1 Introduction

On low-lying land in the Ganges Delta which covers around 4.7 Mha, most farms are restricted to a single monsoon rice crop, and land remains fallow in the dry (‘rabi’) season (Fig. 1). Cropping systems intensification is hampered by waterlogging, salinity, poor soil structure, and natural calamities in the rabi season. In addition, the growth of long-duration traditional wet (‘aman’) season rice and its late harvesting delay the establishment of rabi season crop (Paul et al. 2020b). The dilemma is that early rabi crop establishment may be delayed by excess soil moisture after harvest of the previous monsoon rice (aman), which constrains timely soil preparation (Paul et al. 2020a), while delays in sowing expose the crop to soil dryness, salinity, and possibly heat stress later in the season. Currently, in areas where double-cropping is practiced, farmers prepare the soil to grow rabi crops such as sunflower and maize with 4–5 passes of rotary tillage beginning when the topsoil has dried below field capacity (Mondal et al. 2015a). Rotary tillage accelerates the drying of the tilled surface soil but delays the sowing time, which can lead to lower yields.

Fig. 1
figure 1

Alternative land uses in the rabi season in Southern Bangladesh. a Fallow land after wet season rice, b sunflower establishment during dry season, and c increased risk from waterlogging from post-monsoon rain after early rabi sowing

Full size image

Due to its low elevation (< 3 m above sea level), much of the coastal zone of the Ganges Delta is protected by peripheral earthen embankments (polders) to prevent tidal inundation and seawater intrusion (Mondal et al. 2015b; Paul 2020). However, due to a shallow watertable, which is temporally dynamic in both depth and salinity, soil and water salinity tend to increase with the progress of the dry season (Paul et al. 2016). Intensive soil puddling, which is used for transplanted rice establishment, further hampers non-rice crop establishment in the dry season because of adverse soil physical constraints (Mitchell et al. 2013). When the soil is too wet immediately after harvesting wet season rice, poor drainage and aeration can negatively affect the emergence of seedlings (Rahmianna et al. 2000). Also, post-monsoon rain may expose early sown crops to waterlogging (Yu et al. 2019). On the other hand, if the time of sowing is delayed, crop growth and development can be affected by late-season soil dryness and compaction (Paul et al. 2020a). In addition to this, late sowings may be damaged by high soil and water salinity and waterlogging from pre-monsoon rains prior to harvest (Zeng et al. 2015). Rather than wait until the soil is dry enough for rotary tillage, early planting at high soil water content (SWC) may be achieved by zero-tilled dibbling (Mondal et al. 2015a).

Sunflower is considered to be a drought-tolerant crop because of its efficient use of soil water. It develops a deep root system (up to 2 m) compared with other species such as sorghum, soybean, and maize, which aids in water uptake by the plant during water stress (Sadras et al. 1989; Seiler 1994). Sunflower is also classed as a moderately salt-tolerant crop (Katerji et al. 2003). In addition, low temperatures (< 25 °C) may delay and reduce germination and emergence (Gay et al. 1991), while high temperatures (> 30 °C) during bud formation and flowering can depress seed number, grain size, and yield (Rondanini et al. 2006).

There is already some information on the effects of early sowing on sunflower yields in Bangladesh. In the coastal saline area, sunflower sown in the first week of January produced larger heads and had greater yields than those sown on 22 February (Rashid et al. 2014). Mondal et al. (2015a) showed that the early establishment of rabi crops such as sunflower after wet season rice improved cropping system productivity. A recent study in this area has demonstrated that rice straw mulch was effective in storing soil water and reducing salinity (Paul et al. 2020b). While these earlier studies emphasized the importance of early sowing, they did not identify the main yield-limiting factors associated with early or late sowing or the climatic risks, nor did they consider the opportunity presented to establish crops early if sown without tillage by dibbling. The present study (i) evaluated the effect of sowing dates on sunflower yield and yield components over two seasons, (ii) examined the effects of mulch on sunflower established by dibbling, and (iii) examined the relative impacts on yield from variations in the prevailing soil and weather drivers: soil water content, EC1:5, the solute potential of the soil solutions, and average seasonal temperature during crop growth.

