Two year field experiment was carried out at Cotton Research Station, D. I. Khan (situated at 31˚49'N latitude and 70˚55'E longitude) during the crop season 2015-2016 and 2016-2017. Four cultivars viz., CIM-600, CIM-616, CIM-622, and CRIS-641 were investigated during the study. These cultivars were cultivated at various times in order to generate variation in temperature, humidity and heat units for growth, development and maturity, thus providing an extensive choice of various weather variables from planting till maturity. The climatological statistics was collected from Arid Zone Research Centre (AZRC), PARC, D. I. Khan. The baseline temperature for cumulative heat units (CHU)/ growing degree days (GDD) calculation was 16˚C. The CHU for different phases were calculated via the equation as outlined by Dwyer and Stewart [8] .

where, T_max, T_min and T_b are maximum, minimum and baseline temperatures respectively below which development ceases.

The soil of the experimental site was bit saline in nature and less fertile having electrical conductivity of 2.66 dS∙m⁻¹, pH of 7.90, organic matter 0.87%, sand 151 g∙kg⁻¹, silt 451 g∙kg⁻¹, clay 400 g∙kg⁻¹ and total N 0.99 g∙kg⁻¹.

The trial was carried out at 12 environments (2 years × 6 sowing times) (Table 1).

The four cotton cultivars (CIM-600, CIM-616, CIM-622, CRIS-641) were sown at 6 diverse planting times in the cropping year 2016 and 2017. These cultivars were sown in randomized complete block (RCB) design having 3 replications. Sowing was done in a four row subplot having 10 m length and a spacing of 0.75 m between and 0.30 m within the hills. Seeds were cultivated in hills having 5 seed which was immediately irrigated. 60 kg∙ha⁻¹ of phosphorus in the form of single super phosphate and 50 kg∙ha⁻¹ of N as urea were used before sowing and 50 kg∙ha⁻¹ N as urea at flowering and 50 kg∙ha⁻¹ at boll formation stage. 20 days after sowing, seedlings were thinned to single plant per hill. Picking was performed after 130 and 150 days of sowing. All the agronomic and plant protection practices were performed as per recommendations.

List of genotypes used in the study.

Data on number of bolls plant⁻¹, boll weight (g) and seed cotton yield (kg∙ha⁻¹) were recorded. The data recorded were subjected to analysis of variance technique appropriate for RCB design as suggested by Steel et al. [9] . After having homogeneity test for error variances by using Bartlett’s tests [10] , combined analysis of variance was performed. To determine the statistical differences in means, least significant difference (LSD) test at 5% level of probability was used when the F value was significant. For this purpose, MSTATC [11] statistical package was utilized. For diagrams and regression analysis [9] , Microsoft excel 2007 was used.

Combined analysis of variance depicted significant (P ≤ 0.01) differences among genotypes (G) and planting times (D) for number of bolls, boll weight and seed cotton yield (Table 2) presenting the incidence of variation amongst the genotypes in addition to environments. Furthermore, it proposes that some of the genotypes were superior to others in these studied traits. Significant mean squares of genotypes also depicted higher genetic variability among the tested genotypes. This variation was recorded for all the studied attributes which can be used as a valuable breeding source for several purposes. The G × D interaction was also significant (P ≤ 0.05) for boll weight indicating that various genotypes performed differently in different environments (Table 2). It further indicates significant genotype × environment interaction. It can be thus concluded that in cotton breeding program, environmental effects in understanding plant growth must be given due consideration. These results are in accordance with the findings of Machado et al. [12] and Ullah et al. [13] who also recorded significant genotype × planting times, genotype × year and genotype × year × location interaction. Significant genotype × environment interaction in upland cotton was also recorded by Unay et al. [14] and Satish et al. [15] . Similarly significant mean squares for G, E, and G × E was also recorded for various morphological traits by Khan et al. [16] and Gul [17] .

Number of bolls plant⁻¹ plays a key role in managing the differences in seed cotton yield on account of the reason that it is the major independent yield component. Hence in cotton improvement selection for more number of bolls plant⁻¹ must receive due emphasis. Analysis of variance revealed that planting times significantly (P ≤ 0.01) affected the number of bolls plant⁻¹ (Table 3). Number of bolls plant⁻¹ increased with delay in planting time and more number of bolls

*, **significant at 5% and 1% level of probability respectively, using the F test.

