Native flora receive more visits than exotics from bees, especially native bees, in an urbanised biodiversity hotspot

Introduction

Bees (Hymenoptera, Anthophila) are highly reliant on flowering resources – pollen and nectar – for nutrition at all stages of their life-cycle. However, not all flowers are equally suitable for bees (Matteson and Langellotto 2011). Flowers can have inferior nectar or pollen qualities, quantities or compositions (Potts et al. 2003). For example, some contain toxic compounds (e.g. alkaloids, saponins, cardiac glycosides and cyanogenic glycosides present at concentrations that can cause harmful effects to bees (Detzel and Wink 1993; Tiedeken et al. 2014)). Other flowers do not provide the correct nutritional profile for bee larval development (Jeannerod et al. 2022) (e.g. some pollen species fail to support successful larval development; such as Tanacetum pollen for generalist megachilids (Sedivy et al. 2011)). Importantly, some native bee taxa have co-evolved with a restricted range of flora, and if their host flowers are not present, they will not be present (Waser and Ollerton 2006).

Conversion of natural landscapes to anthropogenic ones, and associated loss of native wildflowers (Brown and Paxton 2009), to be replaced by exotic flora (Vanbergen et al. 2018), is considered a major threat to wild bees. Urbanisation heavily contributes to loss of native vegetation, with cities expected to expand across the globe (Güneralp et al. 2013). However, cities can also present benefits to bees. Compared with agricultural and some natural habitats, cities often have the highest species richness of plants (McKinney 2008). Hülsmann et al. (2015) found that for bumblebees, an increased plant diversity associated with urbanisation compensated for the negative effects of large areas of impervious surfaces that accompany urbanisation.

However, increased flowering species may not result in benefits for bees. Flowering plants in urban gardens include horticulturally modified varieties unsuitable for bees e.g. owing to petal modifications that reduce accessibility to floral resources (Corbet et al. 2001). For example, horticultural modifications can change features that have co-evolved with insect visitors, leading to a loss in attractiveness or nutritional value. This can include loss of a functional spur (and therefore loss of nectar), whilst alterations in size, form or colour of the corolla that can obstruct access to nectar/pollen, come at the cost of reducing investment in nectar/pollen quality or quantity, or result in disruption of co-evolved preferences (Comba et al. 1999).

Therefore, the general finding of greater plant diversity in cities (McKinney 2008) may still result in a loss of preferred floral species. Specialists are especially vulnerable to urbanisation if vital mutualisms are severed with native flora reductions (Frankie et al. 2005; Prendergast et al. 2022a). Exotic flowering plants may end up benefitting exotic generalists – in particular, European honey bees (Apis mellifera), at the expense of native bees (Richardson et al. 2000; Prendergast et al. 2021). In natural habitats it has been demonstrated that native bees tend to prefer native flora, whereas a plant community dominated by exotic, invasive flora is favoured by the exotic honey bees (Abe et al. 2011). The range of floral resources a species uses is also constrained by competition; if honey bees outcompete native bees, native bees will be excluded from certain resources (Prendergast et al. 2023). This may lead to a narrowing of niche breadth (Smith 1982; Hudewenz and Klein 2015). However, there exists evidence that exotic flora may support a greater diversity of bees than native flora (Petanidou et al. 2008; Montero-Castaño and Vilà 2017). Horticultural varieties with large floral displays and high nectar crops may be even more attractive to bees (Comba et al. 1999).