2 Materials and methods

2.1 Description of the study site

Two field experiments were conducted in a farmer’s field at Pankhali Village, Dacope Upazila, Khulna District, Bangladesh (22°37′55″ N latitude and 89°30′10″ E longitude), where the land is low-lying, and wet season rice (aman) is generally grown from July to January. The area belongs to the Ganges Tidal Floodplain agro-ecological zone, AEZ13 (Paul et al. 2020a). The climate is sub-tropical with an average annual rainfall of 1800 mm, a cool dry winter from November to February, and a hot and humid summer from March to June. Initial soil samples were collected at 15 cm increments to a depth of 60 cm (before crop establishment) to measure soil physical and chemical properties (Paul et al. 2020a). The soil texture was silty clay (0–30 cm) and overlying clay (30–60 cm). The soil (0–15 cm) had a bulk density of 1.4 g cm−3, a pH of 7.5, an EC1:5 of 0.45 dS m−1, an organic carbon content of 12 g kg−1, a total nitrogen content of 1.3 g kg−1, an available phosphorus of 3 mg kg−1, and an extractable potassium concentration of 338 mg kg−1. To use the land for sunflower establishment in November, a medium duration high-yielding aman rice was transplanted in the first week of August and the field drained of standing water (using open ditches: 15–20 cm deep and 20–25 cm wide) before the harvest in the third week of November.

2.2 Experimental design and crop management

The experiment was laid out in a randomized complete block design with three replications in plots 10 × 4 m in 2016–2017 and 9 × 5 m in 2017–2018. Experimental treatments comprised five sowing dates in 2016–2017 (23 November, 30 November, 10 December, 20 December, and 30 December) and a further five dates in 2017–2018 (25 November, 14 December, 25 December, 10 January, and 25 January). In each year, there were two mulching treatments (RS = applied rice straw after crop emergence at 5 t ha−1 and RR = retained 15–20% of rice residue, which is approximately 1 t ha−1). Sunflower seeds (cv. Hysun-33) were sown into wet soil (soil water content varied from 43 to 50%, w/w in 2016–2017 and from 31 to 51%, w/w in 2017–2018) by dibbling with a round stick to a depth of 2–3 cm with two seeds per hole except for the fifth sowing in 2017–2018, where rotary tillage was done (five passes to achieve a suitable soil tilth) by two-wheel tractor before dibbling as soil surface dried. The row to row distance was 70 cm, and plant to plant distance was 40 cm. Plants were thinned to one per hill at the 4–6 leaf stage (20 days after emergence). Fertilizer (urea-triple superphosphate-muriate of potash-gypsum-zinc sulphate-boric acid) was applied at 200–200–170–170–10–12 kg ha−1. All fertilizers except 75% of the urea were applied immediately after crop emergence by placing in holes (5–7 cm deep) on both sides of the plant (~ 5 cm distance) along the rows, and the rest of the urea was broadcast between the rows in three splits at seedling stage (30–40 days after sowing [DAS]), maximum vegetative period (50–55 DAS), and before flowering (60–65 DAS) in both seasons. Light irrigation was applied by spraying the soil immediately after urea top dressing to wash the urea into the soil. Insecticide Nitro (Cypermethrin Chlorpyriphos) was applied 2–3 times to control hairy caterpillar. In each plot, 7 m2 was selected to monitor the emergence rate, crop growth, and development, bud initiation, first flowering, and physiological maturity. In 2016–2017, the crop was harvested on 15 March, 17–20 March, 29 March to 2 April, 5–8 April, and 11–14 April 2017 for the first, second, third, fourth, and fifth sowing, respectively. In 2017–2018, the crop was harvested on 19–23 March, 7–9 April, 10–13 April, 23–26 April, and 30 April to 3 May 2018 for the first, second, third, fourth, and fifth sowing, respectively. Ten plants were randomly selected to measure plant height, head diameter, head weight, number of seeds per head, and 1000 seed weight at maturity. Seeds were threshed manually from heads and sun-dried for 2–3 days to calculate the seed yield (t ha−1) at an adjusted moisture content of 9% (w/w).

