Effect of row spacing, seeding rate and nitrogen fertilization on yield and yield components of soybean

Soybean is cultivated because of its high content of oil and protein for many uses in the food and feed sector (Zeller, 1999). The protein content is especially higher in soybean than that of the other major grain legumes cultivated in Central Europe (pea, faba bean and lentil) (cf. Vollmann, 2016). Soybean meal is an important protein source in farm animal nutrition with a high digestibility and additional energy content (Florou-Paneri et al., 2014).

Production of grain legumes in the European Union occupies only about three percent of the arable land, producing only about thirty percent of protein crops consumed as animal feed in the European Union. Consequently, due to the high deficit of vegetal protein for livestock, a high share of protein feedstuffs has to be imported, especially soybean meal and soybeans (Glycine max (L.) Merr.), mainly from Brazil, Argentina and the US (Henseler et al., 2013). These massive imports are connected with a number of environmental problems, including the destruction of natural habitats over wide areas, for example, in Brazil where soybean production is expanded on both land converted from pasture and tropical forest (Fearnside, 2001).

Taking into account production and transportation, locally produced and supplied food is recognized as more environmentally friendly, for example, due to the reduction of greenhouse gas emissions compared to imported food, and could therefore help to reduce greenhouse gas emissions and energy input (Blanke, 2008; Michalský and Hooda, 2015). Similarly, the use of regionally produced protein-rich feedstuffs for dairy production in Austria was found to result in lower greenhouse gas emissions than feeding dairy cattle with soybean meal imported from South America, predominantly due to high emission resulting from land use changes (Hörtenhuber et al., 2011). Additionally, there is an increasing interest of consumers in locally grown food for reasons such as taste and concerns regarding food safety or genetically modified food (Halweil, 2002).

All those reasons call for the reinforcement of home grown grain legumes in Central Europe. As the yields of commonly used grain legumes in Central Europe, which are mainly spring crops, are more variable than the non-legume yields in Europe (Cernay et al., 2015), new approaches have to be followed for increasing the grain legume production, especially under the conditions of climate change with higher year-round temperatures and a lower amount of total annual precipitation with a shift of the rainfall pattern to more rainfall in winter and early spring but less in summer (Trnka et al., 2011). Such approaches include winter grain legumes (Neugschwandtner et al., 2015b; 2015c), heat-tolerant new grain legumes such as chickpea (Neugschwandtner et al., 2013; 2014) and the reinforcement of the cultivation of soybean. In a large mega-environment study for Europe, Kurasch et al. (2017) have shown that the adaptive plasticity of soybean might enable its cultivation throughout Europe even at high-latitude locations in northern Europe, concluding that thereby soybean might become an integral part of a sustainable agriculture in Europe.

Soybean is a relatively young crop in the Austrian and European agriculture. First experiments with soybean in Europe were started in Austria in 1875 (Haberlandt, 1878). Further experiments have been conducted since the 1970s for grain use in Austria (Gretzmacher, 1978; Liebhard, 1980) and for fodder use in Germany (Schuster and Jobehdar-Honarnejad, 1977). In the Austrian national crop survey, soybean was first recorded in 1990. In that year, 9,271 ha soybean were cultivated in Austria, whereas 64,467 ha have been cultivated in 2017 (Statistik Austria, 2018). In 2017, Austria was ranking as number 8 among the soybean producing countries in Europe. Compared to the neighboring countries, the cultivation in 2017 was lower in Austria as compared with Italy (322,417 ha), but higher as compared to Slovakia (43,900 ha), Germany (19,000 ha), Czech Republic (15,344 ha), Hungary (7,727 ha) and Switzerland (1,695 ha) (FAOSTAT, 2017). In Austria, a nationwide survey of long-term (2003–2016) grain yield data showed that soybean had the smallest yield gap between conventional and organic production systems among major arable crops with an average relative crop yields of 93.4% in organic compared to the conventional production (mean: 2.57 t ha⁻¹) (Brückler et al., 2017).

Row spacing and seeding rate has been of high interest in soybean research, especially in the United States in the last decades. For example, the row spacing in Iowa was in the 1960s and 1970s ≥ 76 cm (with a traditional 102 cm row spacing according to Taylor, 1980), but a trend towards row spacings < 76 cm occurred since the 1990s (cf. De Bruin and Pedersen, 2008). Minimizing plant population but still achieving optimal yields has become an important topic due to high seed costs of soybean (Board and Kahlon, 2013).

