Skip to main content
NCBI home page
AoB Plants logoLink to AoB Plants
. 2025 Sep 17;17(5):plaf052. doi: 10.1093/aobpla/plaf052

Plant reproductive strategies and pollinator attributes differ in small-scale habitat heterogeneity

Dongzhou Deng 1, Juanli Chen 2, Li He 3, Dawei Li 4, Dechao Chen 5, Wuxian Yan 6, Junpeng Mu 7,2,3,4,✉,5
PMCID: PMC12480739  PMID: 41036205

Abstract

Habitat variability critically influences plant reproductive strategies and pollinator attributes. However, studies on intraspecific variation in vegetative and floral traits, pollinator attributes, and seed traits remain limited in the context of small-scale habitat heterogeneity, particularly meadows interspersed with sandy patches. On the Tibetan Plateau, discrete sandy patches (some as small as 10 m2) occur within alpine meadows. We hypothesized that distinct plant reproductive strategies and pollinator attributes exist between meadows and sandy habitats at a microhabitat scale. To test this hypothesis, we conducted a field experiment to investigate variation in floral traits, pollinator attributes, and seed traits in a Tibetan alpine herb (Astragalus purpurinus) across meadow and sandy habitats. Our results show that meadow populations produced fewer nectar-enriched flowers with high sugar concentrations, fewer and larger seeds, and were pollinated primarily by bumble bees. In contrast, sandy-habitat populations produced numerous nectar-poor flowers with low sugar concentrations, more numerous small seeds, and relied on mason bees for pollination. Our results demonstrate that micro-scale habitat heterogeneity drives divergent plant reproductive strategies and pollinator attributes within a single species. These findings reveal novel mechanisms by which small-scale environmental variation shapes reproductive adaptation in alpine ecosystems.

Keywords: alpine meadow, bee behavior, reproductive strategies, land restoration, pollination

Habitat variability critically influences plant-pollinator interactions. However, we know little about how intraspecific variation in plant reproductive strategies and pollinator attributes responds to small-scale habitat heterogeneity, such as meadows interspersed with sandy patches. To address this, we performed a field experiment in Tibetan meadows and sandy habitats to examine the impact of these differences on floral, pollinator, and seed traits. Our data indicate a notable disparity: meadow populations produced less nectar-abundant flowers and larger seeds, predominantly pollinated by bumble bees. Conversely, populations in sandy habitats produced a higher number of nectar-deficient flowers and smaller seeds, depending on mason bees for pollination.

Introduction

Climate change and human activity are modifying the habitats of flowering plants. Human-caused habitat heterogeneity can influence various plant reproductive traits, including floral production and display, flower size, the quantity and quality of nectar and pollen, and flowering phenology (Aizen and Vazquez 2006, Jacquemyn et al. 2012, Xiao et al. 2016, Gargano et al. 2017, Librán-Embid et al. 2021, Chen et al. 2025). Additionally, an increase in habitat heterogeneity leads to changes in pollinator attributes, including total visitation rates, pollinator diversity, and the composition of the pollinator fauna (Aizen and Vazquez 2006, Ekroos et al. 2013, Neumüller et al. 2020). These disparities in plant traits and pollinator attributes, resulting from habitat heterogeneity, influence plant reproductive strategies (Aguilar et al. 2006, Evans et al. 2017, Reverté et al. 2019). For instance, in resource-abundant habitats, plants produce larger, nectar-rich flowers to attract pollinating insects, thereby increasing seed yield and reproductive success (David et al. 2019, Dai et al. 2022, Gao et al. 2023). In contrast, plants in resource-limited habitats (e.g. low water, poor nutrients) confront significant challenges to survival and reproduction. To maximize fitness and ensure species persistence, they have evolved diverse specialized reproductive strategies. These often involve trade-offs, diverting energy from vegetative growth towards reproductive functions like flower size, pollen production, and seed output (Waser and Price 2016, Mueller et al. 2020, Dai et al. 2022). The majority of research focuses on the effects of large-scale (regional scale) habitat heterogeneity on plant-pollinator interactions, with only a few studies examining the implications of small-scale habitat heterogeneity (local scale).

At the local-patch scale, small-scale habitat heterogeneity might create distinct micro-niches, potentially influencing plant–pollinator interactions. Recent studies indicate that small-scale habitat heterogeneity can increase field border density (Hass et al. 2018). This increased density, in turn, enhances the quality and quantity of floral rewards (e.g. pollen and nectar), increases flower abundance, and promotes pollinator richness and abundance (Hass et al. 2018, Battle et al. 2021, Berthon et al. 2021, Bueno and Iglesias 2021). Consequently, these improvements elevate the diversity of plant–flower visitor interactions and increase plant reproduction. Our previous study shows that small-scale habitat heterogeneity, like changes in soil moisture over short distances, causes plants to evolve specific floral traits (e.g. different flower color and flower size) to entice bees and flies for pollination (Dai et al. 2022).

Arid and wet habitats constitute a prominent form of habitat heterogeneity, which in turn affects vegetation traits, floral traits, and seed yields (Tadey et al. 2009, Dai et al. 2022, Rauschkolb et al. 2022). Plants that grow in arid habitats usually evolve traits that help them survive in dry conditions, like having low above-ground biomass and allocating limited energy to nectar and flower production (Li et al. 2005, Hassan et al. 2017, Kuppler et al. 2021). Plants with reduced flower resources are associated with specific insect pollinators, such as arid plants that rely on ants and beetles for pollination (Raine et al. 2002, Phillips et al. 2018). Furthermore, arid plants often produce a larger number of small seeds that are dispersed great distances by the wind, facilitating the establishment of new habitats (Venable et al. 2008). In contrast, plants growing in meadows produce larger, nectar-enriched flowers, attracting more effective pollinators and generating high-quality seeds (Petanidou et al. 1999, Gallagher and Campbell 2017, Dai et al. 2022). Larger seeds supply seedlings more energy throughout the germination process, which helps them compete with other species and enhance their establishment (Jakobsson and Eriksson 2000). However, most studies on plant reproductive strategies and pollinator attributes in arid and wet habitats focus on large-scale effects, with limited attention to small-scale effects. There is a scarcity of studies investigating how alterations in the reproductive strategy of the same species respond to small-scale changes in their habitats.

The Tibetan Plateau is characterized by habitat heterogeneity, which includes sandy habitats due to climate change and human activities (Xue et al. 2009, Li et al. 2021). The Tibetan alpine meadow contains interspersed sandy habitats, with individual sandy patches measuring under 10 m². Habitat conditions alter significantly over short distances (Gao et al. 2016, Wang et al. 2023), resulting in variations in plant traits and pollinator attributes. In addition, the low temperature and high precipitation contribute to significant species diversity in the meadows, comprising sedges, grasses, and forbs, including Asteraceae, Gentianaceae, and Fabaceae (Mu et al. 2014, Hu et al. 2024). However, species diversity in sandy habitats is low (Wang et al. 2023, 2025). Most research in this region focuses on soil physical and chemical properties and community diversity (Wang et al. 2023, 2025). There is limited understanding of the mechanics of species adaptation, especially how the same species acclimatizes to different habitats over short distances. To our knowledge, limited data have examined plant reproductive strategies within the same species across alpine meadows and alpine sandy habitats at a small scale. This study addresses three questions: (i) Do the flower traits of the studied plant populations differ in response to small-scale habitat heterogeneity? We predicted that plants from the alpine meadow would produce nectar-rich, high-quality flowers, whereas plants from sandy habitats would produce nectar-poor, low-quality flowers. (ii) Do pollinator richness and visitation rates differ under small-scale habitat heterogeneity? Our prediction was that bumble bees would be attracted to flowers in alpine meadows, while mason bees would be attracted to those in sandy habitats, reflecting their preferences for distinct flower traits. (iii) Do the seed traits of the studied populations differ between Tibetan alpine meadows and sandy habitats on a small scale? We predicted that populations from the alpine meadows would produce few, large seeds, whereas populations from the alpine sandy habitats would produce numerous small seeds.

