ARTICLE IN PRESS
ECOLEC-03506; No of Pages 14
Ecological Economics xxx (2009) xxx–xxx
Contents lists available at ScienceDirect
Ecological Economics
j ourna l homepage: www.e lsev ie r.com/ locate /eco lecon
1
2
3
4
5
6
7
8
9
10
11
12
13
1546
17
18
19
20
21
22
42
41
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Q2Analysis
Regionalization of climatic factors and income indicators for milk production
in Honduras
Peter Lentes a,⁎, Michael Peters b, Federico Holmann b
a International Centre for Tropical Agriculture (CIAT), Tegucigalpa, Honduras
b International Centre for Tropical Agriculture (CIAT), Cali, Colombia
F
⁎ Corresponding author. Present address: Alley 67, Ho
Hanoi, Vietnam.
E-mail address: geo.lentes@gmx.de (P. Lentes).
0921-8009/$ – see front matter © 2009 Published by E
doi:10.1016/j.ecolecon.2009.09.001
Please cite this article as: Lentes, P., et al
Ecological Economics (2009), doi:10.1016/a b s t r a c t
O
a r t i c l e i n f o23
24
25
26
27
28
29
30
31
32
33Article history:
Received 6 March 2009
Received in revised form 29 August 2009
Accepted 1 September 2009
Available online xxxx
Keywords:
Climate
Dry season
Honduras
Livestock
Regionalization34
35
36
37The temporal and spatial distribution of dry and wet seasons is drastically limiting forage and agricultural
production in Honduras. A regional overview on how these patterns influence the income of different types of
milk producers was non-existent and would be a beneficial tool for targeting policies and development
interventions. Thispaper examines the regionalized incomesderived frommilkproductionby relatingdry season
length to milk production parameters for dairy farms. Cattle farms were assessed using two samples. Milk
production in the dry andwet seasonswas characterizedbymonthlynet income frommilkper cow. SampleA (97
farms)was classified according to a) herd size classes and b) performance in dry seasonmilk production. Sample
B (30 farms) assessed advanced farms that used more forage technologies than the others.
The income from milk was related to environmental conditions by means of a countrywide map based on dry
season length. Themapwas created by estimating thewater balance for eachmonth in a GIS. Yearly income from
milk/cow was regionalized for the farm classifications and combined with agricultural census data.
Results of the GIS analysis show a detailed zoning of dry season length and yearly income per cow from milk.
Climate-income maps quantify the income ranges of the examined groups of farms.
Climate change models predict temperature rise and decreasing precipitation for Honduras. In view of these
trends the results can be used for an interpretation of farm vulnerability and resilience to climate change.
ED PRO
38
use 32 a, To Ngoc Van, Tay Ho,
lsevier B.V.
., Regionalization of climatic factors and inc
j.ecolecon.2009.09.001© 2009 Published by Elsevier B.V.3490T
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
771. Introduction
Large parts of Honduras are characterized by a prolonged dry
season, varying in length between the moist zones of the North,
seasonally dry livestock zones in the center and the dry South. This
temporal and spatial seasonality is limiting forage and agricultural
production gradually and as a consequence, the income of farmers
depends on climatic conditions. Thus an interdisciplinary research
approach is needed when it comes to relate specific climatic
conditions to economic indicators for milk production.
Detailed information on climatic patterns in Honduras is important,
because Central America's milk production in the dry season is about
40% lower than in the rainy season, when feed resources from green
pasture are abundant (Argel, 1999; Holmann, 2001). Low quality and
quantity of feed aswell as the low genetic potential for milk production
of the commonly used dual-purpose cattle (i.e. cattle for beef and milk
production) lead to the sharp decline in milk production during the dry
season. (Suttie, 2000; Fujisaka et al., 2005).
UNCORREC78
79
80
81
82Farmer's live histories tell that milk production systems in
Honduras mainly originated from extensive ranching systems. In the
past when land was abundant in Honduras ranching enabled farmers
to cope with difficult ecological conditions of prolonged dry seasons.
In ranching, the use of labor is considerably less intensive than in
other agricultural land use purposes (Williams, 1986). However, over
the past years a high demand for dairy products has resulted in a
general change of farming systems from ranching, with its primary
product beef, to increased importance of milk production. Between
2001 and 2003, milk production in Honduras lagged 14% behind
consumption (FAO, 2005) and projections to 2020 foresee an annual
growth of milk demand by 2.9% for developing countries (Delgado,
2005). Such conditions may be an opportunity for smallholder farmers
to increase their incomes but low market participation (Kyeyamwa,
et al, 2008) and the technological level of their production temper the
optimism.
Yet, cost efficient milk production under the given climatic
conditions of Honduras is much more demanding than ranching. So
far, many farmers have shifted to milk production but did not yet fully
account for this in herd management and feeding strategies. In both,
traditional andmodern farming systems of Honduras, the profitability
of milk production depends on climatic factors. Moreover, for many
farms the income from milk sales provides the only continuous cashome indicators for milk production in Honduras,
ARTICLE IN PRESS
2 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
83 144
84 145
85 146
86
87 114478
88 149
89 150
90 151
91 152
92 153
93 154
94 155
95 156
96 157
97 158
98 159
99 160
100 161
101 162
102 163
103 164
104 165
105 166
106 167
107 168
108 169
109 170
110 171
111 172
112 173
113 174
114 175
115 176
116 177
117 178
118
119 179
120 180
121 181
122 182
123 183
124 184
185
125
186
126
127 187
128 188Q3
129 189
130 190
191
131 192
193
132 194
133 195
134 196
135 197
136 198
137 199
138 200
139 201
140 202
141 203
204
142 205
143 206
C
flow, which allows investing in other farm activities such as the
cultivation of cash and subsistence crops, general improvements of
the livestock system, the adoption of improved forage options or the
improvement of cattle breeds.
Regionalization of income disparities is able to efficiently visualize
and present the complex situations. Policy and development inter-
ventions can be planned easier when the situation and possible
impact of changes is modeled spatially.
Only few papers used regionalization for the case of Honduras and
none of them related climatic factors to the income from milk. Jansen
et al. (2006) used a combination of biophysical factors to regionalize
livelihood strategies of rural families in Honduras. Land use change
models (Munroe et al., 2002) were set up linking panel data on land
cover changes derived from satellite imagery to socio-economic
conditions.
To assess regional trends, specific socio-economic indicators need to
bemade available across larger regions, however data collection is often
restricted to surveys in limited study areas. Regionalization of socio-
economic data tackles these scale related constraints by taking into
account that farms act in their spatial setting which is determined by a
sum of conditions, making up the frame for production (Lentes, 2004,
2006). Many of these factors are physical site conditions, like climate,
soil quality, terrain, slope and water availability throughout the year.
Regionalization makes use of the interplay between economic and
ecological systems, following the assumption that indicators derived
from surveys are similar in other areas with similar physical site
conditions. For example an income indicator for milk production in the
dry season that was assessed in an area with prolonged dry season can
be used to represent income in other areas with similar ecological
constraints. The site conditions, represented by spatial variables, can be
used for regionalization if a dependencywith socio-economic indicators
can be established. Then, socio-economic indicators can be extrapolated
to the coverage area of the spatial variables.
Against this background the objective of this paper is to relate the
effect of dry season length to the income frommilk per cow for farms of
distinct cattle herd sizes andperformance classes and to regionalize these
data. A further objective is to demonstrate how the average income from
milk to be expected in a department can be assessed when agricultural
census data, dry season length and survey results are combined.
The spatial spread of the profitability of dairy production is
mapped and enables regional targeting of forage options considering
specific groups of farms in the regions.
2. Material and Methods
The approaches for regionalization presented in this paper use the
length of the dry season as a spatial and temporal variable. The returns
frommilk during the dry andwet seasonswere assessed on 127 farms to
distinguish socio-economic systems and for the approximation of the
yearly incomedependingondry season length (Lentes et al., 2006, 2007).
2.1. Climate Data Generation and Water Balance
The minimum of meteorological data required for setting up a
water balance model consists of monthly mean temperatures and
mean monthly rainfall (Schöninger and Dietrich, 2003).
Available climate datasets (Mitchell and Jones, 2005; CGIAR-CSI,
2006) are designed for continental scale analyses andare thus too coarse
for the requirements of this study. Although the Ministry of Natural
Resources and the Environment of Honduras SERNA (2005) published a
map of annual rainfall for Honduras, itwas not available in a processable
form and further data gaps on monthly mean temperatures had to be
filled. This was achieved by combining three data sources, which are:
1. Climate station data provided from SERNA (2007a,b) and the
national meteorological institute (SMN, 2007).
