Solanum aethiopicum Shum is a nutrient rich leafy vegetable consumed in Sub Saharan Africa1. It is thus a potential remedial to hidden hunger2,3,4,5,6,7,8. Smallholder farmers and vendors find the crop a largely affordable enterprise because of low input requirements along the production-marketing value chain1,9. In Uganda only, at least 4,000,000 people are directly engaged in either production or vending of the S. aethiopicum Shum1,10,11; justifying need for research investment. High population involvement has as well been reported elsewhere3,7,12,13,14. Extent of contribution of the crop to alleviation of hidden hunger is however, threatened by biotic and abiotic constraints15,16,17. Postharvest deterioration and drought stress are the two major moisture-related abiotic challenges directly affecting market vendors and farmers, respectively1,16,18. At the market place, non-turgid leaves either sell for low price or they are entirely rejected by consumers, attracting a direct loss to the vendors9,18. Similarly, drought stress reduces leaf size thereby reducing the harvest index19,20 to the detriment of the farmer. These issues point to a need for identification of long storage and/or drought tolerant varieties1,2,9. However, S. aethiopicum is a research neglected crop5 and inadequate breeding clues for drought improvement exist17,20,21,22.

An absence of a reliable drought phenotyping protocol for S. aethiopicum Shum group curtails the development of improved varieties23. Optimum drought screening conditions in terms of effective crop growth stage for selection against drought stress, wilting points and discerning parameters had not been thoroughly investigated. Studies by Sseremba et al.17, Banik et al.24, and Kesiime et al.25; as well as reviews by Osakabe et al.26 and Fahad et al.27 opined that parameters namely leaf relative water content (LRWC), plant height and number of green leaves per plant can discriminate among genotypes under drought stress at 25 percent field capacity. Though, comparisons across growth stage and continuous moisture depletion were not made. Whereas the LRWC is one of the most important measure of plant-water status17,23,24 in leafy vegetables as an indicator of leaf quality17,28,29, the parameter is semi-destructive24,25 and too time-consuming to be used in high throughput and large scale phenotyping. The current study took care of three growth stages namely seedling, vegetative and reproductive whereby watering was withheld permanently at each stage. Further, non-destructive sensor-based monitoring of soil moisture content, chlorophyll content and stomatal conductance was made; in addition to leaf wilting scores, leaf size19,30 and leaf number27.

It is reported that watering depletion impairs growth as a result of decreased cell turgor and leaf wilting, chlorophyll content, stomatal conductance for CO2/O2 and reduced photosynthetic efficiency23,30. The reduction in photosynthetic efficiency is an intermediate effect through a signal31,32. Either an abscisic acid (ABA) dependent or ABA independent signal in a feedback mechanism of relative soil-plant water tension32,33,34,35,36, stimulates stomata closure (reduces gas exchange) in order to prevent further water loss. Otherwise, a continued gas exchange that maximizes photosynthetic efficiency under drought stress results in leaf rolling (wilting)23,37. Closed stomata attract exhaustion in levels of CO2 for photosynthesis at the expense of photorespiration (due to accumulated O2 relative to CO2)19,26,31,32. Photorespiration produces reactive oxygen species which damage the thylakoids; thereby reducing the chlorophyll content. The devastation in photosynthetic apparatus and thus reduced cell division and dry matter accumulation continues to worsen till plant death unless re-watering29,30 is done.

As mentioned already, consequent to compromised photosynthetic efficiency, drought effects in leafy vegetables are commonly morphologically manifested as plant stunting, reduced leaf size, reduced number of leaves, reduced yield and leaf wilting/rolling27. As a plant grows, adaptive structures such as extensive root system and waxing of leaf surfaces (cell wall remodeling) also develop38. Vegetative stage in grain crops is however, less critical to drought than flowering/reproductive stage in almost all flowering plants.

In leafy vegetables, it is the vegetative stage that matters most to growers and severe drought affects such crops18,27,28. This study took interest in verifying water depletion effects on leaf wilting, number of green leaves per plant and leaf blade width in S. aethiopicum Shum. The main objective of the study was therefore to identify growth stage-specific watering thresholds in S. aethiopicum Shum and specifically to determine the gradual tendency of leaf morphological traits amidst declining soil moisture.