2.3 Soil water content measurement

Soil samples were collected to measure gravimetric SWC at 0–7, 7–15, and 15–30 cm depth every 10–15 days between sowing and harvest in both seasons. A hand-held auger was used to collect soil samples from each depth. The weight of the samples was measured immediately, and samples were then oven-dried to constant weight. The gravimetric SWC was then calculated (Cresswell and Hamilton 2002). The SWC at field capacity (− 33 kPa) and wilting point (− 1500 kPa) was also measured using a pressure plate apparatus (Eijkelkamp pF set) on undisturbed soil cores collected before sowing at each of the depths.

2.4 Soil electrical conductivity (EC1:5) measurements and calculation of soil solute potential

The EC1:5 of the soil was measured at 0–7, 7–15, and 15–30 cm depths at 15–20-day intervals during the growing season. Measurements of electrical conductivity were made in a 1:5 soil:water suspension using a portable EC meter. The solute potential (SP) of the soil solutions was calculated (Paul et al. 2020a) from the equation:

$$ {\Psi}_{\mathrm{s}}=\frac{-22580\times \mathrm{EC}1:5}{W} $$

where Ψs is the solute potential (kPa), EC1:5 is the electrical conductivity (dS m−1) of the 1:5 soil:water extract, and W is the gravimetric SWC (%, w/w).

2.5 Statistical analysis

Statistical analyses were done using STAR software version 2.0.1. The effects of sowing dates and mulch on grain yield and yield components were determined using a two-way analysis of variance (ANOVA). The significance of effects of sowing dates and mulch on SWC, EC1:5, and SP were determined using a three-way factorial ANOVA model that also took account of the effects of sowing dates and mulch with soil depth or DAS as repeated measures. The differences between means were tested using the least significant difference (LSD) at the 95% confidence level. The relationship between yield and air temperature was tested by single factor regression analysis. The relative contribution (weighting) of the prevalent physical drivers for crop performance was examined by performing multiple linear regressions of the observed data, using several different factors. Factors selected for their explanation of yield variation were (i) a 2-factor (average seasonal temperature and average seasonal soil SP in the root zone) regression analysis and (ii) a 3-factor (average seasonal temperature, SWC, and soil EC1:5) regression analysis. We considered that the magnitude of the coefficient assigned to each of the factors represented the relative influence that variable had in determining grain yield.

3 Results and discussion

3.1 Yield and yield components of sunflower

In 2016–2017, early sowing on 23 and 30 November produced the highest yield (~ 4 t ha−1), later sowings had lower yields, and the last sowing on 30 December had the lowest yield (2.9 t ha−1) (Table 1). By contrast, in the 2017–2018 season, the crop sown on 25 November had a lower yield than crops sown on 14 and 25 December as the former sowing was affected by waterlogging (Table 2). The highest yield (2.9 t ha−1) was from crop sown on 14 December, and the lowest yield was from crop sown on 25 January (1.3 t ha−1) (Table 2). Overall, there were lower yields in the second season compared to the first season. In the first season, mulch did not affect yield, but in the second season, the RS treatment had a significantly higher yield than the RR treatment (Tables 1 and 2).

Table 1 Yield and yield components of sunflower at Pankhali, Dacope, Khulna, Bangladesh in 2016–2017
Full size table
Table 2 Yield and yield components of sunflower at Pankhali, Dacope, Khulna Bangladesh in 2017–2018
Full size table

Early sowing in both seasons was associated with a larger head diameter. The head diameter showed a decreasing trend with delayed sowing. In both seasons, the RS treatment increased the head diameter by 3–4% relative to RR treatment (Tables 1 and 2). The number of seeds per head decreased with delayed sowing in both years. Mulch did not affect the number of seeds in the first season, but there was a significant effect on the number of seeds in the second season. In 2016–2017, early sowing (23 November, 30 November, and 10 December) had significantly greater seed weight (90–97 g) than later sowing on 20 and 30 December (71–85 g). In 2017–2018, sowing on 14 December produced the highest seed weight compared to the other sowing dates. There were significant interactions between sowing dates and mulch on seed weight in the second season (Table 2). In both years, the differences of yield were correlated with head diameter and thousand seed weight, but the greater correlation was found with seeds per head.

The time of sowing had a marked effect on sunflower yield, but the responses differed between years.

