Light Interception and Radiation Use Efﬁciency of Three Cassava Genotypes with Different Plant Types and Seasonal Variations

: The yield potential of cassava might be increased by enhancing light interception and the ability to convert energy into biomass and yield, which is described as radiation use efﬁciency (RUE). The objective of this study was to determine light interception, extinction coefﬁcient ( k ), and RUE of three cassava genotypes (Kasetsart 50 (KU50), Rayong 11 (RY11), and CMR38-125-77) under seasonal variations. The ﬁeld experiments were conducted in a randomized complete block design with four replications, using two planting dates for 2 years at Khon Kaen, Thailand. Data were recorded for weather conditions, light interception, leaf area index (LAI), and biomass. Solar radiation interception, RUE, and k were calculated. Light interception of the crop planted in May sharply increased in the early growth stage, whereas the crop planted in November slowly increased and could maintain higher light interception from the mid–late growth stages. Light interception and LAI had a moderate to high coefﬁcient of determination (R 2 = 0.61–0.89) for three cassava genotypes and all planting dates. The k values ranged from 0.59 to 0.94, varying by genotypes and planting dates, indicating that the leaf orientation of the three cassava genotypes was horizontally oriented. The relationship between biomass accumulation and cumulative solar radiation produced a high value of R 2 (0.86–0.99). The RUE for biomass (RUE bi ) varied by genotype and planting date, ranging from 0.66 g MJ − 1 to 0.97 g MJ − 1 . However, the RUE for storage root dry weight (RUE sr ) ranged from 0.29 g MJ − 1 to 0.66 g MJ − 1 . The RUE bi and RUE sr in each genotype on each planting date were signiﬁcantly different. The highest RUE bi and RUE sr were found at 4–6 and 7–9 MAP for almost all genotypes and planting dates, except for the crop planted in November 2015, when both RY11 and CMR38-125-77 had the highest RUEbi at 10–12 MAP. RY11 had a lower LAI compared to other genotypes, which contributed to lower light disruption and lower RUE bi and RUE sr . KU50 and CMR38-125-77 could maintain canopy light interception during canopy development and storage root accumulation stages and had high RUE bi and RUE sr , resulting in high biomass and crop yield.


Introduction
Cassava (Manihot esculenta Crantz) is an important starch crop worldwide and is mostly grown in tropical and sub-tropical regions (~30 • N to~30 • S). The products from cassava roots and leaves not only serve human consumption but are also used as animal (Phenacoccus manihoti Matile-Ferrero). Stalks were incubated in a gunnysack for 3 days before planting. Single stalks were buried to half their length on the ridge row. An overhead sprinkler irrigation system was installed to supply water for the cassava crop from planting to harvesting. Weeds were controlled by hand weeding at 1 and 2 MAP. Fertilizer was applied 2 times after weed control by calculating minimized rates of N:P 2 O 5 :K 2 O following Howeler [27]. Pests and diseases were controlled as necessary.

Weather Conditions
Weather data were recorded by an automatic data logger (Watch Dog 2700) (Watchdog, PCE group, PCE Germany, Meschede, Germany) that was installed in the field experiments (i.e., minimum and maximum temperature, daily rainfall, relative humidity, and photosynthetically active radiation (PAR, 400-700 nm)). Daily solar radiation was converted from daily PAR each day [28]. Maximum and minimum temperatures for cassava planted in May of 2015 ranged from 16.4 to 43.9 • C and from 8.9 to 29.5 • C, respectively, and relative humidity ranged from 22.9 to 92.8%. For November 2015, the maximum temperatures were 16.8 to 43.9 • C, the minimum temperatures were 8.9 to 29.5 • C, and relative humidity ranged from 22 In the year 2016, solar radiation ranged from 5.4 to 24.5 MJ m −2 d −1 , and the total was 6120.0 MJ m −2 (May planting); however, the November planting crop had solar radiation ranging from 7.7 to 24.8 MJ m −2 d −1 , and the total amount of solar radiation was 5974.5 MJ m −2 ( Figure 1).