Effect of watering depletion on leaf traits

There was a very highly significant difference (p < 0.001) in stages, watering regime by plant growth stage and day since last watering by plant growth stage for all the parameters measured (SMC, LWS, LPP and LBW) (Table 1). Under drought stress (DS) over time (days since last watering), seedling stage had the wettest soil (21.2% SMC) followed by vegetative (20.6%) and the least was under flowering stage (11.8% SMC) (Table 2).

Table 1 Mean squares for soil moisture content and leaf traits measured.
Table 2 Mean soil moisture content at different growth stages over time.

Across days, the highest LWS under DS was observed at flowering stage (2.6) followed by vegetative (2.4), and seedling stage had the lowest score (2.0) (Table 3). At seedling stage, leaf wilting was visible after day 6. For both vegetative and flowering stages, leaf wilting was visible after day 2. The differences between LWS from date of visible leaf wilting to the extreme LWS attained was lower for the flowering stage (difference of 3.3 attained on the 10th day since stress imposition) than for the vegetative stage (difference of 3.5 attained on the 8th day since stress imposition). The extreme LWS observed was 3.9 on 15th day followed by 4.3 on the 10th day and 4.5 on the 8th since stress imposition for seedling, flowering and vegetative stages, respectively. As expected, the LWS under WW was always 1.0 (no leaf wilting).

Table 3 Mean leaf wilting scores at different growth stages over time.

Number of green leaves per plant (LPP) under DS were highest (LPP = 12) at flowering followed by seedling (4) and vegetative (4) (Table 4). At seeding, vegetative and flowering, a reduction in LPP was observed after day 13, day 3 and day 3, respectively. Under well watering (WW) regime, LPP was always higher than that under DS at each stage of plant growth.

Table 4 Mean number of green leaves per plant at different growth stages over time.

As naturally may be expected, LBW was highest at flowering stage (10.8 cm) followed by vegetative (10.0 cm) and then seedling stage (8.9 cm) (Table 5). Within plant growth stage, LBW started declining after day 6 for the seedling stage. For vegetative and flowering stages, the decline in LBW was observed after day 3 and day 2, respectively (the measurement was taken on most fully open but non-wilted leaves).

Table 5 Mean leaf blade width over time nested within watering regime nested within growth stage.

Stage-specific watering thresholds

Moisture decline over time at each stage

Standard curve for seedling stage was above that for vegetative followed lastly by that for flowering stage (Fig. 1). The moisture decline was steepest for flowering stage followed by vegetative and relatively gentle for seedling stage. At flowering stage, within two days of drought stress, the soil moisture content (SMC) dropped from 45.2% to 17.2% and it continued to drop to 3.3% by the 8th day and continued to decline to 1.6% by day 10. For the vegetative stage, SMC declined quite fast from 46.5% (day 1) to 4% by day 8. The SMC declined the slowest from a high level (50.0%) at day 1 to 15.3% (day 8); then the moisture continued with the slow decline to 4.3% at ~day 15. The coefficients of determination (R2) were very high for both seedling (98.9%) and vegetative (97.6%), and high for flowering (96.1%) stages based on exponential models.

Figure 1
figure 1

Soil moisture depletion curve at seedling, vegetative and flowering stages during the experiment.

Leaf wilting point at each growth stage

The standard curves touched the SMC curve at critical wilting points below whose moisture levels leaf wilting causes economic loss. The seedling, vegetative and flowering stage curves touched the SMC curve at 16.5, 18 and 16 percent moisture levels respectively (Fig. 2). Clearly visible leaf wilting symptoms (LWS ≥ 2.0) occurred at 20% SMC by day 8, 4 and 3 for plants at seedling, vegetative and flowering stages, respectively. The highest leaf wilting score was attained at SMC 9% (seedling), 8% (vegetative) and 11% (flowering stage) by the 15th, 8th and 10th day. The generated standard curves were exponential with R2 = 0.87, 0.97 and 0.96 for seedling, vegetative and flowering stages, respectively.

Figure 2
figure 2

Leaf wilting curves for seedling, vegetative and flowering stages under declining soil moisture content over time.

Critical soil moisture content for preventing leaf loss

Generally, the leaf loss curve at flowering was above that for vegetative followed by that for seedling stage (Fig. 3). During seedling stage, LPP were stable from ~46.0% SMC up to 15% SMC. However, the curve for seedling exhibited a relatively gradual decrease in number of leaves after 8th day. At vegetative stage, LPP began gradually reduce after the 3rd day while a fast decrease in number of leaves was observed after the 4th day when the SMC was 16%. On withholding water, LPP started reducing after a day however the decrease was gradual. Further, early decline in SMC was detectable through corresponding reductions in LPP for both vegetative and flowering stages. When SMC decline continued beyond 16%, it was clear that all leaves could most likely be lost under vegetative followed by seedling and the highest leaf retention (LPP at extreme moisture deficit) was observed at flowering stage. The generated standard curves were polynomial with R2 = 0.67, 0.96 and 0.59 for seedling, vegetative and flowering stages, respectively.

