Temporal and spartial changes in land use patterns and biodiversity in relation to farm productivity at multiple scales in Tigray,Ethiopia

- Lukas Brader Fellowship

Ethiopia is one of the mega centres of diversity


TIGRAY REGIONAL STATE(The study region)

Tigray is located in the northern highlands of Ethiopia, covering 80,000 square km

Topography: 500 - 4000 meters above sea level

Population: 4 3 millionPopulation: 4.3 million

Crop calendar:

June to September rainy season; October & November harvest season; and May & early June land preparation and sowing.y y p p g


Study AreaStudy Area


Research objectives:

1 To detect land use and land cover (LULC) changes based on a time series1. To detect land use and land cover (LULC) changes based on a time series of remote sensing data and identify drivers of the changes at a regional scale.

2. To identify and analyze factors affecting agro-biodiversity and soil erosion, focusing on relationships between agro-biodiversity, physical environment, crop (farm) characteristics and measures of wealth at farm and regional scales.

3. To study spatial and temporal variation in agro-biodiversity and soil d d i i l i f d i i l h i l d ldegradation in relation to farm, productivity, wealth, social, development and topographic characteristics between 2000 and 2005 at farm and regional scales.

4. To investigate the effects of F. albida based land use systems on crop productivity at field and landscape scales.


+

How was the research done?

Reg

iona

l

Component 2 (2000)-Road & town mapsEl ti d lR

e

Component 1 (1964-2005)-LULC change detection-Driving factors of change

-Elevation and slope-Farm & wealth(151 farms) Component 3

(2000-2005)-Road & town maps

Farm

Fiel

d sc

al

oad & to aps-Elevation and slope-Farm, wealth &social (151 farms)

F

+

Component 4 (2005)Road & town maps

Fiel

d

-Road & town maps-Elevation and slope-Field (77) and farm (81) productivity & soil

Productivity Agro-Biodiversity Land use

F

Level of study

Remote sensing based land use/ land cover change detection and associated driving factors for the period 1964 – 2005 in the highlands of Tigray, Ethiopia.


Problem statement

N d di f h h d h l d /l d (LULC)No understanding of where, when and why land use/land cover (LULC) changes took place in relation to drivers of the changes which may have serious implications on biodiversity loss, land degradation and declining agricultural productivityproductivity.

Changing land use policies with changing governments/regimes (three different land use policies in the whole study period: 1964 – 2005)land use policies in the whole study period: 1964 2005).

Challenge on how to ensure food security while conserving biodiversity and minimizing land degradation. g g

Objectives

To assess the dynamics of LULC for the period 1964 – 2005 in the highlands of Tigray, northern Ethiopia using remote sensing techniques, and

To identify and quantify the drivers associated with LULC changes.


Specific study area description

The specific study area is located in Tigray, northern Ethiopia (40 82’ - 50 10’ N p y g y, p (and 150 66’ - 150 28’ E), and covers an area of 30 x 40 km at an elevation of 1300 - 2800 metres above sea level (m.a.s.l.).


Data Year + month Path/row Resolution/

Materials used:

Data Year + month Path/row Resolution/Scale

Landsat ETM+ 2005,10 169/050 30 meterLandsat TM 1994,10 169/050 30 meterAerial Photograph 1964 and 1994,11Topographic map 1994Topographic map 1994Shuttle Radar Topographic Mission (SRTM)

2000 90 meter

Mission (SRTM)

Softwares:

ERDAS IMAGINE 9 1ERDAS IMAGINE 9.1ArcGIS 9.2SAS statistical package


SN

Class Name

DescriptionLand use/land cover classes used in the classification

1 Woodland It is composed of trees, bushes, shrubs and herbs. Canopy cover of this unit is estimated to be 65%.

2 Grassland This is open grassland with some shrubs and occasional trees.

3 Shrub land Land supporting a stand of shrubs usually not exceeding 3m in height with a canopy cover of more3 Shrub land Land supporting a stand of shrubs, usually not exceeding 3m in height, with a canopy cover of more than 30%.