This study was conducted on the alpine meadows on the Zoige Plateau in China, where sandy habitats are interspersed within the meadow. Previous studies indicate that the levels of soil moisture and soil nutrient levels are higher in alpine meadows compared to the sandy habitats (Mu et al. 2014, Dai et al. 2022, Wang et al. 2025). We conducted field experiments using a Tibetan alpine herb (Astragalus purpurinus) to examine the effects of small-scale habitat heterogeneity on plant reproductive strategies and pollinator attributes. We measured the quantity and quality of nectar, the number of flowers per plant, the number of inflorescences per plant, plant height and above-ground biomass, seed mass and the number of seeds per plant, reproductive allocation, and pollinator richness and visitation rates between alpine meadows and alpine sandy habitats. These findings will elucidate the mechanisms through which small-scale habitat heterogeneity influences plant reproductive strategies and pollinator attributes.

Materials and methods

Study site and species

The study sites were located in the Zoige Plateau, within the eastern Tibetan Plateau (32°20′–34°05′N and 101°36′–103°55′E). The elevation is between 3400 and 3900 m. This climate is characterized by prolonged winters and short summers. The mean annual temperature is recorded at 1.73°C, with total annual precipitation of 756 mm. The yearly average relative humidity is ∼69%, and there are no definitive frost-free periods. The primary soil types in the region consist of subalpine meadow soil and mountain meadow soil (Mu et al. 2014, Wang et al. 2023).

The species richness differs between meadows and sandy habitats. Meadows encompass over 30 species per m2, including Saussurea nigrescens, Anemone trullifolia, Deschampsia caespitosa, Festuca ovina, Blysmus sinocompressus, A. purpurinus, and Kobresia setchwanensis (Dai et al. 2022, Hu et al. 2024). Vegetation coverage surpasses 95%, with an average plant height of around 30 cm (Dai et al. 2022). However, sandy habitats have 10 species per m², including Kobresia myosuroides, Elymus dahuricus Turcz, Elymus sibiricus, Heracleum hemsleyanum, Acanthocalyx alba, Aconitum gymnandrum, and A. purpurinus (Huang et al. 2023, Wang et al. 2023).

We selected three study sites in the Zoige Plateau (Fig. 1). Site 1 was located in Waqie (33°10′40.8″N 102°37′11.0″E), Site 2 was located in Maixi (33°51′46.8″N 102°32′37.3″E), and Site 3 was located in Xiaman (33°43′36.3″N 102°26′23.8″E). Each site comprised two subsites: meadows and sandy habitats. Meadows are solely grazed by cattle throughout winter, while the sandy regions remain ungrazed (Wang et al. 2023, Mu et al. 2014). Each subsite comprised 10 plots. The plots were 2 m × 2 m. In total, there were 3 sites, 6 subsites, and 60 plots.

Figure 1.

Alt Text: Graphical representation of study site (include ten meadow and ten sandy plots for each site), plant species, and pollinator species.

Open in a new tab

The figure shows the study sites and plant population and bee species of alpine meadows and sandy habitats. There were three sites, and each site included 10 meadow (red square) and 10 sandy plots (yellow square). Each plot was 2 m × 2 m. Totally, there were 3 sites, 60 plots, and 180 plants. Photo Credits: J. Mu

Astragalus purpurinus, a perennial herbaceous species within the Fabaceae family, was used as the study species. It typically thrives at elevations between 2500 and 4000 m. Plants reach a height of 30–50 cm, producing 10–20 flowers per inflorescence. This species becomes green in mid-May each year and undergoes senescence in mid-September. Plants typically bloom from late-June to August and bear fruit from July to September.

Measurement of vegetation and flower traits

During the peak of the flowering period in 2022, the vegetative and floral traits of each plant were assessed, including plant height, nectar volume per flower, nectar concentration, number of flowers per inflorescence, number of inflorescences per plant, pollinator richness, and visitation rates. Upon fruit maturation, we collected fruits, stems, and leaves. The stems and leaves were dried at 65°C for 48 hours. We determined the reproductive allocation by dividing the seed mass by the total dry mass of stems and leaves for each plant. The above-ground biomass was the sum of leaf mass plus stem mass. Each plot contained 3 plants, and each subsite had 30 plants, totaling 180 plants of A. purpurinus.

We randomly selected three flowers from each plant. Subsequently, we encased each flower in cylindrical metal netting for 24 hours to exclude pollinators before quantifying the nectar volume and concentration per flower (Real and Rathcke 1991). We quantified the nectar volume per flower using 1- or 2-µl micropipettes. A hand refractometer (Eclipse, Stanley and Bellingham, Basingstoke, UK) with a precision of 0.5% was used to measure nectar concentration (Brix scale). The measurement was conducted on a total of 540 flowers.

We assessed the number of flowers per inflorescence and the number of inflorescences per plant during the flowering period. We quantified the total flower number per plant by multiplying the number of flowers per inflorescence by the number of inflorescences per plant daily at noon. Flowers that were counted were marked with strings to ensure that only new flowers were counted each day. We used a rule to measure plant height, defined as the height of the photosynthetic tissue from the ground surface (Heady 1957). Because the calyx tube length is associated with pollinator diversity (Dohzono et al. 2004), we determined the average flower size per plant by measuring the calyx tube length (cm) of three flowers.

Measurement of richness and visitation rates of pollinators

We netted bumble bees and mason bees, which we could not identify in the field (McFrederick and LeBuhn 2006). The specimens were identified in the laboratory. All samples were collected during the peak of the flowering period.

We measured the pollinator visitation rate as the number of pollinator visits per flower each hour. We initially recorded the number of flowers for each sample plant. Observers documented pollinator visitation to the flowers from a distance of three meters. We evenly distributed the observation intervals throughout the day, from 09:00 to 17:00. During each hourly interval, a single observer scrutinized each distinct plant for 2 minutes from a stationary position. With six observers, each plant was observed twice per hour. Observations recorded the number of flowers per plant that received bee visits during each observation period (Vaudo et al. 2014). We conducted the observations exclusively on sunny days, repeating them five times for each plant in 2022. We monitored each individual plant for a total of 90 minutes per year. We calculated the visitation rates (R) per flower per hour by dividing the total number of visits (Nv) per hour by the number of flowers (Nc), which we represented as R = Nv/Nc (Arroyo et al. 1985).

Measurement of seed trait with pollinators and hand-supplemented pollen

We meticulously picked, counted, and weighed each fruit (from August to September) to the nearest 0.001 g upon maturation. We regarded empty seeds as an indication of failure. We defined ‘seed yield’ as the mass of seeds produced per flower. The average seed mass (mg) was calculated by dividing the total seed mass per flower by the number of seeds per flower.

In 2022, we conducted hand-supplemental pollen experiments because pollinator density and foraging rates influenced seed yield during the peak of the flowering season. For hand-supplemented pollen experiments, one flower was selected from each sampled plant. This included 30 plants in alpine meadows and 30 plants in alpine sandy habitats within each subsite. We collected fresh pollen from the anthers of 30 distinct plants for manual supplementation once stigmas matured. Using a fine paintbrush, we carefully transferred the pollen grains to the stigmas of flowers. Pollen transfer occurred exclusively among plants within the same populations. After the fruits ripened, we collected the pods, measured the seed number and seed mass, and calculated seed set as previously described.

