Presentation3

SITE-SPECIFIC CROP PRODUCTION BASED ON FARMERS’ PRODUCTION EXPERIENCES IN COLOMBIA. CASE STUDIES ON ANDEAN BLACKBERRY

(Rubus glaucus Benth) AND LULO (Solanum quitoense Lam)

Daniel Ricardo Jiménez Rodas

Farmers’ production experiences

Principles of participatory and

operational research

Modern information technology

SSCP

Environmental characterization of the production system

Analysis of the Observations to optimize the system

Kg/tplant Temperature Age

Observations made by farmers according to their particular circumstances

- 3 0 . 1

3 0 . 5

M e a n a n n u a l

t e m p e r a t u r e ( º C )

0

1 2 0 8 4

A n n u a l

p r e c i p i t a t i o n ( m m )

publicly-available environmental databases

Site-Specific Crop Production (SSCP)

2

Jimenez, Daniel (CIAT)

Comentar lo de top-down que no se puede financiar proyectos a largo plazo para experimentación con perennes, so farmrers still lack on these..research is short-term... no hay información de que factores ambientales, de manejo o socio economicas influyen en productividad


There is not DDSAT, not enough knowledge...no deterministic, and are perennial

Objectives

The objectives of this thesis are to:

• Demonstrate that the principles of operational and participatory research can be applied to Andean blackberry and lulo, and provide growers with insights into how yield varies

• Evaluate modelling methodologies developed for sugarcane, to determine their suitability as tools for modelling Andean blackberry and lulo yield

• Use these methods to identify the conditions that are most suitable for the production of Andean blackberry and lulo

3

• Modern information technology can be used to combine information on farmers’ production experiences with publicly-available environmental databases

• Principles of operational and participatory research facilitate the task of collecting, characterizing and interpreting cropping events that occur under a wide range of conditions

The hypotheses that this research seeks to verify are:

4

Methods

Collecting farmers’ production experiences

Participatory research • Consultative mode• Collaborative mode

Guide form based on a calendar77


Data recorded on guides included a description of each plot, its location, species and variety or eco-type, events or harvesting experiences and some management practices were registered on the calendars

Collecting Farmers’ production experiences

Calendars developed to capture harvest events

Cropping events

88


Data recorded on guides included a description of each plot, its location, species and variety or eco-type, events or harvesting experiences and some management practices were registered on the calendars

Mostly estimates physical soil properties:Texture, Drainage, Effective soil depth, Structure, Colour

Collecting Farmers’ production experiences

Soil information

9


mostly estimates physicalsoil properties such as slope, stoniness, and mottling, which change less with time compared to chemical properties that change with each fertilizer application or as nutrients are extracted by harvested crops

- 3 0 .1

3 0 .5

M e a n a n n u a l

t e m p e r a t u r e ( º C )

0

1 2 0 8 4

A n n u a l

p r e c i p i t a t i o n ( m m )

SRTM : The Shuttle Radar Topography Mission (high-resolution topographical and landscape information )

WorldClim: Monthly data (precipitation, temperature)

TRMM : Tropical Rainfall Measuring Mission

Publicly-available environmental databases

1010


As the information provided by the satellite is not ground-based, researchers refer to this measure as either precipitation estimates or precipitable wat

Analytical approaches

V1 V2 V3 V4 V5 … V60 L 2 L 3 L 4 L 5 … Kg/plot

Obs 1 0.1 18 3 312 0.3 … 89 0 1 0 1 0 … 2.39

Obs 2 0.2 15 4 526 0.1 … 52 1 0 0 0 1 … 30.35

Obs 3 0.6 14 1 489 0.2 … 64 0 1 1 1 1 … 42.25

Obs 4 0.05 19 2 523 0.5 … 13 0 0 0 0 1 … 52.50

Obs 5 0.4 13 3 214 0.6 … 57 1 1 1 1 1 …

Obs 6 0.8 12 4 265 0.4 … 24 1 1 0 1 0 … 82.25

Obs 7 0.2 15 1 236 0.8 … 26 0 0 1 0 0 … 89.28

Obs 8 0.1 17 3 541 0.1 … 35 0 1 1 1 0 … 125.0

Obs9 0.6 16 2 845 0.3 … 51 0 0 1 1 0 … 142.8

Obs10 0.1 18 1 126 0.1 … 43 1 1 0 0 1 … 150.0

… … … … … … … … … … … … … … …

Obs3000 0.04 15 3 235 0.6 … 85 1 1 1 1 0 … 180

70.52

L 1

Supervised models

Independent variables/ Inputsdependent/output(known)