In the first year, sunflower yields were highest (~ 4 t ha−1) with the earliest sowing (23 and 30 November), but in the second year, due to the waterlogging by post monsoon rainfall, the earliest sowing on 25 November decreased yields by 30%, while with the second sowing (14 December) that occurred after the plots had been drained yield was 2.6–3.1 t ha−1. Hence, provided there is not heavy rainfall, early sowing enhances sunflower yield, but it is risky, especially in the absence of rapid and adequate drainage. In the Ganges Delta (over the last 40 years), rainfall data show an increasing trend of intense rainfall (Bell et al. 2019). According to Yu et al. (2019), in the coastal zone at Khulna, there can be > 20 mm rainfall events in 50% of years and > 50 mm in 25% of years in November, and there is a chance of heavy rainfall > 40 mm day−1 from December to February which can hamper early sowing and cause waterlogging. Surface drainage (open ditches: 15–20 cm deep and 20–25 cm wide) installed the day after the heavy rainfall event on 10 December was effective as a means of averting the failure of early sown crops from waterlogging. A more comprehensive assessment of the frequency of yield-damaging waterlogging events could be obtained from well-calibrated crop simulation models such as APSIM (Gaydon et al. 2017). In particular, it is critical to predict the optimum window for early sowing in this environment to identify over the long term when the risk of waterlogging becomes acceptably low. Even though sunflower yield was higher in the first year compared to the second year, the maximum yield (2.9–4 t ha−1) in both years was from planting between 20 November and 15 December (Fig. 2).

Fig. 2
figure 2

The mean sunflower seed yield associated with five dates of sowing over two seasons (2016–2017 and 2017–2018), also showing the major constraints to grain yield with time. Error bars indicate standard error of the mean. The vertical dotted line indicates sowing before 15 December has an average yield of 2.9–4 t ha−1

Full size image

3.2 Seasonal rainfall and temperature, and sunflower growth

In 2016–2017, total rainfall throughout the cropping season (November to 15 April) was 173 mm, but two-thirds fell in March during the flowering stage (Fig. 3a). By contrast, in 2017–2018, 51 mm rain fell during 8–10 December, inundating plots previously sown on 25 November and causing waterlogging stress (Fig. 3f). This heavy rain also delayed the second sowing until 14 December. There was no rain in the following months until 96 mm fell in April (Fig. 3f). The results of long-term data analysis (1970–2017) indicate that the chances of sudden heavy rainfall have increased during the rabi season in recent years (Yu et al. 2019). This study also showed that from 2008 to 2017, the 10-day accumulated rainfall was around 50 mm in December and 100 mm in January and February. In both years, air temperature declined in December/January and increased again from February (Fig. 3b, g). The average winter temperature (December–February) in 2017–2018 was warmer than in 2016–2017. The lowest minimum temperature was 8 °C in January 2018 and 9.5 °C in January 2017. The monthly maximum temperature was similar in both seasons apart from higher values in March and April in 2017–2018. The maximum temperature in December and January was less than 30 °C in both seasons. Temperatures began to increase from February, with the maximum temperatures being 36 and 37 °C in March and April (2018), respectively.

Fig. 3
figure 3

Variation of seasonal weather and average soil data at 0–30 cm depth: a, f daily rainfall, b, g daily temperature, c, h soil water content, d, i electrical conductivity in a 1:5 (EC1:5), and e, j solute potential of soil solution. Parts a–e are for 2016–2017, and parts f–j are for 2017–2018. Nov = November; Dec = December; Jan = January. Lines are fitted with polynomial cubic relationships for c r2 = 0.84, d r2 = 0.65, e r2 = 0.81, h r2 = 0.91, i r2 = 0.49, and j r2 = 0.75 at P < 0.001

Full size image

In both temperature and sub-tropical regions, delayed planting of sunflower after the normal sowing time caused a yield reduction (Barros et al. 2004; De La Vega and Hall 2002; Unger 1980), which has been mainly attributed to the warmer temperature at reproduction and grain filling stages. The temperature for optimum growth and development of sunflower was between 26 and 29 °C (Rondanini et al. 2006). In the present study, early sown crops (between 20 November and 15 December) in both years were exposed to temperature < 30 °C during flowering, but later sowing after 15 December exposed the crop to > 30 °C during flowering and grain filling stages. Thus, the effect of high temperature in later sowings shortened the reproductive period and induced physiological maturity 5–9 days earlier in the first year and 15–20 days earlier in the second year, respectively, which may be related to lower yield under late sowing (data not presented). Some studies also reported that in subtropical conditions, late-sown crops encounter heat stress during the reproductive stage, decreasing seed numbers, grain unit weight, and grain quality (Chimenti et al. 2001; Rondanini et al. 2003).

