Friday, July 15, 2011

As Bt Cotton turns 10, observational data certifies it a Super-Flop: Part II

Is Bt either a necessary or sufficient condition for cotton productivity increases?
Click for Part I Part III

The average productivity for the cotton crop for the period 2004-10 is 507 kgs per hectare, up 20% than the base year viz. 2003-04 as seen in the graph to the left.  Annualized productivity for eight years since Bt’s commercialization is around 4%, higher than seen in crops like rice or wheat.  More importantly, since the introduction of Bt, from a net importer of cotton, India catapulted to a net exporter of cotton as illustrated by the graph in the right. The Bt lobby in this country use both these graphs to claim Bt a runaway success as a technology. 
In 2003, David Zilberman and Matin Qaim, published a paper "Yield Effects of Genetically Modified Crops in Developing Countries" (Science, Feb 7, 2003, Vol. 299) that claimed that Bt introduction will increase cotton yields over 80 percent and went further to the extent of extrapolating this "phenomenal growth in production" to cover the entire range of GM crops for the developing countries. This paper whipped up  a lot of passion with NGOs and environmentalists then denouncing the paper as fraudulent. With Bt 10 years old in India, NGOs and environmentalists should feel vindicated for their rage against the paper for over-hyping Bt’s potential yield.
But a 20% decadal jump in productivity is still 20% jump, which is whopping in agriculture by any yardstick. But is the Bt gene either a necessary or a sufficient explanation for increased in cotton productivity rates as observed during the last decade? The Indian Central Institute for Cotton Research Director Dr Keshav Kranthi in his recent article in Cotton24/7 admitted that that Bt cotton technology may not have been singularly responsible for the dramatic improvement of cotton fortunes in India. The following discussions provide some of the alternate explanations.

1. Bt piggybacked on an already rising yield curve

In 1951-52, cotton productivity was a mere 92 kg/lint per hectare but increased to 400 kgs/lint per hectare by 2003-04. So even before the introduction of Bt, the underlying long-term productivity trendline was fairly buoyant (See above graph).  This was fuelled by factors such as expansion of hybridization in cultivation in the country; improvement of management practices; better quality inputs; expansion of irrigation etc. Since Bt genes are backcrossed with local elite hybrids, the introduction of Bt Cotton varieties only gave further impetus to hybrid adoption within the country.    
Can non-Bt hybrids bring about a quantum productivity jump? We were told they can’t and that was the reason for introducing the Bt variety. So let’s have a look at observational data.
(Please note that in 2003-04 Bt’s share in area of cultivation was 1.1% - a share that reflects lack of critical mass to impact the overall productivity of the country and so treated as such by this paper. Besides yield potential of Bt hybrids at the time of introduction was relatively lower than as now. Desegregated yield data Bt and non-Bt validates this treatment by this paper).

So what does the data tell us? YoY 2002-03 to 2003-04 the productivity jump was a mind boggling 34%. Bear in mind this is an overwhelming performance by non-Bt hybrids. And further this happens to be an all-time YoY record jump - a performance a Bt inserted hybrid has not matched to date. The very next year, (when Bt’s share is only 5.7% and still lacking a critical mass), the increase in productivity was 18%. So even before Bt attained a critical mass to significantly impact overall productivity rates of cotton in this country, productivity rates spiked 52% in just two years!
So do we really need to insert a Bt gene into our hybrids or better-off without it?
Bennett et al (2005) who analyzed the effect of varietal differences showed that the host germplasm played a very significant role in yield increases. A NGO, Gene Campaign conducted an experiment that compared Bt cotton varieties Bt. 162 and Bt. 184 belonging to Mahyco-Monsanto and local hybrids Brahma and Banny and found yields of the latter 15-17% higher than the Bt varieties compared.

The experiment proved that the Bt gene is neither a necessary nor sufficient condition for productivity increases in the cotton crop. It is equally possible that productivity increase observed during the last decade had been entirely brought about by the spread of hybrids with little or no contribution by the inserted Bt gene.
Has India this potential to increase productivity via hybrids? Why not? We were the first to develop a cotton hybrid and still the leaders of the world in the area of cotton hybrid development.

2. Poor statistical correlation between Bt and productivity 

In 2004-05 the Bt share in area cultivated stood at mere 5.7%. Contrast this to in 2009-10 where it was a whopping 85%. Yet in both years productivity rate stood the same - within the 470-475 kgs/lint per hectare band! So where is the much touted correlation that links Bt expansion in the country with productivity gains? (See table above).