Measurement of soil temperature, moisture, and nutrition

To test the soil moisture and temperature between two habitats, we used Hobo Data Loggers (Pro V2) to monitor the daily soil temperature and moisture levels of the soil throughout the experiment. After collecting all the plant samples, soil samples were collected from 0 to 20 cm depth using a 20 cm diameter soil auger. We analyzed total soil nitrogen, phosphorus, and carbon in the laboratory (Wang et al. 2023).

The effects of nectar traits or community composition on pollinator richness and visitation rates

We conducted a pot experiment to evaluate the impact of community composition on pollinator richness and visitation rates in 2023. During the peak of the flowering period, 30 individuals of A. purpurinus were randomly selected from meadows and sandy habitats, respectively. All plants were transferred into pre-prepared containers containing indigenous soil. The pots (each 27.5 cm in diameter and 30.0 cm in height) were sterilized with 70% ethanol (Gao et al. 2023). Each pot had a single plant. We categorized the pots into two groups, including 15 pots from meadows and 15 pots from sandy habitats. One group was placed in alpine meadows, and the other group was placed in sandy habitats. We assessed pollinator diversity and visitation rates using the aforementioned approaches. The observations were performed solely on sunny days. In each hourly interval, a single observer examined each plant for 4 minutes from a fixed location. Each individual plant received a total of 96 minutes of field observations annually.

We performed an experiment to assess the impact of the quantity of nectar supply on pollinator richness and visitation rates in alpine meadows versus sandy habitats. During the peak of the flowering season in 2023, we randomly selected 10 individuals of A. purpurinus from both meadows and sandy habitats. We randomly selected ten flowers from each plant. Prior to the nectar addition experiment, we used capillaries to collect all nectar previously secreted from the flowers. We added 0 (purified water), 0.1, 0.2, 0.3, and 0.4 µl of a 40% sucrose solution to flowers 1 through 5, while providing equivalent volumes of a 20% sucrose solution to flowers 6 through 10. We delivered the sucrose solution into the spur's tip using a 10 µl microsyringe (Johnson et al. 2004). We evaluated the pollinator visitation rate using the previously described methods. Recognizing that bumble bees and mason bees prefer different nectar concentrations (Cnaani et al. 2006, Ahrenfeldt et al. 2019), two separate nectar concentrations (40% and 20%) were used throughout the experiment.

To assess whether nectar quality affected pollinator attributes, we conducted a nectar addition experiment using different sucrose solution concentrations at Site 1 in 2023. We randomly selected 10 individuals of A. purpurinus from both meadows and sandy habitats during the peak of the flowering season. We randomly selected 6 flowers from each plant. Before the nectar addition experiment, we used a 10 µl microsyringe to collect all previously secreted nectar from the flowers. Flowers labeled 1 to 6 received 0.2 µl of sucrose solutions at concentrations of 0%, 10%, 20%, 30%, 40%, and 50%, respectively. We observed that the base of the calyx tube is where plants store their nectar. Hence, we added the nectar to the base of the calyx tube. The sucrose solution was delivered into the spur's tip using a 10 µl microsyringe (Johnson et al. 2004). We evaluated the pollinator visitation rate using the previously described methods.

Statistical analysis

In this study, the data for each trait were averaged for each plant prior to analysis. All statistical analyses were conducted using R (R version 4.4.1; R Core Team 2024).

We used Wilcoxon rank-sum tests to assess the effects of small-scale habitat heterogeneity on vegetation and floral traits, pollinator visitation rates, and seed traits. This non-parametric approach, suitable for comparing differences between two independent groups, was implemented using the Wilcox test function from an R package. Specifically, vegetation and floral traits (e.g. plant height, nectar volume, and number of inflorescences per plant) and reproductive success metrics (e.g. seed set and fruit-to-flower ratio), pollinator visitation rates, and species richness were all compared across habitat types using this test.

Principal coordinate analysis (PCoA) based on Bray–Curtis distance (functions cmdscale, ordiplot, and ordiellipse from the Vegan package) was used to examine the patterns in soil properties, vegetation and flower traits, and pollinator visitation rate across alpine meadows and alpine sandy habitats (Oksanen et al. 2025). The two primary components that most effectively explained the variation in community structures were used to construct the PCoA plots. Z-score standardization was performed prior to PCoA.

We analyzed the effects of soil temperature and soil properties on plant traits using generalized linear mixed models (GLMMs). The fixed effects encompassed soil temperature and soil properties (soil moisture, TN, TP, and TC), while the plots and the plants were considered random effects. First, we modeled the flower number, inflorescence number, and seed number with a Poisson model (log-link) using Laplace approximation to estimate parameters. Second, we modeled the seed set with a negative binomial model (logit-link) using Laplace approximation to estimate parameters. Finally, we modeled the nectar volume and concentration and flower size with a Gaussian model (logit-link) using Laplace approximation to estimate parameters.

We used random forest models analysis using the ‘randomForest’ R package (Rigatti 2017) to ascertain the principal parameters affecting pollinator visitation rates. The principal parameters considered included nectar volume and concentration, flower size, plant height, and total flower number per plant.

We analyzed correlations among all variables using the R package ‘linkET’. Relationships between pollinator visitation rates and nectar rewards (such as nectar volume per flower and nectar concentration), between seed set and pollinator visitation rates, as well as between seed mass and seed number, were analyzed using the method of least squares.

Results

Do the soil properties vary across the habitats?

Soil properties differed between alpine meadows and sandy habitats. Alpine meadows had higher total nitrogen (W = 8100, P < 0.001, n = 90), phosphorus (W = 8100, P < 0.001, n = 90), and carbon (W = 8100, P < 0.001, n = 90) compared to alpine sandy habitats (see Supplementary Figure S1). Compared to alpine sandy habitats, meadow habitats had higher soil moisture (W = 8100, P < 0.001, n = 90) but lower soil temperature (W = 0, P < 0.001, n = 90; see Supplementary Figure S1).

Do the vegetative and flower traits and the relative coverage of A. purpurinus vary across the habitats?

Small-scale habitat heterogeneity significantly influenced vegetative and floral traits (Supplementary Figure S2). Plants in alpine meadows exhibited greater plant height (W = 8100, P < 0.001, n = 90), higher nectar volume per flower (W = 8100, P < 0.001, n = 90), higher nectar concentration (W = 8100, P < 0.001, n = 90), and larger flower size (W = 8083, P < 0.001, n = 90) compared to those in alpine sandy habitats (W = 8100, P < 0.001, n = 90; Fig. 2). However, plants in sandy habitats exhibited a greater number of flowers per inflorescence (W = 859, P < 0.001, n = 90) and a greater number of inflorescences per plant (W = 31.5, P < 0.001, n = 90; Fig. 2).

Figure 2.

Alt Text: The violin plots showing meadow plants produced fewer nectar-enriched flowers with high sugar concentrations, while sandy-habitat plants produced numerous nectar-poor flowers with low sugar concentrations.

Open in a new tab

The violin plots show nectar volume per flower (a), nectar concentration (b), flower size (c), number of flowers per inflorescence (d), number of inflorescences per plant (e), and number of flowers per plant (f) of Astragalus purpurinus between alpine meadows and alpine sandy habitats. Violin plots show the density (width), interquartile range (hinges), and 1.5 times the interquartile range (adjacent lines). The line within the box represents the median of the responses, and the black star within the box represents the mean of the responses. Different letters above the boxes denote significant differences among treatments (P < 0.05).

