Trial was conducted during the 2018-2019 growing season, from October 25, 2018 to March 14, 2019 at Mvuazi research center of the National Institute for Agronomic Study. Mvuazi is located at 14˚54' East longitude and 5˚21' South latitude, at 470 m of altitude (Figure 1).

The soil of Mvuazi belongs to the Sudano-Guinean climatic zone of AW4 type [23]. This soil is characterized by a low organic matter content and a low water retention capacity, resulting to a low availability of nitrogen [24] and the orthic soil type [25] (Table 1). The climatic data recorded during the experimental period are presented in Table 2.

Varieties Mudishi-1 (M1) and Mudishi-3 (M3) selected by the National Maize

Tmax = Temperature maximal, Tmin = Temperature mi[10]Nimal, Rain = Rainfall, Rel.Hum = Relative humidity, T.mean = Temperature mean.

Program of INERA, are cultivated in the most of agro-ecological zones of the DRC. They are preferred for their resistance to leaf diseases, to root and stem lodging, their yield potential of 4 to 6 t/ha [26] and their high rate in lysine and tryptophan [27].

The factorial experimental design with 3 replications and 3 factors was used (Figure 2). The first factor (percentage of damage by cutting the blade of leaves) consisted of five levels, namely 0% (no damage, which corresponds to the controls), 25% (leaves having lost the 25% of the blade), 50% (leaves having lost the half of the blade), 75% (leaves having lost 75% of the blade) and 100% (leaves having lost all of the blade). The second factor (growth stage) consisted of two levels: V3 and VT as defined by McWilliams et al. (1999), V3 which corresponds to defoliation at the 2 - 4 week after emergence and VT which corresponds to defoliation at the stage of flowering. The third factor (variety) was made up of two levels: M1 and M3. The maize plants have been sown at 0.75 m × 0.50 m spacing with 2 seeds per hill in the 1.50 m × 2 m plots.

The combination of percentage of damage (factor in sub-sub-plot), stages of growth (factor in sub-plot) and variety (factor in plot) allowed a total of twenty (20) treatments:

T0.1 (M1:0%:V3), T0.2 (M3:0%:V3), T1.1 (M1:25%:V3), T1.2 (M3:25%:V3), T2.1 (M1:50%:V3), T2.2 (M3:50%:V3), T3.1 (M1:75%:V3), T3.2 (M3:75%:V3), T4.1 (M1:100%:V3), T4.2 (M3:100%:V3), T5.1 (M1:0%:VT), T5.2 (M3:0%:VT), T6.1 (M1:25%:VT), T6.2 (M3:25%:VT), T7.1 (M1:50%:VT), T7.2 (M3:50%:VT), T8.1 (M1:75%:VT), T8.2 (M3:75%:VT, T9.1 (M1:100%:VT), T9.2 (M3:100%:VT).

Data collected concerned observations made on some parameters relating to yield performance were:

- Number of plants per useful plot (without borders)

- Number of ears harvested per useful plot

- Percentage of harvest: Number of ears of corn harvested from number of plants per plot multiplied by one hundred

- Percentage of marketable ears based on quotation of appearance: number of healthy ears (not attacked) on number of ears harvested multiplied by one hundred, the quotation scale of appearance of ear is from 1 to 5 where 1 = excellent, 2 = good, 3 = fairly good, 4 = not good and 5 = poor) [26]

- Ear diameter (mm)

- Ear length (cm)

- Yield: was measured in kilogram/m² following the formula here beyond.

Yield = (FW * (100 − FGM/100 − DGM) * (GW/EW) * (1/UA)) (1)

where:

FW: Field Weigh

FGM: Field Grain Moisture

DGM: Dry Grain Moisture

GW: Grain Weigh

EW: Ear Weigh

UA: Useful Area

- Grain loss (%): weight of the grains of a treatment with damage on weight of grains of the control without damage multiplied by a hundred.

