With the population of humans continually increasing, the demands placed upon agriculture to supply enough food will remain a great challenge for the foreseeable future [56]. Increasing crop yields has often been highlighted as a potential solution for meeting the challenge of feeding our growing population [52, 57]. Soybean, as one of the primary sources of edible oil and protein for both humans and animals, has become an increasingly important agricultural commodity crop [50, 51, 55, 58]. In general, crop yields realized under field conditions depend strongly on shoot architecture traits. Soybean is a typical short-day flowering crop, and its shoot architecture is affected considerably by three main types of factors, namely the soybean genetic background, agro-meteorological factors, and soil properties. In spite of the uncontrollable agro-meteorological factors, soybean shoot architecture might be remolded through pyramiding of excellent genetic backgrounds in soybeans onto management efforts aimed at optimizing soil properties for crop productivity. However, information concerning the controllable and uncontrollable factors capable of altering soybean shoot architecture is limited and fragmented. Therefore, further exploration of shoot architecture related factors will likely facilitate soybean breeding efforts aiming to maintain high grain yields under varied environmental conditions.
In contrast to environmental factors, genetic factors can be easily predicted and manually designed through traditional or modern techniques, such as cross-breeding or genetic modification. Furthermore, once genetic factors have been established, further monitoring of markers is unnecessary. Therefore, mining favorable alleles of QTLs conferring development of ideal plant heights became one of the most economic strategies employed to promote crop yield. Over recent decades, many researchers have attempted to identify stable QTLs regulating soybean plant height under varied environments, with a subset of these efforts seeking to clone the underlying genes [59–67]. For example, two well documented loci, Dt1 and Dt2, act as stable regulators of soybean plant height, and each of them play critical roles in nearly all known environments [17, 30]. Dt1 was also identified in the current study, where it was found to be closely linked to the SNP markers near the physical location Chr19.44628812, which displays wide ranging impacts on SH, SNN and AIL in all of the four tested environments (Additional file 1: Table S1). On the other hand, Dt2 was not detected in this study, most likely because the two parents contain the same allele of Dt2.
To date, more than 304 QTLs have been documented in Soybase (https://www.soybase.org), however, many of the reported effects could not be confirmed in different environments, or their additive effects declined considerably in different conditions [13, 59, 63]. This reinforces the point suggested herein that QTLs effects depend on the specific environment conditions present where the soybeans are being grown. Therefore, it is unsurprising that only 2 loci (20 QTLs) out of the identified 19 loci (51 QTLs) were detected across all of the four distinct environments (Additional file 1: Table S1), and that the 51 detected QTLs could not explain a majority of the phenotype variation observed among RILs grown in the 4 diverse environments (Fig. 3a). Unfortunately, these “environmental QTLs”, which might play critical roles under specific environmental conditions, have been typically neglected in previous studies, possible due to more attention being devoted to detecting QTLs that remain stable under varied environmental conditions. Hence, exploring and incorporating environmental factors that can regulate effective QTLs into breeding efforts should facilitate the development of new varieties selected through marker assistant selection (MAS) that are adapted to produce grains in wide ranges of environmental conditions.
In order to facilitate the development of such breeding programs, various ecological environments have been classified and characterized throughout the main soybean producing countries [68–72]. For example, photoperiod and temperature are critical environmental factors that influence soybean shoot architecture development [29, 35, 36, 73–76]. In soybean, the effect of photoperiod on a variety of developmental processes has been well described, and more than 10 genetic loci sensitive to photoperiod changes have been cloned [36–38, 40, 77]. The sensitive alleles of these loci may enhance the duration of the soybean juvenile phase under long-day conditions, which leads to taller plants. Moreover, these photoperiod sensitive alleles have also been shown to play critical roles in the process of domestication and improvement, due of their ability to alter shoot architecture and enhance grain yields [40].
