Phenological imbalance in the supply and demand of floral resources: Half the pollen and nectar produced by the main autumn food source, Hedera helix , is uncollected by insects

1. Declines in floral resources, pollen and nectar, are considered one cause of pollinator decline. However, the supply and demand of floral resources can vary tempo-rally. In Britain, autumn has been suggested as a period of high floral resource availability due to the flowering of ivy ( Hedera helix ), a common native plant, combined with fewer insects active during this season. Here, we directly quantified the proportion of pollen and nectar produced by ivy, the primary autumn food source, which is uncollected by the flower-visiting insect community. 2. We quantified the proportion of nectar produced but uncollected by comparing the mass of nectar sugar accumulated in insect-accessible versus inaccessible ivy flowers and by surveying the presence of wasted, crystalised, nectar on flowers. Pollen wastage was quantified by comparing pollen counts on anthers at the start of anthesis versus anthers dropped from ivy flowers. 3. Approximately, half the floral resources, 59% nectar and 44% pollen, were uncollected by the flower-visiting insect community in autumn. As ivy flowers supply most of the available nectar and pollen in autumn, our results show that a large proportion of all floral resources are wasted in autumn. 4. Our results are the first to show that a season can be characterised by a large surplus of floral resources relative to collection by flower-visiting insects. These results demonstrate the importance of considering seasonal variation in floral resources in the conservation of bees and other flower-visiting insects


INTRODUCTION
Nectar and pollen are essential resources for bees and other flowervisiting insects, and reductions of these floral resources are considered a primary cause of flower-visiting insect decline (Baude et al., 2016;Carvell et al., 2006;Ollerton et al., 2014;Potts et al., 2010). These reductions have many potential causes, often resulting from land use changes. In agricultural landscapes, which is c.70% of UK land area (DEFRA, 2020a), many changes resulting from intensification have reduced floral resources.
In any area, floral resources are not constant but rather vary according to season and month. In temperate areas such as Britain, floral resources are effectively absent in the winter. However, this is not a problem as few flower-visiting insects are active then. But even in the rest of the year, when bees and other insects are active, the relative availability of floral resources (Balfour et al., 2018;Couvillon et al. 2014aCouvillon et al. , 2014bTimberlake et al., 2019) and associated changes in exploitative competition (Wignall, Campbell Harry, et al., 2020b) varies greatly due to the phenology of flowering pants and the number of competing insects. In British agricultural landscapes, late summer (July-August) is suggested to be a period of relative shortage of floral resources and high competition for flower visiting insects (Couvillon et al. 2014a(Couvillon et al. , 2014bRundlöf et al., 2014;Timberlake et al., 2019), due to disproportionate declines in late summer flowering plants, and most insects, c.62%, reaching peak activity during this period (Balfour et al., 2018). Late summer consequently has lower per capita resource supply (Timberlake et al., 2019), with insects competing for minute quantities of nectar (0.1 μL, Balfour et al., 2015Balfour et al., , 2021Williams, 1998;Wignall, Campbell Harry, et al., 2020b). Pollen can also be rapidly depleted from flowers within a matter of hours (c.62% of presented pollen within 3 h, Schlindwein et al., 2005), or even as a result of a single insect visit (Larsson, 2005). Conversely, other periods of the year may contain excesses of pollen and nectar, where supply can outstrip demand (Timberlake et al., 2019), and where pollen and nectar produced may go uncollected as there are not enough active insects to collect the available floral resources (Elliott, 2009). Effective conservation strategies should ideally be underpinned by a sufficient understanding of temporal changes in resource supply to maximise benefits for flower-visiting insects by targeting periods which are more likely to have shortages (Schellhorn et al., 2015).
