species are in fact shared across them. In this context, adding the less common animal-centred method reduces beta interaction turnover since it detects the shared interactions in heterogeneous landscapes such as in the Pantanal Floodplain. However, pollen identification in highly biodiverse tropical habitats for animal-centred sampling, can be a challenge (
Ramírez-Burbano et al., 2017). Thus, the choice of the best method should take into account the costs (time, financial) as well as the logistics, keeping in mind the limitation of the results.
It has been argued that animal-centred sampling could help unveil rare interactions not detected by direct observations for both rare and common plant species (Dorado et al., 2011). Such additional detection blurs the “artificial” separation of different networks imposed by the sampling units, revealing a lower dissimilarity of interactions. Our results show that interaction turnover, rather than rewiring, was the dominant driver of interaction dissimilarity. This pattern was consistent across both methods and the combined networks, providing evidence that turnover plays an important role on the dynamics of ecological networks in space (Trøjelsgaard et al., 2015), which indicates that species composition, mainly of plants, can change the distribution and occurrence of interactions in the landscape. At the same time, we also show that animal-centred sampling showed higher rewiring of interactions than plant-centred networks, indicating that pollinators are more flexible in changing their interactions in the landscape than assumed by only observing plants. Thus, adding pollen-data reinforces the plasticity of plant-pollinator interactions.
In our study, the lower number of pollinators in the animal-centred network was caused by the absence of pollen in the body of many legitimate visitors. Although not a rule (Vianna
et al., 2014; Ramírez-Burbano
et al., 2017; Zhao
et al., 2019), previous studies have found that subsequently calibrating interaction data with “better” estimates of potential pollination, i.e. pollen loads, pollen deposition on stigma and fruit set, will increasingly render only the subsets of interactions constrained by specific co-adaptations and higher estimates of interaction specialization (Bosch
et al., 2009; Popic
et al., 2013; Santiago-Hernandez
et al., 2019; Zhao
et al., 2019). Likewise, when adding pollen data, we showed that both specialization and modularity increased in the combined network in relation to the plant-centred network (Fig. 5). Considering only the visitors that contact the reproductive structures of flowers is a first step in including only potential pollination interactions, but not all such visitors actually carry and deposit pollen (Mesquita‐Neto
et al., 2018; Santiago-Hernandez
et al., 2019; Minaar
et al., 2019), leading to distinct estimates of specialization according to the currency used to estimate interactions. Although pollen in the body of pollinators may have other fates than deposition on stigmas and not all insect touching stigmas are pollinators (Bosch
et al., 2009;
Ballantyne et al., 2015; Minaar
et al., 2019), calibrating the interaction networks with currencies that better approximates pollination will more likely reflect the most effective pollinators that drive convergent evolution of floral forms (Rosas-Guerrero
et al., 2014). Refined estimates of interactions in pollination networks are certainly important to improve our understanding of the community level and evolutionary drivers of plant-pollinator associations (Phillips
et al., 2020). And this should also include measures of beta diversity of interactions (Trøjelsgaard
et al., 2015).
The differences according to method observed at the network level were partially reflected at the species level, and importantly, were related with some species traits. Oil flowers had both higher degree and higher specialization than some other pollination systems such as non-oil bee pollinated plants and species associated with small diverse insects that characterize more opportunistic pollinator groups (Lima et al., 2020). Besides that, oil rewarding plants also showed greater differences on specialization values depending on the method when compared to plants pollinated by small bees. Specifically, these results indicate that oil rewarding plants had more bee species as pollinators than plants with other systems, but these are a specialized subset of bees not observed in other plants. Floral oil is a crucial resource for solitary bees that feed the larvae with a mixture of oil and pollen and is also used for nest construction (Simpson & Neff, 1981). Oil-collecting species of the genus Centris and Epicharis are the main pollinators of Neotropical oil producing plants and show two main traits: (i) usually large body sizes, and (ii) a particular behaviour when handling the flowers, completely embracing them and touching their reproductive structures (Barônio et al., 2020). Some of the interactions between oil-flowers and bees were only recorded by pollen analysis, and it may be especially relevant for oil-collecting Centris bees which are often difficult to be observed in the field due to their agility and speed during the flight. Hence, pollen-based methods could be especially interesting for detecting interactions for this group of oil plants and their pollinators (Dorado et al., 2011; Rabelo et al., 2015). Gullet and flag floral types were associated with higher differences in specialization levels between methods than others such as brush and dish types, which may be caused by the fact that the former floral types show floral morphologies that are often more restrictive to floral visitors (Olesen et al., 2007; Westerkamp & Claßen-Bockhoff, 2007). Hence, pollen and nectar collection from flag and gullet flowers may require both specialized pollinator behaviour and a specific body size, restricting the visitor fauna mainly to large bees (Olesen et al., 2007). This can mean that even with adequate pollination behaviour during visits, some insects may adhere little pollen grains to their body surface, hindering detection and increasing the difference between the methods.
