Abstract
Elevated atmospheric carbon dioxide (eCO2) can affect plant growth and physiology, which can, in turn, impact herbivorous insects, including by altering pollen or plant tissue nutrition. Previous research suggests that eCO2 can reduce pollen nutrition in some species, but it is unknown whether this effect is consistent across flowering plant species. We experimentally quantified the effects of eCO2 across multiple flowering plant species on plant growth in 9 species and pollen chemistry (%N an estimate for protein content and nutrition in 12 species; secondary chemistry in 5 species) in greenhouses. For pollen nutrition, only buckwheat significantly responded to eCO2, with %N increasing in eCO2; CO2 treatment did not affect pollen amino acid composition but altered secondary metabolites in buckwheat and sunflower. Plant growth under eCO2 exhibited two trends across species: plant height was taller in 44% of species and flower number was affected for 63% of species (3 species with fewer and 2 species with more flowers). The remaining growth metrics (leaf number, above-ground biomass, flower size, and flowering initiation) showed divergent, species-specific responses, if any. Our results indicate that future eCO2 is unlikely to uniformly change pollen chemistry or plant growth across flowering species but may have the potential to alter ecological interactions, or have particularly important effects on specialized pollinators.
Similar content being viewed by others
Introduction
Carbon dioxide (CO2), which continues to rise, is recognized as a major greenhouse gas contributing to climate change1. CO2 is also the sole supplier of carbon (C) for plant photosynthesis, and as such, its ongoing increase directly affects plant productivity, essentially providing ‘C-fertilization’. Rising CO2 levels, for example, can alter plant growth and development2,3,4, chemistry5,6, and reproduction and floral biology7,8. Increasing CO2 and corresponding carbon availability to plants leads to an increase in carbon-based compounds and a decrease in the concentration of compounds rich in other elements, such as nitrogen-rich proteins9. While a decrease in protein has been observed across multiple plant species and tissue types10,11,12, some plants may maintain sufficient nitrogen (N) levels and avoid becoming N-limited (i.e., N-fixing legumes), ultimately leading to no or little reductions in protein content13. This suggests that life history traits may influence the degree to which plant species are impacted by elevated CO2 (eCO2) levels.
Initial findings suggest that eCO2 can also affect the plant nutritional rewards for pollinators, which can, in some cases, rely exclusively on floral resources8. For nectar, eCO2 can increase or decrease nectar quantity in some plant species, though for most studied plants, nectar sugar content was not affected by increasing CO214,15,16. Notably, changes to pollen nutrition in response to eCO2 are largely unknown. Ziska et al.17 demonstrated that elevated atmospheric CO2 levels (280–500 ppm) reduced pollen protein content in goldenrod (Asteraceae: Solidago canadensis L.) in both historical specimens and experiments simulating historic and near-future CO2 scenarios. Surprisingly, even short-term exposure (24 h) to eCO2 (500, 1000, and 3000 ppm) in black maple (Sapindaceae: Acer negundo L.) resulted in reduced soluble pollen protein levels18. However, in the same study on black maple, the authors found the opposite trend of increased pollen protein with 6 h of exposure to eCO218, demonstrating the complexity of plant responses to eCO2. Because bees and many other pollinating insects rely on pollen as their primary source of protein, reduced pollen protein under eCO2 could significantly affect pollinator nutrition and health19,20,21, with subsequent impacts on pollinator populations and pollination services22. However, we have little understanding of the generality of these changes in floral nutrition as few flowering species have been studied under eCO2.
Just as changes to atmospheric CO2 can alter plant protein levels, eCO2 is likely to affect the secondary metabolites and amino acids found in pollen (and in other plant tissues23), with potential knock-on effects for pollinators. Some secondary metabolites can, for example, facilitate learning and memory formation in bees (e.g., caffeine24), and the amino acid content of pollen can structure bee foraging preferences25,26. Pollen is chemically complex, and possible effects of most of their constituent compounds on flower visitors are unknown. Still, research on ragweed (Asteraceae: Ambrosia artemisifolia L.) pollen found that eCO2 can alter the expression of secondary plant metabolites27 and has been shown to increase flavonoid metabolites28. Flavonoids are largely responsible for color production in plant tissues, which can facilitate plant-pollinator interactions by attracting pollinators or signaling when flowers have already been pollinated29. As the importance of secondary metabolites and amino acids are elucidated for pollinators, understanding if and how these compounds change in composition and abundance under a future climate with elevated CO2 levels is required.
For vegetative tissues, eCO2 growing conditions typically result in increased growth4. Leaves increase in both biomass and area under eCO2 growing conditions30,31 and plants grow taller, increasing above-ground biomass with exposure to eCO230. Similarly, plants generally produce more flowers, fruits, and seeds when grown under increasing CO2 levels7. However, Jablonski et al.7 found that response to eCO2 varied between crop and wild plant species, with crop plants exhibiting a greater increase in fruits and seeds than wild species when exposed to eCO2. A recent review on eCO2 (mean eCO2 of 730 ppm vs mean ambient CO2 (aCO2) of 360 ppm) and flowering biology found that plants flower nearly four days earlier on average under eCO28. For organisms that rely on plant floral resources, like pollinators, changes to plant growth patterns or flowering phenology could disrupt plant-pollinator interactions, potentially leading to reduced phenological overlap between plants and pollinators32.
Here, we conducted two greenhouse experiments to evaluate how eCO2 growing conditions impact pollen chemistry (nutrition as measured by pollen %N a direct proxy for protein content33, secondary metabolites, and amino acids) and plant growth in fourteen flowering plant species spanning nine plant families and two pairs of functional groups (1): plants that can and cannot fix nitrogen and 2: crop and non-crop plants). We predicted that eCO2 growing conditions: 1) will reduce pollen nutrition (%N) in plant species that cannot fix N due to limited N availability, but would observe no changes in pollen %N in N-fixing plant species which have greater access to N through symbiotic microbes; (2) will alter the content of amino acids and secondary metabolites in pollen34; (3) plants would grow taller with more leaves as increased CO2 increases plant growth rates2, but that flower number will not differ between treatments8,16; and 4) that plants will flower earlier when compared to plants grown in aCO28.
Methods
Two greenhouse studies were conducted to characterize the impacts of eCO2 on pollen chemistry and plant growth in several species of flowering plants. The first study was conducted in Cambridge, MA (USA), and quantified the impacts of eCO2 on pollen or anther chemistry (nutrition, secondary metabolites, and amino acids); if flowers were small, whole anthers were collected and analyzed as a proxy for pollen in some species (as in Ziska et al.17). The second study was conducted in Madison, WI (USA), and quantified impacts on pollen or anther chemistry (nutrition) and plant growth. Focal plant species across these experiments were selected for (a) their importance as forage plants for bees across a range of crop and non-crop flowering plants, (b) taxonomic diversity, and (c) functional diversity (species with root-symbiotic N-fixing bacteria vs non-N-fixing species; crop vs non-crop plant species) (Table 1). Both greenhouse studies were conducted in compliance with relevant institutional, national, and international guidelines and legislation.
Experiment 1-experimental design
We grew plants from seed (source detailed in Table 1) in two adjacent greenhouse chambers each 42.2 m2 in size at Harvard University (Cambridge, MA, USA), one aCO2 and one eCO2 (elevated to a target of 650 ppm CO2) chamber. 650 ppm CO2 was chosen as our eCO2 treatment as this was approximately 200 ppm above current aCO2 levels and within the predicted CO2 range for 210035; CO2 treatments were initiated from planting. Two experimental rounds were conducted with CO2 treatments switching chambers between rounds to account for chamber effects (round 1 was planted in September 2020 and round 2 in January 2021). We planted between 15 plants of each focal species per chamber per round (n = 60 total plants per species; Table 1).
Focal species included (1) buckwheat, (2) melon, (3) poppy, (4) sunflower, (5) tomatillo, (6) tomato, and (7) squash (Table 1). Plants were grown from seed in 6-inch (15.2 cm) diameter pots under 16:8 light:dark photoperiod with 23 and 17 °C day and night temperatures, respectively. Ventilation was provided by manually operated vents and plants were grown in Premier Pro-Mix BX 3.8 ft3/bale with mycorrhizae and provided slow-release fertilizer (1.8 kg/m3 14-14-14 Osmocote with a 3–4 month longevity). Plants were watered until water dripped through the pots either every day or every other day and treated with systemic pesticides for thrips as needed throughout the study.