In Britain, spring (March-May) and autumn (September-November) generally have lower levels of resource competition due to highly abundant food sources including mass-flowering crops in spring (Westphal et al., 2009), and fewer insects on the wing (Balfour et al., 2018). Models of landscape-scale nectar supply and demand have estimated a net oversupply of nectar during spring, and a deficit during summer (Timberlake et al., 2019). Autumn is also suggested to have increased per capita resource supply, given that decoded honeybee waggle dance distances indicate a decline in mean foraging distances from summer (c.2 km, July-August) to autumn (c.1 km, September-October), indicating relatively better foraging conditions in autumn (Couvillon, Schürch, & Ratnieks, 2014b). This is largely due to the flowering of ivy, Hedera helix (Couvillon, Schürch, & Ratnieks, 2014b), a widespread plant common in both urban and rural landscapes and is the primary pollen and nectar source during autumn (Garbuzov & Ratnieks, 2014b;Timberlake et al., 2019). Due to increased resource supply and reduced demand during spring and autumn (Couvillon, Schürch, & Ratnieks, 2014b;Timberlake et al., 2019;Wignall, Campbell Harry, et al., 2020b), a large proportion of floral resources produced by ivy may be uncollected by the flowervisiting insect community and be 'wasted'. Oversupply or 'wastage' of floral resources is a phenomenon which has implications for flower-visiting insects, as oversupply would indicate reduced resource competition and a relief of resource bottlenecks which may occur during other seasons such as late summer (Elliott, 2009;Timberlake et al., 2019;Wignall, Campbell Harry, et al., 2020b). Indeed, observations of uncollected nectar accumulating and evaporating to form crystals visible to the naked eye beside the nectaries have been reported (Garbuzov & Ratnieks, 2014b), which shows that more nectar was secreted than collected.
This project aims to quantify the wastage of floral resources in autumn. Although previous research has identified seasonal changes in per capita floral resource supply, most studies investigating supply and demand of resources have been indirect. These include modelling estimated nectar supply and demand in common bumblebee species (Timberlake et al., 2019), using honeybee waggle dance distances as indicators of local floral resource availability (Couvillon, Schürch, & Ratnieks, 2014b), and experimental exclusion of insects during different seasons to quantify exploitative competition (Balfour et al., 2015;Wignall, Campbell Harry, et al., 2020b). No study has yet attempted to directly quantify resource production and utilisation through direct measurement of nectar and pollen. We did this by studying ivy, Hedera helix, which is the predominant autumn nectar and pollen source and blooms for an extended period, from late August to November. Previous studies have shown that c.90% of the pollen collected by honeybees in the study area during autumn is from ivy (Garbuzov & Ratnieks, 2014b;Hennessy, Uthoff, et al., 2021b). Moreover, ivy is extremely abundant, has unrestrictive flowers, attracts all types of insects (Garbuzov & Ratnieks, 2014b) and flowers when little else is in bloom (Timberlake et al., 2019). Due to its importance as an autumn nectar source, modelling the loss of ivy from agricultural landscapes predicted a 33% decline in bumblebee colony density (Timberlake et al., 2020). Ivy, therefore, presents a unique opportunity to directly quantify overall floral resource production and utilisation across the entirety of Autumn, a whole season, by studying a single plant species. We determined: the proportions of (i) nectar and (ii) pollen produced by ivy which are uncollected, and therefore

Study sites
Data were collected from three sites in Sussex, southeast Britain, close to the University of Sussex during autumn 2020 and 2021 (7 September -18 November). The sites were approximately 1 km apart and in Falmer village (50.8646, À0.0785), Ridge farm (50.8731, À0.0818), and Stanmer Park (50.8685, À0.0926). Data were collected on days with weather suitable for insect foraging (>10 C, low wind, no rain) between 10:30 and 17:00. Because the study was carried out in autumn when weather conditions are becoming less favourable, data on resource wastage were only gathered on days when weather was suitable for insect foraging to ensure that any wastage was not simply due to unfavourable weather conditions.
Land-use within a 500 m radius from the centre of each study site was characterised using the UKCEH land cover map 2020 (Morton et al., 2020) using QGIS (QGIS Development Team, 2022). Land-use was mixed, consisting of improved and calcareous grassland, arable fields, deciduous woodland, and urban/suburban areas ( Table 1). The proportion of urban/suburban area, which has greater ivy abundance (Garbuzov & Ratnieks, 2014b), varied between the study sites, ranging from c.5% (Ridge Farm) to 41% (Falmer Village) of local land area (Table 1).

Quantifying ivy bloom
Ivy bloom was tracked using methods adapted from Hennessy, Uthoff, et al. (2021b) and Garbuzov and Ratnieks (2014b). A transect of approximately 2 km along the road between Falmer village and Ridge farm was walked weekly between 7 September and 9 November.
Twenty discrete patches of ivy were allocated to 1 Â 1 m (1m 2 ) patches and the percentages of inflorescences in bloom were estimated by eye. The same ivy patches were used each week. All researchers involved in data collection worked together to ensure consistency. For analyses, the average percentage of ivy in bloom per week was used.

