In the Sudan Gezira, some 240,000 feddans of Egyptian-type cotton (Gossypium barbadense) are grown annually by gravity irrigation from the Sennar Dam, and this is sprayed with DDT, 6–10 weeks after sowing, in order to control the cotton jassid, Empoasca lybica (de Berg.), which is considered to reduce the yield. The seasonal yield response to DDT was estimated by comparing the yields of single sprayed and unsprayed 90-feddan fields, pairs of which were selected in each of the 40 (now 44) Blocks into which the cotton estate cropped by the Sudan Gezira Board is divided for administrative purposes. In order to establish that this estimate adequately represented that of the estate as a whole, it was necessary to show the validity of three assumptions: that the sample was representative; that the unsprayed plots, though surrounded by sprayed cotton, were sufficiently large to behave in respect of yield as if their environment had not been sprayed; and that the selection of the fields was effectively at random.
Examination of the yield history of the fields that were sprayed showed that there had been a slight bias towards the selection of fields that yielded better than the mean of the locality, and although this bias did not change from place to place or year to year, it did increase with level of yield. It is not, however, considered big enough to invalidate the first assumption.
The second assumption was tested by a special experiment reported elsewhere, which produced no evidence to suggest that a 90-feddan field would be affected in respect of yield and insect pests by proximity to a differently treated neighbouring field.
The third assumption cannot be proved, but during two seasons when fields were selected by a random procedure, the results did not differ seriously from those in years when the choice was not completely random. Moreover, when the yields of fields selected for treatment in one year were examined for the season when the fields were last cropped with cotton, and spray treatments were identical, they were found not to differ significantly, although the differences were increased and became significant in the year when sprayed and unsprayed treatments were applied.
Accordingly, the yield of the unsprayed sample is taken to indicate what the yield of the whole estate would have been had all the cotton been unsprayed.
The yield response to DDT spray of Gezira cotton, as thus estimated over the period 1949–50 to 1956–57, was shown to vary from season to season. A highly significant part of this variance was correlated with the amount of ram falling in July, some six weeks before the crop was sown, but the response was also significantly affected by both seasonal and site effects that were also correlated with this early rain. A series of experiments in 1956–57 showed that Domains Sakel cotton, which is grown in the drier, northern part of the Gezira, tended to give an increased response to DDT spray at higher levels of nitrogenous fertiliser, and that a very highly significant part of this response was due to the control of jassid, E. lybica, and thrips, (Caliothrips spp.). X1730A, grown in the wetter, southern part of the Gezira, gave no over-all response to spraying, nor was the response to nitrogenous fertiliser affected by spraying. When the yields of these varieties were examined separately over eight seasons, it was found that Domains Sakel gave a mean yield response to spraying of 1·11 kantars of seed cotton per feddan, which did not vary significantly from season to season, but that X1730A gave a yield response which was less when pre-sowing rains were good and nitrogenous fertiliser was applied. Although the regression on July rains and on fertiliser of the response of X1730A to spray failed to reach significance at P = 0·05, it is considered that further data should enable a prediction of yield response of X1730A to DDT spray to be made by examination of these two factors.
Since 1935–36, yields of cotton in the Sudan Gezira have been positively correlated with the amount of rain falling between 1st July and 15th August of the year of sowing. Comparison of the coefficient of regression of yield on this rainfall in the period (1935–36 to 1947–48) before sprays or fertiliser were applied, with that of a more extended period (1935–36 to 1954–55) including six seasons in which sprays and fertiliser were applied, shows that the latter is only about one half the former. After allowing for differences in varietal response to spraying and fertiliser, it can be shown that this decrease in the deleterious effect of a deficiency in pre-sowing rains, is almost exactly accounted for by the beneficial effect of DDT spray.
It is concluded that at least half of the deleterious effect of poor pre-sowing rains on Gezira yields is a pest effect which can be eliminated by DDT spray.
The study thus provides a means whereby DDT spray may be applied selectively to those areas of the Sudan Gezíra where its effect is likely to be most profitable and suggests, moreover, that such a policy would reduce the considerable seasonal fluctuations in cotton yields which in the past have characterised cropping in this area.