The reintroduction of cotton to the Australian semi arid tropics (SAT) was prevented by insect pests

that are dominant during the wet (summer) season and a perception that the crop could only be

grown in the wet season. Growing cotton during the dry (winter) season could avoid these pests

provided an integrated pest management system was adopted. However the photothermal pattern of

the dry season is the reverse of the wet season and that of spring sown cotton in temperate

latitudes. Cold night temperatures are possible mid season and high temperatures are likely early

and late in the season. Solar radiation is 20% less than at temperate latitudes mid season and

could also limit crop growth. It was not known what yield or fibre quality was possible or whether

the crop could be reliably sown and picked within the confines of the dry season.

Over three seasons two Gossypium hirsutum (upland) Bt transgenic cultivars and one Gossypium

barbadense cultivar were sown from March to June in field experiments at the Ord River (15.5oS)

in Western Australia. A pot experiment was conducted at Katherine, Northern Territory (14.5oS)

where biotic stresses were removed and over two seasons: ambient and ambient plus 5 to 6 oC night

thermal conditions were imposed from 1 wk prior to first flower to 2 wk after last effective flower.

Day temperatures were the same. Average ambient minimum temperature for the treatment period

was 2 to 4 oC less than the Ord River at the same growth stage. The OZCOT cotton simulation

model was validated then applied to simulate gross margin, yield and quality, with enhancements to

predict fibre length and colour grade developed in this thesis, at the Ord River using 53 years of

historic climatic records for sowing dates from March 1st to May 25th.

Experimental results at the Ord found for the upland cultivars, the highest lint yields of 1900 to

2300 kg/ha were for March and April sowings and were at the high end of Australian and

international benchmarks. The lint yield of the Gossypium barbadense cultivar was highest at a

March sowing, at least 87% of the upland cultivars, which is comparable with temperate climates.

For the March and April sowings both the lower temperatures and radiation during early boll

growth reduced the crop growth rate during this phase compared with cotton grown at temperate

latitudes. However, assimilate supply was adequate because boll demand was also lower due to

early flowers having slower boll development, lower retention and smaller bolls. Increasing late

season temperature and radiation permitted yield compensation via an extended flowering period

and a greater contribution to yield from later pollinated flowers on the top and outside of the plant.

The number of temperatures >35oC and or <11oC affected time to squaring, requiring modification

of development models derived in temperate climates. Radiation use efficiency (RUE) was similar to

Gossypium hirsutum grown at temperate latitudes. The RUE measured for the Gossypium barbadense cultivar was the first reported for this species. The linear decline of RUE with average

temperature up to first flower has not been reported previously in cotton and explains some of

variation in RUE measured here and elsewhere. Due to cool temperatures during fibre development

fibre length and strength at March and April sowings were low to marginal compared with market

preference values. The cultivar differences observed here suggest wider screening may identify

upland cultivars with suitable fibre length and strength in these conditions. The commercial

prospects for Gossypium barbadense are doubtful unless longer and stronger fibre types are identified.

The pot experiment confirmed that flowers were damaged by low ambient minimum temperatures

near anthesis which led to shedding or reduced boll size due to lower seed number. The latter could

be due to poor pollination and competition for assimilates. Importantly this experiment

demonstrated that full yield recovery from minimums <11oC during flowering and boll growth is

possible provided they are episodic.

The OZCOT cotton simulation model was validated for lint yield and average time-to-maturity in

response to sowing date and N fertiliser rate. Further research was required to reduce the

variability of maturity predictions. There was only a 14 day sowing period from March 19th where

the simulated gross margin (GM) was maximised at $2378/ha. Poor trafficability combined with the

Bt resistance sowing window, reduced the number of sowing days in the optimum period. Hence to

reliably sow a commercial area it was likely sowing would extend beyond April 3 and reduce

median GM by 9-15% due to lower fibre quality. Future research should apply this type of analysis

throughout the Australian SAT.

Cotton management in the dry season should aim to increase the flowering period, to ensure yield

compensation from later flowers and to adopt practices that can improve trafficability within the

optimum window e.g. minimum tillage.