Collection of pollen samples from honeybee hives
To quantify the importance of ivy versus alternative food sources, pollen was collected from three honeybee hives located in Falmer village (50.8643, À0.0782) once per week between from 17 September to 29 October. Standard pollen traps with 5 mm mesh (E.H Thorne UK) at hive entrances caused pollen to be knocked off returning foragers into a tray below (Hennessy, Uthoff, et al., 2021b). Traps were in place 10:00-17:00 on days with favourable foraging conditions (>10 C, low wind, no rain). Pollen was then removed and stored at À14 C.
Pollen samples from particular hives for particular days were analysed. For each, 50 pollen pellets were taken per sample and sorted by colour. In each colour group, one pellet per five was analysed with a compound light microscope. A small amount of each pellet was mounted onto a glass slide with glycerine jelly and stained with Fuchsin dye. The pollen was identified as either ivy or not ivy at 400Â magnification using a pollen identification guide (Sawyer & Pickard, 1981) and reference samples taken from ivy flowers. Pollen pellets were assumed to contain pollen from a single flower species due to honeybees being flower constant, visiting one flower species per foraging trip (Free, 1963). One sample had only 13 pellets so all were used.

Nectar crystallisation surveys
The presence of crystals on ivy flowers indicates that nectar was being wasted, having not been collected by the flower visiting insect community resulting in crystal formation via evaporation.

Direct quantification of nectar wastage by the flowervisiting insect community
Nectar is secreted onto disc-shaped nectary glands which are open and easily accessible to insect visitors (Konarska, 2014;Pacini et al., 2003;Smets, 1986;Vezza et al., 2013). Nectar remaining in the flower will not, therefore, be due to inaccessibility. To estimate T A B L E 1 Percentage area cover of land use types within a 500 m radius from the centre of each study site. directly how much nectar produced by ivy is wasted by nonconsumption by the flower-visiting insect community, nectar sugar accumulation was quantified in flowers which were either accessible or inaccessible to insects.
During each survey day, five discrete ivy patches per site were selected and within each a 1 Â 1 m area containing flowers in full bloom was defined. At the start of the day between 10:30 and 13:00, when weather conditions became suitable for insect foraging (see above), pairs of inflorescences in full bloom were selected within each patch. To reduce variation between pairs, inflorescences were selected visually to be at the same stage of flowering. On each inflorescence, one flower was cleared of any nectar present by pipetting 1 μL of water onto the nectary and removing it after 15 s using microcapillary pipette tubes (Drummond Microcaps 5 μL, 32 mm). Each inflorescence was randomly assigned one of two treatments: (i) covered with a fine gauze bag (inaccessible to insects) and (ii) unbagged (accessible to insects).
After, approximately, 3 h, the volumes and concentrations of nectar in the test flowers were quantified. A microcapillary tube was used to remove the nectar and the volume calculated from the length of liquid (Corbet, 2003). Nectar sugar concentration (% Brix) was calculated by ejecting the nectar onto a handheld refractometer (Bellingham and Stanley™, 0%-50% Brix). Nectar sugar mass (g) was calculated for each individual flower by converting % Brix into mass/ volume (Weast, 1972) and multiplying by the estimated nectar volume (Corbet, 2003). This gave a single ecologically relevant measure of nectar as the dissolved sugar mass. This was then converted to the rate of sugar mass secretion during the period between initial nectar removal and subsequent collection of newly secreted nectar.
Greater relative nectar demand would manifest itself as increased frequency of insect visits to ivy and reduce the nectar accumulation in unbagged flowers (Gill, 1978). Therefore, nectar wastage was Directly quantifying pollen wastage by flower-visiting insect community Ivy anthers are fully exposed to foraging insects. When flowers have accessible anthers, up to 100% of pollen can be removed from the flower by foragers (Stanghellini et al., 2002;Wolf et al., 1999). Therefore, any uncollected pollen will be due to insufficient demand.
To quantify pollen wastage, samples of ivy anthers were collected approximately weekly from 19 October to 18 November from the same 1 Â 1 m patches used for the nectar sampling. Naturally abscising anthers were obtained by gently shaking the ivy patch so that anthers fell onto a piece of paper, from where they were transferred to Eppendorf tubes. Pollen remaining in these anthers was considered to be wasted as it would not be collected by flower-visiting insects, rather falling to the ground. Flowers at the start of anthesis, and so with a full complement of pollen, were collected from the ivy patches using forceps and stored in Eppendorf tubes. All anthers were stored at À14 C.
