Table 1.
Geographical coordinates, sowing dates, total number of lines planted and number of lines for which anthesis dates were observed for all site-year combinations.
Table 2.
Parameter definitions, ranges, and number of unique values from Sobol sequence generation.
Fig 1.
Parameter estimation using database method.
L1…N are the lines in the data set.
Fig 2.
Simulated and observed anthesis days of all 5,266 lines from 11 site-year combinations.
Fig 3.
The extent of equifinality (number of ties) for 2,254 lines.
(a) 2153 lines with 40 or fewer ties and (b) ties for the remaining 101 lines. The traces (right axes) show the average number of site-years for lines having different numbers of ties.
Fig 4.
a) simulated and b) observed values of anthesis dates for site-years NY6 and NY7. The red line (copied in both panels) separates observations that the model could vs. could not reproduce. NY6 was cooler than NY7 (Table A in S1 File). The symbol color shows the extent of P1 equifinality on a log10 scale.
Fig 5.
Empirical distribution of selected GSP estimates (main diagonal), pairwise scatterplots (upper right) and estimated Pearson correlation coefficients, regression coefficients, and p-values (lower left). Each symbol in the scatter plots represents a pair of GSP estimates from a single maize line.
Fig 6.
Phenotype space plots for simulated and observed anthesis dates for five site-year pairs.
Blue regions outline simulated anthesis date pairs using Sobol database GSP estimates. The data symbols depict expressible (yellow) and inexpressible (red) observed anthesis dates.
Table 3.
Extended range of parameter values used for DE search.
Fig 7.
NY6/NY7 anthesis date results from DE (dark blue) and Sobol database (light blue) searches.
Fig 8.
Scatterplot of P1 vs. P2O NY6/NY7.
Estimates using (a) Sobol database searches and (b) DE.
Fig 9.
NY6/NY7 phenotype space plot in numeric form.
P1 estimates from the database search (black numbers) and the numbers of lines with inexpressible observations (red numbers) arranged in a tableau organized as a phenotype space plot corresponding to the center portion of Fig 8. The inexpressible line at the green arrow would receive a P1 estimate of 254. Horizontal and vertical yellow strips are the anthesis dates for NY6 and NY7, respectively.
Fig 10.
P2O and PHINT scatter plots (top row) and P2O cumulative density functions (bottom row). Parameter estimated by the database method using (a, e) 3 southern site-years, (b, f) 8 northern site-years, and (c, g) all site-years; and (d, h) the same as (a, e) but obtained via DE.
Table 4.
Estimated log likelihood, fit statistics, selected summary measures, and a likelihood ratio test for competing statistical models fitted on GSP estimates with and without the random effect of site-year subset, based on GSP estimates for the base group (N = 60,834).