Comparison of soybean crop performance under tropical environment between tropical and temperate cultivars with adjusted growth duration

ABSTRACT Considerable variation among soybean genotypes has been observed in growth and yield adaptations to tropical environments. However, the response to daylength and, hence, different growth durations are confounding factors in most studies. In this study, we conducted field experiments in a tropical environment with daylength treatment to compare the crop performance of temperate and tropical cultivars with different daylength sensitivities in terms of biomass production, seed yield, and apparent seed quality. A tropical cultivar, ‘Tanggamus’, which has been highly productive in Indonesia, was grown for two years in a field at Bogor, Indonesia (lat. 6.5° S) with four cultivars from temperate regions. The proposed long-day treatment (LD) decreased the negative effects of short daylength in low-altitude areas on the vegetative growth and yield of soybean cultivars from temperate regions. With LD, the differences in the yield and above ground dry weight between temperate cultivars and ‘Tanggamus’ was less than without LD. However, the ratio of wrinkled seeds of temperate cultivars with LD was higher than that of ‘Tanggamus’. There were differences in the seed-filling rate and the growth rate of pod plus seed during latter half of the seed-filling period between them, indicating a higher ability in the supply of assimilates to seeds under high-temperature conditions during the seed-filling stage. These results suggest that temperate cultivars remain to be improved in biomass and seed production under environments with higher temperatures than the climate where they are cultivated, and that ‘Tanggamus’ will be worth utilizing in breeding programs. GRAPHICAL ABSTRACT


