April 11, 2023

Implications of coffee obituary notices on tree

abundance and richness on coffee farms of Mt Kenya.

Sammy Carsan

Collaborators': Stroebel A*, Munyi A, Kindt R. Pinard F. and Jamnadass R.

*University of the Free State, South Africa

ICRAF Seminar 11th November 2011

April 11, 2023

1) Background: Evolution of Kenya’s coffee smallholder sector2) Study Context: Why coffee agro-forests?

3) Statement of objectives & hypotheses4) Methods: sampling approach

a) Coffee farms typologiesb) Whole farm tree inventoriesc) Tree diversity analysis

6) Results & DiscussionsConclusions

Presentation outline

April 11, 2023

• Kenya’s coffee previousely estimated to offer livelihood support to over 5 million people directly and indirectly

• Developments summed during and after the ICA market regime

• Living standards, incomes, food security in coffee growing negatively affected post ICA

• Uncertainties in international market and loss of coffee productivity have affected overall coffee profitability

• Are smallholders ready for incentives to shift from traditional cash crop systems? Is AF tree cultivation a good incentive?

• Smallholder enterprise choice will be driven by resource availability and market value for cash crops

Evolution of Kenya’s smallholder coffee

April 11, 2023

Compared to robustas the arabica price differential (premium) is about US$ C 60 (s.d 34.63, n = 129) per pound in the last ten years.

Mean price for Columbian Milds, New York Composite and Robusta coffee from 2000 to 2010

Source: ICO Statistics, 2010

arabica

robusta

April 11, 2023

Kenya’s coffee exports fell by over 50% between 2000 and September, 2010; world market share declined from 3.1% in 1986 to 0.6% by 2006 (ICO, 2010).

Columbian mild coffee exports by: Columbia, Kenya and Tanzania

Source: ICO statistics, 2010

KenyaTanzania

Columbia

April 11, 2023

2001/02 2002/03 2003/04 2004/05 2005/06 2006/07

Area in Hectares (Ha) ‘000

Cooperatives 128 128 128 128 128 120.7

Estates 42 42 42 42 42 42

Total 170 170 170 170 170 162.7

Production (tonnes) ‘000

Cooperatives 28.8 34 30 25.5 27 28.4

Estates 23.1 21.4 18.5 19.7 21.3 25.

Total 51.9 55.4 48.4 45.2 48.3 53.4

Average yield (kg/Ha)

Cooperatives 198.8 265.8 234 199.2 211.3 235

Estates 537 509.9 439.8 469 506 595

Coffee production in Kenya year 2001-2007 (Source: CBK, 2008)

April 11, 2023

• ‘Shaded’ coffee as opposed to open ‘sun’ coffee’-a more sustainable production approach (Mas and Diestch, 2004). Coffee AF systems act as reservoirs of indigenous tree species (Perfecto et al., 2005). Trees yield complimentary products e.g. fruits, timber and firewood which diversify, diet and stabilize farmer incomes

• Peeters (2003) :coffee shaded with any density of Cordia alliodora has better benefit-cost ratio than un-shaded estates although yields were lower. Simplifying these systems was disadvantageous even if coffee production increases

• Structurally complex habitats support more diverse fauna (Garcia et al., 2009).

• Coffee AF seen as an approaches to build alliances between ecologically sustainable agriculture and conservation efforts in protected areas. Trees contribute ecological services similar to those provided by forest e.g. soil protection, nutrient cycling, water retention and carbon capture (Chazdon et al., 2009)

• Farmers benefit culturally by maintaining biological diversity that ensure productivity (Lengkeek et al. 2005)

• Genetic diversity helps farmers to manage their inputs in more efficient ways- e.g. a mix of fast growing and slow growing timber grown for different markets; fruit species with different fruiting phenology to contribute to HH food security (Dawson et al., 2009)

Why coffee agro-forests?

April 11, 2023

Research challenges…• Not clear the extent to which farmers are willing to conserve tree diversity given constraints

e.g. land (Lengkeek & Carsan, 2004)

• It is challenging to study all the factors affecting biodiversity simultaneously

• Impact of converting natural forests to different AF have not been directly compared for many agricultural landscapes (Fitzerbert et al., 2008; Asase and Tetteh, 2010)

Characterizing farm tree demographics is useful for identifying shortcomings that may underlie tree based systems- An understanding of the structure and densities of tree population on could help determining the viability of trees on agricultural landscape even for genetic resource supply and conservation.

