A reaction norm for flowering time plasticity reveals physiological footprints of maize adaptation

Understanding how plant phenotypes are shaped by their environments is crucial for addressing questions about crop adaptation to new environments. This study focused on analyzing the genetic variability underlying genotype-by-environment interactions and adaptation for flowering time in maize. We developed the use of a physiological reaction norm for flowering time plasticity (PRN-FTP) modeled from multi-environment trial networks, with genotype-specific parameters putatively linked to different regulatory modules for flowering time. We show how genotype-specific differences in developmental responses to temperature fluctuations condition differences in photoperiod perceived among genotypes. This occurs not only across but also within common environments, as the perception of photoperiod is altered by variation in rates of development and durations for becoming sensitized to photoperiod. Using a new metric for envirotyping sensed photoperiods for maize, it was found that, at high latitudes, different genotypes in the same environment can experience up to hours-long differences in photoperiod. This emphasizes the importance of considering genotype-specific differences in the experienced environment when investigating plasticity. Modeling the PRN-FTP for global breeding material of maize showed that tropical and temperate germplasm occupy distinct territories of the trait space for PRN-FTP parameters, supporting that the geographical spread and adaptation of maize was differentially mediated by distinct pathways for flowering time regulation. Our results have implications for understanding crop adaptation and for future crop improvement efforts.

Methods
Datasets (“D”) from multiple projects and experiments were assembled to: (D1) select a temperature response function for thermal time estimation; (D2) estimate variation in the thermal time duration between crop emergence and tassel initiation (Ttem-ti) in order to develop an index for sensed photoperiod (DLs); (D3) envirotype DLs in maize fields across the Northern Hemisphere; (D4) demonstrate methods for modeling a physiological reaction norm for flowering time plasticity (PRN-FTP); and (D5) compare the trait space of parameters estimated from the PRN-FTP in a diversity panel constituting separate breeding pools of maize. The current study focused on days to anthesis (male flowering) as the final characteristic.

(D1) Three temperature response functions (Methods S1, Eq. S1-S3) for measuring thermal time were compared using MET data with 24 hybrids tested in 25 to 26 field environments, spanning latitudes from 30.5371° N to 44.2089° N (File S1; data sourced from the Genomes-To-Fields Initiative). For each environment, the sowing date, field coordinates, and best linear unbiased estimates (BLUEs) for calendar days from sowing to anthesis were obtained from McFarland et al., 2020 and Rogers et al., 2021. Using temperature data, genotype-specific BLUEs for calendar days to anthesis were converted to thermal time.

(D2) A collection of 284 maize inbred lines (panel 1) was used to estimate Ttem-ti for each genotype using a physiological formalism (Eq. 1) based on experimental data for crop emergence, phyllochron, and final leaf number recorded in two experiments with short-day (SD) conditions that (File S2). Crop emergence and phyllochron were measured in a greenhouse phenotyping platform at the LEPSE of INRAE (PhenoArch, Montpellier, FR; Cabrera-Bosquet et al., 2016) under well-watered conditions, while final leaf number was recorded in a field experiment in Puerto Vallarta, MX, also in well-watered conditions. For both experiments, daylengths throughout the growth cycle never exceeded 12.5 h. Additional details for these experiments can be found in Supplemental Information (Methods S2).

(D3) Computational envirotyping of DLs was performed for fields across the Northern Hemisphere (latitudinal range of 13.7578° N to 54.2900° N) where maize was previously grown. For this, the geographical coordinates and planting dates of 671 field environments were assembled from past publications and ongoing projects (File S3).

(D4) To test modeling procedures for the PRN-FTP, anthesis data from 37 field trials were combined into a multi-environment trial (MET) dataset (File S4). This dataset maximized the number of field trials available for a common subset of genotypes also present in dataset D2 (so that genotype-specific values for DLs could be determined), including seven temperate inbred lines (B37, B73, M37W, Mo17, Oh43, LH123Ht, and 2369) and seven tropical inbred lines (CML10, CML258, CML277, CML341, CML373, Tzi8, and Tzi9). Still, representation of these lines across the 37 environments was not fully balanced. The temperate lines were present in a minimum of 15 and a maximum of 22 environments. The tropical lines were present in a minimum of 18 and a maximum of 37 environments.

(D5) To examine the relationship between PRN-FTP parameters associated with flowering time per se and photoperiod sensitivity, MET data for a collection of 111 temperate, 52 tropical, and 51 admixed inbred lines constituting separate breeding pools (panel 2; Flint-Garcia et al., 2005) was assembled from previous experiments (File S5). These lines were observed across a maximum of three SD environments (maximum photoperiod <12.3 h) and 16 long-day (LD) environments (minimum photoperiod of >14.5 h) spanning 18.00° N to 42.76° N latitude.