Soil Physical and Chemical Properties
The soil samples at depths of 0-30 cm were taken to analyze the soil's physical and chemical properties. The Yasothon soil series (Typic Paleustult) with loamy sand and sandy loam was found in all experimental plots. Soil pH ranged from 7.01 to 7.41 except for the November 2015 planting date, when soil pH ranged from 6.27 to 6.50. The soil of the four planting dates had organic matter ranging from 0.44 to 0.53% and total nitrogen ranging from 0.013 to 0.037%. The available phosphorus in all four planting dates ranged from 51.6 to 88.5 mg kg −1 . The critical level of available phosphorus in the soil is about 4-6 mg kg −1 (Bray II-extractable phosphorus) [27]. Thus, the soil in this experiment had sufficient phosphorus for all planting dates. For soils used in all four planting dates, exchangeable potassium ranged from 30.8 to 54.6 mg kg −1 at 0-30 cm soil depth. Therefore, the soil in this experiment had insufficient exchangeable potassium.

Light Interception and Cumulative Solar Radiation
Light interception was defined by light penetration through the plant canopy. The light penetration could be calculated by measuring PAR light above the canopy and below the canopy (under the plant and between row) by using a 1.0 m line quantum sensor (Licor 191) and data logger (LI-1500, LI-COR) [29]. A set of data consisted of three measurements at three times per plot. The data were recorded under clear sky conditions at near noon (11.30 am-1.00 pm) at monthly intervals after 1 MAP until harvested (12 MAP). Light interception percentages were calculated by [PAR (above the plant canopy)-PAR (average below the plant canopy)] × 100 divided by PAR (above the plant canopy).

Soil Physical and Chemical Properties
The soil samples at depths of 0-30 cm were taken to analyze the soil's physical and chemical properties. The Yasothon soil series (Typic Paleustult) with loamy sand and sandy loam was found in all experimental plots. Soil pH ranged from 7.01 to 7.41 except for the November 2015 planting date, when soil pH ranged from 6.27 to 6.50. The soil of the four planting dates had organic matter ranging from 0.44% to 0.53% and total nitrogen ranging from 0.013% to 0.037%. The available phosphorus in all four planting dates ranged from 51.6 to 88.5 mg kg −1 . The critical level of available phosphorus in the soil is about 4-6 mg kg −1 (Bray II-extractable phosphorus) [27]. Thus, the soil in this experiment had sufficient phosphorus for all planting dates. For soils used in all four planting dates, exchangeable potassium ranged from 30.8 to 54.6 mg kg −1 at 0-30 cm soil depth. Therefore, the soil in this experiment had insufficient exchangeable potassium.

Light Interception and Cumulative Solar Radiation
Light interception was defined by light penetration through the plant canopy. The light penetration could be calculated by measuring PAR light above the canopy and below the canopy (under the plant and between row) by using a 1.0 m line quantum sensor (Licor 191) and data logger (LI-1500, LI-COR) [29]. A set of data consisted of three measurements at three times per plot. The data were recorded under clear sky conditions at near noon (11.30 am-1.00 pm) at monthly intervals after 1 MAP until harvested (12 MAP). Light The cumulative solar radiation intercepted was calculated from the amount of solar radiation in each month (recorded by weather station) multiplied by the percentage of intercepted light of the cassava canopy for each genotype in each month's measurement. The data of solar radiation intercepted for each genotype were summarized.

Leaf Area Index (LAI) and Extinction Coefficient (k)
LAI in each plot was measured at one-month intervals in the 6 × 7 m 2 area by using the LAI-2000 Plant Canopy Analyzer (LI-COR, Lincoln, NE, USA) under overcast and/or clear sky conditions [30]. Three readings were taken in each subplot, and the data were averaged. The relationship between the percentage of light interception and the LAI was calculated using the equation proposed by Monsi and Saeki [26]; where I = light received at below the plant canopy, Io = incoming light just above the plant canopy, k = extinction coefficient, which indicates the leaf arrangement in a canopy [31,32].