Small-scale habitat heterogeneity significantly influenced the relative coverage of A. purpurinus and the average community height. Plants in alpine meadows had lower relative coverage (about 5%) in contrast to alpine sandy habitats (around 15%; W = 8100, P < 0.001, n = 90). In alpine meadows, the community average height was 30 cm, whereas in alpine sandy habitats, it was 15 cm (W = 8100, P < 0.001, n = 90).

Do pollinator attributes vary across the habitats?

The richness and the visitation rates of pollinators varied among the habitats (Fig. 3; see Supplementary Figure S2). Bombus was the principal pollinator with high visitation rates in alpine meadows (W = 8100, P < 0.001, n = 90), including Bombus friseanus, B. supremus, and B. filchnerae. In alpine sandy habitats, mason bees served as the principal pollinator with relatively low visitation rates (W = 0, P < 0.001, n = 90).

Figure 3.

Alt Text: The violin plots showing meadow plants were pollinated primarily by bumble bees, while sandy-habitat plants relied on mason bees for pollination.

Open in a new tab

The violin plots show visitation rates of bumble bees (a) and mason bees (b) of Astragalus purpurinus between alpine meadows and alpine sandy habitats. Violin plots show the density (width), interquartile range (hinges), and 1.5 times the interquartile range (adjacent lines). The line within the box represents the median of the responses, and the black star within the box represents the mean of the responses. Different letters above the boxes denote significant differences among treatments (P < 0.05).

Do seed traits vary across the habitats?

Small-scale habitat heterogeneity significantly influenced reproductive traits, including seed mass, the number of seeds per plant, and reproductive allocation (Fig. 3; see Supplementary Figure S3). Plants in alpine meadows exhibited larger seeds (W = 7882, P < 0.001, n = 90) and an increased seed set (W = 7951, P < 0.001, n = 90) following hand-pollination compared to plants in alpine sandy habitats. Plants in alpine meadows produced more seeds (W = 160, P < 0.001, n = 90) and had a greater reproductive allocation (W = 8100, P < 0.001, n = 90) compared to those in sandy habitats (see Supplementary Figure S3). However, seed set in nature did not differ significantly between habitats (W = 4283, P = 0.505, n = 90; Fig. 4).

Figure 4.

Alt Text: The violin plots showing meadow plants produced fewer and larger seeds, while sandy-habitat plants produced more numerous small seeds in nature. Although seed set was similar in both habitats under natural conditions, meadow plants had a higher seed set than sandy-habitat plants when provided with supplementary pollen.

Open in a new tab

The violin plots show seed mass (a), number of seeds per plant (b), seed set in nature (c), and seed set by supplementary pollen (d) of Astragalus purpurinus between alpine meadows and alpine sandy habitats. Violin plots show the density (width), interquartile range (hinges), and 1.5 times the interquartile range (adjacent lines). The line within the box represents the median of the responses, and the black star within the box represents the mean of the responses. Different letters above the boxes denote significant differences among treatments (P < 0.05).

Do the trait relationships vary among the habitats?

The correlations between the attributes of flora and pollinators varied across habitats (see Supplementary Figure S4). In alpine meadows, we observed a substantial correlation between nectar volume per flower and nectar concentration. However, this relationship was absent in alpine sandy habitats. In both habitats, strong positive correlations were found between pollinator visitation rates and nectar volume per flower. A negative association was observed between seed number and seed mass. Furthermore, a positive relationship between seed set and pollinator visitation rates was found in both alpine meadows and sandy habitats (Fig. 5). The random forest model further indicated that nectar volume per flower and nectar concentration significantly influenced visitation rates of bees in both alpine meadow and sandy habitats (Fig. 6).

Figure 5.

Alt Text: The graphs compare pollinator attributes and seed traits between meadow and sandy habitats, illustrating positive correlations between visitation rates and both nectar quantity and quality, as well as seed set in nature and visitation rates. Additionally, there was a negative correlation between seed mass and the number of seeds per plant, accompanied by detailed annotations and statistical values.

Open in a new tab

The correlations between visitation rates and nectar volume per flower (a), visitation rates and nectar concentration (b), seed set and visitation rates (c), and seed mass and seed number per plant (d) in alpine meadows and alpine sandy land. The red circles, lines, and words refer to the trait relationships in the alpine meadows, and the blue pluses, lines, and words refer to the trait relationships in the sand habitats.

Figure 6.

Alt Text: The bar plots comparing relative importance of five plant traits on pollinator visitation rates between alpine meadow and sandy habitats. All comparisons are non-significant (ns) except for one trait in each panel showing a significant difference. The x-axis represents relative importance, and groups are differentiated by color.

Open in a new tab

Random forest analysis identified vegetative and flower traits that dominate changes in visitation rates of bumble bees (a) and mason bees (b). * P < 0.05, ** P < 0.01. ‘nvf’ refers to nectar volume per flower, ‘nc’ refers to nectar concentration, ‘fz’ refers to flower size, ‘height’ refers to plant height, and ‘fnp’ refers to flower number per plant. %IncMSE indicates the increase of the Mean Squared Error when a given variable is randomly permuted. The larger the value of IncMSE, the greater the influence of the characteristic variable on the target. Blue bars indicate statistical significance, and green bars indicate statistical insignificance.

Soil moisture, soil temperature, and soil total nitrogen significantly influenced nectar volume per flower (Table 1). Nectar concentration was considerably influenced by soil moisture, soil temperature, soil total phosphorus, and soil total carbon (Table 1). Soil moisture markedly affected flower size and the number of flowers per plant (Table 1). Soil total phosphorus significantly impacted plant height (Table 2).

Table 1.

Effects of soil properties on flower traits of Astragalus purpurinus were analyzed using generalized linear mixed models.

Predictors Nectar volume per flower Nectar concentration Flower size Number of flowers per plant
Estimates CI P Estimates CI P Estimates CI P Estimates CI P
(Intercept) −1.98*** −2.04 to −1.92 <0.001 3.33*** 3.31 to 3.36 <0.001 0.63*** 0.61 to 0.66 <0.001 5.18*** 5.12 to 5.25 <0.001
Soil moisture 0.12* 0.01 to 0.22 0.029 0.09* 0.01 to 0.17 0.021 1.12** 1.05 to 1.20 0.001 −0.27* −0.53 to −0.02 0.036
Soil temperature −0.12* −0.24 to −0.01 0.038 −0.12** −0.20 to −0.04 0.003 0.98 0.91 to 1.04 0.454 −0.09 −0.25 to 0.06 0.247
Total soil nitrogen 0.14* 0.01 to 0.27 0.030 0.10 −0.01 to 0.20 0.065 1.08 0.99 to 1.19 0.081 −0.29 −0.69 to 0.11 0.149
Total soil phosphorus 0.16 −0.00 to 0.32 0.053 0.18** 0.05 to 0.31 0.006 0.97 0.88 to 1.07 0.599 −0.11 −0.38 to 0.16 0.413
Total soil carbon −0.06 −0.19 to 0.07 0.390 −0.11* −0.21 to −0.00 0.042 1.04 0.96 to 1.13 0.336 0.04 −0.26 to 0.34 0.796
Random effects
 σ2 0.00 20.28 0.01 3722.35
 τ00 0.00plot 0.00plot 0.00plot 0.00plot
0.00plant 0.00plant 0.00plant 0.00plant
 ICC 0.61 – 0.05 –
 N 3plant 3plant 3plant 3plant
10plot 10plot 10plot 10plot
Observations 180 180 180 180
Marginal R2/Conditional R2 0.991/0.996 0.007/NA 0.828/0.837 0.000/NA
Open in a new tab

CI, the confidence intervals; σ2, variance; τ00, intercept variance; ICC, the intra-class correlation coefficient; N, sample number.