In summary, sampling method affects not only how plant-pollinator network structures are characterized (Bosch
et al., 2009; Popic
et al., 2013; Zhao
et al., 2019), but also the beta diversity of interactions among networks. Moreover, we show that some plant traits contributed to method-related differences on estimates of species level specialization, including pollination systems and floral morphology. Importantly, our results highlight that plant-centred sampling reflects the spatial limits defined by the researcher, which is usually smaller than the flight capacity of many of the interacting floral visitors. If fact, because distinct pollinator species and groups will have distinct mobilities, choosing a unique “correct” sampling unit may not be possible. In this regard, even when considering distinct habitat types in the studied heterogeneous landscape, animal-based sampling still contributes to overall lower levels of beta diversity of interactions among networks. Hence, flying pollinators show considerable flexibility to change their interactions in space. Since understanding the variation in architecture of interaction networks and flexibility of species to change interactions across environmental gradients has important implication for conservation (
Burkle et al., 2016) and coevolutionary dynamics (
Tylianakis & Morris, 2017) of plant-pollinator communities, using distinct and complementary methods to estimate interaction network structure and beta diversity of interactions may be desirable. This is likely most relevant in interactions involving organisms with considerable differences in mobility, such as plants and their pollinators. Nevertheless, the choice of method to sample interactions should also reflect the scale at which interaction patterns are investigated. By strongly reflecting local scale variation in plant composition, plant-centred approaches may be especially suitable for investigating the flexibility of plants and pollinators in changing their interactions is space, reflecting niche partitioning related to local competition and trade-offs associated with the optimal foraging theory (Bartomeus
et al., 2016; Guimarães Jr., 2020). The choice of the method, or the decision to use them in combination, should thus always reflect the appropriate scale addressed on the question about the factors determining species interactions.
References
Alho CJR. 2008. Biodiversity of the Pantanal: response to seasonal flooding regime and to environmental degradation. Brazilian Journal of Biology 68: 957–966.
APG IV. 2016. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG IV. Botanical Journal of the Linnean Society 181: 1–20.
Ballantyne G, Baldock KC, Willmer PG. 2015. Constructing more informative plant–pollinator networks: visitation and pollen deposition networks in a heathland plant community. Proceedings of the Royal Society B: Biological Sciences 282(1814): 20151130.
Barônio GJ, de Brito VLG, Nogueira A, Rech AR. 2020. Beyond robbery: the role of upside‑down behaviour performed by small oil‑collecting bees in Malpighiaceae. Arthropod-Plant Interactions 14: 613–621.
Bartomeus I, Gravel D, Tylianakis JM, Aizen MA, Dickie IA, Bernard‐Verdier M. 2016. A common framework for identifying linkage rules across different types of interactions. Functional Ecology 30(12): 1894-1903
Baselga A. 2010. Partitioning the turnover and nestedness components of beta diversity. Global Ecology and Biogeography 19: 134–143.
Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67: 1–48.
Beckett SJ. 2016. Improved community detection in weighted bipartite networks. Royal Society Open Science 3: 140536.
Blüthgen N, Menzel F. Blüthgen N. 2006. Measuring specialization in species interaction networks. Ecology 6: 1–12.
Bosch J, Martín González AM, Rodrigo A, Navarro D. 2009. Plant-pollinator networks: adding the pollinator’s perspective. Ecology Letters 12: 409–419.