Narrow row spacing generally were reported to increase yield. In a multi-year data survey over three geographic regions in the United States, using data from 4879 soybean fields, narrow (38 cm) rows had overall higher yields than wide (76 cm) rows. In two regions, the differences were larger with shorter vegetation period, whereas in the third region, the differences were low likely as environmental conditions favored canopy closure irrespective of row spacing (Andrade et al., 2019).

Diverging results have been reported for seeding rate: Cox and Cherney (2011) observed similar yields with three seeding rates in the northeastern United States and therefore recommended lower seeding rates. Contrary to this, increases in plant density caused higher yields in Austria (Gretzmacher, 1978) and eastern Germany (Stock et al., 1996). Comparing different seeding rates on three locations in Iowa, maximum grain yield was attained with the highest plant population, but the yield increase was not so high; thus, the yield benefit with the highest seeding rates was overcompensated by increased production costs (De Bruin and Pedersen, 2008). Also, in northern Nigeria, higher fodder and grain yields have been observed with higher than currently recommended seeding rates (Kamara et al., 2014). Reasons given for better yields with narrow rows and higher plant densities are the enhanced early crop growth and the higher leaf area index and light interception (Shibles and Weber, 1966; Envi, 1973; Willcott et al., 1984; Bullock et al., 1998). Further, a decrease of weed infestation and weed biomass production has been shown by both planting in narrower rows (Orlowski et al., 2012) and increasing plant densities (Liebert and Ryan, 2017).

The N demand of soybean is partly fulfilled by biological nitrogen fixation (BNF). In a meta-analysis on BNF of soybean, Salvagiotti et al. (2008) reported for unfertilized soybeans that 58% of total N uptake derived from BNF (with 46 and 74% at the 25–75% percentiles and a maximum of 98%). N fixation contributed to 40–52% of overall N uptake in organically grown soybean in Central Europe (Schweiger et al., 2012).The further demand derives from soil mineral N uptake and eventually from N fertilizer. The effect of fertilization with manure and mineral fertilizer on soybean grown in Iran was shown to be heavily influenced by the weather conditions of the experimental years (Gretzmacher et al., 1994).

Knowledge of soil coverage, yield and yield formation of soybean as affected by row spacing, seeding rate and nitrogen fertilization under Pannonian climate conditions is limited. All yield components of soybean can be more or less influenced by various agronomic measures (Stock et al., 1996). Therefore, grain yield and formation of yield components as affected by the mentioned parameters were assessed to find the appropriate cultivation techniques for enhancing the grain yield of soybean.

The crop stand height of Lenka was the highest in both years, but in 2016, it showed the strongest lodging among varieties resulting in higher crop stands of Naya at harvest. Also, the highest seeding rate resulted in 2016 in lodging and thus a reduced crop stand height. With widest row spacing, highest crop stand was obtained from mid-August up to harvest in 2016. De Bruin and Pedersen (2008) did not observe differences in the plant height between the row spacings of 38 and 76 cm of soybean tested with different seeding rates on three locations in three years in Iowa. Cox and Cherney (2011) observed a higher crop stand with 19 cm than with 76 cm but no effects of seeding rate on crop stand height.

A fast soil coverage by plants is important for radiation interception and soil protection against erosion processes (Klima et al., 2016). Among varieties, Xonia covered the soil faster than the other two. N fertilization did not affect soil coverage but both a higher seeding rate and narrower row spacings resulted in a faster soil coverage. In 2016, the widest row spacing attained just in mid-July, thus two weeks after the other treatments and two months after seeding, a soil coverage of > 90%. Similar to the soil coverage, a higher plant density and narrower row spacing was reported to increase the leaf area index and to reduce the days to reach high solar radiation interception (Shibles and Weber, 1966). Variety effects on soil coverage have already been reported for spring barley and winter faba bean. Further, seeding date, seed size and year can affect soil coverage (Neugschwandtner et al., 2015a; 2015b; 2019a; 2019b). Grain loss during combine harvesting is a major problem in soybean cultivation. This is mainly related to the first pod high. Stock et al. (1996) have reported that high grain losses can just be avoided if the first pod height is over 20 cm. In our experiment, the first pod height differed between years and was higher for Lenka and Xonia than that of Naya in one year. Also, Popovic et al. (2012) reported for soybean that the first pod height differed between varieties. The first pod height was highest in the treatment with the double amount of sown seeds and lowest with the widest row spacing. Similar to our observations, a strong increase of the first pod high with higher plant densities has been shown by Stock et al. (1996). Further seeding date can influence the first pod height as shown for chickpea, where with earlier seeding, the plant height and thereby the FPH increased (López-Bellido et al., 2008).

