ARTICLE IN PRESS
Environmental Hazards 4 (2002) 113–127
Climate variability, political crises, and historical population
displacements in Ethiopia
Joshua Comenetz*, Ce! sar Caviedes
Department of Geography, University of Florida, P.O. Box 117315, Gainesville, FL 32611, USA
Received 24 May 2002; received in revised form 3 August 2003; accepted 28 August 2003
Abstract
El Niño events from the 1970s through the 1990s caused extended droughts in Ethiopia. These droughts were followed by famine
and political turmoil that resulted in radical changes of government, secession, and a massive program of population redistribution.
Cartographic analysis of Ethiopian census data from 1984 and 1994 shows changes in demographic patterns. The consequences of
government-imposed migration policies, whose catalyst was the climate variability caused by repeated El Niño events, were changes
in the ethnic composition of certain Ethiopian regions and changes in the geographic pattern of population growth.
r 2003 Elsevier Ltd. All rights reserved.
Keywords: Climate; El Niño; Ethiopia; Migration; Cartographic representation
1. Introduction
Climate variability over the last three decades of the
20th century resulted in droughts, famines, and locust
outbreaks in several countries of Africa—all of them
being catalysts of decisive changes in the political
development of such countries and causes of dramatic
population displacements. Climatic oscillations commonly have been given only a subordinate role as the
triggers of historical events and political changes,
perhaps to avoid conveying an air of determinism to
the assumed free acts of humans. Nevertheless, the
persistence and dimensions of recent environmental
crises have swayed some researchers away from this
traditional interpretation of recent history. It is now
more easily accepted that in vulnerable traditional
societies, climate or environmental crises have a
compelling influence over social structures and political
stability (McNeill, 2000). When describing the aftereffects of the widespread El Niño of 1972–73, Caviedes
(1982) observed that the fall of Ethiopia’s emperor Haile
Selassie and his replacement by a Marxist military
regime had been prompted by the social unrest resulting
*Corresponding author. Tel.: +1-352-392-0494; fax: +1-352-3928855.
E-mail address: [email protected]fl.edu (J. Comenetz).
1464-2867/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.hazards.2003.08.001
from the droughts in sub-Saharan Africa (Sahel) and
eastern Africa that had been induced by that climate
anomaly. The political and social developments unleashed in Ethiopia by drought and famine did not end
in the subsequent years, but were, on the contrary,
exacerbated by further El Niño events in 1976–77, 1982–
83, and 1991–93. When the effects of El Niño 1991–93
raged over the drought- and famine-stricken country,
the communist rulers, in order to subdue unruly
populations, introduced draconian measures such as
ousting the people, including the forced migration of
emancipationist ethnic groups. After the 1991–93 El
Nino, dissent and uprisings in Ethiopia were enormously magnified by the lack of rain, consequent food
shortages, and the severe resettlement policies taken by
the Marxist government during the previous two
decades (Pankhurst, 1991).
This paper assesses the spread and intensity of the
droughts and famines caused by El Niños from
the 1970s to the 1990s, which served as the context for
the relocation of populations undertaken by the regime
of General Mengistu Haile Mariam. It also analyzes the
effects of these climate crises and the forcible relocations
of inhabitants on the demographic structure of contemporary Ethiopia. The arguments made in this paper
rest on historical data that reveal the deep impacts that
past droughts had on nature and Ethiopian populations.
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
114
natural catastrophes through cartography breaks new
ground.
Using cartographic analysis of recent census data, this
research also examines the consequences that political
repression had on the distribution of population across
the country.
This is a case study of the consequences a climatic
crisis can have when coinciding with social unrest and
political caprice. Not many African countries are suited
for this type of study due to the lack of appropriate
statistical data, scarce documentation of the impact of
climatic crises, and the unpredictability of political
developments. Ethiopia has the advantage of having a
relatively straightforward history of events, central
policies that are applicable country-wide, and a statistical base that—with careful consideration of its pitfalls—can offer a fair picture of the consequences that
political and natural events have had on the demographic makeup of the country (Rubenson, 1991). This
study enriches the underdeveloped field of demographic
studies in African nations that have been repeatedly
affected by El Niño. Studies that quantify the influence
of El Niño on population change in Africa do not exist:
these countries have neither the infrastructure nor the
human or monetary resources to document with hard
data the demographic impact of such catastrophes.
Ethiopia is the only country in semi-arid Africa with the
basic statistical resources (census data) necessary to
assess the impact of droughts and political events on
demographic trends. In this sense, the country and this
study are unique. Cartographic analysis of population
change is also rare or absent in these countries. Though
existing literature on population movements is abundant, it mostly reports on qualitative changes. Thus, our
attempt to quantify population movements related to
2. Persistent droughts and famines in the Horn of Africa
Located at a climatic crossroad in northeast Africa,
where interannual variations of precipitation depend on
the seasonal displacements of the Inter-Tropical Convergence Zone (ITCZ), the dominance of the drying
harmattan during the winter months, and the time of
inception and intensity of the summer monsoons
(Sirocko, 1996), Ethiopia accurately reflects whatever
variation takes place when one of these three major
climatic controls fluctuates (Nicholson, 1997). Since the
main contributor to Ethiopian rainfall is the summer
monsoon (74 percent of annual rain), its weakness or
failure to occur has disastrous consequences for eastern
Africa and the Sahel. Early studies from the 1960s tied
the decrease of precipitation in sub-Saharan Africa to
the occurrence of the warm phases of the El Niño/
Southern Oscillation (ENSO)—in other words, to the
development of sea surface warming in the equatorial
Pacific.