The flowering species composition may be of overriding importance for bee assemblages, including in urban areas (Hülsmann et al. 2015). In cities, much of the original flowering community is lost and remains as fragmented wildland remnants (Stenhouse 2004), with other greenspaces being determined by practical and aesthetic preferences of the homeowner or manager (Hoyle 2021), such that urban flower composition represents a diversity of flora based on both the natural pool of flowers, as well as those introduced for gardening or urban landscaping purposes, or unintentionally as weeds (Cane 2005). Which flower species are present, and their particular nectar and pollen chemistry, impacts bee populations and communities This importance of flower composition has been supported by numerous studies – both manipulative and mensurative (Weiner et al. 2010; Eckhardt et al. 2014; Vanderplanck et al. 2014). Wild bees in Australia can be predicted to be particularly sensitive to flower origin and composition: as Australia is a continent that has been biogeographically isolated from the rest of the globe for millions of years, many species are endemic, and have co-evolved in isolation (Hopper 2009). Examples include the diverse and predominantly oligolectic Euryglossinae, and Stenotritidae (Houston 2018). In contrast, generalist bees like the introduced European honey bee Apis mellifera may be indifferent, or even capitalise on, increased representations of exotic flora, especially if exotic plant species offer good quality sources of nutrition (Miller et al. 2015; Drossart et al. 2017).

The response of bees to flower resources can be quantified in terms of niche breadth. Numerous indexes exist in the ecological literature, but generally relate how many resource states a species uses out of all resource states available in a given time and space (Krebs 2009). Such metrics however ignore taxonomic identity, which is critical in terms of categorising bees in the mellitological literature. Bees are categorised as specialists if they forage for pollen from flowers in just one family across the species’ range, or as oligolectic, if they forage for pollen from flowers across multiple families (Michener 2007).

Supporting and conserving bee populations requires identifying which flower species – be they exotic or native – are major resources. Lists and schemes for supplying evidence-based, region-specific bee-friendly flowers to support native bees in cities is an area of great interest. Most attention to date into creating pollinator habitats however has focused on agricultural areas, and on managed, generalist pollinators (Decourtye et al. 2010). Moreover, success of such schemes may be limited to generalist bees (Wood et al. 2015, 2017), and flowers marketed as ‘pollinator-friendly’ may not be supported by empirical evidence or provide food to native bees (Garbuzov et al. 2017).

Understanding how the shift from a native vegetation landscape to that with high diversity of exotics influences bee assemblages is important as we continue to design and manage urban greenspaces, given the increasing proportion of urbanised landscapes across the globe (Prendergast 2022a). Identifying major nutritional foraging resources for wild bees can allow targeted plantings, and furthermore, unlike in production dominated landscapes, we have a greater opportunity to actively design gardens with native pollinators in mind. This is important due to wild bee declines, increasing urbanisation, and the potential for urban habitats to represent either conservation hubs or areas of bee decline depending on their management (Prendergast 2020a). Nurseries and gardens, and the flora they contain, are a major means by which citizens can improve managed greenspaces for bee conservation under urbanisation (Egerer et al. 2020; Prendergast 2020a, 2022a). Discerning which plants are particularly attractive for native bees, i.e. ‘bee-friendly’, is of great interest from both a conservation and socioeconomical–ecological perspective (Wollaeger et al. 2015), yet empirical data to support flowering plants that are key ‘bee-friendly’ flowers is currently largely absent (Turo and Gardiner 2019; Prendergast 2020a, 2020b).

This study sought to clarify foraging niches of native bees and honey bees in an urban region with respect to plant community and plant origin of the flowers visited, and whether this differed between native vegetation remnants and residential gardens. Key plants supporting the majority of bees in this urbanised biodiversity hotspot were listed to assist future management practices.

It was hypothesised that:

Methods

Results

Niche breadth

Smith’s index of niche breadth

In both years, niche breadth varied widely (Figs 1 and 2). In Year One, the narrowest niche breadth was 0.015 (Lipotriches at a residential site in November), whereas the maximum niche breadth was 1.00 (honey bees at a residential site in February). Niche breadth similarly covered a large range in the second year, ranging from a minimum of 0.002 (Lasioglossum at a residential site in November), to a max of 0.987 (Hylaeinae bees at a bushland site in November). With the median niche breadth being 0.470 in Year One, and 0.399 in Year Two, overall the bee assemblages in this urbanised region have fairly narrow niche breadths.

Fig. 1.

Niche breadth (mean ± standard error) of native bees and honey bees (Apis mellifera) according to habitat type in (a) Year One and (b) Year Two. ΔAICc, corrected Akaike information criterion.

