Introduction

Trees constitute an ecologically and aesthetically important component of diverse ecosystems (Czaja et al. 2020). In urban ecosystems, they improve city climate, absorb stormwater, reduce air pollution and smog, store carbon, decrease the energy that reaches the ground, and enhance biocontrol (Nowak et al. 2018; reviewed in Donkersley 2019). There is also evidence that trees provide health benefits for local residents, e.g. support convalescence after stress and illness and reduce mental fatigue (Ulrich et al. 1991; Li and Sullivan 2016).

Recently, attention has been paid to the importance of trees in maintaining the diversity and abundance of insect pollinators (Bożek et al. 2023 and literature cited therein). Pollinator conservation contributes to global management efforts, as pollinator decline is reported all over the world (Ayers and Rehan 2021). The potential threats to pollinators are derivative of the loss and fragmentation of habitats, agricultural intensification and use of chemicals (pesticides, herbicides), intensification of urbanization, environmental pollution, climate change, diseases, and pests (Baldock 2020). However, the loss of floral resources and homogenization of diet are mentioned as critical factors responsible for the pollinator decline (Goulson et al. 2015). The availability of food resources demonstrates seasonal and annual variation and differs substantially between regions and landscapes (Timberlake et al. 2019; Baldock 2020). These differences are mainly derivative of variations in the abundance and diversity of flowering plants and the differences in flowering phenology (flowering time and duration) (Daniels et al. 2020; Bożek et al. 2023). A serious problem in many types of landscapes is ‘food gaps’ (i.e. periods with no available floral resources) (Vaudo et al. 2015). Recently, it has also been indicated that plant species with an optimal “nutritive value” are necessary for the proper development of pollinators (Filipiak et al. 2017; Filipiak 2019).

Various traits of floral reward may have an impact on attractiveness for pollinators, e.g. floral morphology, features of the nectar (amount, sugar concentration, volatile composition; e.g. Nicolson and Thornburg 2007; Masierowska et al. 2018; Tew et al. 2021). In general, plant species differ in the amount of produced resources and their nutritional value (Somme et al. 2015; Vaudo et al. 2015; Bożek 2021; Dmitruk et al. 2022). Nectar and pollen are the dominant food resources for pollinators (e.g. Nicolson 2011). Nectar is regarded as the main energy source. It is composed of three common carbohydrates (glucose, fructose, and sucrose) (Nicolson 2011). However, nectar characteristics are more complex and include the amount of secretion, the nectar sugar concentration, the proportion of sugars, and the content and composition of amino acids or secondary metabolites (Bernardello 2007). Nectar traits (mainly the nectar amount and the sugar concentration) can be seriously affected by environmental factors (e.g. water availability, air temperature and humidity, light); therefore, intra-species variations in nectar characteristics are reported [e.g. (Corbet et al. 1979; Farkas and Zajácz 2007; Donkersley 2019)].

Pollen is regarded as the prime nutrient resource that provides insects with proteins, lipids, macro- and microelements, vitamins, hormones, carotenoids, and flavonoids (Roulston and Cane 2000; Nicolson 2011; Di Pasquale et al. 2016). Individual flowers differ considerably in pollen productivity, which is related to androecium morphology (i.e. number of anthers per flower, anther size) (Szklanowska and Teper 1999; Masierowska et al. 2018; Dmitruk et al. 2022). Also, the pollen nutritional value varies substantially from species to species; for example, the crude protein content is in the range of 2–60% (Crailsheim et al. 1992; Roulston and Cane 2000). Both the amount of produced pollen and its chemical composition are important for the proper development of pollinators and brood rearing (e.g. (Di Pasquale et al. 2016).

Planting trees/shrubs is especially recommended for the improvement of food resources for pollinators as a highly cost-effective and long-term approach (e.g. Filipiak 2019; Donkersley 2019; Tew et al. 2021; Bożek et al. 2023). Trees are strongly recommended, as they are suggested to offer more critical food sources than herbaceous plants (Geslin et al. 2013). The superlative role of trees is related to the large number of developed flowers that can provide pollinators with abundant nectar and pollen (Baude et al. 2016; Daniels et al. 2020). Trees can also help to complement the pollinator food base and fill the seasonal ‘food gaps’ (Donkersley 2019). Several authors have also emphasized that woody species offer high-value nectar and pollen resources (Vaudo et al. 2015; Hall et al. 2017; Filipiak 2019). However, there are also reports that indicate that trees may provide inadequate or even toxic floral resources (Pawlikowski 2010; Koch and Stevenson 2017).

Activities towards the management of pollinator food resources require the creation of a list indicating attractive plant species for diverse groups of pollinators (Theodorou et al. 2020). Such a list should contain top nectar and pollen producers (Hall et al. 2017; Ayers and Rehan 2021; Tew et al. 2021). The list should also be supplemented with other information, e.g. the time and duration of blooming, groups of pollinators that use floral resources, plant status (native, exotic, naturalized, or invasive), and other values (decorative, ecological) (Dmitruk 2019; Ghosh et al. 2020; Dmitruk et al. 2022). The information is essential for policy makers and can help to optimize sustainable bee conservation (Phillips et al. 2020; Baldock 2020; Tew et al. 2021; Braman and Griffin 2022).

