https://orcid.org/0000-0002-0797-9261
Figures
Abstract
Pollinators are threatened by land-use and land-cover changes, with the magnitude of the threat depending on the pollinating taxa, land-use type and intensity, the amount of natural habitat remaining, and the ecosystem considered. This study aims to determine the effect of land use (protected areas, plantations, pastures), land cover (percentage of forest and open areas within buffers of different sizes), and plant genera on the relative abundance of nectivorous birds (honeyeaters), bees (native and introduced), and beetles in the mixed-use landscape of the Tasman Peninsula (Tasmania, Australia) using mixed-effect models. We found the predictor selected (through model selection based on R2) and the effect of the predictors varied depending on the pollinating taxa. The land-use predictors were selected for only the honeyeater abundance model with protected areas and plantations having substantive positive effects. Land-cover predictors were selected for the honeyeater and native bee abundance models with open land cover within 1500 m and 250 m buffers having substantive negative and positive effects on honeyeaters and native bees respectively. Bees and beetles were observed on 24 plant genera of which only native plants (and not invasive/naturalised) were positively associated with pollinating insects. Pultenaea and Leucopogon were positively associated with native bees while Leucopogon, Lissanthe, Pimelea, and Pomaderris were positively associated with introduced bees. Leptospermum was the only plant genus positively associated with beetles. Our results highlight that one size does not fit all—that is pollinator responses to different landscape characteristics vary, emphasising the importance of considering multiple habitat factors to manage and support different pollinator taxa.
Citation: Diengdoh VL, Brook BW, Hunt M, Ondei S (2023) Association between land use, land cover, plant genera, and pollinator abundance in mixed-use landscapes. PLoS ONE 18(11): e0294749. https://doi.org/10.1371/journal.pone.0294749
Editor: Kleber Del-Claro, Universidade Federal de Uberlandia - Campus Umuarama, BRAZIL
Received: August 25, 2021; Accepted: November 8, 2023; Published: November 22, 2023
Copyright: © 2023 Diengdoh et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: BWB was the recipient of the Australian Research Council [grant numbers FL160100101 and CE170100015]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Pollinators (animal vectors of pollen) are important ecosystem components and are estimated to globally pollinate 70% of crops [1] and 87% of wild plants [2]. The abundance, diversity, distribution, evenness, population, and richness of pollinators such as birds (e.g., hummingbirds), butterflies, bees (wild), hoverflies, and fruit bats have declined/are predicted to decline, albeit with exceptions [3–9]. The decline of different pollinator species is due to the interactions and synergies between land-use and land-cover change, introduced and invasive species, agrochemicals, and climate change [10–14]. The importance of the aforementioned threats as drivers of pollinator decline varies across different regions of the world [14]; while land-use and land-cover change have been globally identified to be among the main causes of pollinator decline, evidence is still inconclusive for some regions, particularly Australia, New Zealand, and Africa [14].
Land use refers to human activity on the land while land cover refers to the natural and artificial structures covering the land [15]. The effect of land-use and land-cover change on pollinator abundance varies depending on the pollinator taxa, land-use type and intensity, and the amount of natural habitat remaining and the ecosystem [16–18]. However, studies generally assess the effect of land use and land cover on a single pollinator taxon–bees [e.g., 19–21], butterflies [e.g., 22–24], or birds [e.g., 25–27], although there are exceptions where multiple taxa are assessed [e.g., 28–30]. When different land-use and land-cover types are included, comparisons are generally made in landscapes dominated by agriculture [e.g., 22, 27, 29] or urban areas [e.g., 26, 28, 31] with small areas of native vegetation. A minority of studies investigated landscapes that are not dominated by a single human activity [e.g., 19, 30, 32, 33].
Land-use and land-cover change drive pollinator decline by reducing plant abundance and diversity and in turn reducing the availability of floral resources on which pollinators depend [34–36]. As such, increasing plant species richness has a positive effect on pollinator richness irrespective of the land-use type [37]. The restoration of pollinator communities can be accomplished by using a subset of all available plant families or species [38, 39] and benefits of flower strips on pollinator populations and visitation can be achieved through careful plant species selection [40, 41]. Identification of such a subset of plants will be useful for improving pollinator abundance in different land use, such as urban areas [42] and even be used to improve crop yield [43].
In this study, we aim to assess the effect of land use, land cover, and plant genera on the relative abundance (count) of nectivorous birds (honeyeaters), bees, and beetles. There are three research questions we aim to address–(1) do protected areas support a higher abundance of pollinators than plantations and pastures? (2) which land cover buffer has the highest predictive capacity and does forest land cover support a higher abundance of pollinators than open land cover? and (3) are native plants associated with a higher abundance of insect pollinators than invasive/naturalised plants and do the effect of these plant genera vary among the insect pollinator groups? In this study, land use included protected areas, plantations, and pasture while land cover included forest and open areas. To answer those questions, we firstly collected field data on pollinator abundance across protected areas, plantations, and pastures using the Tasman Peninsula (Tasmania, southern-temperate Australia) as a case study of a diverse, mixed-use landscape. Secondly, we assessed land cover from satellite imagery. Thirdly, we used mixed-effect models to evaluate the effect of the aforementioned landscape characteristics on pollinator abundance. We discussed the associations between predictors and pollinator abundance and which predictors could be managed to improve the abundance of pollinator communities.
Materials and methods
Study area
The Tasman Peninsula, located in the south-eastern region of the large island of Tasmania, Australia (Fig 1), covers an area of 660.4 km2 with elevation from 0 to 582 m a.s.l. (meters above sea level). It is characterised by a mix of dry and wet sclerophyll eucalypt forest and dry coastal vegetation and supports a third of the vascular plants found in Tasmania with 556 vascular plants, of which there are 336 dicotyledons [44]. Overall, the study area is classified as a rural area with a mixture of grazing pasture, forest plantations, and protected areas land use types [45], with the latter covering over a quarter (26.7%) of the study area [46]. The vegetation also varies within the same land use; for example, protected areas cover dry and wet eucalypt woodland and forest and at a higher level, covers Eucalyptus obliqua forest with broad-leaf shrubs and Eucalyptus amygdalina forest and woodland [47].