Contemporary warm ENSO events, those that can be
documented with instrumental meteorological data,
have revealed the remarkable simultaneity that droughts
in Ethiopia and the Sahel have with droughts in India,
Indonesia, Australia, and southeast Asia (Hastenrath,
1990). The association between monsoon failures in the
Indian Ocean and droughts in the Horn of Africa can be
understood through examination of the interannual
variation of rainfall at Addis Ababa (Fig. 1). Prior to the
250
Rainfall, mm
3.00
2.00
200
150
100
50
0
J F MAM J J ASO N D
Z-score
1.00
0.00
-1.00
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
1968
1966
1964
1962
1960
1958
1956
1954
1952
1950
1948
1946
1944
1942
-3.00
1940
-2.00
Fig. 1. Interannual rainfall variability at Addis Ababa (1940–94): annual deviation from mean in standard deviation units. Darkened bars indicate
coincidence with warm ENSO events (El Niño) in the tropical Pacific. Inset: averaged monthly rainfall for the stations Addis Ababa, Gondar, and
Awassa in the highlands of central Ethiopia. Source: Global Historic Climate Data, NOAA National Climatic Data Center (NCDC). Asheville, NC,
2003.
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
existence of instrumental data (before the second half of
the 19th century) historical sources and anecdotal
references to times of dryness and famine are very
frequent in historical accounts of Ethiopia and allow a
fair identification of the effects of past El Niños in the
context of worldwide historical variability (Whetton and
Rutherford, 1994). A useful list of famines and
epidemics, reaching back to the 1400s, was compiled
by Pankhurst (1990). Using these materials, Caviedes
(2001) was able to relate the occurrence of the most
devastating drought/famine events in Ethiopia to the
occurrence of past El Niños in the Pacific realm. These
events also coincided with droughts in Asia and the rest
of Africa, such as those that occurred in the wake of the
warm ENSO episodes of 1540, 1567–68, 1618–19, 1633–
35, 1747, 1772–73, 1812, 1828, 1888–92, 1916–18, and
1927–28. After the El Niño of 1957–58, the coincidence
of warm ENSO episodes and droughts in Ethiopia in
1965, 1972–73, 1982–83, 1986, 1991–93, and 1997–98
became quite evident, indicating that these phenomena
were becoming more frequent and intense, probably due
to contemporary global warming.
Detailed documentation offered by Wolde-Mariam
(1986) presents irrefutable evidence that the occurrence
of recent El Niños had deleterious consequences for
agricultural and pastoral communities in Ethiopia.
Agricultural activities, whether for subsistence or for
cash crops such as coffee, sorghum, maize, and cereal
are concentrated in the central highlands of the country.
Due to good rainfall and the relatively low prevalence of
diseases such as malaria (Kloos and Adugna, 1989a), the
central highlands are where the highest concentration of
population is found. The agricultural calendar reflects
115
the annual distribution of rain (inset in Fig. 1). The early
rains (belg) are used to grow basic staples and to prepare
the ground for the main planting season that depends on
the June-to-September (kiremt) rains. Whenever the
summer monsoon is weak or late to arrive—typical
occurrences during El Niño events in the tropical
Pacific-crops planted in the early summer are stunted
in their growth or fail altogether. With most of the
cultivating activities being conducted with oxen, the
failure of rains also causes critical shortages of fodder
that lead to cattle mortality (Wolde-Georgis et al.,
2001).
The most extensive cultivated area is the core region
of the country, Shewa (where Addis Ababa is located),
and the northern mountain regions of eastern Gondar
and Gojjam (Fig. 2; Ethiopian Mapping Authority
(EMA), 1988). In the uplands bordering this central
core, stretch the drought-prone regions of Wello,
western Gojjam, Keffa, and Sidamo, where subsistence
agriculture mixes with mountain pastoralism and the
density of population diminishes remarkably. The
northern region that comprises Tigray and the now
independent, former Ethiopian province of Eritrea is
less densely populated than the core (Fig. 2) and
increasingly more drought-stricken. So too is the eastern
region of Hararge and Bale (Somali area in Fig. 2) that
borders Djibouti and Somalia. Desert conditions predominate and pastoral activities are restricted to a few
watercourses descending from the Harar Mountains.
The Western Uplands, including Wellega, Keffa,
western Gojjam, and Illubabor, gradually descend onto
the plains of Sudan and are drained by the Blue Nile
and its tributaries. Increased humidity, caused by the
Fig. 2. Population density by wereda, 1984, with old provincial boundaries.
ARTICLE IN PRESS
116
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
northern shift of the ITCZ in summer, permits the
spread of more arboreal vegetation in this region than in
the rest of the country, although the recent wars, human
occupation, and progressive dryness during the last
decades of the 20th century are converting this large
sector of Ethiopia into a hardwood-savanna environment. Here pastoral activities are mixed with subsistence
agriculture.