Xonia had in 2016 a higher grain yield, pod wall yield and stems and leaves yield, and thus, the highest AGDM yield compared to the other varieties, whereas the AGDM and grain yield did not differ between the varieties in 2017. Seeding and fertilization treatments had no effect on ADGM and the grain yield in 2016 but in 2017, the ADGM was highest with narrow row spacing and decreased with increasing row spacings. Grain yield did not differ between treatments in both years. Pod wall yield was highest with wide spacing and lowest with narrow row spacing in 2016 and the other way around in 2017. Yield of stems and leaves was in both years highest with narrow row spacing and lowest with wide row spacing. Thus, no differences in total residue yield occurred in 2016 but in 2017, it was highest with narrow and lowest with wide row spacing. Consequently, the lowest HI with highest seeding rate was a result of the tendency, that with higher seeding rate, the lowest grain yield was obtained (as AGDM was in both years high with the higher seeding rate) and the highest HI in 2017 with wide row spacing was a result of the low AGDM yield (and no differences between grain yields).

Narrower row spacing and higher seeding rate have been reported to increase the yield due to a higher leaf area index and less days to reach high solar radiation interception (Shibles and Weber, 1966; Willcott et al., 1984) and enhanced early growth (Bullock et al., 1998). Contrary to that, we found no difference for grain yield in both years and a lower AGDM with wider row spacing just in one year. Also, contrary to our observations that no interaction of experimental factors occurred, Willcott et al. (1984) reported for the grain yield of soybean a variety × row spacing × seeding density interaction.

The harvest index was the lowest with highest seeding rate in both years and narrower row spacing in one year. With a higher plant density and narrower plant spacing, the HI was reported to decrease as full radiation interception during seed formation is required for maximum seed yield, and conditions resulting in a high seasonal interception like higher plant density may not result in highest seed yield (Shibles and Weber, 1966). Further, the crop growth rate but not the relative growth increases with higher plant densities, resulting in a higher AGDM with increasing seeding rate but lower grain yield and impaired formation of yield components (Envi, 1973).

The grain yield of the varieties did not differ in 2017, but in 2016, the grain yield of Xonia was higher than that of the Lenka and Naya, which was a combination of the general higher plant density (of all varieties) in 2016 than in 2017 and a higher number of pods plant⁻¹ of Xonia in both years. Although grain yield did otherwise not differ, the yield formation through yield components differed between the three varieties: Xonia had the highest number of pods plant⁻¹ and thus also the highest PoD and due to more grains pod⁻¹ also the highest grain density. Lenka had the lowest number of grains pod⁻¹ and also the lowest number of grains plant⁻¹ (as also pods plant⁻¹ and the pod density were lower than that of Xonia but similar to that of Naya ). Naya had in 2017 the highest number of grains pod⁻¹ but the lowest TKW.

Among treatments with 30 sown seeds, the widest row spacing resulted in the lowest plant density, presumably due to higher intra-row competition compared to plants sown in narrower rows. Whereas pods plant⁻¹, TKW and grains plant⁻¹ and grains pod⁻¹ (in 2017) of that treatment were the highest. Consequently, a similar grain density but a lower pod density was achieved. The single plant biomass was the highest in 2017, the SPlY in both years and single pod grain yield in 2017. Similar to Ferreira et al. (2018), who reported a higher production of pods and grains in branches with narrower row spacing, we also observed more pods plant⁻¹ and more grains plant⁻¹ with wider row spacings. Contrary to our observations, Cox and Cherney (2011) reported for three row spacings (19, 38 and 76 cm) no differences in the plant density but a lower AGDM and grain yield with wider row spacing due to a lower pod density, a lower grain density and a lower number of grains plant⁻¹, whereas grains pod⁻¹ increased as also observed in our study with wider row spacing. Anyhow, yield components of individual plants were shown to be more affected by plant spacing within a row than by the row spacing itself (Worku and Astatkie, 2011).