Means followed by similar letters in the same column are not significantly different at 5% level of probability.

plant⁻¹ (23.17) was recorded in the crop sown on 15^th April whereas further delay in planting reduced the number of bolls plant⁻¹ and thus minimum was recorded in the crop planted either very early (18.83) on 15^th March or delayed (14.25) on 1^st June. This reduction in number of bolls plant⁻¹ in early and delayed planting might be attributed to the variation in temperature prevailed by the crop growth. The genotypes studied also bared different number of bolls plant⁻¹. The mean comparison of the genotypes depicted significant differences at 5% probability and thus CIM-616 has the maximum (22.97) number of bolls plant⁻¹ followed by CIM-622 (18.50) while the lowest was recorded in genotype CRIS-641 (17.69).

Statistical analysis of the data depicted significant differences (P ≤ 0.01) in boll weight among various planting times. Maximum boll weight of 2.91 g was recorded in a crop sown on 15^th April which was followed by the crop planted on 1^st April (2.80 g) or 1^st May (2.78 g) (Table 3). Boll weight enhanced with delayed planting time from 15^th March to 15^th April. However further delay in planting reduced the boll weight which might be due to unfavourable photoperiod and high temperature at early growth and development stage that forced the crop to end up the life cycle rapidly at the cost of reduction in yield and yield related attributes. Different genotypes also produced significantly different boll weight. Highest boll weight of 2.89 g was recorded in CIM-616 which was followed by CIM-622 and CRIS-641 (Table 3). Seed cotton yield in cotton was highly affected directly and indirectly both by boll numbers and boll weight [18] [19] .

Planting times and suitable genotypes plays the most significant role in yield and yield related components. The cotton crop planted on 15^th April depicted maximum seed cotton yield of 2096.70 kg∙ha⁻¹ on account of having maximum number of bolls plant⁻¹ boll weight, and most favourable environmental conditions in comparison to other planting times investigated in the study. The crop planted on 1^st April ranked second with 2026 kg∙ha⁻¹ regarding the seed cotton yield (Table 3). It is evident from the table that planting times and genotypes significantly influenced the seed cotton yield. The seed cotton yield enhanced from 1792.50 to 2096.70 kg∙ha⁻¹ as planting time delayed from 15^th March to 15^th April while further delay in planting time reduced the seed cotton yield and delayed planting from 15^th April to 1^st June reduced the seed cotton yield to 1109.30 kg∙ha⁻¹ reflecting 304.20 kg∙ha⁻¹ increment up to 15^th April and then 987.40 kg∙ha⁻¹ decrements in seed cotton yield respectively (Table 3). Determination of optimum planting time and suitable cultivar for certain region is the major agronomic factor and utmost important to obtain higher seed cotton yield. The yielding ability of the present genotypes was also the result of the different interaction (D × G, G × Y, D × G × Y) factors [16] . Different growing conditions resulted in variation in seed cotton yield from year to year variation. Both genotypes and environment contributed to the variation in seed cotton yield however the environmental complex depicted primary yield effects in comparison to genotypes [20] [21] . The finding of the present study was also in accordance with and the environmental factor was predominant for the yield variation.

Significant difference was also observed for seed cotton yield among different genotypes. The maximum seed cotton yield of 2083.60 kg∙ha⁻¹ was recorded for CIM-616 which was followed by CIM-622 (Table 3). Cotton genotypes revealed significant differences through various environments and their interaction for seed cotton yield, indicating considerable variation among genotypes as well as environments [14] [17] .

Regression analysis of planting time and seed cotton yield also depicted significant relation between these two attributes. Besides, R² (coefficient of determination) revealed that up to 99% of the increase till 15^th April and later on decrease in seed cotton yield might be due to the influence of planting time. The regression between seed cotton yield and planting time was negative and regression analysis suggested that increase in one unit (15 days) of planting time increased/decreased seed cotton yield by 93.76 kg∙ha⁻¹ (Figure 1).

Environmental factor, particularly temperature was the main attribute that influenced the plant growth and development. Significant variation among genotypes regarding growing degree days were recorded which explained that different genotypes have different maturing period. Accumulation of maximum growing degree days/heat units for the 15^th April and 1^st May planting during both the years provided a clue that 15^th April to 1^st May is the best planting time for better cotton crop to harvest good yield (Table 4). The yield of 1^st April was also comparable. The accumulation of GDD/CHU determined the yield and

Means followed by similar letters are not significantly different at 5% level of probability.

yield related attributes because temperature is the primary factor which governs the crop growth rate [22] .