One greenhouse chamber served as the ambient CO2 (aCO2) chamber. In the second chamber, CO2 was elevated to a target of 650 ppm (eCO2). CO2 levels were maintained in the eCO2 room using a LiCor Li-850 gas analyzer and Arduino-based controller. The LiCor Li-850 monitored CO2 levels, relaying this to the Arduino-based controller which, if low (50 ppm or more below the target CO2 levels), would initiate flow from a CO2 tank (AirGas; via a solenoid-operated flow regulator) to maintain the target CO2 concentration. If CO2 levels were at or above 650 ppm, the Arduino-based controller would not initiate CO2 flow. Observed median CO2 concentrations in the elevated CO2 chambers were 647 and 642 ppm for experimental rounds 1 and 2, respectively (Fig. S1a). CO2 concentrations were quantified in the aCO2 chambers using an identical LiCor Li-850 gas analyzer. Median CO2 concentrations in the uncontrolled chamber were 445 and 438 ppm for rounds 1 and 2, respectively (Fig. S1a).
Experiment 1–pollen collection and processing
In both rounds, pollen and/or anthers (a proxy for pollen17; hereafter referred to as pollen samples) were collected from flowering plants: for the two buzz-pollinated species with poricidal anthers (tomato and tomatillo), pollen was collected using a tuning fork, for melon and buckwheat, anthers were collected using forceps. For poppy, sunflower, and squash, the flowers were gently shaken over weighing paper as they readily dropped pollen. We began collecting pollen samples when plants started flowering and continued to collect pollen throughout bloom, up to five times per individual plant; therefore, pollen samples contain a mix of pollen throughout bloom stages. Pollen samples were collected into a new microcentrifuge tube during each collection event and stored in a −80 °C freezer after collection. Once pollen was collected, samples were pooled across days by individual plant. Pollen samples were allocated for two analyses: targeted metabolomics, where pollen samples were kept frozen and analyzed fresh, and elemental C and N analyses, where pollen samples were dried in a drying oven at 50 °C for 48 h.
Experiment 2–experimental design
We grew plants from seed (source detailed in Table 1) in four 10.2 m2 greenhouse chambers on the campus of the University of Wisconsin–Madison with two chambers experiencing aCO2 and two with eCO2 (with a target of 690 ppm, selected as a realistic 2100 prediction35). We systematically assigned CO2 treatments to the four greenhouse chambers alternating between eCO2 and aCO2. Two experimental rounds were conducted to increase sample size and account for chamber effects, though because rounds 1 and 2 overlapped, chambers were not switched between rounds (round 1 was planted in November 2021 and round 2 in January 2022). We planted 12 pots per focal plant species per greenhouse chamber per round (total plants per species detailed in Table 1) and eCO2 was initiated 14 days after seeds were planted.
Nine flowering plant species with known value to pollinating insects were studied (Table 1): (1) buckwheat, (2) borage, (3) lacy phacelia, (4) red clover, (5) partridge pea, (6) nasturtium, (7) dwarf sunflower, (8) dandelion, and (9) sweet alyssum. All pots (2 in3, or 0.3 L) were sown with at least two seeds and thinned to the strongest seedling on day 14. Plants were re-potted to 6 in (15.2 cm) diameter, 1.9 L, pots on an as-needed basis (i.e., roots growing out of the bottom of the pot) and watered until water dripped through the pots either every day or every other day. Pests were monitored by greenhouse staff; plants were treated for thrips two times throughout the study (once with Hachi-Hachi, Radiant, and Pradia and once with Pylon and Talstar).
In each greenhouse, lighting was set to a 16:8 light:dark photoperiod, and temperatures were set to a minimum of 18 °C during the day, and 15 °C at night. If chambers were experiencing temperatures greater than 27 °C, cooling was automatically provided by vents in the side and ceiling of each chamber. All seeds were sown into ProMix HP Biofungicide + Mycorrhizae medium with a medium feeding rate (84 g/sq ft. of soil) of controlled-release fertilizer (Nutricote Total, 13-11-11, Type 100). Two greenhouse chambers were unmanipulated and served as the aCO2 chambers while in the remaining two chambers, CO2 was elevated to a target of 690 ppm for the eCO2 treatment using the same setup as in experiment 1. Additionally, in each chamber, a CO2 sensor (Adafruit SCD-30, run by Arduino UNO) collected a reading every ten minutes. For the eCO2 chambers, observed median CO2 concentrations were 645 ppm in both rounds and the aCO2 chambers had observed median CO2 levels of 456 and 443 for rounds 1 and 2, respectively (Fig. S1b).
Experiment 2–plant growth data collection
Starting on day 14, and every 7 days thereafter, the number of leaves and the plant’s height was collected for each individual plant. Once plants began producing flowers, flower numbers were recorded daily. Due to time constraints, we stopped counting flowers for sweet alyssum after week 10 (when a plant had > 500 flowers) and buckwheat after week 11 (when a plant had > 100 flowers).
Flower diameter data was collected from five open flowers (or inflorescences in the case of red clover and sunflower) per individual plant for all species except sunflower which only produced a single inflorescence per plant. We measured flower diameter within the first three days of a flower being open. For each flower, three diameter measurements were taken and then these were averaged per flower, then the flower diameter values were averaged for each individual plant, resulting in one flower diameter value per plant. We collected flower diameter data for a minimum of five plants per species in each greenhouse chamber and round.
As CO2 is known to impact plant growth, we destructively sampled above-ground biomass on day 60 for half of the plants grown in round 1 (up to 6 plants per species per greenhouse chamber). Many plants in round 1 had not yet flowered when biomass samples were collected, reducing the total possible number of pollen samples. As a major focus for this study was to investigate how pollen nutrition (in the form of %N) changed in response to elevated CO2, we opted to forego the destructive above-ground biomass sampling in round 2, maximizing the number of plants available for pollen sampling. To measure biomass, plants were cut at ground level, placed into a pre-weighed paper bag, and then dried at room temperature until the weight stabilized. The final measurements of dry plant biomass were obtained following another 14 days of storage at room temperature.
Experiment 2–pollen collection and processing
One pollen or anther sample (pooled across days) was collected from freshly opened (within three days of opening) flowers per round. For borage and partridge pea flowers, we used a VegiBee Sonic Garden Pollinator to release pollen from poricidal anthers directly into microcentrifuge tubes. For buckwheat, lacy phacelia, and sunflowers, when flowers were open and producing pollen, they were held over a plastic weigh boat and gently tapped with a paintbrush, and any pollen dropped was brushed into a microcentrifuge tube. For nasturtium, red clover, and sweet alyssum, whole anthers were collected directly from flowers using fine forceps; for these plant species, we use anthers as a proxy for pollen (as in Ziska et al.17) and refer to these samples as pollen samples hereafter. Dandelions in this study never flowered. Pollen samples were stored in a −4 °C freezer until sample sizes were large enough for pollen to fill the bottom 3–5 mm of the microcentrifuge tube (pooled across collection days if needed), then were dried at 60 °C for 48 h in a drying oven.
Pollen C and N analysis
To collect C and N composition data, pollen samples were flash-analyzed in the Jackson Lab at the University of Wisconsin–Madison. Pollen and anther samples were processed and analyzed in the same manner. Pollen samples were placed into tin capsules and the samples were carefully rolled to enclose all contents. Next, samples were flash combusted using a Thermo Flash EA1112 Carbon Nitrogen analyzer to obtain values for the %N and %C present in each sample. The %N value serves as an approximation for protein present within a pollen sample33. In experiment 1, 1.0 ± 0.5 mg samples of dried pollen were analyzed, while samples analyzed from experiment 2 contained a dry pollen mass of 1.5 ± 0.5 mg.
Pollen secondary chemistry
Targeted metabolomics were performed on pollen or anther samples for a subset of plant species (buckwheat, sunflower, squash, tomato, and poppy) to assess how pollen secondary chemistry is affected by changing CO2 levels. 5 mg samples of fresh pollen were prepared following the methods in Palmer-Young et al.64: the pollen sample was first combined with 1 ml of methanol then incubated for 10 min in an ultrasound bath, then overnight at room temperature. Samples were centrifuged for 10 min at maximum speed and then the supernatants were transferred to a 1.5 ml glass autosampler vial. Next, samples were dried under N2 flow and then resuspended in methanol (at a rate of 500 ul/5 mg). Secondary chemistry was quantified using a ThermoFisher IDX LC-MS using a Kinetex F5 column at the Harvard Center for Mass Spectrometry (Cambridge, MA, USA). 5 µL of each sample was injected on a Kinetex F5 column (2.6 um, 150 × 2 mm, Phenomenex), maintained at 30 °C. Mobile phases were A: water, 2.5 mM ammonium acetate and B: Methanol:Isopropyl Alcohol 1:1. The flow rate was maintained at 0.2 mL min-1. The gradient was as follows: 5 min at 0% B, then to 100% B in 20 min, followed by 10 min at 100% and finally 6 min re-equilibration at 0% B. Data was acquired in MS1 for all samples, with switching polarities at 120,000 resolution and a scan range of 65 to 1000 m/z. Retention time and ions were confirmed with standards and for all targeted compounds. Tracefinder (ThermoFisher Scientific) was used to integrate the corresponding peaks in each sample.