The genus Tilia includes ca. 60 species that are native to the temperate zone of the Northern Hemisphere. However, there are also species cultivated in southern Africa, e.g. T. americana, T. × euchlora, T. cordata, T. platyphyllos, and T. tomentosa (Glen 2002). Moreover, there is a great number of cultivars and hybrids of Tilia species (Seneta and Dolatowski 2012). Native and exotic Tilia species tend to be used across large geographical areas due to their aesthetic value (Ignatieva 2011; Grote et al. 2016). Furthermore, they are listed among the 10 top phytoremediants with a high capacity for the reduction of dust pollutants (in particular T. platyphyllos and T. cordata) and are recommended for use in urban areas (Popek et al. 2013). They support the mitigation of the ‘heat island effect’ and improve city climate (Nowak et al. 2018). The other advantageous traits of Tilia species are their tolerance to temperature changes as well as drought- and pest-resistance (Sadowiec and Gawroński 2013; Bayer et al. 2018; Kusiak et al. 2022). As very popular urban trees that produce a large number of flowers, lindens have a great potential impact on pollinators within diverse landscape types, including urban areas (Baude et al. 2016; Daniels et al. 2020). Therefore, there is a need to evaluate their usefulness as a food base.

The current study focused on the determination of (i) the time of flowering and (ii) production of nectar sugars and pollen in the flowers and per 1 m2 of tree crown of eight species from the genus Tilia. The profile of nectar carbohydrates and the quantity of pollen proteins were also evaluated to complete the floral reward characterization. We also attempted to assess the attractiveness of individual species for insect visitors (based on the composition and abundance of pollinator taxonomic groups).

Material and methods

The observations were conducted in 2018–2019 and in 2021 in Lublin, SE Poland (51°08’–51°18’N and 22°27’–22°41’E; Fig. 1). Lublin—the capital of Lublin Voivodeship, is one of the largest cities in Poland with ca. 350,000 inhabitants. It ranks fifth in the general national classification of sustainable cities (Borys et al. 2021). The city is characterized by a significant share of green infrastructure coverage (ca. 40%), including parks, urban meadows, forests (deciduous and coniferous pine), tree avenues, greenways, and private green spaces). The region is located in a warm-summer continental climate zone (Dfb in Köppen classification; Chen and Chen 2013). The mean annual temperature is + 8.9 °C (48.0 °F) and the mean annual rainfall is 750 mm. The coldest month is January (mean air temperatures = 3.6 °C; 38.4°F) and July is the warmest month (mean = 18.6 °C; 65.5°F). Snow cover remains for 60–80 days per year. The vegetation period in the Lublin area lasts ca. 209 days (https://pl.climate-data.org). The annual concentration of PM2.5 of approx. 12.1–35.4 μg/m3 is higher than in many European cities (https://www.iqair.com).

Fig. 1
figure 1

Location of the study area and distribution of examined trees in Lublin, Poland. Explanations: 1 – Tilia platyphyllos, 2—T. americana, 3—T. dasystyla, 4—T. cordata, 5—T. × euchlora, 6—T. amurensis, 7—T. tomentosa ‘Petiolaris’, 8—T. japonica; orange starlike symbol indicates urban apiaries placement

Full size image

The distribution of individual trees of linden species is shown in Fig. 1. We selected eight species from the genus Tilia L. (i) native to Polish flora (T. platyphyllos Scop., T. cordata Mill.) and (ii) exotic species: T. americana L. (native to Canada and the central and eastern United States), T. dasystyla Steven (occurring in areas from the western part of Russia to Iran), T. amurensis Rupr. (native to eastern Asia), T. tomentosa Moench ‘Petiolaris’ (occurring in southeastern Europe and southwestern Asia), T. japonica (Miq.) Simonk (native to eastern China and Japan), and Tilia × euchlora (= T. dasystyla × T. cordata; cultivated, not occurring in the wild). The number of individuals included in the observations was n = 4 (T. platyphyllos, T. americana, T. cordata), n = 3 (T. × euchlora, T. amurensis, T. tomentosa ‘Petiolaris’), and n = 2 (T. dasystyla, T. japonica).

According to the Cronquist classification system, Tilia spp. belong to the family Tiliaceae Juss. (linden family). In turn, in the APG IV classification, the genus belongs to the family Malvaceae Juss. (mallow family). The species from the genus Tilia are deciduous trees. Most of these species are native to the Northern Hemisphere (mainly Europe or North America). The flowering of linden trees propagated generatively begins at the age of about 25–30 years, but those originating from vegetative sprouts blossom at about 10–15 years (Radoglou et al. 2009; Seneta and Dolatowski 2012). Linden flowers are fragrant, yellow, and easily accessible for insect visitors (Anderson 1976). They offer nectar and pollen (Szklanowska and Teper 1999).