Image is licensed under a CC BY 4.0 license.
Ethics
We received written consent from the Animal Ethics Committee, University of Tasmania under Project No: A0016856 to carry out bird surveys. We received written consent from the Department of Primary Industries, Parks, Water and Environment (DPIPWE) Tasmania, under Authority No’s. FA17315 and FA18185 to carry out surveys in protected areas. We received verbal consent to carry out fieldwork in plantations and pastures from the respective owners.
Pollinator abundance survey
Surveys were conducted under low-wind and non-rainy conditions during the September to December Austral spring-summer months of 2018. Surveys were repeated monthly to account for any temporal changes in flowering vegetation and pollinator presence, and provide more reliable estimates within sites, although at the cost of spatial extent [48]. Surveys were conducted in 2-ha circular plots which were stratified across three different land-use types–protected areas (n = 18), plantations (n = 12), and pastures (n = 6; Fig 1). The unequal number of plots was due to limitations on accessibility. The plots were randomly distributed to be at least 400 m apart and at a minimum distance of 80–100 m from a road. Within a plot, different sampling methods were used for different pollinator taxa, because a pilot study we carried out suggested that no one method worked for all pollinators.
Bird abundance was recorded using the standard protocol recommended by BirdLife [49]. This involved visually identifying bird species and recording their count from the centre of the 2-ha plot for 20-minutes which took place between 07.00–11.00 h and 16.00–19.00 h. We focused on honeyeaters (family Meliphagidae) as they are the dominant group of nectivorous birds in Australia [50]. Birds were identified at the species level using the field guide by Simpson and Day [51]. We also recorded the presence/absence of eucalyptus trees, which we identified at the species level using a field guide by Wiltshire and Potts [52]. The trees were grouped into the sub-genera Symphyomyrtus and Eucalyptus (formerly Monocalyptus) [53] as they have been shown to influence honeyeater presence [54, 55]. In total, we completed 144 surveys for birds (36 2-ha plots × 4 months).
Bees and beetle abundance was recorded using a visual method, a commonly used method for detecting bee abundance [56, 57]. This involved photographing and recording the count and species of bees and beetles that landed on the stigma of 4 ground-level flowering plants within the 2-ha plots for 10-minutes each (i.e., a total of 40-minutes) between 10.00–15.00 h. We chose 40 minutes because it allowed us to collect enough data while remaining logistically feasible and splitting it across 4 flowering plants (10 minutes each) allowed us to capture local variation. The flowering plants were randomly selected and where possible, four different species of flowering plant were chosen; if none was present no observation was made. Bees and beetles were identified using the iNaturalist website (https://inaturalist.org/) while flowering plants were identified to genus level using the University of Tasmania key to Tasmanian vascular plants (www.utas.edu.au/dicotkey/dicotkey/key.htm). We also aimed to record butterfly abundance using this method, however, none were observed. In total, we completed 393 surveys for bees and beetles, respectively, out of a total of 576 surveys (36 2-ha plots × 4 flowering plants × 4 months) because 183 surveys had no flowering plants.
Land-cover analysis
The land-cover classes considered in this study are ‘forest’ (areas dominated by trees, including plantations), ‘open’ (areas with low-lying vegetation, including shrubs and grasses), ‘barren’ (areas lacking vegetation), and ‘water’. The allocation of land-cover classes was inferred from Sentinel 2 imagery, which has a spatial resolution of 10 × 10 m [58].
We carried out a pixel-based classification of the imagery following the method outlined by Diengdoh [59]. This method involved, firstly, creating training and validation data for the above-mentioned land-cover classes in QGIS [60]. The training and validation data were used for fitting and tuning the algorithms, and accuracy assessment, respectively. Secondly, machine-learning algorithms (hereafter algorithms)–support vector machine, Random Forests, k-nearest neighbour, and naïve Bayes were trained and used to classify the satellite imagery using the caret R package [61]. Thirdly, the images classified by the four algorithms were averaged using an unweighted ensemble algorithm [59].
Accuracy was assessed by comparing the output to the validation data, which consisted of 100 points/pixels per land-cover class, randomly selected from the classified image and visually compared to imagery from Google Earth and field data for accuracy assessment. The output metrics we report include the overall accuracy (OA) and the true-skill statistic (TSS) for the classified image, and the sensitivity and specificity of each land-cover class, where: OA is the number of correctly classified pixels divided by the total number of pixels examined [62]; TSS is equal to sensitivity plus specificity minus 1, where sensitivity is observed presences that are correctly predicted, and specificity is the observed absences that are predicted as such [63]. The classification analysis was done in R [64].
To determine the effect of land cover on the abundance of birds, bees, and beetles, the percentage of forest and open land-cover types was calculated within buffers of different sizes– 100, 250, 500, 750, 1000, 1500, and 2000 m radius from the centre of the 2-ha plot. These buffer sizes are similar to those from other studies, e.g., birds [65] and bees [66].
Statistical analyses
We assessed the effect of the predictors–land use, land cover, and plant genera on the abundance of each pollinator group using generalised mixed-effect models to account for the repeated surveys at each plot. The mixed-effect model was implemented using the lme4 R package [67].