Fig. 2.

Niche breadths of native bee taxa and the honey bee (Apis mellifera) in (a) Year One and (b) Year Two. The box represents the range between the 1st and 3rd quartiles, and the whiskers represent the values above and below the interquartile range.

Niche breadth of the overall bee community did not differ between urban vs residential garden habitats in the first year of surveys (mean bushland remnant: 0.551 ± 0.027 vs mean residential garden: 0.437 ± 0.023; X² = 2.70, P = 0.100), however in the second year niche breadth was significantly different between habitat types for the overall bee assemblage, being significantly greater (broader) in native vegetation than residential sites (mean bushland remnant: 0.505 ± 0.353 vs mean residential garden: 0.353 ± 0.02; X² = 5.78, P = 0.016) (Fig. 1).

No significant differences in niche breadth were found between honey bees and the group of native bees in Year One (X² = 2.24, P = 0.134) (Fig. 1a). However, there was a significant interaction between native/honey bee niche breadth with habitat type (X² = 12.0, P < 0.001), such that differences in niche breadth between native bees and honey bees varied depending on habitat type. Niche breadth of native bees was higher in bushland remnants than in residential gardens, whereas there was no difference between habitats for honey bee niche breadth (Fig. 1a). Native bees and honey bees did not differ in niche breadth in bushland remnants, however honey bees had significantly broader niche breadths than native bees in residential gardens (Fig. 1a). In Year Two there was no interaction effect (X² = 0.109, P = 0.742), with honey bees have significantly broader niche breadths than native bees across habitats (Fig. 1b).

Niche breadth did not vary significantly by taxa in Year One (X² = 20.6, P = 0.057) (Fig. 2a), whereas niche breath varied significantly by bee taxa in Year Two (X² = 56.3, P < 0.0001) (Fig. 2b). In the second year pairwise differences were significant between Amegilla–Lipotriches (P = 0.007) and Amegilla–Homalictus (P < 0.001), Homalictus–Exoneura (P = 0.003), Homalictus–Lasioglossum (P = 0.003), Homalictus–Hylaeinae (P = 0.030) and Homalictus–Megachile (P = 0.002) (Appendix 2).

In the first year of surveys, native bee niche breadth was significantly positively correlated with honey bee abundance (estimate = 0.261, s.e. = 0.061, z-value = 4.25, P < 0.001, R² = 0.357). In the second year, however, niche breadth was unrelated to honey bee abundance (estimate = −0.006, s.e. = 0.048, z-value = −0.121, P = 0.905, R² < 0.001).

Lecty niche breadth

The number of plant families visited by bees ranged from one to a maximum of 12 per taxon per survey. There was a significant main effect of habitat (X² = 53.7, P < 0.0001), bee taxon (X² = 503, P < 0.0001), and a significant interaction between habitat × bee taxon (X² = 41.0, P < 0.0001) (Fig. 3). There was no interaction with year (X² = 0.002, P = 0.982). There were fewer plant species visited in bushland remnants (1.6 ± 0.06) than residential gardens (2.4 ± 0.14). The average number of families visited per survey was greater for honey bees than all other native bee taxa (P < 0.001). Megachilidae also visited a greater number of families on average compared with Euryglossinae (P = 0.02) and Hylaeinae (P = 0.02). No other pairwise comparisons between the native bee taxa were significantly different (P > 0.05). Whilst there were clear differences by habitat for honey bees, Amegilla, and Exoneura, these were not apparent for the other native bee taxa (Fig. 3). Indeed as a whole, native bees did not differ in the number of families visited between bushland remnants and residential gardens (X² = 0.39, P = 0.530), whereas honey bees visited more plant families in residential gardens (X² = 13.9, P = 0.0002).

Fig. 3.

Niche breadth in terms of average number of families visited (i.e. lecty) of bee taxa in bushland remnants and residential gardens. Niche breadth expressed as box and whisker plots.

