DNA fingerprinting of plant material from farmers fields:What have we learned? James Stevenson
1. DNA fingerprinting of plant material
from farmers fields: What have we learned?
James Stevenson (ISPC Secretariat)
work with Talip Kilic, John Ilukor, Mywish Maredia, and many others...
2. “As has often been remarked, probably no two individuals are
identically the same… when the eye is well practiced, the
shepherd knows each sheep, and man can distinguish a
fellowman out of millions and millions of other men”
Charles Darwin, 1868,
The variation in animals and plants under domestication, vol. 1, p. 361
3. “As has often been remarked, probably no two individuals are
identically the same… when the eye is well practiced, the
shepherd knows each sheep, and man can distinguish a
fellowman out of millions and millions of other men”
Charles Darwin, 1868,
The variation in animals and plants under domestication, vol. 1, p. 361
• Darwin may have been considerably less optimistic about our
ability to distinguish variation within a population had he
studied maize in farmers’ fields in sub-Saharan Africa…
4. Content
• Adoption of improved varieties: How do we normally get these
data?
• DNA fingerprinting: Overview of field implications of methodology
• Methods and results from 8 new empirical field-based studies
• Implications for CGIAR
• For study of adoption and impacts of new varieties
• For genebanks?
5. • Adoption of improved varieties: How do we normally get these
data?
• DNA fingerprinting: Overview of field implications of methodology
• Methods and results from 8 new empirical field-based studies
• Implications for CGIAR
• For study of adoption and impacts of new varieties
• For genebanks?
Content
6. Varietal identification used to be easy
Early period of the Green Revolution underwritten by a huge turnover
of genetic material in farmers’ fields.
Semi-dwarf varieties spread rapidly through the irrigated wheat and
rice production systems of a number of Asian countries
Adoption in this case represented a very significant shift: the
improved varieties were immediately noticeable to the naked eye -
they looked different
8. Varietal identification used to be easy
Reliable data on adoption of improved varieties has long been
recognized as the cornerstone of any assessment of the impact of
investments in plant breeding (Dalrymple, 1978; Walker and Crissman,
1996; Evenson and Gollin, 2003; CGIAR Science Council, 2008; Walker
and Alwang, 2015)
In the decades since the Green Revolution, the focus of breeding has
diversified significantly in two dimensions:
• diversification across crops
• diversification of the targets for breeding
Poses deep challenge to the process of understanding the adoption of
new varieties in farmers’ fields
9. Current practice in adoption studies
Methodology has traditionally fallen into one of two categories:
1) “Expert opinion” elicitation in focus groups
+ Bulk of the literature has relied on this method
+ Only feasible way of getting large coverage across crops and
countries where varieties have been released (e.g. SPIA is heavily
invested in this method):
DIIVA (2010 – 2012) 115 crop-country combinations in Africa
SIAC (2013 – 2017) 127 crop-country combinations in S, SE, E Asia
- Unable to rigorously link these estimates to development
outcomes
- Obvious concerns about accuracy and possible bias
10. Current practice in adoption studies
2) Household surveys with farmers self-reporting their varieties
+ Potential to sample in a representative manner
+ Combination of adoption status with other variables in survey
allows for impact evaluation
- Unclear to what extent farmers are able to correctly identify their
own varieties
- Very poor track record of setting up surveys to be:
• statistically representative at policy-relevant scale
• revisited in a panel
• provide public goods beyond “the project”
12. Content
• Adoption of improved varieties: How do we normally get
these data?
• DNA fingerprinting: Overview of field implications of
methodology
• Methods and results from 8 new empirical field-based
studies
• Implications for CGIAR
• For study of adoption and impacts of new varieties
• For genebanks?