Introduction
The IPCC Sixth Assessment Report states that climaterelated risks to crop production will increase owing to future global warming (IPCC, 2023).This includes soybean production in temperate regions (Iizumi et al., 2017;Rose et al., 2016).In Japan, negative effects on yield and apparent seed quality, such as 100 seed weight, seed coat cracking, and seed shriveling have recently been reported in warm years (Matsuda et al., 2011;Matsunami et al., 2013;Uchikawa et al., 2003).Several field experiments have shown that elevated temperature has negative effects on soybean production and production processes: temperature increases of 1-3°C from ambient one decrease yield (Tacarindua et al., 2013), seed growth (Tacarindua et al., 2012), and gas exchange activity of canopy (Nakano, Tacarindua, et al., 2015).An increase of 3°C in temperature during seed filling and maturity decreases yield, single-seed weight, and nitrogen fixation activity (Mochizuki et al., 2005;Shiraiwa et al., 2006), and temperature above 31°C during this period decreases yield and related traits (Oh-E et al., 2007).Short-duration heat treatment by 6°C, imitating a heat wave, affects photosynthetic activity even after treatment and decreases yield (Siebers et al., 2015).Zhang et al. (2016) reported that a mean temperature increase of 0.7°C decreases photosynthetic rate and yield.Several chamber experiments conducted at well-controlled temperatures have reported various effects.A temperature higher than 32°C for a few weeks irreversibly affects leaf structure and chloroplast (Carrera et al., 2021;Djanaguiraman et al., 2011).High temperatures during blooming decrease pollen activity (Djanaguiraman et al., 2013), and those during seed filling and maturity alter seed composition, for example, increased oil content at around 30°C (Nakagawa et al., 2020;Thomas et al., 2003), reduced protein content (Nakagawa et al., 2020) and reduced isoflavone content (Tsukamoto et al., 1995).Keigley and Mullen (1986) reported that a longer duration of high temperature after seed filling negatively affected seed apparent quality.For stable soybean production in temperate regions (such as Japan) under future elevated temperatures, cultivars adapted for these regions need to be improved to adapt to high-temperature conditions.
There is little information about candidates for adaptive cultivars in temperate regions to high-temperature environments, except for Bellaloui et al. (2017), who evaluated exotic germplasms for high yield and seed quality in high-temperature environments in the Midsouth US.Conversely, considerable variation in growth and yield performance among soybean genotypes has been recognized in tropical environments (Puteh et al., 2013;Saryoko et al., 2017Saryoko et al., , 2018)), suggesting an opportunity for breeding to adapt to hightemperature conditions.Saryoko et al (2017Saryoko et al ( , 2018)).demonstrated higher yield and biomass production and better apparent seed quality in Indonesian cultivars than in cultivars from Japan and the US.Tropical environments are characterized by constant high temperatures (above 25°C), and thus Indonesian cultivars are expected to have high yield potential under hightemperature conditions.A detailed examination of Indonesian cultivars can reveal the adaptation to hightemperature environments in terms of crop performance and its related traits.
However, the comparison of crop performance between cultivars from temperate and tropical regions has certain limitations.One of the major factors in the comparison among cultivars of different maturity groups is the difference in vegetative growth duration and time to maturity derived from photoperiod sensitivity.Daylength in tropical environments, compared to temperate environments, is consistently shorter (approximately 12 h duration).Cultivars from temperate regions are not adapted to such short daylengths, unlike cultivars from low-altitude regions, including Indonesia (Li et al., 1996;Lu et al., 2017;Nagata, 1961), which are characterized by shorter growth durations, poorer canopy development, and lower yield performance under short-day conditions (Guiamet & Nakayama, 1984;Shanmugasundaram, 1979).Correcting this large confounding factor, i.e. different growth durations, is necessary to evaluate adaptation to a hightemperature environment in yield production.
There are several experimental methods for comparing crop performance among cultivars with different maturities.One solution to this issue is to use the maturity-corrected yield index (Voldeng et al., 1997) used in Saryoko et al. (2017), which standardizes the amount of biomass by the growth period or the cumulative radiation energy received by the crop canopy.Voldeng et al. (1997) applied the correction to a relatively narrow range of maturity groups, from MG 0 to MG 000.Here, we attempted to make comparisons with largely different maturity groups, such as cultivars from temperate and tropical regions.In this case, the comparison of crop performance by correcting solely the maturity-corrected yield index is insufficient.In addition, long-and shortgrowing cultivars are exposed to different environments throughout the growing period because of the additional growing period in the longer-growing cultivar.This may influence the comparisons as well.
To equalize a growth period, a short-daylength treatment can be applied to accelerate the progress of PLANT PRODUCTION SCIENCE development in photoperiod-sensitive cultivars (i.e.tropical cultivars) in long-daylength environments.However, the short-daylength treatment requires a strict shading equipment, restricting field-scale experiments.That is one of the reasons why only limited number of short-day experiments have been reported (Dong et al., 2021;Lu et al., 2017).Alternatively, a longdaylength treatment can be applied to slower the progress of development in cultivars with low photoperiodsensitivity (i.e.temperate cultivars) in short-daylength environments.
For the current experiment, we used a long-day treatment (LD) in tropical environments.Artificial lighting in the field can extend days to the blooming stage and the entire growth period (Kantolic & Slafer, 2005;Sameshima et al., 2011;Sinclair, 1993).For instance, Nemoto et al. (2011) conducted LD at the lower-latitude sites of Morioka, Iwate (lat.39.7°N), where it is warmer and has shorter daylengths in summer than Sapporo, Hokkaido (lat.43.0°N) to evaluate the effects of increased temperature on the heading date of rice.LD can be adjusted specifically for the cultivars according to their photosensitivity so that the cultivars have a similar duration of vegetative growth.The light intensity for extending vegetative growth duration by LD is 130 lx (Sinclair, 1993), approximately 1.2 μmol m −2 s −1 as photosynthetic photon flux density (PPFD) calculated by the value in Thimijan and Heins (1983) and Tibbitts et al. (1983), and it is easier to lighten a large part of the plots above this range of values.A carefully designed LD for temperate cultivars in low-latitude regions can be used to evaluate crop performance after adjusting for the confounding effects of growth duration and the adaptation ability of crop performance to high-temperature conditions.
The objective of this study was to compare the soybean crop performance of temperate and tropical cultivars with different daylength sensitivities in a tropical environment in terms of biomass production, seed yield, and apparent seed quality after adjustment for vegetative growth duration using LD and to examine the traits associated with crop performance.We further demonstrated the different abilities of tropical and temperate cultivars to adapt to tropical environments, leading to soybean breeding for adaptation to increasing temperatures.
The seeds were sown on 11 March 2017, and 6 March 2018, with a spatial arrangement of 50 cm between rows and 20 cm within rows.Fertilizer (N: P 2 O 5 : K 2 O = 3: 10: 10 g m −2 ) was added to the rows at sowing.In 2017, each plot of all cultivars had five rows with 8 m rows.In 2018, the number of rows with 8 m was 8 for 'Enrei', 12 for 'DS25-1', 10 for 'Fukuyutaka', and 6 for 'Tanbaguro'.The number of rows with 4 m was 7 for 'Tanggamus'.Two replicates were performed in both years.Water conditions were maintained using natural precipitation and irrigation.The total amount of precipitation recorded at a nearby meteorological station (Bogor Climatology Station, latitude 6.5° S, longitude 106.8°E) during cultivation was 1,317 mm in 2017, and because a large amount of data for 2018 was not recorded, we did not aggregate rainfall for that year.Weeding and spraying of pesticides and insecticides were conducted as necessary during cultivation to maintain optimal conditions, except for the late seed-filling stage in 2017, when a severe stink bug attack on pods and seeds occurred.Because of this limitation, the data from 2017 were subjected to analysis and discussion after excluding the data for seed fertility and singleseed weight.