Statistical analyzes:

The collected data were analyzed using packages agricolae and FactomineR of R3.1.3 software to determine difference among treatments by analysis of variance following the linear model (aov(y~Trait3*Trait2* Trait1+error (replication/Trait princ)) with respect to various growing and yield parameters. The means of treatment that exhibited significant differences were separated using the least significant difference (LSD) post-hoc test. A principal component analysis (PCA) to detect the correlations between the variables studied.

Table 3 shows the effects of factors and their interaction on characteristics parameters observed during plant growth and yield parameters.

With regard to Table 3, it can be seen that the damage caused by defoliation, the growth stage and their combined effect influenced significantly (p < 0.05) the number of ears harvested, the ear sizes (diameter and length), the harvest rate, the percentage of marketable ears, the yield and the yield losses. No significant difference (p > 0.05) was observed between treatments for the number of plants per plot. However, the variety factor and most of its effects interacting with the other factors did not directly influence (p > 0.1) the yield as well as the components of the yield (loss, harvest rate, marketable ears, size of ears).

PCA was performed on 8 active variables characterizing the treatments based on the data of the combination of factors. More than 87% of variability are represented on the plane formed by two exes Dim1 and Dim2 (Figure 3(A)). Dim1 is characterized by strong correlations between variables. This dimension benefits from the strong contributions of variables linked to productivity, including loss, yield, harvest rate, ear size, number of ears harvested. Losses are greatest when yields are low, however the other variables correlate positively with each other variables. Dim2 is characterized only by the number of plants

Sig[10]Nif. codes: 0 “***” 0.001 “**” 0.01 “*” 0.05 “.” 0.1 “ ” 1, Var = Variety; Dmg = Damage; Gstp = Growht stape; V3: 3 to 4 weeks after sowing; VT: Flowering stage.

per plot (Figure 3(C)). Less dispersion of varieties was observed around two axes (Figure 3(B)). However, there was a large dispersion of the damage rate factor compared to two axes. Based on the contributions of the factors on Dim1 (Figure 3(B)), it seems that the two varieties M1 and M3 combine similar values of the variables studied. However, there is a wide dispersion in damage rates.

For all factors taken individually and in interaction of factors, the number of plants per plot varied from 5.33 to 9.66 plants respectively for T3.1 and T1.2 (Table 4). The results of analysis of variance showed non-significant effects (p > 0.05) of factors and interactions between factors on the number of plant per plot. However, there were noticeable variations between the factors resulting in a large coefficient of variation. These variations should be linked to variety effects (p = 0.1) by looking at Table 3.

Regardless of variety, the number of ears harvested per plot varied significantly between the factors and interactions studied. The highest number, 9.66 ears per m², was recorded at T1.1 and the lowest number, 0 ears per m², was recorded at T9.2. In general, the low numbers of ears harvested were recorded for two varieties subjected to 100% of defoliation at flowering stage (VT). High numbers of

ears harvested were associated with low damage rates at three weeks after sowing (V3). No damage (0%) to flowering resulted in the maintenance of a high number of ears harvested per plot. Decrease in number of grain on the ear (cob) was probably the result of floral primordial or a slight pollination of the dichogamy [28]. Both varieties had a high yield in treatments without defoliation. This situation can be explained by the fact that in maize, the grain yield is determined by the quantity of silk and by the length of the period during which the sensitivity is very high [29] [30] [31].

The averages of diameters of the ears were appreciably affected by the factors caused damage, stage of growth and their interaction. These averages varied between 0 mm and 47.1 mm respectively for T4.2 and T3.2. Low values of ear diameter were recorded for high damage rates during the flowering stage (VT). Similarly to the ear diameter, the ear length was significantly influenced by the same factors. Length values varied between 0 and 17.29 cm respectively for T9.2 and T5.2. [32] [33] observed that defoliation decreases the size of the ears and it seems that the weight of the grains is more dependent on the genetic factor than on the environmental factor. Following [33], ear length and weight had a positive and significant correlation with all treatment. However, [34] observed that defoliation had non-significant effect on ear length.

Based on results, length ear decreased by 12% to 100% in VT treatments, where as other treatments do not have significant effect on ear length (Table 4). These results are in agreement with the finding of [35]. It is indicated that ear length is most affect by some factor such as defoliation time, intensity of defoliation and position of leaves on the plant [36] [37].