3.3 Relative influence (weighting) of temperature, EC1:5, SWC, and SP in determining grain yield

In the salt-affected Ganges Delta, the differences in the final yield of sunflower may be attributed to the establishment time and method, average seasonal air temperature, in-season rainfall, and conditions faced later in the growing season such as low soil water, rising soil salinity, and increasing crop water demand. The early sowing may be affected by waterlogging, whereas the later sowing can be negatively impacted by higher temperature, salinity, and soil water depletion (Fig. 2).

In both years, grain yield was negatively correlated (single factor relationship) with increasing minimum and maximum temperature averaged over the season (Fig. 4). In the first year, minimum and maximum temperature accounted for 79 and 78% of the crop yield variation, whereas in the second year, these accounted for 69 and 68% of the variation, respectively. The mean percentage reduction in yield for increased average minimum and maximum temperature from 15 and 28 °C at sowing to 18 and 30 °C at harvesting was 27% in the first year. In the second year, the percentage yield decrease was 54% for increased temperature from 17 and 29 °C at sowing to 22 and 33 °C at harvest (Fig. 4). We also explored the relative weighting of other driving factors affecting the variation of yield in both seasons by testing multiple linear regression analyses considering 3-factor (average temperature, SWC, and EC1:5) and 2-factor (average temperature and SP) relationships (Figs. 5 and 6). In the first season, adding the SP factor to the 2-factor analysis had minimal effect on explaining yield variation (Fig. 5c, d). By contrast, the 3-factor analysis (average temperature, SWC, and EC1:5) explained 91% of the yield variation with the major contribution from EC1:5 and a smaller contribution from temperature (Fig. 5a, b). Moreover, seed yield was mostly attributable to variation in EC1:5 during the later half of the growing season. In the second season, the 3-factor analysis (average temperature, SWC, and EC1:5) explained 72% of observed yield (Fig. 6a); the 2-factor analysis (average temperature and SP) explained only 53% of yield variation, and as in the previous year, average temperature had the dominant effect (Fig. 6c, d). While the driving factor for determining the yield in the second season was EC1:5 and to a lesser extent temperature, as in the first year, these factors had most influence on sunflower yield in the first half of the season. The 3-factor analysis explained more of the variance in yield (higher r2 values) than the 2-factor analysis in both seasons.

Fig. 4
figure 4

Relationship between sunflower yield and temperature during the growing season a average minimum temperature, b average maximum temperature in 2016–2017, c average minimum temperature and d average maximum temperature 2017–2018. For a, y = − 0.3418x + 9.1553, r2 = 0.79; for b, y = − 0.4563x + 16.665, r2 = 0.78. For c, y = − 0.3542x + 9.11103, r2 = 0.69; for d, y = − 0.4464x + 16.111, r2 = 0.68 at P < 0.001

Full size image
Fig. 5
figure 5

Relationship between observed and predicted yield and weighting of factors a–b for 3-factor and c–d for 2-factor analyses in 2016–2017 season (Avg = average; SWC = soil water content; EC1:5 = electrical conductivity of the 1:5 soil:water solution; SP = solute potential). For a, y = 0.9118x + 0.307, r2 = 0.91; for c, y = 0.8058x + 0.6757, r2 = 0.81 at P < 0.001

Full size image
Fig. 6
figure 6

Relationship between observed and predicted yield and weighting of factors a–b for 3-factor and c–d for 2-factor analyses in 2017–2018 season (Avg = average; SWC = soil water content; EC1:5 = electrical conductivity of the 1:5 soil:water solution; SP = solute potential). For a, y = 0.7175x + 0.623, r2 = 0.72; for c, y = 0.5264x + 1.0445, r2 = 0.53 at P < 0.001