UNCORREPlease cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.0012. Digital elevation data accessible from CSI-CGIAR SRTM.
3. Climate data generated for 412 points with databank included in
the software MarkSim.
MarkSim is a computer tool that generates simulated weather data
for crop modeling and risk assessment for the tropics. “MarkSim
works from a set of interpolated climate surfaces to fit a Markov
model to the estimated climate data. It uses a third order model with a
special stochastic resampling of the model parameters to realistically
simulate the rainfall and temperature variances for almost anywhere
in the tropics.” (Jones, 2001). For a good estimation, MarkSim requires
the coordinates of the point and its respective elevation information.
Elevation information was obtained from a digital elevation model
(DEM), (CGIAR-CSI, 2004). This DEM has a resolution of approxi-
mately 90 m and the inherent error of the elevation information is
specified not to exceed 16 m.
To provide simulated weather data for Honduras, a set of 383
points, which corresponds to the resolution of MarkSim's climate grid
surface was generated, using GIS. To represent the area around these
points, the mean elevation inside an 8200 m buffer was calculated
from the DEM. For areas with stee^p gradients of rainfall and
temperature, 29 additional points were selected and fed to the
climate model. The output of MarkSim was made accessible for
calculations with spreadsheet software by means of a small
application.
The model results were compared to data, which was available
from the meteorological stations of SERNA (2007a,b) and SMN
(2007), using their locations and altitudes as model input. For mean
monthly rainfall this was done for 17 stations. Measured mean
monthly temperature data are scarce. Only six stations measure
temperature but linear correlations betweenmeasured and simulated
temperatures are highly significant and were used to correct the
model output.
The dataset of mean monthly temperature and rainfall for 430
points contains:
• 412 input points for MarkSim (383 regularly spaced and 29
additional), with mean monthly rainfall and corrected mean
monthly temperatures.
• 7 points frommeteorological stations withmeasuredmeanmonthly
temperature and mean monthly rainfall.
• 11 points from stations with measured mean monthly rainfall and
MarkSim generated and corrected temperature data.
2.2. Dry Season Length Approximation
The length of the dry season is the period in which evapotrans-
piration (Et) exceeds precipitation i.e. the period in which the amount
of available water is not sufficient for the growth of vegetation. To
enable the dry season assessment for livestock holders pastures were
selected as reference plants for dry season assessment. For compre-
hensive descriptions of methodologies to estimate evapotranspiration
and definitions for the water balance, see Allen et al. (1998), or
Schöninger and Dietrich (2003).
The empirical method of Thornthwaite (1948, cited in Schöninger
and Dietrich 2003) was applied for the countrywide Et assessment,
creating calculation routines in Excel and applying them to each
location for which the climate data was generated. The Thornthwaite
method copes with the minimum data requirements, relying on
empirical relations between reference evapotranspiration and air
temperature, based on measurements from various climate zones
(Schöninger and Dietrich, 2003). Other methods for evapotranspira-
tion calculations, like the FAO Penman-Monteith (Allen et al, 1998)
require data of meteorological elements not available for many
development countries (Pereira and Pruitt 2004). It is known that the
Thornthwaite method tends to underestimate Et0 under arid
TED PROOFc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 3
207 264
208 265
209 266
210 267
211 268
269
270
271
2123 272
273
274
2154 275
276
2176 277
278
279
2189 280
281
220 282
221 283
222 284
223 285
224 286
225 287
226 288
227 289
228 290
229 291
230 292
231 293
232 294
295
296
2334 297
235 298
236 299
237 300
301
302
2389 303
240 304
241 305
242 306
243 307
244 308
245 309
246
247
248 310
249
250 311
251 312
252 313
253 314
254 315
255
256 Table 1 t1:1
Herd size definition, according to the number of cattle per farm.
257
t1:2
258 Herd size category Number of cattle per farm t1:3
259 Very small 1–9 t1:4
260 Small 10–19 t1:5
261 Medium 20–49 t1:6
Large 50–99 t1:7
262
Extra large N100 t1:8
263conditions (Pelton et al., 1960; Stanhill, 1961) and that it over-
estimates Et0 under the equatorial humid climate of the Amazon
region (Camargo et al., 1999). Those studies mainly focused on daily
Et0 estimation. Since only monthly averages were used for the
regionalization, the inaccuracy of the method was tolerated.

 
10T a
Et0 = 16c i ð1Þ
I
N
I = ∑ ðT 5Þ1:514i = ð2Þ
i=1
a = 6 7 107I3: ⁎  7 5 2
:71⁎10 I + 1:79⁎102 + 0:49 ð3Þ
c = ðd = 30Þ⁎ðh= 12Þ ð4Þ
Where:
Et0 reference evapotranspiration mm per month
Ti mean surface air temperature in month i (°C)
I heat index defined in Eq. (2)
a in Eq. (1) is a function of the heat index (I)
c correction factor for month length and daylight duration
Eq. (4)
d length of month in days
h hours of daylight at the 15th of the month.
To obtain crop specific evapotranspiration (5) (Etcrop), Et0 was
corrected using a crop specific correction factor (Kc). For the scope of
this study the Kc for rotated grazing land higher than 15 cm, as
provided by (Allen et al., 1998) was used.
Etcrop = Et0⁎Kc ð5Þ
Water surplus (6) is the difference between rainfall and evapo-
transpiration of the respective land cover. Whenever water surplus
was negative, the month was defined as dry.
Water surplus = Rainfall Etcrop ð6Þ
Formulae (1)–(6) were applied to the mean monthly rainfall and
temperature data of the 430 sample points that cover Honduras.
Kriging interpolation was used to fill the information gaps between
points for which climate data were generated. Thus it was possible to
create climate and dry season length surfaces from the sample points.
Kriging interpolation is a linear estimation procedure introduced
by Matheron (1963). In Kriging the value of the variable at the
location of estimation is calculated from the weighted mean of the
surrounding sample points. The weights of the sampled points are
calculated to perform optimally to reach the smallest variance in the
estimation error. For the interpolation, the Kriging plug-in of Boeringa
(2000) for ArcView GIS was used. The grids were calculated
considering the variance of the 12 neighboring sample points and
their distances to the point of estimation. A linear trend in the sample
data was assumed for the model.
2.3. Sampling and Calculation of Socio-Economic Indicators
The data used for this paper were collected by means of a
comprehensive socio-economic questionnaire, which covered all
parts of the farming system (e.g. family members, education levels,
employment, land use inventory, perennial and annual crops,
pastures, cut-and-carry forages, forage cultivation, forage conserva-
tion, beef production, milk production, poultry and off-farm work).
This enabled to take into account the diverse structures of farms and
the different feeding strategies.
UNCORRECPlease cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001The total number of cattle farms in Honduras is reported to be
86,829. Their main focus of production lay in beef (5.8%), milk (44.2%),
beef and milk (33.5%) and others (16.5%) (INE, 2001). The sampling
plan applied for the collection of micro level farm data covered two
study areas in representative zones in the departments of Olancho
and Yoro. These study areas were selected after consultation of local
experts to be typical in terms of herd composition andmanagement in
parts of Honduras with prolonged dry seasons. The income indicators
used for regionalization were assessed in 2005 and 2006 from the two
sub-samples A and B.
In sub sample A the economic conditions of the typical livestock
holderwere assessed for randomly selected farms. The sample covers
69 farms in Olancho and 28 in Yoro.
For sub-sample B, 30 farms, referred to in the text as positive
deviances were selected using expert knowledge provided by local
extension staff. In this study, the term positive deviance does not
exclusively mean “success story”, as it is used by Biggs (2008). On these
farms adoption of diverse forage options is more obvious than on the
typical farms fromsampleA.However, the advanced use of forage options
did not necessarily mean that the farms took full advantage of the
technologies adopted and that this would translate into higher income.
Forage technology adoption is seen as a necessary entry point for
cattle farms to improve resource use efficiency but not as the sole
technology necessary to reach an integrated development of the
farms. Extra large farms (N100 cattle head) were not accepted as
positive deviances, because the availability of financial resources was
not comparable to the typical Honduran farm.
The emphasis of this paper lies on the dairy enterprise, yet other
parts of the farming system (beef and crops) were also considered in
order to characterize the systems and to highlight the importance of
milk production.