Data analysis
Statistical analyses were carried out in R v. 4.2.265. For both experiments, we included plant species in our analyses that had at least three data points per treatment.
When we used models (linear models, linear mixed effect models, generalized linear mixed models) to compare the effects of CO2 on pollen chemistry or plant growth, model fit was evaluated using the package DHARMa66. For linear models, the statistical significance of the fixed effects was assessed using a Type III Analysis of Variance with Satterthwaite’s method using the function “anova” with the package lmerTest67 to evaluate main and interactive effects for model terms. We report significant fixed effects using Type II analysis of deviance (Type II Wald Chi-Square tests) for generalized linear mixed-effect models. Specific models for each variable measured are outlined below. To investigate species-specific responses to CO2 treatment, we used post-hoc pairwise comparisons (function “pairs” in emmeans68).
In the text of our results, alongside significant raw p-values, we report the strength of effects using the language of evidence (per Fig. 1 of Muff et al.69): little or no evidence (1 > p > 0.1), weak evidence (0.1 > p > 0.05), moderate evidence (0.05 > p > 0.01), strong evidence (0.01 > p > 0.001), and very strong evidence (0.001 > p). Where little or no evidence or weak evidence (p > 0.05) was found, we do not report the p-value in text and only report the relevant language of evidence69.
Experiment 1. After omitting small sample sizes (plant species with fewer than three data points per treatment), we analyzed 110 pollen samples from six plant species grown in experiment 1. For melon (pollen samples: eCO2: n = 3, aCO2: n = 3), poppy (pollen samples: eCO2: n = 13, aCO2: n = 14), sunflower (pollen samples: eCO2: n = 6, aCO2: n = 12), tomatillo (pollen samples: eCO2: n = 9, aCO2: n = 16), and tomato plants (pollen samples: eCO2: n = 6, aCO2: n = 6), we only had sufficient pollen samples from round 1 to analyze, while for squash, we had enough samples from rounds 1 and 2 to analyze both (pollen samples: eCO2: n = 11, aCO2: n = 11). To evaluate the effects of CO2 on pollen nutrition (%N, %C, and C:N), we used linear models (“lm” in the package lme470) with CO2 treatment, round, plant species, and greenhouse chamber as fixed effects and evaluated the interaction between CO2 treatment and plant species (i.e., lm(%N ~ CO2* Plant_species + Round + Chamber)).
In total, 119 pollen samples were analyzed for changes to secondary compounds using targeted metabolomics, including 67 samples from round 1 (buckwheat: aCO2: n = 5, eCO2: n = 5; poppy: aCO2: n = 10, eCO2: n = 10; squash: aCO2: n = 4, eCO2: n = 3; sunflower: aCO2: n = 10, eCO2: n = 8; tomato: aCO2: n = 6, eCO2: n = 6) and 52 samples from round 2 (buckwheat: aCO2: n = 10, eCO2: n = 10; squash: aCO2: n = 8, eCO2: n = 9; sunflower: aCO2: n = 8, eCO2: n = 7). To evaluate changes to pollen secondary chemistry using targeted metabolomics, we performed principal components analysis on nine amino acids (arginine, histidine, isoleucine + lycine, lysine, methionine, phenylalanine, threonine, tryptophan, and valine). Then, for each experimental round, we ran a MANOVA to determine if CO2 treatment, plant species, or the interaction between the two influenced amino acid composition. In round 1, samples from tomato and squash were lacking data for several amino acids and could not be included in amino acid analyses. For the remaining secondary metabolites, we analyzed changes in each metabolite separately since many metabolites were only detected in a subset of plant species. Details on which plant species and experimental round(s) were analyzed for each compound are provided alongside the model outputs. To investigate changes in metabolites across CO2 treatments and plant species, we used a t-test (for nicotine, only detected in buckwheat pollen in round 1) or linear models (for caffeine, chlorogenic acid, cinnamic acid, eugenol, gallic acid, kaempferol, p-coumaric acid, and quercetin). The linear models used to evaluate secondary metabolites included CO2 treatment, plant species, experimental round (where possible), and greenhouse chamber as fixed effects and investigated the interaction between CO2 treatment and plant species (i.e., lm(gallic_acid ~ CO2*Plant_species + Round + Chamber)). To improve data normality, the values of caffeine, chlorogenic acid, cinnamic acid, kaempferol, p-coumaric acid, and quercetin were natural-log-transformed before analysis.
Experiment 2. We collected two rounds of plant growth data for borage, buckwheat, red clover, lacy phacelia, nasturtium, sweet alyssum, and sunflower. Partridge pea exhibited high mortality rates during round 2, thus only data from round 1 were analyzed. Dandelion did not flower during our experiment and, as a result, were also removed from round 2. However, data on dandelion growth (height, biomass, leaf number) are included from round 1.
We analyzed 328 pollen samples across eight species in experiment 2 after omitting groups with small samples: borage (eCO2: n = 23, aCO2: n = 25), buckwheat (eCO2: n = 27, aCO2: n = 25), red clover (eCO2: n = 14, aCO2: n = 12), lacy phacelia (eCO2: n = 29, aCO2: n = 32), nasturtium (eCO2: n = 20, aCO2: n = 19), sweet alyssum (eCO2: n = 15, aCO2: n = 23), and sunflower (eCO2: n = 33, aCO2: n = 31).
For biomass, we analyzed 194 samples for borage (eCO2: n = 11, aCO2: n = 10), buckwheat (eCO2: n = 10, aCO2: n = 11), red clover (eCO2: n = 11, aCO2: n = 12), dandelion (eCO2: n = 11, aCO2: n = 12), lacy phacelia (eCO2: n = 12, aCO2: n = 12), nasturtium (eCO2: n = 10, aCO2: n = 12), partridge pea (eCO2: n = 7, aCO2: n = 6), sweet alyssum (eCO2: n = 12, aCO2: n = 12), and sunflower (eCO2: n = 12, aCO2: n = 11).
We analyzed 380 samples for flowering start date, in analyses borage (eCO2: n = 21, aCO2: n = 22), buckwheat (eCO2: n = 36, aCO2: n = 34), red clover (eCO2: n = 23, aCO2: n = 15), lacy phacelia (eCO2: n = 24, aCO2: n = 22), nasturtium (eCO2: n = 24, aCO2: n = 17), partridge pea (eCO2: n = 5, aCO2: n = 6), sweet alyssum (eCO2: n = 45, aCO2: n = 29), and sunflower (eCO2: n = 29, aCO2: n = 28).
Flower diameter analyses included measurements from 349 plants borage (eCO2: n = 21, aCO2: n = 22), buckwheat (eCO2: n = 27, aCO2: n = 28), red clover (eCO2: n = 24, aCO2: n = 17), lacy phacelia (eCO2: n = 22, aCO2: n = 20), nasturtium (eCO2: n = 18, aCO2: n = 16), partridge pea (eCO2: n = 5, aCO2: n = 3), sweet alyssum (eCO2: n = 33, aCO2: n = 34), and sunflower (eCO2: n = 31, aCO2: n = 28).
For analyses on pollen nutrition (%N, %C, and C:N), biomass, flowering start date, and flower diameter, for which we have one data point per individual plant, we used linear models (“lm” in the package lme470) with CO2, plant species, round, and greenhouse chamber as fixed effects and evaluated the interaction between CO2 treatment and plant species (i.e., lm(%N ~ CO2*Plant_species + Round + Chamber)). Flowering start date and flower diameter were natural-log transformed prior to performing linear model analyses.
Flower number, height, and leaf number data were collected multiple times on the same individual plant (each week for height and leaf number, daily from flowering onset for flower number) and for these data sets, to account for plant growth over time, we additionally include a temporal measure in analyses: weeks since planting. To account for variation within individual plants, we also include individual plant identity (a unique identifier for each plant) as a random effect in our models.
To evaluate trends in flower number under eCO2, we used a subset of our data, only including one data point per individual plant per week (the value on day 7 and every seven days after that). Since flower number was zero-inflated count data, we used a generalized linear mixed model with a negative binomial distribution (“glmer.nb” in lme470) to evaluate CO2, plant species, round, and greenhouse chamber as fixed effects and weeks since planting as a random effect, we also investigated whether there was an interaction between CO2 treatment and plant species (glmer.nb(flowers ~ CO2*Plant_species + Round + Chamber + (1|Week)); when we included individual plant identity in our model to evaluate flower number, the model did not converge, therefore, we removed this random effect from the model.
For plant height and leaf number, data were natural log-transformed (i.e., ln(height)) to normalize the data sets. Then, evaluated using linear mixed-effect models with CO2, plant species, round, weeks since planting, and greenhouse chamber as fixed effects along with the interaction between CO2 treatment and plant species and individual plant identity as a random effect (i.e., lmer(ln(height) ~ CO2*Plant_species + Round + Week + Chamber + (1|Plant_identity))).