Full size image

Previous studies have pointed out that SP better explained effects of tillage and mulch on grain yield (Paul et al. 2020a, b); however, in the present study, EC1:5 explained the highest yield variation (r2 = 0.91 and 0.72) in both years using a multivariate regression analysis that also included temperature and SWC. Moreover, in a 2-factor analysis using temperature and SP, temperature was the main driver of yield, and SP had minimal effect. The apparent lack of influence of SP on yield relative to other factors across the multiple times of sowing may be because both SWC and EC had contrasting impacts on yield. Higher SWC may have a positive effect on yield under drying conditions later in the growing season, while more water can negatively affect yield due to waterlogging in the early part of the season. Both of these effects were evident in the 2 years of the experiments. Therefore, combining SWC and EC1:5 into the single variable of SP did not improve the predictability of the model which involved multiple times of sowing. Based on the best fit of the line in the 3-factor analysis (high r2 values), EC1:5 was the strongest driver of grain yield followed by temperature. Since the 2 years of experimentation with sunflower were quite different, a long-term assessment of the optimum sowing window for sunflower and the major drivers of yield is needed. Crop simulation models such as APSIM, calibrated to account for waterlogging, salinity, solute potential, and heat stress would provide a clearer insight into the long-term variability in response of sunflower yield to sowing date in the Ganges Delta.

3.4 Seasonal dynamics of SWC, EC1:5, and SP

For both seasons, the effects on SWC, EC1:5, and SP for five sowing dates are reported in the Supplementary Materials (Section S2 and Figs. S3–5). All these soil variables changed as the dry season progressed and were affected by seasonal rainfall. In both seasons, the observed data for SWC, EC1:5, and SP over the 0–30 cm soil depth for all dates of sowing were best fitted with a polynomial cubic relationship (Fig. 3c, d, e, h, i, and j). For all dates of sowing, time accounted for 84 and 91% (P < 0.001) of the decrease in SWC during the first and second seasons, respectively (Fig. 3c, h). Irrespective of sowing treatments, the SWC was around 42–46% (w/w) at the beginning of the season (last week of November) and declined to 17–22% by the end of the season. In the second season, the SWC at the last sowing (25 January) was slightly higher because the land was cultivated by a rotary cultivator before sowing by dibbling.

Early sowing of sunflower was achieved by dibbling even though SWC exceeded field capacity (40–50%, w/w). However, the higher SWC was apparently not limiting for emergence as there was no inhibition of emergence percentage (emergence: 95–98%) relative to the second and third sowing with lower SWC. Emergence after dibbling was drastically reduced at the fourth sowing because of decreased SWC (33%), but emergence was satisfactory in the last sowing even though SWC was 31%, as the land was cultivated with five passes in the second season (Supplementary Materials Section S1; Figs. S1 and S2). The reason for lower emergence in fourth no-till dibbled sowing may be the detrimental effects of surface drying, soil hardening, surface crusting, and high soil strength. The hardened soils may have reduced the availability of soil water surrounding the seed zone (Mitchell et al. 2013). In the first season, SWC during the early sowing (between 23 November and 15 December) was apparently higher than the wilting point (26%, w/w) from sowing to harvest, but SWC in the last two sowings (20 and 30 December) dropped to 3 and 5% below wilting point in the grain filling stage. In 2017–2018, SWC was much lower than the wilting point from maximum vegetative stage to maturity (4–10% < wilting point) in the later sowings (between 25 December and 25 January). The higher soil water availability during the early sowing treatments may contribute to better plant growth and development and higher yield compared to the late sowings. Although sunflower is considered to be a moderately drought-tolerant crop, water stress in early flowering to grain filling can cause up to 50% yield loss (Aboudrare et al. 2006; Alahdadi et al. 2011). However, in this study, in the 3-factor (temperature, SWC, and EC) multivariate regression analysis, SWC did not have a significant impact on the yield variation (Figs. 4b and 5b).

The EC1:5 increased with time for all dates of sowing at 0–30 cm soil depth (Fig. 3d, i). About 65 and 49% of increased values of EC1:5 can be explained by time after sowing in the first and second seasons (Fig. 3d, i) using the polynomial cubic relationship. The EC1:5 was around 0.5 dS m−1 at the start of the season, and it increased to 0.8 and 0.6 dS m−1 at the end of the season in the first and second years, respectively. In the second year, overall EC1:5 was lower due to rainfall in the beginning and later part of the season.

