Are we measuring what we think we are measuring? Results of two pilot studies using DNA fingerprinting for varietal identification

Transcription

1 Are we measuring what we think we are measuring? Results of two pilot studies using DNA fingerprinting for varietal identification Mywish K. Maredia Professor, Dept. of Agricultural, Food and Resource Economics, Michigan State University Research seminar presented at ICRISAT, Hyderabad, India June 11, 2015

2 Collaborators Byron Reyes (Formerly at Michigan State; now at CIAT) For Cassava study: Joe Manu, Ghana Crops Research Institute, and Awere Dankyi, Agriculture Innovations Consult (AIC) Peter Kulakow, Ismail Rabbi, Elizabeth Parkes, Tahirou Abdoulaye and Gezahegn Girma from IITA Ramu Puna (Cornell) For Beans study: Petan Hamazakaza and Kennedy Mui Mui, Zambia Agricultural Research Institute Enid Katungi, Bodo Raatz, Clare Mukankusi and Allan Male, CIAT 2

3 3 Acknowledgement Funding support from the Standing Panel on Impact Assessment (SPIA) of the Independent Science and Partnership Council under the Strengthening Impact Assessment in the CGIAR (SIAC) program Partial funding support from the CRP on Roots and Tubers and PABRA, and technical support from IITA and CIAT ALL the enumerators, field staff and farmers / respondents who participated in the survey and shared their time and knowledge

4 4 Disclaimer All opinions and errors in data interpretation remain with the presenter

5 5 Outline 1. Rationale 2. Methodology 3. Results 4. Discussion

6 1. RATIONALE 6

7 7 Measuring varietal technology adoption: Importance Crop germplasm improvement research remains one of the important investment foci of national and international public research systems Historically, it has been a flagship program of CGIAR as measured by outputs in terms of varietal releases by NARS for major staple crops Tracking the adoption of these outputs and assessing its impact remains important to: Meet the accountability needs of public sector investments in plant breeding research (level and magnitude of adoption); and Learning needs of researchers and stakeholders (who is adopting, not adopting, why? etc.)

8 Methodologies used for measuring varietal adoption: Past and present Secondary sources (published government reports, available data sets) Seed sales and seed multiplication/distribution data Expert opinion / key informant interviews (e.g., Dalrymple 1980s; Evenson and Gollin 1990s; DIIVA and TRIVSA 2010; SIAC 2014) Community level surveys (i.e., focus group discussions) Farmer elicitation: Asking farmers what varieties they planted (name and / or type) and collecting this and supplemental information as part of a household survey 8

9 9 Methodologies used for measuring varietal adoption: Challenges Each method has pros and cons Method Pros Cons Surveys - Farmer Considered most rigorous Resource intensive (time, level Considered more evidence money and skill) based Requires careful planning Provides opportunity to other HH level data for impact analysis Can provide representative data to making inferences about the population - Community Relatively more cost-effective Adoption rate as a level then farmer level survey percentage of area harvested may be more challenging to estimate

10 10 Methodologies used for measuring varietal adoption: Challenges (cont d) Method Pros Cons Seed sales data Expert elicitation Low cost Lends itself to periodic tracking and estimation of varietal adoption Relatively easy to implement Low cost Can be used to periodically to update estimation of varietal adoption Requires the following conditions which may be difficult to observe in most developing countries for staple crops: A mature seed industry A majority of farmers relying on the formal seed system Farmers using fresh seeds on a regular basis Accessing data from private sector may be difficult (information may be proprietary or guarded from competition) Not considered most reliable Estimated values of adoption rates per variety generally have higher confidence intervals Provides perception of relative adoption of different varieties rather than accurate estimates May be prone to personal biases of experts Need to find good 'experts' to obtain the best estimates possible

11 11 Methodologies used for measuring varietal adoption: Challenges (cont d) Usually there is a tradeoff between cost and accuracy; and between cost and frequency Methods considered most credible are costly, and thus conducted less frequently at a representative scale Methods that are less costly can be conducted more frequently but have high confidence interval Farmer elicitation (i.e., HH survey) is currently considered the most rigorous and thus the gold standard for estimating varietal adoption But is it?