A higher seeding rate at 13 cm row spacing resulted in a higher plant density but a lower grain yield, as the pod density was lower with the higher seeding rate. Also, the single plant biomass, the SPlY and pods plant⁻¹ were considerably reduced, whereas seeding rate had no influence on grains pod⁻¹, TKW and the single pod grain yield. Cox and Cherney (2011) had already reported for four seeding rates a similar grain yield as single plant biomass and side branches plant⁻¹, pods plant⁻¹ and grains plant⁻¹ were lower with the higher plant density, whereas grains pod⁻¹ and TKW were not affected by seeding rate. Contrary to that, Envi (1973) observed a decrease of the TKW of side branches with increasing seeding rate. Ferreira et al. (2018) reported that with increasing seeding rate, the number of pods plant⁻¹ was reduced in a greater magnitude at the branches than at the main stem whereas grain pod⁻¹ was neither altered in the pods located at the stem or at the branches when the plant arrangement changed. Similar to these observations for soybean, also in faba bean, different plant densities tended not to affect the pod density (Patrick and Stoddard, 2010), as competition between plants generally reduced pods plant⁻¹ rather than grains pod⁻¹ or seed size (Adisarwanto and Knight, 1997).

Different morphological responses (e.g., through branching) between varieties to different seeding rates have already been shown by Board and Kahlon (2013). The higher SPB and SPlY of Xonia compared to Lenka and Naya and the higher TKW of Lenka and Naya compared to Xonia with 30 S/76 cm but not with other treatments indicate that there are also different morphological responses of varieties to the altered row spacing. Similar to these observations, Aufhammer and Kübler (1997) also reported for silage maize a decrease of the single plant yield but no effects on dry matter yields per area with increasing plant density.

No yield response was observed due to the application of 5.4 g N m⁻² at flowering. Although with N fertilization, less pods plant⁻¹ and a lower TKW was observed in 2016 as compared to the unfertilized treatment with the same row spacing; no differences occurred for pod density, grain density, grains plant⁻¹ and grains pod⁻¹ in both years. Also, the single plant biomass, the single plant grain yield and the single pod grain yield were similar for both treatments in both years. Soil mineral nitrogen at seeding was in both years very high. So obviously, soybean could fulfill its N demand through the available soil N and N fixation, needing no further mineral N. In experiments performed in Argentina and Nebraska (USA), which had a wide range of seed yields, a yield response to N fertilizer was just observed in environments with a high yield potential, where indigenous N sources were insufficient to fully satisfy the crop N demand (Cafaro La Menza et al., 2017). Further, N application at flowering might not be the best time for N fertilization. In a synthesis-analysis comparing experiments over the United States, N fertilization had a measurable, but small effect on soybean yield. Across all experiments, split N application was more effective than a single N application and N application before the reproductive stage was more effective than N application during the reproductive stage (Mourtzinis et al., 2018).

Correlation analysis showed that grain yield increased with an increase of the AGDM, harvest index and the plant and pod density. TKG and grain pod⁻¹ did not affect the grain yield but their combined effect (SPoY) had a positive effect on it. A higher SPlY resulted from more pods plant⁻¹ and grains pod⁻¹ whereas TKW had no effect; and a higher SPoY resulted from more grains pod⁻¹ and a higher TKW. Similar to our observations, Sudarić and Vratarić (2002) reported a positive correlation of grain yield with pods plant⁻¹, grains plant⁻¹, SPlY, HI but they also reported a positive correlation with TKW. They also found no correlation of pods plant⁻¹ with TKW and HI. Also, Stock et al. (1996) reported a positive correlation of grain yield with pods plant⁻¹ but also contrary to our results, a positive correlation of grain yield with grains pod⁻¹ and TKW and a negative correlation with plant density. TKW was in our experiments just positively correlated with HI, with no other parameters affecting it. Contrary to that, Weatherspoon and Wentz (1934) have found a negative correlation of TKW and pods plant⁻¹. Our correlation results and similarities or differences to other experiments highlight the high phenotypic plasticity of soybean to undergo changes in morphology during adaption to distinct environmental conditions (cf. Ferreira et al., 2018).

Crop management techniques such as choice of variety, seeding rate, row spacing and N fertilization might affect crop stand height and soil coverage. But grain yield of soybean remained stable over a wide range of row spacings and also between different seeding rates and with N fertilization. Soybean is obviously highly adaptable to different conditions through compensation or competition processes during the formation of yield components. Consequently, from an agronomic point of view, cultivation with a low seeding rate, a wider row spacing and no N fertilization can be recommended.