The SP decreased (i.e., became more negative) with the progress of the dry season due to increased soil salinity and reduced soil water content. The variation of the SP with time in all times of sowing can be well explained by the polynomial cubic relationship (r2 = 0.81 and P < 0.001 in the first season and r2 = 0.75 and P< 0.001 in the second season) (Fig. 3e, j). In both seasons, the SP dropped from around − 250 kPa at the beginning of the season to − 860 kPa at the end of the season.

In the present study, the time of sowing and seasonal rainfall altered the average and maximum soil EC1:5 in the soil profile within and between the two seasons. During establishment, the EC1:5 ranged from 0.5 to 0.68 dS m−1 and from 0.4 to 0.48 dS m−1 in the first and second years, respectively, but these values had no obvious effects on emergence and germination. Some previous studies have shown that sunflower germination is less sensitive to salinity than growth during the early seedling stage (Delgado and Sánchez-Raya 2007; van Hoorn 1991). At flowering and grain filling stages, high salt concentration can reduce seed numbers, seed size, and yield (Caterina et al. 2007). Across the season, early sowing before 15 December had lower EC1:5 (0.4–0.68 dS m−1) than later sowing until 10 January (0.6–0.78 dS m−1), which suggests that there was higher salt stress for sunflower growth after the late sowing.

In previous studies with a sunflower in this environment, SP of the soil solutions was an effective integrator of water and salt stress on plant growth and development (Paul et al. 2020a, b). When SWC decreased and soil salinity increased, SP decreased (became more negative), which inhibits water and nutrient uptake by plants, resulting in poor plant growth and development (Munns and Tester 2008). In the present study, there was a substantial variation of SP among the different sowing date treatments. In both seasons, early sowing had comparatively higher SP (varying from − 240 to − 620 kPa) at surface soil 0–7 cm depth from seedling to flowering, which was far above from the wilting point (− 1500 kPa), suggesting that water uptake would not be limiting. By contrast, in the later sowing treatments, plants experienced lower average SP mainly from flowering to maturity (varying from − 980 to − 1145 kPa), which was close to the wilting point. The higher SP for early sowing can be attributed to adequate soil water and limited salt accumulation in soils. By contrast, the low SP value that approached wilting point at flowering and grain filling stages, especially in the later sowing treatments, may have led to plant water stress and excess Na accumulation, which could explain the reduced number of seeds per head, the small grain size, and lower yield.

3.5 Effect of mulch on SWC, EC1:5, and SP

In both seasons, the RS treatment significantly increased the SWC for all times of sowing, and there was a significant interaction between sowing, mulch, and time (DAS) on SWC (Supplementary Materials Fig. S6a–j). On average, throughout the season, the RS treatment significantly increased SWC by 3–5% and 2–3% (w/w) over the RR treatment in the first and second years, respectively.

In the first season, there was no difference in EC1:5 for early sowing on 23 and 30 November between the two mulching treatments (Supplementary Materials Fig. S7a and S7b), but the EC1:5 was significantly lower under the RS treatment compared to the RR treatment after sowing on 10, 20, and 30 December (Supplementary Materials Fig. S7c, d and e). There was a significant interaction between mulch and time, but no interaction between sowing and mulch or sowing, mulch, and time at P > 0.05 (Supplementary Materials Fig. S7f–j).

In both seasons, SP decreased with the progress of the season in both mulching treatments (Fig. 7a, b). The SP under the RS treatment was significantly greater than with the RR treatment in both seasons.

Fig. 7
figure 7

Effects of mulch on the solute potential of soil solutions for five dates of sowing in a 2016–2017 and b 2017–2018 under the RS (rice straw) (blue circle) and RR (rice residue) (red circle) treatments. For RS, y = − 0.0007x3 + 90.266x2 – 4 × 106x + 6 × 1010, r2 = 0.81, and for RR, y = – 0.0011x3 + 143.6x2 – 6 × 106x + 9 × 1010, r2 = 0.82 in 2016–2017. For RS, y = 0.0007x3 – 88.236x2 + 4 × 106x – 5 × 1010, r2 = 0.72, and for RR, y = – 0.0016x3 – 202.43x2 + 9 × 106x – 1011, r2 = 0.73 in 2017–2018