In the laboratory, ten anthers per patch per treatment were Quantifying flower-visiting insect density on ivy To assess how insect density, number of insects per flower, varied throughout the study period, on the same study days as the nectar samples were taken the flower visiting insect community was surveyed for each of the 1 Â 1 m patches. Insect surveys were carried out twice per patch per study day by making a near instantaneous (<10 s) 'snapshot' count of the insects foraging in each patch (Garbuzov & Ratnieks, 2014a). Most insects were identified to species where possible using field guides (Brock, 2021;Falk, 2018). Where possible insects which could not be identified in the field were taken back to the laboratory for identification. In addition to recording the number of insects in the ivy patches, the number of flowering ivy inflorescences in the 1 Â 1 m patch was estimated by eye.
Although the distance between the study fields, 1 km, is small relative to the maximum foraging distance of honeybees and bumblebees (Beekman & Ratnieks, 2000;Redhead et al., 2016), it is greater than mean bumblebee foraging distances (B. terrestris c.600 m, B. pascuorum c.300 m, Darvill et al., 2004;Redhead et al., 2016) and comparable to the mean honeybee foraging distance at the time of year the study was carried out (1-1.5 km, Couvillon, Schürch, & Ratnieks, 2014b). Further, honeybees and bumblebees comprised a small proportion, 16%, of total insects observed on ivy in the study area (see results). Therefore, although it is possible that the same honeybee or bumblebee may visit multiple study sites, insects at different study sites will largely be different individuals.

Statistical analysis
All analyses used R studio version 1.3.1 (R Core Team, 2018). For all models, linear and quadratic polynomial models with differing family link functions were tested for best fit. Models were built and the best models were selected as having the lowest Akaike information criterion (AIC) values using the R package bbmle (Bolker, 2020). Variables were transformed where appropriate. Model assumptions were checked visually using the R package DHARMa (Hartig & Lohse, 2020) and p values were obtained through stepwise removal of independent variables.
To analyse how flower treatment (bagged vs unbagged) affected nectar sugar accumulation, nectar sugar mass accumulation rate was included as a dependent variable and flower treatment as predictor variable. Daily mean nectar sugar content for each site was calculated for each treatment to improve heteroscedasticity.
To analyse how anther age affected pollen count, log-transformed pollen count was included as a dependent variable and anther treatment (start vs. end of anthesis) as predictor variable.
To assess how nectar accumulation in unbagged flowers was affected by insect density, mean insect density per ivy patch was calculated per study day. In the linear model, the square root of nectar sugar accumulation per unbagged flower was included as a dependent variable and square root of mean insect density was included as predictor variable.
Temporal changes in insect density and nectar accumulation per unbagged flower were analysed in separate models. Quadratic polynomial models fitted the data better than the linear models (nectar accu-

RESULTS
A total of 14 pollen samples were collected from honeybee hives from 17 September to 29 October. As in previous studies, ivy predominated at 83.6% of total pollen collected from returning foragers. Ivy bloom started in early September, peaked in late September and early October and was mostly over by early November (Figure 2).
Thirty nectar crystal surveys were carried out in 2020 and 2021.
The proportion of ivy inflorescences with nectar crystals present per survey was 30.3% ± 31.8% (mean, SD). A total of 1038 insects were recorded in 357 insect counts. The most frequent was the ivy bee Colletes hederea (27%, respectively. Other Diptera and butterflies were also observed foraging on the ivy (Table 2). Approximately, half the ivy bees (47.3%) and honeybees (45.5%) had pollen in their scopae and corbiculae. Mean F I G U R E 3 Nectar sugar accumulation (mean ± standard error) for ivy flowers which were covered with a fine gauze bag (n = 152) and inaccessible to the flower visiting insects or unbagged (n = 144) which allowed insects to access the nectar (linear model: F 1,64 = 9.64, p < 0.005). Overall data for the study period 7 September-2 November 2021 are shown. Mean nectar sugar accumulation for each treatment is shown above each bar. Flowers were initially drained when weather became suitable for insect foraging and nectar accumulation was quantified approximately 3 hours later.