Surveys of the famine-proneness of Ethiopia by
Wolde-Mariam (1986) and Wolde-Georgis et al. (2001)
reveal that the most affected regions are those of the
northeast, the eastern plains bordering Somalia, and the
south adjacent to Kenya. Struck by less severe droughts
and ensuing famines are the regions located in the
central highlands and in the uplands descending to
Sudan. On the basis of drought occurrences from 1958
to 1977, Wolde-Mariam (1986) recognized a pattern of
progression: in general, droughts and famines begin to
be felt in the arid mountains of Eritrea and Tigray
before they propagate southward along the Awash
Valley into the southern regions of Hararge and Bale.
As the seriousness of a drought mounts, the central
highland regions of Wello, Shewa, Arssi, Keffa, and
Wellega become involved. The last regions to fall under
the influence of dryness, livestock mortality, and
increased lack of staples are the western regions of
Gojjam, Gondar, Wellega, and Illubabor. The stress
produced by this internally spiraling progression of
drought and famine first affects the peripheral regions of
the central core, forcing people fleeing disaster to seek
refuge in regional centers before moving into the more
populated heartland. This helps to explain why both
Emperor Haile Selassie and General Mengistu were
reluctant to admit that an impending drought (and
social unrest) was brewing in the northeast section of the
country, and why, when they reacted, it was too late to
defuse the militance of secessionist forces (Keller, 1992).
3. The human and environmental impact of the 1970s
droughts
It was not until the occurrence of the 1972–73 warm
ENSO episode that the negative consequences of these
anomalies in certain regions of the tropics became clear
(Caviedes, 1982). Globally, the 1972–73 El Niño was not
extremely severe when compared with subsequent
episodes, but regionally, the consequences in Africa
were the most calamitous of recent decades. This event
deeply upset the hydro-meteorological regimes of eastern and sub-Saharan Africa, initiating the infamous
drought and subsequent famines of the Sahel (Kates,
1981). In Ethiopia, the precipitation of the kiremt
(planting season rains) were so scant or absent in some
districts that the subsequent harvest season of the
autumn of 1972 was seriously reduced, particularly in
northern Wello. As 1973 arrived and the ENSO episode
became more acute, the failure of the kiremt to
materialize extended the drought further into the
northern highlands. Agricultural dryness also expanded
into the eastern and southern regions of the country to
encompass 55 percent of all the 102 awrajas (regions),
into which the country was divided before the administrative restructuring of the 1980s. By mid-1974, the
continued drought and failed summer harvests grew to
comprise 60 percent of the awrajas as the specter of
famine covered most of the country, with the exception
of the western regions (Wolde-Mariam, 1986). Though
he was alerted to the threats that the expanding famine
posed in terms of social unrest and flows of refugees into
urban places, Emperor Haile Selassie was poorly
advised as to the seriousness of famine in different
regions. State food was therefore not distributed to all
hungry populations in all needy regions. His delays were
used by a group of Addis Ababa army officers known as
the Derge (Committee) to stage a coup that deposed the
emperor in September 1974 and forced him into exile
(Hancock, 1985).
The Marxist roots of the officers who removed the
emperor were evident in the name Derge, the authoritarian policies implemented to deal with the emergency,
and with popular unrest. Food was distributed mostly to
rural dwellers sympathetic to the new regime (Kebbede,
1988) and drastic measures were adopted by the military
to restrict the mobility of those who were abandoning
scorched fields and trying to escape famine-stricken
areas (Keller, 1992). In 1975 the Derge regime, now
firmly in the hands of General Mengistu Haile Mariam,
proceeded to decree the nationalization of all agricultural lands in the country with the purpose of creating
rural communes in the Soviet style (Clay and Holcomb,
1986). Creating rural communes in the middle of a
disastrous food shortage caused by the drought set the
conditions for disaster, but then the military ruler
enacted a forced agricultural program to stimulate
productivity. At this point agricultural productivity
was severely impaired by climatic limitations and
military conscription or joining of agricultural teams
further exacerbated an already volatile situation. Finally, all measures originated from the ethnic-Amharic
military power in Addis Ababa, so that provincial
resistance was not long in coming. When the next El
Niño (1982–83) provoked a drought of even larger
magnitude and more prolonged effects than that of
1972–73, the ethnic rebellion flared up again (Tareke,
2002).
To understand the roots and dynamics of the
opposition to the Derge during the 1980s that resulted
in the collapse of the regime in 1991 and the secession of
Eritrea in 1993, it is necessary to outline the ethnic and
cultural complexity of Ethiopia (Henze, 1995). Through
the decades of imperial rule by Haile Selassie, Ethiopia
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
117
Fig. 3. Main ethnic groups in the 1980s, with old provincial boundaries.