4 Scrubland It is mainly characterized by strata of shrubs and grasses or herbs growing here and there.

5 IntensivelyCultivated

It is estimated that of this mapping unit over 70% of the land is under annual and perennial crops Cultivated land

6 Moderately cultivated land

It is estimated that of this mapping unit 40-70% of the land is under annual and perennial crop.

7 Sparsely Cultivated land

It is classified as sparsely cultivated (only 20-40%) of the entire mapping unit is under cultivation.

8 Water body Water in Micro Dams

9 Settlement Residential/industrial areas with a population of more than 2000 households


Description of Land Use/Land cover classes of Tigray, Ethiopia

Woodland

Shrub land

Scrubland

Sparsely cultivated

Moderately cultivated

Intensively cultivated


Cont…

GrasslandGrassland

Waterbody

Settlement


MethodsTopographic map of the

areaCorrected ETM+ 2005 image

Aerial photographs (1964) SRTM

Geo-referenced topomap

ScanningFurther correctionfrom Topomap & field data Scanning

DTM Geo-referencing

Geometric correction

Geo-referenced aerial Geo-referenced recent

image (2005)

Image-to-image registration

photographs

Digitize ground control points

Unsupervised classes

Topographic

1994

Unsupervised

Corrected aerial photos

Resampled aerial photos

Resample

Normalized images

Topographic Normalization

pclassification

Gluing

-Training sample collection from field for the signature editor-Supervised classification by MLKH classifier

Accuracy assessment (2005 & 1994)

Overall accuracy

Transfer interpreted aerial photographs to Ortho-photo

mosaicLandover map of 1964

Vectorize Change Statistical l i d

Land cover map of each year

Kappa statistics detection analysis and interpretationLand cover map of

each year (2005 & 1994)


Cont…Spatially explicit multiple logistic regression model was used to estimate the probability of occurrence of LULC class change as affected by a set ofprobability of occurrence of LULC class change as affected by a set of independent variables:

•elevation (continuous)•slope (continuous)p ( )•distance to major river (buffered)•distance to major road (buffered)•distance to settlement (buffered) and• population density (continuous)

Dependent LULC classes (a total of 2000 samples: 1000 changed; 1000 h d)unchanged)

•Woodland (binary: 0 - 1)•Shrub land (binary: 0 - 1)•Scrubland (binary: 0 1)•Scrubland (binary: 0 - 1)•Agricultural land (binary: 0 - 1)

The general formula of the multiple logistic regression model was:The general formula of the multiple logistic regression model was:Logit (p) = log [p/1-p] = α + β1 X1 +β2 X2 … + βn Xn


ResultsAcross 41 years (1964 - 2005), the results reveal a sharp reduction in natural habitats and an increase in agricultural land.1964 1994 2005

g

droad


Regional scale - results1964

Sparsely cultivated

Shrub land

Woodland

Shrub land


Regional scale - results1994

Moderately cultivated1994

Scrubland

Sparsely cultivated


Regional scale - results

2005 Intensively cultivated2005

Shrub land

Moderately cultivated


Cont…In 1964, shrub land was dominant (covering 46% of the study area) followed by woodland (covering 28% of the study area)woodland (covering 28% of the study area).

In 1994 and 2005, agriculture was dominant covering 34 and 40 % of the study area respectively.p y

45.0050.00

25 0030.0035.0040.00

enta

ge

19641994

10.0015.0020.0025.00

Perc

e 19942005

0.005.00

Wd Sh Sc SCu MCu ICu Gr W Se

Land use/land cover type


Cont…

Over the study period (1964 – 2005), there was conversion of one land cover type to another For example 32 4 and 33 1 % of shrub land was converted intotype to another. For example, 32.4 and 33.1 % of shrub land was converted into combined agricultural land in 1964 – 1994 and 1994 – 2005, respectively.

Moreover, 59.3 and 50.1 % of grassland was converted into agricultural land in , g g1964-1994 and 1994-2005, respectively.

There was even conversion of sparsely cultivated into moderately cultivated by 27.7 % in 1964-1994 and 37.3 % in 1994-2005.


Cont…

Accuracy assessment

V lid ti i d t f d lid ti i t ll t d f fi ldValidation was carried out from random validation points collected from field for the 2005 Landsat ETM+ and from the same spatial and temporal scale of 1994 aerial photographs for the 1994 Landsat TM.