To obtain the net income of a production system, all production
costs were deduced from the gross income. Production costs include
all purchased inputs and farm inputs, costs for renting machinery,
services and the opportunity cost of family labor. This means that the
income for each personworking on the farm is valuedwith equivalent
wages like the wages paid for hired labor.
The indicators net income per cow from milk for the dry and for
the wet season was chosen to measure the performance of the dairy
enterprise in both seasons. Another indicator, the production cost per
liter of milk in both seasons was used to underline the cost of milk
production in the groups.
Classifications according to cattle herd size and performance serve
to make farms comparable throughout systems and sizes. Farms from
sample A were classified, compared to each other and to farms of
sample B.
TED PROOF
2.4. Classification Procedure
Two classificationmethodswere applied to farms of sub-sample A:
herd size and performance in dry season milk production.
Table 1 shows 5 herd size classes based on a modification of the
classes used by SECPLAN (1994) and a class of positive deviances,
(sub-sample B) which contains farms of various herd sizes.c factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
4 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
316 374
317 375
318 376
377
319
378
320
379
321
380
322
381
323
382
324
383
325
384
326
385
327 386
328 387
388
329 389
330
390
331
332
391
333
392
334
393
335
394
336
337
338 395
339
340 396
341 397
342 398
343 399
344 400
345 401
346 402
347 403
348 404
349 405
350 406
351 407
352 408
353 409
354 410
355 411
356 412
357 413
358 414
359 415
360 416
361 417
362 418
363 419
364 420
365 421
366 422
367 423
368 424
425
426
427
36790 428
429
430
431
3712 432
433
373 434
C
Performance in dry season milk production was based on the dry
season net income frommilk per cow per month. Performance classes
were defined as follows.
• Very low performers (31 farmers): Cost of milk production
exceeded the revenue.
• Low performers (17 farmers): Positive observations below the
median.
• Medium performers (29 farmers): Observations between the 50 and
80% percentile.
• Top performers (20 farmers): Observations above the 80%
percentile.
The positive deviances (30 farmers of sub-sample B) were
considered separately.
2.5. Regionalization of Indicators
For the regionalization of income from milk production, the
seasonality of the net income plays a crucial role.
For the performance and herd size groups, the indicator net
income from milk per cow per year is the sum of the dry and wet
season income per cow. Where dry and wet season income were
calculated by multiplying the corresponding average income figures
with the number of months in each season.
The regionalization of the average net income per cow per year in
the departments of Honduras used the last complete agricultural
census (SECPLAN, 1994) to determine the share of each herd size class
in the each department. The spatial units of this publication are the
departments. Five years later, INE (2001) published agricultural
statistics for 7 representative regions covering Honduras. This
publication has the disadvantage that the spatial resolution is more
coarse, compared to the 1994 census.
The 1994 Census data were collected before hurricane Mitch. In the
year after the disaster, cattle population had declined to 82.5% of the
1994 population. Annual growth rates are reported to be 2.5% for the
postMitch years between 1999 and 2001. Supposing that from 2001 on
till 2005, the year of the socio-economic survey undertaken for this
study, growth rates have been similar, the livestock population would
have reached the pre Mitch level again by 2005. If we further suppose
that this growthhas not lead to a drastic shift in herd size composition of
farms, the inaccuracy of the data from 1993 can be tolerated. Although
there is uncertainty about this development, the 1993 data are still the
best available information on herd size composition in the departments
of Honduras. To make the analysis more reliable two factors were
considered: a) Instead of using the numbers of cattle reported in the
statistics, only the numbers reported for farms in herd size classes were
used. b) The seven herd size classes of SECPLAN and INE were
aggregated to 5 classes. This was done by merging classes for very
small farms (1–4 and 5–9 cattle) to the class 1–9 cattle and by merging
the classes for the very large farms (100–499 and N500 cattle) to N100
cattle. Together with the result of the productivity assessment from the
farming systems survey, census data were used to extrapolate the
income situation of the dry and wet season from the survey population
to the population of the department.
The department wide average net income/cow/month was
calculated as follows:
N=4 F
Y i
dry = ∑ *I
F idry ð7Þ
i=1 tot
N=4 F
Ywet = ∑ i * I ð8Þ
i=1 F iwet
tot
where:
Ydry Region wide average income/cow/month in the dry season
UNCORREPlease cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001Ywet Region wide average income/cow/month in the wet season
Fi Number of farms in farm size class i
Ftot Total number of farms
Iidry Net income/cow/month of dry season for farm size class i
Iiwet Net income/cow/month of wet season for farm size class i.
Formulae (7) and (8) yield the average income values for the dry
and wet season for each data point (Grid cell). These depend on the
proportion of each herd size class in the department's cattle farmer
population. The region wide average income per month of dry season
was calculated for each department by creating two grid themes: a) the
respective value forYdry for each department andb) the respective value
for Ywet for each department. These two grid themes were processed
with the grid obtained for dry season length to calculate the average
yearly income from milk per cow as described for the herd size and
performance classifications.
3. Results
Results are presented in three sections: a) The assessment of the
dry season length, b) the classification of sampled farms according to
farm size and the performance indicator and c) results of the three
regionalization approaches.
3.1. Dry Season Length
Temperature and rainfall data of stations were compared to the
corresponding result for their locations as generated with MarkSim. A
set of linear regression models, one for each month, was created with
SPSS to correct the MarkSim data with station data. These regressions
on temperature yielded high explanatory qualities in terms of R-
square, since altitude is of major importance when explaining
temperatures. On what concerns the rainfall data, the differences
between the model results and the measured rainfall are on average
small and tolerable.
Fig. 1 shows the annual rainfall distribution for mainland Honduras
as modeled with MarkSim, interpolated and mapped with GIS.
While the north and especially the northwest receive most rain,
the central departments of Honduras are marked by annual rainfall
sums between 1400 and 1000 mm. Moisture islands inside the
territory consist of mountain areas shared between Comayagua and
Santa Barbara, where higher elevations yield more rain and the area
around Lake Yohoa. In the slipstream areas behind the coast parallel
mountain ranges of the North, there is an abrupt drop of annual
rainfall sums. A distinct moisture gradient is to observe in Olancho
from the southwest to the northeast and further throughout the
departments of Gracias a Dios and Colon to the Caribbean coast, where
annual mean temperatures are also higher than inside the country
(Fig. 2). Although favored by high rainfall sums, much of this area is a
protected biosphere reserve and in most of the unprotected part
access is highly limited. In some areas on the Caribbean coast rainfall
sums map turned out not precise, according to field experience. These
estimation errors can be attributed to an edge effect of the
interpolation. Such estimation errors occur along the geographic
margins of the input datasets, e.g. when gradients between the last
measurement points on are steep. Although the edge effect would not
have affected the results greatly, dry season length was adjusted to
surrounding areas using field experience of local experts.
The dry season lengths (Fig. 3 and Table 2) were calculated from
the difference between evapotranspiration, as assessed with method
of Thornthwaite (1948) and the annual rainfall.
Dry seasons shorter than 3 months cover about 15% of the land and
are characteristic for the northern part of the country near the coast,
where elevations are below 200 m (Fig. 3). Short dry seasons inside
the country are characteristic for mountain areas e.g. those shared
TED PROOFc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 5
Fig. 1. Annual rainfall.
435 452
436 453
437 454
438 455
439 456
440 457
441 458
442 459
443 460
444 461
445 462
446 463
447 464
448 465
449 466
450 467
451 468
OOFbetween Comayagua and Santa Barbara, where higher elevations yield
more rain and the area around Lake Yohoa.
In the slipstream areas behind the coast parallel mountain ranges,
there is an abrupt drop of annual rainfall sums and an increase of dry
season length.
About 80% of Honduras was mapped with dry seasons lengths
between 3 and 7 months. In the central departments of Yoro,
Francisco Morazan, Comayagua El Paraiso and in most of Olancho,
as well as in the eastern departments Ocotepeque, Lempira and La Paz
dry a dry season length of 4 to 7 months is most characteristic. Where
the dry season is shorter, cooler temperatures and increasing rainfall
are due to higher elevations.
About 4% of Honduras was mapped with dry seasons longer than
7 months. The driest areas are found in intra mountain valleys, e.g. in
the South of El Paraiso, bordering Choluteca, or in rain shadow-
influenced environments, such as in the South east of Olancho.
Although the South of Honduras shows higher annual rainfall than e.g.
C
Fig. 2. Annual mean
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORRE
the central departments, it has longer dry seasons, because rainfall is
concentrated on short periods of the year, in which heavy rainfall
events occur. This region also has higher temperatures than the
central departments (Fig. 2).