F I G U R E 4
Haemocytometer pollen counts (mean ± standard error) of anthers at the start (n = 42) and end (n = 47) of anthesis (linear model: F 1,87 = 18.7, p < 0.001). Counts were made from 1 μL from a suspension of 10 anthers in 10 μL of 70% ethanol. Mean nectar sugar accumulation for each treatment is shown above each bar. Overall data from anthers collected 19 October to 18 November 2021.
T A B L E 2 Insect species observed foraging on ivy during autumn (7 September-18 November) 2021 (n = 346 counts).   (Table 3), reducing to a minimum density in early October, near peak ivy bloom, before increasing again as bloom began to go over ( Figure 6a). Nectar sugar content in unbagged flowers likewise showed temporal variation throughout the ivy flowering period ( Figure 6b), with date having a significant effect on nectar sugar content (Table 3). Nectar sugar content per unbagged flower increased to a peak in mid-October, before declining for the remainder of the ivy bloom period (Figure 6b).
T A B L E 3 Final summary outputs for statistical models analysing the effect of date on nectar sugar availability in unbagged flowers and insect density.  F I G U R E 6 Temporal changes in (a) daily mean insect density per site and (b) daily mean unbagged flower nectar sugar accumulation per site in autumn 2021. Each datapoint represents the daily mean value at a particular study site as fitted in the models (mean insect density calculated from n = 2 surveys; mean nectar sugar accumulation calculated from n = 5 measurements). Dashed lines represent quadratic polynomial models. In (a) the square root of insect density is plotted.

DISCUSSION
Our results show that approximately half the floral resources, 59% of nectar and 44% of pollen, produced by ivy flowers in the autumn is wasted as it is uncollected by foraging insects (Figures 3 and 4).
Nectar crystals, which also indicates wastage, were commonly observed. Ivy was the most important foraging source for honeybees (83% of total pollen collected from returning foragers). Given that honeybees are extreme generalists (Biesmeijer & Slaa, 2006), this indicates that ivy was the main source of floral resources in autumn.
These results strongly support the hypothesis that autumn is a period of reduced resource competition between foraging insects in the study area (Couvillon, Schürch, & Ratnieks, 2014b;Wignall, Campbell Harry, et al., 2020b). It goes beyond this by showing an actual surplus of pollen and nectar for the flower-visiting insects active in autumn.
As such, our results add an important additional dimension to understanding the ecology and conservation of flower-visiting insects, namely that floral resources available to all flower-visiting insects can be in such abundance that a considerable proportion is wasted for an extended period.
Data were collected in mixed farmland in southeast England. In the study area, we have observed nectar crystals regularly over the past decade, showing that it is not an occasional phenomenon, and nectar crystals have been reported at other locations in south England (Garbuzov & Ratnieks, 2014b) and Ireland (Ratnieks et al. 2022). More broadly, ivy is widely distributed in the British Isles and Europe and is also common (Metcalfe, 2005;Small, 2019), having been recorded in 91% of unique hectads (10 km Â 10 km squares) in Britain (Hill et al., 2004). It is, therefore, likely that autumn is a season of nectar and pollen surplus in many areas. However, factors such as landscape context and management will likely affect the degree of resource wastage (discussed below). Our results nevertheless provide a valuable starting point for understanding supply and demand of floral resources.
59% of nectar secreted by ivy flowers was uncollected. Greater overall nectar demand by the flower visiting insect community should increase the frequency of visits to flowers, leading to unbagged flowers containing a lower fraction of the secreted nectar than bagged flowers (Gill, 1978). This was shown by our results with nectar accumulation being significantly negatively related to insect density ( Figure 5), although with a low correlation coefficient, indicating that greater competition increased the proportion of nectar collected. This is consistent with previous results showing that greater exploitative competition causes greater nectar depletion (Torné-Noguera et al., 2016;Wignall, Campbell Harry, et al., 2020b). Moreover, there was intra-seasonal variation in insect density and nectar depletion ( Figure 6). Density decreased as ivy bloom increased, with the statistical model showing a nine-fold drop between early bloom (0.018 insects per inflorescence, 7 September) and peak bloom (0.002 insects, 5 October). In accord with this, nectar accumulation in unbagged flowers increased three-fold between early bloom (0.32 μg per minute, 7 September) and peak bloom (0.92 μg per minute, 5 October). This intra-season variation in insect density and nectar availability suggests that as ivy reached peak bloom insect foragers were spread over a greater number of flowers, reducing nectar depletion per flower. This is analogous to the 'dilution effect' observed in mass-flowering crops (Holzschuh et al., 2011;Riedinger et al., 2014).