12 12 Is all that glitters gold? There are practical challenges of farmer elicitation (through HH survey) method of varietal adoption in a setting where the formal seed system is non-existent or ineffective, and farmers mostly rely on harvested grain as the main source of planting material. Eliciting information from farmers may not provide the most accurate adoption of varietal technology by name or type (i.e., IV vs. local), esp., when: Varieties have local names and are not identifiable on the official list of varietal releases Farmers cannot name or identify the varieties Varieties may be misidentified Farmer may not know the type of variety (i.e., improved vs. local) Seeds may be recycled several seasons and have thus lost their genetic identity (esp. cross-pollinated crops and hybrid varieties)

13 Varietal adoption: Challenges and Opportunities Due to these confounding factors, adoption estimates based on farmer surveys may be biased The direction of this bias and standard errors of existing diffusion estimates remain unknown Other methods can be used to address the shortcomings of farmer elicitation method But they require time intensive data collection such as visiting the field to observe plant characteristics, or collecting sample materials (i.e., photos, seeds/plant tissues) from the farmers for later verification by experts Each of these approaches has implications on the cost of data collection and the accuracy with which they can correctly identify a variety 13

14 DNA fingerprinting as a method of varietal identification in adoption studies DNA fingerprinting offers a reliable method to accurately identify varieties grown by farmers The use of this method can: Increase the accuracy and credibility in the interpretation of results of economic analysis based on household surveys that estimate the causal link between the adoption of improved varieties and the impact on crop productivity and income Serve as a benchmark against which to compare the effectiveness of other potential methods for scaling up 14

15 15 DNA fingerprinting as a method of varietal identification in adoption studies (cont d) However, despite the advantages, DNA fingerprinting has not been used widely for tracking varietal adoption Questions related to sampling, logistics, and costeffectiveness of using this innovative method remains to be explored Under the SIAC project, pilot studies in Ghana and Zambia were conducted to: Explore and understand some of the practical challenges of using DNA fingerprinting as a method of varietal identification as part of a farmer survey To test the effectiveness of different methods of varietal identification against the benchmark of DNA fingerprinting To come up with lessons learned and recommendations on methods / approaches that can be scaled up

16 Crop and country combinations for this pilot study Cassava (Manihot esculenta) in Ghana Beans (Phaseolus vulgaris) in Zambia 16 These were selected based on the: Opportunities presented to piggy back on a planned adoption survey (e.g., beans in Zambia) Desire to cover diverse crops in different countries Interest from national and CGIAR centers to collaborate on this study, which required a multi-disciplinary team of researchers

17 2. METHODOLOGY 17

18 Methods of varietal identification tested in the pilot countries Methods V DNA fingerprinting (used as a benchmark to compare/validate other methods) Ghana (cassava) X Zambia (beans) A Farmer elicitation (name and type of variety) \a X X B C D E F Farmer elicitation based on series of photographs of plants and later identifying varieties based on morphological characteristics Farmer response on type of variety he/she had planted that match seed samples presented by the enumerators Trained enumerators/experts visiting the field and: 1. Recording observations on varietal characteristics (phenotyping); and 2. Identifying the variety based on observation (phenotyping) Taking photos of the plant in the field or seeds harvested by farmers for latter identification by experts (i.e., breeders, agronomists, etc.) Collecting harvested seeds from farmers for latter identification by experts (i.e., breeders, agronomists) \a In Zambia method A also included a small sample of bean vendors in two local markets in Kasama X X X 18 X X X X

19 19 Sampling and data collection Ghana: The pilot study was conducted in three regions which account for 61% of cassava production in the country-- Brong Ahafo, Ashanti and Eastern. A total of 500 households across 100 villages (5 farmers in each village) were targeted for the survey using a multistage cluster sampling method (district, village and farmer samples selected randomly) Survey conducted in October-November 2013 Survey was coordinated by a research team led by the cassava breeder from the Ghana Crops Research Institute and a socio-economist from the Agriculture Innovation Consult

