Citizen science photos and AI reveal how switchgrass adapts across a continent

Millions of Photos Unlock Plant Survival Secrets

Every summer, nature enthusiasts upload wildflower photos to apps like iNaturalist without realizing they're contributing to groundbreaking research. A team led by Professor Jianming Yu at Iowa State University has demonstrated the power of integrating citizen science with experimental data by analyzing nearly 44,000 crowd-sourced images to understand how plants adapt across vast landscapes1

1

2

2

. The study, published in the journal Cell, focused on four common North American perennial grasses: switchgrass, big bluestem, Indiangrass, and little bluestem, which grow from Texas to the Canadian border1

1

.

AI Tool to Analyze Photos Reveals Flowering Patterns in Prairie Grasses

Tracking when plants flower across an entire continent would typically require years of fieldwork. Instead, the researchers built an AI tool called FLORIST to scan a massive online photo archive from GBIF, the Global Biodiversity Information Facility, largely fed by iNaturalist contributors1

1

. FLORIST identified around 5,000 observations of plants caught mid-bloom from the nearly 44,000 photos, each tagged with GPS coordinates and dates. The pattern was immediately clear: across all four species, plants growing farther north flowered earlier in the year, which made intuitive sense given that northern summers are short and winters arrive fast1

1

.

Controlled Experiments Revealed a Puzzling Contradiction

The same team spent two years growing switchgrass at 10 research sites across the Midwest and Gulf regions, with hundreds of genetically distinct plants growing side by side under identical conditions1

1

. In these controlled experiments, the trend flipped completely: plants from northern origins flowered later than plants from southern origins, by about 2.3 days per degree of latitude northward2

2

. "We had to wrap our minds around that," said Yu1

1

. Two datasets about the same plant were telling completely different stories, presenting a scientific puzzle that demanded resolution.

Genetic Basis for Adaptive Responses Explains the Mystery

By analyzing the DNA of hundreds of switchgrass plants from across their native range, researchers identified three key genes that control flowering time: GI, Hd1, and FTL11

1

. These genes act like a biological clock that reads environmental signals and decides when to bloom. The team discovered three haplotypes—combinations of variants of the three underlying genes—with each primarily found in geographic clusters, including a variant specific to the Midwest (H2) and one common in Gulf Coast states (H1)2

2

. The two versions respond to their environment differently: H2 plants are highly sensitive to spring temperatures and flower quickly when April and early May run warm, while H1 plants flower much later regardless of that spring warmth1

1

.

How Plants Adapt: Temperature Drives Phenotypic Plasticity

Analyzing the expected flowering time with a model that included genetic and environmental data showed the temperature from April 25 to May 5 had the strongest correlation with flowering time, speeding up flowering by 3.4 days for each degree Celsius2

2

. However, H1 tended to flower 45 days later in all instances than H2, and H2 flowering was more sensitive to temperature during the critical late April and early May period2

2

. This phenotypic plasticity is not random—it's plant adaptation in action. In the north, flowering early is a survival necessity, and the researchers saw this directly during the severe winter of 2018, when H1 plants at northern garden sites suffered devastating losses while H2 plants came through largely unscathed1

1

. In the south, switchgrass pollen is remarkably fragile in heat, surviving less than 10 minutes at 32 degrees Celsius (89.6 degrees Fahrenheit), so H1 plants delay flowering until late summer after the worst heat passes1

1

.

Large Public Datasets Transform Ecological Research

"With this study, we have connected our quantitative genetic and genomic research with ecology, evolution and adaptation over a large-scale landscape," said Yu, the Pioneer Hi-Bred Distinguished Chair in Maize Breeding and director of the Raymond F. Baker Center for Plant Breeding2

2

. The framework demonstrates that neither dataset was wrong—the citizen science photos captured real plants already equipped with the gene version suited to their location, while the controlled experiments mixed all gene versions across all environments, revealing the underlying genetic baseline rather than what actually plays out in nature1

1

. "Our study highlights the power of combining citizen science observations with designed experiments to uncover mechanisms of adaptation across spatiotemporal scales," wrote the study's 27 authors2

2

. This approach matters because understanding genetic differences and how plants respond to environmental pressures will become increasingly critical as climate patterns shift, potentially informing breeding strategies for crops and conservation efforts for native species across diverse geographies.