The overall accuracy and overall Kappa statistic for the Landsat 1994 image were 78 and 71 %, respectively.

For the Landsat 2005 image, overall accuracy and Kappa statistic were 74 and 70 %, respectively.


Cont…

Drivers of LULC changeDrivers of LULC change

In the first period (1964 – 1994), distance to road was important driver of LULC change The further thedistance to road was important driver of LULC change. The further the location was from a road so much the greater was the probability of change (reductions) in wood and shrub lands and associated increase in scrublandscrubland.

Change in location (increase) in agricultural land was primarily associated with an increase in human population density. p p y

In the second period (1994 – 2005), woodland locations changed (decreased) primarily by settlement, particularly at high elevation and steep slopes.

Similar to the first period, agricultural land changed (increased) by l ti d itpopulation density.


Conclusion

• The study over a period of 41 years (1964 -2005) reveals LULC changesThe study over a period of 41 years (1964 2005) reveals LULC changes particularly expansion and intensification of agricultural lands at the expense of natural habitat reductions.

•Reductions in extent and location of natural habitats (woodland and shrub land) was higher as locations were further from a road in the first study period (1964-1994).

•In the second period (1994-2005), natural habitats were reduced closer to settlements, particularly at high elevation and steep slopes.

•Expansion and intensification of agricultural lands was associated with an increase with human population.

•This study provides a spatially explicit approach that can help to improve the understanding of LULC dynamics in relation to their drivers in heterogeneous landscapes of tropical highlands.heterogeneous landscapes of tropical highlands.


Agro-biodiversity and soil erosion on farmlands in Tigray, northern Ethiopia.


Problem statement

Previous research results, in northern Ethiopia, indicated that there is expansion and intensification of agriculture (even in steep slopes) at the expense of

l b f h i d d LULC hnatural components because of human induced LULC changes.

However, there was no information on status and spatial distribution of agro-biodiversity and soil erosionbiodiversity and soil erosion.

Objective

To identify and analyze factors affecting agro-biodiversity and soil erosion, and relate them to physical environment, farm characteristics, wealth characteristics and topographic/development drivers in Tigray, northern Ethiopia.


Materials and Methods

Study area


Soil type

Non-spatial datasetData collection and analysis

Farm chara.per farm Insect species

Crop type

Weed species

Agro-biodiversityCrop selectionCriteria

Inorganic fertilizer

per farmCrop diversity

Wealth chara.per farm

fertilizer

Number of livestock holding

Number of creditSoil erosion per

Tree/shrub species Statistical analysis

•Multiple regression•Chi-square test

Number of credit sources

Erosion classes

Spatial dataset

farm •Correlation analysis•Redundancy analyses (RDA)

p

Topographic/Development

Distance from road

Distance from DevelopmentDrivers

town

Elevation of farms

Results

Factors related to agro-biodiversityFactors related to agro biodiversity

- The higher the number of tree and shrub species, the higher was the crop diversity.

Th th il i th l di ifi d b th th d t / h b i- The worse the soil erosion, the less diversified were both the crops and tree/shrub species.

-Farmers with few credit sources planted a great variety of crops (χ2 = 18.6, DF = 6, P=0.01).

- Crop selection criteria was positively associated with crop diversity.

I l l d d ht i t1620

f far

ms

HighlandIntermediateIn lowlands, drought resistance was

first choice

In highlands, straw quality was mosti t t

048

12Yield e lue ity lity lity

Num

ber o

f IntermediateLow land

important

In intermediate altitude, esp. close to towns, high market value was first

l ti it i

YieInsect r

esistance

Weed resis

tanceMarke

t value

Early matu

rity

D rought r

esistance

Treshab ility

Beverage s

uitab ility

Straw qua lity

Selection criteriacrop selection criterion. Selection criteria


Cont…Development drivers and altitude

a ba ba

- Tree/shrub species diversity and crop diversity decreased as buffer distance of farms from roadsdistance of farms from roads decreased.

- Higher agro-biodiversity was observed in farms far from roadsobserved in farms far from roads.

