INTRODUCTION
Cotton (Gossypium hirsutum L.) is a source of income for more than 10 million people in West and Central Africa (Baffes and Bank, Reference Baffes2004) where the cotton sector is or was supervised by development companies such as Sodecoton in Cameroon. There, cotton is grown under rainfed conditions (Sultan et al., Reference Sultan, Bella-Medjo, Berg, Quirion and Janicot2010), hence the tremendous importance of water availability during crop cycle on final production (M'Biandoun and Olina, Reference M'Biandoun and Olina2006). Sodecoton supplies seeds, fertilisers and pesticides to farmers, holds a monopoly in buying seed cotton from farmers, ginning it and selling fibre mainly on the international market. Whereas farmers aim at producing a lot of seed cotton per unit land area to increase their incomes, Sodecoton is interested in maximising the production of high-quality fibre for securing outlets on the international market. In this context, the IRCT (Institut de Recherches du Coton et des Textiles Exotiques) initiated a breeding program in Cameroon in 1950 with the objectives of improving fibre yield, resistance to pests and diseases and fibre quality (Levrat, Reference Levrat2010). For 60 years, this breeding program has been carried out considering the specifications of Sodecoton, and then breeding several cultivars among which the Sodecoton picked the ones to be released to farmers. By 2009, more than 20 cultivars were released. Despite the breeding efforts, Naudin et al. (Reference Naudin, Gozé, Balarabe, Giller and Scopel2010) showed that on-farm seed cotton yield has been decreasing steadily since the 80s in Cameroon. This could be explained either by agronomic factors (soil fertility, rainfall pattern, cultural practices) or by the performance of genetic material.
Past studies have evaluated genetic improvement of yield on cotton (Campbell et al., Reference Campbell, Chee, Lubbers, Bowman, Meredith, Johnson and Fraser2011; Schwartz and Smith, Reference Schwartz and Smith2008). In most of them widely grown cultivars were compared in field experiments and linear or break-linear regressions on the year of release (YR) of each cultivar were estimated. In other studies, the interactions between genetic improvement and several aspects of the environment were studied. Significant interactions between genetic improvement and environment were observed on seed cotton yield in the USA (Campbell et al., Reference Campbell, Chee, Lubbers, Bowman, Meredith, Johnson, Fraser, Bridges and Jones2012) and Australia (Liu et al., Reference Liu, Constable, Reid, Stiller and Cullis2013). In Cameroon where cotton is grown in a wide range of environmental and crop management conditions, resulting mainly from various onsets of rains and planting dates, genetic improvement should be assessed under a diverse set of production situations.
To our knowledge, genetic improvement and its interaction with planting date have never been evaluated on seed cotton and fibre yields in Africa. Therefore, this paper aims at:
-
i. estimating the rate of genetic gain (GG) in yield and its components on a set of cotton cultivars released between 1950 and 2009 in Cameroon;
-
ii. determining whether or not the rate of GG is affected by the planting date, main source of environmental variation of cotton crop in Africa.
MATERIALS AND METHODS
Plant material
We chose 10 cotton cultivars widely cultivated and released in Cameroon between 1950 and 2009. Most cultivars grown in Cameroon are derived from Allen Commun and bulk with N'Kourala and foreign cultivars as shown in Table 1.
Table 1. Description of cotton cultivars used in field experiments in Cameroon 2012.
Field conditions
Field trials were conducted in 2012 in two locations in Cameroon: Maroua (Far North Region, GPS: 10.652, 14.410, and altitude: 380 m) and Garoua (North Region, GPS: 9.246, 13.471, and altitude: 250 m). In Maroua, a randomised complete block design was used with one factor (cultivar) and three replicates; in Garoua a split-plot design was used with two factors (main plot: planting date, subplot: cultivar) and three replicates. In total, there were four environments defined by the location, a planting date and its corresponding level of fertilisation according to the recommendations of Sodecoton. In Maroua, there was one late planting date. In Garoua, the earliest planting date was G0, delayed planting date G1 and the latest planting date G2 (Table 2). For all sites, plots were 32 m², spacing 0.8 × 0.4 m (31250 plants ha−1). Weed and pest control was maximised. Weather including solar radiation was recorded daily with a synoptic weather station within 10 km from each site, while rainfall was recorded on each site (Figure 1). In Maroua, the rainy season accumulated 671 mm of rainfall from planting to harvest with 84 days available to the crop from planting to last rain > 10 mm. In Garoua, it was 1116 mm with 121 days in G0, 1063 mm with 108 days in G1, 921 mm with 94 days in G2. In Maroua and in G2, the crop met the end of the rainy season before the 94 days after emergence necessary to achieve its physiological maturity (Gérardeaux et al., Reference Gérardeaux, Sultan, Palaï, Guiziou, Oettli and Naudin2013).
