Searching for Suitable Cultivation System of Swiss Chard (Beta vulgaris subsp. cicla (L.) W.D.J.Koch) in the Tropical Lowland

Soil moisture was measured using a soil moisture meter (Lutron PMS-8714) at a depth of 5 cm below the upper surface of the substrate. The dry weight of plant materials was measured after drying in an oven at 100 °C for 24 to 48 hours, depending on the size or thickness of the samples. The harvest was carried out three times, i.e., 22, 25, and 28 days after transplanting (DAT). The essential data that have to be collected destructively were carried out during the final harvest.

For the development of the leaf area (LA) estimation model, a range of leaf sizes from the smallest to the largest was deliberately selected and the distribution of the leaf size was sought to be as even as possible. The length of the midrib of the leaf (L) as the length of the blade, the width of the blade measured as its dimensions at its widest point (W), and the L×W multiplication were used as predictors in the development of the LA estimation model. Two regression models were used, i.e., the zero intercept linear regression using L×W as a predictor and the power regression using L, W, and L×W as predictors (Kartika et al. 2021; Meihana et al. 2017). The goodness of fit of statistical models was proxied based on the coefficient of determination (R²) value to the value of LA measured using the digital image analysis software (LIA32, Kazukiyo Yamamoto, Nagoya University, Japan).

This study was arranged in a randomized block design with two factors. The first factor was the cultivation systems, consisting of conventional, floating, and BWCs. The second factor was the cultivars of Swiss chard ‘Red Ruby’, ‘Yellow Canary’, and ‘Pink Passion’. Each combination of two factors consisted of three replications, each replication consisted of three pots, and each pot was planted with three seedlings.

All data were organized and analyzed with the analysis of variance (ANOVA). Furthermore, significant differences between cultivars of the Swiss chard and between cultivation systems were assessed using the least significant difference (LSD) test. The value of P < 0.05 was referred to as statistically significant differences among cultivars or cultivation systems. Trend analysis between predictor and estimated LA was used in both the zero-intercept linear and power regressions.

The floating culture system exhibited the best Swiss chard growth comparable to conventional growth system after 14 DAT. At 21 DAT, the leaves were mostly the largest in plants grown in the floating system (Table 1). All pot systems take up little space in urban area and are considered as easy to manage, requiring no electrical energy, and of the low cost compared to hydroponic systems (Baiyin et al. 2020); thus, it is more likely to be adopted in the urban community, including people whose financial capacity is limited.

A similar reaction to the culture conditions was observed in the analysis of fresh and dry weight. At 22 DAT, fresh and dry weight leaf blades and petioles did not differ between the conventional and the floating system, but at 25 and 28 DAT the highest values were noted in plants grown in floating culture (Table 2). Also, the leaf numbers were higher and comparable to the conventional system, except for 25 DAT (Fig. 2).

Figure 2

The growth of Swiss chard plants cultivated with the BWC system was slightly inhibited in, comparison to plants cultivated with the other two systems. However, there was no significant difference in leaf dry weight in all three cultivation systems (Table 2). Consequently, the differences between the cultivation systems were associated with leaf water content (Table 3). In other words, plants cultivated in the floating system absorbed more water than those in other cultivation systems, especially over a longer period of culture. The high water content improves the freshness of the vegetables at harvest.

Air and soil temperatures in the tropical lowlands are higher than conditions in the temperate zone where Swiss chard is generally cultivated. High air temperatures can limit plant growth and development due to disturbances in metabolic processes (Zhu et al. 2021), while the limited availability of oxygen in the growing medium due to the presence of a water-saturated soil layer limits growth and reduces yields (Sharma et al. 2018).

For terrestrial plants including Swiss chard, the presence of a water-saturated layer at the bottom of the growing substrate in floating and bottom-wet cultivation systems was a limiting factor for root growth and development. The roots of plants cultivated in the floating system and BWCs were significantly shorter at 25 and 28 DAT than the roots of conventionally cultivated plants. Nevertheless, the fresh and dry weight of the roots was higher in floating cultivated plants. This indicated that despite the short main roots, plants cultivated with a floating system formed a more intensive lateral root system (Table 4).

The same phenomena were also observed in chili pepper (Siaga et al. 2018) and common beans (Lakitan et al. 2018). Meier et al. (2020) explained that plants could modulate their root system architecture to improve water and nutrient acquisition from the soil. Nitrogen application stimulated the accumulation of shoot-derived auxin in the root vasculature and promoted lateral root emergence to build a highly branched root system. In our study, nitrogen was supplied via a compound NPK fertilizer.

