Ear, tassel, and yield characteristics in different breeding eras
In this study, we investigated the evolutionary trends of maize tassel, ear, and yield characteristics in six various environmental conditions, especially during flowering (15 days bracketing the silking stage; Extended Data Fig. 1). A total of 323 elite maize inbred lines that were released in different eras in China (167 inbred lines) and the United States (156 inbred lines) were collected based on a previous study24. The Chinese inbred lines were classified into three groups based on the date of release and the use in breeding: 29 inbred lines released in the 1960s and 1970s (CN1960&70s), 88 lines released in the 1980s and 1990s (CN1980&90s), and 50 lines released after the year 2000 (CN2000&10s). The American elite inbred lines consisted of 69 public lines (Public-US) and 87 commercial lines (Ex-PVP) with expired Plant Variety Protection Act Certificates; the latter were mainly released after 2003 (Extended Data Fig. 2). Based on the genetic background, these inbred lines are divided into five groups: Stiff Stalk Synthetic (SS), Nonstiff Stalk (NSS), Iodent (IDT), Huangzaosi (specific to China), and Mix group24. In each of the different breeding-era germplasm, the inbred lines have at least three out of five genetic backgrounds to avoid genetic impacts (Supplementary Table 1).
We recorded four grain-related parameters, i.e., seed set, kernel number per ear (KN/ear), and thousand kernel weight (TKW), four tassel-related parameters, i.e., spikelet number per tassel (spikelet/tassel), spikelet opening angle, pollen viability, and pollen shedding duration, and six ear-related parameters, i.e., kernel row number per ear, floret number per row, floret number per ear, emerged silk number per ear, anthesis silking interval (ASI), and anthesis silking overlap (ASO) duration. The recorded values of these parameters at different environments were fit by a linear mixed model to obtain the best linear unbiased estimator (BLUE) values. The mean seed set, kernel/ear, TKW, and grain yield of CN2000&10s lines were all higher than that of CN1980&90s and CN1960&70s lines (Fig. 1). In particular, the seed set of CN2000&10s lines was 22.4% higher than that of CN1980&90s lines and 22.3% higher that of CN1960&70s lines. Likewise, the mean seed set, kernel/ear, TKW, and yield of Ex-PVP lines were all significantly higher, compared to Public-US lines (Fig. 1a).
In maize tassel, CN1960&70s, CN1980&90s, and CN2000&10s lines showed continuously declining trends in the mean spikelet/tassel, spikelet opening angle, and pollen shedding duration. Ex-PVP lines also had significantly lower values in these tassel parameters than Public-US lines. The reductions in spikelet/tassel from early to new inbred lines were larger (37.6%) than in other tassel parameters in both Chinese and US lines. Pollen viability showed no significant difference between inbred lines of different breeding eras (Fig. 1b).
In maize ear, floret row per ear, floret number per row, and especially emerged silk number per ear showed significantly increases from CN1960&70s lines to CN2000&10s lines. Floret row number per ear and floret number per row both had no significant differences between Public-US and Ex-PVP lines, but emerged silk number per ear was 36.8% higher in Ex-PVP than in Public-US lines. ASI and ASO duration were considered as ear traits due to the large effects of silking time on them (Extended Data Fig. 3). ASI significantly decreased from 3.6 days in CN1960&70s lines to 2.4 days in CN2000&10s lines and from 3.8 days in Public-US lines to 3.0 days in Ex-PVP lines. ASO duration increased with breeding eras in both Chinese and US lines (Fig. 1c).
The effects of temperature during flowering on tassel, ear, KN/ear and seed set
Growing environments especially temperature during the 15 days bracketing the silking stage had large effects on maize seed set4, 16(Extended Data Fig. 4). Seed set values of all the lines were negatively correlated with Tmax during flowering, with newly released lines (i.e., CN2000&10s and Ex-PVP lines) having higher correlation coefficients. The decrease in seed set was 9.2% in CN1960&70s lines, 9.8% in CN1980&90s lines, and 12.8% in CN2000&10s lines for each 1°C increase in Tmax; the decrease was 7.5% in Public-US lines and 10.4% in Ex-PVP lines (Fig. 2).
To determine how Tmax affects seed set via tassel and ear, Tmax during flowering was correlated with tassel and ear flowering parameters (Extended Data Fig. 5). Spikelet opening angle, emerged silk number per ear, and ASO duration were all significantly correlated with Tmax. Spikelet opening angle and emerged silk number per ear had higher correlation coefficients. With each 1 °C increase in Tmax, spikelet opening angle on average decreased by 1.2º in CN1960&70s lines, 1.8º in CN1980&90s lines, and 2.3º in CN2000&10s lines; the decrease was 1.3º and 2.0º in Public-US and Ex-PVP lines, respectively (Fig. 3a, b). Similarly, emerged silk number per ear decreased by 16.9, 8.2, and 8.0 silks per ear in CN1960&70s, CN1980&90s, and CN2000&10s lines, respectively, with each 1 °C increase in Tmax; and the decrease was 16.1 and 13.0 silks per ear in Public-US and Ex-PVP lines, respectively (Fig. 3c, d).