Discussion

Native bees, in abundance and diversity, are more strongly associated with the relative availability of native plant species than honey bees. Community composition was important in influencing native bee abundance and species richness, but not honey bee abundance. In line with predictions, native bees visited more native flowers than exotic flowers. Whereas, contrary to predictions, the introduced European honey bee also followed this pattern, native bees were more strongly associated with native flowers. Native bees also appeared to be impacted by honey bee competition, as reflected by altering niche breadth (Fig. 4). Thus native bees had stronger preferences for native flora than honey bees, whilst honey bees foraged at greater abundances, and on a broader range, of exotic flora than did native bees.

Fig. 4.

Schematic of main findings of the preference for native flowers by both native bees and European honey bees (Apis mellifera), as a subset of total available flora, and the influence of A. mellifera on native bee niche breadth. Arrows connect variables, with +v meaning a positive association. Photographs and illustrations by K. S. Prendergast.

These results add to the growing body of evidence that native plants are preferred over exotic plants, and native plants are especially important for native bees (Pawelek et al. 2009; Rollin et al. 2013). The preference for visiting native flowers in this study can be explained by how the native bee fauna of Australia have co-evolved with the flora for millions of years in isolation (Hopper 2009). An urban pollinator study on the east coast of Australia likewise emphasised the importance of native vegetation, as opposed to a high diversity of flowering plants (Threlfall et al. 2015). Other studies that have failed to support the idea that non-native flowers are relatively unattractive to bees relative to native flowers tend to be from regions where there has been the opportunity for dispersal and interchange of biota e.g. Europe and the USA (Cane 2005; MacIvor et al. 2014, Salisbury et al. 2015; Lowenstein et al. 2019). It may be that biogeographic history of a region and the degree of co-evolution of the bees and flowering plant species will influence the relative attractiveness of introduced flora. This has been suggested by the theory that conservation of biodiversity requires different approaches between old, climatically-buffered, infertile landscapes (OCBILs), such as that in south-west Western Australia, compared with young, often disturbed, fertile landscapes (YODFELs), located mainly in the Northern Hemisphere, and from which most conventional ecological theory and practice is derived (Hopper 2009). As such, my results suggest that the biogeographic system is important when devising strategies for pollinator conservation, and ecosystems that have evolved in relative geological isolation, including Australia, as well as New Zealand, South Africa’s Western Cape and Madagascar, may need different considerations from strategies developed for pollinators in the northern Hemisphere. It is worth noting that there is a major research bias towards bee conservation in the global north, with the global south being substantially under-represented (see for example, Prendergast et al. 2022a on urban bee research). Supported by the OCBIL theory of the influence biogeographic history of a landscape has on its biota (Hopper 2009), the dominance of research of pollinator preference in the norther hemisphere may not be generalisable on a global scale, pointing to greater research attention needed in these under-represented regions. In the context of native bee conservation in urbanised Australia, unlike in the Northern Hemisphere, many native flowering plant species are not herbs but rather trees or shrubs – which tend to be coniferous in the Northern Hemisphere rather than mass-blooming like the Australian eucalypts (Armstrong 1979). In addition, Australia has been biogeographically isolated from other continents, allowing a long period of co-evolution with minimal disruptions from glaciation events in old, climatically-buffered infertile landscapes (Hopper et al. 2021). This therefore creates present ecological interactions that have a different evolutionary history from plant–pollinator networks in the Northern Hemisphere.

Although the hypothesis of a preference for native flowers by native bees was supported, the prediction that honey bees, in contrast, would prefer exotic flowers was not. The highly polylectic behaviour of honey bees means they are less constrained to foraging on species from their region of origin. In addition, as with native bees, honey bees foraged prolifically on Myrtaceae – six (Year Two) and eight (Year One) of the 10 most-visited plants being from this family. This preference for native flowers, driven by Myrtaceae, can be linked to how most Myrtaceae – the dominant native plant family in Australia – have highly-accessible open-cup flowers, they flower prolifically, and produce high quantities of nectar and pollen, and furthermore the main visited species here provide high-quality resources with high antibacterial properties, and a high protein content, with beneficial impacts on bee health (Coleman 1962; Frias et al. 2016; Manning 2011).