-Road type was also important- Both tree/shrub and crop diversity were reduced close to

a ba b b

diversity were reduced close to all weather roads than dry weather roads.

-Diversity of tree/shrub and crops-Diversity of tree/shrub and crops were negatively influenced by proximity of farms to urban areas.

-Both diversity components were

c dc d d

-Both diversity components were higher at higher altitude.


Cont…RDA analysis

RDA analysis clearly separated diversity, soilerosion and other explanatory variables.

-The lowland region (region 3) was distinctg ( g )from the others because of minimal

agricultural activities and sparse naturalvegetation.

-Region 5, with the highest altitude, was also separated from the intermediate regionand had high agro-biodiversity, as it waslocated far from towns and roads.

1.0

R1

-Region 1, located close to town, was also somehow separated from the others because of relatively high inorganic fertilizer use.

Var/6yr

ErosionNo.SoilT

crops/yrcrops/6y

SelCr

Fert/kg

Weed_yr

weed_6yrpest ind

SPECIES

ENV. VARIABLES

SAMPLESy g g

-RDA analysis showed agro-biodiversitywas significantly (p<0.001) related to each ofthe explanatory variables, but mainly 0

TotTSppT_Ratio

Var/yrSelCr

Town Dist

Road Dist SAMPLES

Region 1

Region 2

Region 3

Region 4

R i 5

R3 R5

p y , ywith distance to road and town (positively)and fertilizer and soil erosion (negatively).

-1.0 1.0

-1. Region 5


Cont…

Relationship with soil erosion

- Soil erosion (measure of un-sustainability) was positively correlated with inorganic fertilizer use ( r = 0.44; P < 0.001),e e use ( 0. ; 0.00 ),

- Soil erosion was negatively correlated with

-Crop diversity ( r = -0.44; P < 0.001),p y ( ; ),

-Tree/shrub species diversity ( r = -0.74; P < 0.001),

-Crop selection criteria ( r = -0.42; P < 0.001) andp ( ; )

-Animals per farm household ( r = -0.21; P < 0.01)


Conclusion

Higher agro-biodiversity was associated with farms located far from road and towns, often associated with indigenous farming , g gpractices.

Agricultural technology packages (inorganic fertilizer) wereAgricultural technology packages (inorganic fertilizer) were important to increase food production

but were often associated with removal of landraces and i l ( d h b i )native plants (trees and shrub species).

Soil erosion was worse on less diversified farms.

Improved agricultural production should, therefore, take in to account locally available land races and native tree/shrub speciesaccount locally available land races and native tree/shrub species.


Spatial variation in agro-biodiversity, soil degradation and productivity in agricultural landscapes in the highlands of Tigray, northern Ethiopia.


Objective

To compare the spatial and temporal variations in agro-biodiversity and soil degradation in relation to agricultural productivity in Tigray, northern Ethiopia between 2000 and 2005.

Hypothesis

Based on previous research results, we hypothesized agro-biodiversity and crop productivity have declined in recent years, whereas soil erosion has increased.

Aim

The aim was to understanding the drivers of agro-biodiversity loss and soil erosion, and relate them to agricultural productivity.



Study areaStudy area


DatasetFarm Wealth/Social

•Soil type, OM, Avail. P, N•Crops planted/year•Animal manure

•Inorganic fertilizer•Livestock holding•Number of credit sourcesFarmer’s education level•Weed, insect species/farm

•Caloric crop yield & sel.sri.•Farmer’s education level•Farmer’s off farm employment opportunity

Agro BiodiversityAgro-Biodiversity

Soil erosion

•Distance to major road•Distance to major

•Elevation of farms•Slope of farms

town/marketSlope of farms

Development drivers Topographic


Topographic

Data Collection Method

Development driversTopographic

Geog. locationsGPS GPS

Development drivers

Feature delineation Aerial photos

Aerial

photos

Slope/AltitudeAlti t / SRTM

Crossing of fieldsDouble trackTransect walk

Slope/AltitudeAltimeter/ SRTM

Wealth/SocialFarm Field Measurement & interview

Agro-Biodiversity &Soil erosion

Wealth/SocialFarm


Data AnalysisSpatial data Non-spatial data

Development drivers

Topographic

Field data

Interviews

Agro-biodiversity

LULC classes

Farm characteristics

GIS

Socio-economicdata

GIS

Overlay Analysis

GIS output mapsData integration

Statistical Analysis- Discriminant analysis (2005) andComparison between 2000 & 2005

Outcome

Co pa so be wee 000 & 005for agro-biodiversity & soil erosion.