Remember the underlying long-term productivity trend? Well we seem to be losing at least some of its underlying buoyancy after Bt’s introduction.  Even worse, for the last 3 years of the last decade witnessed cotton productivity performing well below its short term trendline.

3. Huge productivity variations calls the bluff of the Bt gene magic

Bt has been hard sold as a sort of a magic remedy call it panacea or silver bullet for all problems confronting cotton cultivation in the country. 8 years into its commercialization, its bluff has been called.

Overall productivity numbers mask wide heterogeneity in its distribution pattern. Apart from huge inter regional variances, even intra- regional data shows that Bt cotton did not have had the same marginal effects in each state.The Bt lobby finds difficulty in explaining why for example cotton yields in Gujarat surged from below the national average to presently occupy the position as leaders in the country in terms of cotton productivity while yield rates in Madhya Pradesh have in fact significantly decreased post Bt introduction? 

Or why for that matter Maharashtra with around 40 lakh hectares under cotton cultivation manages a productivity rate of only 351 kgs/lint per hectare while Gujarat with 26 lakh hectares under cotton cultivated possess the highest in the country? Gujarat’s productivity has often been within reach of the average, international productivity rate (725 kgs/lint per hectare).

So in the central region that comprises of these three states and accounting for nearly 60% of the country’s cotton production, Bt performance is characterized by a high degree of unevenness. In more adverse conditions, the so-called Bt signature in productivity increase is either undetectable (Maharashtra) or negatively correlated with yields (MP). 

So among the top three cotton states that account for the great majority of the country’s production, it is only Gujarat where Bt can theoretically claim success.  However even here, as the discussions in the next section illustrates, there are other conditions including Gujarat specific conditions that offer more convincing alternate explanations.

4. Alternate Explanations for Productivity Increases

a. Seed Quality:
Poor seed quality is a pervasive problem in cotton cultivation. Only about 35 percent of cotton area is estimated to be sown with certified seed with assured varietal purity and germination before the entry of Bt varieties. Commercially available seeds are often of poor quality, with sale of uncertified, substandard, and second generation (F2) hybrid seeds not uncommon. Although supplies of certified seed are generally available, financial constraints lead most farmers to use retained seeds or opt for lower priced uncertified seeds from the market. Consequently, poor seed quality emerged a significant constraining factor for realization of the full yield potential of hybridization of cotton cultivation. 

Now the Bt gene is backcrossed with local varieties. But why then are desi (traditional) varieties excluded?  It has low productivity and is not responsive to agronomic practices in terms of yield. So they are instead backcrossed with our elite local hybrids. Why? Their responsiveness to external inputs is much higher. 

Now hybrid seeds yield can be as high as 800 kg/lint per hectare (in the US it goes up to 1,200 kg/lint per hectare), under the right agronomic conditions and practices. So the introduction of Bt Cotton by default effect fastened the pace of hybridization of cotton cultivation in the country. Farmers began to get better access to quality and unadulterated seed materials that reflected in higher yields.

To give the devil its due, Bt entry catalysed the improvement of cotton seed in the country in relative terms which in turn reflected itself as higher yields. That said there are as many as 700 brands of Bt Cotton in the market whose performance in the field and regions vary markedly. Performance depends into which elite germplasm the Bt gene is inserted to as they could vary widely in yield potential and agro-climatic adoption traits. The real problem is that the presence of so many Bt brands reduces the capacity of the Indian state to regulate Bt in the country.  

b. Seed treatment:
Dr Keshva Kranthi pointed out that the chemical imidacloprid, popularly known as Gaucho, used as protection against leaf hopper, had been applied as a seed treatment. And since 2000 (two years before Bt introduction) every cotton seed had been so treated. Now imidacloprid is observed to generate 20-30% yield increases in the crop according to the Indian Central Cotton Research Institute.  Wait a minute. What's the growth rate seen during the last decade? 20-30%. What's the contribution of Gaucho? 20-30%? So what exactly has Bt gene contributed?