was a conglomerate of ethnic groups united by an iron
rule exerted by Amharic elites from the central highlands, where Addis Ababa is located. The central
mountain region is the spine of historical Abyssinia
and the heartland of the country, populated by the
Amhara peoples, a group of Semitic language and
Christian Orthodox belief, that has traditionally dominated the upper ranks of the political, military and
cultural elites of Ethiopia (Fig. 3). The northern extreme
of the country is inhabited by peoples of Tigre ethnic
roots, who differ historically as well as religiously
(Muslim), from the Amhara of the central mountains;
they make up the ethnic core of modern Eritrea. On the
high plateau south of the Eritrea border live the
Tigrayan peoples, of Semitic language like the Amhara,
Orthodox Christians (Kaplan et al., 1971), and subsistence cultivators. Their traditional independent leanings often have placed them at odds with the centralizing
and authoritarian Amhara elites (Tareke, 1991). In the
arid lands to the east of Tigray and into the shores of the
Red Sea live the Affar, members of the Cushite linguistic
family, a group consisting of Muslim herders having
much in common with the Affar tribes located in coastal
Eritrea. Populating most of eastern Wello, eastern
Shewa, and northern Hararge, the Affar have long
struggled against Amharic centralism. During the Derge
regime, and even more so when Eritrea broke from
Ethiopia, they fought to achieve autonomy. The dry
uplands and plains of the southeast, comprising the
territories of Hararge and Bale, are in lifestyle, language,
culture, and Muslim religion, very similar to their
southern neighbors in Somalia. The Oromo, established
mostly in the provinces of Sidamo and Gamo Goffa,
south of the region inhabited by the Amharic population, are closely related by language (Cushite) and
religion (Muslim) to their Kenyan neighbors, and are
estimated to make up 40 percent or more of the total
population of Ethiopia (Clay and Holcomb, 1986;
Keller, 1995). The western fringe of the country,
adjacent to Sudan, is inhabited by semi-nomadic tribesmen of Nilotic language and customs, predominantly
Muslim. It is against this ethnic backdrop that the
sweeping changes following the dethronement of Haile
Selassie, and the repressive policies of General Mengistu
are reviewed.
4. Population relocation after ENSO 1982–83
As early as May 1982, anomalous ocean and atmospheric conditions typical of El Niño began to develop
in the tropical Pacific. By September and October, these
anomalies had propagated into the eastern hemisphere
as shown by a delayed onset of the southeast Asian
monsoons. This earlier-than-usual development of the
warm ENSO event meant that in Ethiopia the rain
needed for planting failed to occur after the commonly
rainless winter months (Mattsson and Rapp, 1991). As
had happened in previous drought-famine occurrences,
the areas struck first were the northern provinces.
However, as dryness persisted during the rest of the
year and continued into 1983, in phase with the
prolonged El Niño episode in the Pacific, the drought
gradually expanded into the central highlands, and from
there progressed into the western semi-arid belt (Kiros,
1991; World Meteorological Organization (WMO),
ARTICLE IN PRESS
118
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
1985). The first reaction of the Derge government was
denial, followed by hasty and unorganized distribution
of stored grain, and then by suppression of information
and repressive measures against desperate peasants and
villagers who were fleeing the scorched countryside
(Adhana, 1991). With the arrival of a third year of
dryness in 1984 and another summer harvest failure, the
government finally acknowledged that catastrophe had
hit rural Ethiopia and that the external world had to
come to relieve the disaster (Webb and von Braun,
1994). It was only in the autumn of 1984 that the
western world came to know the misery that ineffective
political measures, bad administration of meager
resources, and extreme climatic conditions had brought
into the country (Clay et al., 1988). By then the excesses
of repression and outright war triggered by the drought
calamity became evident.
As is usual in emergency situations in Ethiopia, lack
of food and the failure to obtain assistance from the
central government erupted in popular restlessness in
the provinces traditionally discontented with the central
government (Tareke, 2002). After an initial government
reaction based on the use of armed force to quell
discontent, the spread of regional resistance against the
military was countered through administrative quickfixes, of which the most long-lasting and destructive was
the reorganization of administrative boundaries. New
districts were artificially created in order to divide
cohesive and well-functioning ethnic communities. In
other cases, new units—particularly in border regions—
were instituted to grant an apparent autonomy to
clamoring ethnic minorities supportive of the Derge
(Adhana, 1991). As Aalen (2002) noted, this process has
continued through more recent administrative restructurings, including the creation of provinces for large
ethnic groups such as the Oromo, and ‘‘special’’ weredas
(similar to counties) to protect smaller minorities within
larger multiethnic provinces.
In almost all cases these measures backfired. The
attempts at dividing territories occupied by Eritreans and
Tigrayans increased their resolve to make themselves
independent from the central power and fueled the
formation of regional militias that started guerrilla
warfare against Addis Ababa (Tareke, 2002). In addition,
Muslim populations from Bale began to gravitate more
towards their Muslim relatives in Somalia, who had
already invaded southeastern Ethiopia in 1977–78 and
from there threatened to take control over most of Bale
and Sidamo (Khalif and Doornbos, 2002). Inhabitants of
part of Gojjam rebelled against the central government
army, while Oromo populations from Wello, who disliked
the deposed emperor Haile Selassie, also challenged the
new Amharic rulers of Addis Ababa (Keller, 1992).
In conjunction with the administrative reorganization, the progressing damage caused by drought and
restlessness triggered by hunger enticed the government
to accelerate the process of relocation of rural communities, a policy instituted in 1978 with the guise of
providing them with more fertile agricultural lands and
better sources of subsistence. Most affected were
peasants from Wello and Tigray who were placed in
hastily instituted agricultural villages (the process hence
called villagization) in lands of northern Bale, southern
Shewa and Arssi, and Hararge (Pankhurst, 1992). So
poorly planned and hastily prepared were these agricultural colonies that when the drought following the El
Niño of 1982–83 hit Ethiopia, the inhabitants of these
colonies contributed heavily to the number of refugees
who fled west trying to escape drought and famine.
At the national level, General Mengistu justified these
forced relocations as a national obligation to colonize
southern lands with inhabitants from the over-populated north, but, as a whole, the villagization program
was regarded as a draconian measure to displace
populations from areas hostile to the central government into distant regions (Clay and Holcomb, 1986).