Relationship between spikelet/tassel and seed set
Under a warming climate, tassel size is an important concern in maize production. Spikelet number per tassel (spikelet/tassel) as a crucial indicator for evaluating tassel size is a key factor (Extended Data Fig. 6). Hence, our emphasis is directed towards elucidating the optimal spikelet/tassel to guarantee ample pollen shed grains at HT. We found that, as spikelet/tassel increased, the seed set showed a linear increase and then retained at a stable level when spikelet/tassel was above ~700, which is close to that of CN1980&90s lines and Public-US lines (Fig. 4a). Based on this threshold, all the lines were divided into two groups, lines having more than 700 spikelet/tassel (large tassel group) and less than 700 spikelet/tassel (small tassel group). The seed set of large tassel group was lower under favorable environments but became larger under HT stress compared to the small tassel group (Extended Data Fig. 7). To test seed set responses of these two groups to HT stress during flowering, the seed set of large tassel group relative to the small tassel group under different sowing dates was calculated. The relative seed set of large tassel group was negative at Tmax below 31°C, and became positive at Tmax above 32°C; the positive value was much larger at Tmax above 35°C (Fig. 4b). To further verify the above findings, a field experiment including five sowing dates and four treatments for removal of tassel branches (tassel treatment) was carried out. Maize hybrid Zhendan958 which has a large tassel size and is widely planted in China was used. The numbers of spikelet/tassel in four tassel treatments were artificially controlled at ~1400 (control), ~700 (T1), ~550 (T2), and ~300 (T3), respectively (Fig. 4c). Treatments of Control and T1 both maintained high seed set values (~90%) in all the treatments of sowing dates, but comparatively, seed set of T2 and T3 were significantly lower once daily Tmax was above 32°C (Fig. 4d, e, f), consisting with results of relative seed set of large tassel group under different Tmax levels (Fig. 4b).
The importance of the key spikelet/tassel threshold in coping with HT stress under a warming climate
To precisely achieve this spikelet/tassel threshold, we calculated the contribution of central spike length, central spikelet density, branch number, branch length, and branch spikelet density to the seed set. Tassel branch length and central spikelet density contributed more than 60% of the seed set, which should be given more focus when improving tassel HT stress tolerance (Fig. 5a).
Each tassel spikelet contains two florets (upper and lower floret), but generally, only the upper floret can release three anthers from the glumes (Fig. 5b). Each anther on average releases ~2,000 pollen grains certainly with a large variation, and pollen viability is around 90% under natural field conditions33 (Fig. 5e). One viable pollen has the potential to reproduce one seed, but seed set would be limited at pollen densities less than 3000 pollen grains per silk32. Based on the above information, 700 spikelet/tassel can produce enough viable pollen grains for 1,260 silks per ear, which is approximately equivalent to the sum of the emerged silk number of two ears in modern maize hybrids (Fig. 5c, d, e). Under HT conditions during flowering (>35°C), pollen shed number is reduced by ~38% and pollen viability is reduced to ~68% (Supplementary Table 4). In such conditions, 700 spikelet/tassel can provide viable pollen grains for ~590 silks, which is equivalent to the emerged silk number of one modern maize ear (Fig. 5c, d, f). This can be an important reference for the selection of and breeding for HT-tolerant varieties.
Besides, suitable varietal distribution at different regions can be recommended, based on the seed set responses of varieties with different tassel sizes to Tmax during flowering. We know from the results above that the seed set of small-tassel maize varieties starts to decrease at Tmax during flowering above 32°C and the decrease becomes much larger at Tmax above 35°C. Therefore, global maize-growing regions were divided into regions with Tmax during flowering below 32°C, 32-35°C, and above 35°C, respectively, based on the daily temperature in the past decade. In the regions with Tmax during flowering below 32°C, small-tassel varieties are recommended, while in other regions especially at Tmax above 35°C, large-tassel varieties are recommended (Fig. 6a). Globally, approximately 23.7% of maize-growing regions is recommended to grow large-tassel varieties, including the North China Plain, part of the central regions of the United States, almost all the regions of Paraguay, regions of north-central Africa, central region of Asia, and some regions in the Middle East. The ratio intends to increase with the warming climate based on the trends in the past 40 years (Fig. 6 b, c, d).