April 11, 2023

(I) Globally, plant diversity in large areas is positively related to increasing productivity

(II) Regionally, diversity in small plots is negatively related to increasing productivity

(III) Evaluation of large or small scale effects differ according to species and given field conditions.

Source: Purvis and Hector (2000)

Plant diversity can be related to productivity at: global, regional and plot level

April 11, 2023

Study objectives• To investigate agroforestry tree species richness and abundances on

smallholder coffee farms showing differences coffee production behaviour such as increasing, decreasing or constant yield trends

• To determine tree diversity assemblages maintained under different coffee agro-ecological zones around Mount Kenya

Hypothesis:

H0 :Farms with decreasing coffee production (yields, density of bushes) support higher levels of tree abundances and richness on farm

H1 :Farms with increasing coffee production (yields, density of bushes) have decreased tree abundances and richness on farm

April 11, 2023

• Cross-sectional survey in three coffee districts of Mt Kenya (Meru , Embu & Kirinyaga)

• The zones are comparable on coffee and other crops production practices and largely representative of smallholder coffee systems in Kenya

• The regions have strong farmer organization by cooperatives and societies

Research Methods

April 11, 2023

• Stratified random sampling used to obtain a representative sampling frame (Stern et al., 2005)

• 10 Farmer Cooperative Societies (FCS) selected through key informant discussions to cover upper and lower coffee zone in 3 target coffee districts

• Farmers selected per society based on cherry deliveries in the last 8 years (2000-2008)

• Farmer produce records was used to cluster sample farmers in three categories: “increasing”, “decreasing” and “constant”

• 5 farmers most fitting each prescribed category were picked selecting 15 farmers per society /factory level

• Later 2 farmers selected per category for HH interviews and farm assessments (due to survey logistics, resources)

Sampling strategy

April 11, 2023

Farms sampling strategy

April 11, 2023

Functional coffee farm typology

constantdecreasing

increasing

April 11, 2023

Field methods• Ground based methods used to

enumerate tree species presence on coffee farms

• Trees were defined as all woody perennials growing to over 1.5 m tall, including exotics (Beentje, 1994; Brown, 1997).

• Tree basal area (tree cross-sectional area measured at breast height) undertaken

• All trees ≥5 cm DBH measured

• Local/common names of trees recorded from local farmer consultations

• All trees were identified to species level according to Beentje (1994) or Maundu and Tengnäs (2005).

April 11, 2023

• Diversity refers to the number of species that can be differentiated, and to the proportions (or relative abundances) of the number of trees in each species.

- diversity refers to both richness and evenness

Source: Kindt and Coe (2005)

Tree diversity analysisIncreasing richness

Increasing evenness

April 11, 2023

• Farm tree assemblages analyzed following BiodiversityR (Kindt & Coe, 2005; R Dev’t Team, 2010)

• Tree abundances and basal area distributions calculated to assess structural composition of current farm tree population (Jongman et al., 1995)

• Diversity indices, species accumulation and rank abundance curves used to compare species richness and evenness

• Rènyi diversity ordering techniques used to rank tree communities from low to high diversity (Legendre & Legendre, 1998; Kindt et al., 2006)

• Renyi profile values (Hα) are calculated from the frequencies of each component species (proportional abundances (Pi ) = abundance of species i/total abundance) and a scale parameter (α) ranging from zero to infinity (Tothmeresz, 1995; Legendre and Legendre, 1998):

 

• Simple linear regressions were used to regress tree diversity measurements against coffee farm categories & agro-ecological zones

Farm tree diversity analysis

April 11, 2023

ResultsSmallholder farms and households characteristics

HH variables

Farm Categories

All Farms(s.d; n)

Increasing(s.d; n)

Decreasing(s.d; n)

Constant(s.d; n)

Farmers age 57 (10.99;61) 58(13.90;57) 60(15.02; 60) 58(13.38;178)

Family size 5.4(2.17;61) 4.8(2.20;59 4.9 (2.47;62) 5.02(2.29;182)

Farm size (Ha) 1.4(0.97;61) 1.1( 0.84;59) 1.3 (1.27;62) 1.2(1.05;182)

AFT Ha-1 182(99;61) 202(152;59) 227 (222;62) 204(166;182)