Table 2. Description of field experiments conditions in Maroua and Garoua, Cameroon, 2012.
Figure 1. Cumulative rainfall amount in field experiments in Maroua (M.) and Garoua (G0,G1, G2), Cameroon, 2012.
Plant measurements
Each growth stage was scored as soon as 50% of the plants in the observed row reached emergence, anthesis or first open boll stage. Cycle duration was measured in growing degree days with a base temperature of 13 °C (GDD). In each plot, three plants were randomly chosen for destructive samplings at 35, 65 and 120 days after planting (DAP). Aerial dry biomass per m² was assessed from dry biomass per plant and actual plant density. Seed cotton yield was measured on two central rows on an area of 12.8 m² per experimental plot. Fibre yield was determined by multiplying cotton yield with the ginning out-turn. Harvest index was determined as seed cotton yield divided by total aerial dry biomass at 120 DAP. In all plots, LAI dynamics was assessed with a LICOR LAI-2200 (LI-COR, USA) during crop cycle. LAI at maximum vegetative stage was measured for Maroua, G0 and G1.
Determination of global radiation use efficiency
The global RUE was defined as the ratio of accumulated aerial dry biomass to intercepted photosynthetically active radiation over the same period (Monteith and Moss, Reference Monteith and Moss1977). The coefficient of light extinction was set at 0.69 (Brodrick et al., Reference Brodrick, Bange, Milroy and Hammer2013) and the maximum percentage of light interception at 95%. Global RUE was calculated between 35 and 95 DAP.
Statistical analyses
Analyses were carried out with SAS 9.3 proc MIXED [SAS Institute Inc., Cary, NC, USA]. For each variable, a first step analysis suitable for split plot and randomised complete block designs was performed in order to estimate adjusted means; then, a second step aimed at estimating the rate of GG, i.e. the slope of a linear regression of the adjusted mean on the YR of the cultivars. Where several planting dates were compared, contrast estimation was also used to test for any difference in rate of GG observed under these different planting dates. Maroua was the least favourable condition, as the late onset and unusually early stop of the rainy season left only 84 days of rainy season for the crop cycle. However, to test the effect of planting date, Maroua was considered separately from Garoua, as conditions other than planting date differed between the two locations (e.g. rainfall pattern, temperatures, global radiation, soil type). The YR was centered at 1980, so that the intercept of the regression is the predicted performance of any cultivar released in 1980.
RESULTS
For all variables, no significant difference in the rate of GG was found between planting dates in Garoua (results not shown). Consequently, the rate of GG in Garoua was considered to be the same irrespective of the planting date. In Maroua and Garoua, there was no significant GG on the duration from emergence to anthesis (Table 3). In Maroua, the duration from emergence to first open boll significantly increased (1.05 GDD year−1) but not in Garoua. In Maroua and Garoua, there were neither significant GG on LAI at any time measured nor on RUE. Similarly, there was neither a significant genetic improvement of the maximum biomass measured at 120 DAP, nor of the harvest index at maturity. However, the fibre yield was significantly increased in Garoua with a yearly rate of 3.3 kg ha−1, but not in Maroua (p = 0.070). The ginning out-turn was significantly increased in Maroua (6.2% in 60 years), and Garoua (3.9% in 60 years). On the contrary, there was no significant GG on boll number per square meter, average boll weight and seed cotton yield in all field conditions.
Table 3. Yearly rate of genetic gain (GG) and prediction at year 1980 in field experiments conducted in Cameroon in 2012.
*α = 0.05; **α = 0.01; ***α = 0.001; ns = Not significant.
†Slope of the linear regression on the year of release.
‡Intercept of the linear regression on year or release centered at 1980, i.e. predicted value for an average cultivar released in 1980. Always significant at the 0.001 probability level.
§Standards errors.
¶GDD: growing degree days with a base temperature of 13 °C.
DISCUSSION
Genetic improvement
There are two possible explanations for the absence of significant GG on seed cotton yield. First, with little or no change at all in cycle duration, no significant GG on LAI and on RUE, no GG on the biomass production was observed in any location. This absence of GG on biomass production was coupled with no significant GG on harvest index. Secondly, no significant GG on the number of bolls produced and on the average weight of each boll were observed. Our results confirmed that breeders aimed at increasing fibre yield, while Sodecoton aimed at increasing fibre yield and the ginning out-turn in their choice of cultivar to release. However, breeding for increased ginning out-turns may lead to higher seed-coat fragments content in lint due to increased brittleness of the seed coat (Bowman, Reference Bowman1996).
The seed cotton yields obtained in this study in Garoua under favourable conditions (G0, G1, Table 3) were higher than 1460 kg ha−1 obtained by Naudin et al., (Reference Naudin, Gozé, Balarabe, Giller and Scopel2010) on average in farmers’ fields. However, the results in late planting conditions (G2 and Maroua) were similar to theirs.