The climate model seems to overestimate dry season length in the
driest parts and it seems to under estimate dry season length in the
wettest parts of the country. However, these over- and undershoots
could not be confirmed by data measured on meteorological stations
and cover a comparatively small area.
From the calculation of dry season length, it can also be assessed in
which months and where consecutive dry months occur. This showed
that the start of the dry season is spatially more variable than the end
of the dry periods. The areas, where the dry season starts first (in
November) are located in the south along a strip oriented from south
east on the border to Nicaragua to the west of the country on the
border to El Salvador and Guatemala. In the central parts of the
country dry season starts between December and January. Where dry
TED PRtemperatures.
c factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
6 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
Q1 Fig. 3. Dry season length and topography.
469 483
470 484
471
472
473
474
475
476
477
478
479
480
481
482
t2:1
t2:2
t2:3
t2:4
t2:5
t2:6
t2:7
t2:8
t2:9
t2:10
t2:11
t2:12
t2:13
t2:14
D PROOF
season is short water balance turns negative from February andMarch
on. In most of the country May and June are the first wet months.485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
C
3.2. Milk Production for Herd Size Classes
Within each herd size class, there was a wide range of
management and general production conditions, like the character-
istics of the land, the genetic potential of the cows, the availability of
improved forages and the knowledge available on the farm to manage
the farm efficiently under the specific circumstances. These differ-
ences made it difficult to characterize herd size groups with an
indicator, because the indicators always included the range of
production conditions of the group. Consequently, indicators on
herd size are subject to comparatively high variation within groups
(Table 3). This was observed clearly on farms with less than 20
animals. Variability was smaller on farms with more than 20 cattle.
ORRE500
501
Table 2 502
Areas and percentage of dry season length classification for Honduras.
503
Dry season length in months Area in Square kilometers Percent of the area 504
0 to 1 17 0.02 505
1 to 2 2244 2.00 506
2 to 3 15189 13.55 507
3 to 4 24527 21.87 508
4 to 5 26968 24.05
5 to 6 26836 23.93 509
6 to 7 11772 10.50 510
7 to 8 2866 2.56 511
8 to 9 980 0.87 512
9 to 10 700 0.62
513
10 and more 35 0.03
514
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCFor more detailed presentation of results from on the farming systems
see Lentes et al. (2006).
Very small and greater herd sizes differedmost in the income/cow/
month (Pb0.01). This was especially striking in the dry season. The
farmswith few cattle generated the lowestmonthly income frommilk
per cow in both seasons. On very small farms, feedwas not available in
sufficient quantity and quality and milk production dropped sharply.
On some farms, commercial concentrates were used to maintain the
cows. Milk production of very small farms was not profitable in the
dry season. Only in the wet season farms with 1 to 9 cattle generated
positive income from milk but this did not compensate the losses
experienced in the dry season.
Small farms generated little income from milk in the dry season
but did not loose on average. In the wet season, small farms generated
about half the income of the other farm size classes but only slightly
more than one third of what positive deviances gained.
The seasonality of income was relevant for all farm sizes (Table 3).
Net income from milk per cow on farms with more than 20 cattle
dropped between 44% and 53% in the dry season. Dry season incomes
per cow were about half the ones in the wet season.
Compared to farms from very small to large, positive deviances
showed a high income from milk per cow in both seasons (highest P-
value 0.053). Their dry season income was comparable to the wet
season income of the farm size classes from 50 cattle upwards. The
income of positive deviances dropped by 23% in the dry season.
The productivity of the milk production systems of very small
farms was the lowest. They earned more from beef than from milk.
Small farms managed to reach a continuous cash flow from their milk
production, which exceeded beef production. Yearly income from
milk of medium size farms was about 3.9 times higher than income
from beef. Among large farms (ranching systems) there were cases
that earned much more from beef than from milk. Extra large farms
TEc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 7
t3:1 Table 3
Income parameters for milk and beef production in herd size classes, Olancho and Yoro in $.
t3:2
t3:3 Very small Small Medium Large Extra large Positive
deviances
t3:4 1 to 9 10 to 19 20 to 49 50 to 99 N100
t3:5 n=16 n=22 n=34 n=16 n=9 n=30
t3:6 A B C D E F
t3:7 Dry season: net income from Mean −7.80 3.14 10.12 11.25 11.95 22.83
t3:8 milk/cow/month Std. Dev. 13.95 20.70 18.60 10.10 7.70 20.62
t3:9 Sig. C**, D***, E***, F*** F**
t3:10 Wet season: net income from Mean 3.47 10.86 21.68 20.95 21.60 29.91
t3:11 milk/cow/month Std. Dev. 18.55 21.19 16.01 9.94 7.06 15.82
t3:12 Sig. C**, D**, E**, F*** C*, D*, F*** F*
t3:13 Net income from milk/farm/year Mean −2.51 528.42 1793.70 3324.82 10134.08 5886.40
t3:14 Std. Dev. 457.41 942.48 1649.00 2683.39 5101.11 4967.22
t3:15 Sig. C***, D***, E ***,F*** C**, D***, D* , E***, F *** E**, F* F*
E***, F ***
t3:16 Net income from beef/farm/year Mean 87.72 300.63 460.61 5240.96 10375.74 1982.17
t3:17 Std. Dev. 136.54 887.60 769.18 11445.17 7733.00 4017.93
t3:18 Sig. D*, E***, F** E***, F** E***, F** E** F***
t3:19 Note: Significance between groups is indicated by letters followed by *Pb0.05, **Pb0.01, ***Pb0.001.
^ ^ ^
515 542
516 543
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
t4:1
t4:2
t4:3
t4:4
t4:5
t4:6
t4:7
t4:8
t4:9
OFwere in equilibrium between the two products, while positive
deviances had a clear focus on milk.544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
5693.3. Milk Production for Performance Classes
To characterize farms four performance classes were built, using
the indicator net income per cow per month of dry season. Groups
differed in the net income/cow/month of dry season (Pb0.01). The
socio-economic and produc^tion conditions of performance classes are
presented in Table 4.
All farms that experienced losses in the dry season were joined to
the class of the very low performers. Even in the wet season, very low
performers did only marginally recuperate their expenses. Wet
season production cost/liter of milk was very high, compared to the
other performance groups (Pb0.001). Milk production of very low
performers was low because few cows of low genetic potential were
milked and cost efficient feed was not available in the dry season. The
low volumes resulted in an under exploitation of family labor force:
Farmers on many very small farms earned less than a worker's salary.
While some low performers had negative yearly incomes but were
close to the breakeven cost, others lost more.
Dry season production costs of milk declined, as the performance
level improved. So did the variability of production costs. Those farms at
the bottom of the performance scale lacked of cost efficient farm feed
and needed higher milk production volumes to produce efficiently.
Farm size distribution in the performance classes differed signifi-
cantly (Pb0.005). Small farms had nearly the samepresence in very low
and medium performers categories (Table 5). A few small farms
performed low (12.7%) and some more performed top (20%). As much
ORREC570
571
572
Table 4
Distribution of herd size classes in performance categories. 573
574
Lowest Low Medium Top 575
% % % % 576
Very small 38.71 17.65 3.45 577
Small 25.81 11.76 27.59 20.00 578
Medium 25.81 29.41 31.03 60.00 579
Large 6.45 23.53 24.14 15.00
580
Extra large 3.23 17.65 13.79 5.00
581
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCas60%of the topperformersweremediumsize farms,while thenumber
of farms from small to large similar in the medium performer's group.
Low and medium performers generated nearly the same net
income/cow during the months of the wet season. Low performers
showed deficiencies in dry season herd management such as
inadequate provision of feed and exaggerated use of purchased
supplements (Lentes et al., 2007). In forage technology adoption,
medium performers were ahead of low performers. Medium
performers used more low-cost farm feed and were better prepared
for the dry season with conserved forage.
Positive deviances lay between medium and top performers in the
income but had comparatively high production costs per liter in both
seasons. The inclusion of positive deviances in the analysis does not
necessarily demonstrate what can be achieved with an appropriate
use of forage technology. The analysis rather showed that an
integrated change of the livestock production system is not yet fully
implemented on these farms. More factors than the availability of
forages have influence on the economic success of dairy production
e.g. the genetic quality of the milking cows (Lentes et al., 2007).
3.4. Countrywide Income Regionalization
The spatial variables used for regionalization were the length of
the dry and wet seasons. For the regionalization of the income
indicators, income values for the categories derived from the socio-
economic sample were used to create income grid surfaces with GIS.