20 20 Sampling and data collection Ghana (cont d) Survey team consisted of Enumerators who were in-charge of completing the household modules Cassava expert in-charge of completing the field survey module DNA sampling expert in-charge of collecting, labeling and storing the plant tissue samples as per the protocol established

21 21 Sampling and data collection Zambia The study was conducted in Muchinga and Northern provinces of Zambia based on the importance of beans (Phaseolus vulgaris) and prior seed dissemination efforts The study was designed to take advantage of an already planned bean varietal adoption and impact study by ZARI with support from PABRA and CIAT A total of seven districts were purposively selected which together represent 59% of the total bean area in Zambia Data were collected from a sample of about 400 farmers across 67 villages (sample size mostly determined based on the available budget) Survey was implemented between August-September 2013

22 22 Sampling and data collection Zambia (cont d) Enumerators were trained by a research team from ZARI, MSU and CIAT on how to: Use the instruments, Take photographs of the seed samples, and Implement the protocol for collecting seeds of each variety the farmer had harvested in the agricultural season and labeling them for proper tracking Each enumerator received a set of seed samples (with codes) representing ten different improved varieties that was presented to the farmer to facilitate in variety identification (method C)

23 Implementation of varietal identification methods based on expert elicitation Methods E and F After the field survey, the seed samples and photographs collected in Zambia, and photographs of plants taken in Ghana were used in varietal identification by a panel of crop experts familiar with the varieties grown in the study area. For beans, the experts included breeders and extension staff from the study districts For cassava, the experts included cassava breeders, experts and field technicians from CRI, IITA and the University of Cape Coast In both the cases, the overall elicitation process was facilitated by a socio-economist either from IITA (in the case of cassava) or from MSU (in the case of beans) who did not participate in varietal identification In Zambia, a consensus name and type of variety (improved, local, mix) was recorded for each sample In Ghana, only type of variety (improved, not improved or mix) was recorded by each individual expert. For analysis, a majority opinion rule was applied to identify each sample by type under Method F 23

24 24 DNA fingerprinting In both the pilots, DNA fingerprinting was used as a benchmark against which alternate approaches were evaluated / validated (method V). This involved first establishing a reference library of DNA fingerprints, and then applying the same or a sub-set of markers used to establish the reference library to genotype the samples (plant tissues or seeds) collected during the farm surveys

25 25 DNA fingerprinting: Sampling method Cassava in Ghana: For each HH, a cassava expert visited ONE cassava field with the largest number of cassava varieties as declared by the farmer For each variety as identified by the farmer, the expert randomly selected ONE representative plant and collected leaf tissues from the youngest (apical) leaves Leaf tissues were collected in a small screw-capped plastic jar with ~20 g dessicated silica gel Samples were also collected from plants that had observed variations in morphological characteristics as assessed by the expert, but were identified by farmers as belonging to the same variety

26 DNA fingerprinting: Sampling method for cassava in Ghana 26

27 27 DNA fingerprinting: Field logistics Cassava in Ghana: DNA extraction and shipment to sequencing laboratory Getting high quality DNA for GBS is a challenge, especially when samples are collected from farmers fields, hundreds of miles from extraction labs A versatile and economical sampling kit was developed Small screw-capped plastic jar with desiccated silica gel Collected samples were shipped to IITA (Ibadan) where a low-cost and high-throughput DNA extraction system was used to isolate DNA from > 1000 samples in less than one week DNA was freeze-dried and shipped to Cornell for GBS

28 DNA fingerprinting: Methodology and Data Analytics Cassava in Ghana: A total of 64 accessions of released varieties (n=18) and popular landraces (n=46) were included in the reference library Samples of these accessions along with the samples collected from farm surveys were all genotyped at 56,849 single nucleotide polymorphisms (SNP) loci. Genetically identical sets of clones were then identified by using distance-based hierarchical clustering and modelbased maximum likelihood admixture analysis (done by researchers at IITA) 28