LD method
In each replicate, four or six 100 W LED lights (LDT-160, GOODGOODS Co., Ltd., Osaka, Japan), whose light color is 6000K, were set 2 m above the ground (Figure 1(a)) and turned on only for the Japanese and US cultivars to extend their growth duration.The angle of the LEDs was adjusted so that the whole plot was lit except at the corners of the plots, where no plants were to be measured and harvested.'Tanggamus' and temperate cultivars grown under natural daylength were protected from the LEDs by placing the corresponding plots at 2 m (in 2017) or 6 m (in 2018) apart from the LD plots and separated by fences with black sheets.The distribution of light intensity at the ground in the plots was measured after seeding in both years.A spectrometer (C-7000, SEKONIC Co., Tokyo, Japan) was placed on the ground, and the intensity was measured as the PPFD every 0.2 m in every planted row (2017) and 1 m (2018) in the dark after sunset.
The LD for temperate cultivars was based on the difference between the target and local daylengths.The target daylength in 2017 was the weekly average in Kyoto, Japan, starting on July 1st, which is the conventional sowing day.In 2018, the constant daylength was estimated using the developmental index model (Nakano, Kumagai, et al., 2015) for the same full bloom stage (R2; Fehr & Caviness, 1977) and beginning seedfilling stage (R5) with 'Tanggamus'.Although no parameters for 'Tanbaguro' and 'DS25-1' are available, the LD setting of 'Tanbaguro' was set to that of 'Fukuyutaka' because of similar maturity stages in Japan, and that of 'DS25-1' was set as the mean value between 'Fukuyutaka' and 'Enrei'.The LD settings were changed every week.The LD continued from sowing until 56 days after sowing (DAS) in 2017, until 52 DAS for 'Fukuyutaka' and 'Tanbaguro', and until harvest for 'Enrei' and 'DS25-1' in 2018.