- Chi-square (Educat. & Employment

P i d t t t (b t 2000 & 2005)Outcome - Paired t test (between 2000 & 2005)


Results

Status of agro-biodiversity in 2005

I. Agro-biodiversity and social characteristicsg y

-More off-farm employment, less agro-biodiversity (χ2 = 30.8, DF = 4, P=0.001).

-Farmer’s education was not significantly associated with agro-biodiversity.g y g y


Cont…II. Agro-biodiversity and quantitative explanatory variables

For the discriminant analysis, combined (both tree/shrub species and crop diversity) average agro-biodiversity was categorized into three classes:

low (<7), medium (7-12) and high (>12).

- Total N (%), - Available P (mg/kg), - Crop types (number/farm),- Animal manure (kg/ha) and - Crop selection criteria (number/farm)

significantly separated (P <0.05) the agro-biodiversity classesand were positively associatedwith the first canonical function.


Cont…

Caloric crop yield (Mcal/farm)Caloric crop yield (Mcal/farm), Animal ownership (number/farm), Farm distance from the nearest town (km) andElevation (m)

also significantly (P<0.05) discriminated agro-biodiversity classes andwere positively associated with the first canonical function.

Compared to low agro biodiversity classes farms with high agro biodiversity class hadCompared to low agro-biodiversity classes, farms with high agro-biodiversity class had52 % higher available P, 39 % higher total N,47 % more crop types,71 % higher animal manure71 % higher animal manure,53 % more animals,42 % more crop selection criteria and 19 % caloric crop yield.

Inorganic fertilizer use (kg/farm) and credit sources (number/farm) werenegatively associated with the first canonical function,but significantly discriminated the three agro-biodiversity classes.


Cont…

III. Agro-biodiversity and land use types

-Low agro-biodiversity class wasstrongly associated with intensively cultivated land use type (Icu).

-High agro-biodiversity class coincidewith the sparsely cultivated land use type (Scu).


Cont…

Status of soil degradation in 2005

Soil erosion classes were categorizedinto four: low (<10 tons/ha), moderate (10-20 tons/ha), high (20-40 tons/ha) and extremely high (>40 tons/ha).

Farm slope (%) was positively associatedwith the first canonical functionand contributed significantly (p<0.001)t th di i i ti f th il ito the discrimination of the soil erosionclasses.

-The higher the slope of farms,th hi h th il ithe higher was the soil erosion.


Cont…

Soil OM (%) and crop selection criteria (number/farm) were negativelyassociated with the first canonical function but significantly (p<0.001) separated the soil erosion classes.

-The higher OM content of farms, the less soil erosion.

-The higher crop selection criteria per farm, the less the soil erosion.


Cont…

Temporal changes (between 2000 and 2005)

I. Changes in farm and wealth characteristics

Paired t-test comparison between 2000 and 2005 resulted in significant decrease in

- crop diversity (Paired t-test, t = 6.46, P < 0.001, n=151)

- animal ownership (Paired t-test, t = 4.23, P < 0.001, n=151)

- crop selection criteria (Paired t-test, t = 2.05, P < 0.05, n=151)

Whereas inorganic fertilizer increased significantly (Paired t test t 3 40 P < 0 01 n 151)Whereas, inorganic fertilizer increased significantly (Paired t-test, t = -3.40, P < 0.01, n=151)

No significant change for the other variables between 2000 and 2005.


Cont…II. Agro-biodiversity and soil degradation (2000-2005)

Agro-biodiversity (2000-2005)

Agro-biodiversity was compared between 2000 and 2005 and categorized into: decrease (<0),no change (=0) and increase (>0).

-Crop type (number/farm), -Crop selection criteria (number/farm),-Animal ownership (number/farm),

i f ( )-Farm distance from the nearest town (km), and -Farm distance from the nearest road (km)

significantly (P<0.05) separated agro-bi di i h l (dbiodiversity change classes (decrease, no change and increase) and were positivelyassociated with the first canonical function.

