Projected climate-driven changes in pollen emission season length and magnitude over the continental United States

The study investigates how climate change will alter the timing (phenology) and magnitude of airborne pollen emissions across the continental United States, with implications for human health. Pollen emissions from wind-pollinated taxa are closely tied to meteorological drivers, especially temperature and precipitation, and contribute to allergic diseases such as rhinitis and asthma. Observations have shown earlier spring pollen seasons and prolonged durations, with late-flowering taxa sometimes showing later starts. However, comprehensive assessments across multiple taxa and regions are limited. This work aims to quantify future changes in pollen season start, end, duration, daily maxima, and annual totals for 13 prevalent taxa under different climate scenarios, and to assess the additional influences of rising CO2 and land cover change.

Prior studies document that warmer temperatures are linked to earlier start dates (3–22 days) for spring-flowering taxa and later starts for some late-flowering taxa (up to 27 days delay), with prolonged seasons observed for multiple genera and families. Precipitation exerts short-term scavenging effects and potential long-term impacts on plant growth. Observation-based analyses are often spatially or temporally limited and typically examine total pollen or a few taxa, leading to uncertainty in continental-scale projections. Some continental-scale studies assess individual taxa or limited drivers, and taxa-specific responses to climate vary. Increases in atmospheric CO2 have been shown in chamber studies to boost pollen production and allergen content, but real-world magnitudes remain uncertain. Land cover and plant community composition are expected to shift with climate and anthropogenic change, potentially affecting pollen emission distributions, but gridded taxa-specific future land cover data are scarce.

The study employs the Pollen Emissions model for Climate Models (PECM1.0/2.0), a prognostic, taxa-specific emission model developed from historical pollen counts (NAB/AAAAI) that simulates 13 prevalent wind-pollinating taxa (Acer, Alnus, Ambrosia, Betula, Cupressaceae, Fraxinus, Poaceae C3 and C4, Morus, Pinaceae, Platanus, Populus, Quercus, Ulmus) over the US at 25 km resolution. Emissions for each taxon E_pol (grains m−2 d−1) are computed as the product of land cover fraction, an annual pollen production factor (pf_annual), a phenology factor (Y_phen) modeled as a Gaussian with mean and width derived from start and end DOY, a precipitation factor (Y_precip), and a CO2 factor (Y_CO2). Phenology (sDOY, eDOY) is linearly related to the previous-year annual average temperature (PYAAT). Annual production factors scale literature-based production (P_annual) by regressions between log annual counts and PYAAT (taxon-specific m_prod, b_prod), normalized to historical temperatures. Precipitation scavenging is included by setting Y_precip=0 on days with precipitation >5 mm d−1. CO2 sensitivity is included as a scenario with Y_CO2=2 (doubling production) for SSP585 end-of-century. Land cover for taxa-based simulations uses BELD v3 for trees and CLM4 satellite-derived land cover for grasses and ragweed (urban/crop proxy). For sensitivity to land cover change, a PFT-based PECM variant uses GCAM-Demeter future land use (2015 vs 2100) across PFTs: deciduous broadleaf (DBL), evergreen needleleaf (ENL), grasses (GRA), ragweed (RAG). Climate inputs (daily temperature and precipitation) are from 15 CMIP6 models, regridded to 25 km. Two periods are analyzed: historical (1995–2014) and end-of-century future (2081–2100), under SSP245 and SSP585. A multi-model average across 15 model-forced PECM runs is used. A Morris method sensitivity analysis (1000 runs per taxon; 9 parameters per taxon) evaluates parameter influence on maximum daily emissions, showing production-related parameters (P_annual, P_norm, m_prod, b_prod) dominate, with phenology parameters also important for taxa with strong duration sensitivity.