The growth and yield of bottom-wet cultivated Swiss chard plants were not as good as the floating cultivated plants. This difference was likely related to the thickness of the WSSLs, i.e., the depth of the bottom substrate whose pores are saturated with water. The thick WSSL (5 cm) as in the BWC negatively affected the root weight and the fruit yield of the ‘Green Apple’ eggplant (Jaya et al. 2019) and the overall growth of the leaf celery (Jaya et al. 2020, 2021).

In pot experiments of the same substrate volume, increasing the thickness of the WSSL means reducing the volume of the aerobic substrate. Therefore, in a floating cultivation system, the thickness of WSSL needs to be minimized but the contact between the bottom of the substrate and the water surface must be continuously maintained so that the water supply by the capillarity force can still take place. The WSSL in floating culture varied from 1 to 3 cm depending on rainfall and air temperature. In the BWC, the WSSL was controlled at 5 cm by continuously opening the outflow valve, and water was added manually if the WSSL fell below 5 cm due to evapotranspiration on a hot, rainless day.

The substrate surface absorbs sunlight energy, thereby increasing the substrate temperature, triggering evaporation, and decreasing soil moisture (Dincă et al. 2018). The advantage of floating and BWC was that the water supply by the capillarity force took place continuously so that the moisture of the growth substrate could be maintained (Jaya et al. 2021), as long as the direct contact between the substrate and the water surface was not interrupted (Kartika et al. 2021).

Among the tested cultivars, the petiole of ‘Yellow Canary’ Swiss chard exhibited the longest blade, the thickest leaf, and the highest ratio of the length of the midrib to the petiole, and therefore the largest leaf blade, because the leaf width was similar to the three cultivars studied. However, this cultivar produced the least number of leaves per plant, which was recorded at 21 DAT (Table 1). Smaller leaves are in some cultivars compensated with a greater number of leaves per plant.

Fresh leaf weight did not differ significantly between all the cultivars evaluated, with the exception of ‘Pink Passion’, which was lower at 22 and 25 DAT. In contrast, the number of leaves on ‘Yellow Canary’ became smaller compared to the other two cultivars at 25 and 28 DAT (Table 5, Fig. 2).

Based on the linear regression, there was a positive trend but a very weak correlation between LA and leaf width. Although statistically insignificant, the trend shown in the 14 and 21 DAT results provided interesting findings, i.e., at a younger age, the LA was still small but thick; then, after the leaves enlarged, the thickness decreased. This phenomenon suggests that lateral growth (elongation and widening) was becoming more dominant and vertical growth (thickness) was halted.

Using length × width (L×W) as a predictor of LA was consistently more accurate compared to using length or width alone for all cultivars studied. In this case, the use of zero-intercept linear regression was consistently more accurate when L×W was used as a predictor; conversely, the use of power regression was more accurate if L or W were used separately as a predictor. Conclusively, the most accurate combination was to use zero-intercept linear regression and L×W as predictors for ‘Red Ruby’ (R² = 0.9807), ‘Yellow Canary’ (R² = 0.9821), and ‘Pink Passion’ (R² = 0.9758), as indicated by the values of their respective coefficients of determination (Table 6).

The development of LA estimation models has attracted the attention of many agronomists and biologists. The resulting models are primarily for further use in nondestructive, rapid, and accurate prediction of the LA during its development, so that leaf growth analysis can be realistically calculated as it uses the same leaves. Most LA estimation models are based on regression equations using morphological traits as predictors, such as tomatoes (Meihana et al. 2017), cocoa (Salazar et al. 2018), tatsoi (Kartika et al. 2021), celery (Lakitan et al. 2021a), and many others.

Many types and combinations of traits have been used in estimation models for the LA, but the type of trait that was always consistent and accurate was the traits that are directly related to the dimensions of the leaf, namely the length and width of the leaf blade. The dimension of the leaves’ thickness is more difficult to measure and should be taken for choice for some species of succulent plants.

Although the normal leaf shape in a single plant looks the same, the length-to-width ratio of a leaf can vary significantly between leaves of the same plant. For this case, the use of the L×W combination as a predictor in the LA estimate will be more accurate than using L or W separately using linear regression. This argument had already been proved in the leaves of Bougainvillea (Fascella et al. 2018), Habanero chili (Lakitan et al. 2022), and Java ginseng (Lakitan et al. 2021b). Since LA should be zero if L or W is zero, then the most appropriate to use is the zero-intercept linear regression model (Lakitan et al. 2021b). If the estimation model uses only a single predictor L or W, then the most suitable regression option is the power regression. Power regression automatically sets LA = 0 if either L or W were zero.