Burkle LA, Alarcón R. 2011. The future of plant-pollinator diversity: Understanding interaction networks across time, space, and global change. American Journal of Botany 98(3): 528-38.
Burkle LA, Myers JA, Belote RT. 2016. The beta‐diversity of species interactions: Untangling the drivers of geographic variation in plant–pollinator diversity and function across scales. American Journal of Botany 103: 118-128.
CaraDonna PJ, Petry WK, Brennan RM, Cunningham JL, Bronstein JL, Waser NM, Sanders NJ. 2017. Interaction rewiring and the rapid turnover of plant–pollinator networks. Ecology Letters 20: 385–394.
Carstensen DW, Sabatino M, Trøjelsgaard K, Morellato LPC. 2014. Beta diversity of plant-pollinator networks and the spatial turnover of pairwise interactions. PLoS One 9(11): e112903.
Carvalho JC, Cardoso P, Gomes P. 2012. Determining the relative roles of species replacement and species richness differences in generating beta-diversity patterns. Global Ecology and Biogeography 21: 760–771 (2012).
Chacoff NP, Vazquez DP, Lomascolo SB, Stevani EL, Dorado J, Padron B. 2012. Evaluating sampling completeness in a desert plant-pollinator network. Journal of Animal Ecology 81: 190-200.
Chao A. 1984. Nonparametric estimation of the number of classes in a population.
Scandinavian Journal of Statistics 11: 265–270.
Colwell RK, Coddington JA. 1994. Estimating terrestrial biodiversity through extrapolation.
Philosophical Transactions of the Royal Society B: Biological Sciences 345: 101–118.
Dafni A. 1992. Pollination ecology: a practical approach, University Press, New York. 272pp.
Dalsgaard B, Schleuning M, Maruyama PK, Dehling DM, Sonne J, Vizentin‐Bugoni J, Zanata TB, Fjeldså J, Böhning‐Gaese K, Rahbek C. 2017. Opposed latitudinal patterns of network‐derived and dietary specialization in avian plant–frugivore interaction systems. Ecography 40(12): 1395-1401.
Danieli-Silva A, Souza JTM, Donatti AJ, Campos RP, Vicente-Silva J, Freitas L, Varassin IG. 2012. Do pollination syndromes cause modularity and predict interactions in a pollination network in tropical high-altitude grasslands? Oikos 121: 35–43.
de Manincor N, Hautekèete N, Mazoyer C, Moreau P, Piquot Y, Schatz B, Schmitt E, Zélazny M, Massol F. 2020. How biased is our perception of plant-pollinator networks? A comparison of visit-and pollen-based representations of the same networks. Acta Oecologica 105: 103551.
Dorado J, Vázquez DP, Stevani EL, Chacoff NP. 2011. Rareness and specialization in plant–pollinator networks. Ecology 92: 19–25.
Dormann CF, Gruber B, Fründ J. 2008. Introducing the bipartite package: analysing ecological networks. R News 8: 8–11.
Dormann CF, Frund J, Bluthgen N, Gruber B. 2009. Indices, graphs and null models: analyzing bipartite ecological networks. The Open Ecology Journal 2: 7–24.
Erdtman G. 1952. Pollen morphology and plant taxonomy-angiosperms. Almqvisit & Wiksell, Stockholm. 539pp.
Fox J, Weisberg S. 2011. An {R} companion to applied regression (2nd ed.). Thousand Oaks, CA: Sage.
Frankie GW, Haber WW, Opler PA, Bawa KS. 1983. Characteristics and organization of the large bee pollination system in the Costa Rican dry forest. In: Handbook of experimental pollination biology. Jones CE, Little RJ. [eds]. New York, Scientific and Academic Editions, New York, 441–448.
Fründ J, McCann KS, Williams NM. 2016. Sampling bias is a challenge for quantifying specialization and network structure: lessons from a quantitative niche model. Oikos 125(4): 502-513.
Graham CH, Weinstein BG. 2018. Towards a predictive model of species interaction beta diversity. Ecology Letters 21: 1299–1310.
Guimarães Jr PR. 2020. The structure of ecological networks across levels of organization. Annual Review of Ecology, Evolution, and Systematics 51: 433-460.