Functional group comparisons. We were only able to analyze pollen nutrition of one species of N-fixing plant, red clover, preventing us from exploring the relationship between N-fixing plants and non-fixing plants with respect to pollen nutrition. To compare frequency of response to eCO2 between crop and non-crop species, we used a Fisher’s exact test.
Results
Experiment 1–impacts of eCO 2 on pollen nutrition
Nitrogen content. When we compared pollen %N for all plant species, we found very strong evidence that plant species predicted %N (p < 0.0001), but no evidence that experimental round, greenhouse chamber, CO2 treatment or the interaction between CO2 treatment and plant species were predictors of %N (summarized in Table 2; effect sizes in Table S1). Consistently, post-hoc tests revealed no evidence that CO2 affected pollen %N for any species (Fig. 1a; statistics summarized in Table S2).
Carbon content. When all species were compared together, we found strong evidence that CO2 treatment (p = 0.001) and very strong evidence that plant species (p < 0.0001) and experimental round (p < 0.0001) affected %C in pollen samples, but no evidence that greenhouse chamber nor the interaction between CO2 treatment and plant species did (Fig. 1b; Table 2; Table S1). Post-hoc tests found no evidence for effects of eCO2 on pollen %C in any plant species (Table S2).
C:N ratios. When the C:N ratio was analyzed, we found very strong evidence that plant species (p < 0.0001) and moderate evidence that the interaction between CO2 and plant species (p = 0.03) affected C:N with no evidence that CO2 treatment, experimental round, or greenhouse chamber influenced C:N (Table 2; Table S1). For individual species, post-hoc tests found strong evidence (p = 0.01) that eCO2 increased C:N ratio in melon pollen (Fig. 1c; Table S2).
Experiment 1–targeted metabolomics
In round 1, we found strong evidence that CO2 treatment and very strong evidence that plant species affected pollen amino acid composition, with no evidence that the interaction between CO2 and plant species influenced pollen amino acids (MANOVA: CO2 treatment: F2, 34 = 10.49, p = 0.0003, plant species: F4, 70 = 105.4, p < 0.000001; CO2 treatment x plant species: F4, 70 = 1.13, p = 0.35; Fig. 2a). When species were compared separately, we found no evidence for an effect of CO2 on amino acid composition (MANOVA: buckwheat: CO2 treatment: F2, 5 = 1.26, p = 0.36; poppy: CO2 treatment: F2, 17 = 3.13, p = 0.07; sunflower: CO2 treatment: F2, 10 = 1.0, p = 0.4). For round 2, we found very strong evidence for an effect of plant species on pollen amino acids, but no evidence that CO2 nor the interaction between the two affected pollen amino acids (MANOVA: CO2 treatment: F2, 37 = 1.72, p = 0.19, plant species: F4, 76 = 78.65, p < 0.000001; CO2 treatment x plant species: F4, 76 = 2.02, p = 0.1; Fig. 2b).
For specific compounds, when we consider the effect of CO2 on secondary metabolites, only caffeine demonstrated evidence for a weak effect of CO2 (p = 0.07) and two compounds, chlorogenic acid and quercetin, demonstrated strong evidence (p ≤ 0.01) for an interaction between CO2 treatment and plant species (Table 3; effect sizes in Table S3). Post-hoc tests (Table S4) revealed strong evidence (p = 0.001) that the amount of chlorogenic acid in buckwheat pollen was affected by CO2 treatment with eCO2 resulting in a decrease in chlorogenic acid (by 75%). For quercetin, post-hoc tests (Table S4) found strong evidence (p ≤ 0.04) that both buckwheat and sunflower pollen were affected by CO2 treatment, altering quercetin in pollen by −58% and + 525%, respectively.
Experiment 2 – impacts of eCO 2 on pollen nutrition
Nitrogen content. We found very strong evidence that plant species (p < 0.0001) and strong evidence that experimental round (p = 0.01) affect %N, but no evidence that greenhouse chamber, CO2 treatment, or the interaction between plant species and CO2 affected pollen %N (Fig. 3a, statistical models summarized in Table 2; effect sizes in Table S5). Post-hoc tests revealed moderate evidence that eCO2 increased %N in buckwheat pollen (by 2.1%) (Table S6).
Carbon content. In the full model, we found very strong evidence that %C was affected by round and plant species (p < 0.0001 and < 0.0001, respectively), but no evidence that greenhouse chamber, CO2, nor the interaction between CO2 and species influenced %C (Fig. 3b, Table 2; Table S5). We found no evidence from post-hoc tests that eCO2 affected the %C in pollen for any individual species (Table S6).
C:N ratios. For C:N ratio in pollen, we found very strong evidence that plant species (p < 0.0001) and strong evidence that greenhouse chamber influenced C:N (p = 0.04) but no evidence that CO2 treatment, experimental round, or the interaction between CO2 and plant species affected C:N ratio (Fig. 3c, Table 2; Table S5). Post-hoc tests revealed strong evidence (p = 0.007) that the C:N ratio in buckwheat pollen decreased (by 9.7%) when grown in eCO2 (Table S6).
Experiment 2–plant growth
Plant height. For plant height, we found moderate evidence that CO2 (p = 0.04) and very strong evidence that plant species, weeks since planting, and greenhouse chamber (all p-values < 0.0001) affected height with no evidence that experimental round or the interaction between CO2 and plant species influenced height (Table 4; effect sizes in Table S7). When comparing plant species with post-hoc tests, four species exhibited at least moderate evidence (all p ≤ 0.03) for an effect of CO2 on height, with five species, borage, red clover, nasturtium, sunflower and partridge pea, demonstrating no effect (Fig. 4, Table S8). At week 10, plants grown in eCO2, were on average, taller for four species: buckwheat (+ 19.6%), dandelion (+ 17.5%), lacy phacelia (+ 15.1%), and sweet alyssum (+ 17.7%).
Leaf number. We found strong evidence that plant species, experimental round, and week (all p < 0.0001), weak evidence that greenhouse chamber (p = 0.08), and no evidence that CO2 treatment, nor the interaction between plant species and CO2 treatment affected leaf number (Table 4; effect sizes in Table S9). When species were analyzed using post-hoc tests, we found moderate evidence that eCO2 reduced leaf number for nasturtium (p = 0.03; average change at week 10: −21%) (Fig. 5; Table S8).
Biomass. For above-ground biomass of plants in round 1, we found very strong evidence that plant species (p < 0.0001), strong evidence that the interaction between CO2 treatment and species affected biomass (p = 0.007), and moderate evidence that CO2 and greenhouse chamber influenced biomass (p = 0.02 and 0.03, respectively) (Table 4; effect sizes in Table S10). In post-hoc tests, partridge pea was the only species to show strong evidence (p < 0.001) for an effect of eCO2 on plant biomass (Fig. 6; Table S8), with plants grown under eCO2 having more (297%) biomass than those grown under aCO2.
Flower number. When we examined flower number across treatments and over time, we found very strong evidence that plant species, round, and the interaction between CO2 treatment and plant species (all p ≤ 0.0001; Table 4; effect sizes in Table S11) and strong evidence that greenhouse chamber (p = 0.006) affected flower number, with no evidence that CO2 did. When species were analyzed using post-hoc tests, we found very strong evidence that buckwheat, red clover, and nasturtium flower number (all p < 0.001; Table S8), strong evidence that partridge pea and lacy phacelia flower number (p = 0.006 and 0.01, respectively), and weak evidence that borage (p = 0.09) flower number was affected by CO2 treatment (Fig. 7; Table S8). Buckwheat and lacy phacelia grown in eCO2 had more flowers while red clover, nasturtium, and partridge pea plants had fewer, compared to those grown in aCO2 (Fig. 7).
Flowering initiation. We examined whether CO2 treatment influenced when plants began flowering (Fig. 8a) and found very strong evidence that plant species (p < 0.0001), weak evidence that CO2 (p = 0.07) and the interaction between CO2 and plant species (p = 0.08), but no evidence that round or greenhouse chamber affected flowering initiation (Table 4; effect sizes in Table S12). When species were analyzed separately with post-hoc tests, we found strong evidence (p = 0.001) that nasturtium flowered later in eCO2 (8.6 days later, 13.4%; Fig. 8a; Table S8).
Flower diameter. We found very strong evidence that flower diameter was affected by plant species (p < 0.0001), strong evidence that flower diameter was affected by CO2 treatment and greenhouse chamber (p = 0.002 and 0.004, respectively), and no evidence that the experimental round or the interaction between CO2 and plant species effected flower diameter (Fig. 8b; Table 4; effect sizes in Table S13). When plant species were analyzed separately using post-hoc tests, we found no evidence for an effect of CO2 on flower diameter for any species (Fig. 8b; Table S8).
Functional group comparisons
We were unable to evaluate changes in pollen nutrition (%N) across either pair of functional groups (N-fixing ability, crop status) as only one species, buckwheat, exhibited a response to CO2 treatment (Table 5).