Wh i i f ili (k /f )Whereas, inorganic fertilizer (kg/farm)was negatively associated with the firstcanonical function but significantly (P < 0.05) separated the agro-biodiversityh lchange classes.


Cont…

Soil degradation (2000-2005)

Classes for changes in soil erosion (decrease, no change and increase)between 2000 and 2005 were not significantly separated by the explanatory variables.


Cont…Spatial distribution of agro-biodiversity (2000 and 2005)

Tree and shrub species diversity didnot change significantly between 2000 and 2005.

Number of crop diversity decreased(a) No. of tree and shrub species

overlaid with road buffers in 2000

(b) No. of tree and shrub species overlaid with road buffers in 2005 Number of crop diversity decreased

significantly mainly on farmslocated close to the nearest major roads.

Proximity of farms to the nearest townProximity of farms to the nearest town was strongly associated with lowagro-biodiversity (mainly with cropdiversity), both in 2000 and 2005.

(c) No. of crop varieties overlaid with road buffers in

(d) No. of crop varieties overlaid with road buffers in

2000 2005


Conclusion

Significant loss of agro-biodiversity, mainly crop diversity, between 2000 and 2005.

Higher loss of agro-biodiversity was contributed from higher use of inorganic fertilizer d hi h b f ditand higher number of credit sources.

Proximity to towns and roads reduced agro-biodiversity, both in 2000 and 2005.

Agro-biodiversity loss was also facilitated by higher soil erosion.

Higher agro-biodiversity was associated with increased caloric crop yield.

Higher agro-biodiversity was favored by sparsely cultivated land use (with higher trees/shrubs).

The information on the relationships among agro-biodiversity, productivity and soil erosion can improve the understandings on increasing food security while maintaining locally available agro-biodiversity resources.


Assessing the effect of Faidherbia albida (F. albida) based land use systems on barley yield at field and landscape scales in the highlands of Tigray, northern Ethiopia.


ObjectiveObjective

To investigate influence of traditional F. albida based land use systems on barley productivity at field and regional scales in Tigray northern Ethiopiabarley productivity at field and regional scales in Tigray, northern Ethiopia.



Study areaStudy area


RRA & PRATopographic map

F lbid densit

Landscape scale-farm datasetTour, interview &

group discussion

Landscape scale

S l ti b l d ith 77 fi ld

F. albida density

Eucalyptus density

Li k d i

Selected site

Selecting sub-landscape with 77 fields Livestock density

Inorganic fertilizer

Land use system

A. albida +Livestock

A. albida aloneStatistical

analysisM d t ti lLow spatial A. albida

Land use system

Di t f t Livestock

High spatial A. albida

Moderate spatial A. albida

Overlay

1 m from A. albida trunk

2 f A

A. albida +Eucalypt

Mixed model

Multiple i

CCAanalysis

Distances from tree

Added Ecosystem Services(Added barley yield)

Overlay

LULC map

Elevation map



regression

Regional scaleField scale

( y y )Elevation map

ResultsProductivity and land use systems at field scale

Barley yield estimate

Th F lbid l d id dThree F. albida land use systems were considered: F. albida alone (AA),F. albida + livestock (AL) and F. albida + Eucalyptus (AE).yp ( )


Cont…Significantly (P < 0.05) higher barleyyield was found at 1 m from A. albida trunk than 25 and 50 m for the AA andAL land use systems.

In contrast barley yield did not changeIn contrast, barley yield did not changesignificantly with distance from the A. albida trunk for the AE land use system.

1400

1600 a a a

LUS

800

1000

1200

1400

ld (k

g/ha

)

AAAL

ab b

a

b b

a

LUS

200

400

600

800Ba

rley

yiel AL

AE

01 25 50

Distance from A. albida trunk (m)


Cont…

Soil properties (Mixed Model Analysis)Soil properties (Mixed Model Analysis)

Interaction effect of F. albida land use systems (AA, AL and AE) and distance from F. albida trunk was significant for g

- Total N ( P < 0.05),- Available P ( P < 0.001),

Soil moisture ( P < 0 001)- Soil moisture ( P < 0.001).