Figure 3
figure 3

Standard curves for leaves per plant and soil moisture content at three growth stages over time.

Critical soil moisture content for preventing shrinkage in leaf blade width

Standard curve at flowering was above that for vegetative below which followed the seedling stage curve (Fig. 4). At seedling and vegetative the critical SMC were at 32% by the 4th day and 38% by the 2.5th day. The critical SMC for flowering stage could not be attained in number of days stipulated for drought stress. Below the critical SMC levels, LBW started declining. A gradual increase in LBW was observed from day 1 to day 8 at SMC 40% at seedling stage beyond which the LBW size begun a gradually decrease. At vegetative stage, leaf expansion was observed between the 1st and 2nd day after which a gradual decline was observed until the 8th day. The generated standard curves were polynomial with R2 = 0.81, 0.92 and 0.72 for seedling, vegetative and flowering stages, respectively.

Figure 4
figure 4

Standard curves for leaf blade width and soil moisture content at seedling, vegetative and flowering stages over time.


Watering depletion gradually led to decline in SMC which correspondingly increased leaf wilting, reduced the number of green leaves per plant and reduced the leaf size (measured as leaf blade with). Fastest leaf wilting, drop in LPP and LBW at flowering than rest of stages points to high plant sensitivity at flowering23. The view of high plant sensitivity to moisture changes at flowering has severally been corroborated in grain crops such as maize19,26,27,28. Overall, crop moisture requirements increase with growth stage23,26. At seedling stage, moisture deficit can lead to complete crop failure when germination and emergence is affected27. In this study however, drought stress was imposed after seedling emergence and so, germination/emergence was studied. The quantities of water required at seedling stage are way below that at subsequent stages27,35,39. This study reported of high SMC at seedling as compared to other stages. The reason is that seedlings do not have the capacity to take up and transpire large quantities of moisture from soil. The main avenue of soil moisture loss at seedling stage can be direct evaporation from the soil surface depending on solar radiation, ambient temperature and relative humidity19,27,35,39. Indeed, when watering was withheld, it took about twice the period as compared to vegetative and seedling stages for soil moisture to drop to same level across stages. Specifically, whereas SMC decline from field capacity to 4.5% for the case of vegetative stage in 8 days and flowering stages required 10 days, the same reduction at seedling stage had taken 15 days. The suggestion is that it would be expensive in terms of time required to observe selectable variation. Focusing on rest of stages (vegetative and flowering), a bigger difference in visible leaf wilting for a shorter period was observed for vegetative than the flowering stage (Table 5). In addition, the highest coefficient of correlation was observed at vegetative stage for all evaluated traits. It implies that screening for drought tolerance at vegetative stage is achievable in the shortest period. Being a crop grown mainly for its leaves at the vegetative stage1, S. aethiopicum Shum variety selection at the harvest could result in direct benefits to most of value chain actors such as farmers, vendors and consumers9.

The S. aethiopicum Shum group seedlings, just like for other crops, require constant watering; lest they quickly become wilted and productivity declines as observed earlier in Miscanthus spp. by38. This study observed that the critical SMC was highest at vegetative followed by seedling and flowering. However, response to water deficit stress was earliest for flowering followed by vegetative and seedling stages. Nonetheless, the plants at flowering stage endured for the longest (without complete leaf wilting) compared to the vegetative stage; suggesting suitability of the latter for drought screening. As seedling stage, plants do not demand a lot of water and there is limited surface area for water loss through transpiration23. Conversely, at advanced growth stages (typically the reproductive), plant structure is well developed to withstand low moisture tension of the soil19. At the flowering stage, plant systems are tuned to survival fitness through maximizing flowering/pollination success/fertilization success and subsequent seed development19,23. During the flowering period, plants express drought stress detection genes for “housekeeping” to safeguard against excessive water loss26; this explains tendency of leaves to wilt at high soil moisture potential23,26 even though complete leaf death is delayed. The potential defensive “housekeeping” against drought stress effects at flowering stage also accounts for the reduction in LPP and LBW at relatively high soil moisture potential26,31,32. There exists a clear difference in critical SMC at seedling, vegetative as compared to flowering stage.