Gaucho's main benefits are as follows: 
  • Control of jassids, aphids and thrips up to at least 45 days after sowing. 
  • No dependence on weather conditions which may prevent or delay spray operations. 
  • Improved emergence of plants due to control of soil pests. 
Since Gaucho was introduced in 2000, we have comparative observational productivity data for non-Bt hybrids and those further treated with imidacloprid to tweeze out the latter’s specific contribution to yield growth. This was presumably the procedure used by Indian Central Cotton Research Institute to conclude that seed treatment through imidacloprid application can contribute upto 20-30% yield increases in cotton.
But in the case of Bt crops, all productivity increases are equated to Bt introduction as if no other factors may have played a contributory part. This helps Bt claim runaway success even if it was to be assumed to play a negative role in respect to productivity increases. That’s the power of the Bt lobby in this country that even research appears subservient to promotion of its agenda.

c. Irrigation:
Cotton is highly responsive to water. But in India, most of the cotton is cultivated under rain-fed condition (Sundaram et. all, 1999). Thus the variability in yields is largely dependent on the monsoon.

This is why most Bt cotton seed packets carry the information - “Best grown in irrigated conditions” a statutory warning, but provided in very small fine print. Union Minister of State for Agriculture and Food KV Thomas replying to a written query in parliament: “In all other countries cotton is an irrigated crop whereas in India hardly 35-40% of the cotton is irrigated. Therefore, the scope of adopting balanced nutrition is limited.” 

In simple terms what the Minister was tacitly admitting that not only the yield potential was lower but the pest infestation potential was higher under dryland cultivation as compared to irrigated conditions.

Though India's decadal average productivity of 502 kgs/lint per hectare may look  abysmally low as compared to the international average of 725 kg/lint per hectare or China's 1260 kg/lint per hectare or even the US's 868 kg/lint per hectare, much of this difference can be explained away by lack of irrigation.

The difference between China and US in productivity rates only reiterate this fact- the relative share of dryland in overall cotton cultivation of the US being considerably higher than China though much lower than India.  If productivity data is accordingly desegregated to reflect this reality, what is found is that both dryland and irrigated cotton productivity of India are not too far away from their respective international productivity averages.

Dr.Keshva Kranti observed: “Bt-cotton hybrids utilize more nutrients and water for higher yields and profits, therefore the soils are getting progressively depleted and need more nutrient recharging.”  Minister PV Thomas commented in the same vein: “The scope of adopting balanced nutrition is limited”  Both were obliquely admitting of Bt’s inappropriateness to dryland conditions that characterise much of our cotton cultivation.  Without water as an input, the degree of freedom is reduced in the choice of soil nutrition solutions, whether organic or inorganic.

Both in Maharashtra and Gujarat most of their cultivation is accounted by dryland. They however offer a striking contrast as case studies. In Maharashtra 97% of 4 million hectares of cotton is grown under rainfed conditions with 95% of its cultivation under Bt varieties. Yields doubled in Maharashtra, from 194 kg/lint per hectare in 2002 to 394 kg/lint per hectare in 2010. Compare this with Gujarat which manages a productivity rate around 700 kgs/lint per hectare.

How can we explain this? Dr Kranti gives us a clue to solving the riddle by revealing that maximum productivity gains were obtained from just  0.6 million to 0.7 million hectares in Gujarat whose land quality was enhanced by a watershed development programme that conferred this swathe of land the benefits of more than 100,000 check dams.

What we can take away from this learning is that we don’t need big irrigation or dam projects or even the Bt gene. Just simple micro-watershed projects are sufficed to bring about quantum jumps in productivity of cotton. Increasing the soil moisture content can make all the difference in how crops adopt to abiotic stresses like droughts. We don’t even need to search every nook and corner of the world for solutions but in Gujarat we have a ready model for replication.

We now know the reason why at least one cotton farmer in Vidarbha daily commits suicide as painfully documented by renowned journalist P Sainath and why Gujarat farmers are enjoying the fruits of their newly found prosperity by going on a consumer spending spree. Now Dr Keshva Kranthi also took a dig on NGOs and environmentalists for blowing up Vidharbha suicides by highlighting that Bt cotton has spread almost 94% of all area cultivation and that productivity rates have more than doubled in Vidarbha. 

Well apart from stating the obvious what he conceals is that Bt Cotton is a highly input intensive crop and that the average productivity of 394 kgs/lint per hectare of the state simply does not offer the same amount of cushion to economically breakeven as compared to Gujarat’s over 700 kgs/lint per hectare, as per capita per day returns show wide variations between these two states. (Please note the productivity rate of Vidarbha is much below Maharashtra’s average). 

Without irrigation, Vidharbha farmers face much more higher risks from abiotic stress and therefore more prone to suicides than a Gujarati farmer. The result was that in an area with a history of indebtedness, the high input costs of Bt cotton acted to further increase indebtedness. A study by Gene Campaign had shown that 70 per cent of small farmers had already lost their landholdings as collateral for loans that they could never repay. 