Measurements
In both years, 12 plants (equal to 1.2 m 2 ) were harvested from each plot at full maturity (R8).The aboveground total dry weight (TDW), seed yield, and harvest index (HI) were measured.The numbers of branches, nodes, flower scars, pods, ovules in pods, and seeds of four representative harvested plants were counted.Dry weight was measured for the aboveground plant parts, stems, pod shells, and seeds, and the TDW was determined.Leaves and petioles attached to plants at R8 were separated and not included in the TDW and HI measurements.Harvested seeds were divided into three classes: full, wrinkled, and damaged, as described by Saryoko et al. (2017).The seeds in each class were counted and weighed, and the moisture content was calculated as the difference between the fresh and dry weights of the full seeds.Yields and 100 seed weight were calculated after conversion to 14% moisture content.The yield capacity (the maximum possible yield when all ovules were assumed to have grown to full seed) was calculated by multiplying the ovule number per meter by the single-seed weight of the full seed as a proxy for potential yield in both years.To evaluate apparent seed quality, the ratio of wrinkled seed weight to total seed weight was evaluated.This ratio was determined after excluding the seeds damaged by insects or diseases.
In 2018, eight plants (equal to 0.8 m 2 ) from each plot were harvested at approximately R1 (beginning of bloom), R5, 10 days after R5 (R5 + 10d), and 20 days after R5 (R5 + 20d, only for temperate cultivars with LD and 'Tanggamus').Four plants were separated into leaves, petioles, stems, pods, and seeds, and each part was dried at 80°C for 48 h to weigh dry matter, whereas the other four plants were subjected to measurement of the whole plant dry weight.The radiation use efficiency until the onset of R5 was calculated, which was determined by the TDW at R5 per cumulative intercepted radiation for the period from emergence to R5.The cumulative intercepted radiation was estimated from the product of the fraction of intercepted radiation (F) and daily solar radiation.To measure F, we used digital images of the plant canopies from above (Purcell, 2000;Shiraiwa et al., 2011).To evaluate seed growth, the seedfilling rate was calculated as the ratio of the actual yield to the yield capacity in 2018.This value reflects the degree of yield realization compared with the capacity calculated from the sink size and biomass supply to the seeds.All sampled plants were harvested only from areas where the plants appeared equal in growth, avoiding the corners and edges of the plots.

Data analysis
To evaluate the effect of LD on vegetative growth and yield production, temperate cultivars under natural daylength and LD were compared.Differences among daylengths, cultivars, and years were tested using analysis of variance (ANOVA).'Tanggamus' under natural daylength and temperate cultivars with LD were compared.Differences between cultivars and years were tested using ANOVA, and the Tukey-Kramer method was used for multiple comparisons.R software (version 3.6.3;R Core Team, 2021) was used for all analyses.The aov function was used for ANOVA, and the 'multcomp' package was used for multiple comparisons.

Weather and daylength conditions
The average temperatures during the growth season at the cultivation site in Bogor were 26.3°C (2017) and 26.4°C (2018).At approximately 70 DAS, the average daily temperature at Bogor was consistently higher than the normal temperature in Kyoto (Figure 2).This implies that temperate cultivars under LD conditions were exposed to temperatures higher than the normal temperatures in Kyoto during the late growing season, whereas conditions during the early half of the growing season were similar to or lower than those in Kyoto.

Effects of LD on growth duration
Figure 3 shows the distribution of light intensity during LD in 2017 and 2018.Almost all plots reached 2 μmol m −2 s −1 as PPFD at the ground.Through LD, the DAS to R1 of temperate cultivars extended from 5 to 27 in 2017 and from 5 to 24 in 2018 (Table 1).DAS to R5 extended from 8 to 36 d in 2017 and from 10 to 28 d in 2018.The difference in DAS to R1 of temperate cultivars with LD from 'Tanggamus' under natural daylength ranged from 18 d before and 10 d after in 2017 and from 12 d before and 9 d after in 2018, and DAS to R5 was from 16 d before and 20 d after in 2017 and from 12 d before and 10 d after in 2018 (Table 1).Through LD, the growth duration of temperate cultivars was closer to or even longer than that of 'Tanggamus' except that of 'Tanbaguro' in 2017, which did not reach maturity when LD was excessive.With LD, temperate cultivars grew well enough to close their canopy as well as that observed for 'Tanggamus' (Figure 1(b)).

Effects of LD on biomass production, yield, and apparent seed quality
Table 2 shows the average values of the 2-year TDW data, traits related to sink size (number of nodes,  scars, and pods), yield capacity, and wrinkled seed rate of temperate cultivars under natural daylength and LD.The TDW at maturity and yield capacity increased under LD (p < 0.01).TDW under LD at R1, R5, and R5 + 10d in 2018 was significantly higher than that under natural daylength (p < 0.01, Table 3).The sink size also increased with LD (p < 0.01).These results indicate that LD increased the biomass production and yield of temperate cultivars.The effect of LD on wrinkled seed rate differed between cultivars and years (Table 2), indicating that LD did not decrease wrinkled seed rates.