Total Biomass and Yield
Data on biomass and storage root dry weight in each genotype were recorded from six plants in the sampling area (6 m 2 ) of each plot at one-month intervals from one month continuing to harvest for the crop planted in May and November in the year 2015. However, for both planting dates in the year 2016, the data for biomass and storage root dry weight were recorded 8 times at 1,3,4,6,8,9,10, and 12 MAP. At the final harvest (12 MAP), the data were recorded from 18 plants (18 m 2 ) in each plot. The plant samples in each sampling date were separated into leaves, petioles, stems, and storage roots. The total fresh weight of each plant part was recorded immediately, and a sample of more than 10% for each plant part was oven-dried at 80 • C for 72 h or until the weight was constant to determine dry weight. Data on the storage root dry weight of each plot were recorded, and the total biomass in each plot was calculated.

Radiation Use Efficiency (RUE)
The coefficient of the linear regression between biomass and cumulative solar radiation could be referred to as RUE for all crop growth cycles. The RUE for biomass (RUE bi ) and RUE for storage root (RUE sr ) in each growth stage were considered. The RUE of each subplot in every 3 months intervals (0-3, 4-6, 7-9, and 10-12 MAP, followed by cassava growth stages as described by Alves [33]), was calculated by biomass and storage root dry weight (g m −2 ) divided by the cumulative solar radiation intercepted (MJ m −2 ) by the canopy of each subplot in each sampling time.

Statistical Analysis
Data on cumulative solar radiation, RUE bi, and RUE sr in each crop growth stage for all genotypes in each planting date and each year were analyzed following a randomized complete block design (RCBD). Mean comparisons based on the least significant difference (LSD) were conducted for all crop growth stages and planting dates. The regression analysis was carried out to determine the relationship between light interception and LAI, accumulation of biomass, and cumulative radiation. The extinction coefficient (k) was determined by the relationship between ln(I/Io) and LAI. All of the statistical analyses were performed using Statistix 10 program [34] and by following the procedure described by Gomez and Gomez [35].

Light Interception
There was a sharp increase in light interception for the crop planted in May 2015 for all three genotypes that peaked at 3-6 MAP. After that, light interception slightly declined at 7 MAP ( Figure 2a). During the late growth stage, RY11 had a higher light interception percentage than KU50 and CMR38-125-77. However, for the November 2015 planting, all three cassava genotypes had slightly increased light interception that peaked at 5 MAP and was then maintained at a light interception of more than 50% until harvest ( Figure 2b). RY11 seems to maintain a higher light interception percentage than the other genotypes.
The pattern of light interception for the crops planted in 2016 was similar to 2015 for both May and November plantings. In May 2016, RY11 had a higher light interception percentage than KU50 and CMR38-125-77 during the late growth stage (Figure 2c). However, the crops planted in November 2016 had slightly increased light interception that peaked at 7 MAP, and after that, the crop maintained light interception of more than 80% until harvest ( Figure 2d).

Light Interception and LAI
A multiple regression analysis was performed to evaluate the relationship between light interception and LAI. The correlation between light interception and LAI was positive and highly significant for all genotypes and planting dates. The determination coefficient (R 2 ) for regression analysis between light interception and LAI was moderate to high (0.61-0.89) ( Figure 3). In May 2015, the light interception for KU50 increased by 24.85% with increasing LAI 1 unit, whereas those of RY11 and CMR38-125-77 increased by 18.99% and 18.52%, respectively ( Figure 3a). However, maximum light interceptions were observed when LAIs were 5.0-6.0. The crops planted in November 2015 had a higher potential to intercept light than the crops planted in May 2015 for all three genotypes. KU50 could increase light interception by 41.01%, while those of RY11 and CMR38-125-77 increased interception by 51.96% and 38.38%, respectively ( Figure 3b). The light interception reached the maximum when LAIs were 3.0-4.0; however, the light interception declined when LAIs were over 4.0. at 7 MAP (Figure 2a). During the late growth stage, RY11 had a higher light intercepti percentage than KU50 and CMR38-125-77. However, for the November 2015 planting, three cassava genotypes had slightly increased light interception that peaked at 5 MA and was then maintained at a light interception of more than 50% until harvest (Figu 2b). RY11 seems to maintain a higher light interception percentage than the other gen types.
The pattern of light interception for the crops planted in 2016 was similar to 2015 both May and November plantings. In May 2016, RY11 had a higher light intercepti percentage than KU50 and CMR38-125-77 during the late growth stage (Figure 2c). Ho ever, the crops planted in November 2016 had slightly increased light interception th peaked at 7 MAP, and after that, the crop maintained light interception of more than 80 until harvest (Figure 2d).