Suman Sahai, Director of Gene Campaign was quoted saying “that despite specific knowledge that Bt cotton would not work in rainfed areas, the government had introduced it in Vidarbha.” She further observed that irrigated farmers performed better in Vidarbha and suicide rates of the latter were much lower than dryland farmers. It follows that if we want to attain quantum jumps in cotn yields, irrigation rather than Bt would be a better and safer bet.

d. Quality land and agronomic conditions:
These are key factors in bringing about quantum jump in productivity rates. In the 0.6 million to 0.7 million hectares in Gujarat where cotton experienced a phenomenal jump in productivity, according to Dr Keshav Kranti farmers previously  grew groundnut crops, which being a legume enhances soil fertility. Cotton on the other hand depletes the soil by decreasing its nutrient content.  Cotton can of course grow on even mediocre soils but throw some soil fertility to it and its yields spike. This held true in the case of Gujarat. The land was fertile to begin which that made it all the more conducive for higher yields.

Further, since cotton was not previously cultivated in these lands, cotton specific pest pressure is low. But unless crop rotation is practice, Gujarat farmers too will experience progressive increase of pest pressure if they continue to cultivate cotton year after year. 

But compare the Gujarat experience with Vidarbha which has over a 1000 year history of cotton cultivation and where cotton was described all through history as white gold, being the principal economy of the region. It’s a no brainer to conclude pest pressures are higher in this region and consequently the challenges in maintaining and increasing yield rates are much higher in Vidarbha than Gujarat. 

Significantly, where yield increases were needed most (Vidarbha), Bt failed to deliver the goods and where it was perhaps least needed (Gujarat), Bt claims credit for bringing about productivity increases!

e. Resource rich farmers: 

A review of literature finds many studies pointing to Bt’s unsuitability for small-scale and resource-poor farming systems. Early adopters tended to be the most prosperous and well-financed farmers, who were any way getting better yields than other farmers even before Bt seeds were adopted. (Stone).  Bt or no Bt they were more adept in increasing yields. And so part of the explanation for increased productivity need to be credited to the ingenuity of this category of farmers.

In hindsight, NGOs and environmental groups campaign against the exorbitant introductory prices for Bt seeds only succeeded in playing right into the hands of Bt seed manufacturers. Lowering of Bt seed prices can be identified as the single most significant factor that led to their rapid adoption within the country. If prices remained exorbitant, normal hybrids (without Bt gene inserts) would have acted as bulwark against the spread of Bt, confining the latter to few irrigated pockets and more resource rich farms.

But since this had not been the case, the high investment for cotton cultivation induces small and marginal farmers to cut corners by trying to procure cheaper but less effective or even spurious pesticides and seeds that are responsible for much of their lower yields, if not crop failures leading to higher suicide rates.


So back to the question: Is Bt either a necessary or a sufficient explanation for increased in cotton productivity?

We can conclude it is a complete perversion to attribute productivity increases solely to the Bt gene though it may have indirectly contributed by catalyzing a quicker pace of hybridization of cotton within the country and relatively improving supply of quality seeds to farmers. Other factors such as irrigation and newer lands played a much more significant role in increasing productivity of the crop than the Bt gene itself. 

Here ends Part II Click here for Part I and Part III. This paper is written by Rajan Alexander, Development Consultancy Group, Bangalore. Rajan is a livelihood consultant and had been in the NGO sector for more than 31 years.  He can be contacted at


  1. good read. how can we consumer protect us? I do not want to buy a Levis Jeans made out of bt cotton. Anke

  2. Hi Anke. For that you need to read the third part. Since you don't want to buy BT Levi jeans, tailor your own jeans. 5% of all cotton grown in this country is organic. Buy it, and stitch into jeans that you want.

    You do two things - make a statement and also reduce demand for Bt Cotton.

    But there is no evidence that by wearing BT Cotton there is a health risk.

  3. I posted in Part I and because I wanted to comment I started a blog. titled "We Are All Connected"
    How can you be so stupid to not consider that there is a big difference from irrigated farming and crops to dry land crops. My grandfather had a ranch just outside Billings, Montana and I heard him say "dry land crops" and "irrigated crops" and with just a little science education in biology you should know that when you change mother natures equation that she formulated over thousands of years that "it is not nice to fool mother nature" because bad and unintended things will happen.