…

11


We mostly usde unsupervised and supervised approaches. What we can see in both sids are databases (no real, I simulated in order to ilustrate the idea) on the left there is a Matrix with continues valuues (lets say temp, ppt) and 1 and 0 thar are for example caterogical variables such as texture, structure that are converted to 1 and 0 based on prescence and abscence. So, observation 1 are conditions that leads to a productivity of 2.39 kg, in obs 10 of 150... with this matrix we are intersted in to know the variability and we employ multivariate analysis such as regressions (linear and non-linear or combination of both, as we are going to see later on) working with these fruits non-linear because we do not know the functional relationships between lulo and Andean blackberry yield and many of the factors considered likely to influencethese crops production

12

L 1

Observations close to each other in the multidimensional/input are located close in the output/visualization layer - clustering and visualization tool

Unsupervised models

Unsupervised models

V1 V2 V3 V4 V5 … V60 L 2 L 3 L 4 L 5

Obs 1 0.1 18 3 312 0.3 … 89 0 1 0 1 0

Obs 2 0.2 15 4 526 0.1 … 52 1 0 0 0 1

Obs 3 0.6 14 1 489 0.2 … 64 0 1 1 1 1

Obs 4 0.05 19 2 523 0.5 … 13 0 0 0 0 1

Obs 5 0.4 13 3 214 0.6 … 57 1 1 1 1 1

Obs 6 0.8 12 4 265 0.4 … 24 1 1 0 1 0

Obs 7 0.2 15 1 236 0.8 … 26 0 0 1 0 0

Obs 8 0.1 17 3 541 0.1 … 35 0 1 1 1 0

Obs9 0.6 16 2 845 0.3 … 51 0 0 1 1 0

Obs10 0.1 18 1 126 0.1 … 43 1 1 0 0 1

… … … … … … … … … … … … …

Obs30000.04 15 3 235 0.6 … 85 1 1 1 1 0

L 1

Analytical approaches…………………………………………………………………………

…………

…………

………………………………


the same matrix, but we do not now or we are insted of being interested in explaining variations.. we want to see how obsertations are related to each other (to see if observation can be put it together as they are similar), Thus, observations close to each other in the multidimensional/input space active neurons which are close in the output/visualization layer Decir que com en Andean vimos un resultado que mostraba la importancia de localidad, en un análisis utilizamos en la mayoria unsupervised y suvevised... pero un hierarchcal approach with mixed models

SSCP = (Participatory & Operational research ) + publicly-available environmental data + analytical approaches + farmers’ production experiences

Crop Departments Geo-referenced

Cropping events Production

Variety and number of

plantsRASTA Complete plots

No offarms

weeklyperiods

No offarms

No offarms

No offarms

No offarms

Andean blackberry

Caldas, Nariño

75 488 35 34 20 20

Lulo Nariño, Others

111 254 54 43 21 21

Total 186 742 89 77 41 41

Results

Summary of the number of Andean blackberry and lulo growers who recorded information via calendars

14

Results - Andean blackberry

Scatter plot displaying MLP predicted yield versus real Andean blackberry yield, using only the validation dataset1715

-0.2 0.3 0.8 1.3 1.8-0.2

0.3

0.8

1.3

1.8

f(x) = 0.892655122481665 x + 0.0157451798619761R² = 0.891999243225333

Predicted

Real yield (kg/plant/week)

Pred

icte

d yi

eld

(kg/

plan

t/w

eek)

Supervised models - Non-linear regressionCoefficient of determination= 0.89

Histogram displaying yield data distribution of Andean blackberry (Kg/plant/week)