Socio-economic data were collected from areas where the dry season
plays an important role and included a wide range of herd
management practices. The income surfaces approximate what the
income would be if herd composition and management would not
differ substantially between those areas and the rest of the country.
3.4.1. Countrywide Income Assessment for Herd Size Classes
The yearly income from milk per dairy cow was mapped for five
farm size classes and the category of positive deviances. Table 3
presents the income characteristics of the dairy enterprise for these
farm categories.
Table 5 shows the yearly income per cow in relation to dry season
length. Very small farms up to 9 cattle head were usually resource
poor farms, which did not put much emphasis on dry season milk
production. The model designates only areas with dry seasons shorter
than 3 months as zones, in which very small livestock herd owners
could make profit form milk production (Fig. 4). These areas are
TED PROc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
8 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
t5:1 Table 5
Income parameters and costs for milk production in performance groups, Olancho and Yoro in $.
t5:2
t5:3 Very low Low Medium Top Positive deviances
t5:4 n=31 n=17 n=29 n=20 n=30
t5:5 A B C D E
t5:6 Dry season: net income/cow/month Mean −15.31 5.54 14.29 27.08 22.83
t5:7 Std. Dev. 11.57 3.41 2.51 5.41 20.62
t5:8 Sig. B***, C***, D***, E*** C***, D***, E** D***
t5:9 Wet season: net income/cow/month Mean 0.95 19.59 18.57 33.01 29.91
t5:10 Std. Dev. 16.22 10.26 9.61 15.24 15.82
t5:11 Sig. B***, C***, D***, E*** D**, E* D**, E**
t5:12 Dry season: milk production cost/liter Mean 0.53 0.19 0.13 0.09 0.18
t5:13 Std. Dev. 0.34 0.07 0.05 0.05 0.06
t5:14 Sig. B***, C***, D***, E*** C**, D*** D*, E** E***
t5:15 Wet season: milk production cost/liter Mean 0.22 0.07 0.07 0.03 0.10
t5:16 Std. Dev. 0.16 0.06 0.05 0.02 0.05
t5:17 Sig. B***, C***, D***, E*** D*, E* D**, E** E***
t5:18 Net income from milk/farm/year Mean −89.39 3699.43 3273.46 3096.26 5886.40
t5:19 Std. Dev. 545.69 3961.11 4646.34 1531.33 4967.22
t5:20 Sig. B***, C***, D***, E** E* D*, E** E*
t5:21 Note: Significance between groups is indicated by letters followed by *Pb0.05, **Pb0.01, ***Pb0.001.
^ ^ ^
Fig. 4. Yearly income per dairy cow according to dry season length for herd size classes and positive deviances.
Please cite this article as: Lentes, P., et al., Regionalization of climatic factors and income indicators for milk production in Honduras,
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORRECTED PROOF
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 9
582 613
583 614
584 615
585 616
586 617
587 618
588 619
589 620
590 621
591 622
592 623
593 624
594 625
595 626
596 627
597 628
598 629
599 630
600 631
601 632
602 633
603 634
604 635
605 636
606 637
607 638
608 639
609 640
610 641
611 642
612 643
t6:1
t6:2
t6:3
t6:4
t6:5
t6:6
t6:7
t6:8
t6:9
t6:10
t6:11
t6:12
t6:13
t6:14
t6:15
t6:16
t6:17
t6:18
t6:19
t6:20
t6:21
t6:22
t6:23
t6:24
t6:25
t6:26
t6:27
t6:28
t6:29
t6:30
t6:31
t6:32
t6:33
t6:34
t6:35
t6:36
t6:37mainly located in the Northern part of the country and in a few
mountain areas inside the country. For the drier parts of Honduras, the
model estimated losses in milk production for the whole year. Model
results of very small farms for all observed dry season lengths differed
significantly (Pb0.01) to the figures for farms with more than 20
heads and positive deviances.
The owners of small herds of 10–19 animals could produce milk
profitably in all regions of Honduras (Table 6). Corresponding to dry
season lengths small cattle farms earned between 53.13 and 122.61$/
year per cow in milk. Small farms had a lower yearly income from
milk/cow thanmedium farmswhere dry seasonwas between one and
six months long (Pb0.05). This difference was also observed between
small and large farms but only in areas with dry seasons shorter than
4 months (Pb0.05). Small, medium and large farms differed (Pb0.05)
when dry season was between one and four months. The income
difference between small farms and positive deviances was significant
for all observed dry season lengths (Pb0.01).
On those farms with more than 20 cattle, income between the
driest and wettest areas varies between 144 and 249$. As in the
survey results (Table 3) the model did not state dramatic income
differences for those groups with more than 20 cattle. Large and
medium size farms differed significantly (Pb0.05) from positive
deviances in most of the observed dry season lengths.
Income depressions in the dry season are great on very small and
small farms. It is apparent, that a better dry season herd management
would help farms to generate more income per cow. Very small farms
would need to improve their dry season feed base and increase the
number of milking cows to be able to work profitably in areas with
more than 3 months of dry season. When compared to the average
Honduran farmer of different herd sizes below 100, positive deviances
yield the highest incomes per cow per year in the whole country.Table 6
Income from milk/cow/year of herd size classes for observed dry season lengths.
Very small Small
1 to 9 10 to 19
n=16 n=22
A B
1 dry month 30.35 122.61
213.78 249.60
C**, D**, E**, F*** C*, D*, F***
2 dry months 19.08 114.89
205.51 245.66
C**, D**, E**, F*** C*, D*, F***
3 dry months 7.82 107.17
197.84 242.43
C**, D***, E**, F*** C*, D*, F***
4 dry months −3.45 99.45
190.83 239.96
C**, D***, E***, F*** C*, D*, F***
5 dry months −14.71 91.73
184.57 238.25
C**, D***, E***, F*** C*, F***
6 dry months −25.98 84.01
179.13 237.33
C**, D***, E***, F*** C*, F**
7 dry months −37.24 76.29
174.58 237.21
C***, D***, E***, F*** F**
8 dry months −48.51 68.57
171.01 237.89
C***, D***, E***, F*** F**
9 dry months −59.77 60.85
168.47 239.36
C**, D***, E***, F*** F**
10 dry months −71.04 53.13
167.01 241.61
C**, D***, E***, F*** F**
Note: Significance between groups is indicated by letters followed by *Pb0.05, **Pb0.01, **
^ ^
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORREC3.4.2. Countrywide Income Regionalization for Performance Classes
In the countrywide maps (Fig. 5) on the income/cow/year for
performance classes, income is a function of dry season^length and the
dry and wet season incomes for each performance class (Table 7). The
degree to which yearly income depends on the dry season length
differs between the performance groups and is determined by the
difference in incomes between the dry and the wet season.
The maps show, that there is only a small area mapped in
Honduras, where very low performers are predicted to recuperate
costs of milk production. Taking into account that the dry season
length estimation could not validated through measurements for the
wettest and driest parts of the country, the minimum and maximum
values from the grid statistics should only be seen as approximations.
However wettest and driest areas cover comparatively small areas.
Although the income of low performers traces the spatial pattern
of dry season length in Honduras, their income is always positive and
lower than the incomes of top performers and positive deviances
(Pb0.01). The maps (Fig. 5) and Table 7 show that the income range
between areas with short and long dry season is the highest in the
very low and low performers categories. Low andmedium performers
differed in areas with more than six dry months (Pb0.05).
Medium and top performerswere considerably less affected by dry
season length. These groups generated comparatively high incomes in
all areas of Honduras. Under all climatic conditions, the income of
medium performers was lower than of top performers (Pb0.001). Top
performers and positive deviances had similar incomes under all
climatic conditions. Top performers showed differences to the other
performance groups in all climate scenarios (Pb0.05). The differences
between positive deviances and medium performers lost strength for
dry seasons of eight (P=0.052), nine (P=0.069) and ten months
(Pb0.081). PROOF
Medium Large Extra large N Positive
deviances
20 to 49 50 to 99 100
n=34 n=16 n=9 n=30
C D E F
248.63 241.71 249.56 351.83
188.73 114.19 83.14 187.18
F* F*
237.06 232.01 239.91 344.75
186.45 109.86 81.94 186.03
F* F*
225.49 222.30 230.27 337.67
185.30 106.41 81.14 186.39
F* F*
213.93 212.60 220.62 330.58
185.30 103.93 80.77 188.25
F* F*
202.36 202.89 210.97 323.50
186.46 102.48 80.82 191.56
F*
190.79 193.19 201.32 316.42
188.75 102.12 81.31 196.24
F* F*
179.22 183.49 191.68 309.34
192.14 102.85 82.21 202.21
F* F*
167.66 173.78 182.03 302.25
196.56 104.65 83.51 209.35
F*
156.09 164.08 172.38 295.17
201.95 107.46 85.21 217.55
F* F*
144.52 154.37 162.73 288.09
208.24 111.22 87.27 226.70
F*
*Pb0.001.