A reduction in SMC below a particular critical level results in manifestation of drought effects24,35,39. For maximum productivity of the crop, it is logical to apply stage-specific optima of water which do not fall below critical levels19,27. Based on LWS, the critical watering intervals (action threshold for watering after a period of drought stress) should be eight, four and three days for seedling, vegetative and flowering stages, respectively. The critical watering intervals based on LWS were the same as those for LPP and LBW for the case of seedling and vegetative stages. This study could not establish the action thresholds for watering at flowering stage based on LPP and LBW since the values of the parameters never shrunk to extreme low levels as SMC declined. This study did not also delve into economic thresholds for watering requirements for S. aethiopicum Shum. During drought phenotyping of germplasm, drought stress treatments in S. aethiopicum Shum group should be set below the identified critical levels. An actual SMC to impose as drought stress treatment would be guided by the diversity of test germplasm23,28,35,37; and the contextual breeding objective in terms of parameters to measure (say, LWS, LPP or LBW) and desired selection intensity. In this study, focus was on three morphological variables (LWS, LPP and leaf size); raising need to corroborate findings when physiological traits as well as molecular underpinnings are also included33,34.

Materials and Methods

Plant material

A farmer preferred genotype E16 (also coded as 184 P, pedigree SAS184/P/2015)17,20,21,22,29 was selected following an earlier on-farm participatory evaluation (results of farmer preference study are unpublished). The genotype E1621 out of ten S. aethiopicum Shum group genotypes was advanced for national performance trial for yield evaluation, and distinctiveness, uniformity and stability (DUS) testing by Uganda Christian University. The farmer desirability of E16 and its relative morphological DUS prompted us to adopt it as a model accession for drought phenotyping.

Study location

An experiment was carried out in screen house at Uganda Christian University, Mukono for 90 days. Screen house weather conditions. Midday screen house temperature (°C), relative humidity (%) and light intensity (lum/ft2) remained relatively stable (slight variations) throughout the period of the study (Fig. 5). The model screen house temperature (Temp), relative humidity (RH) and light intensity (LI) was 38 °C, 38.6% and 48.0 (122.0) lum/ft2, respectively (Table 6). The weather data was recorded using a data logger (LGR S/N: 10280023, SEN S/N: 10280023).

Figure 5
figure 5

Variation in mid-day temperature, relative humidity and light intensity during period of the experiment. S, V and R; stages at which watering depletion was imposed (S, seedling stage; V, vegetative stage; R, reproductive stage). Temp, temperature; RH, relative humidity; LI, light intensity; ln(LI)*10; natural logarithm (ln) of light intensity (LI) multiplied by a constant 10.

Table 6 Summary of screen house weather data during the experiment.

Experimental design

A split-plot arrangement was visualized in a nested treatment structure. The main plot and subplot factors were watering regime (WR) and day since last watering by plant growth stage (Day), respectively. Two WRs namely well-watering (WW) and drought stress (DR); and three growth stages (seedling, vegetative and flowering) were evaluated. The seedling, vegetative and flowering stages were defined as 4 weeks after planting (WAP), 6 WAP and 8 WAP, respectively. For all stages, planting was carried out the same day in December 2017 (12th December 2017).

For each stage, 20 pots (filled with a potting substrate comprising a mixture of loam soil and manure in a ratio 3: 1) were allocated per replication. The pots were of 10 kg soil capacity. Field capacity (100% FC) of the potting substrate was established following a procedure described in Sseremba et al.17 and Nakanwagi et al.20. It was found to be ~3.0 litres of water per 10 kg of potting substrate. Therefore, the WW regime was at 100% FC.

Two replications per WR per stage were used. Each pot was planted with four directly sowed seedlings which were thinned to two plants per pot at three weeks after planting. The replication was not taken as a factor (no blocking) and the experiment was conducted in the screen house at Department of Agricultural and Biological Sciences, Uganda Christian University. The actual drought treatment experiment lasted 15, 8 and 10 days for the seedling, vegetative and flowering stages, respectively.

In effect, at each plant growth stage and WR, 80 plants were evaluated. The DS regime was introduced following a WW pre-watering until the respective evaluation stage. That is to say, experimental pots were continuously well-watered until they reached designated growth stages at which watering was withheld (hence DS). The WW controls and DS experimental pots were then maintained throughout the study. Lighting, temperature and humidity conditions of the screen house used as well as agronomic practices for potted plants were similar to what was described in our earlier study29 and as updated using the weather data presented in Fig. 5 and Table 6.