Hartig F. 2020. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R package version 0.3.3.0. http://florianhartig.github.io/DHARMa/
Hothorn T, Bretz F, Westfall P. 2008. Simultaneous inference in general parametric models. Biometrical Journal 50: 346–363.
Hsieh TC, Ma KH, Chao A. 2014. CRAN - Package iNEXT: iNterpolation and EXTrapolation for species diversity. R package version 2.0. Available at: http://chao.stat.nthu.edu.tw/blog/software-download.
Jordano P. 2016. Sampling networks of ecological interactions. Functional Ecology 30: 1883–1893.
Kessler M, Abrahamczyk S, Bos M, Buchori D, Putra DD, Gradstein SR, Höhn P, Kluge J, Orend F, Pitopang R et al. 2009. Alpha and beta diversity of plants and animals along a tropical land‐use gradient. Ecological Applications 19: 2142–2156.
Legendre P. 2014. Interpreting the replacement and richness difference components of beta diversity. Global Ecology and Biogeography 23: 1324–1334.
Lima GO, Leite AV, Souza CS, Castro CC, Bezerra ELS. 2020. A multilayer network in an herbaceous tropical community reveals multiple roles of floral visitors. Oikos 00: 1–11.
Machado IC, Lopes AV. 2004. Floral traits and pollination systems in the Caatinga, a Brazilian Tropical Dry Forest. Annals of Botany 94: 365–376.
Maglianesi MA, Blüthgen N, Böhning-Gaese K, Schleuning M. 2015. Functional structure and specialization in three tropical plant–hummingbird interaction networks across an elevational gradient in Costa Rica. Ecography 38: 1119–1128.
Magrach A, González-Varo JP, Boiffier M, Vilà M, Bartomeus I. 2017. Honeybee spillover reshuffles pollinator diets and affects plant reproductive success. Nature Ecology & Evolution 1(9): 1299.
Maruyama PK, Vizentin‐Bugoni J, Oliveira GM, Oliveira PE, Dalsgaard B. 2014. Morphological and spatio‐temporal mismatches shape a neotropical savanna plant‐hummingbird network. Biotropica 46: 740–747.
Mesquita‐Neto JN, Blüthgen N, Schlindwein C. 2018. Flowers with poricidal anthers and their complex interaction networks - Disentangling legitimate pollinators and illegitimate visitors. Functional Ecology 32: 2321–2332.
Minnaar C, Anderson B, de Jager ML, Karron JD. 2019. Plant–pollinator interactions along the pathway to paternity. Annals of Botany 123(2): 225-245.
Noreika N, Bartomeus I, Winsa M, Bommarco R, Öckinger E. 2019. Pollinator foraging flexibility mediates rapid plant-pollinator network restoration in semi-natural grasslands. Scientific Reports 9: 1–11.
Novotny V. 2009. Beta diversity of plant–insect food webs in tropical forests: a conceptual framework. Insect Conservation Diversity 2: 5–9.
Nunes da Cunha C, Junk WJ. 2009. A preliminary classification of habitats of the Pantanal of Mato Grosso and Mato Grosso do Sul, and its relation to national and international wetland classification systems. In: The Pantanal: Ecology, biodiversity and sustainable management of a large neotropical seasonal wetland. Junk WJ, Da Silva CJ, Nunes da Cunha C, Wantzen KM. [eds]. Pensoft Pub, Sofia, 127–141.
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Wagner H. 2018. Package ‘vegan’ - Community Ecology Package.
Olesen JM, Dupont YL, Ehlers BK, Hansen DM. 2007. The openness of a flower and its number of flower‐visitor species. Taxon 56: 729–736.
Patefield WM. 1981. Algorithm AS 159: An efficient method of generating random R x C tables with given row and column totals. Journal of the Royal Statistical Society: Series C (Applied Statistics) 30: 91–97.
Phillips RD, Peakall R, van der Niet T, Johnson SD. 2020. Niche perspectives on plant–pollinator interactions. Trends in Plant Science 25: 779–793.
Poisot T, Canard E, Mouillot D, Mouquet N, Gravel D. 2012. The dissimilarity of species interaction networks. Ecology Letters 15: 1353–1361.