Num

ber o

f obs

erva

tions

Eff

Dep

th

Tem

pAvg

_1

Na_

un_c

hica

l

Na_

un_c

usba

Tem

pAvg

_0

Tem

pAvg

_2

Tem

pAvg

_3

Ext

Dra

in

Pre

cAcc

_1

Trm

m_3

Nar

-Cal

Cal

_rio

su_z

r

Srt

m

Slo

pe

Pre

cAcc

_0

Trm

m_2

Na_

un_c

usal

Trm

m_0

Pre

cAcc

_3

Tem

pRan

g_0

Tem

pRan

g_2

AB

_Tho

rn_N

Na_

un_l

ajac

Pre

cAcc

_2

Trm

m_1

IntD

rain

Tem

pRan

g_3

Tem

pRan

g_1

12 20 3 5 17 23 26 11 22 16 2 7 8 9 19 15 4 13 28 18 24 1 6 25 14 10 27 21

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

% S

ensi

tivity

Sensitivity distribution of the model with respect to the inputs

Jiménez, D., Cock, J., Satizábal, F., Barreto, M., Pérez-Uribe, A., Jarvis, A. and Van Damme, P., 2009. Computers and Electronics in Agriculture. 69 (2): 198–208

Sensitivity Matrix Results - Andean blackberry

16

- 3 0 .1

3 0 .5

M e a n a n n u a lt e m p e r a t u r e ( º C )

0

1 2 0 8 4

A n n u a l p r e c i p i t a t i o n ( m m )

Effective soil depth

Temperature averages

Geographic location


(a) Kohonen map displaying the resultant 6 clusters and their labels according to yield values (b) Component plane of Andean blackberry yield, the scale bar (right) indicates the range value of productivity in kg/plant/week The upper side exhibits high values of yield, whereas the lower displays low values

Unsupervised model - Visualization – component planes - SOM

17

Andean blackberry yieldKohonen map – 6 clusters

(a) (b)


Component plane of effective soil depth. The scale bar (right) indicates the range value in cm of soil depth: the upper side of the scale exhibits high values, whereas the lower displays low values

18




Components planes of the temperature averages. In all figures, the scale bar (right) indicates the range value in ◦C of temperature. The upper side exhibits high values, whereas the lower displays low values

19



Component planes of the specifics geographic areas Nariño–La Union–Chical alto (left) and Nariño–La union–Cusillo bajo (right). The highest values indicate presence and the lowest absence as they are categorical variables

Visualization – component planes - SOM

20

Nariño - La Union – Chical Alto Nariño - La Union – Cusillo bajo

Results - Lulo

Distribution of R2 obtained with each model

Regression R2

(mean)Confidence

interval (95%)

Robust (linear) 0.65 0.63 - 0.66

MLP (non-linear) 0.69 0.67 - 0.70

Both models explained more than 60% of variability in Lulo production

2321

Histogram displaying yield data distribution of lulo (g/plant/week)

R2 pr ovide d by e ach appr oach

MLP

Robus t regres s ion

0.2877 0.3545 0.4214 0.4883 0.5552 0.6221 0.6889 0.7558 0.82270

2

4

6

8

10

12

14

16

18

20

22

24

26

Nu

mb

er o

f o

bse

rvat

ion

sN

umbe

r of o

bser

vatio

ns

Num

ber o

f obs

erva

tions

Supervised modelling

Results - LuloThe Sensitivity Matrix

effD

epth

tem

pAvg

_0slo

pe

Na_un

_chi

cal

srtm

Na_un

_jac

trmm

_2

Na_ca

_san

Na_un

_ba

Tem

pRan

g_1

Tem

pRan

g_0

trmm

_1

int_

extD

rain

Tem

pRan

g_2

trmm

_0

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

% S

ensi

tivity

Jiménez, D., Cock, J., Jarvis, A., Garcia, J., Satizábal, H.F., Van Damme, Pérez-Uribe, A., and Barreto, M., 2010. Interpretation of Commercial Production Information: A case study of lulo, an under-researched Andean fruit. Agricultural Systems. 104 (3): 258-270

22

Sensitivity distribution of the model with respect to the inputs


Temperature averages

Slope

(a) U-matrix displaying the distance among prototypes. The scale bar (right) indicates the values of distance. The upper side exhibits high distances, whilst the lower displays low distances; (b) Kohonen map displaying the 3 clusters obtained after using the K-means algorithm and the Davies–Bouldin index