Comparison of biomass production and yield between 'Tanggamus' under natural daylength and temperate cultivars with LD
Table 4 shows 2-year data of TDW and yield components.Among the yield components, filling rate and single-seed weight were considerably affected by a stink bug attack in 2017 and were therefore excluded from the analysis.TDW of 'Tanggamus' (475 g m −2 , 2-year average) was similar to or higher than that of temperate cultivars with LD, except that of 'Enrei' with LD (294 g m −2 ).There was a significant difference (p < 0.05) among cultivars in the number of nodes (m −2 ), the number of scars per node, the number of pods per scar, and the number of ovules per pod.Yield capacity differed (p < 0.10) among cultivars, although that of 'Tanggamus' (370 g m −2 in 2017) was higher than that of 'Enrei' (180 g m −2 in 2017) only.
Table 5 shows the seed yield, HI, yield capacity, filling rate, number of seeds per ovule, and single-seed weight in 2018.The yield, HI, yield capacity, and seed number per ovule of 'Tanggamus' were not significantly different from those of temperate cultivars with LD.Single-seed weight of 'Tanggamus' (0.083 g seed −1 ) was significantly lower (p < 0.05) than that of 'Enrei' (0.259 g seed −1 ) and 'Tanbaguro' (0.253 g seed −1 ) with LD.
Figure 4 shows the changes in TDW for 'Tanggamus' and the four temperate cultivars under LD in 2018.The TDW at the R5 stage and radiation use efficiency until the onset of R5 were not significantly different (data not shown).The maximum TDW was in the order 'Fukuyutaka', 'DS25-1', 'Tanbaguro', 'Tanggamus', and 'Enrei'.There was no significant difference in TDW at R8 in 2018, although the reduction in TDW until R8 tended to be less in 'Tanggamus' than in the temperate cultivars.Although the yield capacity of 'Tanggamus' was not different from that of 'Fukuyutaka' and 'Tanbaguro', the filling rate (ratio of actual seed yield to yield capacity) of 'Tanggamus' was significantly higher (p < 0.05) than that of 'Tanbaguro' with LD (Table 5).

Comparison of apparent seed quality and its associated traits between 'Tanggamus' under natural daylength and temperate cultivars with LD
The wrinkled seed rate of 'Tanggamus' (5.4%) was significantly lower than that of the four temperate cultivars with LD (from 54.7% to 99.3%, Table 4).Figure 5 shows the changes in the growth rates of pod and seed in 2018.When the values at different time points were pooled, the growth rate ranged from −0.89 to 14.2 (g m −2 d −1 ) in the temperature cultivars with LD and from 9.8 to 14.1 (g m −2 d −1 ) in 'Tanggamus' throughout the period from R1 to R8.The values before R5 (during the flowering and pod-setting stages) were not significantly different among cultivars.In contrast, from R5 to R5 + 10d, the growth rate of pod plus seed of 'Tanggamus' was significantly higher (p < 0.10) than that of 'Enrei' with LD, and from R5 + 20d to R8, it was higher (p < 0.10) than that of 'Fukuyutaka' with LD.These results showed that 'Tanggamus' maintained a relatively high rate of pod plus seed growth until the late stage of the seed-filling period.