We found an increase of 3.3 kg ha−1 year−1 (0.38 % year−1) in fibre yield (Table 3). In the USA, Schwartz and Smith (Reference Schwartz and Smith2008) observed a rate of GG on fibre yield of 8.7 kg ha−1 year−1 (0.93 % year−1). In Australia, Liu et al. (Reference Liu, Constable, Reid, Stiller and Cullis2013) observed a rate of fibre yield GG of 12.3 kg ha−1 year−1 (0.67 % year−1). Also in Australia, Rochester and Constable (Reference Rochester and Constable2015) observed a rate of GG of 28.8 kg ha−1 year−1 (1.2 % year−1) on fibre yield and of 0.14 % year−1 on ginning out-turn. However, these authors found no change in harvest index despite a joint increase in aerial biomass and seed cotton yield. Our rate of GG is very low as compared to that obtained in the USA and in Australia. This is probably because cotton breeding in Cameroon is targeting less favourable growing conditions (poor soil fertility, harsh climate, no irrigation). From our results, we concluded to no improvement of seed cotton yield in Cameroon. However, under similar conditions, breeding has increased yield of other important crops such as corn in Nigeria (Badu-Apraku et al., Reference Badu-Apraku, Oyekunle, Menkir, Obeng-Antwi, Yallou, Usman and Alidu2013) and in Kenya (Beyene et al., Reference Beyene, Semagn, Mugo, Tarekegne, Babu, Meisel, Sehabiague, Makumbi, Magorokosho, Oikeh, Gakunga, Vargas, Olsen, Prasanna, Banziger and Crossa2015), and cowpea in Nigeria (Kamara et al., Reference Kamara, Tefera, Ewansiha, Ajeigbe, Okechukwu, Boukar and Omoigui2011). A GG in seed cotton yield is probably attainable if targeted in priority by the cotton sector.
Non-genetic limiting factors
The decreasing seed cotton yield that has been observed from the 1980s in Cameroon's farms (Naudin et al., Reference Naudin, Gozé, Balarabe, Giller and Scopel2010) was very likely due to agronomic factors, as we did not find a clear genetic explanation for this negative trend (Table 3). This yield decline in Cameroon could be attributed to limiting factors such as low soil fertility, low water availability during the crop cycle (M'Biandoun and Olina, Reference M'Biandoun and Olina2006), lack of training for new farmers (Cao et al., Reference Cao, Oumarou, Gawrysiak, Klassou and Hau2011), use of fertilisers below recommended rates, cultivation of infertile plots and to late planting dates (Cao et al., Reference Cao, Oumarou, Gawrysiak, Klassou and Hau2011) explained by priority to food crops and late onset of rainy season.
Perspectives
An optimisation of examined cotton breeding strategy seems possible for seed cotton yield improvement with a major shift in the ranking of breeding criteria. For example, breeding for more boll numbers should efficiently increase seed cotton yield since the boll number and the seed cotton yield are highly positively correlated in water-limited conditions (Rahman et al., Reference Rahman, Ullah, Ahsraf, Stewart and Zafar2008). In addition, some physiological variables related to crop yield in water-limited conditions were successful for yield improvement: stability of metabolic response on peanut (Singh et al., Reference Singh, Collakova, Isleib, Welbaum, Tallury and Balota2014), greater global RUE and high leaf assimilation rate on cereals (Fischer and Edmeades, Reference Fischer and Edmeades2010), high Δ13C on emmer wheat (Konvalina et al., Reference Konvalina, Capouchova and Stehno2012), and leaf enhanced RuBisCo activity on cotton (Plaut and Federman, Reference Plaut and Federman1991). These additional physiological traits indirectly linked to high yields in drought conditions should be targeted as early as in generation F5 when there are still many different lines to test and already a population of plants for evaluating yield.
CONCLUSION
In conclusion, the breeding program in Cameroon has improved cotton fibre yield by 200 kg ha−1 in 60 years entirely due to increased ginning out-turn. Since growers are paid based on their seed cotton production, it seems that it is mainly the development company that has reaped the benefits of these breeding efforts. In order to increase seed cotton yield for the farmers, breeding efforts should target cultivars with more boll numbers or include physiological measurements such as those relative to photosynthesis capacity. For example, the highest leaf chlorophyll content and the lowest specific leaf area could easily be targeted in campo.
Acknowledgements
This study was funded by an agreement between Sodecoton, IRAD (Institut de Recherche Agricole pour le Développement) and CIRAD (Centre de Coopération Internationale en Recherche Agronomique pour le Développement). The authors thank Ousmanou Kanti, and Adama Mana for their major contribution to data collection.