Light Interception and LAI
A multiple regression analysis was performed to evaluate the relationship betwe light interception and LAI. The correlation between light interception and LAI was po tive and highly significant for all genotypes and planting dates. The determination coe cient (R 2 ) for regression analysis between light interception and LAI was moderate to hi For the crops planted in May 2016, the correlation between LAI and light interception was positive and significant for all genotypes. However, the highest light interceptions were observed when LAIs were 3.0 to 4.0 ( Figure 3c). For RY11, a 1 unit increase in LAI resulted in a 71.40% increase in light interception, which was higher than KU50 (51.69%) and CMR38-125-77 (53.82%). In November 2016, the highest light interception reached the peak when LAI was 2.5 for CMR38-125-77, while the highest for KU50 and RY11 were 3.0 to 4.0, respectively ( Figure 3d). CMR38-125-77 had the highest performance of intercepted light.

Extinction Coefficient (k)
In this study, the linear regression plotted against LAI and Ln (I/Io) showed a moderate to high determination coefficient (R 2 = 0.58-0.93). The slope of regression is referred to as k. The k values in this study ranged from 0.59 to 0.94 depending on genotype and planting date. For the crop planted in May 2015, RY11 (0.59) had a lower k value than KU50 (0.70) and CMR38-125-77 (0.76) (Figure 4a). The results indicated that the angle between leaf and sunlight of RY11 was smaller than for KU50 and CMR38-125-77. However, for the Novem- was positive and significant for all genotypes. However, the highest light interceptio were observed when LAIs were 3.0 to 4.0 ( Figure 3c). For RY11, a 1 unit increase in L resulted in a 71.40% increase in light interception, which was higher than KU50 (51.69 and CMR38-125-77 (53.82%). In November 2016, the highest light interception reached t peak when LAI was 2.5 for CMR38-125-77, while the highest for KU50 and RY11 were to 4.0, respectively ( Figure 3d). CMR38-125-77 had the highest performance of intercept light.

Extinction Coefficient (k)
In this study, the linear regression plotted against LAI and Ln (I/Io) showed a mo erate to high determination coefficient (R 2 = 0.58-0.93). The slope of regression is referr to as k. The k values in this study ranged from 0.59 to 0.94 depending on genotype a planting date.

Cumulative Solar Radiation Interception
The cumulative solar radiation interceptions were calculated from the amount of lar radiation in each crop growth stage and the percent intercepted over time. Genotyp were not significantly different for the total amount of solar radiation interceptions with planting date for all 4 planting dates, however, different growth stages within genotyp were significantly different (Table 1).
For the crop planted in May 2015, the cumulative solar radiation interceptions amo

Cumulative Solar Radiation Interception
The cumulative solar radiation interceptions were calculated from the amount of solar radiation in each crop growth stage and the percent intercepted over time. Genotypes were not significantly different for the total amount of solar radiation interceptions within planting date for all 4 planting dates, however, different growth stages within genotypes were significantly different (Table 1).
For the crop planted in May 2015, the cumulative solar radiation interceptions among genotypes were not significantly different in each crop growth stage. However, among the crop growth stages within genotypes, a significant difference for cumulative solar radiation interceptions were observed in all three genotypes. The highest cumulative solar radiation interceptions were observed during canopy development (4-6 MAP) for all three genotypes (1429-1474 MJ m −2 ).
For the crop planted in November 2015, genotypes were not significantly different for cumulative solar radiation interceptions in each crop growth stage. However, the cumulative solar radiation interceptions in each crop growth stage within genotypes were significantly different. All three genotypes showed the highest cumulative solar radiation interceptions during canopy development (4-6 MAP) (1448-1480 MJ m −2 ) and at the storage root accumulation stage (7-9 MAP) (1367-1537 MJ m −2 ).