Land use was a categorical predictor with 3 levels–protected area, plantation, and pasture; we transformed it into dummy variables (i.e., binary variables) thereby resulting in 3 land-use variables. Land cover was a numeric predictor consisting of the percentage of forest and open areas within different buffer sizes– 100, 250, 500, 750, 1000, 1500, and 2000 m and log-transformed. For honeyeaters, plant sub-genera Symphyomyrtus and Eucalyptus are both categorical predictors. For insects, plant genus is a categorical predictor with 24 levels–Acacia, Anopterus, Arctotheca, Bauera, Bedfordia, Bursaria, Carduus, Daviesia, Epacris, Goodenia, Hibbertia, Leptospermum, Leucopogon, Lissanthe, Lomatia, Melaleuca, Olearia, Pimelea, Pomaderris, Prostanthera, Pultenaea, Taraxacumm, Trifolium, and Zieria (these were the flowering plants from the surveys, identified to genus level) and were similarly transformed into dummy variables thereby resulting in 24 plant genera variables. These transformations were carried out to better fit the models i.e., reduce issues with model convergence.
Instead of using all predictors for model fitting, we only selected those with high R-squared values (R2) to find the best predictive variables and ensure model convergence. To do so, we ran two sets of mixed-effect models. Firstly, we fitted a mixed-effect model for each predictor with ‘plot’ set as the random variable, maintaining all other parameters constant. We then calculated the R2 values of the model and the fixed and random effects using the rsq R package [68] and finally selected the predictor with non-zero R2 values—most predictors had zero R2 values indicating that they do not contribute to explaining model variation and were subsequently removed. For categorical predictors (land use and plant genera), we compared the R2 between different land use and plant genera and then selected the ones with non-zero values. For continuous predictors (percentage of forest and percentage of open land cover), we compared the R2 between different buffers and chose the buffer with the highest R2 for both the percentage of forest or open land cover. Finally, after reducing the number of predictors, we fitted a model including all selected predictor variables. We then assessed the predictor estimates and the R2 of the model and the fixed and random variables. All statistical analyses were carried out using the R software [64].
Results
Pollinator abundance
We observed a total of 297 honeyeaters belonging to eight different species, of which four are endemic to Tasmania (127 individuals). All eight species were recorded in protected areas while five species were recorded in plantations and none in pastures (S1 Table in S1 File). We observed a total of 511 bees which we grouped into two categories–native and introduced bees with 284 and 227 individuals respectively. Native bees included 184 individuals of the Exoneura genus, 48 individuals of the Lasioglossum genus and 52 individuals classified as ‘other bees’ while introduced bees included 211 honey bees and 16 bumble bees. The Exoneura and Lasioglossum native bee genera, as well as honey bees and bumble bees, were found in all land-use types (S2 Table in S1 File). We observed a total of 423 beetles belonging to nine families, with eight families recorded in protected areas, four in planation, and one in pastures (S3 Table in S1 File).
Pollinator abundance and land use
The median count (along a logarithmic axis) of different pollinator groups varied across the three land-use types and four months (Fig 2, S1, S2 Figs in S1 File). For honeyeaters, it was highest in protected areas (S1 Fig in S1 File), with no major temporal variations (Fig 2). No honeyeaters were observed on pastures (Fig 2, S1, S2 Figs in S1 File). The median count (along a logarithmic axis) of native and introduced bees was highest in plantations and protected areas, respectively (S1 Fig in S1 File). However, during October, the median count of native bees was lower in plantations than protected areas and pastures (Fig 2), while for introduced bees it was higher in pastures than protected areas and plantations (Fig 2). The median count (along a logarithmic axis) of beetles was highest in pastures (S1 Fig in S1 File); however, this was only observed only during the month of September, with no beetles observed in pastures during the other months (Fig 2). Although the median count of beetles was lowest in protected areas (S1 Fig in S1 File), protected areas supported beetles during all four months (Fig 2).
Pollinator abundance and land cover
The classified image (Fig 1) had an overall accuracy of 79.5% with a 95% CI [75.2, 83.4] and a true skill statistic score of 77.4%. The confusion matrix of the classification results with sensitivity and specificity of the land-cover classes are included in S4, S5 Tables in S1 File.
Both honeyeaters and beetles were more abundant in areas with a higher forest cover and lower open land cover (Fig 3, the 100m buffer size roughly corresponds to the size of the 2-ha plot). The count of native and introduced bees was similar across the different percentages of land cover (Fig 3).
Pollinator abundance and plant genera
Bees and beetles were observed visiting 24 plant genera (Fig 4). Of those, only Acacia was observed across all land use (Fig 4). The genera Bedfordia, Bursaria, Daviesia, Epacris, Hibbertia, Leucopogon, Lissanthe, Melaleuca, Pomaderris were observed exclusively in protected areas (Fig 4). The genera Anopterus, Bauera, Lomatia, and Prostanthera were found exclusively in plantations (Fig 4) while the genera Arctotheca, Carduus, Taraxacum, and Trifolium were observed only in pastures (Fig 4). The genera Goodenia, Leptospermum, Olearia, Pimelea, Pultenaea, and Zieria were observed in both protected areas and plantations (Fig 4). The only flowering exotic/naturalised genera found in this study were Arctotheca, Taraxacum and Trifolium (Fig 4). Of all the plant genera observed, no one genus was observed across all four months (Fig 4). The months of October and November had a higher number of genera than September and December (Fig 4). The median count (along a logarithmic axis) of insect pollinators varied on plant genera and across land use and months (Fig 4).
In protected areas, the median count (along a logarithmic axis) of native bees was highest in Leucopogon in September, Daviesia in October, and Pultenaea in November (Fig 4). In plantations, it was highest in Zieria in September, Olearia and Acacia in October, Goodenia and Lomatia in November, and Goodenia in December (Fig 4). In pastures it was highest in Acacia in September and October (Fig 4). For introduced bees, the median count (along a logarithmic axis) in protected areas was highest in Pomaderris and Pultenaea in September, Pimelea, Daviesia, and Zieria in October, Melaleuca in November, and Leptospermum in December (Fig 4). In plantations, it was highest in Zieria in September, Anopterus in October, and Goodenia in November and December (Fig 4). In pastures it was highest in Acacia in September and October (Fig 4). The median count (along a logarithmic axis) of beetles was highest in Leucopogon in September, Pomaderris in October, and Leptospermum in December (Fig 4). In plantations, it was highest in Leptospermum in November and Prostanthera in December (Fig 4). In pastures, only Acacia supported beetles (Fig 4).