Discussion
The objective of this study was to compare the crop performance of temperate and tropical cultivars with different daylength sensitivities in a tropical environment in terms of biomass production, seed yield, and apparent seed quality after adjustment for vegetative growth duration using LD, and to examine the traits associated with crop performance.The proposed LD decreased the negative effects of short daylength in lowaltitude areas on the vegetative growth and yield of soybean cultivars from temperate regions.After adjustment of growth duration with LD, the differences in yield and TDW between 'Tanggamus' and temperate cultivars were not as large as those described in a previous study (Saryoko et al., 2017).In contrast, the ratio of wrinkled seeds in 'Tanggamus' was lower than that in temperate cultivars with LD.This difference seems to be associated with the efficiency of seed development during the late seed-filling stage, considering that biomass production until R5 was not superior in 'Tanggamus'.The LD proposed in the current research enabled the extension of the growth duration of temperate cultivars  4 The yield capacity of 'Enrei' with LD in 2018 and 'Tanbaguro' with LD in 2017, and the wrinkled seed rate of 'Tanbaguro' with LD in 2018 could not be calculated owing to absence of full seeds and are represented as NA. 5 'Fukuyutaka' with LD in 2017 did not complete their full maturity and its yield was calculated by the mid filling seeds.and allowed realistic comparisons among cultivars of different maturity.In the LD, the light intensity was approximately 2 μmol m −2 s −1 or greater at ground level, except at the corner of each plot (Figure 3).This value approximately meets the conditions established in earlier studies conducted in the field (Kantolic & Slafer, 2005;Sinclair, 1993).Thus, LD in this study provided sufficient light intensity to extend the growth duration uniformly over the entire area of the respective plots.
The potential effects on crop performance of sudden change of daylength like our LD were not fully known and have not been referred (Kantolic & Slafer, 2005;Nico et al., 2016).Physiological response in plants to sudden change of daylength may be fairly rapid (Graf et al., 2010) and thus we assumed that the change of daylength in LD would little influence crop performances.In this experiment, LD was applied before flowering.This is because the growth durations from R1 to R5 are similar between 'Tanggamus' and the temperate cultivars even under natural conditions (Saryoko et al., 2017).Thus, extending the growth duration by LD mitigated the negative effect on the growth of temperate cultivars before R5 under short daylengths and made it possible to evaluate crops with a similar duration of vegetative growth.The growth duration of temperate cultivars with LD was similar to or longer than that of 'Tanggamus' grown under natural daylength in Indonesia.LD before the beginning of flowering extended to R1 and R5; however, the duration between R5 and R8 decreased.A previous study (Saryoko et al., 2017) showed that the duration between R5 and R8 of temperate cultivars was longer than that of tropical cultivars, and this was associated with lower seed growth rates in Japanese cultivars, which is considered one of the factors.In the present study, temperate cultivars with natural daylength showed abnormal maturity.After they set pods and start filling seeds normally, they exhibit symptoms of green stem disorder (Furuya et al., 1988;Hobbs et al., 2006) and stop maturing.In contrast, temperate cultivars in the LD plot proceeded smoothly to maturity.The 2 Wrinkled seed rate is the ratio of weight of wrinkled seeds to total seed weight without damaged seeds.
3 'Tanbaguro' with LD in 2017 did not reach R5 (beginning of seed filling) and no seeds were harvested owing to excess long growth duration by LD. 4 The yield capacity of 'Enrei' with LD in 2018 could not be calculated owing to absence of full seeds and is represented as NA.Different lowercase letters represent significant differences according to Tukey -Kramer's method (p < 0.05).
1 Yield capacity is calculated by ovule number (m −2 ) and single seed weight of full seed.
2 Seed filling rate is the ratio of actual yield to yield capacity, which was calculated using the ovule number (m −2 ) and single seed weight of full seed.
3 Yield capacity and seed filling rate of 'Enrei' could not be calculated owing to absence of full seeds and is represented as NA.
shorter duration after seed filling under LD was thought to be a consequence of the restored normal developmental progress in temperate cultivars.
We distinguished the effects of short daylength and other factors on the adaptation to tropical environments in field-scale experiments.The temperature in the experimental field before R5 was not much different from that in Kyoto, Japan, in the temperate climate (Figure 2), where three Japanese cultivars ('Enrei', 'Fukuyutaka', and 'Tanbaguro') are commercially grown.However, the average temperature during seed filling in this experiment was stable and higher than that in Kyoto (Figure 2).These facts imply that, based on the results of this experiment, we can evaluate the performance of 'Tanggamus' and the temperate cultivars under consistently hightemperature conditions during seed filling with less negative effects of short daylength.Separating the effect of daylength by LD can make low-altitude regions suitable for the evaluation of hightemperature conditions or screening for adaptation to high temperatures.In previous experiments by Saryoko et al (2017Saryoko et al ( , 2018)), the growth stages of tropical and temperate cultivars were too large to compare at the same growth stage at the same time.For example, the DAS to R1 of Japanese cultivars on average was 26.7, and the duration between R1 to R5 was 15.0 compared to 35.3 and 20.0 of Indonesian-modern cultivars, respectively (Saryoko et al., 2017).Growth duration, especially after R1, is largely related to yield formation in soybeans (Dunphy et al., 1979;Kantolic & Slafer, 2005;Nico et al., 2016), although Saryoko et al. (2017) addressed this issue by standardizing TDW, yield, and the number of nodes by the cumulative radiation energy received by the crop canopy during the growth period.LD to temperate cultivars adjusted their growth duration to almost the same or even larger than that of 'Tanggamus'.This means that the disadvantage to biomass production and yield of temperate cultivars derived from short daylength was minimized with LD and that they could be compared under more realistic conditions.