Looking at the number and place of origin of the
‘‘villagized’’ population reveals that the program was
conducted with the purpose of taking dissidents from
the provinces of Tigray and Wello into regions where
they became disenfranchised ethnic minorities (Fig. 4).
As dryness and hunger gradually spread in the wake
of the 1982–83 ENSO event, forcible resettlement—
different from villagization in that peasant families from
the highlands were simply transported to remote areas
of Metekel (western Gojjam), Wellega, Illubabor, and
Keffa—was imposed by the Mengistu regime. This
policy had the negative result that native populations
from these regions were displaced by the newcomers,
and violent encounters occurred between natives and
resettled families. This was to be expected, because most
‘‘resettlements were located in areas troubled by
insurgent activities’’ (Gebre, 2002, p. 268) and thus the
government intended to tackle two problems at once
(famine and ethnic tension). It was not long before
famine encroached upon local populations as well as
resettled peasants, and they began to flee into neighboring Sudan, this time as drought-famine refugees. Similar
developments also occurred in the villages of Bale and
Hararge, where it was ethnic-Somali Ethiopians who
were escaping deprivation and hunger (Khalif and
Doornbos, 2002).
The climatic component of the Ethiopian tragedy is
obvious when considering the impact that the droughts
and famines (associated with the two El Niños: 1972–73
and 1982–83) had in influencing political developments
and altering population flows in the country. However,
this was not the end of the duress and deprivation.
During the prolonged warm ENSO event ensued from
1991 to 1993, even though precipitation was not
critically low, the distress caused by more than 12 years
of dryness and food scarcity—added to the excesses of
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
119
Fig. 4. Participants in resettlement, 1984–86, showing settlers’ source (origin) and destination weredas and old provincial boundaries. Based on
government data on the origins of 365,000 settlers and the destinations of 425,000 settlers. Shaded weredas experienced in- or out-migration.
the Mengistu regime and the fatigue caused by constant
warfare—caused a famine that affected 6.7 million
people (Wolde-Georgis et al., 2001).
The political faction that deposed General Mengistu,
the Ethiopian People’s Revolutionary Democratic Front
(EPRDF), was not to be exempted from the vagaries of
climate. Consecutive droughts in 1991–92 and 1993–94
caused as much distress as during the 1980s. Moreover,
in 1997–98, a drought induced by yet another El Niño
cost nearly $28 million in damages. This drought was
followed by catastrophic flooding in 1998 that compounded the shortage of food (Wolde-Georgis et al.,
2001). As usual, there was rural exodus and city inmigration, both exacerbated by the displacement of
nearly 350,000 Ethiopians who fled from the border
region with Eritrea when boundary disputes led to
armed hostilities between the two countries (WoldeGeorgis et al., 2001). Once again, a climatic crisis had
worsened a political conflict to bring more distress to the
Ethiopian population.
5. Changes in population distribution: data and caveats
To determine whether the spatial patterns of Ethiopian population change reflect the geographical variation of the drying effects of El Niño or are the
consequence of forced migration policies, it was
necessary to create a map with sufficient geographical
resolution to display regional variations. Ethiopia
conducted a census in 1984, just at the wake of the
major 1982–83 El Niño and before the moderate event
of 1986. There was a second census in 1994, which
probably captures the effects of a less pronounced event
in 1991–93 and the results of nearly 17 years of forced
population displacements.
Though the 1994 census was partially incomplete, due
to political problems in the eastern part of the country,
the final enumeration was completed in 1996–97. This
information was used to extrapolate values for certain
administrative units in 1994. The census also omitted
parts of the regions enumerated in 1994, with the
information deficit made up by estimation, whose
accuracy it is impossible to determine. For example, in
Gambella region, ‘‘a certain rate of growth of population was utilized’’ to project data from the 1984 census
forward (Central Statistical Authority (CSA), 1995, p.
6). However, these are the only two ‘‘full, modern’’
national censuses for Ethiopia according to the US
Census Bureau’s listing of international census dates
(previous Ethiopian demographic reports were based on
sample surveys only) (Central Statistical Office (CSO),
1971; US Census Bureau, 2001). The simplest way to
look at demographic change over time is to map change
by province between 1984 and 1994 assuming no
changes to the geography of the enumeration unit.
However, the administrative boundary reorganization
precluded this. Therefore, the following discussion is
partly dependent on the authors’ processing of census
data into a usable format.
ARTICLE IN PRESS
120
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
6. Data and cartographic methods: reconciling historical
demography
Though Ethiopia conducted censuses in 1984 and
1994, tracking demographic change through this period
proves very difficult because the Ethiopian government
restructured its administrative boundaries twice in the
intervening decade. For the sake of simplicity, all
geographic references in this paper are to the 13 familiar
older provinces (Fig. 5). Conversion of data between
different sets of administrative boundaries is a common
problem in demographic mapping, and can only be
accurately accomplished when either (a) one knows the
location of every person or (b) all administrative units in
one data set can be exactly amalgamated or translated
into units in the other data set. Otherwise, one must
either aggregate old units to new unit boundaries by a
weighting method (Simpson, 2002) or create a third set
of units that subsumes both original sets. Because the
Ethiopian reorganization affected not only the highestlevel administrative unit (provinces), but also the next
level (awrajas or regions), the third level (weredas), and
possibly even the fourth level (kebeles, or peasant
associations), re-aggregation of data from one census
into the provinces existing at the time of the other census
proved impractical. Even if it were possible, the number
of provinces has always been too small for detailed
analysis: omitting now-independent Eritrea, there were
13 in 1984 and 11 in 1994.