Flowering

We observed the phenology of flowering, i.e. the trees were monitored every 2–3 days from mid-May to the end of July, for 2–4 individuals depending on the species. We followed the procedure described in detail by Dmitruk et al. (2022). In brief, the date of the onset of flowering and the date of the end of flowering were recorded (onset—5% of flowers in bloom; end—ca. 80% of flowers wilted).

The average number of flowers per 1 m2 of tree crown was estimated based on the number of flowers in an inflorescence (randomly selected, n = 10 per species, per year) and the number of inflorescences in an area of 0.1 m2 (n = 6). An area of 0.1 m2 was determined using a ring with a diameter of 36.7 cm (Denisow 2011); then, the number of flowers was recalculated per 1 m2.

Nectar measurements

To establish the amount of secreted nectar, the branches/inflorescences on different trees (n = 2–4, depending on the species) were bagged with isolators (tulle type) to exclude pollinators. Nectar was removed from randomly selected flowers using glass pipettes with known mass (Jabłoński 2002). Nectar from each species was sampled twice per season in three replications. The flowers were picked at the same time of the day (in the morning hours 9.00–10.00 GMT + 2 h) and immediately transferred to the laboratory. Each replication consisted of nectar from 10 to 12 flowers. The nectar samples were weighed using an analytical balance and expressed in mg. Its sugar concentration was measured with a digital refractometer (in % w/w). Then, the nectar sugar amount was determined (in mg).

Nectar carbohydrate analysis

The technique of high-performance liquid chromatography (HPLC) was employed (Shimadzu system, USA) to identify and quantify the types of nectar carbohydrates. Mean values of carbohydrates were calculated and were shown as a percentage of total sugars (± SD). Next, the sugar ratios were calculated (S/G + F and G/F; where: S—sucrose G—glucose, F—fructose) (Bernardello 2007; Nicolson and Thornburg 2007).

Pollen measurements

The amount of produced pollen was established using the method described in detail by Denisow (2011). Buds (randomly selected) from different individuals (n = 2–4, depending on the species) were collected. Each year, the samples (n = 6; with 100 anthers per sample) were gathered per species per year. The anthers (well-developed but before dehiscence) were extracted from the buds and transferred into glass containers with known tare weight. Then, the glass containers with the samples were transferred into a dryer for 78 h (temperature ca. 31–34 °C). After the anthers opened, the pollen was washed out using 2–6 mL of ether and 3–6 mL of 70% ethanol. The procedure was repeated 4 times. The glass containers with pollen separated from anther tissues were placed in a dryer again (for one-two weeks). To establish the pollen dry weight, the samples were reweighed. The mass of produced pollen was established per one flower. To this end, the number of anthers was counted (n = 15 flowers).

The pollen quality was evaluated by means of assessment of the protein content according to the Kjeldahl method (Bradford 1976). First, the total nitrogen – N was measured in dry pollen samples (n = 4 per species). A factor of 6.25 was used to calculate crude protein content (Roulston and Cane 2000).

Insect visitors

In 2019 and 2021, insect visitors were observed to establish whether they feed on the floral reward provided by the Tilia species studied. Insect visitors were counted taking into account particular taxonomic groups: Apis mellifera, Bombus spp., solitary bees, and Diptera. These observations were made for three days in the full-bloom phase. The counts were made on branches accessible for observation. We counted the insect visitors in different periods of the day (9.00–10.00, 12.00–13.00, and 17.00–18.00 GMT + 2 h; n = 6 counts per day for circle areas with a diameter of 36.7 cm, i.e. 0.1 m2; n = 18 counts per season, per species; Denisow 2011). The census of observations was made for ca. 10 min. To compare the insect groups abundance, the data from diverse individuals of the same species and periods were summed. Insect monitoring was conducted in suitable weather conditions (no wind, no rain, air temperature 20–28 °C).

Statistical analyses

All data are presented as mean ± SE. For phenological data analyses, the recorded date of flowering phenology (start and end of flowering) was converted into a Julian date—day of the year (Jabłońska et al. 2015). We applied the non-parametric Kruskal–Wallis H test (for more than two independent samples with different sample sizes) for comparing the start and end dates of flowering and flowering period durations between species and between years.

The data on the flowering abundance, number of anthers, and the nectar and pollen production were verified for normality (the Shapiro–Wilk test) before the analyses were performed. These data were log transformed, as the normal distribution was not confirmed. Next, the two-way analysis of variance (ANOVA) was applied to test the differences (for the species and the year effects and for the interaction between these two factors) in the analyzed traits (abundance of flowering, number of anthers per flower, nectar and pollen data). Post-hoc means were compared with the use of Tukey's test (α = 0.05).

The results on the protein content in pollen were verified with the Lilliefors test, which indicated the lack of homogeneity of variance. Therefore, the non-parametric Kruskal–Wallis H test was used to check differences in the protein content in the pollen between the species.