The two N-fixing species, red clover and partridge pea, did not exhibit consistent nor unique patterns in response to eCO2 (Table 5): partridge pea in eCO2 increased in biomass, but not height, while red clover in eCO2 demonstrated the opposite pattern: increased height but no change in biomass. Both species had fewer flowers in eCO2, but so did nasturtium, a non-N fixing plant.
When we compare the number of variables that responded to eCO2 between crop vs non-crop species, we found no evidence (Fisher’s exact test; p = 0.14) that crop plants (n = 5 species) responded to eCO2 (7.7% of the time) more or less often than non-crop plants (21.4%; n = 9 species) (Table 5).
Discussion
We found species-specific effects on pollen composition and plant growth that were only partially consistent with our initial predictions, painting a complex picture of the effects of eCO2 on plant physiology, and, in turn, anticipated broader ecological effects. Across the variables we measured, and the plant species studied, some species and variables showed strong responses to eCO2 (i.e., buckwheat, plant height), while others demonstrated few or no responses (i.e., %C; sunflower) (Table 5). Pollen nutrition (%N), which we predicted would decrease for non-N fixing plants, did not change for the sole N-fixing plant (red clover) evaluated, nor for ten of eleven non-N fixing species (Figs. 1, 3; Table 5), requiring further study to elucidate potential trends across these functional groups. While we expected both pollen amino acids and secondary metabolites to change in response to eCO2, only two secondary metabolites (out of nine) were affected by CO2 treatment in up to two species (Table S4; Table 5). As we predicted, nearly half of the studied species studied here (44%) grew taller in eCO2 (Fig. 4). However, our prediction that flower number would not change across CO2 treatments was not true for most (63%) species (Fig. 7). Among species that did show significant effects of eCO2 on flower number, the direction of those responses diverged (3 species with fewer and 2 with more flowers). Plants in our study did not flower earlier or have changes in flower diameter when grown in eCO2, contrary to our prediction and previous research on other plant species8,71,72. The results here underscore the complexity of the effects of eCO2 on plant growth and physiology and are consistent with broad variability in CO2 responses found in the literature.
Pollen nutrition (%N, %C, and C:N) for only two plant species (buckwheat and melon) across both experiments were affected by CO2 treatment. Specifically, pollen %N (a direct proxy for pollen protein33) was only significantly affected by eCO2 for one plant species. Despite the increased carbon available in eCO2 growing conditions, we also found no evidence for carbon-loading (increased %C) in the pollen samples (Figs. 1b, 3b) nor many differences in C:N ratios (only buckwheat and melon; Fig. 1c, Fig. 3c) suggesting that eCO2 is not consistently altering the composition of compounds present in pollen (e.g., the ratio of C-rich:N-rich compounds was consistent across CO2 treatments) and is consistent with our findings from targeted metabolomic analyses (Fig. 2; Table 3). The general lack of C-loading in pollen we report here (Table 5) does not follow the well-documented trends of C-loading in other vegetative plant tissues10,11,12 nor the findings of Ziska et al.17, which found that under eCO2, goldenrod (Solidago canadensis) had higher C:N ratios, indicative of C-loading. Given the role of pollen in plant reproduction, pollen nutrition may exhibit stronger compensatory regulation than the chemistry of other plant tissues.
Changes to flowering patterns and floral reward nutrition are relevant for pollinator health and the critical pollination services they provide (e.g., flowering patterns73; pollen nutrition74). Our results suggest that the protein content of pollen may not show widespread changes under eCO2 (given the diverse and inconsistent responses across species; Fig. 3) but may alter the composition of amino acids in pollen (Fig. 2) and some secondary metabolites (Table 3). Although changes to amino acids were small (Fig. 2), amino acid composition structures bee foraging preferences25,26, therefore, even small changes to the relative abundance of these compounds in pollen could potentially modify pollinator floral preferences. When we investigated the effects of CO2 on secondary metabolites in pollen, which can guide associational memory in pollinators24 or chemically defend pollen64,75,76, we found that two chemical defense compounds, chlorogenic acid and quercetin, were affected by eCO2 in two plant species. Changes to pollen chemical defenses may reduce pollination success by making pollen grains more vulnerable to colonization by microbes or less competitive against conspecific pollen grains76. For pollinators, while the role of many secondary metabolites or amino acid composition in pollen are largely unknown, changes to these compounds could have important consequences for plant-pollinator interactions and require further study77,78.
In addition to the effects on pollen nutrition, other changes to flowering may have important consequences for pollinators. For example, changes to flowering time8 (i.e., later flowering by nasturtium, Fig. 8a) can result in phenological mismatches, where plants and pollinators are no longer active at the same time32. However, our findings on nasturtium flowering are inconsistent with another previous study on this species that found when exposed to 750 ppm eCO2 (vs. 380 CO2 ppm in aCO2 treatment) there was no difference in nasturtium flowering time (n = 43 plants16). The same study by Lake and Hughes16 found no difference in the number of flowers nasturtium produced, though our study found that this plant had fewer flowers under eCO2 (Fig. 7), though CO2 treatments and growing temperatures varied slightly between studies (day and night temperatures: 22 and 16 °C, respectively), which can influence CO2 responses31,79. In our study, red clover produced fewer flowers (a finding supported by Rusterholz and Erhardt15; n = 40 plants) as did partridge pea and nasturtium, but buckwheat and lacy phacelia plants produced more flowers under eCO2 (Fig. 7). Flower size changes may alter the potential pollinator community as changes to flower size may result in inaccessible floral resources for pollinators (i.e., flowers become too deep for pollinators to reach nectar, documented in bumble bees80). Here, we found no changes to flower size in any species (Fig. 8b), which contradicts findings for sunflower by Maia et al.71 where increased CO2 led to larger sunflower inflorescences (hybrid BRS 323; eCO2 of 800 ppm vs. 400 ppm in aCO2) and for pumpkin by Hoover et al.72 finding that flowers were smaller in eCO2 (Cucurbitaceae: Cucurbita maxima ‘Little Cutie’, eCO2 of 700 ppm vs. 360 ppm in aCO2). While we, as well as previous research, report changes to flowering patterns and floral rewards under eCO28, these findings are not uniform across species (Table 5), making it hard to draw general predictions on how future CO2 levels are likely to impact pollinators via changes to their food resources. However, variable responses across species may still have important ecological impacts. For instance, when specialist bees are considered, they may be more vulnerable to these changes if their preferred floral resources are affected by eCO2 (e.g., goldenrod17, Solidago sp. specialists81).
We had initially predicted that in eCO2 growing conditions, other nutrients, like N, would become limited and thus result in reductions in N-rich compounds like proteins. Therefore, we included plant species in this study that can and cannot fix N (e.g., sunflowers cannot, while red clover can). We had high mortality rates with partridge pea, preventing us from including this species in pollen nutrition analyses and from comparing changes in %N across functional traits (N-fixers vs non-N-fixers). While red clover, which can fix N, did not show an effect of eCO2 on pollen %N, ten of eleven non-N-fixing plant species also were unaffected by eCO2 when pollen %N was considered (Table 5; Fig. 3a). However, these findings may be confounded by the nitrogen fertilization that we used. In the study by Ziska et al.17 that demonstrated the negative effects of eCO2 on pollen protein in goldenrod, plants were grown outside either as wild, unmanaged plants or, outdoors in FACE (Free-Atmospheric Carbon Experiment) plots without the addition of N fertilizer. As both experiments included in our study took place in a greenhouse setting, they required an addition of some N fertilizer to sustain plant growth in pots over time. While the feeding levels utilized in this study were low, it is possible that the addition of even limited N fertilizer was enough to overcome the negative impacts of eCO2 on pollen protein. The interactive effects of eCO2 and N fertilization on plant growth and pollen protein content are not well understood and warrant further investigation.
Crop species are often bred for characteristics desirable to humans like climate resilience82, palatability83,84, and quick growth85, and as such, may respond to environmental stimuli differently than their wild counterparts. When we considered crop and non-crop plants in this study, we found that non-crop plants were no more likely to respond to CO2 treatment than were crop species (Table 5), though for most crop species studied, we were only able to evaluate data on pollen chemistry and not growth. It is also possible that the crop plants, due to selective breeding, may already be growing at their limit and cannot grow faster even with increased CO2 availability. In contrast, wild plants, which have not been intentionally bred for fast growth, may still be able to leverage the excess CO2 to increase their growth. Furthermore, previous work has shown that crops grown in eCO2 experience a greater increase in fruits and seeds than wild species7, which was not investigated here. When crop species are bred into cultivars, they have a distinct set of traits86,87,88, which may not be present across cultivars, meaning that our findings on specific cultivars of crop species here may not be generalizable across cultivars. Future research investigating the role of eCO2 on crop species should incorporate cultivar to explore this relationship.