TBA Ha-1 2.76(1.69;61) 2.93(3.00;59) 2.81(1.71;62) 2.83(2.19;182)

Coffee bushes Ha-1 596(495;60) 578(697;59) 496 (377;61) 556(536;180)

Cherry kg Ha-12544.6

(1943.9;60)1682.8

(1743.6;59)2185.8

(2054.8;61)2140.5

(1941.7;180)

Cherry val. Ha-171331.1

(55717.38;61)47960.8

(49693.76;59)62295.2

(58560.30;61)60667.89

(55371.4;181)

Banana val. Ks Ha-129675.7

(34682.5;61)58416.6

(106229.7;59)53498.6

(80646.6;61)47072.9

(79773.8;181)

Maize val. Ks Ha-115746.8

(13196;.62)21248.3

(18096.5;59)19933.6

(16999.3;61)18951.1

(16292.2;181)

TLU Ha-1 3.9(2.9;60) 4.3(4.9;59) 5.4(5.5; 61) 4.5(4.6;180

TLU dairy -1 3.84(2.08;53) 3.36(1.68;46) 3.68 (4.23;53) 3.64(2.92;152)

Milk val day-1 211.5(178.3;48) 147.8(109.1;37) 152.2(114.1;47) 172.5(141.87;132)

TLU=tropical livestock units; AFT= agroforestry trees; TBA=tree basal area; s.d=standard deviation; Val in Ks, 1 US$ = Ks 80

April 11, 2023

Just how much coffee and trees is

present on farms…?

• 75% (156) farms cultivate 250-750 bushes ha-1

• 61%(110) produce 1000-2000 kg cherry ha-1 yr-1

• 41% (75) farms, tree density : 100-200 trees ha-1

• 30% (54) farms: TBA class of 1.1-1.9 m2 ha-1

• 22% (40) farms: TBA class of 2.0-2.9 m2 ha-1

• 35% (66) farms : TBA class of 3-5 m2 ha-1

• Average tree volume : 36.31 (31.1-41.5 ) m3 ha-1

April 11, 2023

Tree species richness and abundance by farm typology

Farm category (n) Farm size Ha. (s.d)

Total richness (mean)

Total abundance

(mean)

Shannon index

Inverse –Simpson index

Constant (60) 1.18 (1.2) 110 (14.4) 10,079 (168) 2.59 4.93

Decreasing (60) 1.12 (0.9) 145 (17.9) 11,149 (186) 2.79 5.49

Increasing (60) 1.37 (0.9) 141 (18.4) 14,592 (243) 2.72 5.51

All farms (180) 1.22 (1.0) 190 (16.9) 35,820 (199) 2.76 5.4

• Tree richness by farm categories: Decreasing>Increasing>Constant (corresponds to Shannon Index)

• Richness within farm in category: Increasing>Decreasing>Constant (corresponds to Inverse Simpson Index)

• Tree abundance: Increasing>Decreasing>Constant

April 11, 2023

• Sample based accumulation curves and Rènyi diversity profiles for the 3 coffee farm types showed overlapping richness patterns for the coffee increasing and decreasing farms

• The constant farm types have smaller species richness, and slightly higher species un-evenness

• Rènyi diversity indices: similar proportions (40%) of the most abundant species for the increasing and decreasing farms. The constant category had higher proportions (43%) of most abundant sp

Species accumulation curves (i) and Rènyi diversity profiles (ii) by farm typology

constantconstant

April 11, 2023

Tree diversity analysis by coffee agro-ecological zones

AEZ Speciesrichness (H0)

Shannon index(H1)

Inverse Simpson

(H2)

Proportion (%) of most dominant species (H∞)

UM1 (n=70) 98 2.56 4.23 0.38

UM2 (n=68) 110 2.78 4.17 0.40

UM3 (n=42) 129 2.57 5.69 0.46

All farms (n=180) 190 2.76 5.40 0.41

Species accumulation curves : UM 3>UM2>UM1

Intersecting Rènyi profiles

April 11, 2023

Linear Regression Analysis

• Tree richness and abundance significantly different by AEZ (P< 0.001)

• Strong evidence that UM3 tree density is different from those in UM1. Abundance in UM2 is however not significantly different from UM1 (P = 0.593)

• Indigenous trees abundance in UM3 significantly different (P<0.001) from UM1 There was weak evidence on differences between UM1 & UM2 (P = 0.096)