The bold font represents statistical test results that reached a significant level, *P < 0.05 **P < 0.01 ***P  < 0.001

Table 2.

Effects of soil properties on plant height and seed traits of Astragalus purpurinus were analyzed using generalized linear mixed models.

Predictors Plant height Seed set Seed mass Number of seeds per plant
Estimates CI P Estimates CI P Estimates CI P Estimates CI P
(Intercept) 2.39*** 2.37 to 2.42 <0.001 −0.71*** −0.75 to −0.66 <0.001 2.25*** 2.07 to 2.44 <0.001 4.48*** 4.39 to 4.57 <0.001
Soil moisture 0.06 −0.01 to 0.14 0.092 0.23*** 0.11 to 0.36 <0.001 0.94 0.75 to 1.19 0.626 −0.07 −0.41 to 0.28 0.707
Soil temperature −0.08 −0.15 to 0.00 0.055 0.10 −0.02 to 0.22 0.095 0.89 0.71 to 1.11 0.292 −0.06 −0.27 to 0.16 0.611
Total soil nitrogen 0.04 −0.06 to 0.13 0.465 0.00 −0.17 to 0.18 0.990 1.30 0.93 to 1.81 0.127 −0.47 −1.00 to 0.07 0.087
Total soil phosphorus 0.20** 0.08 to 0.32 0.001 −0.25** −0.43 to −0.06 0.009 1.53* 1.09 to 2.16 0.015 −0.25 −0.61 to 0.11 0.171
Total soil carbon −0.02 −0.11 to 0.08 0.748 0.13 −0.03 to 0.28 0.105 0.82 0.61–1.10 0.189 0.19 −0.21 to 0.58 0.357
Random effects
 σ2 2.66 0.01 0.11 1689.94
 τ00 0.00plot 0.00plot 0.00plot 0.00plot
0.00plant 0.00plant 0.00plant 0.00plant
 ICC 0.03
 N 3plant 3plant 3plant 3plant
10plot 10plot 10plot 10plot
Observations 180 180 180 180
Marginal R2/Conditional R2 0.046/NA 0.276/NA 0.730/0.739 0.000/NA
Open in a new tab

CI, the confidence intervals; σ2, variance; τ00, intercept variance; ICC, the intra-class correlation coefficient (ICC); N, sample number.

The bold font represents statistical test results that reached a significant level. *P < 0.05, **P < 0.01, ***P < 0.001.

Are differences in pollinator visitation rates due to variation in plant community or nectar concentration?

Plants in alpine meadow habitats exhibited high visitation rates of bumble bees when transplanted in alpine meadow and sandy habitats (see Supplementary Figure S5a, b). Similarly, plants in alpine sandy habitats exhibited a high visitation rate of mason bees when transplanted in both alpine meadow and sandy habitats (see Supplementary Figure S5c, d).

The nectar addition experiment revealed that bumble bees chose flowers with high nectar concentrations (40%–50% sugar; Fig. 7a) but mason bees favored flowers with low nectar concentrations (10%–20% sugar; Fig. 7b). The visitation rates of both bees exhibited a positive correlation with nectar volume per flower (Fig. 7c–f).

Figure 7.

Alt Text: Graph comparing visitation rates of bees with different nectar volumes and concentrations (sugar %), which were the flowers from alpine meadow and sandy habitats.

Open in a new tab

The figures show the visitation rate of bumble bees (a) and mason bees (b) at the nectar concentration (0%–50% sugar) addition treatment, the visitation rate of bumble bees in alpine meadows (c) and in alpine sandy habitats (d), as well as the visitation rate of mason bees in alpine meadows (e) and in alpine sandy habitats (f) at the nectar volume (0–0.4 µl) addition treatment. The blue circle represents the mean of the responses. The line represents the standard error. The star denotes significant differences among treatments (P < 0.05).

Discussion

Our results demonstrated that meadow populations and sandy populations of A. purpurinus have different vegetative and floral traits, pollinator attributes, and reproductive strategies. In the meadow populations, elevated individual height and nectar incentives had increased bumble bee visitation, correlating with the production of fewer and larger seeds. However, sandy populations had reduced individual height and nectar incentives, which enticed mason bees and was associated with the production of a greater number of small seeds. These two distinct mechanisms enable A. purpurinus to adapt to differences in their respective habitats.

Small-scale habitat heterogeneity modified various plant traits, including plant biomass and plant height, flower morphology and display, floral rewards, and flowering phenology (Aizen and Vazquez 2006, Hass et al. 2018, Battle et al. 2021, Berthon et al. 2021, Bueno and Iglesias 2021). In this study, although the sandy habitats were situated <10 m from the meadows, soil moisture and nutrient levels differed significantly. Alpine meadow soil had higher nutrients and moisture than alpine sandy soil (Gao et al. 2016, Wang et al. 2023). Plants in alpine meadows exhibited higher nectar rewards and increased individual height (Fig. 2; see Dai et al. 2022). In contrast, plants in alpine sandy habitats exhibited reduced nectar production, lower nectar concentrations (Fig. 2), and decreased individual height. It is important to note that nectar production often incurs significant nutritional and carbohydrate costs (Heinrich and Raven 1972, Pyke 1991, Mu et al. 2015). Plants with nectar-deficient flowers may thus gain a competitive advantage by reallocating resources to enhance other vegetative or floral traits. Previous studies report a strong inverse relationship between nectar production and both the number of flowers per plant and flower size (Pleasants and Chaplin 1983, Mu et al. 2014). Our analysis confirmed this trade-off, revealing that diminished nectar production and concentrations were associated with increased flower and inflorescence number. Alpine meadow populations produced flowers with abundant, high-quality nectar but fewer flowers and inflorescences. In contrast, alpine sandy populations increased flower and inflorescence number while reducing both nectar amount and quality. This suggests that small-scale habitat variability drives floral trait trade-offs.

The diversity of habitat-mediated traits enables plants to adapt to varying habitats. In alpine meadows, populations producing abundant nectar attract pollinators, promoting pollination and increasing seed set (Dai et al. 2022, Wang et al. 2024). Astragalus purpurinus—a rare species with <5% relative abundance in the meadows—faced competition for pollinators from co-flowering species like Saussurea nigrescens and S. stella (Mu et al. 2014, Su et al. 2022). Bumble bees prefer high-nectar rewards due to their larger body and high metabolic demands (Biernaskie and Gegear 2007, Frasnelli et al. 2021). Alpine meadow population allocated resources to increased height (intensifying light competition), and the resulting carbon surplus enabled significant nectar rewards (Heinrich and Raven 1972, Pyke 1991, Mu et al. 2015), attracting bumble bees and improving pollination and seed set. In nutrient- and moisture-limited alpine sandy habitats, plants expend energy to withstand arid conditions while allocating relatively scarce resources to reproduction. Consequently, they produce less-concentrated nectar in smaller volumes (Waser and Price 2016, Mueller et al. 2020), attracting mason bees—which prefer low-sugar flowers (Ahrenfeldt et al. 2019)—ensuring adequate pollination. Simultaneously, A. purpurinus in sandy habitats invests energy in producing more flowers and inflorescences, securing sufficient seed production.