Given the well-documented positive relationship between increasing CO2 and plant biomass30,31, we expected to find an increase in above-ground biomass for all studied plant species. Here, only partridge pea demonstrated increased biomass when grown under eCO2 (Fig. 6) despite the increase in height for seven of nine species grown in eCO2 (Fig. 4; Table 5), and established positive relationship between plant height and biomass89. Height was repeatedly measured each week throughout the study allowing us to investigate trends over time, while biomass was measured once, on day 60, which may not have been enough time in eCO2 growing conditions for the species in this study to have significantly altered biomass. We report no effect of eCO2 on biomass for sunflowers, which contradicts the findings of two recent studies (hybrid BRS 323, eCO2: 800 ppm vs. aCO2: 400 ppm71; hybrid KBSH-1, eCO2: 550 or 700 ppm vs. aCO2: 380ppm90). However, our findings on dandelion biomass do align with findings from Thomas and Bazzaz91, which also found no change in dandelion biomass under eCO2 (700 ppm vs. aCO2: 350 ppm). While our findings on above-ground biomass partially align with existing literature, intraspecific responses to eCO2 are known92.
Responses to CO2 can be modulated by other environmental factors (e.g., soil nutrients93), which could account for some of the observed intraspecific differences between studies. For example, temperature is known to interact with CO2; in combination with warmer temperatures, eCO2 increases plant growth when compared to plants grown in either eCO2 or warmer temperatures (Solanaceae: Capsicum sp.31; Fabaceae: Stylosanthes capitata79). Here, experiment 2 took place during winter, which in northern North America, has relatively short amounts of daylight, and it is possible that variation in growth conditions (i.e., sunlight, temperatures) modulated plant responses to eCO2 in this study, even within a controlled greenhouse setting. Such modulation by environmental conditions would likely further complicate the challenge of predicting the effects of eCO2 on pollen nutrition and should be further investigated, especially given the future climate predictions of increased CO2, more extreme weather, and warmer temperatures35.
While plant responses to eCO2, both in the forms of growth and chemistry, are well-documented in the literature2,3,4,5,6,7,8, this study, in contrast, reports few responses to eCO2 that were often small and species-specific (Table 5). When considering all plant species together, this study found a significant effect of round for 70% (7 of 10) of the measured variables with data available across rounds (Table 5). The strong effect of round documented in this study indicates that even slight shifts in environmental conditions (e.g., planting dates six weeks apart even within a controlled, greenhouse environment) can influence plant growth. Notably, both studies described in this paper took place in greenhouses but outside of the natural growing season for North America (fall through early spring), and the corresponding reduced natural sunlight (despite grow lights in the greenhouse) and lower temperatures may have influenced plant responses to eCO2. In contrast, the study by Ziska et al.17 not only took place during the summer, but plants were grown wild, outdoors, and unconstrained by pots, which could also have impacted growth patterns and plant responses to eCO2. There may also be a CO2 saturation point where responses to eCO2 are diminished with increased CO2 exposure, which could also explain differential findings across studies and should be considered in future research. Another potential contributor to differential responses in plant growth could be differences in sample sizes, or relatively small sample sizes (e.g., n of ~ 40 plants per species15,16 vs n of ~ 48–96 plants per species in this study; Table 1) possibly leading to divergent results. Lastly, the duration of CO2 exposure also appears to affect the directionality and size of plant responses. For example, Silva et al.18 found that in black maple, very short (6 h) exposure to eCO2 (500, 1000, 3000 ppm) either did not affect, or increased pollen protein, but exposure to the same CO2 treatments for 24 h, led to a decrease in pollen protein. The findings from our study, when contextualized within the existing literature, underscore the sensitivity of plants to CO2 levels and the complexity of studying the effects of eCO2 on plant growth and chemistry.
A potential limitation of our study design, which measured multiple individual plants within greenhouse chambers, is the possibility of pseudoreplication94. In ecology, potential non-independence of data (e.g., arising from blocked treatments or sampling) is a widespread challenge94,95,96 and can lead to spurious findings of false positive results (i.e., finding a significant relationship when there isn’t one, or Type I error). However, there is mixed evidence for the importance of pseudoreplication95,96,97,98,99, with some authors arguing that sampling many individuals within a population may result in non-independence among samples, but not necessarily pseudoreplication100, or that measuring multiple individuals within a given experimental setting can help to hone estimates of variance101. To account for potential sources of non-independence in our data, we performed multiple experimental rounds and included greenhouse chamber as a fixed effect in all models to assess chamber-level variance. Given that we report few significant responses of eCO2 by any plant species studied here (17 responses out of 96 measured), we consider it unlikely that we are over-reporting the impacts of CO2 on plant growth or pollen nutrition.
Overall, we documented the effects of eCO2 growing conditions on the growth habits, and pollen nutrition of fourteen flowering plant species spanning phylogenetic and functional diversity, the largest species-level evaluation of plant characteristics to date. Responses to eCO2 are small where present and vary across species. Pollen protein does not appear to uniformly decrease across the species studied here, nor are other flowering traits consistently affected. While our findings suggest that elevated atmospheric CO2 may not have widespread, consistent negative effects on pollinator nutrition, divergent responses across species may still have important effects on plant-pollinator communities, as well as particular (e.g., specialist) pollinator taxa. Similarly, patterns in plant growth under eCO2 were only consistent for plant height. The findings of this study highlight the need for further studies exploring how plant growth and floral nutrition are likely to change in a future with increasing levels of CO2 and the consequences for plant-pollinator interactions.
Data availability
All data produced in this study can be found on Dryad (https://doi.org/10.5061/dryad.70rxwdc5n). Code used for analysis and figure-making can be found on GitHub (https://github.com/Crall-Lab/eCO2_SciReports.git).
References
Crutzen, P. J. The “Anthropocene.” In Earth System Science in the Anthropocene (eds Ehlers, E. & Krafft, T.) (Springer, 2006). https://doi.org/10.1007/3-540-26590-2_3.
Pritchard, S. G., Rogers, H. H., Prior, S. A. & Peterson, C. M. Elevated CO2 and plant structure: A review. Glob. Change Biol. 5, 807–837 (1999).
Woodward, F. I. Potential impacts of global elevated CO2 concentrations on plants. Curr. Opin. Plant Biol. 5, 207–211 (2002).
Bhargava, S. & Mitra, S. Elevated atmospheric CO2 and the future of crop plants. Plant Breed. 140, 1–11 (2021).
AbdElgawad, H., Zinta, G., Beemster, G. T. S., Janssens, I. A. & Asard, H. Future climate CO2 levels mitigate stress impact on plants: Increased defense or decreased challenge?. Front. Plant Sci. https://doi.org/10.3389/fpls.2016.00556 (2016).
Dong, J., Gruda, N., Lam, S. K., Li, X. & Duan, Z. Effects of elevated CO2 on nutritional quality of vegetables: A review. Front. Plant Sci. 9, 367384 (2018).
Jablonski, L. M., Wang, X. & Curtis, P. S. Plant reproduction under elevated CO2 conditions: A meta-analysis of reports on 79 crop and wild species. New Phytol. 156, 9–26 (2002).
Crowley, L. M., Sadler, J. P., Pritchard, J. & Hayward, S. A. L. Elevated CO2 impacts on plant-pollinator interactions: A systematic review and free air carbon enrichment field study. Insects 12, 512 (2021).
Peñuelas, J. & Estiarte, M. Can elevated CO2 affect secondary metabolism and ecosystem function?. Trends Ecol. Evol. 13, 20–24 (1998).
Cotrufo, M. F., Ineson, P. & Scott, A. Elevated CO2 reduces the nitrogen concentration of plant tissues. Glob. Change Biol. 4, 43–54 (1998).
Taub, D. R., Miller, B. & Allen, H. Effects of elevated CO2 on the protein concentration of food crops: A meta-analysis. Glob. Change Biol. 14, 565–575 (2008).
Loladze, I. Hidden shift of the ionome of plants exposed to elevated CO2 depletes minerals at the base of human nutrition. eLife https://doi.org/10.7554/eLife.02245 (2014).
Rogers, A., Ainsworth, E. A. & Leakey, A. D. B. Will elevated carbon dioxide concentration amplify the benefits of nitrogen fixation in legumes?. Plant Physiol. 151, 1009–1016 (2009).
Davis, A. R. Influence of elevated CO2 and ultraviolet-B radiation levels on floral nectar production: A nectary-morphological perspective. Plant Syst. Evol. 238, 169–181 (2003).
Rusterholz, H. P. & Erhardt, A. Effects of elevated CO2 on flowering phenology and nectar production of nectar plants important for butterflies of calcareous grasslands. Oecologia 113, 341–349 (1998).
Lake, J. C. & Hughes, L. Nectar production and floral characteristics of Tropaeolum majus L. grown in ambient and elevated carbon dioxide. Ann. Bot. 84, 535–541 (1999).
