The Research Priorities Study

The research priorities study was developed after the rice biotechnology study had been completed. This study sought SPE estimates both for research potential (RP) and for time to achievement. Perhaps its major innovation was that it used a combined RPA and research technique (RT) format, enabling comparisons among conventional breeding, wide-crossing, and transgenic and marker-aided breeding.

A formal questionnaire designed to elicit ratings of research potential and timing was administered to a total of 17 rice scientists (nine from IRRI, eight from NARS). Each scientist was asked to provide four numbers for research problem areas where they considered themselves to be qualified. The four numbers were:

1. A 'rating' of the potential (RP) for a research contribution to the RPA-RT problem area. Ratings were on a 1-5 scale.

2. A rating of the achievement to date (RA) by research on the RPA-RT problem area.

3. An assessment of the date (years from now) by which either a 25% achievement of the remaining potential (RP minus RA) would be achieved or by which there was a 25% likelihood of achievement.

4. An assessment of the date by which either a 75% achievement of the remaining potential was expected or by which a 75% likelihood of achievement was expected.

Table 19.3. Time assessment: disease resistance response distribution.

Year of achievement

Biological tool

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2015 2020 2030 2050 Range

Median

II. Bacterial leaf 25% prob. 75% prob.

Respondent group (1994) IARC scientists 25% prob. 75% prob. Less developed countries' scientists 25% prob. 75% prob. Developed countries' scientists 25% prob. 75% prob.

12 1

12 1

12 1

18 11

11 11

13 11

14 11

995-2005 995-2010

986-2000 998-2010

996-2000 998-2010

995-1999 998-2002

Table 19.4. Time assessment: abiotic stress tolerance response distribution.

Year of achievement

Table 19.4. Time assessment: abiotic stress tolerance response distribution.

Year of achievement

Biological tool 1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006 2007

2008

2009

2010

2015

2020

2030

2050

Range

Median

I. Drought

1

3

1

5

6

20

1

1

2

1

6

6

1

1998-2005

(2000)

25% prob.

1

1

4

3

5

1

9

3

2

9

7

1

2000-2015

(2005)

75% prob.

II. Flood

25% prob.

2

1

6

6

13

1

1

2

1

3

1

1998-2000

(2000)

75% prob.

3

1

4

1

10

1

1

6

6

3

1

2003-2015

(2005)

III. Cold

25% prob.

1

4

6

6

10

5

2

1

2

1

7

1997-2005

(1 999)

75% prob.

3

2

3

4

1

7

1

7

7

5

1

1

1

2000-2015

(2005)

IV. Salt

25% prob. 1

2

5

7

2

7

10

1

4

1

5

2

1

11

1997-2005

(2000)

75% prob.

4

1

5

1

3

4

6

1

15

4

1

1

2000-2010

(2005)

V. Nutrient deficiency

25% prob.

3

6

4

6

1

2

2

3

1998-2005

(2000)

75% prob.

5

1

3

6

3

5

34

2

1

2000-2015

(2010)

Respondent group (1994)

IARC scientists

25% prob.

4

2

2

10

3

1

5

4

1

75% prob.

1

1

1

8

1

1

7

4

1

Less developed countries'

scientists

25% prob.

1

3

5

8

7

1

1

7

3

6

1

2

75% prob.

1

1

2

6

3

3

1

5

4

Developed countries'

scientists

25% prob.

1

4

1

7

3

17

1

2

4

1

1

10

7

75% prob.

5

1

4

10

1 1

10

10

8

Table 19.5. Time assessment: yield enhancement response distribution.

Year of achievement

Biological tool

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2015 2020 2030 2050 Range

Median

I. Male sterility 25% prob. 75% prob.

II. Starch metabolism 25% prob.

III. Photosynthetic effect 25% prob.

V. Nitrogen fixation 25% prob. 75% prob.

Respondent group (1994) IARC scientists 25% prob. 75% prob. Less developed countries' scientists 25% prob. 75% prob. Developed countries' scientists 25% prob. 75% prob.

12 3

7 12

14 4

9 11

10 7

12 12

4 11

5 10

1995-2000 1997-2010

1996-2000 2000-2006

1997-1999 2005-201 5

2000-2010 2000-2015

1 2010-2020 6 2010-2050

The elicitation of these four numbers was based on the following principles:

1. Scientists are more comfortable with a rating scale (1-5) than with a specific estimate of a productivity level. Rating scales linked to achievement were provided to scientists. These were:

• less than 10% achievement of loss elimination (or increase in biological efficiency);

• 10-25% achievement of loss elimination (or increase in biological efficiency);

• 25-50% achievement of loss elimination (or increase in biological efficiency);

• 50-75% achievement of loss elimination (or increase in biological efficiency);

• 75%+ achievement of loss elimination (or increase in biological efficiency).