29 29 DNA fingerprinting: Sampling and logistics Beans in Zambia: For each household, during the household interview, enumerators collected seed samples (10-15 grains) of each variety grown by the farmer in the 2012/2013 season On a small scale, ZARI collaborators also collected seed samples of different varieties from bean vendors in two local markets (in Kasama) Seed samples were collected and stored in small labeled (with HH id, and variety name and id) manila envelopes and kept under dry conditions Few seeds were germinated in a greenhouse by ZARI staff (June 2014) Young leaves from 3-4 week-old plants were collected by ZARI and CIAT staff, put into a 96-well and shipped to LGC Genomics (a private lab in the UK) for analysis About 20% of the bean samples were germinated at MSU and sent to LGC Genomics for inclusion as replicates to compare the results

30 DNA fingerprinting: Methodology and Data Analytics Beans in Zambia: 13 accessions specific to Zambia (including 11 released varieties and two landrace Kabulengeti market classes) and 723 accessions from the East/Southern Africa region (that were genotyped by CIAT as part of another project) were included in the reference library as the background materials to compare the samples collected from farm surveys. The farmer samples were genotyped using 66 assays/markers selected as a sub-set of ~800 SNPs used for the reference library The 66 SNP assays were made up of 4 groups, each of which has more or less the same power to differentiate released varieties from each other and from the background genotypes CIAT researcher helped with the analysis of GBS data for varietal identification 30

31 Sample characteristics and number of samples collected in Ghana and Zambia to test different methods Details Zambia Ghana (Beans) (Cassava) Number of farmers surveyed Average number of plots on which the crop was planted (range) 1.36 (1-4) 2.05 (1-25) Average number of varieties planted per household (range) 2.08 (1-6) 1.92 (1-7) Average number of varieties planted per plot visited (range) (1-5) Number of varietal data points reported in farmer and market survey (for method A and B) Number of samples genotyped (DNA fingerprinted) (for method V) Number of samples photographed and available for varietal identification (for method E)

32 32 3. RESULTS 1. Zambia

33 33 Farmer reported groupings of bean varieties by type (Method A2) Variety type reported by farmers Frequency Improved 106 Local 612 Don't know 84 All 802

34 34 Varietal identification by name based on different methods Asking farmers Method A Showing seeds to farmers Method C Showing photos to experts Method E Showing seeds to experts Method F Released bean varieties in Zambia Chambeshi Lukupa Lyambai Kalungu 207 Kabulangeti Kapisha Kabale Lwangeni Kalambo Mbereshi 2 Planted variety matched more than one seed sample of IV Response does not match any released variety Total

35 Classification of farmer samples and accessions from reference library in to unique variety clusters based on DNA fingerprinting Results for beans in Zambia: 5 unique varietal clusters Unique variety group based on DNA fingerprinting # of accessions from farmer samples Variety Accessions from reference library that fall in the variety group Released varieties Kabulangeti set 3 Landraces / Local varieties Classification of farmer samples that fall in this cluster group IMPROVED VARIETY Variety 2 13 IMPROVED Lukupa VARIETY Variety 3 3 IMPROVED Lwangeni VARIETY Variety 4 1 IMPROVED Lyambai VARIETY Variety 5 24 Kablanketi Local variety 35 Other 690 No MATCH NO MATCH Local variety

36 Measures of effectiveness of different methods of varietal identification: Comparison of outcomes against the benchmark of DNA fingerprinting Measures of effectiveness Varietal data points compared Outcome: Number of data points classified as released (or improved) varieties Type I Error A local variety incorrectly identified as IV Type II Error An improved variety incorrectly identified as a local variety or by an incorrect IV name Accuracy of name: Data points correctly identified as IV by name Accuracy of category: Data points correctly identified as IV by Type Method A1 Method A2 Method C Method E Method F Benchmark (DNA) N N % 4% 13% 71% 18% 15% 16% N % 3% 13% 67% 16% 12% N % 91% 84% 48% 73% 70% N % 9% 52% 27% 30% N % 16% 81% 29% 32% 36