Ziska, L. H. et al. Rising atmospheric CO2 is reducing the protein concentration of a floral pollen source essential for North American bees. Proc. R. Soc. B Biol. Sci. 283, 20160414 (2016).
Silva, M., Ribeiro, H., Abreu, I., Cruz, A. & Esteves da Silva, J. C. G. Effects of CO2 on Acer negundo pollen fertility, protein content, allergenic properties, and carbohydrates. Environ. Sci. Pollut. Res. Int. 22, 6904–6911 (2015).
Di Pasquale, G. et al. Influence of pollen nutrition on honey bee health: Do pollen quality and diversity matter?. PLoS ONE 8, e72016 (2013).
Frias, B. E. D., Barbosa, C. D. & Lourenço, A. P. Pollen nutrition in honey bees (Apis mellifera): Impact on adult health. Apidologie 47, 15–25 (2016).
Roger, N., Michez, D., Wattiez, R., Sheridan, C. & Vanderplanck, M. Diet effects on bumblebee health. J. Insect Physiol. 96, 128–133 (2017).
Jones, J. & Rader, R. Pollinator nutrition and its role in merging the dual objectives of pollinator health and optimal crop production. Philos. Trans. R. Soc. B https://doi.org/10.1098/rstb.2021.0170 (2022).
Decker, L. E. et al. Elevated atmospheric concentrations of CO2 increase endogenous immune function in a specialist herbivore. J. Anim. Ecol. 90, 628–640 (2021).
Stevenson, P. C. For antagonists and mutualists: The paradox of insect toxic secondary metabolites in nectar and pollen. Phytochem. Rev. 19, 603–614 (2020).
Cook, S. M., Awmack, C. S., Murray, D. A. & Williams, I. H. Are honey bees’ foraging preferences affected by pollen amino acid composition?. Ecol. Entomol. 28, 622–627 (2003).
Kriesell, L., Hilpert, A. & Leonhardt, S. D. Different but the same: Bumblebee species collect pollen of different plant sources but similar amino acid profiles. Apidologie 48, 102–116 (2017).
Rauer, D. et al. Ragweed plants grown under elevated CO2 levels produce pollen which elicit stronger allergic lung inflammation. Allergy 76, 1718–1730 (2021).
El Kelish, A. et al. Ragweed (Ambrosia artemisiifolia) pollen allergenicity: SuperSAGE transcriptomic analysis upon elevated CO2 and drought stress. BMC Plant Biol. 14, 176 (2014).
Samanta, A., Das, G. & Das, S. Roles of flavonoids in plants. Int. J. Pharm. Sci. Technol. 6, 12–35 (2011).
Ainsworth, E. A. & Long, S. P. What have we learned from 15 years of free-air CO2 enrichment (FACE)? A meta-analytic review of the responses of photosynthesis, canopy properties and plant production to rising CO2. New Phytol. 165, 351–372 (2005).
Pereyda-González, J. M. et al. High temperature and elevated CO2 modify phenology and growth in pepper plants. Agronomy 12, 1836 (2022).
Forrest, J. R. K. Plant–pollinator interactions and phenological change: What can we learn about climate impacts from experiments and observations?. Oikos 124, 4–13 (2015).
Keller, I., Fluri, P. & Imdorf, A. Pollen nutrition and colony development in honey bees: Part 1. Bee World 86, 3–10 (2005).
Qaderi, M. M., Martel, A. B. & Strugnell, C. A. Environmental factors regulate plant secondary metabolites. Plants 12, 447 (2023).
Collins, M. et al. Long-term climate change: projections, commitments and irreversibility. In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessement Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) (Cambridge University Press, 2013).
Shpak, Y. V., Leheta, U. V. & Sosnovskyi, K. S. Importance of pollinator diversity for sunflower seed production. Agrology 6, 79–85 (2023).
Larson, J. L., Kesheimer, A. J. & Potter, D. A. Pollinator assemblages on dandelions and white clover in urban and suburban lawns. J. Insect Conserv. 18, 863–873 (2014).
Muñoz, A. A. & Cavieres, L. A. Sharing of pollinators between the invasive Taraxacum officinale and co-flowering natives is not related to floral similarity in the high-andes. Flora 261, 151491 (2019).
Gorenflo, A., Diekötter, T., Kleunen, M. van, Wolters, V. & Jauker, F. Contrasting pollination efficiency and effectiveness among flower visitors of Malva sylvestris, Borago officinalis and Onobrychis viciifolia. J. Pollinat. Ecol. 21, 62-70 (2017).
Stawiarz, E., Wróblewska, A., Masierowska, M. & Sadowska, D. Flowering, forage value, and insect pollination in borage (Borago officinalis L.) cultivated in SE Poland. J. Apic. Sci. 64, 77–89 (2020).
Williams, I. H. & Christian, D. G. Observations on Phacelia tanacetifolia Bentham (Hydrophyllaceae) as a food plant for honey bees and bumble bees. J. Apic. Res. 30, 3–12 (1991).
Giovanetti, M., Malabusini, S., Zugno, M. & Lupi, D. Influence of flowering characteristics, local environment, and daily temperature on the visits paid by Apis mellifera to the exotic crop Phacelia tanacetifolia. Sustainability 14, 10186 (2022).
Hogg, B. N., Bugg, R. L. & Daane, K. M. Attractiveness of common insectary and harvestable floral resources to beneficial insects. Biol. Control 56, 76–84 (2011).
Bosch, J., Retana, J. & Cerdá, X. Flowering phenology, floral traits and pollinator composition in a herbaceous Mediterranean plant community. Oecologia 109, 583–591 (1997).
Simao, M.-C.M., Matthijs, J. & Perfecto, I. Experimental small-scale flower patches increase species density but not abundance of small urban bees. J. Appl. Ecol. 55, 1759–1768 (2018).
Rodrigo Gómez, S., Ornosa, C., Selfa, J., Guara, M. & Polidori, C. Small sweat bees (Hymenoptera: Halictidae) as potential major pollinators of melon (Cucumis melo) in the mediterranean. Entomol. Sci. 19, 55–66 (2016).
Tschoeke, P. H., Oliveira, E. E., Dalcin, M. S., Silveira-Tschoeke, M. C. A. C. & Santos, G. R. Diversity and flower-visiting rates of bee species as potential pollinators of melon (Cucumis meloL.) in the Brazilian cerrado. Sci. Hortic. 186, 207–216 (2015).
Roldán-Serrano, A. S. & Guerra-Sanz, J. M. Reward attractions of zucchini flowers (Cucurbita pepo L.) to bumblebees (Bombus terrestris L.). Eur. J. Hort. Sci. 70, 23–28 (2005).
Rundlöf, M., Lundin, O. & Bommarco, R. Annual flower strips support pollinators and potentially enhance red clover seed yield. Ecol. Evol. 8, 7974–7985 (2018).
Palmer-Jones, T., Forster, I. W. & Clinch, P. G. Observations on the pollination of montgomery red clover (Trifolium pratense L.). N. Z. J. Agric. Res. 9, 738–744 (1966).
Damgaard, C., Simonsen, V. & Osborne, J. L. Prediction of pollen-mediated gene flow between fields of red clover (Trifolium pratense). Environ. Model. Assess. 13, 483–490 (2008).
Campbell, J. W., Irvin, J. H. & Ellis, J. D. Bee contribution to partridge pea (Chamaecrista fasciculata) pollination in florida. Am. Midl. Nat. 179, 86–93 (2018).
Frankie, G. et al. Native and non-native plants attract diverse bees to urban gardens in California. J. Pollinat. Ecol. 25, 16-23 (2019).
Colteaux, B. C., McDonald, C., Kolipinski, M., Cunningham, J. B. & Ghosh, S. A survey of pollinator and plant interactions in meadow and grassland habitats of Marin county California. Bios 84, 1–7 (2013).
Anic, V., Henríquez, C. A., Abades, S. R. & Bustamante, R. O. Number of conspecifics and reproduction in the invasive plant Eschscholzia californica (Papaveraceae): Is there a pollinator-mediated allee effect?. Plant Biol. 17, 720–727 (2015).
Taki, H., Okabe, K., Makino, S., Yamaura, Y. & Sueyoshi, M. Contribution of small insects to pollination of common buckwheat, a distylous crop. Ann. Appl. Biol. 155, 121–129 (2009).
Jacquemart, A. L., Gillet, C. & Cawoy, V. Floral visitors and the importance of honey bee on buckwheat (Fagopyrum esculentum Moench) in central Belgium. J. Hortic. Sci. Biotechnol. 82, 104–108 (2007).
Natsume, K., Hayashi, S. & Miyashita, T. Ants are effective pollinators of common buckwheat Fagopyrum esculentum. Agric. For. Entomol. 24, 446–452 (2022).
Delgado-Carrillo, O. et al. Pollination services to crops of watermelon (Citrullus lanatus) and green tomato (Physalis ixocarpa) in the coastal region of Jalisco Mexico. BioRxiv https://doi.org/10.1101/2024.03.18.585619 (2024).