A trade-off between seed number and seed mass is common, which influences plant adaptation to diverse habitats (Harper et al. 1970, Greene and Johnson 1993, Guo et al. 2000). Our study demonstrated that the quality and quantity of seeds varied among different habitats. The population in alpine meadows produced fewer, larger seeds. Larger seeds provide more energy for germination, enhancing seedling competitiveness and establishment success (Jakobsson and Eriksson 2000). In sandy habitats, however, populations produced a larger number of small seeds, which facilitated long-distance dispersal and colonization. Astragalus purpurinus has evolved two distinct reproductive strategies to adapt to small-scale habitat heterogeneity.

The quantity and quality of floral rewards affect the diversity and choice of pollinators (Fowler et al. 2016, Su et al. 2022). For example, bumble bees prefer flowers with higher nectar concentrations (Roubik et al. 1995, Cnaani et al. 2006). A. purpurinus was pollinated by two distinguishable pollinators: bumble bees in alpine meadows and mason bees in alpine sandy habitats. Pollinators exhibit differentiation across small-scale habitat heterogeneity due to the varying concentrations and quantities of nectar provided by flowers. Our nectar addition experiment further demonstrated that bumble bee visitation frequency significantly increased when nectar concentration exceeded 30% but declined dramatically below 30%. Conversely, mason bee visitation peaked at 10%–20% nectar concentration, aligning with findings that mason bees prefer strawberry flowers with 17%–34% sugar content (Ahrenfeldt et al. 2019). Overall, the frequency of bee visits to flowers positively correlated with increased nectar volume. Our findings also showed that bumble bees preferred flowers with a nectar concentration of 40%, while mason bees favored those with 20%. Although we observed that both mason bees and bumble bees carried pollen on their bodies, both bees appear to gather pollen passively. They invest considerable time foraging for nectar.

Community composition also affects pollinator visitation rates (Kelly and Elle 2020, Allen-Perkins et al. 2025). Our data demonstrated that populations from alpine meadows had a high visitation rate of bumble bees, irrespective of their location in alpine meadows or sandy habitats. Similarly, the populations from alpine sandy habitats had the greatest frequency of floral visits by mason bees, irrespective of their location in alpine meadows or sandy habitats. The findings suggest that intrinsic floral traits outweigh community context in determining pollinator preferences. Flower scent is a crucial factor in pollinator behavior (Farré-Armengol et al. 2016). However, our findings indicated that neither bumble bees nor mason bees exhibited a preference for flowers treated with 0.2 µl purified water, implying that flower scent is not associated with pollinator attributes in this context.

Conclusion

In summary, our findings indicate that A. purpurinus exhibits different vegetative and floral traits, pollinator richness and visitation rates, as well as reproductive strategies within small-scale habitat heterogeneity. These two distinct mechanisms enable plants to adapt to various environments. Our findings provide new insights into how small-scale habitat heterogeneity influences plant reproductive strategies and pollinator attributes.

Supplementary Material

plaf052_Supplementary_Data
plaf052_supplementary_data.zip (973.8KB, zip)

Acknowledgements

We thank Zhiqian Zheng, Rui Gao, Ruolan Wang, Yuanxin Lou, Yali Chen, Peiyue Che, Weiwei Chen, Maoping Yan for the field and lab assistance.

Contributor Information

Dongzhou Deng, Ecological Restoration and Conservation on Forest and Wetland Key Laboratory of Sichuan Province, Sichuan Academy of Forestry Sciences, 18 Xinghui West Road, Chengdu 610081, China.

Juanli Chen, Ecological Security and Protection Key Laboratory of Sichuan Province, Mianyang Normal University, 166 Mianxing West Road, Mianyang 621000, China.

Li He, Ecological Restoration and Conservation on Forest and Wetland Key Laboratory of Sichuan Province, Sichuan Academy of Forestry Sciences, 18 Xinghui West Road, Chengdu 610081, China.

Dawei Li, Ecological Security and Protection Key Laboratory of Sichuan Province, Mianyang Normal University, 166 Mianxing West Road, Mianyang 621000, China.

Dechao Chen, Ecological Restoration and Conservation on Forest and Wetland Key Laboratory of Sichuan Province, Sichuan Academy of Forestry Sciences, 18 Xinghui West Road, Chengdu 610081, China.

Wuxian Yan, Ecological Restoration and Conservation on Forest and Wetland Key Laboratory of Sichuan Province, Sichuan Academy of Forestry Sciences, 18 Xinghui West Road, Chengdu 610081, China.

Junpeng Mu, Ecological Security and Protection Key Laboratory of Sichuan Province, Mianyang Normal University, 166 Mianxing West Road, Mianyang 621000, China.

Author contributions

D.D., W.Y., and J.M. conceived the idea. J.C., D.L., L.H., and J.M. generated data and materials. D.D., J.C., D.C., J.M. analyzed the data. D.D. and J.M. wrote the first version of the manuscript and all authors contributed to it.

Supplementary data

Supplementary data is available at AoB Plants online.

Funding

This study was funded by National Natural Science Foundation of China (32371604), and by the open project from the Ecological Restoration and Conservation on Forest and Wetland Key Laboratory of Sichuan Province (2023SKJT05-1), by the science and technology innovation project of the Sichuan Forestry and Grassland Bureau (510201202109731), and by Research and Application Demonstration of Comprehensive Management Model for Severe Desertification Grassland Caused by Wind-water Combined Erosion in the Northwest Yellow River Basin (2024YFNH0005).

Data availability

Raw data and R code are available online at https://zenodo.org/records/17054532.

Ethics approval

The species sampled is not endangered or protected. Fieldwork was conducted in locations where no permission was required.

Consent for publication

Not applicable.