Data collection

Data was collected at seedling, vegetative and vegetative stages starting from day 1 since withholding of watering until severe leaf wilting symptoms become apparent. Individual pots were the observational unit for the case of soil moisture content (SMC). The leaf morphological traits namely leaf wilting score (LWS), number of green leaves per plant (LPP) and leaf blade width (LBW) were measured on individual plants. Decision to concentrate on the four parameters was based on the insights provided by Banik et al.24 and Sseremba et al.17,29. The SMC was measured in individual pots in percentage volume by volume using a portable digital soil moisture meter (model SKZ111K-1B, SKZ Industrial Co., Limited, D-11, No.9 Lanxiang road, Tianqiao District, Jinan, Shandong province, China). The LWS was taken on individual plants using a scale of 1 to 5 where 1 = no visible leaf wilting symptoms, 2 = 1–25% of leaves are showing slight wilting; 3 = 25–50% of leaves are wilted; 4 = 50–100% of leaves are wilted but still attached to plant; and 5 = complete defoliation and stem scorching. The LPP and LBW are measured as described in Sseremba et al.17.

Statistical analysis

Effect of watering depletion on leaf traits

A nested design model was analyzed using GenStat 12th edition (VSN International, Hemel Hempstead, UK). At experimental setup, the number of days since last watering (Day) was nested within watering regime (WR) and plant growth stage (Stage). The following analysis of variance model was thus considered:

$${Y}_{ijk}=\mu +Stag{e}_{i}+Stag{e}_{i}(W{L}_{j}(Da{y}_{k}))+{\varepsilon }_{ijk}$$

where \({Y}_{ijk}\) stands for observed measurement (any of SMC, LWS, LPP and LBW); \(\mu \) stands for grand mean; and \({\varepsilon }_{ijk}\) stands for random error. \(Stag{e}_{i}\), \(W{L}_{j}\) and \(Da{y}_{k}\) stand for the \({i}^{th}\) plant growth stage, \({j}^{th}\) watering regime nested within growth stage, and \({k}^{th}\) day since last watering nested within watering regime and growth stage. Differences (least significant differences) among treatments were declared at 1% error margin.

Identifying stage-specific watering thresholds

Taking growth stage as a grouping factor, a non-linear exponential regression model of form \(Y=a+b{r}^{X}\) was analyzed using GenStat Release 12.1. The \(Y\) and \(X\) standard for response and explanatory variates; while a, b and r are non-linear regression parameters. The r parameter is elaborated as \(r={e}^{-k}\); that is to say, \(\,k=-\,lo{g}_{e}r\). Hence, the exponential model can as well be expressed as \(Y=a+b{e}^{-kX}\). The regression model for each pair of variates (Table 7) was considered significant at 5% error margin. Alongside exponential regressions in GenStat, iterations for alternative models (linear, logarithmic, polynomial and power) were conducted in spreadsheets using Pivot Chart with a view of identifying a model for each pair of variates where mean squares (MS) are minimized. Minimum MS across model iterations was judged based on proportion of variance explained (R2). A model resulting in maximum R2 was selected and appended on the standard curve generated from the exponential model \(Y=a+b{r}^{X}\).

Table 7 Model variate pairs for moisture depletion related questions in S. aethiopicum Shum.


Watering depletion negatively affects leaf morphological traits in S. aethiopicum Shum group. The plants continue to wilt, lose leaves and reduce leaf size as long as the water deficit treatment is sustained. The leaf wilting point at seedling, vegetative and flowering stage were identified as 16.5%, 18% and 16% SMC, respectively. Visible leaf wilting symptoms (LWS ≥ 2.0) for seedling, vegetative and flowering stages can be observed by the 8th, 4th and 3rd day respectively. We decline to conclude on watering intervals required to prevent soil moisture from dropping to below wilting points or visible leaf wilting symptoms at each stage because of likely seasonal differences in ambient temperature and relative humidity which might influence rates of evapotranspiration; an issue for further research. Since observable leaf wilting symptoms generally occur at SMC levels below those for drop in LPP and LBW, drought screening treatments should include at least a treatment of <20% SMC for the evaluated stages. These findings are useful guide for drought phenotyping of germplasm in S. aethiopicum Shum as well as related species.