Biomass and Cumulative Solar Radiation
The linear regressions between biomass accumulation and cumulative solar radiation are presented in Figure 5. The slope of regression could be referred to as RUE bi for the cassava crops. Biomass accumulation increased with increasing cumulative solar radiation with a high determination coefficient (R 2 = 0.86-0.99), indicating that the cumulative solar radiation could predict the total biomass of cassava with high precision. In May 2015, all three cassava genotypes had higher determination coefficients (R 2 = 0.86-0.95) (Figure 5a). However, biomass accumulation of RY11 increased 0.72 g MJ −1 , was lower than KU50 (0.98 g MJ −1 ) and CMR38-125-77 (0.80 g MJ −1 ), respectively. Likewise, in the November 2015 planting, KU50 (0.87 g MJ −1 ) and CMR38-125-77 (0.85 g MJ −1 ) had higher efficiency to convert solar radiation to biomass than RY11 (0.74 g MJ −1 ) (Figure 5b).
The patterns of regression for biomass and cumulative solar radiation in the year 2016 were similar to the year 2015 for both May and November plantings. In May 2016, the cassava genotypes were slightly different in RUE bi . CMR38-125-77 (0.81 g MJ −1 ) had higher performance in converting solar radiation to biomass than KU50 (0.72 g MJ −1 ) and RY11 (0.66 g MJ −1 ) (Figure 5c). However, in November 2016, KU50 (0.97 g MJ −1 ) and CMR38-125-77 (0.97 g MJ −1 ) were more efficient in converting solar radiation to biomass than RY11 (0.74 g MJ −1 ) (Figure 5d).

Storage Root Dry Weight and Cumulative Solar Radiation
In this study, the cumulative solar radiation was significantly (p < 0.01) and positivel correlated with storage root accumulation for all three genotypes with high R 2 value (0.79-0.99). The storage root dry weight increased with increasing cumulative solar rad ation. The patterns of regression between cumulative solar radiation and storage root dr weight for all three genotypes were similar in most planting dates except for May 201 when KU50 had a lower converting ability. For the crop planted in May 2015, CMR38 125-77 (0.42 g MJ −1 ) had a higher ability to convert solar radiation to storage root tha KU50 (0.37 g MJ −1 ) and RY11 (0.29 g MJ −1 ) (Figure 6a). In November 2015 planting, bot KU50 (0.45 g MJ −1 ) and CMR38-125-77 (0.44 g MJ −1 ) had a higher converting ability tha

Radiation Use Efficiency for Biomass (RUEbi)
For the crop planted in May 2015, RUEbi ranged from 0.44-1.31 g MJ −1 (Table Among genotypes, the RUEbi was not significantly different in each crop growth sta except for the canopy development stage (4-6 MAP) when KU50 (1.09 g MJ −1 ) and CMR3 125-77 (0.97 g MJ −1 ) had higher RUEbi than RY11 (0.68 g MJ −1 ). However, among the cr growth stages, only KU50 showed a significant difference for RUEbi. The highest RU was 1.31 g MJ −1 at 7-9 MAP and 1.09 at 4-6 MAP.

Radiation Use Efficiency for Biomass (RUE bi )
For the crop planted in May 2015, RUE bi ranged from 0.44-1.31 g MJ −1 ( Table 2). Among genotypes, the RUE bi was not significantly different in each crop growth stage, except for the canopy development stage (4-6 MAP) when KU50 (1.09 g MJ −1 ) and CMR38-125-77 (0.97 g MJ −1 ) had higher RUE bi than RY11 (0.68 g MJ −1 ). However, among the crop growth stages, only KU50 showed a significant difference for RUE bi . The highest RUE bi was 1.31 g MJ −1 at 7-9 MAP and 1.09 at 4-6 MAP.
The RUE bi among genotypes within growth stages were not significantly different for both May and November plantings in 2016. However, in May 2016, CMR38-125-77 showed a significant difference in RUE bi among crop growth stages. This genotype showed higher RUE bi 4-6 (1.16 g MJ −1 ) and 7-9 MAP (0.77 g MJ −1 ) than those in 0-3 and 10-12 MAP. Among the crop growth stages in November 2016 planted crops, there were significant differences for RUE bi for almost all genotypes, except for KU50. RY11 had the highest RUE bi during 7-9 MAP (1.16 g MJ −1 ), whereas CMR38-125-77 exhibited higher RUE bi at 4-6 (1.19 g MJ −1 ) and 7-9 MAP (1.05 g MJ −1 ). ns, *, ** = non-significant and significant at p < 0.05 and p < 0.01 probability levels, respectively. Means with the same capital letters in each column and small letters in each row are not significantly different by least significant difference (p < 0.05).