37 37 Main findings from Zambia case study There is no one method that stands out to be most effective across all measures When simply asked for the name and type of bean variety a farmer planted, they under reported the adoption of improved varieties However, when shown the seed samples of IV, they substantially over reported the adoption of varieties matching the seed samples (which in this case were all IVs) Methods based on expert elicitation ex post of the survey based on photos or seed samples collected from the farmers gave adoption estimates closest to the benchmark But these methods had high type II errors or low accuracy rates when the outcome is compared for each data point

38 38 3. RESULTS 2. Ghana

39 39 Varieties names as reported by farmers About 180 variety names reported across 914 accessions collected from farmers fields Names of collected accessions are shown on the right. The font size is proportional to the counts Most common variety names are Debor and Ankra

40 Farmer reported groupings of cassava varieties by type (Method A2) 40 Variety type reported by farmers Frequency Improved 51 Local 796 Don't know / no response 67 All 914

41 List of farmer reported varieties identified by them as improved varieties (Method A1) Variety name Frequency Variety name\a Frequency AGRIC 18 AKOROFUOMPE 1 BANKYE_HEMAA 4 AMENFI 1 NKABOM 4 AMPONG 1 KUFFOUR 3 ANKRA 1 DEBOR 2 BANKYE_FITAA 1 ESI 2 BOSOMENSIA 1 MEDUMEKU 2 ESAM_BANKYE 1 MPOHOR 2 ESIABAYAA 1 ABASA_FITAA 1 GBEZEH 1 ABOSOMNSIA 1 SANTOM_BANKYE 1 AFISIAFI 1 SIKA_BANKYE 1 Total = 51 Total count of variety names that match a released variety = 20 8 of which are reported by farmers as local varieties 41

42 Varietal identification by field visiting cassava expert (Method D) Field visiting experts identified varieties by 86 unique names Most frequently reported names were: Debor 98 Bosomensia 62 Bankye Kokoo 43 Aben Woha 36 Bankye Pole 36 Ankra 23 For 299 observations, field experts could not identify varieties by name Variety type reported by Experts across 914 data points Improved = 47 Local = 805 Don t know / missing data points = 62 42

43 Varietal identification by type based on experts review of photos taken by survey team (Method E) 43 Variety type reported by experts (expost) Frequency Improved 142 Local 644 Don't know / missing information 128 All 914

44 44 Classification of farmer samples and accessions from reference library in to unique variety clusters based on DNA fingerprinting: Results for cassava in Ghana: 11 unique varietal clusters Unique variety group # of accessions from farmer samples Accessions from reference library that fall in the variety group Released varieties Landraces / Local varieties Variety (12) ADW2000_003; ADW2000_004; ANKRA; BOSOMENSIA_1; DEBOR 1; DEBOR_KAAN ; DMA2000_002 ; DMA2000_66 ; KSI2000_126 ; OFF_2000_019 ; OFF_2000_023 ; UCC2000_111 Variety (2) IFAD; UCC (7) TUMTUM ; DWA2000_070 ; ELISHA ; WCH2000_020 ; KW_2000_010 ; KWANWOMA ; OFF_2000_134 Variety 3 65 (1) NKABOM (1) DEBOR 2 Variety 4 17 (1) AFISIAFI (3) ABUSUA; MONICA; UCC2001_449 Variety 5 57 (2) ADE2000_182 ; DMA2000_031 Variety 6 37 (2) KW_2000_148; UCC2001_399 Variety 7 20 No match No match Variety 8 21 (1) BANKYE_BRONI_1 (1) UCC20001_464; Variety 9 13 No match No match Variety No match No match Variety No match No match