Mixed effect models
Based on the R2 values, no one land-use predictor was selected for all the models (Table 1, Fig 5A and 5C). Land cover was not selected as a predictor for the beetle model and the buffer sizes of land cover differed between the honeyeater and bee models (Table 1, Fig 5). Although both the eucalyptus subgenera were selected for the honeyeater models (Table 1, Fig 5A), only a handful of plant genera were selected for the insect models (Table 1, Fig 5B–5D). Land use was positively associated with honeyeater abundance (Fig 5A) while land cover was negatively and positively associated with honeyeater and native bee abundance respectively (Fig 5A and 5B). The presence of the sub-genus Eucalyptus had a negative effect on honeyeater abundance (Fig 5A) while the other plant genera were positively associated with insect abundance (Fig 5B and 5C). The correlation between the fixed effects for each model is included in S6-S9 Tables in S1 File. The honeyeater model had high explanatory power, mostly associated with the fixed effects (Fig 5), whereas the insect models had low to nearly zero explanatory power of which most was attributable to unmeasured site-specific influences (the random effect; Fig 5).
Along with the R2 values of the models.
Discussion
We assessed the effect of land use, land cover, and plant genera, on the abundance of multiple pollinator groups across a mixed-use landscape. We found that protected areas did not support a higher abundance of pollinators than plantations and pastures. The predictive capacity of land cover varied depending on the pollinating taxa and forest land cover did not support a higher abundance of pollinators than open land cover. Native plants were associated with a higher abundance of insect pollinators than invasive/naturalised plants with specific, preferred plant genera (e.g., Leptospermum, Leucopogon, Lissanthe, Pimelea, Pomaderris, Pultenaea) having a positive effect on the abundance of native and introduced bees and beetles. Other than for honeyeaters (birds), the models were descriptively poor (i.e., low R2) with fixed effects having lower R2 than the random effect. Nonetheless, for the insect pollinators, there were clear dependencies on specific plant genera, which would indicate a sensitivity to plant-community composition.
We found that the abundance of different pollinator groups varied spatially across the different land-use types and temporally across months. However, protected areas and plantations were considered as relevant predictors only for honeyeater abundance model; here both the land-use types had a positive effect. This is consistent with previous studies which have found that bee abundance does not vary significantly between different land uses [28, 33] and that land uses such as plantation and pasture can have a positive effect on pollinator abundance [16]. Land cover might thus be the primary driver for the difference in honeyeater abundance. The major difference observed between protected areas, plantations, and pastures is the latter lacking trees. This would explain why no honeyeaters were recorded in pastures and why the model i.e., honeyeater abundance ~ pasture failed to converge. Nectarivorous birds are positively associated with woody vegetation cover [27] and paddocks with remnant trees have been found to have a higher abundance and richness of birds including honeyeaters than those without remnant trees [69]. The month-to-month abundance of insect pollinators was highly variable across land use. Pollinators are inherently dependent on plants for floral resources and, given our observation of the difference in plant genera across land use and month, it is likely that plant genera were the spatial and temporal driver of insect pollinator abundance. The abundance of bees has been previously found to vary across land use and time [33] and pollinator visitation also varies across plant species and time [70]. The availability of natural habitats and floral resources present in and around different land-use types are more likely to affect pollinators than land use itself [69, 71–73].
The predictive capacity of land cover within different buffer sizes depended on the pollinating group, which is consistent with the literature [73–75]. Birds, being more mobile across the landscape, can cover larger distances, and as a result forests within larger buffers sizes were expected to be a better predictor. However, this proved not to be the case, with forests within 100m being the better predictor. This could be because honeyeaters in Tasmania are generally breeding residents and some species, such as Yellow and Little wattlebirds, are highly sedentary [76]. Open areas have a negative effect on honeyeater abundance which is expected as trees and woody vegetation are essential for honeyeaters as previously mentioned. However, forest areas did not have a distinct effect on honeyeater abundance which is odd as we expected forests to have a positive effect. The lack of a distinct effect could be due to the abundance of different honeyeater species grouped as a single group, as a result, the nuances of the effect are lost. Woody vegetation has different effects depending on the honeyeater species [27]. Although we found open cover within a 250 m buffer to be the better predictor compared to other buffer sizes, for both native and introduced bees, the predictive capacity of buffer sizes can vary depending on bee genus and species [20].
Although the partial R2 for most plant genera were zero in the insect models, some had positive albeit very small R2 values, suggesting a potential weak relationship between insect abundance and some plant genera. These relationships have plausible biological underpinnings. Pultenaea and Pomaderris are pollinated by bees, beetles, and butterflies [77–79]. Further, the zygomorphic shape of Pultenaea flowers could be important for native bees, as this taxon has been shown to exhibit bee-pollination syndrome (i.e. flower features evolved in response to natural selection driven by bees), although pollinator syndromes have been found to be an inadequate explanation for pollinator visitation in Tasmania [80]. Pimelea spp. have narrow tubular odourless flowers which are known to be visited by introduced bees–Apis mellifera and Bombus terrestris–which have long tongues [80]. Leptospermum, in addition to providing nectar and pollen, is also a rich source of fruits and grass-root material which beetles and their larvae feed upon [81, 82]. The actinomorphic shape of Leptospermum could facilitate beetle visitation, as they find it easier to land on simple dish- and bowl-shaped blossoms [79].