References

  1. Aguilar  R, Ashworth  L, Galetto  L  et al.  Plant reproductive susceptibility to habitat fragmentation: review and synthesis through a meta-analysis. Ecol Lett  2006;9:968–80. 10.1111/j.1461-0248.2006.00927.x [DOI] [PubMed] [Google Scholar]
  2. Ahrenfeldt  EJ, Sigsgaard  L, Hansted  L  et al.  Forage quality and quantity affect red mason bees and honeybees differently in flowers of strawberry varieties. Entomol Exp Appl  2019;167:763–73. 10.1111/eea.12820 [DOI] [Google Scholar]
  3. Aizen  MA, Vazquez  DP. Flower performance in human-altered habitats. In: Harder  LD, Barrett  SCH (eds.) Ecology and Evolution of Flowers. Oxford, UK: Oxford University Press, 2006, 159–79. [Google Scholar]
  4. Allen-Perkins  A, Artamendi  M, Montoya  D  et al.  Untangling the plant reproductive success of changing community composition and pollinator foraging choices. Ecography  2025;2025:e07240. 10.1111/ecog.07240 [DOI] [Google Scholar]
  5. Arroyo  MTK, Armesto  JJ, Primack  RB. Community studies in pollination ecology in the high temperate Andes of central Chile II. Effect of temperature on visitation rates and pollination possibilities. Pl Syst Evol  1985;149:187–203. 10.1007/BF00983305 [DOI] [Google Scholar]
  6. Battle  KF, de Rivera  CE, Cruzan  M. The role of functional diversity and facilitation in small-scale pollinator habitat. Ecol Appl  2021;31:e02355. 10.1002/eap.2355 [DOI] [PubMed] [Google Scholar]
  7. Berthon  K, Meyer  ST, Thomas  F  et al.  Small scale habitat conditions are more important than site context for influencing pollinator visitation. Front Ecol Evol  2021;9:703311. 10.3389/fevo.2021.703311 [DOI] [Google Scholar]
  8. Biernaskie  JM, Gegear  RJ. Habitat assessment ability of bumble-bees implies frequency-dependent selection on floral rewards and display size. Proc R Soc Lond B Biol Sci  2007;274:2595–601. 10.1098/rspb.2007.0705 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bueno  SS, Iglesias  MM. Small-scale spatial heterogeneity and diversity of plant-flower visitor interactions in high mountains. Ecosistemas  2021;30:2131. 10.7818/ECOS.2131 [DOI] [Google Scholar]
  10. Chen  M, Wang  Z, Ma  X  et al.  Impact of fragmented habitats on reproductive success in dominant shrubs: natural selection on floral display and pollinator visitation. Front Plant Sci  2025;16:1522871. 10.3389/fpls.2025.1522871 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cnaani  J, Thomson  JD, Papaj  DR. Flower choice and learning in foraging bumblebees: effects of variation in nectar volume and concentration. Ethology  2006;112:278–85. 10.1111/j.1439-0310.2006.01174.x [DOI] [Google Scholar]
  12. Dai  W, Yang  Y, Patch  HM  et al.  Soil moisture affects plant–pollinator interactions in an annual flowering plant. Philos Trans R Soc B Biol Sci  2022;377:20210423. 10.1098/rstb.2021.0423 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. David  TI, Storkey  J, Stevens  CJ. Understanding how changing soil nitrogen affects plant–pollinator interactions. Arthropod Plant Interact  2019;13:671–84. 10.1007/s11829-019-09714-y [DOI] [Google Scholar]
  14. Dohzono  I, Suzuki  K, Murata  J. Temporal changes in calyx tube length of Clematis stans (Ranunculaceae): a strategy for pollination by two bumble bee species with different proboscis lengths. Am J Bot  2004;91:2051–9. 10.3732/ajb.91.12.2051 [DOI] [PubMed] [Google Scholar]
  15. Ekroos  J, Rundlöf  M, Smith  HG. Trait-dependent responses of flower-visiting insects to distance to semi-natural grasslands and landscape heterogeneity. Landsc Ecol  2013;28:1283–92. 10.1007/s10980-013-9864-2 [DOI] [Google Scholar]
  16. Evans  TM, Cavers  S, Ennos  R  et al.  Florally rich habitats reduce insect pollination and the reproductive success of isolated plants. Ecol Evol  2017;7:6507–18. 10.1002/ece3.3186 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Farré-Armengol  G, Penuelas  J, Li  T  et al.  Ozone degrades floral scent and reduces pollinator attraction to flowers. New Phytol  2016;209:152–60. 10.1111/nph.13620 [DOI] [PubMed] [Google Scholar]
  18. Fowler  RE, Rotheray  EL, Goulson  D. Floral abundance and resource quality influence pollinator choice. Insect Conserv Divers  2016;9:481–94. 10.1111/icad.12197 [DOI] [Google Scholar]
  19. Frasnelli  E, Robert  T, Chow  PKY  et al.  Small and large bumblebees invest differently when learning about flowers. Curr Biol  2021;31:1058–64. e3. 10.1016/j.cub.2020.11.062 [DOI] [PubMed] [Google Scholar]
  20. Gallagher  MK, Campbell  DR. Shifts in water availability mediate plant–pollinator interactions. New Phytol  2017;215:792–802. 10.1111/nph.14602 [DOI] [PubMed] [Google Scholar]
  21. Gao  R, Hu  B, Yuan  Y  et al.  Nitrogen addition affects floral and vegetative traits, reproduction, and pollinator performance in Capsicum annuum L. Ann Bot  2023;132:1131–44. 10.1093/aob/mcad121 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gao  Y, Ma  X, Cooper  DJ. Short-term effect of nitrogen addition on nitric oxide emissions from an alpine meadow in the Tibetan plateau. Environ Sci Pollut Res Int  2016;23:12474–9. 10.1007/s11356-016-6763-5 [DOI] [PubMed] [Google Scholar]
  23. Gargano  D, Fenu  G, Bernardo  L. Local shifts in floral biotic interactions in habitat edges and their effect on quantity and quality of plant offspring. AoB Plants  2017;9:plx031. 10.1093/aobpla/plx031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Greene  DF, Johnson  EA. Seed mass and dispersal capacity in wind-dispersed diaspores. Oikos  1993;67:69–74. 10.2307/3545096 [DOI] [Google Scholar]
  25. Guo  Q, Brown  JH, Valone  TJ  et al.  Constraints of seed size on plant distribution and abundance. Ecology  2000;81:2149–55. 10.1890/0012-9658(2000)081[2149:COSSOP]2.0.CO;2 [DOI] [Google Scholar]
  26. Harper  JL, Lovell  P, Moore  K. The shapes and sizes of seeds. Annu Rev Ecol Syst  1970;1:327–56. 10.1146/annurev.es.01.110170.001551 [DOI] [Google Scholar]
  27. Hass  AL, Kormann  UG, Tscharntke  T  et al.  Landscape configurational heterogenity by small-scale agriculture, not crop diversity, maintains pollinators and plant production in Western Europe. Philos Trans R Soc B Biol Sci  2018;285:20172242. 10.1098/rspb.2017.2242 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hassan  AM, Giovanetti  M, Raweh  HS  et al.  Nectar secretion dynamics of Ziziphus nummularia: a melliferous species of dry land ecosystems. Saudi J Biol Sci  2017;24:1470–4. 10.1016/j.sjbs.2017.01.059 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Heady  HF. The measurement and value of plant height in the study of herbaceous vegetation. Ecology  1957;38:313–20. 10.2307/1931691 [DOI] [Google Scholar]
  30. Heinrich  B, Raven  PH. Energetics and pollination ecology: the energetics of pollinators may have wide implications in floral biology and community ecology. Science  1972;176:597–602. 10.1126/science.176.4035.597 [DOI] [PubMed] [Google Scholar]
  31. Hu  X, Wu  X, Zhou  Q  et al.  Warming causes contrasting spider behavioural responses by changing their prey size spectra. Nat Clim Chang  2024;14:190–7. 10.1038/s41558-023-01918-8 [DOI] [Google Scholar]
  32. Huang  J, Shi  Y, Li  B  et al.  The effects of combined application of chemical fertilizer, functional bacteria and plant hormones on vegetation restoration of desertification land in Zoige Plateau. Plant Soil  2023;491:355–68. 10.1007/s11104-023-06116-1 [DOI] [Google Scholar]