As to the ultimate fate of the uncollected nectar, the open structure of ivy flowers and nectaries should make accumulated nectar wasted either by being washed off by rain or, if dry, to crystallise via evaporation (Garbuzov & Ratnieks, 2014b). In both cases, the nectar is not available to insects. Indeed, we found c.30% of ivy inflorescences had nectar crystals present, indicating that crystallisation via evaporation is common. The rapidity and occurrence of nectar crystallisation are likely affected by variation in weather conditions such as temperature, humidity, dew formation, and rainfall. Weather conditions are already known to affect the foraging success of flowervisiting insects (Bentrup et al., 2019;Hennessy et al., 2020;Hennessy, Harris, et al., 2021a;McNaughton, 1988;Norton, 1988), and if accumulated nectar is affected in this way, it may add a further challenge to flower-visiting insects. For example, rapidly crystalising ivy nectar may reduce the nectar available for foraging insects, increasing competition.
Pollen is also wasted. Our results show that a similarly high proportion, 44%, of the ivy pollen available at anthesis remained uncollected on the anthers at abscission. This contrasts with previous work on other plant-pollinator systems which show that pollen can be rapidly depleted by flower-visiting insects. For example, 61.6% of pollen produced by Campanula rapunculus is removed by wild bees within 3 hours, and only 0.2% remains by the end of anthesis (Schlindwein et al., 2005). On Knautia arvensis, 26%-79% of pollen on inflorescences can be removed by a single visit by an insect forager (Larsson, 2005). Our results show that, as with nectar, autumn pollen supply from ivy flowers outstrips demand, with almost half of pollen initially available remaining on the anthers at the point of abscission.
Colletes hederae, the ivy bee, was the most frequently recorded species on ivy flowers (27%), outnumbering honeybees (13%) which are the only other bee present in large numbers in autumn. C. hederae almost exclusively forages on ivy (Hennessy, Uthoff, et al., 2021b) and was first recorded in the UK in 2001 and is now widespread throughout Britain (BWARS, 2016;Cross, 2002). A colonising species could be a major competitor for native species which also forage on ivy in autumn, and ivy bee abundance is increasing. A decade ago (2011,2012), C. hederae only comprised 3% of insects observed on ivy flowers in the same area (Garbuzov & Ratnieks, 2014b), whereas in 2020, it was 26% (Hennessy, Uthoff, et al., 2021b). Our data indicate, however, that exploitative competition with ivy bees is unlikely to harm native species, at least at present ivy bee numbers, because of the large quantity of floral resources uncollected on the ivy in our study area (Wignall, Campbell Harry, et al., 2020b). Moreover, ivy bee activity stops before the end of ivy bloom, indicating that honeybees, which continue foraging into December if the weather permits, may not be competing with the ivy bee during the end of ivy bloom (Hennessy, Uthoff, et al., 2021b). Previous work suggests that competitive alien species can increase competition and cause resource depletion (Dupont et al., 2004;Gross, 2001;Gross & MacKay, 1998).
For example, in the Canary Islands, introduced honeybees can displace native bees through nectar depletion on Echium wildpretii spp. (Dupont et al., 2004) and increases in bumblebee nectar collection in colonies located closer to honeybee hives is likely due to increased local nectar depletion (Thomson, 2004). Our results suggest that the presence of ivy bee does not necessarily lead to nectar or pollen depletion. However, the effect of ivy bees on exploitive competition and the success of other foraging insects should be examined (Hennessy, Uthoff, et al., 2021b), for example, via exclusion experiments (Balfour et al., 2015;Wignall, Brolly, et al., 2020a) (Wignall, Brolly, et al., 2020a).
This may typically be the case as many insect species are generalists and plant-pollinator networks are often well connected (Memmott, 1999;Waser et al., 1996). Flowering of ivy, therefore, may also reduce exploitative competition on alternative plant species. Furthermore, the effect of ivy flowering on insect visitation on coflowering plant species will likely be influenced by differences in nutritional quality between available food sources (Fowler et al., 2016).