The R2 of the random effect was higher than the fixed effects for all models except for the honeyeater and beetle model and overall, the models are descriptively poor with low R2. The higher R2 of the random effect (i.e., plots) indicates that there probably are plot-specific characteristics that are more likely to influence pollinator abundance such as micro-climatic conditions [83], nesting habitat [84], or abundance of floral resources [85, 86], which were not recorded and a limitation of this study; future studies should ensure such data are collected. Data paucity potentially due to the visual sampling method could be another factor for the low R2 values. Future studies should consider using both active and passive sampling techniques such as acoustic recorders for birds [87], image recorders for insects [88], and a combination of different traps and nets for insects [56, 57] to obtain a more detailed representation of pollinator communities and allow for the testing of species-specific responses to the predictor variables. A limitation of the study is the lack of collection of voucher specimens for bees and beetles which might reduce the confidence of the conclusions drawn. Introduce bees are easy to identify and thus conclusions drawn are reliable. Native bees are more difficult to identify and might be confused with wasp, however, the identification through photographs was limited to those that could be identified as bees and we only included Exoneura and Lasioglossum genera and an uncategorised group of bees. Beetles were identified at the family level and since we did not assess family-specific response, the identification of a photographic specimen as a beetle and the conclusion drawn are reliable.
Overall, we found that one size does not fit all i.e., the effects and predictive capacity of land use, land cover, and plant genera in a mixed-use landscape varied depending on the pollinating taxa. Indeed, our results highlight the complexity of pollinator-landscape interactions and that different taxa require different conservation and management policies. Based on our results for the Tasman Peninsula, plants species belonging to the native genera Leptospermum, Leucopogon, Lissanthe, Pimelea, Pomaderris, and Pultenaea could potentially be used for promoting and sustaining the abundance of bees and beetles in anthropogenic landscapes, such as agriculture and urban areas, in Tasmania and mainland Australia where applicable. However, further empirical tests are required to test the effectiveness of these plants to sustain and promote bee and beetle populations. Although a past-present comparison would be particularly beneficial for assessing the impacts of land-use and land-cover changes on pollinators, we lacked baseline data on pollinator abundance, making comparative space-for-time assessments a logical substitute. Future studies should consider resampling previously sampled areas to make past-present comparisons to better understand how changes in land use and land cover and even temperature and precipitation influence pollinator abundance and other diversity metrics.
Supporting information
S1 File. Contains Supporting tables and Supporting figures.
https://doi.org/10.1371/journal.pone.0294749.s001
(DOCX)
Acknowledgments
The authors acknowledge the Pydairrerme people, traditional custodians of the land where fieldwork was carried out. We thank the different landowners for the permission provided to carry out fieldwork on their properties. We would also like to thank the volunteers who assisted with fieldwork.
References
- 1. Klein AM, Vaissiere BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C, Tscharntke T. Importance of pollinators in changing landscapes for world crops. Proceedings of the Royal Society B: Biological Sciences. 2007;274(1608):303–13. https://doi.org/10.1098/rspb.2006.3721
- 2. Ollerton J, Winfree R, Tarrant S. How many flowering plants are pollinated by animals? Oikos. 2011;120(3):321–6. https://doi.org/10.1111/j.1600-0706.2010.18644.x
- 3. Minachilis K, Kougioumoutzis K, Petanidou T. Climate change effects on multi-taxa pollinator diversity and distribution along the elevation gradient of Mount Olympus, Greece. Ecological Indicators. 2021;132:108335.
- 4. Welbergen JA, Klose SM, Markus N, Eby P. Climate change and the effects of temperature extremes on Australian flying-foxes. Proceedings of the Royal Society B: Biological Sciences. 2007;275(1633):419–25. https://doi.org/10.1098/rspb.2007.1385
- 5. English SG, Bishop CA, Wilson S, Smith AC. Current contrasting population trends among North American hummingbirds. Scientific Reports. 2021;11(1):1–9. https://doi.org/10.1038/s41598-021-97889-x
- 6. Koh I, Lonsdorf EV, Williams NM, Brittain C, Isaacs R, Gibbs J, Ricketts TH. Modeling the status, trends, and impacts of wild bee abundance in the United States. Proceedings of the National Academy of Sciences of the United States of America. 2016;113(1):140–5. pmid:26699460
- 7. Warren MS, Maes D, van Swaay CA, Goffart P, Van Dyck H, Bourn NA, Wynhoff I, Hoare D, Ellis S. The decline of butterflies in Europe: Problems, significance, and possible solutions. Proceedings of the National Academy of Sciences. 2021;118(2):e2002551117. pmid:33431566
- 8. Powney GD, Carvell C, Edwards M, Morris RK, Roy HE, Woodcock BA, Isaac NJ. Widespread losses of pollinating insects in Britain. Nature Communications. 2019;10(1):1018. pmid:30914632
- 9. Biesmeijer JC, Roberts SP, Reemer M, Ohlemuller R, Edwards M, Peeters T, Schaffers AP, Potts SG, Kleukers R, Thomas CD, Settele J, Kunin WE. Parallel declines in pollinators and insect-pollinated plants in Britain and the Netherlands. Science. 2006;313(5785):351–4. pmid:16857940
- 10. Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin WE. Global pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution. 2010;25(6):345–53. pmid:20188434
- 11. Vanbergen AJ, Insect Pollinators Initiative. Threats to an ecosystem service: pressures on pollinators. Frontiers in Ecology and the Environment. 2013;11(5):251–9. https://doi.org/10.1890/120126
- 12. Regan EC, Santini L, Ingwall-King L, Hoffmann M, Rondinini C, Symes A, Taylor J, Butchart SHM. Global Trends in the Status of Bird and Mammal Pollinators. Conservation letters. 2015;8(6):397–403. https://doi.org/10.1111/conl.12162
- 13. Potts S, Imperatriz-Fonseca V, Ngo H. IPBES: The assessment report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services on pollinators, pollination and food production.Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services: Bonn, Germany; 2016.