^
c factors and income indicators for milk production in Honduras,
TED
ARTICLE IN PRESS
10 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
Fig. 5. Yearly income per dairy cow according to dry season length for performance group.
644 662
645 663
664
665
6467 666
667
6498 668
669
6510 670
671
6523 672
673
674
6545 675
656 676
657 677
658 678
659 679
660 680
661 681
ECTED PROOF
Using the average values of income per farm, the regionalization of
income parameters according to climate yields the equations:
Verylowperformers :Y = 11:4 16:26x ð10Þ
Lowperformers : Y = 235:08 14:05x ð11Þ
Mediumperformers : Y = 228:84 4:28x ð12Þ
Topperformers : Y = 396:120 5:93x ð13Þ
Positivedeviances : Y = 358:909 7:079x ð14Þ
where:
Y net income/cow/year
^
x months of dry season.
The income gradients, as shown in Eqs. (10) and (11) of very low
and low performers are considerably steeper than for the other
performance classes. This means, that these two classes are affected
UNCORRPlease cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001more seriously by dry season length than the others. This can also be
seen from the income range between the wettest and driest parts of
the country on Fig. 6.
Lowandmediumperformerswouldgeneratenearly the same income
under conditionswithout dry season constraints (Eqs. (11) and (12)). For
eachmonth of dry season, the gradient of lowperformerswas nearly 10$
steeper than the one of the medium performers. Medium performers
incomeper cowdeclined4.28$ for eachmonthof dry season (Eq. (12)). If
therewerenodry season, topperformerswouldhave thehighest income.
In the conditionswithdry season, thedecline of the incomeper dairy cow
per month of dry season was a little steeper than among medium
performers. Positive deviances showed comparatively higher costs
during the dry season than top performers. Their yearly income/cow
declined more rapid/steeply for each month of dry season (Eq. (14^)).
3.4.3. Average Income Assessment for Farm Size Class Proportions for
Each Department
According to SECPLAN (1994), the distribution of herd size classes
was uneven throughout the country (Fig. 6). The Western and
Southern departments had a high share of farms with very small herd
sizes of less than 10 cattle. The maximum share of very small herdsc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 11
t7:1 Table 7
Income from milk/cow/year of performance classes for observed dry season lengths.
^ ^
t7:2
t7:3 Lowest Low Medium Top Positive deviances
t7:4 n=31 n=17 n=29 n=20 n=30
t7:5 A B C D E
t7:6 1 dry month Mean −4.84 221.02 218.54 390.24 351.83
t7:7 St. Dev 185.01 113.78 106.03 170.45 187.18
t7:8 Sig B***, C***, D***, E*** D**, E** D***, E**
t7:9 2 dry months Mean −21.10 206.98 214.26 384.31 337.67
t7:10 St. Dev 175.99 104.54 96.86 158.11 186.03
t7:11 Sig B***, C***, D***, E*** D**, E** D***, E**
t7:12 3 dry months Mean −37.35 192.93 209.99 378.37 337.67
t7:13 St. Dev 167.62 95.45 87.79 145.96 186.39
t7:14 Sig B***, C***, D***, E*** D**, E** D***, E**
t7:15 4 dry months Mean −53.61 178.88 205.71 372.44 330.58
t7:16 St. Dev 160.00 86.55 78.84 134.05 188.25
t7:17 Sig B***, C***, D***, E*** D***, E** D***, E**
t7:18 5 dry months Mean −69.87 164.83 201.43 366.51 323.50
t7:19 St. Dev 153.24 77.90 70.05 122.45 191.56
t7:20 Sig B***, C***, D***, E*** D***, E*** D***, E**
t7:21 6 dry months Mean −86.12 150.79 197.16 360.57 316.42
t7:22 St. Dev 147.46 69.61 61.51 111.25 196.24
t7:23 Sig B***, C***, D***, E*** C*, D***, E*** D***, E*
t7:24 7 dry months Mean −102.38 136.74 192.88 354.64 309.34
t7:25 St. Dev 142.78 61.81 53.33 100.59 202.21
t7:26 Sig B***, C***, D***, E*** C**, D***, E*** D***, E*
t7:27 8 dry months Mean −118.64 122.69 188.60 348.70 302.25
t7:28 St. Dev 139.31 54.71 45.70 90.66 209.35
t7:29 Sig B***, C***, D***, E*** C***, D***, E** D***
t7:30 9 dry months Mean −134.89 108.65 184.33 342.77 295.17
t7:31 St. Dev 137.15 48.64 38.94 81.73 217.55
t7:32 Sig B***, C***, D***, E*** C***, D***, E*** D***
t7:33 10 dry months Mean −151.15 94.60 180.05 336.83 288.09
t7:34 St. Dev 136.35 44.00 33.59 74.15 226.70
t7:35 Sig B***, C***, D***, E*** C***, D***, E** D***
t7:36 Note: Significance between groups is indicated by letters followed by *Pb0.05, **Pb0.01, ***Pb0.001.
^ ^ ^
682 690
683 691
684 692
685 693
686 694
687 695
688 696
689 697
PROOF
was found in Intibuca with 81% of the farms. On country average the
majority of cattle farms had very small herds. These were the farms
that were affected most by a prolonged dry season and which were
least developed in forage options.
The average income per cow per department was dependent on
the herd size composition given for each department. As it was shown
in Table 3 and Fig. 4, each herd size class had distinct incomes from
milk for the dry and wet seasons.
C
Fig. 6. Cattle herd size distributio
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORREThe corresponding average distribution of these classes in each
department and their respective values for the indicator income per
cow in the dry and wet season were used to calculate the average
yearly performance of the dairy enterprises per department. This
regionalization approach was suitable to compare the profitability of
milk production in departments. Based on the presence of cattle herd
sizes, the income per cow of the average farm in this department was
calculated and the dry season length was considered.
TEDn in departments for 1993.
c factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
12 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
698 734
699 735
700 736
701 737
702 738
703 739
704 740
705 741
706 742
707 743
708 744
709 745
746
710 747
748
711 749
712 750
713 751
714 752
715 753
716 754
717 755
718 756
719 757
720 758
721 759
722 760
723 761
724 762
725 763
726 764
727 765
728 766
729 767
730 768
731 769
732 770
733 771As it can be read from Fig. 7, the central part of Honduras showed
incomes between 20 and 40$ in the areas with 4 to 5 months of dry
season, but dropped to 0 to 20$/cow/year in much of the Northern
part of Francisco Morazán. For most of the mountainous areas of e.g.
Olancho and Yoro, which had between 3 and 6 months of dry season,
the model estimations were between 20 to 80$/cow/year. In the
North along the Caribbean or in the east of Olancho, incomes per cow
rose to values between 80 and 120$/cow/year, while in small areas
income may reach up to 130$. Although Gracias a Dios is one of the
areas with most rainfall in Honduras, the income level of livestock
keepers was estimated low, because of a very high share of very small
farms in the population.
4. Discussion and Conclusions
Dry season length was calculated from evapotranspiration gener-
ated with the method of Thornthwaite (1948). The weather
simulation software, MarkSim (Jones, 2001) provided the tempera-
ture and rainfall input data. Temperatures were corrected with station
data. The resolution of the dry season assessment is one month.
Experienced local experts agreed with the final dry season map
produced, although it tends to over and undershoot in extreme
conditions, like in the wettest and driest parts of Honduras, which
cover comparatively small areas (less than 4% of the area).
With the income regionalization maps^we localized gradual
changes from low to high income for herd size and performance
classes on country scale. Model results showed clear impacts of the
dry season length on the income per cow per year.
When based on herd size classes, income indicators had the
inevitable disadvantage of comparatively high standard deviations
(Table 3). The standard deviations represent a measure for the
representativeness of mean values (Bamberg and Baur, 2002). It
showed that within each herd size class, there were farms with higher
and lower incomes, as compared to the mean. The reason why the
classification in herd size classes was used despite the high variability
of the indicator was that herd size can be easily assessed in the field
and is easily understood by farmers, extension workers and policy
makers. Income from milk per cow per year of extra large farms wasFig. 7. Income distribution derive
Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORRECsimilar to the values for positive deviances under all climatic
conditions. However positive deviances had higher income per cow
than farmswith less than 100 cattle where dry seasonwas up to seven
months long (Pb0.05). These conditions covered 96% of Honduras.