Spearman’s correlation coefficients were established to describe the statistical dependence between different variables (sugar mass per flower vs. nectar mass per flower; sugar mass per flower vs. nectar sugar concentration; total sugar/pollen yield per 1 m2 vs. nectar traits vs. flowering abundance). For interpretation of the correlations, the following guide was applied: 0.0–0.19—very weak; 0.20–0.39 – weak; 0.40–0.59 – moderate; 0.60–0.79 – strong, and 0.80–1.0 – very strong. The data presented in the tables and graphs are untransformed means (± SD, standard deviation). These analyses were performed using Statistica software (v. 13.3, Statsoft Poland).

To assess the dominant flower visitor groups for Tilia species in our study sites, we used the non-parametric Kruskal–Wallis H test for independent variables. Pearson’s chi-square tests for independence and goodness of fit were used to examine the distributional properties of the insect visitor data. We tested whether the observed distributions of insect visitors for particular Tilia species fit the uniform distribution. These calculations and graphics were performed in the R environment (R ver 4.3.1).

Results

The flowering periods of the Tilia species varied (Fig. 2, Table S1- Supplementary materials).

Fig. 2 
figure 2

Flowering period of eight Tilia species observed in 2018–2019 and 2021 (in SE Poland)

Full size image

Most Tilia species were at the flowering peak in June.

The onset of flowering differed between the species (H (7, N = 75) = 31.82, P < 0.001). Four species (T. platyphyllos, T. americana, T. dasystyla, T. cordata) were classified in the early flowering group. Also, there were differences in the flowering onset date between the years of the study (H (2, N = 75) = 37.59, P < 0.001). Acceleration of the flowering onset was documented in 2018, in which the flowering start was registered 20.1 days earlier than in 2021 and 11.2 days earlier in relation to 2019. The disparity in the flowering end ranged from 9.9 to 18.9 days between the years. No difference was revealed for the flowering duration (species: H (7, N = 75) = 13.74, P = 0.056; year H (2, N = 75) = 2.68, P = 0.269). On average, the duration of linden flowering was 11.8 days.

Significant differences in the number of flowers developed per 1 m2 were recorded between the Tilia species (F7,120 = 25.00, P < 0.001), between the study years (F2,120 = 36.37, P < 0.001) and for species x years interaction (F14,120 = 2.37, P = 0.006; Table S2). The number of flowers produced per Tilia species was on average 3.6 ± 2.9 thousand per m2 of tree crown. The largest number of flowers was recorded in T. cordata (7.3 ± 4.5 thousand). Less abundant flowering was noted for T. tomentosa ‘Petioralis’ and T. amurensis (1.3 ± 0.6 and 1.6 ± 0.9 thousand, respectively). The highest number of flowers was established in 2019, a ca. 30% lower abundance of flowers was noted in 2018, and the fewest flowers were recorded in 2021.

Nectar sugar and pollen production in flowers

The nectar traits varied across the Tilia species and the years of the study (for nectar mass secreted per flower—F7,120 = 53.02, P < 0.001, F2,120 = 13.42, P < 0.001, species x years F14,120 = 2.97, P < 0.001; for the nectar sugar concentration—F7,120 = 3.30, P = 0.003, F2,120 = 22.48, P < 0.001, species x years F14,120 = 4.43, P < 0.001; for nectar sugar per flower—F7,120 = 35.44, P < 0.001, F2,120 = 1.69, P < 0.001, species x years F14,120 = 1.30, P = 0.213; Table 1).

Table 1 Nectar traits in the flowers of eight Tilia species in 2018–2019 and 2021 (SE Poland). Means and ± SD (standard deviation) are presented. Untransformed data are shown
Full size table

In terms of the nectar amount, the flowers of T. amurensis and T. japonica produced low amounts of nectar (1.74 ± 1.2 mg and 4.74 ± 1.8 mg per flower), In contrast, T. tomentosa “Petiolaris’, T. × euchlora, T. americana, T. platyphyllos, and T. dasystyla were classified as high nectar producers (they secreted 7.97 ± 3.5 mg—11. 47 ± 3.2 mg of nectar). The comparison of the sugar concentration in the nectar showed a range of 22.24%–77.38% between the different species. The highest amount of sugars was produced in the flowers of T. platyphyllos, T. × euchlora, T. tomentosa ‘Petiolaris’, and T. americana, while the lowest amount of sugars was recorded in T. amurensis. In every species, the lowest sugars mass was found in 2021. The sugar mass per flower was strongly correlated with the nectar mass secreted in the flowers (R = 0.8526, P < 0.05) and moderately correlated with the nectar concentration (R = 0.4354, P < 0.05).