Popic TJ, Wardle GM, Davila YC. 2013. Flower‐visitor networks only partially predict the function of pollen transport by bees. Austral Ecology 38: 76-86.
Pott A, Oliveira AK, Damasceno-Junior GA, Silva JSV. 2011. Plant diversity of the Pantanal wetland. Brazilian Journal of Biology 71: 265–273.
R Development Core Team. 2019. R: a language and environment for statistical computing. R Foundation for Statistical Computing. Available at <https:// www.r-project.org/>. Access on 22 April 2020.
Rabelo LS, Vilhena AM, Bastos EM, Aguiar CM, Augusto SC. 2015. Oil-collecting bee–flower interaction network: do bee size and anther type influence the use of pollen sources? Apidologie 46: 465–477.
Ramírez-Burbano MB, Stiles FG, González C, Amorim FW, Dalsgaard B, Maruyama PK. 2017. The role of the endemic and critically endangered Eriocnemis mirabilis in plant-hummingbird networks of Colombian Andes. Biotropica 49: 555–564.
Ramsar & MMA. 2010. Cuidar das Zonas Úmidas – uma resposta às mudanças climáticas. MMA, Brasília, 28 p.
Rosas-Guerrero V, Aguilar R, Martén-Rodríguez S, Ashworth L, Lopezaraiza-Mikel M, Bastida JM, Quesada M. 2014. A quantitative review of pollination syndromes: do floral traits predict effective pollinators? Ecology Letters 17: 388–400.
Santiago-Hernandez MH, Marten-Rodrıguez S, Lopezaraiza-Mikel M, Oyama K, Gonzalez-Rodrıguez A, Quesada M. 2019. The role of pollination effectiveness on the traits of interaction networks: from floral visitation to plant fitness. Ecology 100: e02803.
Sazatornil FD, Moré M, Benitez-Vieyra S, Cocucci AA, Kitching IJ, Schlumpberger BO, Oliveira PE, Sazima M, Amorim FW. 2016. Beyond neutral and forbidden links: morphological matches and the assembly of mutualistic hawkmoth–plant networks. Journal of Animal Ecology 85: 1586–1594.
Simpson BB, Neff JL. 1981. Floral rewards: alternatives to pollen and nectar. Annals of the Missouri Botanical Garden 68: 301–322.
Soininen J, Heino J, Wang J. 2018. A meta‐analysis of nestedness and turnover components of beta diversity across organisms and ecosystems. Global Ecology and Biogeography 27: 96–109.
Souza CS, Maruyama PK, Aoki C, Sigrist MR, Raizer J, Gross CL, Araujo AC. 2018. Temporal variation in plant–pollinator networks from seasonal tropical environments: Higher specialization when resources are scarce. Journal of Ecology 106: 2409–2420.
Trøjelsgaard K, Jordano P, Carstensen DW, Olesen JM. 2015. Geographical variation in mutualistic networks: similarity, turnover and partner fidelity. Proceedings of the Royal Society B: Biological Sciences 282(1802): 20142925.
Tylianakis JM, Morris RJ. 2017. Ecological networks across environmental gradients. Annual Review of Ecology, Evolution, and Systematics 48: 25–48.
Vianna MR, Luz CFP, Kleinert AMP. 2014. Interaction networks in a Brazilian cerrado: what changes when you add palynological information to floral visitor data? Apidologie 45: 418-430.
Vizentin‐Bugoni J, Maruyama PK, Debastiani VJ, Duarte LDS, Dalsgaard B, Sazima M. 2016. Influences of sampling effort on detected patterns and structuring processes of a Neotropical plant–hummingbird network. Journal of Animal Ecology 85: 262–272.
Vizentin-Bugoni J, Maruyama PK, Souza CS, Ollerton J, Rech AR, Sazima M. 2018. Plant-pollinator networks in the tropics: A review. In: Ecological networks in the tropics. Dáttilo W, V Rico-Gray [eds.]. Dordrecht, the Netherlands: Springer, 73–91.
Westerkamp C, Claßen-Bockhoff R. 2007. Bilabiate flowers: the ultimate response to bees? Annals of Botany 100: 361–374.