- 14. Dicks LV, Breeze TD, Ngo HT, Senapathi D, An J, Aizen MA, Basu P, Buchori D, Galetto L, Garibaldi LA. A global-scale expert assessment of drivers and risks associated with pollinator decline. Nature Ecology & Evolution. 2021:1–9. https://doi.org/10.1038/s41559-021-01534-9
- 15.
Anderson JR, Hardy EE, Roach JT, Witmer RE. A land use and land cover classification system for use with remote sensor data. Washington DC US Government Printing Office; 1976.
- 16. Millard J, Outhwaite CL, Kinnersley R, Freeman R, Gregory RD, Adedoja O, Gavini S, Kioko E, Kuhlmann M, Ollerton J. Global effects of land-use intensity on local pollinator biodiversity. Nature Communications. 2021;12(1):1–11. https://doi.org/10.1038/s41467-021-23228-3
- 17. Montero-Castaño A, Vilà M. Impact of landscape alteration and invasions on pollinators: a meta-analysis. Journal of Ecology. 2012;100(4):884–93. https://doi.org/10.1111/j.1365-2745.2012.01968.x
- 18. Winfree R, Aguilar R, Vazquez DP, LeBuhn G, Aizen MA. A meta-analysis of bees’ responses to anthropogenic disturbance. Ecology. 2009;90(8):2068–76. pmid:19739369
- 19. Kaluza BF, Wallace H, Heard TA, Klein AM, Leonhardt SD. Urban gardens promote bee foraging over natural habitats and plantations. Ecology and Evolution. 2016;6(5):1304–16. pmid:26848387
- 20. Brown J, Barton PS, Cunningham SA. Flower visitation and land cover associations of above ground-and below ground-nesting native bees in an agricultural region of south-east Australia. Agriculture, Ecosystems & Environment. 2020;295:106895. https://doi.org/10.1016/j.agee.2020.106895
- 21. Brown J, Groom SV, Rader R, Hogendoorn K, Cunningham SA. Land cover associations of wild bees visiting flowers in apple orchards across three geographic regions of southeast Australia. Agriculture, Ecosystems & Environment. 2022;324:107717. https://doi.org/10.1016/j.agee.2021.107717
- 22. Sagwe RN, Muya SM, Maranga R. Effects of land use patterns on the diversity and conservation status of butterflies in Kisii highlands, Kenya. Journal of Insect Conservation. 2015;19(6):1119–27.
- 23. Wurz A, Grass I, Lees DC, Rakotomalala AA, Sáfián S, Martin DA, Osen K, Loos J, Benasoavina E, Alexis T. Land‐use change differentially affects endemic, forest and open‐land butterflies in Madagascar. Insect Conservation and Diversity. 2022. https://doi.org/10.1111/icad.12580
- 24. Koh LP, Sodhi NS. Importance of reserves, fragments, and parks for butterfly conservation in a tropical urban landscape. Ecological Applications. 2004;14(6):1695–708. https://doi.org/10.1890/03-5269
- 25. Major RE, Christie FJ, Gowing G. Influence of remnant and landscape attributes on Australian woodland bird communities. Biological Conservation. 2001;102(1):47–66.
- 26. Davis RA, Gole C, Roberts JD. Impacts of urbanisation on the native avifauna of Perth, Western Australia. Urban Ecosystems. 2013;16(3):427–52. https://doi.org/10.1007/s11252-012-0275-y
- 27. Bain GC, MacDonald MA, Hamer R, Gardiner R, Johnson CN, Jones ME. Changing bird communities of an agricultural landscape: declines in arboreal foragers, increases in large species. Royal Society Open Science. 2020;7(3):200076. pmid:32269823
- 28. Baldock KC, Goddard MA, Hicks DM, Kunin WE, Mitschunas N, Osgathorpe LM, Potts SG, Robertson KM, Scott AV, Stone GN. Where is the UK’s pollinator biodiversity? The importance of urban areas for flower-visiting insects. Proceedings of the Royal Society B: Biological Sciences. 2015;282(1803):20142849. pmid:25673686
- 29. Rader R, Bartomeus I, Tylianakis JM, Laliberté E. The winners and losers of land use intensification: Pollinator community disassembly is non‐random and alters functional diversity. Diversity and Distributions. 2014;20(8):908–17.
- 30. Campbell AJ, Carvalheiro LG, Maués MM, Jaffé R, Giannini TC, Freitas MAB, Coelho BWT, Menezes C. Anthropogenic disturbance of tropical forests threatens pollination services to açaí palm in the Amazon river delta. Journal of Applied Ecology. 2018;55(4):1725–36. https://doi.org/10.1111/1365-2664.13086
- 31. Tam KC, Bonebrake TC. Butterfly diversity, habitat and vegetation usage in Hong Kong urban parks. Urban Ecosystems. 2016;19(2):721–33. https://doi.org/10.1007/s11252-015-0484-2
- 32. Proesmans W, Bonte D, Smagghe G, Meeus I, Decocq G, Spicher F, Kolb A, Lemke I, Diekmann M, Bruun HH. Small forest patches as pollinator habitat: oases in an agricultural desert? Landscape Ecology. 2019;34(3):487–501.
- 33. Harrison T, Gibbs J, Winfree R. Forest bees are replaced in agricultural and urban landscapes by native species with different phenologies and life‐history traits. Global Change Biology. 2018;24(1):287–96. pmid:28976620
- 34. Baude M, Kunin WE, Boatman ND, Conyers S, Davies N, Gillespie MA, Morton RD, Smart SM, Memmott J. Historical nectar assessment reveals the fall and rise of floral resources in Britain. Nature. 2016;530(7588):85. pmid:26842058
- 35. Nicolson SW, Wright GA. Plant–pollinator interactions and threats to pollination: perspectives from the flower to the landscape. Functional Ecology. 2017;31(1):22–5. https://doi.org/10.1111/1365-2435.12810
- 36. Paton DC. Disruption of bird‐plant pollination systems in southern Australia. Conservation Biology. 2000;14(5):1232–4.