The clear preference of both native bees and honey bees for foraging on native flora was further supported when looking at the 10 most-visited species, which comprised over 70% of visits. These 10 most-visited flower species provide clear guidance for citizens and policymakers, as just by planting these 10 species, the majority of native bees and honey bees can be supported. These top 10 flowers for native bees were from the two most common plant families, indicating the co-evolution between dominant plants in Australia’s ecosystems and their insect visitors. Some overlap in the most preferred flora between honey bees and native bees suggests competition may occur over these resources, especially if honey bees are allowed to increase in abundance (Prendergast et al. 2022a). This is not to suggest that other flora should be excluded from consideration in plantings, and planting designs should also include hosts of known specialists in the region. As the most-visited species by honey bees in the second year – Schinus terebinthifolius – is an environmental introduced weed that is dioecious and insect-pollinated (Hussey et al. 1997; Cuda et al. 2004), this is cause for concern. These two introduced species may be favouring each other’s success, as has been found for other invasive weeds and A. mellifera in Australia (Goulson and Derwent 2004).

Identifying the top-10 most-visited plants is valuable from a management perspective (Hegland et al. 2010). Given that 70% of visits were to 10 highly attractive flora, this allows gardeners, land managers and urban planners in the region a guide for restoration, urban greening and gardening, since the majority of bees can be catered for with this selection. Further research is required if the aim is to protect a particular species of interest; however, with systematic bee biodiversity surveys only recently being conducted (Prendergast 2020a, 2020c, 2021a, 2021b, 2021c, 2022a), many species remaining undescribed (see the recent new species description of the oligolectic Leioproctus zephyr (Prendergast 2022c), and only six species having been formally assessed for legislative protection (Department of Biodiversity Conservation and Attractions 2022), these top-10 flowers provide a heuristic (Rollings and Goulson 2019).

It would be of interest to evaluate which plant traits were associated with particularly attractive plant species (and conversely, unattractive plant species). Due to the hundreds of plant species present, and lack of funding available to measure many plant traits, this was not possible. Potential variables include flower shape (Mačukanović-Jocić et al. 2011), petal colour (de Camargo et al. 2019), flower colour and pattern (de Ibarra et al. 2015; Dyer et al. 2015), nectar volume and sugar concentration and composition (Nicolson 2022), pollen nutritional composition (Nicholls and Hempel de Ibarra 2017), floral scent (Dötterl and Vereecken 2010), and secondary plant metabolites in nectar or pollen (Stevenson 2020).

Previous research in this study system found that native bee abundance and species richness were positively related to the proportion of native flora at a site, yet – unexpectedly – negatively associated with native flower species richness (Prendergast et al. 2022b). It was hypothesised this was a result of the restrictive foraging preferences of native bees in the south-west Australian biodiversity hotspot (Prendergast et al. 2022b). This explanation is supported by the current results where overall, number of flowering plant species visited were fairly low relative to the total number of plant species available at a site (Prendergast 2020b), and reflected in the calculations of niche breadth, which explicitly considers the resources used in relation the total resources present. Many flowering plants present were not visited, even those that had a high relative abundance of flowers (Prendergast 2020b), indicating that general calls to ‘plant more flowers’ for bees may be misguided if flower identity is not taken into account. Instead, what is required are particular preferred plant species at high relative abundances, indicated by the consistent positive association between proportion of native flora and native bee abundances (Prendergast 2020b).