Medium size farms earned more than small farms whe^re dry season
was six months and shorter (85% of Honduras). Small and very small
farms were the most hit by a lon^g dry season.
The classification on performance yielded more representative
mean values and was more precise for regionalization. Performance
indicators are beneficial tools for assisting effective decision making
aimed at improving business performance (Wilson et al., 2005). The
disadvantage of the performance indicator used was that it was not as
quickly accessible in the field when compared to herd size.
Dry season impact on income for low performers was greater than
for medium performers where the dry season length exceeded six
months (Pb0.05), i.e. on 16.5% of the territory of Honduras. Very low,
low and medium performers had lower incomes than top performers
under all observed dry season lengths (Pb0.01). On 96% of the area of
Honduras (up to 7 months of dry season), medium low and very low
performers had lower incomes than positive deviances (Pb0.05).
For the regionalization of average income/cow/year per department,
the paper made use of the available da^ta and demonstrated the
methodology, estimating total livestock population for 2005 from
annual growth rates in the post-Mitch period. Since the agricultural
census data (SECPLAN, 1994)were old it wouldmake sense to apply the
method again once a new census becomes available. When the average
herd size composition of departments was considered, regions with a
high share of small herd sizes showed low incomes per cow.
The regionalization of positive deviances (in Fig. 5) showed the
state of farms that were developing towards more intensive cattle
management and better use of forages. Sharp dry season income
drops (44–53%), as observed on farms with more than 20 cattle could
be avoided with a better use of forage technologies and intensifica-
tion. The even sharper income drops on farms with less than 20 cattle
could be mitigated through adequate low-cost measures that need to
be based on as much as possible farm produced feed. One
recommendation is the subsidized introduction of well-adapted
improved grasses (e.g. B. brizantha cv Toledo) and their conservation.
TED PROOFd for departments for 2005.
c factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx 13
772 834
835
773 836
774 837
775 838
839
776 840
777 841
778 842
843
779
844
780 845
781 846
782 847
848
783 849
784 850
785 851
852
786 853
787 854
788 855
856
789 857
790 858
791 859
860
792
861
793 862
794 863
864
795
865
796 866
797 867
798 868
869
799 870
800 871
801 872
873
802 874
803 875
804 876
877
805 878
806 879
807 880
881
808
882
809 883
810 884
885
811
886
812 887
813 888
889
814
890
815 891
816 892
817 893
894
818 895
819 896
820 897
898
821 899
900
901
822
902
903
823 904
905
824
906
825 907
826 908
827 909
910
828 911
829 912
913
914
830 915
916
831 917
832 918
833 919Another possibility would be the improvement of maize stover with
Lablab purpureus as a legume (Lentes et al., 2007).
Interpreting the maps on the performance classes as stages of
intensification, it can be demonstrated to farmers and policy makers
how much and where in Honduras an upward movement between
performance classes is likely to increase income per cow. Intensification
of production is an important solution for resource-poor farmers (Peters
et al., 2001) and for a self-sufficient milk production in Honduras.
The adoption of new crops and improved technologies is
constrained substantially where the availability of working capital is
limited (Van Keulen, 2007). Financial bottlenecks are important
constraints for adoption of forage technologies and genetic improve-
ments of the herds on small and very small farms. These farms lack of
capital at the end of the dry season and their priority is to secure
subsistence crop production. Without an increase of working capital it
is unlikely that resource poor farms in such a situation invest in
forages of better nutritive value and their conservation during the
rainy season, because their crop production requires the investment.
Without investments or efforts for intensification, these farms will
remain on low-income levels. More off-farm employment would help
alleviate the lack of capital since the additional income could be
invested in more capital-intensive technologies (Van Keulen, 2007).
Such opportunities are rare and usually far from being available to the
rural poor in Honduras. Nevertheless some innovative and motivated
individuals undertook low-cost efforts and improve slowly.
On farms with more than 20 head of cattle, the probability for
changewas higher. These farms are able to accumulate some capital to
reinvest in the farm e.g. in forages, their conservation or in cow breeds
with better genetic potential for milk production.
The resource use efficiency of farms was related to the length of
the dry season and the technological level of the farms. Where the dry
season was very long, farmers with low technological level generated
little to reinvest and were thus cash constrained. A higher level of dry
season adaptation was required to sustain production and income
with increasing dry season length.
The climatic data used for this paper are estimations for long-term
averages derived from the past. It is however known (e.g. from farmers
experience and a few climate stations, where measurements are done)
that precipitation and temperatures and also dry season length is
variable between years. Long-term climate change scenarios for
Honduras show trends of increasing temperatures and decreasing
precipitation (IPCC, 2007). In the decadal climate risk index for 1998–
2007 of Harmeling (2008), Honduras is listed as the most vulnerable
country, followed by Bangladesh and Nicaragua. Taking these factors
into account, themaps produced in this paper can also be interpreted as
vulnerability maps for climate change and natural disasters. Those
farmers that are already seriously affected under average dry season
conditions are more vulnerable to climate change and natural disasters.
Those farmers that aremore efficient under average climatic conditions
are more resilient to the effects of natural disasters and climate change.
Acknowledgements
This publication is a result of a research project funded by BMZ
(German Ministry of economic Cooperation and Development) under
the German PostDoc Programme of BEAF-GTZ. (Advisory Service on
Agricultural Research for Development). Thanks are due to the
Partner organization DICTA (Dirección de Ciencia y Tecnología
Agropecuaria), the Honduran national agricultural research institute
for their support during the fieldwork.
References
Allen, R.G., Pereira, L.S., Raes, D., Smith,M., 1998. Crop evapotranspiration—Guidelines for
computing crop water requirements — FAO Irrigation and drainage paper No. 56.
Food and Agriculture Organization of the United Nations (FAO), Rome.
UNCORRECPlease cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001Argel, P., 1999. Caracterización de sitios con potencial de adopción de germoplasma
forrajero mejorado por pequeños y medianos productores de doble propósito en
Honduras. Informe de Consultoría. Centro Internacional de Agricultura Tropical
(CIAT), Cali.
Bamberg, G., Baur, F., 2002. Statistik. Oldenbourgs Lehr- und Handbücher der
Wirtschafts- und Sozialwissenschaften. Oldenbourg, Munich.
Biggs, S., 2008. Learning from the positive to reduce rural poverty and increase social
justice: institutional innovations in agricultural and natural resources research and
development. Experimental Agriculture 44, 37–60.
Boeringa, M., 2000. Nieuwland Kriging Interpolator, Plugin for ArcView GIS. [3.2].
Nieuwland Automatisering, Wageningen.
Camargo, A.P., Marin, F.R., Sentelhas, P.C., Picini, A.G., 1999. Adjust of the Thornthwaite's
method to estimate the potential evapotranspiration for arid and superhumid
climates, based on daily temperature amplitude. Revista Brasileira de Agrometeor-
ologia 7 (2), 251–257 in Portuguese with English summary.
CGIAR-CSI (CGIAR — Consortium for Spatial Information), 2004. SRTM 90m Digital
Elevation Data. http://strm.csi.cgiar.org/ accessed on 30 January 2009.
CGIAR-CSI (CGIAR — Consortium for Spatial Information), 2006. CRU TS 2.1 Climate
Database. http://cru.csi.cgiar.org/ accessed on 30 January 2009.
Delgado, C.L., 2005. Rising demand for meat and milk in developing countries:
implications for grassland-based livestock production. Pp. 29–39. In: McGilloway,
D.A. (Ed.), Grassland — A Global Resource. Wageningen Academic Publishers.
FAO, 2005. FAO statistical country profile Honduras. — From FAO statistics division. http://
www.fao.org/es/ess/yearbook/vol_1_2/pdf/Honduras.pdf accessed on 30 January 2009.
Fujisaka, S., Holmann, F., Peters,M., Schmidt, A.,White,D., Burgos, C.,Ordoñez, J.C.,Mena,M.,
Posas,M.I., Cruz,H., Davis, C.,Hincapié, B., 2005. Estrategias paraminimizar la escasezde
forrajes en zonas con sequías prolongadas: Honduras y Nicaragua. Documento de
Trabajo No. 201. CIAT (Centro Internacional de Agricultura Tropical), Cali.