The standard alternative, assigning weights to parts of
administrative units, also introduces error because
weighting methods depend upon assumptions about
population distribution within enumeration units, e.g.
that population is concentrated in cities. In order to
minimize this error, two gridded world population data
sets have been created: the Gridded Population of the
World (GPW) version 2 (Center for International Earth
Science Information Network (CIESIN), International
Food Policy Research Institute, and World Resources
Institute, 2000) and Landscan version 3 (Oak Ridge
National Laboratory (ORNL), 2001). Instead of manipulating administrative units, these divide the world into
regular grid squares, and assign populations to each
square based on weighting methods. Unfortunately,
available gridded data are not suited to the present
analysis due to problems of accuracy and consistency.
The GPW for Ethiopia is based on 1984 census data,
projected to subsequent years based on estimated annual
growth rates. The Landscan program uses the latest
available projections provided by the US Census
Bureau. Thus, neither data set is derived from unedited
census data, but rather on census-based projections that
introduce additional error (data quality issues are
acknowledged for Ethiopia on the GPW website).
Equally important, in order to detect change over the
desired time period, one would need to combine GPW
and Landscan because the former is based on the 1984
census while the latter is based on newer data. This is
impossible because (1) grid cell sizes differ, and (2) GPW
and Landscan use different methods for assigning
population to grid cells: GPW assumes that population
density is constant within weredas, while Landscan uses
a weighting method based on road proximity, urban
area location, land cover, and other variables. Thus, use
of gridded data would neither improve accuracy nor
encompass the desired inter-censal period (1984–94).
Fig. 5. Old provincial boundaries (pre-1987).
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
The method chosen to address the problems caused
by administrative boundary restructuring was to design
a set of new geographic units, approximating real
political boundaries as much as possible but deviating
where necessary due to changes in the boundaries of
lower-level administrative units. For the construction of
the new map, the following data sources were used:
*
*
*
Base maps of 1984 provinces, regions, and weredas in
paper (Ethiopian Mapping Authority (EMA), 1988;
Tareke, 1991) and GIS formats (US Agency for
International Development Famine Early Warning
System Network (USAID FEWS NET), 2002);
Base maps of 1994 provinces and weredas, in paper
and graphic file format (Graham, 1997; United
Nations Development Program (UNDP), 1999);
Population figures for provinces and weredas from
the 1984 and 1994 censuses (Central Statistical
Authority (CSA), 1998; US Agency for International
Development Famine Early Warning System Network (USAID FEWS NET), 2002)
The ultimate purpose was to aggregate weredas from
both 1984 and 1994 into a set of new regions whose
boundaries could be held constant for the analysis of
data from both censuses, thus permitting the sketching
of an accurate map of population change. The task was
difficult because, with the rearrangement of administrative boundaries, some weredas were split or combined, others were partly annexed to neighboring
weredas, and a few changed name. As some new
weredas spanned the old provincial boundaries, and
vice versa, it was impossible to simply re-aggregate
weredas to the other set of provincial boundaries. Use of
121
provincial units is further complicated by the inconsistency of different sources in their delineation of
Ethiopian administrative boundaries for any single time
period (Grepperud, 1996).
Fortunately, the boundaries of many weredas remained constant between 1984 and 1994. These served
as a starting point for amalgamation of 1984 weredas
into a new set of 30 regions called ‘‘Merged Zones’’,
accomplished through geographic analysis using base
maps and GIS. As no digital map of 1994 boundaries
was available, weredas in the 1984 map were combined
to create larger units that also encompassed groups of
1994 weredas. This was eased by the fact that many
weredas did not change between 1984 and 1994,
according to inspection of all available digital and
paper map sources. Weredas that did change were
combined with their neighbors until the border of the
combined region (Merged Zone) could be held constant
between 1984 and 1994. The Merged Zones (MZs) have
the advantage of encompassing certain populations,
because they are all composed of entire weredas. No old
or new wereda is split between two MZs. Every old or
new wereda that crosses a new or old provincial
boundary is included within a single MZ, thus eliminating the problems inherent in weighting. Creation of the
MZs is illustrated in Fig. 6, where dotted lines indicate
wereda boundaries, solid lines show region/awraja
boundaries, and heavy black lines represent 1984
provincial boundaries. Five sample MZs are numbered
and shaded in gray (for all MZs, see Fig. 7). Whenever
possible, MZs were designed to be equivalent to a
region/awraja (MZ 1), a province (MZ 2, equivalent to
Arssi province), or combinations of awrajas (MZ 3). For
Fig. 6. Construction of MZs, showing examples cited in text.