Foraging patterns differed between the residential garden habitats and the native vegetation remnant habitats, indicating that environmental context is influential. No previous studies have compared flower preferences of native bees based on whether they are foraging in urban remnant native vegetation patches or in gardens within a city. However, there is evidence that landscape context influences foraging behaviour in other landscapes. For example, floral plantings have been found to attract bumblebees to a greater extent in landscapes that have greater proportions of intensively farmed areas (Carvell et al. 2011). In an urban corridor, greater habitat heterogeneity in the surrounding landscapes was associated with bees exhibiting higher foraging specialisation, whereas greater semi-natural habitat was associated with fewer visits to flowering species (Pardee et al. 2023). Research has also found that varying levels of urbanisation and presumably stressors associated with urbanisation intensity can modulate the attractiveness of native vs exotic plant species (Buchholz and Kowarik 2019).

The hypothesis that honey bees had a wider niche breadth than native bees was partially supported: honey bees had significantly greater niche breadth in the second year of study, and in residential gardens in the first. A wider niche breadth is consistent with honey bees being super-generalists (Giannini et al. 2015). Interannual variation in results may relate to the extra months surveyed in the second year: by extending the period of observations, this included flowering plants whose phenology was not captured in the first year, meaning there were additional resources options to forage on in the second year. The role of the range of resources available in influencing honey bee niche breadth was also evident in how, in the first year, honey bees had wider niche breadths only in residential gardens, which offered a greater range of potential resources that the super-generalist honey bee could exploit compared with bushland remnants. Niche breadth varied considerably within and between native bee taxa, with sometimes counter-intuitive results i.e. the wide niche breadth of euryglossines (oligoleges), and the narrow niche breadth of Amegilla (polyletic) (Houston 2000). Although there are many indexes that have been created, there is no consensus over which one is ‘best’ (Krebs 2009). Whilst the formula used here has advantages in that it takes into account the availability of resources present, it has drawbacks in being sensitive to the number of individuals observed foraging in each taxon, and does not control for phylogeny i.e. euryglossines foraged on a number of species that were in just a few plant families, whereas Amegilla foraged on plant families encompassing a wide taxonomic range (Prendergast 2020b). The impracticality of identifying most native bees to species-level in the field, and the rarity of most species, make statistical calculations of niche breadth of most species impossible; so for this purpose native bees were grouped into taxonomic categories. This does mean that whilst native bees were represented by >1 species, honey bees were represented by a single species. Yet, despite native bee taxonomic groups representing two to tens of species based on reliable identification by observation in the field, their niche breadth was in fact narrower than the single honey bee species in most instances.

Niche breadth also was calculated from total flower visits. When expressed in terms of lecty, this focuses solely on pollen, and thus is largely based on female foragers (Michener 2007). Whilst bees have preferences for nectar (Parachnowitsch et al. 2019), species are often more flexible in their nectar than pollen foraging (Cane and Sipes 2006) – explaining broader niches of kleptoparasitic species that do not need to collect pollen for their brood (Michener 2007). Determining the sex of a bee in the field was not possible for most species (especially not the tiny Euryglossinae and Hylaeinae), nor was distinguishing between pollen vs nectar foraging (again, likely for these small colletids, which comprised a large proportion of all visits). Consequently, to utilise the full dataset, these finer details could not be assessed but should be borne in mind when interpreting these results.

The hypothesis of niche breadth being broader in the residential gardens than native vegetation remnants was not supported; indeed, the opposite occurred. Although a greater range of floral species available in residential gardens could be inferred to allow bees to expand the range of flowers they forage on, my finding of a narrowing of niche breadth in residential gardens relative to bushland remnants can be explained by the fact that many native bee species are restrictive in the flora they visit. Thus, of the flora they actually will forage on, there is a broader suitable range of preferred plant species present at native vegetation sites, contrasting with a narrow range of preferred flora present at residential sites.

The hypothesis that high honey bee abundances would cause native bee taxa to decrease their niche breadth, predicted as a means of avoiding competition, was not supported. Instead, in Year One, the niche breadth of native bees was broader when honey bee abundances were higher. This may arise if native bees under competition with honey bees expand their niche breadth (Fontaine et al. 2008; Pianka 2011). If native bees forage on a wider range of resources, especially those not utilised by honey bees, and focus on resources that are in relatively high abundance, then the intensity of competition is reduced. (This consideration is accounted for in Smith’s measure of niche breadth used here). If native bees are excluded from preferred flowers and must switch to and forage on a greater range of potentially less suitable flowering resources, this may have fitness consequences (Pyke 1984). There is also the possibility that under higher honey bee abundances, specialised native bees were outcompeted, leaving a community of largely generalist species.