The pollen mass in the Tilia flowers varied between the species (F7,120 = 21.25, P < 0.001; Table 1) and was positively correlated with the number of anthers per flower (R = 0.3687, P < 0.05; Table 2). The species (F7,216 = 61.8, P < 0.001), year (F2,216 = 48.6, P < 0.001) and species x years (F14,216 = 12.0, P < 0.001) effects where significant for the number of anthers developed per flower. The lowest number of anthers were found in T. japonica and T. cordata (23.8 ± 2.6 and 27.1 ± 3.3, respectively), whereas the highest number was recorded in T. americana (58.6 ± 5.2; Table S2). The amount of pollen per flower varied from 0.20 ± 0.08 mg to 1.34 ± 0.30 mg. On average, one Tilia flower produced 0.78 ± 0.46 mg of pollen. The highest amount of pollen was produced in the flowers of T. × euchlora and T. tomentosa ‘Petiolaris’ (1.23 ± 0.33 mg and 1.18 ± 0.30 mg per flower, respectively). A small amount of pollen was recorded in the flowers of T. cordata and T. japonica (0.37 ± 0.39 mg and 0.32 ± 0.27 mg per flower, respectively). We did not observe the year effect (F2,120 = 0.762, P = 0.4690) and any interaction between the species and the study years (F14,120 = 1.605, P = 0.087) for milligrams of produced pollen.

Table 2 Relationship of (1) the total sugar yield to nectar parameters and flowering abundance and (2) the total pollen yield to the number of developed anthers, pollen production per flower, and flowering abundance (calculated across the study species and the years of the study), based on Spearman's rank coefficients (R), at P < 0.05
Full size table

Total sugars and pollen yield

The total sugar yield per 1 m2 of tree crown differed across the Tilia species studied (F7,120 = 94.95, P < 0.0001; Fig. 3A). The Tilia species produced on average 8.5 ± 0.8 g of sugars per 1 m2 of tree crown. T. × euchlora and T. platyphyllos were the top-ranked producers (19.1 ± 10.4 g per 1 m2 and 14.7 ± 9.0 mg per 1 m2, respectively). The lowest amount of total sugars was calculated for T. amurensis (0.95 ± 0.7 g per 1 m2). The year effect for the yield of sugars was significant (F2,120 = 79.80, P < 0.001). The lowest total sugar yield was calculated for 2021 (2.9 ± 1.1 g per 1 m2), and it was ca. 3–fourfold lower than in the other study years. The species x years interaction was also significant (F14,120 = 3.43, P < 0.001). The sugar production per 1 m2 was strongly positively correlated with the mass of sugars secreted by the flowers (R = 0.8918 P < 0.05), and there was a moderate positive correlation with the number of flowers (R = 0.367, P < 0.05) (Table 2).

Fig. 3 
figure 3

Total yield of nectar sugar (A) and pollen (B) in eight Tilia species in the years of the study (in SE Poland). Means marked by different letters differ significantly: A-C between species, X–Z between years, a-d species x years; Tukey’s HSD test performed for log transformed data P < 0.05. The presented values are untransformed; the whiskers show SD- standard deviation

Full size image

The total amount of pollen produced per 1 m2 of Tilia crown differed between the species (F7,120 = 13.17, P < 0.001; Fig. 3B). On average, 1 m2 of Tilia tree crown produced 2.2 ± 1.5 g of pollen. T. × euchlora, yielding 4.1 ± 1.9 g of pollen per 1 m2 of crown, was the top-ranked producer. The other species yielded ca. 50–60% less pollen per 1 m2 of crown. The total yield of pollen differed between the study years (F2,120 = 70.56, P < 0.001; species x years interaction (F14,120 = 2.84, P < 0.001). The highest amount was noted in 2019 (3.3 ± 1.5 g per 1 m2). In the other years, a 30–55% decrease in the total pollen yield was recorded. The total pollen yield per 1 m2 was positively correlated with the number of developed flowers (R = 0.6117, P < 0.05) (Table 2).

Nectar quality

The HPLC analyses detected only hexoses (glucose and fructose) and sucrose in the nectar of the studied Tilia species (Table 3). The concentrations of glucose and sucrose in the nectar differed significantly between the species (glucose: F7,8 = 28.37, P < 0.001; sucrose: F7,8 = 11.67, P < 0.001), while the fructose content did not differ (F7,8 = 1.31, P = 0.355). The sucrose/hexose ratio was species dependent (F7,8 = 9.43, P = 0.002) and ranged from 0.36 ± 0.08 to 1.49 ± 0.76.

Table 3 Nectar sugar composition, sugar ratios (r), and hexose ratios (rh) in floral nectar from eight Tilia species (means calculated from samples collected in 2021)
Full size table

The protein content in the pollen was in the range of 7.2–16.8% (11.2 ± 5.4%). We found no significant differences in the content of proteins for individual species (chi-squared = 11.483, df = 7, P = 0.1189) (Fig. 4).