ARTICLE IN PRESS
122
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
Fig. 7. Population change in MZs, 1984–1994 (national average +34%).
example, MZ 1 is equivalent to an awraja within
Hararge province, while MZ 3 is equivalent to a group
of four awrajas within Bale province. MZs 4 and 5 are
more complex. MZ 4 lies entirely within Gondar
province, but does not consist of entire awrajas. One
awraja is split between MZ 4 and another MZ. The split
portion of the awraja is indicated by cross-hatching. This
was necessary when weredas, boundaries were altered so
that they crossed pre-existing awraja boundaries. In the
case of MZ 5, new weredas crossed old provincial
boundaries resulting in an MZ spanning parts of three
old provinces. MZs 4 and 5 do, however, lie entirely
within new provinces, and are approximately equivalent
to new second-level administrative units. Finally, the
largest MZs (Fig. 7) represent areas where weredas could
not be amalgamated within any configuration of either
old or new provincial or regional boundaries and it was
necessary to create more extended zones to avoid splitting
weredas. Thus, all MZ borders represent constant
boundary lines—lines that served as wereda borders in
both 1984 and 1994 and that were not affected by wereda
splits or mergers. Where possible, these lines also coincide
with existing regional or provincial boundaries, and MZs
are therefore not arbitrary divisions of territory. Beyond
the unavoidable problems inherent in census statistics,
e.g. miscounting, it is likely that the populations of the 30
MZs are as good approximations of truth as one can
achieve using available data.
7. Interpretation of mapped changes
Ethiopian population can be divided into three
geographic regions: the central uplands that contain
the majority of the country’s population; the relatively
unpopulated and arid south; and the northern and
central areas with moderate population density. Between 1984 and 1994, the areas of fastest growth (Fig. 7)
were mostly located in the far west and east, in Hararge
and Illubabor provinces. These were zones of very low
population density, so this rapid change involved a
relatively small number of people, but it is still indicative
of actual growth relative to neighboring areas. Regions
in the eastern half of the country grew faster than the
national average of 34%, while the western half
experienced slower growth. Increased growth also
occurred around the capital and in Gamo Goffa and
western Gojjam. Demographic decline or stagnation
took place primarily in several southwestern regions,
along with western Wello and the eastern part of
Hararge adjacent to Somalia.
To test the hypothesis that the enumerated demographic change between 1984 and 1994 reflects the
distant effects of ENSO-induced droughts and decisions
made in the period, a geographic comparison of
population change is appropriate. The regions affected
by droughts and famines have not been consistently
portrayed, as one observes by comparing the maps in
Ethiopian Mapping Authority (EMA) (1988), WoldeMariam (1986), and Webb and von Braun (1994). This
means that the demographic changes shown on Fig. 7
are not likely to be the product of continued, long-term
trends prompted by previous drought episodes. This is
confirmed by the analysis of Kloos and Adugna (1989a),
who found that population growth in the years before
1984 was highest in central and western Ethiopia. Fig. 7
shows high growth in the east and lower growth in all
western areas, with the notable exception of those areas
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
affected by government resettlement programs. This
shift occurred, at least in part, as consequence of the
ENSO episodes of 1982–83 and 1986, and was
aggravated by subsequent political actions and resettlement policies. The latter include forced migration or
population relocation and villagization of formerly
nomadic or scattered rural inhabitants.
According to Wolde-Mariam (1986), the most ‘‘famine-prone’’ regions of Ethiopia are the northeast, much
of the south, and some central areas, with the largest
regions corresponding roughly to Tigray and Wello and
southern and eastern Hararge, Bale, and Sidamo. Webb
and von Braun’s map of areas ‘‘most affected by famine
and drought’’ in the 1980s roughly corresponds to this,
with the addition of Illubabor and portions of the
central part of the country. However, these areas also
include most of the regions of fastest population growth
detected by our mapping, suggesting that drought alone
did not have a significant effect on net demographic
change. Studies by Dejene (1990) and Grepperud (1996)
show that population pressure (the result of an annual
growth rate approaching 3% for the past several
decades) has negatively affected soil quality, thus
reducing agricultural productivity. More striking, Ezra
(2001) found that land has become so scarce in Wello
and Tigray (the origin regions of most settlers in 1984–
86) that not only were some people willing to migrate,
but there has been a decline in the fertility rate relative
to other parts of the country. In this situation, drought
only exacerbated the pre-existing long-term problem of
over-intensive land use, without necessarily having a
major effect on population change within the 1984–1994
period.
It is more likely that the mapped changes reveal a link
between population change and the resettlement/villagization that occurred primarily between 1984 and 1986,
a period that corresponded with generalized rainfall
anomalies in eastern Africa, and when the movement of
‘‘drought-affected population’’ in Ethiopia reached its
peak (Webb and Habtu, 1994). According to Clay and
Holcomb (1986), the Ethiopian government moved
people from the northern part of the country to the
west, mostly toward the provinces of Keffa, Illubabor,
and Wellega. A second flow took place between the
north and northwest, primarily toward Gondar and
western Gojjam. Some resettled people returned to their
northern homes, often through Sudan, landing in or
passing through the same northwestern area. Pankhurst
(1991) estimates that up to a quarter of settlers left their
new villages, but does not offer an estimate as to how
many of them returned to their original homes. Thus, a
large majority of settlers remained outside their regions
of origin and it is no surprise that areas of governmental
resettlement (Fig. 4) and population growth (Fig. 7)
show some correlation. As noted above, most relocated
migrants originated from Tigray or Wello (Ethiopian
123
Mapping Authority (EMA), 1988; Dejene, 1990), and
our map shows either decline or below-average population growth in these regions.