Radiation Use Efficiency for Storage Root Dry Weight (RUE sr )
The RUE sr for each genotype and each crop growth stage are shown in Table 3. The RUE sr in this study ranged from 0.08-0.94 g MJ −1 . For the crop planted in May 2015, there were significant differences among genotypes for RUE sr at 4-6 and 7-9 MAP. KU50 and CMR38-125-77 had higher RUE sr than RY11 at 4-6 MAP, whereas during 7-9 MAP, KU50 had the highest RUE sr (0.94 g MJ −1 ). However, the data showed significant differences in RUE sr among crop growth stages in each genotype. The highest RUE sr were found at 7-9 MAP for KU50 (0.94 g MJ −1 ) and CMR38-125-77 (0.56 g MJ −1 ), whereas RY11 had the highest RUE sr during 10-12 MAP (0.71 g MJ −1 ).
For the November 2015 planting, there were no significant differences among genotypes for almost all the crop growth stages except for 4-6 MAP. After this growth stage, KU50 (0.39 g MJ −1 ) and CMR38-125-77 (0.41 g MJ −1 ) had higher RUE sr than RY11 (0.25 g MJ −1 ). Comparisons for RUE sr among growth stages in each genotype showed significant differences. KU50 had higher RUE sr for almost all crop growth stages, except for during 0-3 MAP. RY11 showed the highest RUE sr at 7-9 and was also high at 10-12 MAP. CMR38-125-77 had the greatest RUE sr at 10-12 MAP (0.66 g MJ −1 ).
For the crop planted in May 2016, a significant difference among genotypes were found at 10-12 MAP, when KU50 had the highest RUE sr (0.55 g MJ −1 ). The RUE sr among crop growth stages were not significantly different for both RY11 and CMR38-125-77, however, KU50 showed the greatest performance in 4-6 (0.56 g MJ −1 ) and 10-12 MAP (0.55 g MJ −1 ). For the November 2016 planting, KU50 and CMR38-125-77 exhibited greater RUE sr than RY11 during 4-6 MAP. However, among the crop growth stages, the RUE sr were significantly different for RY11 and CMR38-125-77. The best performance for RUE sr was shown at 4-6 (0.47 g MJ −1 ) and 7-9 MAP (0.61 g MJ −1 ) for RY11, and 4-6 MAP for CMR38-125-77 (0.90 g MJ −1 ). ns, *, ** = non-significant and significant at p < 0.05 and p < 0.01 probability levels, respectively. Means with the same capital letters in each column and small letters in each row are not significantly different by least significant difference (p < 0.05).

