Cultivation of cotton cultivars The study was conducted in the research fields of Bangabandhu Sheikh Mujibur Rahman Agricultural University (BSMRAU, 25°25′N, 89°5′E), Gazipur, Bangladesh from July 2013 to June 2014 with the five cotton cultivars CB1, CB3, CB5, CB8 and CB12. The study location is a subtropical region having a dry season (February to May), a wet season (June to September) and a short winter (December and January). Annual mean relative humidity, rainfall, and temperature are 65.8%, 237.6 cm, and 24.5°C, respectively. The five cotton cultivars were cultivated following the recommendations of the Cotton Development Board of Bangladesh, excluding insecticide use. Treatments were assigned to a randomized complete block design (RCBD) comprising 4.0m × 4.0m plots. The spacing between all blocks and plots was 1.0m × 1.0m. Cotton seeds were sown 50 cm apart on 02 July 2013 in rows. The distance from row to row was 1.0m and each plot contained 32 plants. Recording of infestation After emergence of seedlings, field inspections were carried out weekly to record the population build up and aphid and jassid infestations. Due to a difficulty in differentiating the two insect pests, the infestation level of both was characterized in a single metric. At each inspection, three plants from each plot were randomly selected and the total number of leaves and the number of infested leaves on these plants were counted. Data collection was started from the first signs of infestation and continued for 5 weeks. When the plants produced a sufficient number of bolls, five plants were randomly selected from each cultivar and the total number of bolls and the number of infested bolls per plant were recorded. Each infested boll was opened and the total number of locules and the number of infested locules were counted. This procedure was carried out three times at 7-day intervals. Determination of starch content during the vegetative phase The starch content of uninfested and infested twigs of each of the five cotton cultivars was determined following the method of McCready et al. (1950). Infested and uninfested plant samples (500mg) of each cultivar were cut into small pieces and dipped into a flask containing 10mL of 80% ethanol. The samples were centrifuged at 2000 × g for 20 min. The supernatant (1.0mL) from each sample was collected in a tube, then phenol solution (5%, 1.0mL) and sulfuric acid (96%, 5.0mL) were rapidly added. The tube was gently agitated during addition of the acid, and allowed to stand in a water bath at 26–30°C for 20 min to facilitate total sugar recovery. Extracted soluble sugar was suspended in 5.0mL water and subsequently 6.5mL of 52% perchloric acid. The contents were centrifuged for 20 min at 2000 × g, and the resulting supernatant was decanted, collected, and the total volume was made up to 100mL with distilled water. This mixture was then filtered through Whatman filter paper (no. 42). A 1.0mL aliquot of each filtrate was analyzed for glucose content. Samples were measured at 630nm in a spectrophotometer and glucose concentration was determined using a standard curve prepared using known concentrations of glucose. Glucose concentrations were multiplied by 0.9 to calculate plant starch content. Quantifying agronomic traits Plant height, number of branches, number of sympodial branches, number of leaves and bolls per plant, number of locules per boll, and boll length, width, and weight were considered agronomic traits of the cultivars. Mature bolls were randomly collected from second node branches and classified as fresh and infested. For each variety, 15 normal and 15 infested bolls were selected. These were weighed using a digital balance (AG204, Mettler Toledo, Switzerland), and their widths and lengths were measured using a digital slide caliper (CD-S15C, Mituoyo, China). At first harvest, 15 plants of each cultivar were randomly selected, and for each the number of leaves, branches, and sympodial branches was counted, and plant height (length from top to base) was measured using an mm-scale tape measure. Quantifying seed cotton yield and crop quality All open bolls (seed cotton) from each plot were handpicked, sun dried, and then separately bagged in brown paper bags. Seed cotton yield was measured using a balance (CANRY, China) and expressed in kg ha-1. The bolls were ginned with a single roller electric gin in the laboratory of the Regional Cotton Research and Extension Centre, Gazipur, Bangladesh. Boll sample data included the following: ginning out-turn (%GOT: percentage of lint obtained from 100g of seed cotton), seed index (weight of 100 seeds), lint index (amount of fiber retained on 100 seeds), and micronaire (fiber fineness and maturity). Micronaire values were tested using a cotton micronaire testing machine (Zhenjiang KDL Machinery Co. Ltd., Jiangsu, China). Germination test The percentage of seed germination indicates the seed quality of the cultivars. A germination test was conducted in homogenous environmental conditions in the laboratory at 25°C. Trays, each 30cm × 30cm × 5cm (L × W × D), were used for this purpose. A single sheet of paper was placed in the bottom of each tray to cover drainage holes. The trays were filled with clean, moist sand, and fresh sand was used for each test. Seeds of each variety were randomly collected from their respective bags and counted. For each variety, 100 seeds were sown on each of five replicate trays in 10 rows of 10 seeds (500 seeds per variety). Seeds were sown at 2-3 cm depth and watered every second day. Only normal seedlings were counted after 10 days when the majority of seedlings had emerged. Diseased, discolored, or malformed seedlings were excluded from counts. The total number of normal, vigorous, and healthy seedlings for each cultivar was used to determine germination percentage. Normal seedlings, abnormal seedlings, and non-germinated seeds were defined and detected according to the protocol adopted by the International Seed Testing Association (ISTA). Data analysis Data were analyzed using one-way analysis of variance (ANOVA). The tstatistic was employed to test for differences between uninfested and infested samples in protein and starch content and boll length, width, and weight for each cultivar. Data were expressed as the mean±SD and the Turkey’s HSD post hoc test was used to test for differences between means.