Full size image

Mulching with rice straw significantly improved the availability of soil water and reduced the soil salinity, which in turn enhanced SP of soil solutions compared to the RR treatment. These benefits which strengthen the conclusions of Paul et al. (2020b) were more pronounced in the later part of the growing season. In addition, due to the higher SWC under the mulch, lower soil strength and cracks may favour root growth and crop development compared to the drier surface soil of the rice residue treatment (Abd El-Mageed et al. 2016; Paul et al. 2021). Clay soils puddled for rice cultivation are prone to becoming dense and hard following the formation of large cracks during the dry season; these effects are associated with reduced infiltration within blocks of soil and higher soil penetration resistance that restricts the emergence, root growth, and crop growth and development (Kirchhof et al. 2000; Mitchell et al. 2013; Paul et al. 2021). One reason for the absence of mulch effect on yield in the first year may be related to the completion of sowing before the end of December so that anthesis and reproductive stages avoided water deficit and salt stress regardless of mulching treatments. Another reason may be associated with 109 mm seasonal rainfall in the first week of March, which provided adequate soil water supply in both mulching treatments during the later growing period. Indeed, RS mulch on the soil surface was especially useful for a later sowing of sunflower (established on 8 January) for increasing yield, seeds per head, and seed weight (Paul et al. 2020b).

3.6 Implications for cropping systems intensification

Across the Eastern Gangetic Plain, most farmers in the last few decades have intensified their cropping systems by adding a second or third crop to the existing monsoon rice crop (Islam et al. 2019). Adding the second or third crop increases the potential profit obtained by smallholder farmers and improves their family livelihood. However, this opportunity has largely bypassed the smallholder farmers in the Ganges Delta due to the perception that salinity and limitation of fresh water make dry season cropping unprofitable and highly risky (e.g., Paul 2020). The present research suggests that a range of constraints may be limiting dry season crop production apart from salinity and drought, including waterlogging, low solute potential in the soil water, and heat stress. Indeed, the limitation from lack of soil water in this region has been disputed (by Bell et al. 2019) who suggested that early sown rabi season crops could escape the most severe effects of drought and salinity by making most of their growth, while the root zone contains relatively fresh soil water. The present study with sunflower supports this proposition by demonstrating higher SP and lower EC1:5 for the early sown crops. Other changes in the cropping system will be necessary to expand the early sowing of rabi season crops. Firstly, monsoon rice cultivars that mature earlier are essential so that excess water can be drained immediately after harvest to allow the early sowing of rabi season crops. Fortunately, there are suitable rice varieties that not only mature 15 to 30 days earlier but have 0.5–1 t ha−1 higher yield than the traditional monsoon rice varieties grown by farmers (Yesmin et al. 2019). Secondly, block planting of high yielding early maturing rice varieties needs to be followed by coordinated drainage of excess water immediately after harvest (Bell et al. 2019). Community coordination of the shift to early rabi season sowing could avoid conflicting demands for water depending on whether a farmer is growing an early maturing or late maturing monsoon rice variety.

4 Conclusions

On a wet, clay-textured soil in the Ganges Delta, we show for the first time that early sown sunflower established by dibbling had higher grain yield relative to later sowing. Moreover, we show that the higher yield from sowing before 15 December was associated with higher SWC, lower EC1:5, and greater SP. The decreasing yield with the late sowing was associated with the lower number of seeds per head and grain weight, and shorter growth duration due to warmer temperature later in the growing season and decreased SP of the soil solutions. Rice straw mulch was effective for increasing yield, especially at the late dates of sowing, by reducing water loss and salt accumulation in the root zone, thereby increasing SP. Yield prediction across all times of sowing by a 3-factor (temperature, SWC and EC) multi-variate regression analysis suggested that EC1:5 was the main driver of yield, followed by average seasonal temperature. Considering the two seasons results, early sowing before 15 December had the highest yield potential for sunflower by enhancing the utilization of residual, non-saline soil water early in the season, by ensuring that flowering is completed before high temperatures inhibit seed set, and by avoidance of high soil salinity and low SP in the upper root zone in the later part of the growing season. We conclude that early sowing of rabi season sunflower, and possibly other crops, could enable growers in the low-lying, salt-affected Ganges Delta to intensify their cropping systems by adding a high-yielding second crop such as sunflower and provided adequate field drainage is installed.