• Exotic trees abundance regressed on coffee AEZ returned a weak model (P = 0.048). There was not strong evidence (P = 0.027) that exotic tree abundance in UM3 was different from UM1

-no evidence that on average exotic tree population in the zone UM2 are different from UM1 or UM3

April 11, 2023

(I) (II)

Rank abundance by tree counts (Relative density) Rank abundance by tree basal area (relative dominance)Rank Species Abundance Proportion

(%)Rank Species Total basal

area (m2)Proportion

(%)1 Grevillea robusta 14923 41 1 Grevillea robusta 223.3 41.92 Eucalyptus sp. 2877 7.9 2 P. americana 40.9 7.73 Macadamia sp. 2445 6.7 3 Mangifera indica 37.7 7.14 Mangifera indica 1402 3.9 4 Cordia africana* 37.5 7.15 Cordia africana* 1086 3 5 Eucalyptus sp. 28.8 5.46 Carica papaya 1059 2.9 6 Macadamia sp. 26 4.97 P. americana 969 2.7 7 C. macrostachyus* 12.7 2.48 Catha edulis* 921 2.5 8 B. micrantha* 9.1 1.79 C. lusitanica 920 2.5 9 C. lusitanica 9.1 1.710 B. micrantha* 722 2 10 Vitex keniensis* 8.4 1.6

Species rank Species rank

Abun

danc

e

Tree abundance and basal area distribution (excluding coffee)

April 11, 2023

Tree species richness by girth class for the three farm categories surveyed

Girth class (cm)

Species richness (tree individuals)

Constant Decreasing Increasing All classes

< 10 99 (7,000) 121 (6,194) 116 (7,374) 160 (20,568)

10.1 – 20 70 (2,864) 79 (2,249) 80 (3,466) 117 (8,579)

20.1 – 30 59 (1,435) 58 (1,325) 70 (1,735) 103 (4,495)

30.1 – 35 32 (417) 23 (267) 34 (471) 49 (1,155)

> 35 40 (471) 42 (492) 49 (635) 64 (1,598)

All categories 121 (12,187) 142 (10,527) 143 (13,681) 190 (36,395)

Farm tree population structure…

April 11, 2023

Simple linear regressions against coffee farm types…

Independent variables measured per farm

P-value Decreasing Increasing

Coffee bushes 0.0004 NS SF***

Cherry kg yr-1 0.0002 NS NS

Tree richness 0.295 NS NS

Tree abundance 0.3939 NS NS

Trees in coffee plot 0.0555 NS SF*

Exotic trees abundance 0.0857 NS SF*

Indigenous tree abundance 0.3993 NS NS

Farm size (Ha) 0.367 NS NS

Manure kg 1 yr-1 0.0345 NS SF*

Fertilizer kg yr-1 0.0003 NS SF***

April 11, 2023

Discussion points• Like in natural ecosystems the type of ecological zones have a major influence on

species richness since abiotic factors such as precipitation, temperature, altitude play a key role

• Data is consistent on rarity of indigenous species compared to exotics for all coffee AEZ. Exotics richness is not different between zones

• Significantly different farm types in a same coffee ecological zone showed no evidence of difference in species richness and only a slight difference in tree abundance

• The marginal coffee zone (UM3) on average has higher species richness but is also most uneven; main coffee zone (UM2) do not contain significantly different species from the UM1 and UM3 indicating likelihood of species sharing

• The ‘coffee increasing farms’ are on average of lower species richness, but higher abundance than the decreasing farms-implies that increased productivity results in competition that displace some species

• Tree size diversity is uneven among the surveyed tree population; the ten most highly ranked species are mainly exotics. Current data have showed likely displacement of some exotics e.g. avocado and indigenous species e.g. cordia, vitex and croton)

April 11, 2023

• Structuring patterns of species diversity to most important factor structuring tree assemblages can help reveal patterns of variations

• Results suggest limited differences exist in tree richness and (abundance) among coffee farm typology

• Low densities of indigenous species on farm could pose a challenge to genetic resources provisioning within highland coffee farming landscapes

-implications on tree reproductive ecology

• Stagnated/ reduced coffee production has potential to reduce tree abundance and richness on farm; increased coffee production does not necessarily lead to more richness but certainly results in higher tree abundance on farm

Conclusions