- 37. Kral-O’Brien KC O’Brien PL, Hovick TJ, Harmon JP. Meta-analysis: Higher Plant Richness Supports Higher Pollinator Richness Across Many Land Use Types. Annals of the Entomological Society of America. 2021. https://doi.org/10.1093/aesa/saaa061
- 38. Campbell AJ, Carvalheiro LG, Gastauer M, Almeida-Neto M, Giannini TC. Pollinator restoration in Brazilian ecosystems relies on a small but phylogenetically-diverse set of plant families. Scientific Reports. 2019;9(1):1–10. https://doi.org/10.1038/s41598-019-53829-4
- 39. Sabatino M, Rovere A, Meli P. Restoring pollination is not only about pollinators: Combining ecological and practical information to identify priority plant species for restoration of the Pampa grasslands of Argentina. Journal for Nature Conservation. 2021;61:126002. https://doi.org/10.1016/j.jnc.2021.126002
- 40. Klatt BK, Nilsson L, Smith HG. Annual flowers strips benefit bumble bee colony growth and reproduction. Biological Conservation. 2020;252:108814. https://doi.org/10.1016/j.biocon.2020.108814
- 41. Campbell AJ, Wilby A, Sutton P, Wäckers FL. Do sown flower strips boost wild pollinator abundance and pollination services in a spring-flowering crop? A case study from UK cider apple orchards. Agriculture, Ecosystems & Environment. 2017;239:20–9. https://doi.org/10.1016/j.agee.2017.01.005
- 42. Threlfall CG, Mata L, Mackie JA, Hahs AK, Stork NE, Williams NS, Livesley SJ. Increasing biodiversity in urban green spaces through simple vegetation interventions. Journal of Applied Ecology. 2017;54(6):1874–83. https://doi.org/10.1111/1365-2664.12876
- 43. Isaacs R, Williams N, Ellis J, Pitts-Singer TL, Bommarco R, Vaughan M. Integrated crop pollination: combining strategies to ensure stable and sustainable yields of pollination-dependent crops. Basic and Applied Ecology. 2017;22:44–60. https://doi.org/10.1016/j.baae.2017.07.003
- 44. Brown M, Duncan F. The vegetation of Tasman Peninsula. Papers and Proceedings of the Royal Society of Tasmania. 1986:33–50.
- 45.
Department of Primary Industries Parks Water and Environment. Tasmanian Land Use 2015. 2015.
- 46.
Australian Bureau of Statistics. Region summary: Tasman (M) 2021 [cited 2020 24/04/2020]. Available from: https://dbr.abs.gov.au/region.html?lyr=lga&rgn=65210.
- 47.
Department of Primary Industries Parks Water and Environment. TASVEG 4.0. Tasmanian Vegetation Monitoring and Mapping Program, Natural and Cultural Heritage Division. 2020.
- 48. Steel D, McLaren C. Design and analysis of repeated surveys. Centre for Statistical and Survey Methodology, University of Wollongong, Working Paper 11–08, 2008, 13p. 2008. https://ro.uow.edu.au/cssmwp/10
- 49. BirdLife. Survey Techniques 2021 [22/04/2021]. Available from: https://birdata.birdlife.org.au/survey-techniques.
- 50. Ford HA, Paton DC. The comparative ecology of ten species of honeyeaters in South Australia. Australian Journal of Ecology. 1977;2(4):399–407. https://doi.org/10.1111/j.1442-9993.1977.tb01155.x
- 51.
Simpson K, Day N. Field guide to the birds of Australia.Viking: Camberwell Victoria; 2010.
- 52.
Wiltshire R, Potts B. EucaFlip-Life-size guide to the eucalypts of Tasmania.University of Tasmania2007.
- 53. Nicolle D. Classification of the eucalypts (Angophora, Corymbia and Eucalyptus) Version 4. https://dn.com.au/Classification-Of-The-Eucalypts.pdf. 2015.
- 54.
Dunkerley GM, Ford H, Danthanarayana W. The fuscous honeyeater: food resources and the bird community [Dissertation]: University of New England; 1990.
- 55. Woinarski J, Cullen J. Distribution of invertebrates on foliage in forests of south‐eastern Australia. Australian Journal of Ecology. 1984;9(3):207–32. https://doi.org/10.1111/j.1442-9993.1984.tb01359.x
- 56. Prendergast KS, Menz MHM, Dixon KW, Bateman PW. The relative performance of sampling methods for native bees: an empirical test and review of the literature. Ecosphere. 2020;11(5):e03076. https://doi.org/10.1002/ecs2.3076
- 57.
Packer L, Darla-West G. Bees: how and why to sample them. Measuring arthropod biodiversity: Springer; 2021. p. 55–83.
- 58.
Sentinel Online. Sentinel-2 Spatial Resolutions 2021 [22/04/2021]. Available from: https://sentinel.esa.int/web/sentinel/user-guides/sentinel-2-msi/resolutions/spatial.
- 59. Diengdoh VL, Ondei S, Hunt M, Brook BW. A validated ensemble method for multinomial land-cover classification. Ecological Informatics. 2020;56:101065. https://doi.org/10.1016/j.ecoinf.2020.101065
- 60. QGIS Development Team. QGIS Geographic Information System. QGIS Association. 3.12.1 ed2021.
- 61. Kuhn M. caret: Classification and Regression Training. R Package version 60–92. 2021. https://CRAN.R-project.org/package=caret
- 62. Foody GM. Status of land cover classification accuracy assessment. Remote Sensing of Environment. 2002;80(1):185–201. https://doi.org/10.1016/S0034-4257(01)00295-4
- 63. Allouche O, Tsoar A, Kadmon R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). Journal of Applied Ecology. 2006;43(6):1223–32. https://doi.org/10.1111/j.1365-2664.2006.01214.x
- 64.