The distribution of these ratings obtained from the sample of respondents were then quantified into a mean percentage achievement measure (the variance was also computed).

2. The distinction between RA and RP was needed to clarify what was meant by remaining research potential. By specifying both RP and RA we attempted to capture more clearly the incremental potential for further gains. In many RPA-RT classes, respondents indicated that while substantial RP for problem solutions existed in the past, research programmes had already achieved all or most of this potential, i.e. they had 'exhausted' much of the potential (see Evenson, 1996, Ch. 5). For research priority setting, we base the future research potential on the remaining potential, i.e. RP —RA.

Achievement-to-date ratings were based on research programming to date. Respondents were asked to visualize the continuation of current research programmes with some strengthening and normal responsiveness to research opportunities in estimating RP and time to achievement. Note that by utilizing these RP —RA concepts in this way we are attempting to rule out the possibility of specifying an arbitrary research programme to obtain RP estimates. Respondents have the experience to rate actual programmes more accurately than hypothetical programmes.

3. Scientists need some scope for expressing the variance in their subjective probability estimates. Our experience with the Rice Biotechnology study and with scientists indicates that eliciting two dates on time-to-achievement was an effective way to obtain a 'distribution' reflecting the degree of uncertainty of scientists.

Tables 19.6-19.10 summarize the scientists' responses to the ratings elicitation. It should be noted that not all respondents completed each block of RPA questions. They did, however, complete each RT question for the RPAs for which they responded. This was designed to achieve comparative consistency over RT.

Table 19.6. Scientists' ratings: insect loss RPAs.

Management research

Y25 Y75 RP-RA SD

Conventional breeding

Y25 Y75 RP-RA SD

Wide crossing

Y25 Y75 RP-RA SD

Transgenic breeding

Y25 Y75 RP-RA SD

Y25 Y75 RP-RA SD

Y25 Y75 RP-RA SD

Y25 Y75 RP-RA SD

Yellow stemborer

5

10

0.24

0.20

8

13

0.16

0.12

9

15

0.22

0.18

7

13

0.54

0.32

Striped stemborer

4

11

0.32

0.22

9

12

0.15

0.10

9

12

0.20

0.20

7

10

0.52

0.46

Brown plant hopper

5

8

0.16

0.22

9

12

0.16

0.15

9

12

0.20

0.26

10

12

0.31

0.28

WB/brown plant hopper

4

10

0.23

0.18

7

11

0.16

0.17

10

13

0.27

0.22

9

14

0.20

0.24

Leaf folder

5

9

0.28

0.18

9

12

0.17

0.15

9

13

0.10

0.12

9

13

0.20

0.36

Hispa

6

13

0.12

0.10

11

15

0.20

0.10

8

14

0.20

0.16

10

12

0.33

0.42

Green leafhopper

5

10

0.17

0.22

7

12

0.20

0.10

10

16

0.30

0.26

9

13

0.30

0.20

Gall midge

5

10

0.30

0.21

7

12

0.30

0.21

9

15

0.28

0.26

9

15

0.32

0.30

Caseworm

6

11

0.30

0.21

8

17

0.16

0.09

11

19

0.15

0.18

10

15

0.36

0.40

Armyworm

6

11

0.30

0.17

9

15

0.16

0.09

11

16

0.15

0.18

10

15

0.36

0.40

Grasshopper

4

6

0.20

0.20

7

9

0.14

0.12

9

11

0.07

0.10

7

10

0.14

0.22

Mealy bug

4

7

0.20

0.10

8

12

0.14

0.12

9

11

0.10

0.10

7

10

0.30

0.42

Rice bug

4

7

0.20

0.14

8

12

0.20

0.16

9

11

0.07

0.10

7

10

0.14

0.22

For each RPA in Tables 19.6-19.10, four numbers are reported for each RT:

• mean years to 25% achievement of remaining potential (Y25);

• mean years to 75% achievement of remaining potential (Y75);

• mean estimated per cent of remaining potential (RP —RA);

• standard deviation of estimated per cent of remaining potential (SD).

Obviously, standard deviations of Y25 and Y75 could also have been computed. However, for purposes of displaying variation in estimated impacts of research programmes, variation in RP —RA is more relevant than variation in Y25 and Y75 which were designed to allow scientists to express their subjective variances. Thus the differences in Y25 and Y75 reflect the 'within scientists' subjective variation in estimates, while the standard deviations reported in Tables 19.6-19.10 reflected variations in estimates between scientists.