Harmeling, S., 2008. Global Climate Risk Index (CRI) 2009 — how extreme weather
events affect countries. Presentation held at COP14, Poznan, 4th December, 2008.
http://www.germanwatch.org/klima/cri09pres.pdf accessed on 30 January 2009.
Holmann, F., 2001. Milkmarket of small scale artisan cheese factories in selected livestock
watersheds of Honduras and Nicaragua. Livestock Research for Rural Development
13, 1. http://lrrd.org/lrrd13/1/hilm131.htm accessed on 30 January 2009.
INE (Instituto Nacional de Estadistica), 2001. Encuesta Agricola Nacional 2000–2001,
Tomo III, Ganadería y Otras Especies Animales. Secretaria del Estado del Despacho
Presidencial, Tegucigalpa, Republica de Honduras.
IPCC, 2007. Summary for policymakers. In: Solomon, S., Qin, D., Manning, M., Chen, Z.,
Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Climate Change 2007: The
Physical Science Basis. Contribution of Working Group I to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge University
Press, Cambridge.
Jansen, H.G.P., Pender, J., Damon, A., Schipper, R., 2006. Rural development policies and
sustainable land use in the hillside areas of Honduras. A Quantitative Livelihoods
Approach. Research report No. 14. International Food Policy Research Institute
(IFPRI), Washington, DC.
Jones, P., 2001. A Computer Tool that Generates Simulated Weather Data for Crop
Modeling and Risk Assessment. http://gisweb.ciat.cgiar.org/marksim/index.htm
accessed on 30 January 2009.
Kyeyamwa, H., Speelman, S., van Huylenbroeck, G., Opuda-Asibo, J., Verbeke, W., 2008.
Raising offtake from cattle grazed on natural rangelands in sub-Saharan Africa: a
transaction cost economics approach. Agricultural Economics 39 (1), 63–72.
Lentes, P., 2004. The Contribution of GIS and Remote Sensing in Micro-economic and
Regional Farming Systems Research in North Western Vietnam. Farming and Rural
Systems Economics, Vol. 52. Margraf Verlag, Weikersheim.
Lentes, P., 2006. Spatial analyses and impact assessment for rural livelihood in
mountain regions of Northern Vietnam. In: Doppler,W., Weber, E. (Eds.), Resources
and Livelihood in Mountain Areas of South East Asia. Farming and Rural Systems in
a changing environment. Margraf Publishers, Weikersheim, pp. 377–409.
Lentes, P., Peters, M., White, D., Holmann, F., Reiber, C., 2006. Assessing and comparing
income generation of livestock holders in Olancho, Honduras. An Analysis Across
Typical Landscapes and Farming System. Tropentag 2006, Prosperity and Poverty in
a Globalized World — Challenges for Agricultural Research October 11–13, 2006.
University of Bonn, Bonn, Germany, p. 144.
Lentes, P., Holmann, F., Peters, M., White, D., Cruz, H., 2007. Management and farm
characteristics that favor or impede efficient resource use in dual-purpose cattle
systems in Central America. In: Lascano, C. (Ed.), Tropical Grasses and Legumes:
optimizing genetic diversity for multipurpose use (Project IP5). Annual Report 2006.
Centro Internacional de Agricultura Tropical (CIAT), Cali, Colombia, pp. 170–186.
http://www.ciat/cgiar.org/forrajes/pdf/report_2006/output_4.pdf accessed on 30
January 2009.
Matheron, G., 1963. Principle of geostatistics. Economic Geology 58, 1246–1266.
Mitchell, T.D., Jones, P.D., 2005. An improved method of constructing a database of
monthly climate observations and associated high-resolution grids. International
Journal of Climatology 25, 693–712.
Munroe, D.K., Southworth, J.Tucker, Tucker, C.M., 2002. The dynamics of land-cover
change in western Honduras: exploring spatial and temporal complexity.
Agricultural Economics 27 (3), 355–369.
Pelton,W.L., King, K.M., Tanner, C.B., 1960. An evaluation of the Thornthwaite andmean
temperature methods for determining potential evapotranspiration. Agronomy
Journal 52, 387–395.
Pereira, A.R., Pruitt, W.O., 2004. Adaptation of the Thornthwaite scheme for estimating
daily reference evapotranspiration. AgriculturalWaterManagement 66 (3), 251–257.
Peters, M., Horne, P., Schmidt, A., Holmann, F., Kerridge, P.C., Tarawali, S.A., Schultze-Kraft,
R., Lascano, C.E., Argel, P., Stür,W., Fujisaka, S.,Müller-Sämann, K.,Wortmann, C., 2001.
The roleof forages in reducingpoverty anddegradationof natural resources in tropical
TED PROOFc factors and income indicators for milk production in Honduras,
ARTICLE IN PRESS
14 P. Lentes et al. / Ecological Economics xxx (2009) xxx–xxx
920 940
921 941
922 942
923 943
924 944
925 945
926 946
927 947
928 948
929 949
930 950
931 951
932 952
933 953
934 954
935 955
936 956
937 957
938 958
939 959
960production systems. ODI Agricultural Research and Extension Network, Network
Paper No. 117.
Schöninger, M., Dietrich, J., 2003. Hydroskript. http://www.hydroskript.de. accessed on
30 January 2009.
SECPLAN (Secretaría de Planificación, Coordinación y Presupuesto), 1994. IV Censo
Nacional Agropecuario 1993. Tomo V, Ganadería y otras especies animales.
Tegucigalpa, Graficentro. Censo Agropecuario 1993.
SERNA, 2005. Informe del Estado y Perspectivas del Ambiente: Geo Honduras 2005/
Honduras, Secretaría de Recursos Naturales y Ambiente (SERNA)1a. ed. Programa de
las Naciones Unidas para el Medio Ambiente (PNUMA)/(SCANCOLOR), Tegucigalpa.
172 pp.
SERNA, 2007a. Monthly rainfall sums in mm. Stations: La Unión (1981–2005), La
Suncuya (1983–2006), El Rosario (1980–2006), La Conce (1982–2007), Guinope
(1966–2007), Taupasenti (1968–2006), Guajinquil (1987–2007), San Lucas (1966–
2007), Texiguat (1967–2006), Villa Ahumada (1975–2006), Oropoli (1972–2007),
Potrerillos (1966–2007), Mariata (1966–2007), Guayabillas (1967–2006): Secre-
taria de Recursos Naturales, Dirección General de Recursos Hídricos, Departamento
de Servicios Hidrológicos y Climatológicos, Tegucigalpa, Honduras.
SERNA, 2007b. Monthly mean Temperatures in °C . Stations: La E.N.A Catacamas (1991–
2006), La Conce (1982–2006), Villa Ahumada (1975–2006), Guayabillas (1967–Please cite this article as: Lentes, P., et al., Regionalization of climati
Ecological Economics (2009), doi:10.1016/j.ecolecon.2009.09.001
UNCORREC2006), Secretaria de Recursos Naturales, Dirección General de Recursos Hídricos.
Departamento de Servicios Hidrológicos y Climatológicos, Tegucigalpa, Honduras.
SMN (Servicio Meteorológico Nacional), 2007. Monthly mean Temperatures in °C and
monthly rainfall sums in mm. Years 1995–2006. Stations: Catacamas, Yoro, La
Esperanza, Jacaleapa. SMN, Departamento de Climatología, Tegucigalpa, Honduras.
Stanhill, G., 1961. A comparison of methods of calculating evapotranspiration from
climatic data. Israel Journal of Agricultural Research 11, 159–171.
Suttie, J.M., 2000. Hay and straw conservation. For small-scale farming and pastoral
conditions. (FAO Plant Production and Protection Series No. 29, Rome, Italy).
Thornthwaite, C.W., 1948. An approach toward a rational classification of climate.
Geographical Review 38 (1), 55–94.
van Keulen, H., 2007. Quantitative analyses of natural resource management options at
different scales. Agricultural Systems 94, 768–783.
Williams, R.G., 1986. Export Agriculture and the Crisis in Central America. The
University of North Carolina Press, Chapel Hill.
Wilson, R.H., Charry, A.A., Kemp, D.R., 2005. Performance indicators and benchmark-
ing in Australian agriculture: synthesis and perspectives. Extension Farming Systems,
1,1 — Research Forum. http://www.csu.edu.au/faculty/science/saws/afbmnetwork/
efsjournal/volume1/number1/EFS_Journal_v01_n05_RossWilson_et_al.pdf accessed
on 30 January 2009.c factors and income indicators for milk production in Honduras,
TED PROOF