In all cases, mean values decreased with increasing distance from the tree for AA and AL land use systems, whereas they were more erratic for AE.

OM was significantly affected only by distance from the tree irrespective of the F. albida land use systems.

Stepwise regression analysis showed soil moisture significantly affected barley yield at the AA (P<0.01) and the AL (P<0.001) land use systems.


Cont…a b

GIS analysis

AA and AL land use systems werei l i d i h lmainly associated with sparsely

cultivated and moderately cultivatedland use classes, respectively.

The AE was not clearly associated with distinct land use class.

Ho e er most AE ere associatedHowever, most AE were associatedwith higher inorganic fertilizer useand irrigation practices.


Cont…

Productivity and farm characteristics at landscape scale

Canonical correspondence analysis (CCA)h d l i bshowed clear separation between

barley yield classes and farmcharacteristics (F. albida density,Eucalyptus dominance, Livestockyp ,density and Inorganic fertilizer).

- Barley yield was positively associatedith F lbid densitwith F. albida density.

- Higher yield (Class 3) at higherF. albida density (HA)

- Barley yield was negatively associated with Eucalyptus dominance (HE), locatedclose to townsclose to towns.


Cont…

Land use classes and spatial distribution of A. albida at regional scalep g

Sparsely cultivated land use class (Scu)was strongly associated with farms havingHigh F albida density but low EucalyptusHigh F. albida density but low Eucalyptusdominance.

Intensively cultivated land use class (Icu)was related with low F. albida density and higher Eucalyptus dominance.

50

60

acia

and

nt

20

30

40

farm

s un

der A

capt

us m

anag

eme

HALAHELE

0

10

20

S M I

Perc

enta

ge o

f E

ucal

y p LE

Scu Mcu Icu

Agricultural land use types


Cont…

Added ecosystem service (barley yield benefit)y ( y y )

Higher overall barley yield benefit (100% at E3) in sparsely cultivated land use type (T1).

Removing trees from inside of the field at random until T2 resulted in a reduction in yield benefit from 100 % in E3 to 40 % in E2.

Further removal of trees down to trees at corners of the fields (T3) and complete clearing resulted in less yield benefits.


Conclusion

The research provides field and regional scale integrated study approach to estimate influence of F. albida land use systems on barley productivity.

It indicates F albida trees should be maintained and promoted in and aroundIt indicates F. albida trees should be maintained and promoted in and aroundfarmlands as a way of increasing crop productivity and soil fertility.

Whereas, Eucalyptus trees did not show both in yield and soil fertility improvement.

Land use types with more trees/shrubs contributed to higher F. albida density which in turn was important to enhance added ecosystem service (addedwhich in turn was important to enhance added ecosystem service (added barley yield).


Overall contributions and implications of the research

The research contributes to the understanding of the relationships among agrobiodiversity (agroforestry)-productivity-soil erosion in agricultural landscapes.

- Relative agro-biodiversity (compared to the maximum of tree/shrub species and crop diversity at 151 farms) was positively correlated with crop productivitycrop diversity at 151 farms) was positively correlated with crop productivity.

- In contrast, soil erosion was higher at lower relative agro-biodiversity (at the 151 farms).


Cont…

Despite the contribution of agroforestry/agro-biodiversity to productivity, expansion and intensification of agricultural lands are continuing at the expense of natural habitats over the past 41 years.

-mainly because of increasing population coupled with an increasing demands for food, feed and construction materials.

Removal of natural habitats and on-farm trees/shrubs can lead to deterioration of soil fertility and enhance soil erosion.

How to increase food production to satisfy the demand of the increasing humanHow to increase food production to satisfy the demand of the increasing human population while minimizing loss of agro-biodiversity is a challenge of land use planners and decision makers in the country.

A f ti t i bl i lt l d ti d f d itAs one way of promoting sustainable agricultural production and food security, agroforestry needs to be considered as a natural capital from which agriculture gains ecosystem services such as increase in productivity, soil fertility, protection against soil erosion, water retention, pollination and pest control., , p p


A O !THANK YOU!