1. Scientists' ratings: insect loss RPA. We turn first to the insect loss RPAs summarized in Table 19.6. We note that there are differences in the RP —RA estimates both by RPA and by RT. Given the small scientist sample and the relatively high standard deviations across scientists, few of these differences are statistically significant. Most standard deviations are lower than the estimated RP —RA terms (note: scientists reported separate ratings for RP and RA). Most standard deviations for RP and RA separately were roughly one-third or so of the mean RP and RA estimates. The standard deviations of the differences, however, are relatively high. Should this be construed to mean that few differences across RPAs actually exist? If so, we can simply use 'congruence' rules to allocate over RPAs (see Evenson, 1996, Ch. 5).

We would argue that the procedure of separately identifying the RP and RA components probably results in an upward bias in the standard deviations and that differences over RTs are meaningful. We also consider differences over research techniques to be meaningful. Here we note that the highest RP —RA estimates are for the transgenic breeding techniques in all but one or two cases. Wide-crossing and tissue-culture techniques tend to be located between conventional breeding and transgenic techniques in these estimates.

Timing estimates also do not vary substantially by RPAs, but clearly do by RT. The management RTs are expected to yield results earlier than the genetic improvement techniques. Interestingly, transgenic techniques do not appear to have very different time estimates from conventional breeding or wide-crossing techniques.

2. Scientists' ratings: disease loss RPAs. Ratings for disease loss RPAs (Table 19.7) show similar patterns of variation over RPAs and RTs to those observed for insect loss RPAs. As with insect loss RPAs, there is more variation in the expected gains from working on the more important diseases, and transgenic techniques generally have the highest expected gains and the longest expected periods to achievement.

3. Scientists' ratings: abiotic stress loss RPAs. Abiotic stress loss RPAs (Table 19.8) again show patterns similar to those for other losses. Management solutions generally have lower expected contributions, however, and tend to have longer expected time-to-achievement estimates.

Table 19.7. Scientists' ratings: disease loss RPAs.

Management research

Y25 Y75 RP-RA SD

Conventional breeding

Y25 Y75 RP-RA SD

Wide crossing

Y25 Y75 RP-RA SD

Transgenic breeding

Y25 Y75 RP-RA SD

CD 3

CD Co

urce

Y25 Y75 RP-RA SD

Y25 Y75 RP-RA SD

Y25 Y75 RP-RA SD

Blast

6

14

0.30

0.20

5

12

0.20

0.26

6

13

0.22

0.20

a

13

0.40

0.2a

Leaf scald

5

10

0.70

0.2a

g

17

0.20

0.10

11

20

0.26

0.12

10

1a

0.14

0.12

Cer leaf spot

5

25

0.26

0.12

10

17

0.30

0.12

11

20

0.30

0.12

11

1g

0.20

0.16

Brown spot

a

15

0.20

0.16

a

12

0.30

0.12

g

15

0.30

0.12

10

1a

0.20

0.16

Sheath rot

10

17

0.2a

0.30

10

17

0.15

0.10

11

1g

0.2a

0.10

10

17

0.10

0.14

Sheath blight

6

15

0.36

0.32

10

16

0.0a

0.10

a

16

0.24

0.20

7

13

0.34

0.26

Stem rot

10

17

0.20

0.10

10

15

0.20

0.10

a

16

0.20

0.10

7

13

0.20

0.10

Bacterial blight

g

12

0.20

0.16

6

13

0.22

0.2a

5

11

0.36

0.22

a

12

0.25

0.20

Bacterial leaf streak

5

10

0.20

0.10

a

13

0.16

0.16

5

10

0.20

0.20

7

11

0.26

0.12

False smut

7

12

0.05

0.05

7

12

0.05

0.05

7

13

0.20

0.10

7

12

0.20

0.10

Glum blight

7

12

0.05

0.05

7

12

0.05

0.05

7

13

0.20

0.10

7

12

0.20

0.10

Tungro

10

17

0.22

0.22

5

14

0.22

0.12

7

14

0.32

0.10

a

15

0.4a

0.40

Ragged stunt

10

17

0.20

0.10

7

12

0.16

0.16

7

14

0.20

0.10

a

15

0.20

0.20

UFRA

10

17

0.20

0.10

7

12

0.16

0.16

7

14

0.20

0.10

a

15

0.20

4. Scientists' ratings: general pest loss RPAs. The RTs specified for the control of weeds and other pests (Table 19.9) differ from those for other crop loss categories. The cultural and mechanical control options are expected to play the major role in weed control. Research has expected contributions to make in terms of biological control methods and bio-pesticides. Transgenic options for control also have some promise.

5. Scientists' ratings: biological efficiency RPAs. It is important that biological efficiency RPAs be included in priority setting. Since they do not have natural 'loss' units, it is sometimes difficult to specify meaningful RPAs. Consultation with scientists indicates that the RPAs in Table 19.10 are meaningful, but the priority setter should be particularly aware that the RPAs are subject to change as new scientific and technological options become available.

Was this article helpful?

0 0

Post a comment