It is of interest to compare niche breadth, in terms of an ecological index of niche breadth, with niche breadth expressed simply in terms of how many plant families were visited. Smith’s index considers both the relative visits to different plant species, irrespective of phylogeny, and their relative availability in the environment. The advantages of Smith’s measure is that it means that the number of visits take into account how much of a particular resource is present, but ignore taxonomic identity. When looking at the number of families visited, results were in line with predictions of the super-generalist honey bee visiting more families than native bee taxa, especially in residential gardens, whereas the oligolectic Hylaeinae and Euryglossinae visited very few families. The flexibility of polylectic taxa such as the honey bee, Megachilidae, Amegilla, and Exoneura, also enabled them to expand their niche breadth in residential gardens, whereas more specialised taxa did not change their niche breadths according to habitat type. The two indexes thus provide a complementary picture of foraging specialisation, depending on whether the focus is on relative resource specialisation vs taxonomic specialisation.

Some hypotheses were supported during one year, but not the other. Interannual variation in ecological interactions is not unusual (e.g. Roubik et al. 1986; Prendergast et al. 2023; Prendergast and Ollerton 2022). These findings indicate that niche breadth and competitive pressure were not invariant across time. Long-term monitoring is required to determine how consistently the patterns observed here arise, and tease apart potential causes such as climatic influences (e.g. Thomson 2016). The finding of native bees increasing their niche breadth under increasing honey bee densities in the first but not second year may relate to how in the second year, October was included – a month where many flowering plants are blooming (Lambers 2014), permitting native bee species to forage on other resources when honey bee abundances were high as a mechanism to enable co-existence under resource competition (Heinrich 1976). Although I found that native bees may expand their niche breadth when honey bee abundances are higher, at least under certain conditions, further research is required to verify this experimentally, and determine if this has fitness consequences (Prendergast et al. 2023).

The data here revealed that community composition strongly influences the native bee assemblage, and thus the identity of which plants occur in urban greenspaces will be important in aiding, or conversely, threatening, the preservation of thriving native bee assemblages. The list of 10 most-visited plants can guide the public and land managers in how to create flowering resources for bees in this region. Obtaining region-specific empirical data for other urban regions would be advisable, so that they likewise have an evidence-based guide for selecting flowers that will maximise bees.

That specific floral resource composition is a strong determinant of native bee abundance and diversity has implications for the potential for competition with honey bees. A study in the present region found that greater overall flower diversity led to increased competitive effects of honey bees on native bees (Prendergast et al. 2021), and a study investigating the potential for honey bee competition to impact upon bumblebee abundance found that the abundance of specific floral resources within habitats was the most important factor limiting bumblebees, and co-existence was permitted due to forage resource partitioning (Franklin et al. 2018). Taken together, these findings suggest that rather than increase overall flower abundance or species richness, to mitigate competition from honey bees, what is required is increasing the availability of flowering species preferred by native bees.

Conclusion

Our study demonstrates that flowers that are native in origin, are important for supporting native bees in urban areas. Despite the high availability and diversity of flowering plant resources, native bees, and even honey bees, foraged on only a small subset. Under honey bee competition, native bees may be forced to expand their dietary breadths to forage on non-preferred resources, potential causing negative impacts on fitness. This study provides key empirical information to guide management of native bees in urban areas.

Supplementary material

Supplementary material is available online.

Data availability

Data are uploaded as Supplementary materials and open access data.

Conflicts of interest

None to disclose.

Declaration of funding

This research received financial support from the Forrest Research Foundation, The Australian Wildlife Society and Hesperia. BeeDay Australia provided financial support for depositing the native bee specimens in the Western Australian Museum.