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 4.0.5 ed. Vienna, Austria: R Foundation for Statistical Computing; 2021.
- 65. Smith AC, Fahrig L, Francis CM. Landscape size affects the relative importance of habitat amount, habitat fragmentation, and matrix quality on forest birds. Ecography. 2011;34(1):103–13.
- 66. Greenleaf SS, Williams NM, Winfree R, Kremen C. Bee foraging ranges and their relationship to body size. Oecologia. 2007;153(3):589–96. pmid:17483965
- 67. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. Journal of Statistical Software. 2014. https://doi.org/10.18637/jss.v067.i01
- 68. Zhang D. Rsq: R-Squared and Related Measures. R package version 25. 2022. https://CRAN.R-project.org/package=rsq
- 69. Kavanagh RP, Stanton MA, Herring MW. Eucalypt plantings on farms benefit woodland birds in south‐eastern Australia. Austral Ecology. 2007;32(6):635–50.
- 70. Herrera CM. Complex long‐term dynamics of pollinator abundance in undisturbed Mediterranean montane habitats over two decades. Ecological Monographs. 2018. https://doi.org/10.1002/ecm.1338
- 71. Daniels B, Jedamski J, Ottermanns R, Ross-Nickoll M. A “plan bee” for cities: Pollinator diversity and plant-pollinator interactions in urban green spaces. PLoS One. 2020;15(7):e0235492. pmid:32667935
- 72. Phillips BB, Gaston KJ, Bullock JM, Osborne JL. Road verges support pollinators in agricultural landscapes, but are diminished by heavy traffic and summer cutting. Journal of Applied Ecology. 2019;56(10):2316–27. https://doi.org/10.1111/1365-2664.13470
- 73. Kennedy CM, Lonsdorf E, Neel MC, Williams NM, Ricketts TH, Winfree R, Bommarco R, Brittain C, Burley AL, Cariveau D. A global quantitative synthesis of local and landscape effects on wild bee pollinators in agroecosystems. Ecology Letters. 2013;16(5):584–99. pmid:23489285
- 74. Steffan-Dewenter I, Münzenberg U, Bürger C, Thies C, Tscharntke T. Scale‐dependent effects of landscape context on three pollinator guilds. Ecology. 2002;83(5):1421–32. https://doi.org/10.1890/0012-9658(2002)083[1421:SDEOLC]2.0.CO;2
- 75. Bartholomée O, Aullo A, Becquet J, Vannier C, Lavorel S. Pollinator presence in orchards depends on landscape-scale habitats more than in-field flower resources. Agriculture, Ecosystems & Environment. 2020;293:106806. https://doi.org/10.1016/j.agee.2019.106806
- 76. Keast A. Seasonal movements in the Australian honeyeaters (Meliphagidae) and their ecological significance. Emu-Austral Ornithology. 1968;67(3):159–209.
- 77. De Kok R, West J. A revision of the genus Pultenaea (Fabaceae). 3. The eastern species with recurved leaves. Australian Systematic Botany. 2004;17(3):273–326. https://doi.org/10.1071/SB02028
- 78. Ogilvie JE, Zalucki JM, Boulter SL. Pollination biology of the sclerophyllous shrub Pultenaea villosa Willd.(Fabaceae) in southeast Queensland, Australia. Plant Species Biology. 2009;24(1):11–9. https://doi.org/10.1111/j.1442-1984.2009.00235.x
- 79. Armstrong J. Biotic pollination mechanisms in the Australian flora—a review. New Zealand Journal of Botany. 1979;17(4):467–508. https://doi.org/10.1080/0028825X.1979.10432565
- 80. Hingston AB, Mc Quillan PB. Are pollination syndromes useful predictors of floral visitors in Tasmania? Austral Ecology. 2000;25(6):600–9. https://doi.org/10.1111/j.1442-9993.2000.tb00065.x
- 81. Andersen AN, New T. Insect inhabitants of fruits of Leptospermum, Eucalyptus and Casuarina in southeastern Australia. Australian Journal of Zoology. 1987;35(4):327–36. https://doi.org/10.1071/ZO9870327
- 82. Stephens J, Molan PC, Clarkson BD. A review of Leptospermum scoparium (Myrtaceae) in New Zealand. New Zealand Journal of Botany. 2005;43(2):431–49. https://doi.org/10.1080/0028825X.2005.9512966
- 83. Fountain-Jones NM, Jordan GJ, Baker TP, Balmer JM, Wardlaw T, Baker SC. Living near the edge: being close to mature forest increases the rate of succession in beetle communities. Ecological Applications. 2015;25(3):800–11. pmid:26214924
- 84. Antoine CM, Forrest JR. Nesting habitat of ground‐nesting bees: a review. Ecological Entomology. 2021;46(2):143–59. https://doi.org/10.1111/een.12986
- 85. Urbanowicz C, Muñiz PA, McArt SH. Honey bees and wild pollinators differ in their preference for and use of introduced floral resources. Ecology and Evolution. 2020;10(13):6741–51. pmid:32724547
- 86. Plascencia M, Philpott S. Floral abundance, richness, and spatial distribution drive urban garden bee communities. Bulletin of Entomological Research. 2017;107(5):658–67. pmid:28245886
- 87. Wimmer J, Towsey M, Roe P, Williamson I. Sampling environmental acoustic recordings to determine bird species richness. Ecological Applications. 2013;23(6):1419–28. pmid:24147413
- 88. Steen R. Diel activity, frequency and visit duration of pollinators in focal plants: in situ automatic camera monitoring and data processing. Methods in Ecology and Evolution. 2017;8(2):203–13. https://doi.org/10.1111/2041-210X.12654