Fig. 4 
figure 4

Quantity of protein in the pollen of eight Tilia species. Mean values are presented; the whiskers indicate the standard deviation. Means marked with the same small letters are not significantly different (Kruskal–Wallis test, P = 0.1189)

Full size image

Insect visitors

We identified four insect groups visiting the flowers the Tilia species studied (Apis mellifera, Bombus species, solitary bees, and dipterans) (Fig. 5). The abundance of the insect visitors differed (H (3, N = 32) = 28.91, P < 0.001). The dominant visitors were Apis mellifera (314.1 ± 43.8), followed by Bombus spp. (81.3 ± 0.7).

Fig. 5
figure 5

Comparison of the abundance of insect visitor groups in Tilia species in Lublin, Poland (means calculated across the species and the study years). Means marked with the same letters are not significantly different (Kruskal–Wallis test, P < 0.001)

Full size image

Using the chi-square test, we found that the frequency of particular insect groups varied significantly between the Tilia species (χ2 = 40.002, df = 21, P = 0.0074). In both study years, the distribution of honey bees differed between the linden species (2019: χ2 = 76.819, df = 7, P = 0.0003; 2021: χ2 = 121.61, df = 7; Fig. 6); however, the preferences changed among the years, e.g. T. amurensis and T. dasystyla were visited the most willingly in 2019, whereas T. platyphylos was preferred in 2021. The Bombus species showed variability in the visit frequencies among the linden species only in 2019. The empirical frequencies were found to match those expected from the uniform model in the following cases: Diptera/2019 (P = 0.4711), Bombus/2021 (P = 0.171), solitary bees/2021 (P = 0.9731), and Diptera/2021 (P = 0.7006), which indicates that there was no preference for one Tilia species over another.

Fig. 6
figure 6

Bubble plot illustrating the distribution of the insect visitor groups among the Tilia species in Lublin, Poland, in (A) 2019 and (B) 2021. Pearson Chi-squared test: Apis mellifera 2019 and 2021 – P < 0.0001, Bombus 2019 – P < 0.0001, Bombus 2021 – P = 0.171; solitary bees 2019 – P =  < 0.0001, solitary bees 2021 – P = 0.9731; Diptera 2019 – P < 0.0001, Diptera 2021 – P = 0.7006

Full size image

Discussion

In the present study, the potential of eight Tilia species as a food resource for pollinators was estimated. We also tried to evaluate the possibility of using these species to improve the insect food base in Poland. Therefore, we characterized the flowering phenology, as it can illustrate the periods of availability of nectar and pollen resources and/or their deficit. We also assessed the amount of available sugar and pollen in flowers and estimated the total resources offered per unit area. Admittedly, our research has limitations, e.g. the study concerned a limited number of individuals (only 2–4 depending on the species) grown in one city. The observations of insect visitors covered only a part of the phenological season, we noted only the insect visitors visible in the observer’s sight and no detailed observations of co-flowering plants were carried out. In the case of the flowering dates, it should be emphasized that the presented data are short-term (they cover only three years).

Flowering periods

In Western and Central Europe, most Tilia species provide nectar and pollen in June and/or July (Szklanowska and Teper 1999; Farkas and Zajácz 2007; Jacquemart et al. 2018; Weryszko-Chmielewska et al. 2019; Czaja et al. 2020), which is also consistent with the flowering time established in our study region. In SE Poland, the availability of woody ornamental plants that can supplement food for insects is limited during June and/or July periods. For example, Robinia pseudoacacia L., Symphoricarpos albus Duhamel, Gleditsia triacanthos L., Lonicera spp., and Irga spp. can support pollinators in June. In turn, the number of woody species flowering in July is even narrower, and linden species are then the main flowering trees. Therefore, linden trees should be promoted more widely to improve the food base for insects in urban environments in Central, Eastern, and Western Europe.

Our short-time analysis has also shown that there may be 2–3-week (11.2–20.1 days) differences in the flowering onset of individual Tilia species between seasons, and flowering may even start as early as the end of May. This disparity is not surprising, since other authors have highlighted that Tilia species flowering is sensitive to weather conditions (e.g. Jabłoński and Kołtowski 1999; Szklanowska et al. 1999; Weryszko-Chmielewska et al. 2019). In particular, prolonged dry weather with air temperature ca. > 20 °C has been indicated to accelerate the flowering onset in Tilia species (Szklanowska and Teper 1999). In fact, the acceleration of the Tilia flowering onset has been observed not only in a short-term period but also in a long-term period. For example, Weryszko-Chmielewska et al. (2019) documented a significant acceleration in the beginning of flowering of Tilia species (by 14 days) over a period of 18 years in SE Poland.

The disadvantageous feature of all the studied Tilia species was the short duration of their flowering, i.e. only 10–14 days. A short flowering period in linden species (T. cordata and T. tomentosa, only 12–14 days) was also reported by Pawlikowski (2010). However, the food availability can be extended by strategic selection of different Tilia species in urban ornamental planning. We assume that the food supply period can be prolonged up to ca. 8 weeks if the species and cultivars studied here are included in complex plant arrangements. As shown by Daniels et al. (2020), if properly chosen, Tilia species can fill the late spring/the early summer ‘food gap’ and support food continuity in the urban landscape for more than a month in Central/Western Europe.