45 45 Results for cassava in Ghana (cont d): several hybrids or admixtures Unique variety group # of accessions from farmer samples Accessions from reference library that fall in the variety group Released varieties Landraces / Local varieties 50% ancestry from variety 1 17 No match No match 50% ancestry from variety 2 11 No match No match 50% ancestry from variety 3 19 (3) ESSAM_BANKYE; BANKYE_HEMAA; TEKBANKYE; DOKU_DUADE 50% ancestry from variety 4_group 1 8 (2) BRONI; KW2000_181 50% ancestry from variety 4_group 2 2 (3) NYERIKOGBA; ABASA_FITAA; OTUHIA 50% ancestry from variety 5 12 No match No match 50% ancestry from variety 6 33 No match No match 50% ancestry from variety 8 21 (4) 12_0236; 12_02Y5 ; CONGO_BATIALION; ESIABAYAA 50% ancestry from variety 9 29 No match No match 50% ancestry from variety 11 5 (2) KW_2000_030; UCC2001_249 Multi-ancestry clones _group (11) 12_0197; ADW2001_051; AFS_2000_050; ANKRA_10_003; AW3_10_008; AW3_10_011; BOSOMENSIA_2; CONGO_BATIALION; DEBOR_BEPOSO; OFF_2000_037 WCH2000_011 Multi-ancestry clones _group 2 2 (6) BANKYE_BRONI_2; AMPONG; FILINDIAKONIA; BANKYE_BOTAN; SIKABANKYE; AGBELIFIA Total

46 46 DNA results for cassava in Ghana Several interesting findings 1. Some improved varieties are genetically identical (e.g., IFAD and UCC) 2. Many released varieties are hybrids or admixtures 3. Library accessions representing both released varieties and landraces fall under the same varietal cluster groups (e.g., variety group 2, 3, 4, 8) The last bullet point (#3) especially poses a challenge for varietal identification The problem is: How to classify farmer samples that fall in these four variety cluster groups? Should they be classified as improved/released varieties or local/landrace varieties?

47 47 DNA results for cassava in Ghana (cont d) To address this dilemma, the analysis of effectiveness of different methods against the benchmark of DNA fingerprinting is done under two scenarios / assumptions Liberal scenario: which assumes that all the farmer samples that fall in a variety cluster in which there is at least one released variety are essentially improved varieties Conservative scenario: assumes the opposite (i.e., farmers samples that match the DNA results of a variety group in which there are both released varieties and landraces, the variety group is considered not-improved). The only exception (under the conservative scenario) is Cluster group 4 (variety Afisiafi), which according to IITA cassava experts is unambiguously a cluster of improved varieties that previously did not exist

48 Number of cassava varietal observations from farmer survey matching DNA fingerprints of released and landrace cluster groups under the two assumptions Liberal Conservative scenario scenario Improved varieties 284 (31%) 40 (4%) Local / Landraces 630 (69%) 874 (96%) Total The estimates of IV adoption data points (4 to 31%) under the two scenarios represent two ends of the spectrum of TRUTH 48

49 49 Measures of effectiveness of different methods of varietal identification: Comparison of outcomes against the benchmark of DNA fingerprinting under the liberal assumption Measures of effectiveness Varietal data points compared Outcome: Number of data points classified as released (or improved) varieties Type I Error A local variety incorrectly identified as IV Method Method Method Method Method (DNA) A1 A2 D1 D2 E N N % 1% 6% 2% 5% 15% 31% N % 27% 36% 74% Type II Error An improved variety incorrectly identified as a local variety or by an incorrect IV name Accuracy of name: Data points correctly identified as IV by name Accuracy of category: Data points correctly identified as IV by Type N % 97% 80% 93% 82% 68% N % 3% 7% N % 13% 11% 13%

50 50 Measures of effectiveness of different methods of varietal identification: Conservative assumption Measures of effectiveness Varietal data points compared Outcome: Number of data points classified as released (or improved) varieties Type I Error A local variety incorrectly identified as IV Method Method Method Method Method (DNA) A1 A2 D1 D2 E N N % 1% 6% 2% 5% 15% 4% N % 44% 57% 20% 60% 86% Type II Error An improved variety incorrectly identified as a local variety or by an incorrect IV name Accuracy of name: Data points correctly identified as IV by name Accuracy of category: Data points correctly identified as IV by Type N % 88% 45% 68% 53% 53% N % 13% 33% N % 55% 48% 48%