  33. Jacquemyn  H, De Meester  L, Jongejans  E  et al.  Evolutionary changes in plant reproductive traits following habitat fragmentation and their consequences for population fitness. J Ecol  2012;100:76–87. 10.1111/j.1365-2745.2011.01919.x [DOI] [Google Scholar]
  34. Jakobsson  A, Eriksson  O. A comparative study of seed number, seed size, seedling size and recruitment in grassland plants. Oikos  2000;88:494–502. 10.1034/j.1600-0706.2000.880304.x [DOI] [Google Scholar]
  35. Johnson  SD, Peter  CI, Ågren  J. The effects of nectar addition on pollen removal and geitonogamy in the non-rewarding orchid Anacamptis morio. Proc R Soc Lond B Biol Sci  2004;271:803–9. 10.1098/rspb.2003.2659 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kelly  T, Elle  E. Effects of community composition on plant–pollinator interaction networks across a spatial gradient of oak-savanna habitats. Oecologia  2020;193:211–23. 10.1007/s00442-020-04661-5 [DOI] [PubMed] [Google Scholar]
  37. Kuppler  J, Wieland  J, Junker  RR  et al.  Drought-induced reduction in flower size and abundance correlates with reduced flower visits by bumble bees. AoB Plants  2021;13:plab001. 10.1093/aobpla/plab001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Li  J, Yao  Q, Zhou  N  et al.  Modern aeolian desertification on the Tibetan Plateau under climate change. Land Degrad Dev  2021;32:1908–16. 10.1002/ldr.3862 [DOI] [Google Scholar]
  39. Li  FR, Zhang  AS, Duan  SS  et al.  Patterns of reproductive allocation in Artemisia halodendron inhabiting two contrasting habitats. Acta Oecol  2005;28:57–64. 10.1016/j.actao.2005.02.005 [DOI] [Google Scholar]
  40. Librán-Embid  F, Grass  I, Emer  C  et al.  A plant–pollinator metanetwork along a habitat fragmentation gradient. Ecol Lett  2021;24:2700–12. 10.1111/ele.13892 [DOI] [PubMed] [Google Scholar]
  41. McFrederick  QS, LeBuhn  G. Are urban parks refuges for bumble bees bombus spp. (Hymenoptera: Apidae)?. Biol Conserv  2006;129:372–82. 10.1016/j.biocon.2005.11.004 [DOI] [Google Scholar]
  42. Mu  J, Peng  Y, Xi  X  et al.  Domesticated honey bees evolutionarily reduce flower nectar volume in a Tibetan lotus. Ecology  2014;95:3161–72. 10.1890/13-2055.1 [DOI] [Google Scholar]
  43. Mu  J, Peng  Y, Xi  X  et al.  Artificial asymmetric warming reduces nectar yield in a Tibetan alpine species of Asteraceae. Ann Bot  2015;116:899–906. 10.1093/aob/mcv042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Mueller  AL, Berger  CA, Schittenhelm  S  et al.  Water availability affects nectar sugar production and insect visitation of the cup plant Silphium perfoliatum L. (Asteraceae). J Agron Crop Sci  2020;206:529–37. 10.1111/jac.12406 [DOI] [Google Scholar]
  45. Neumüller  U, Burger  H, Krausch  S  et al.  Interactions of local habitat type, landscape composition, and flower availability moderate wild bee communities. Landsc Ecol  2020;35:2209–24. 10.1007/s10980-020-01096-4 [DOI] [Google Scholar]
  46. Oksanen  J, Blanchet  FG, Kindt  R  et al.  The Vegan Package. Community ecology package; R package version 2.7-1. Vienna, Austria: R Core Team, 2025. Available from: https://cran.r-project.org/web/packages/vegan/index.html. [Google Scholar]
  47. Petanidou  T, Goethals  V, Smets  E. The effect of nutrient and water availability on nectar secretion and nectary structure of the dominant Labiatae species of phrygana. Syst Geogr Pl  1999;68:233–44. 10.2307/3668604 [DOI] [Google Scholar]
  48. Phillips  BB, Shaw  RF, Holland  MJ  et al.  Drought reduces floral resources for pollinators. Glob Chang Biol  2018;24:3226–35. 10.1111/gcb.14130 [DOI] [PubMed] [Google Scholar]
  49. Pleasants  JM, Chaplin  SJ. Nectar production rates of Asclepias quadrifolia: causes and consequences of individual variation. Oecologia  1983;59:232–8. 10.1007/BF00378842 [DOI] [PubMed] [Google Scholar]
  50. Pyke  GH. What does it cost a plant to produce floral nectar?  Nature  1991;350:58–9. 10.1038/350058a0 [DOI] [Google Scholar]
  51. Raine  NE, Willmer  P, Stone  GN. Spatial structuring and floral avoidance behavior prevent ant-pollinator conflict in a Mexican ant-acacia. Ecology  2002;83:3086–96. 10.2307/3071844 [DOI] [Google Scholar]
  52. Rauschkolb  R, Li  Z, Godefroid  S  et al.  Evolution of plant drought strategies and herbivore tolerance after two decades of climate change. New Phytol  2022;235:773–85. 10.1111/nph.18125 [DOI] [PubMed] [Google Scholar]
  53. R Core Team . R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2024. [Google Scholar]
  54. Real  LA, Rathcke  BJ. Individual variation in nectar production and its effect on fitness in Kalmia latifolia. Ecology  1991;72:149–55. 10.2307/1938910 [DOI] [Google Scholar]
  55. Reverté  S, Bosch  J, Arnan  X  et al.  Spatial variability in a plant–pollinator community across a continuous habitat: high heterogeneity in the face of apparent uniformity. Ecography  2019;42:1558–68. 10.1111/ecog.04498 [DOI] [Google Scholar]
  56. Rigatti  SJ. Random forest. J Insur Med  2017;47:31–9. 10.17849/insm-47-01-31-39.1 [DOI] [PubMed] [Google Scholar]
  57. Roubik  D, Yanega  D, Buchmann  S  et al.  On optimal nectar foraging by some tropical bees (Hymenoptera: Apidae). Apidologie  1995;26:197–211. 10.1051/apido:19950303 [DOI] [Google Scholar]
  58. Su  R, Dai  W, Yang  Y  et al.  Introduced honey bees increase host plant abundance but decrease native bumble bee species richness and abundance. Ecosphere  2022;13:e4085. 10.1002/ecs2.4085 [DOI] [Google Scholar]
  59. Tadey  M, Tadey  JC, Tadey  N. Reproductive biology of five native plant species from the Monte Desert of Argentina. Bot J Linn Soc  2009;161:190–201. 10.1111/j.1095-8339.2009.01001.x [DOI] [Google Scholar]
  60. Vaudo  AD, Patch  HM, Mortensen  DA  et al.  Bumble bees exhibit daily behavioral patterns in pollen foraging. Arthropod Plant Interact  2014;8:273–83. 10.1007/s11829-014-9312-5 [DOI] [Google Scholar]
  61. Venable  DL, Flores-Martinez  A, Muller-Landau  HC  et al.  Seed dispersal of desert annuals. Ecology  2008;89:2218–27. 10.1890/07-0386.1 [DOI] [PubMed] [Google Scholar]
  62. Wang  R, He  L, Chen  J  et al.  Soil carbon, nitrogen, and potassium regulate herbaceous community stability in the restoration phases of desertification land. Ecol Res  2025;40:539–48. 10.1111/1440-1703.12546 [DOI] [Google Scholar]
  63. Wang  LL, Huang  ZY, Dai  WF  et al.  Mixed effects of honey bees on pollination function in the Tibetan alpine grasslands. Nat Commun  2024;15:8164. 10.1038/s41467-024-52465-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wang  R, Lou  Y, He  L  et al.  Plant functional groups and soil properties vary with the restoration periods in Zoige desertification land. Front Ecol Evol  2023;11:1331618. 10.3389/fevo.2023.1331618 [DOI] [Google Scholar]
  65. Waser  NM, Price  MV. Drought, pollen and nectar availability, and pollination success. Ecology  2016;97:1400–9. 10.1890/15-1423.1 [DOI] [PubMed] [Google Scholar]
  66. Xiao  Y, Li  X, Cao  Y  et al.  The diverse effects of habitat fragmentation on plant–pollinator interactions. Plant Ecol  2016;217:857–68. 10.1007/s11258-016-0608-7 [DOI] [Google Scholar]
  67. Xue  X, Guo  J, Han  B  et al.  The effect of climate warming and permafrost thaw on desertification in the Qinghai–Tibetan Plateau. Geomorphology  2009;108:182–90. 10.1016/j.geomorph.2009.01.004 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

plaf052_Supplementary_Data
plaf052_supplementary_data.zip (973.8KB, zip)

Data Availability Statement

Raw data and R code are available online at https://zenodo.org/records/17054532.

Articles from AoB Plants are provided here courtesy of Oxford University Press

RESOURCES