The three most relevant variables were used to train a Kohonen map and identify clusters of Homogeneous Environmental Conditions (HECs)

Results - LuloUnsupervised model - Clustering – component planes - SOM

23

U-Matrix Kohonen map – 3 clusters

Results - LuloClustering – component planes - SOM

A mixed model with the categorical variables of three HECs, location and farmer explained more than 80% of variation in lulo yield

Parameters Estimate (g/plant/week)

StandardError

%of total variance

Model including categorical variables of 3 HECs, location and farm

HEC 1.85 2.01 61.2%

Location 0.07 0.20 2.5%

Site-Farm 0.57 0.21 19.0%

Error 0.52 0.04 17.3%

Total 100.0%

Variance components of the mixed model estimations

24

Variable ranges HEC

Slope (degrees) EffDepth (cm) TempAvg_0 (°C)

5-14 21-40 15 -16.5 18-15 32-69 15 -18.9 213-24 40-67 15.8 -19 3

HEC 3 yielded 41 g/plant/week more fruit than average

Results - Lulo

1 2 3

-30.00

-20.00

-10.00

0.00

10.00

20.00

30.00

40.00

50.00

Effects of clusters of environmental condi-tions

Lu

lo y

ield

(g

/pla

nt/

we

ek

)

25

Results - Lulo

Farm 7 and 9 in HEC 3. Farm 7 produced 68 g/plant/week less than average, whilst farm 9 produced 51 g/plant/week more than average

1 2 3 4 5 8 17 5 6 8 10 11 12 13 15 16 17 19 20 7 9 14 18 19 20 211 2 3

-80.00

-60.00

-40.00

-20.00

0.00

20.00

40.00

60.00

Effects of farms across clusters of environmental conditions

Lu

lo y

ield

(g

/pla

nt/

we

ek

)

1 2 3

26

Jiménez, D., Cock, J., Jarvis, A., Garcia, J., Satizábal, H.F., Van Damme, Pérez-Uribe, A., and Barreto, M., 2010. Interpretation of Commercial Production Information: A case study of lulo, an under-researched Andean fruit. Agricultural Systems. 104 (3): 258-270

Conclusions

27

• Most suitable environmental conditions for producing Andean blackberry are: Average temperature between 16 and 18 °C Minimal effective soil depth between 40 and 65 cm

• Most suitable environmental conditions for producing lulo are: Average temperature between 15.8 and 19°C Effective soil depth between 40 and 67 cm Slope between 13 and 24 degrees

• Farmers who properly manage their fields were identified

• Yield differences Andean blackberry – localities Lulo - yield gap between farms in similar environmental conditions

Conclusions

• Key role of farmers (186 registered information on 742 cropping events)

• Analytical approaches explained more than 80% of variability for both crops

• Farmers’ production experiences and publicly-available environmental data can be analysed as long as it is possible to collect sufficient data on how the growers manage their crop, and how much they produce

• The biggest challenge is not the analysis of information… rather the collection of data

• The data collection and the analysis seem to be promising tools to develop a SSCP for other crops or regions where there is neither information on climate nor on soils

• This is the first time that this methodology has been implemented for under-researched crops in general and in Colombia in particular

28

Limitations of the research

• Quality of the data collected

• Information on management practices

• Black-box / traditional models? In some cases in general agreement

• HECs constructed under the assumption of environmental variables that are constant over the time

• The results found here cannot be extrapolated outside the ranges of the variable values appearing in the collected datasets

29

Contributions

• Use of farmers’ production experiences (commercial data) for understanding variability

• To turn farmers' day-to-day activities into experiments

• Introduction of novel analytical approaches in LAC for analyzing information

• Provides scientific evidence on the factors that drive productivity for highly under-researched fruits

• First formal research study that evidences the yield gap between farmers under similar climatic conditions in Colombia

• More than 3000 farmers in Colombia are willing to increase productivity and taking benefit of this doctoral research

• Provides a sound basis for transferring technology between localities and farms

30

Questions

Presentation3

Technology

real andean blackberry

number of andean blackberry

lulo nario

lulo growers

modern information technology

monthly data precipitation

physical soil properties

modelling methodologies