Discussion
The plants with different canopy architectures led to differences in the area of intercepted light and affected the RUE. In this study, the crop planted in May (3-6 MAP) could reach maximum light interception earlier than the crop planted in November (5 and 7 MAP for 2015 and 2016, respectively). The different peaks for light interception due to variation of crop canopy varied by climatic factors during the crop growth stages. The crop planted in May grew rapidly in the early growth stages because they were subjected to high temperature, high relative humidity, and high solar radiation during the early growth stages, whereas, during early growth for the crop planted in November, the canopy establishment was slower caused mainly by low temperature, low humidity and low solar radiation [36]. The previous studies reported that high temperature and high solar radiation increased canopy size due to increasing new leaf emergence, leaf growth, and expansion, and increasing in LAI [37][38][39][40][41]. Canopy development of cassava planted in May grew faster than those of November plantings in the early growth stages resulting in greater light interception earlier than the crop planted in November. However, light interception for the crop planted in May declined slowly during the mid and late growth stages, which was mainly due to low temperature and low RH, resulting in the falling of old leaves and delaying the formation of new leaves, contributing to smaller canopy size and LAI [35,40]. Although in the May planting, the three genotypes were not significantly different for light interception during early growth stages, at the mid and late growth stages, RY11 had higher light interception than other genotypes. This might be because RY11 maintained a higher canopy size and LAI at the mid and late growth stages than KU50 and CMR38-125-77. However, for the crop planted in November, all three genotypes were not significantly different in light interception for all crop cycles, although RY11 tended to have a higher light interception than other genotypes during the late growth stage. Although light interception was correlated to LAI, the largest LAI does not result in the highest light interception in several crops [11,40].
In this study, the relationship between LAI and light interception could be explained by quadratic regressions, with a moderate to high degree of determination (R 2 = 0.61-0.89), varying by genotypes, planting dates, and years. The light interception was increased when increasing LAI. KU50 and CMR38-125-77 planted in November had higher increases in light interception than the crop planted in May of both years. For RY11, the crop planted in May 2016 had a higher performance than other planting dates. Moreover, the maximum light interceptions appeared when LAI was 2.5-4.0 for most genotypes and planting dates, except for May 2015, when LAI was 5.0-6.0. The highest LAI occurred in May 2015 due to a larger canopy size because the crop was subjected to higher temperatures and higher solar radiation than the crop planted in May 2016 [36].
After reaching the peak, light interception declined for all genotypes and all planting dates, even though LAIs increased. In the 2015 May planting, increasing LAI more than 5.0-6.0 decreased light interception by 1.26-2.13%, and KU50 had a higher reduction of light interception than other genotypes. However, in May 2016, the light interception was reduced when LAI was higher than 3.5 for KU50 and CMR38-125-77, whereas in RY11, the light interception was reduced earlier, when LAI was 2.5. In November 2015, the crop's light interception declined when LAI was higher than 4.0 for all three genotypes. However, in November 2016, the LAI of CMR38-125-77 could reach the maximum light interception earlier than other genotypes, when LAI was 2.5. After that, the ability to intercept light declined, having a higher reduction of light interception (13.05%) than KU50 (7.66%) and RY11 (6.45%). In terms of optimum LAI, the results agreed with other reports, which found that the optimum LAI was 3.0-3.5 [2,35,42]. Cassava intercepts about 90% of the total solar radiation when it reaches LAI of 3.0-4.0 [9,40]. However, the same LAIs may not provide the same light interception [9] due to leaf positioning, leaf curving, and leaf angle differences among genotypes.
The relationships between light transmission (ln I/Io) and LAI were explained by the k value, which is referred to as the leaf angle of plants [26]. In this study, k values ranged from 0.59-0.94, varying with genotype and planting date. Likewise, other reports revealed that k values in cassava generally fall between 0.60 and 0.88 when full canopy cover has been reached [39,40,42]. A report by Pellet and El-Sharkawy [9] showed k values ranging from 0.50-0.78 depending on varieties and fertilizer applications. CM 523-7 and M Col 1684 cassava genotypes exhibited increased k values when fertilized. In the present study, the crops planted in May 2015 for all genotypes had lower k values than those the crops planted in November 2015, whereas in the year 2016, the crops planted in May had higher k values than those planted in November, except for KU50. Furthermore, the k values were more than 0.5, indicating that the leaf position for all three genotypes in this study displayed a horizontal position. According to Boote and Loomis [43], large