Whittaker RH. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecological Monographs 30: 280–338.
Zhao YH, Lázaro A, Ren ZX, Zhou W, Li HD, Tao ZB, Xu K, Wu ZK, Wolfe LM, Li DZ et al. 2019. The topological differences between visitation and pollen transport networks: a comparison in species rich communities of the Himalaya–Hengduan Mountains. Oikos 128: 551–562.
Acknowledgements
We thank the Rufford Foundation for a grant (RSG: 21366-1); CNPQ and CAPES for scholarships to CSS (Finance Code 001; FUNDECT/CAPES – n°03/2008 - PAPOS-MS, Process: 23/200.383/2008 and n° 44/2014 – PAPOS-MS, Process: 23/200.638/2014), the São Paulo Research Foundation for a Postdoctoral grant to PKM (FAPESP, Process: 2015/21457-4) and CNPq for a Productivity grant to ACA (310999/2018-9) and IGV (313801/2017-7). Special thanks also to Ignasi Bartomeus for the script and code provided for calculating beta diversity of interactions and the three anonymous reviewers for comments that greatly improved the quality of the manuscript.
Author Contributions statement
CSS, ACA, CLG and PKM conceived the ideas and designed methodology; CSS and KCBSS collected and organized the data; CSS analysed the data; CSS and PKM wrote the first draft of the manuscript; CSS, PKM, KCBSS, IGV, CLG and ACA contributed critically to the drafts and gave final approval for publication.
Competing Interests statement
The authors declare no competing financial interests.
Figure legends
Figure 1. (a) Location of the 36 studied networks in Southern Pantanal, Brazil and typical physiognomies of the three studied habitats (b): Riparian forest; (c): flooded savannah dominated by Byrsonima cydoniifolia (“Canjiqueiral”) and (d): flooded savannah dominated by Tabebuia aurea (“Paratudal”).
Figure 2. Interaction networks and their sampling methods. We used the frequency of visits for plant-centred network and the frequency of pollinator individuals with pollen grains of a given plant species for the animal-centred network as currencies of interactions. The two methods were aggregated to form the combined networks.
Figure 3. β diversity of species and interactions according to sampling methods (animal-centred, plant-centred and combined method). a) Interaction β diversity, βOS - rewiring, βST – turnover, βWN – beta diversity of interactions and βWN.repl – which removes effects of differences on richness between the networks; b) species β diversity for the pollinators and plants. Box plots show the median (horizontal line), the lower and upper bounds of each box plot denote the first and third quartiles (the 25th and 75th percentiles, respectively), and whiskers above and below the box plot indicate 1.5 times the interquartile range. The points located outside of the whiskers of the box plot represents the outliers. Statistical analysis was performed by two-way ANOVA (type II) followed by a Tukey post hoc test (p values are shown above the box plots).
Figure 4. β diversity of interactions and species measured within and between the three different studied habitats (Canjiqueiral, Riparian forest and Paratudal), considering the combined networks: a) rewiring - βOS; b) turnover - βST; c); β diversity of interactions - βWN; and species turnover for d) pollinators and e) plants. Box plots show the median (horizontal line), the lower and upper bounds of each box plot denote the first and third quartiles (the 25th and 75th percentiles, respectively), and whiskers above and below the box plot indicate 1.5 times the interquartile range. The points located outside of the whiskers of the box plot represents the outliers. Statistical analysis was performed by two-way ANOVA (type II) followed by a Tukey post hoc test (p values are shown above the box plots).
Figure 5. Network metrics describing the structure of interaction networks between plants and pollinators in the Pantanal. We used visitation (plant-centred) and pollen load (animal-centred) to sample interactions. The differences in specialization (H2' and z H2') and modularity (Qw and z Qw) of the networks using the different methods and their combination are shown in figures (a), (b), (c) and (d). Box plots show the median (horizontal line), the lower and upper bounds of each box plot denote the first and third quartiles (the 25th and 75th percentiles, respectively), and whiskers above and below the box plot indicate 1.5 times the interquartile range. The points located outside of the whiskers of the box plot represents the outliers. Statistical analysis was performed by two-way ANOVA (type II) followed by a Tukey post hoc test (p values are shown above the box plots).