Food resource potential

Not surprisingly, we found that the Tilia species differed in floral reward traits (nectar amount, nectar sugar concentration, nectar sugar proportion, and pollen production). Generally, Tilia species are classified among very good nectar-yielding plants (e.g. Farkas and Zajácz 2007; Sun et al. 2020; Sultanova et al. 2022). In our study, T. tomentosa “Petiolaris’, T. × euchlora, T. americana, T. platyphyllos, and T. dasystyla were the top nectar and sugar producers. In every year of our investigations, T. amurensis was a particularly poor nectar-producing species (only 0.31–0.94 mg of sugars per flower). This may indicate that the species is sensitive to some conditions of the urban environment, as it produced high amounts of nectar (c.a. 3.58 mg per flower) in Korea, where it is a native species (Kim et al. 2014; Adgaba et al. 2017 and references therein). However, other causes of such disparity are also possible (e.g. the methods of measurements). The year-to-year variability in nectar/sugar amount (even 2–threefold) was also shown in other linden species. As shown in several earlier studies, nectar secretion in Tilia flowers is extremely unstable in time and may vary considerably between days and even throughout the same day (Jabłoński and Kołtowski 1999; Illies and Muehlen 2007).

The nectar sugar concentration was > 30%, which classifies lindens among pollinator-attractive species (Nicolson 2011; Kim et al. 2014; Ion et al. 2018). However, we found that nectar was highly concentrated (e.g. > 70%—T. platyphyllos in 2018 or T. amuriensis in 2019), which may make it difficult to collect by pollinators. Other researchers also pointed out that, in some years, the nectar of Tilia species may be too concentrated or too diluted and may thus be avoided as unattractive (e.g. Jabłoński and Kołtowski 1999).

Various authors propose that the amount of sugars produced should be taken into account when selecting species to enhance pollinator food resources (Baldock 2020; Tew et al. 2021). This recommendation is related to the high energy demands of insects. For example, one honey bee requires ca. 41.5 mg of nectar/day (Rodney and Purdy 2020). Therefore, 4–8 nectar-rich or ca. 25 nectar-poor flowers must be foraged in order to meet the daily energy needs of individual honey bees.

In addition to the nectar sugar variability in the flowers, our study showed considerable differences in pollen production per flower between the Tilia species. The pollen production in the flowers was related to the interspecies differences in the number of developed anthers. The variability in pollen production in flowers may also be associated with the size of anthers, as documented by Denisow (2011) or Dmitruk et al. (2022), but this trait was not considered in the current study.

An important feature when evaluating the value of woody species as food resources for insects is the total yield of resources (sugars and/or pollen) available per unit area of tree/shrub crown (e.g. Jabłoński and Kołtowski 1999; Farkas and Zajácz 2007; Dmitruk et al. 2022). In view of the total nectar sugar production per 1m2 of tree crown in the Tilia species studied, we in particular recommend T. platyphyllos and T. × euchlora. Additionally, T. americana and T. tomentosa ‘Petiolaris’ are valuable and ensure energy resources for insect visitors in urban green spaces. However, it has to be stressed that the yield of available sugars fluctuates considerably between years. Year-to-year total sugar yield variations are derivative of the differences in the number of developed flowers, which was emphasized by Jabłoński and Kołtowski (1999). For example, in Tilia cordata, they reported considerable differences in the number of developed flowers (from 121 000 to 1, 473 000 flowers per tree) and fluctuations in total sugars (from 0.3 to 2.0 kg per tree) between seasons. Therefore, a significant decrease in the available energy resources derived from linden trees can be expected in some years. In extreme conditions (e.g. short and weak blooming, poor nectar secretion, excessively concentrated or diluted nectar), the linden honey harvest is limited (Farkas and Zajácz 2007; Vercelli et al. 2021) and even starvation of wild pollinators (bumblebees) is observed in areas with a high local density of linden trees and a lack of alternative nectar sources during their blooming (Illies and Muehlen 2007). Hence, the instability of linden resources has to be taken into consideration by urban green planners. For example, the introduction of annual/perennial plants in the vicinity of linden trees to supplement possible nutritional deficiencies during the period of poorer linden nectar production should be considered.

Regarding the total pollen yield per 1 m2 of tree crown and the similar pollen value (measured by the protein content in pollen), we mainly propose T. platyphyllos and T. × euchlora as high-pollen-producing species. However, the other species are also valuable, given the considerable decline in pollen resources during the summer period in various landscape types (e.g. Timberlake et al. 2019; Baldock 2020). As in the case of sugar resources, pollen resources may vary considerably among years. In general, the high annual fluctuations in pollen resources in woody species are related to alternating flowering, which was emphasized previously (e.g. Szklanowska and Teper 1999; Dmitruk 2019; Sultanova et al. 2022).