Marketable ear rates varied between 0% and 90.47% respectively for T9.2 and T3.1, T3.2. Most of the healthy ears were harvested under V3. Figure 4 shows the appearance quotations of the ears rated from 1 to 5 compared to the treatment corresponding to flowering stage. The ears of the plants that did not loosed leaves (0%) had a quote = 1, which is marketable. Plants treated with 25% damage produced ears with a quote score of 1.5 to 2. Plants treated with 50% damage showed ears quoted 2.5 to 3.5. As for the plants treated at 75% of damage, the quotation was 3.5 to 4, while the full defoliated plants (100%), ears had a quotation greater than 4.5, reducing significantly the rate of ear marketable.

Regarding Table 4, is appears that the best harvest rates were recorded after treatments before flowering stage or at V3. The damage caused at stage V3 had no negative effects (p > 0.05) on the production of grain. Low harvest rates (0.00% and 3.70%) were recorded respectively in treatments T9.2 and T9. [35] also reported damage to the silk of maize at 50% until 10 days later, which reduced yield by reducing the number of ears after 20 days or more, the yield was reduced by reducing the weight of a thousand grains. Various reports also indicated that a partial defoliation of about 25% to 33% did not change the weight of the seeds at any stage of seed development.

Estimated per m², this yield varied between 0.00 kg/m² and 0.35 kg/m² respectively for T9.2 and T5.2 (Table 4). The best productivity was recorded at the control treatment corresponding to 0% damage at V3 and VT. Productivity per m² was significantly affected (p < 0.05) by damage rate and growth stage. Varieties used did not influenced the yield after two periods of simulated attack.

Several biotics and abiotics stresses affect the yield of maize through defoliation. [38] reported that the critical rate of defoliation in maize was 50%; this rate did not affect grain yield and its components.

In accordance with the study made by [39] on sunflowers (Helianthus annuus L.), the yield has been greatly reduced by defoliation at the pre-flowering stage. This is similar to complete defoliation was more detrimental and caused yield loss of 6.4% to 82% compared to partial defoliation with loss of yield of 1.5% to 32.7%. These losses varied with the time of application of the treatment.

Still according to [35], the yield component most affected by leaf loss, for the entire treatment period, was the grain weight decrease from 12.7% to 53%. However, the number of grains was considerably reduced by 62.3% when all the leaves were removed 10 days after the appearance of the silk. Thus, the partial leaf removal affected the number of grains much less, approximate decline, 20%. According to [33], the maize plant in complete defoliation had the lowest grain yield, ear weight, number of rows per ear, ear weight and 100 grain weight, but had a percentage, higher grain germination rate and high vigor.

According to [40], the reduction in grain yield was associated with the number of defoliated leaves and this reduction was linked to the decrease in grain number. Based on the results of the trial, early defoliation in the reproductive phase causes yield losses by reducing the number of grains [41].

The average of grain yield losses (loss) varied between 0% and 100% regardless to varieties. These values corresponded respectively for 0% of damage and 100% of damage occurring at two growth stages. The losses observed were associated with the high damage rates and they are more accentuated with the damage caused during the flowering stage (VT).

According to [42], percentage yield loss is depending on factors such as on the amount of removed leaves, leaf position on plant and also defoliation time. This hypothesis can be explain by [43], who suggested that the near leaves to ear are main factor of increasing dry matters and growth rate during the development.

This study showed that a significant loss of leaves at the beginning of silk and up to at least 10 days after 50% of silk had resulted in a reduction in grain yield due to the reduction in the number of grains and regardless of the severity, defoliation in the days after the appearance of the silk considerably reduced the accumulated dry matter [35]. However, the yield declines associated with less severe leaf loss or leaf loss at least 20 days after 50% silk were largely related to a decrease in grain weight. In addition to providing a better understanding of the patterns of dry matter accumulation, these data can be useful for individuals who are required to estimate grain yield after significant loss, due to natural causes, of the leaf area of the plant [35].