Jankiewicz, L. S. & Borkowski, J. The development of tomatillo (Physalis ixocarpa brot.) in polish conditions. II. Flowering and fruiting. Acta Agrobot. 43, 11–23 (1990).
Cooley, H. & Vallejo-Marín, M. Buzz-pollinated crops: A global review and meta-analysis of the effects of supplemental bee pollination in tomato. J. Econ. Entomol. 114, 505–519 (2021).
Duenas-Lopez, M. A. Tropaeolum majus (nasturtium). CABI Compend. https://doi.org/10.1079/cabicompendium.54181 (2022).
Goulson, D. et al. Choosing rewarding flowers; perceptual limitations and innate preferences influence decision making in bumblebees and honeybees. Behav. Ecol. Sociobiol. 61, 1523–1529 (2007).
Palmer-Young, E. C. et al. Chemistry of floral rewards: Intra- and interspecific variability of nectar and pollen secondary metabolites across taxa. Ecol. Monogr. 89, e01335 (2019).
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/ (2022).
Hartig, F. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. (2022).
Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest package: Tests in linear mixed effects models. J. Stat. Softw. 82, 1–26 (2017).
Lenth, R. V. emmeans: Estimated Marginal Means, aka Least-Squares Means. (2023).
Muff, S., Nilsen, E. B., O’Hara, R. B. & Nater, C. R. Rewriting results sections in the language of evidence. Trends Ecol. Evol. 37, 203–210 (2022).
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models Using lme4. J. Stat. Softw. 67, 1–48 (2015).
Maia, R. A. et al. Sunflower physiological adjustments to elevated CO2 and temperature do not improve reproductive performance and productivity. Environ. Exp. Bot. 213, 105448 (2023).
Hoover, S. E. R. et al. Warming, CO2, and nitrogen deposition interactively affect a plant-pollinator mutualism. Ecol. Lett. 15, 227–234 (2012).
Ramos-Jiliberto, R., Moisset de Espanés, P., Franco-Cisterna, M., Petanidou, T. & Vázquez, D. P. Phenology determines the robustness of plant–pollinator networks. Sci. Rep. 8, 14873 (2018).
Vaudo, A. D., Dyer, L. A. & Leonard, A. S. Pollen nutrition structures bee and plant community interactions. Proc. Natl. Acad. Sci. 121, e2317228120 (2024).
Kostić, A. Ž, Gercek, Y. C. & Bayram, N. E. Phenolic acids in pollen. In Pollen Chemistry & Biotechnology (eds Ecem Bayram, N. et al.) (Springer, 2023).
Forrest, J. R. K. & Rivest, S. Defence compounds in pollen: Why do they occur and how do they affect the ecology and evolution of bees?. New Phytol. 225, 1053–1064 (2020).
Barberis, M., Calabrese, D., Galloni, M. & Nepi, M. Secondary metabolites in nectar-mediated plant-pollinator relationships. Plants 12, 550 (2023).
Chau, K. D. & Rehan, S. M. Nutritional profiling of common eastern North American pollen species with implications for bee diet and pollinator health. Apidologie 55, 9 (2024).
Alzate-Marin, A. L. et al. Warming and elevated CO2 induces changes in the reproductive dynamics of a tropical plant species. Sci. Total Environ. 768, 144899 (2021).
Goulson, D., Hanley, M. E., Darvill, B., Ellis, J. S. & Knight, M. E. Causes of rarity in bumblebees. Biol. Conserv. 122, 1–8 (2005).
Fowler, J. Specialist bees of the Northeast: Host plants and habitat conservation. Northeast. Nat. 23, 305–320 (2016).
Marsh, J. I., Hu, H., Gill, M., Batley, J. & Edwards, D. Crop breeding for a changing climate: Integrating phenomics and genomics with bioinformatics. Theor. Appl. Genet. 134, 1677–1690 (2021).
Drewnowski, A. & Gomez-Carneros, C. Bitter taste, phytonutrients, and the consumer: A review123. Am. J. Clin. Nutr. 72, 1424–1435 (2000).
Kaiser, N. et al. The role of conventional plant breeding in ensuring safe levels of naturally occurring toxins in food crops. Trends Food Sci. Technol. 100, 51–66 (2020).
Wallace, D. H. et al. Improving efficiency of breeding for higher crop yield. Theor. Appl. Genet. 86, 27–40 (1993).
Canavar, Ö., Ellmer, F. & Chmielewski, F. M. Investigation of yield and yield components of sunflower (Helianthus annuus L.) cultivars in the ecological conditions of berlin (Germany). Helia 33, 117–130 (2010).
Ban, D., Goreta, S. & Borošić, J. Plant spacing and cultivar affect melon growth and yield components. Sci. Hortic. 109, 238–243 (2006).
Van Ploeg, D. & Heuvelink, E. Influence of sub-optimal temperature on tomato growth and yield: A review. J. Hortic. Sci. Biotechnol. 80, 652–659 (2005).
Proulx, R. On the general relationship between plant height and aboveground biomass of vegetation stands in contrasted ecosystems. PLoS ONE 16, e0252080 (2021).
Lakshmi, N. J. et al. Effect of CO2 on growth, seed yield and nitrogen uptake in sunflower. J. Agrometeorol. 19, 195–199 (2017).
Thomas, S. C. & Bazzaz, F. A. Elevated CO2 and leaf shape: Are dandelions getting toothier?. Am. J. Bot. 83, 106–111 (1996).
Aspinwall, M. J., Juenger, T. E., Rymer, P. D., Rodgers, A. & Tissue, D. T. Chapter 6 intraspecific variation in plant responses to atmospheric CO2, temperature, and water availability. In Photosynthesis, Respiration, and Climate Change (eds Becklin, K. M. et al.) (Springer, 2021).
Lenka, N. K. & Lal, R. Soil-related constraints to the carbon dioxide fertilization effect. Crit. Rev. Plant Sci. 31, 342–357 (2012).
Hurlbert, S. H. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54, 187–211 (1984).
Davies, G. M. & Gray, A. Don’t let spurious accusations of pseudoreplication limit our ability to learn from natural experiments (and other messy kinds of ecological monitoring). Ecol. Evol. 5, 5295–5304 (2015).
Oksanen, L. Logic of experiments in ecology: Is pseudoreplication a pseudoissue?. Oikos 94, 27–38 (2001).
Schank, J. C. & Koehnle, T. J. Pseudoreplication is a pseudoproblem. J. Comp. Psychol. 123, 421–433 (2009).
Hurlbert, S. H. The ancient black art and transdisciplinary extent of pseudoreplication. J. Comp. Psychol. 123, 434–443 (2009).
Hurlbert, S. H. On misinterpretations of pseudoreplication and related matters: A reply to oksanen. Oikos 104, 591–597 (2004).
Colegrave, N. & Ruxton, G. D. Using biological insight and pragmatism when thinking about pseudoreplication. Trends Ecol. Evol. 33, 28–35 (2018).
Marshall, D. J. Principles of experimental design for ecology and evolution. Ecol. Lett. 27, e14400 (2024).
Acknowledgements
The traditional custodians of the land on which this research was conducted are the Massachusett people at Harvard University and the Ho-Chunk people at the University of Wisconsin-Madison. This research was funded by a USDA NIFA Postdoctoral Fellowship awarded to OMB (# 2021-67034-35229), a Winslow Foundation grant to JDC, and the Star-Friedman Challenge for Promising Scientific Research to JDC, SSM, NMH, and BdB. We thank the Charles Vidoudez at the Harvard Center for Mass Spectrometry for assistance with secondary chemistry analyses, the Jackson lab, specifically Erin Pierce, for helping us flash analyze our pollen samples, the staff at the Walnut Street Greenhouse on the University of Wisconsin-Madison’s campus, Ben Erdman, Isaac Kabera, and Deena Patterson, Brett Graham for help in developing environmental sensors and CO2 control for experiment 1, and Janet Sherwood, and the Harvard Greenhouse staff, for helping us to successfully grow the plants used in this study. We thank two anonymous reviewers for their useful contributions to this manuscript.
Author information
Authors and Affiliations
Contributions
JDC, AJ, BDB, SSM, and NMH designed and conceived experiment 1. OMB, JDC, and LHZ conceived, and OMB and JDC designed experiment 2. Data was collected by OMB and AJ. Data was processed and analyzed by OMB and JDC. The original draft was written by OMB with subsequent contributions from all authors. Final approval for publication was given by all authors.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Bernauer, O.M., Jain, A., de Bivort, B. et al. Elevated atmospheric CO2 has small, species-specific effects on pollen chemistry and plant growth across flowering plant species. Sci Rep 14, 13760 (2024). https://doi.org/10.1038/s41598-024-63967-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-63967-z
Keywords
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.