- Phenology shifts under warming: Under SSP585 with 4–6 K warming, spring-flowering taxa start 10–40 days earlier; short-day, late-flowering taxa (e.g., Ambrosia, C4 grasses, late Ulmus) start and end later by 5–15 days, lengthening seasons. Season duration increases by 2–19 days for most deciduous/conifer taxa (Category 1), 10–14 days for short-day taxa (Category 3), and remains roughly unchanged for Betula (Category 2). Duration increases scale with scenario (SSP245: 1–10 days; SSP585: 2–19 days). - Daily maximum emissions (Epol,max) with climate-only: Competing effects of longer duration (which flattens peaks), temperature-driven changes in annual production (pf_annual), and precipitation scavenging yield Epol,max changes from –35% to +40% (SSP585). Precipitation increases in spring/winter reduce Epol,max up to 40% in NE and PNW. Regional patterns: DBL Epol,max increase 10–30% in southern regions but decrease up to 20% in NE and up to 40% in Mountain regions; ENL increase 10–40% over most of US; grasses decrease 10–40% nationwide (strongest in north); ragweed increases ~10–20% across CONUS. - Regional taxa interactions: In NE and MT, earlier phenology of dominant DBL taxa converges, but temperature-driven decreases in pf_annual for some taxa (e.g., Populus, Acer) reduce or offset total DBL Epol,max. In PNW, Alnus and Quercus phenologies diverge, mitigating production increases and yielding a modest ~30% increase in total Epol,max. - Annual totals (Epol,ann) with climate-only: Increase 16–40% over the US. Smaller increases in NE (+16%) and PNW (+26%) due to negative temperature correlations for several regional taxa; larger increases (up to 40%) in SE and MT where dominant taxa (Quercus, Cupressaceae) have positive temperature dependence. - CO2 effects: Assuming doubled pollen production (Y_CO2=2) at end-of-century SSP585, Epol,max increases up to 200% and Epol,ann up to 250% across the US, potentially counteracting climate-only peak decreases. Spatially, DBL and GRA increases are greatest in the South; ENL and ragweed increase broadly. - Land cover change (PFT-based sensitivity): Projected PFT shifts (GCAM-Demeter) cause relatively small regional changes compared to climate or CO2. DBL/ENL Epol,max increase up to 6% in SE and CA and decrease up to 7% in MT and PNW; grasses change −18% to +5%; ragweed maxima decrease up to 32% in NE and SE due to cropland reductions. Overall land cover effects are smaller (<10% typical; range −32% to +6% regionally) than climate or CO2 effects. - Health-relevant implication: Longer seasons and higher annual totals imply increased exposure and allergy risk.

The findings demonstrate that climate warming systematically advances or delays pollen phenology depending on taxa-specific temperature responses, lengthening seasons and altering overlaps among taxa. These phenological shifts, together with temperature-driven changes in annual pollen production and precipitation scavenging, shape daily peak exposures and annual totals. Regionally, convergence of spring taxa can amplify overlapping exposure, while divergence (e.g., in the PNW) can dampen peaks despite higher production. Climate-only impacts on daily maxima are moderate (−35% to +40%), but annual totals robustly increase (16–40%). Inclusion of plausible CO2 fertilization effects leads to much larger increases in both peaks and totals, highlighting CO2 as a potentially dominant future driver of pollen burden. Land cover changes at PFT level have comparatively minor effects at regional scales in this framework. Overall, the study addresses the research question by quantifying taxa- and region-specific changes in timing and magnitude, underscoring substantial implications for aeroallergen exposure and public health planning.

This study provides a continental-scale, taxa-resolved projection of pollen phenology and emissions under future climate scenarios. Key contributions include: (1) quantifying earlier spring and later summer/fall phenologies with overall season lengthening; (2) showing climate-only increases in annual totals (16–40%) and moderate changes in daily maxima (−35% to +40%) shaped by taxa-dependent production responses and precipitation; (3) demonstrating that potential CO2 fertilization could substantially elevate both peaks (up to 200%) and totals (up to 250%); and (4) finding that land cover changes likely exert smaller regional effects than climate and CO2 within this modeling framework. Future research should focus on improving parameterizations of pollen production (including interannual variability), expanding observational networks beyond urban sites, refining CO2 response estimates with multi-species, field-based studies, and developing gridded, taxa-specific future land cover datasets to better represent community shifts and their impacts on pollen emissions.

Key limitations include: (1) uncertainties in pollen production parameterizations, derived from limited field studies and sparse pollen observations concentrated at urban sites; (2) limited and chamber-based data on CO2 effects across few taxa, introducing large uncertainty in production scaling; (3) omission of other meteorological drivers (e.g., wind, humidity) beyond temperature and precipitation scavenging due to lack of coupled atmospheric dynamics; (4) lack of gridded, taxa-specific future land cover projections, necessitating a higher-uncertainty PFT-based sensitivity test; and (5) uncertainties in future plant community shifts and species distributions under climate and biotic stresses, which may alter regional taxa composition and emissions.