The estimation of the amount of resources (nectar sugars and/or pollen) in individual trees is essential for reasonable management and improvement of nectar and pollen resources in urban areas. In general, precise evaluation of food resources available in trees is difficult, as the trait depends on many different variables (crown shape, tree age, resources per flower, abundance of flowering). In this study, we calculated the amount of sugar and pollen per 1 m2 as the first step necessary for the assessment of the availability of resources per tree. Having average sugar/pollen yield data calculated for 1 m2 of crown surface, one can estimate the total resources per tree using the data on the crown surface area. In Poland, the crown surface data of different trees can be easily obtained from the MapaDrzew™ application (https://www.mapadrzew.com/).

In view of the nectar carbohydrate composition, the nectar in most Tilia species is sucrose rich (Nicolson and Thornburg 2007), with a sucrose/hexose ratio in the range of 0.5–1.0 and sucrose content between 26 and 50%. Sucrose-dominant nectar was produced by T. × euchlora (sucrose/hexose ratio > 1; sucrose content 51–100%). Similarly, only three types of sugars were detected by Baal (1994) or Somme et al. (2016), who analyzed the carbohydrate profiles in different Tilia species (T. tomentosa, T. platyphyllos, T. cordata, T. × europea, T. × euchlora). No mannose carbohydrate was found in the nectar of Tilia species in the present study. Mannose was regarded to be a toxic component in linden nectars by Madel (1977) and Pawlikowski (2010), who even found dead bees under linden trees. However, based on the results of multiple scientific studies, Koch and Stevenson (2017) indicated that there was no evidence for Tilia nectar toxicity. We recorded diverse insect visitors (mainly honey bees but also bumble bees, solitary bees, and flies) collecting nectar and pollen in the Tilia species, but we did not observe any dead insects under the trees or in the surrounding area. Nevertheless, Tilia nectar still needs investigations towards evaluation the possible presence of chemicals (e.g. nicotine) that are considered toxic (Singaravelan et al. 2006; Fossen et al. 2018; Lande et al. 2019).

Regarding the pollen qualitative value, we found low protein content in the Tilia species compared to other tree species, e.g. Prunus > 20% (Prđun et al. 2022), Aesculus carnea (89.8%), or Acer pseudoplatanus (71.0%) (Somme et al. 2016). In general, the quantity of protein positively correlates with the health and development of bee larvae, and pollen with > 20% protein content is considered the best (Vanderplanck et al. 2014). However, other bioactive compounds (e.g. sterols, vitamins) and micro- and macroelements contained in pollen are necessary in the pollinator diet. Therefore, a diversity of plant species is recommended in urban arrangements to cover the nutritional requirements of pollinators (e.g. Filipiak et al. 2017; Daniels et al. 2020; Baldock 2020). There is also a need for evaluation of pollen quality in as many plant species as possible in the future.

Insect visitors

In agreement with other research (e.g. Anderson 1976; Pawlikowski 2010), the linden flowers in our study region were visited by honeybees, bumblebees, solitary bees, and flies. As in the study conducted by Jacquemart et al. (2018), honey bees dominated the linden flowers. However, we found that the frequency of honey bees differed between the Tilia species. Moreover, the distribution of honey bees for particular linden trees varied among the study years. In our study, the predominance of honey bees can be related to the presence of urban apiaries within the area where the examined trees were grown (see Fig. 1). As reported by Jacquemart et al. (2018), honey bees are constant visitors to linden flowers regardless of whether there are apiaries nearby. Surprisingly, we found that T. amuriensis (poor sugar and pollen producer) was willingly visited by honey bees, although in only one study year. It is accepted that insect assemblage and frequency to particular plant species are related to multiple factors (floral reward amount, chemistry of nectar and pollen, floral volatiles). Moreover, local insect pollinator diversity and density determine their abundance and visitation pattern to flowering plants (Rollings and Goulson 2019). In addition, various abiotic and biotic factors (co-flowering plans, competition between insect groups) can potentially modify plant-pollinator relationships (Illies and Muehlen 2007; Garbuzov and Ratnieks 2014; Ghosh et al. 2020). The attractiveness of flowers to insects is complex and still poses multiple challenges in the elucidation of the variability of spatio-temporal interactions and mechanisms of the effects of floral resources on insect visitor assemblages and frequencies in individual Tilia species.

In conclusion, in the temperate zone, Tilia trees can support insect pollinators with both nectar and pollen mainly in June/July. In particular, T. × euchlora and T. platyphyllos should be considered for the improvement of food resources. In conservation schemes, high year-to-year fluctuations of total sugar and pollen resources should be taken into account. Therefore, landscape managers ought to consider the introduction of other flowering plants with more stable availability of sugars and pollen to cover the food requirements of insects during the period of temporary shortages/lack of nectar sugars/pollen in linden trees. Diverse pollinators (honey bees, bumble bees, solitary bees, and flies) can benefit from Tilia species. Linden trees provide also other environmental benefits in urban areas (e.g. pollutant absorption, improvement of microclimate, reduction of heat island effects) as well as esthetic values.