For ivy plants, nectar production presumably represents a significant energy cost, potentially reaching 37% of available energy in some species (Pleasants & Chaplin, 1983;Southwick, 1984), which can reduce energy available for other functions such as seed production (Pyke, 1991). With half of the total ivy nectar uncollected, nectar production would seem to be more costly than necessary. Nectar production should theoretically maximise individual plant fitness (Pyke, 1981;. Plant-plant competition for high, even surplus, nectar production is hypothesized to be selected for in seasons when plants compete for relatively scarce pollinators  as increased investment in nectar increases the frequency of pollinator visits (Fowler et al., 2016;Heinrich, 1979;Pleasants & Chaplin, 1983;Wetherwax, 1986) and reproductive success (Larson & Barrett, 2000;Neiland & Wilcock, 1998). This is suggested to result in positive feedback leading to greater per capita nectar availability in seasons of relative pollinator shortage . Our results further support the idea that autumn in the study area is such a season. As greater proportions of ivy flowers set fruit in response to higher insect visitation (Jacobs et al., 2009(Jacobs et al., , 2010, the excess of nectar in our study is consistent with these predictions. Why do insects not alter their flight season to coincide with periods of resource abundance, thus reducing the imbalance between supply and demand? Thermal windows (Lefebvre et al., 2018) and other character traits (Junker et al., 2013) may act as constraints on selection and prevent phenological matching .
Resource wastage in autumn will likely be affected by the local prevalence of suitable surfaces for ivy growth, which is generally greater in urban than agricultural areas (Garbuzov & Ratnieks, 2014b).
Local management, such as whether ivy is removed, will also potentially influence autumn resource supply. Moreover, insect density, which will also affect resource wastage, will be influenced by the availability of other potentially limiting resources such as nest-sites, larval food plants, or even pollen and nectar supply during previous seasons (Benadi, 2015;Rundlöf et al., 2014;Schellhorn et al., 2015).
Land-use composition around the three study sites were mixed, consisting of woodland, grassland, arable fields, and urban/suburban areas (Table 1). Although this is typical for the local area, field data in other locations would be desirable to determine how widespread autumn floral resource wastage is and how this varies between different landscape contexts. One relatively simple way to gather such data would be via a citizen science project asking participants to observe ivy and record the presence of nectar sugar crystals in the flowers.
Flower-visiting insect communities are limited by both floral resource quantity (Baude et al., 2016;Timberlake et al., 2019) and diversity and quality (Potts et al., 2003;Vaudo et al., 2015). Although in autumn, there may be wastage of floral resources in certain areas, flower-visiting insect fitness may still be limited by the diversity and quality of available food sources. Indeed, relatively little is known about the nutritional value of ivy pollen and nectar. Agricultural land often provides a large quantity of pollen and nectar but is often from a single or few sources, such as mass-flowering crops (Vaudo et al., 2015). These landscapes are often unable to support healthy bee populations due to reduced fitness arising from monotonous diets (Di Pasquale et al., 2013;Tasei & Aupinel, 2008;Timberlake et al., 2019). Future work should, therefore, aim to determine whether periods which have been suggested to have sufficient quantity of floral resources also provide sufficient diversity.
Overall, our results show that in autumn, a significant proportion of pollen and nectar produced by the predominant source of these floral resources, ivy, is uncollected by the flower-visiting insect community in our study area. Whereas previous empirical studies (Couvillon, Schürch, & Ratnieks, 2014b;Timberlake et al., 2019;Wignall, Campbell Harry, et al., 2020b) and models suggest that certain seasons could have excesses of nectar , our study is the first to directly demonstrate that a significant proportion of both nectar and pollen produced in an ecosystem is uncollected and, therefore, wasted. More broadly, our results show that floral resources may not be in short supply all year round and supports previous work indicating that in the study area, autumn is a season favourable for floral resource collection (Couvillon, Fensome, et al., 2014a;Couvillon, Schürch, & Ratnieks, 2014b;Wignall, Campbell Harry, et al., 2020b). Our results should give pause to consider strategies aimed at enhancing floral resources, in particular whether these would be more effective if targeted at seasons of resource shortage such as late summer months (June-August), where more insects are active, competition is greater, and floral resource supply may not meet demand (Balfour et al., 2018;Timberlake et al., 2019;Wignall, Campbell Harry, et al., 2020b). In the study area, foraging bumblebees and honeybees compete strongly in the summer (Balfour et al., 2015) and use approximately half the energy obtained in the nectar gathered merely to fuel their foraging activity within the flower patch .

CONFLICT OF INTEREST
The authors declare that they have no financial/personal relationships that may be considered as potential competing interests.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in Mendeley Data at https://data.mendeley.com/datasets/ 78dsgxfx49/1.