leaf size and horizontal leaf position could receive higher solar radiation than small leaves in a vertical position. The smaller k values in the May plantings might result in lower light interception than in the November plantings and may contribute to lower biomass and yield. However, the variation in k values was not consistent but depended on genotype and seasonal variations [44]. A study by O'Connell et al. [45] reported that the effects of seasonal conditions were minimal on the k values and RUE for wheat and pea. However, both wheat and pea are short-cycle crops, whereas cassava has a long-life cycle and is subjected to various seasons throughout the year. Additionally, cassava leaves adjust petioles and leaf blades to track or avoid solar radiation [46]. The seasonal variation might affect the k values, and the responses of cassava with different canopy structures are different.
In this study, RY11 had higher light interception than KU50 and CMR38-125-77; however, the penetration into the lower part of the canopy was lower than other genotypes [47]. This is presumably due to RY11 being an early forking type, starting about 57 days after planting, and the forking levels at least two times more than the other genotypes [48]. Although forking produced more apexes and more leaves, leaf size after forking is smaller, with shorter petioles than the leaves on the main stem [33]. Moreover, RY11 had a shorter internode length than the other genotypes, resulting in a dense canopy and contributing to lower light penetration. Even though both KU50 and CMR38-125-77 were forking types similar to RY11, their first forking occurred later than RY11 [48]. As a result of forking behavior, the light penetrated the lower parts of the canopy, contributing to all leaves in the canopy having more to intercept light and greater photosynthetic capacity in KU50 and CMR38-125-77.
Due to variations in canopy structure, LAI and k in different planting dates affected the ability to intercept light and contributed to cumulative solar radiation interception. Several reports revealed that cumulative solar radiation interception was positive and highly correlated to biomass production and crop yields, which is referred to as RUE. The RUE bi in this study ranged from 0.66 to 0.97 g MJ −1 , and RUE sr was from 0.37 to 0.66 g MJ −1 . A previous study reported that RUE bi was 1.15-2.30 g MJ −1 [9] and RUE sr was 0.69-0.94 g MJ −1 and varied by genotype [41,49]. For crops planted in May for both 2015 and 2016, the RUE bi and RUE sr for the total crop cycle in all three genotypes were not significantly different, although RY11 tended to be lower than the other genotypes. However, for the November plantings for both years, KU50 and CMR38-125-77 had higher RUE bi and RUE sr than those of RY11. For the crop planted in May 2015, KU50 and CMR38-125-77 had higher RUE bi than RY11 during the canopy development stage (4-6 MAP) with significant differences.
Mahakosee et al. [47] reported that crops planted in May had lower biomass and yield than November plantings for most genotypes except RY11. Although in the early growth stage, the crop planted in May had a larger canopy size and reached light interception earlier than the crop planted in November. In contrast, the crop planted in November could maintain canopy size and light interception during the mid to late growth stages, and the greatest cumulative solar radiation interception started from canopy development and continued to the storage root accumulation stage. This provided for higher biomass and yield in the November planting. For the May planting, three cassava genotypes were not significantly different for biomass and yields, whereas the November planting, KU50, and CMR38-125-77 had higher biomass and yields than RY11, even though the LAIs, light interception, and cumulative solar radiation interception for all three genotypes were similar. Due to the denser canopy, less light was available for lower canopy leaves of RY11, resulting in low leaf photosynthesis capacity.

Conclusions
Cassava genotypes with different canopy structures differed in light interception capacity and RUE. The cumulative solar radiation interception was significantly different among crop growth stages. For the May planting, the highest radiation interception occurred during 4-6 MAP, whereas, when planted in November, the crop canopy intercepted more solar radiation during 4-6 and 7-9 MAP. The RUE bi varied by genotype and planting date and ranged from 0.66 (RY11, May 2016) to 0.97 (KU50 and CMR38-125-77, November 2016). However, the RUE sr ranged from 0.29 (RY11) to 0.66 (CMR38-125-77). The RUE bi and RUE sr in each genotype on each planting date were significantly different. The highest RUE bi and RUE sr were found at 4-6 and 7-9 MAP for almost all genotypes and planting dates, except for the crop planted in November 2015, when both RY11 and CMR38-125-77 had the highest RUE bi at 10-12 MAP. KU50 and CMR38-125-77 could maintain canopy light interception during canopy development and storage root accumulation stages and had high RUE bi and RUE sr , resulting in high biomass and crop yield. Our study suggests that the selection of appropriate cassava canopy during canopy development and storage root accumulation stages with a high light interception and RUE could be used as a criterion for selecting genotypes in breeding programs.