The LD experiment suggested an important trait of 'Tanggamus', a better ability to supply assimilate to the seeds under high temperatures during the seed-filling stage, an ability that may contribute to the high yield and highly stable seed quality.In the present experiment, the growth rates of the pod plus seed were high during the latter half of the seed-filling period of 'Tanggamus' (Figure 5), and the seed-filling rate was higher than that of 'Fukuyutaka' and 'Tanbaguro' with LD (Table 5).High temperatures reduce nodulation and nitrogen fixation (Munévar & Wollum, 1981;Shiraiwa et al., 2006), and high temperatures after the seed growth period (R5.5) accelerate leaf senescence and maturity and decrease seed size and yield (Egli & Wardlaw, 1980).The results of the current study suggest that 'Tanggamus' may have a better ability to supply assimilates and nitrogen to seeds under high temperatures during the seed-filling stage.Unfortunately, the present study has a low power of testing owing to the small number of replications and damage from insects or diseases.Further experiments and physiological investigations are required to better understand these adaptive traits.
The wrinkled seed ratio of 'Tanggamus' was lower than that of Japanese cultivars (50-100%) and 'DS25-1' (30-79%), even after LD (Table 4).The value of Japanese cultivars in this experiment was notably high compared to the ranges reported in Japan, typically 10-20% or lower (Inoue & Takahashi, 2006;Matsunami et al., 2013;Sugimoto et al., 2020).In contrast to previous experiments in a tropical environment (Saryoko et al., 2017), the wrinkled seed ratio of 'DS25-1' by LD was higher than that of 'Tanggamus', even though both are smallsized seed cultivars.High temperatures generally lead to an increase in the number of wrinkled seeds, which is associated with a smaller final seed size (Tacarindua et al., 2012;Thomas et al., 2010).As mentioned above, the temperature during the seed-filling stage was consistently high in the current study, and it is suggested that high-temperature conditions may be a factor for the frequent occurrence of wrinkled seeds in temperate cultivars.There are several factors responsible for the formation of wrinkled seeds, including poor development of the protein body, insufficient carbon supply to seeds (Chen et al., 1998), and a shortage of photosynthates or nitrogen (Inoue & Takahashi, 2006;Nagumo et al., 2010;Sekiguchi et al., 2008).Temperate cultivars with LD showed senescence earlier than 'Tanggamus' (data not shown), which implied that they were in a supply shortage in the late growth stage.High growth rates of the pod plus seed during the latter half of the seed-filling period of 'Tanggamus' (Figure 5) may have contributed to its low wrinkled seed rate under high-temperature conditions.These results suggested that 'Tanggamus' have the traits of more stable seed quality under hightemperature conditions compared to 'DS25-1'.To clarify the adaptative characteristics for seed quality of 'Tanggamus' under high temperatures in detail, other aspects such as seed composition (Thomas et al., 2010) and seed germination and vigor (Egli et al., 2005) should be investigated.
Current commercial cultivars in Japan generally have large seed traits, in contrast to 'Tanggamus'.Kawasaki et al (2016Kawasaki et al ( , 2018)).reported that biomass production during seed filling was lower in Japanese cultivars than in US cultivars, owing to less partitioning of assimilates to pods and seeds, despite their large seed size.Because of this trait, Japanese cultivars seem to be more affected by a decline in assimilation during seed filling; hence, an increase in temperature increases the risk of decreased seed quality for Japanese cultivars.Several studies have reported that cultivars with large seeds are more susceptible to high temperatures than those with small seeds (Choi et al., 2016;Puteh et al., 2013).To retain the large seed trait and appearance quality of Japanese cultivars under high temperatures, the stable seed quality traits of 'Tanggamus' will be worth utilizing in breeding programs.The long-term supply of assimilates to seeds is conferred by slow defoliation during mid-to late-seed-filling stages (Board et al., 1994;Matsuda et al., 2012) or by staygreen traits (Kumudini et al., 2001).Further research focusing on the seed-filling stage is required to confirm its effectiveness.
In the current study, we performed LD experiments in a tropical environment to reduce the negative effect of short daylength on temperate cultivars and compare crop performance between 'Tanggamus' under a natural daylength and temperate cultivars with LD.LD to temperate cultivars resulted in smaller differences in biomass production and yield in temperate cultivars from those in 'Tanggamus' than those reported by Saryoko et al (2017Saryoko et al ( , 2018)).However, our results do not contradict the results of the previous study; rather, they indicated that the difference was related to seed filling.In our experiments, 'Tanggamus' was supposed to have a better ability to supply assimilate to the seed under high temperatures during the seed-filling stage.The LD in lowlatitude regions can be a useful method to compare cultivars of different origins in terms of canopy performance in the field and to screen candidate cultivars with good adaptation to high-temperature environments.