In contrast to the number of genes known to be photoperiod sensitive, temperature effects, though well documented, have not yet been adequately explained, and genetic loci sensitive to temperature remain rare. In this study, in order to explain the effects of temperature on soybean plant height, four temperature factors and three plant height traits were observed along with day-length. Interestingly, AMaT appeared to exert influence over the three tested plant height traits, whereas, AMiT, EAT and AT exhibited relatively small impacts (Fig. 4a). In addition, AMaT also affected AIL more than AD (Fig. 4a), which led to considerable impacts of AMaT on the QTLs of AIL (Fig. 5a). On the other hand, while the vector of SNN in PCA grouped with the vectors of SH and AIL (Fig. 4a), the vector of qSNNs was very distinct from those of qAILs and qSHs (Fig. 5a). These results indicate that variation of SNN across the tested environments is mainly regulated by environment × genotype interactions, but not environment × QTL interactions.
Higher temperatures are known to facilitate soybean node development. For instance, soybean node numbers increased from 18 to 29 and to 40 per plant when the temperature was increased from 30/22 °C to 38/30 °C and to 42/34 °C day/night regimes, respectively [78]. It has also been reported that the number of main stem nodes, plant height and mean internode length of crops increases with increasing temperature [79, 80]. However, no research has yet been conducted to determine the effects of diurnal temperature changes on soybean. For soybean, regions with large diurnal variations in temperature, such as Xinjiang Province in China, typically produce higher soybean yields [81]. In this study, we found that AMiT had a positive impact, and AMaT had a negative impact on enhancing the additive effects of QTLs for SNN. This might help to explain why large fluctuations in diurnal temperature can be beneficial for increasing soybean yield, though further work is needed to reveal the underlying molecular and genetic mechanisms.
According to published data, in 2016, 81.3% of global soybean production was occurring in three countries in North and South America, including the United States, Brazil and Argentina. On the other hand, China is the largest consumer of soybeans, despite the fact that China only accounts for 3.57% of the global soybean production [58]. In China, the major soybean production areas include six disparate regions [82], each with soil properties that are distinct from the other regions. Most soils in the South China region belong to acidic soil types with low pH values and poor nutrient conditions, which is similar to soil conditions in Brazil and Argentina [37, 38, 58, 83]. Soil from the Huanghuai Hai region and the lower-middle reaches of the Yangtze River basin tends to have higher pH values and more available nutrients than South China counterparts, which makes them similar to many soil types found across the USA [58, 84]. Despite these similarities between Chinese soils and soils found elsewhere, and in spite of Chinese farmers applying plentiful and, at times, excessive fertilizers in the field, average soybean yields in China (1.8 t ha− 1) are far lower than the average yields obtained in the USA (3.51 t ha− 1), Argentina (3.02 t ha− 1), or Brazil (2.91 t ha− 1) [58]. These situations imply that neither soil properties nor the amount of fertilizers applied are limiting factors for soybean yield in China. This suggests that fertilizer management, which is typically neglected by Chinese farmers and breeders, might be the critical factor for increasing soybean yields to levels in line with the yields reported from leading soybean producing countries.
Based on the present results, soil pH values appear to exert extensive influence over plant height (Fig. 5b and Table 6), possibly due to the fact that soils with low pH values offer limited bioavailability of N and P. Therefore, on acid soils, fertilizers that can increase soil pH values should be first considered. In contrast, alkaline soils tend to have better nutrient availability conditions, and higher biological nitrogen fixation (BNF) capacities for soybean than their acidic counterparts. Over 70% of the N required for soybean growth can be derived from BNF [85], and excess N fertilizer input not only impairs the BNF capacity for soybean (Yang et al. 2019), but also leads to taller plants (Fig. 4b), which leads to poor lodging resistance. In addition, long-term fertilization with excessive amounts of N causes soil acidification [86–88], which often leads to deteriorating soil conditions. Therefore, in regions harboring alkaline soils, the amount of N fertilizers should be strictly controlled, whereas, fertilizers rich in K and P, which can inhibit increases in soybean plant height (Fig. 4b and Fig. 5b), should be considered for more extensive application.