51 51 Main findings from the cassava case study Like Zambia study, no one method stands out to be most effective on all measures On aggregate level adoption of IV, methods based on farmer elicitation and field observations by an expert provide closest estimates under the conservative scenario, but with high error rate No methods come closer to the truth in adoption estimates under the liberal scenario Identifying cassava varieties accurately by NAME in a setting where hundreds of variety names exist is a challenge across all the methods tested Adoption estimates by the experts (based on photos) are substantially higher than other methods and has much higher type I error (false positives)

52 4. DISCUSSION 52

53 53 1. Key methodological implications for tracking varietal adoption Method to measure variety specific adoption Results for both beans and cassava indicate that when there is a diversity of names by which farmers call their varieties, the traditional method of farmer elicitation will give an underestimate of adoption of improved varieties by names Thus the current gold standard of eliciting varietal adoption from farmer surveys may not be an accurate method for measuring varietal turnover and assessments of type II benefits of plant breeding research (i.e., benefits from varietal replacement)

54 2. Key methodological implications for tracking varietal adoption Method to measure variety specific adoption (cont d) Showing the seed samples to elicit farmers response on variety specific adoption is prone to overestimate adoption of improved varieties if ONLY improved varieties are included. This was a limitation of the bean study. 54 More studies are needed to test whether this upward bias of this method can be reduced by including some popular landraces in the visual samples

55 3. Key methodological implications for tracking varietal adoption Methods to measure adoption of improved varieties Farmers and experts are better able to give an aggregate assessment of the adoption of improved varieties as a category than variety specific adoption. However, all the methods evaluated are prone to both type I and type II errors which has implications on the accuracy of any adoption analysis conducted using such data at the farmer level (more analysis to test this hypothesis needs to be done) 55

56 4. Key methodological implications for tracking varietal adoption Potential of non-traditional methods of measuring varietal adoption None of the alternative and non-traditional methods tested emerged as most effective on all measures of effectiveness; although in the case of beans, methods that involved experts opinion and interpretation were generally closer to the benchmark estimates However, given the time and logistics of implementing these methods, the scalability of some of these methods remains questionable on the grounds of cost-effectiveness and feasibility 56

57 1. Emerging conclusions on DNA fingerprinting: Molecular markers (GBS) is a useful tool for determining the genetic identity of varieties grown by farmers It provides a true picture of what is in farmers fields But as shown by the case of cassava, the implication of this truth in estimating the adoption of improved varieties can be ambiguous There could be different shades of truth if the DNA analysis reveals that: Multiple released varieties are essentially identical; and Some accessions considered to be released varieties share the same DNA fingerprints as accessions considered to be landraces Because of this surprising finding in the case of cassava, identification of farmer adopted varieties by name or by type with 100% certainty remains inconclusive 57

58 58 2. Emerging conclusions on DNA fingerprinting Potential for scaling up as part of household surveys will depend on several factors: Logistics of collecting, tracking, storing and transporting the samples from farmers fields to a lab facility to get high quality DNA Cost of DNA fingerprinting which includes establishing the reference library, DNA extraction, and genotyping service. In this study the estimated cost per data point was ~$30 Capacity to do high volume DNA fingerprinting within the country or easy access to such capacity internationally (i.e.., no government restrictions on the shipment of plant tissues or DNA samples to other countries for analysis)

59 3. Emerging conclusions on DNA fingerprinting Given these challenges, the use of DNA fingerprinting as part of large scale representative HH surveys may be long way from becoming routine Potential ways to reduce the cost and to make the logistics more manageable would be to use DNA fingerprinting as a method of validation on a random sub-sample of households rather than all the households More studies on different crops and country settings are needed to generate an experience base and derive generalizable conclusions 59

60 THANKS WELCOME QUESTIONS AND DISCUSSION