Figure 1 .
Figure 1.Long-day treatment (LD) in the present experiment.(a) Four or six LED lights lit up before sunrise and after sunset as scheduled.These lights covered almost the entire planted area.(b) The appearance of the canopy of 'Fukuyutaka' (within the box area) and 'Tanggamus' for comparison under natural daylength (left) and with LD (right).

Figure 2 .
Figure 2. Daily average temperature at the experimental site in 2018 (black line) and at Kyoto (gray line, average value during 1981-2010 from July 1).

Figure 4 .
Figure 4. Change in aboveground total dry weight (TDW) of 'Tanggamus' under natural daylength and of the temperate cultivars with long-day treatment (LD) in 2018.R5 in this figure shows the timing of R5 (beginning of seed filling) in each cultivar.

Figure 5 .
Figure 5. Growth rate of pod plus seed of 'Tanggamus' under natural daylength and the temperate cultivars with long-day treatment (LD) in 2018.Different lowercase letters represent significant differences according to Tukey-Kramer's method (p < 0.10).

Table 1 .
Days after sowing (DAS) to R1 (beginning of bloom), R5 (beginning of seed filling) and R8 (full maturity) of four temperate cultivars with long-day treatment (LD) and their differences from the values under natural daylength (ND), and from values of 'Tanggamus' under ND.

Table 2 .
Comparison between natural daylength (ND) and long-day treatment (LD) of characteristics at maturity in the temperate cultivars.The values are of average of 2017 and 2018, except for yield.Yield capacity is calculated by ovule number (m −2 ) and single seed weight of full seed. 2Wrinkled seed rate is the ratio of weight of wrinkled seeds to total seed weight without damaged seeds. 3'Tanbaguro' under natural daylength in 2017 and 2018 have no full seed and could not calculate yield capacity and is represented as NA. 1

Table 4 .
Yield performance of 'Tanggamus' under natural daylength and temperate cultivars with long-day treatment (LD) in 2017 and 2018 and the results of ANOVA.

Table 5 .
Yield, harvest index, yield capacity, filling rate, the number of seed per ovule, and single seed weight of 'Tanggamus' under natural daylength and temperate cultivars with long-day treatment (LD) in 2018 and results of ANOVA.