13. DNA fingerprinting - steps
1: Sample of
plant material
taken: leaf or
grain
2: Each sample
placed in own
kit with
desiccant (for
leaf) and unique
identifier (bar
code or ID
number)
3: Laboratory
in-country
extracts DNA
(if using grain,
each sample is
first dried and
then ground)
4: DNA from
each sample
placed in 96-
well plates and
shipped for
genotyping
5: Each sample
is compared at
multiple alleles
in the genome
against a
reference
library of
varieties
14. Some (tough) lessons learned
• All this is new to survey enumeration teams
• Threats to samples (to mould; weevil attack) need
vigilance and clear protocol
• Data “cold chain” on the samples essential for
linking to other data (e.g. in a CAPI questionnaire)
• Training… field practice…. (revise field protocol if
needed)…. training… more practice… fieldwork starts
• Having sufficiently high density of genotyping assay
is essential: need to be smart consumers of
commercial lab services
15. Content
• Adoption of improved varieties: How do we normally get
these data?
• DNA fingerprinting: Overview of field implications of
methodology
• Methods and results from 8 new empirical field-based
studies
• Implications for CGIAR
• For study of adoption and impacts of new varieties
• For genebanks?
16. Survey-based methods
Crop Sample Sample
drawn from
DNA
from
A: Expert
opinion
B: Ask is this
improved?
C: Ask for
name
D:
Phenotypic
protocol
Maize,
Uganda
550 2 districts;
random
Grain Yes Yes Yes Yes
Sweet potato,
Ethiopia
231 Wolayita;
snowball
Leaf Yes Yes Yes
Cassava,
Malawi
1,200 National;
random
Leaf Yes Yes Yes
Beans, Zambia 855 2 provinces;
random
Seed Yes Yes
Cassava,
Nigeria
2,500 National;
random
Leaf Yes
Cassava,
Vietnam
1570 National;
random
Leaf Yes Yes Yes
Rice,
Indonesia
798 Lampung
province;
random
Seed Yes Yes
Cassava,
Ghana
914 3 regions;
random
Leaf Yes Yes Yes
18. False positives, false negatives
• Aggregate breakdown of adopters vs non-adopters of any
improved variety is only relevant for limited number of
questions
• Typically want to link adoption status (“treatment”) to some
dependent variable of interest (e.g. productivity; HH income)
conditioned by a bunch of covariates
Farmer states IV Farmers states local
/ traditional
Fingerprint = IV Correct positive False negative
Fingerprint = not IV False positive Correct negative
19. False negatives, false positives
% false negatives % false positives
Maize, Uganda 43 0
Sweet potato, Ethiopia 20 30
Cassava, Malawi 21 0.2
Cassava, Nigeria 28 13
• No consistent deflation or inflation possible across all cases:
context-specificity
• If there are more farmers growing improved varieties that
don’t realize it, why aren’t yields higher?
• Noise? Or bias?
20. Genotype Farmer-stated
Maize in Uganda: SPIA / LSMS-
ISA / UBoS / Diversity Arrays
• Data from 540 HHs in 45
enumeration areas
• Enumerators from UBoS
trained for 1 month
• CAPI-based survey + grain-
based highly-quantitative
DArTSeq genotyping
• 2% of farmers were correct
about the variety they were
growing
22. Share of primary genetic
constituent
Green = >70%
Amber = 60 – 70%
Red = <60%
23.
24. Competing theories
• Deliberate mixing by farmers
Farmers skillfully use seeds of different varieties together in the
same plot
• Counterfeiting / deception
Growing literature on input quality as a drag on agricultural
development (e.g. Bold et al, 2015)
• General chaos / informality of seed system
Things start out fine from seed companies but errors accumulate in
different stages in seed supply chain
2016/17 extension: Testing these theories for maize in Uganda
Second visit to the farms in the sample; mystery shopping from agro-
dealers; and testing samples from seed companies / NARS
25. Content
• Adoption of improved varieties: How do we normally get
these data?
• DNA fingerprinting: Overview of field implications of
methodology
• Methods and results from 8 new empirical field-based
studies
• Implications for CGIAR
• For study of adoption and impacts of new varieties
• For genebanks?