Existing sources do not coincide in determining
exactly how many people were resettled. An official
government survey (Central Statistical Authority (CSA),
1991) gives a figure of 425,000, while independent
sources place the number at 500,000 (Adugna, 1989)
or nearly 600,000 (Pankhurst, 1992; Gebre, 2001),
though, according to Woldemeskel (1989) and Adugna
(1989), the original plan was to resettle as many as
1,500,000, or about 20% of the total number of famine
victims. This uncertainty reflects the fact that there has
always been seasonal and inter-regional migration in
Ethiopia (Kloos and Adugna, 1989a; Berhanu and
White, 2000) largely on a voluntary and non-governmentally sponsored basis and government resettlement
programs began only after the earlier famine in 1973–74
(Dejene, 1990; Ethiopian Mapping Authority (EMA),
1988). In some cases, resettlement meant only local
moves, but for most migrants, the journey involved
crossing several provincial boundaries, often into
regions with significantly different climatic and agricultural conditions (Pankhurst, 1992; Colchester and
Luling, 1986). Though the settlement program has been
described as coerced, many settlers moved quasivoluntarily, deceived by government propaganda promising a ‘‘land of milk and honey’’ with new housing
and tractors (Pankhurst, 1991). Aside from intracountry movement, Ethiopia has been ‘‘one of the most
important refugee-generating countries of the world for
the past 20 years’’ (Bariagaber, 1997, p. 30). A further
large, but uncounted number of people crossed into
Sudan as a result of resettlement and other government
policies in the 1980s (Kibreab, 1996), but once again, the
number of individuals involved remains unknown.
The geographic distribution of the destinations of
migrants according to the government statistical office
survey (Central Statistical Authority (CSA), 1991) is
shown in Fig. 4. Though these data are certainly
incomplete, the geographic patterns agree with those
depicted in Kloos and Adugna (1989b), whose data
came from a different Ethiopian government agency
(the former Relief and Rehabilitation Commission, now
the Disaster Prevention and Preparedness Commission).
Most migrants were sent to west-central areas, including
several low-density regions reporting above-average
population change between 1984 and 1994 (Fig. 7). In
the context of general internal migration, small and/or
sparsely populated regions are more likely to display
higher rates of population change because they have
lower base populations. Thus, it is not surprising that
Illubabor and western Gojjam registered more rapid
growth than other areas receiving large numbers of
migrants. Though settlement areas were ostensibly
selected for their good, unoccupied agricultural land,
ARTICLE IN PRESS
124
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
Fig. 8. Settler proportion of the population. Participants in resettlement in 1984–86 as percent of 1984 population of their source (origin) and
destination weredas, with old provincial boundaries. Based on same source as Fig. 7. Wereda boundaries shown only for regions experiencing in- or
out-migration.
these areas were, in fact, already inhabited by indigenous groups, and did not have sufficient resources to
support greatly expanded populations, not to mention
the presence of diseases unfamiliar to settlers from
highland areas (Woldemeskel, 1989). In Metekel (western Gojjam), for example, the settler population
reached about 82,000 by 1988, but then declined
sharply, as food shortages led to high mortality and
out-migration, falling below 30,000 by the 1994 census
(Gebre, 2001). The effect of migration on receiving
weredas (based on government data only) is shown in
Fig. 8 as a percent of 1984 census population. Migration
had the most effect on sparsely populated areas near the
Sudanese border, reflecting, in part, the return of the
peoples who fled to that country as refugees. Similarly,
since Wello is cited by Pankhurst (1992) as the largest
source region of migrants, it is not surprising to see a
demographic decline in western Wello (Fig. 7).
The second type of migration involved a transfer of
sparse population from rural to non-rural settlements.
Villagization of formerly non-urban population in
1984–87 mostly occurred in the central, relatively
densely populated regions around Addis Ababa (Clay
et al., 1988), away from the sparsely inhabited border
areas that served as resettlement destinations. For this
reason, population change was mostly near the national
average in this region, with above average growth near
the capital and areas of decline in the north (Wello) and
south (Fig. 7). Population growth in the Somali region
reflects the inflow of escapees from the areas of
villagization. Another reason for the population increases in certain troubled spots is cityward migration,
which in most developing countries has swollen urban
centers in the past few decades. In the case of Ethiopia,
this trend was reinforced by the migration of former
rural-dwellers escaping government coercion. For example, one way to escape villagization and other forms
of resettlement was by migrating directly to towns or
cities (Berhanu and White, 2000). In addition, one has to
take into consideration that famine claimed the lives of
one-half to one million people (Dejene, 1990; Ezra,
2001). The famine led to not only a great increase in
mortality but also to a substantial decline in fertility
(Kidane, 1989), so that the net effect on population
change was greater than if only one component was
affected.
8. Migrants and ethnic change
The main ethnic/linguistic groups in Ethiopia as of
the mid-1980s, along with the old provincial boundaries,
are shown in Fig. 3. Ethiopians are much more diverse
in ethnicity, language, and religious affiliation than can
be indicated by any single map. No region of the
country is culturally homogeneous, though comparisons
with Fig. 9, demonstrates that the most recent redrawing
of provincial boundaries was intended to create regions
with more cohesive ethnic identities (compare to map in
Keller, 1995). Data from the 1994 census do not show
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
125
Fig. 9. New provincial boundaries (post-1991).
dramatic changes in the national ethnic pattern, but
resettlement caused very significant changes at the local
level, even if some migrants subsequently fled into
Sudan or returned to their original homes. Figs. 2 and 3,
show the destination regions of most resettlement are
dominated either by smaller minority ethnic groups (of
the Nilotic or Omotic language groups), or by the
Oromo, or do not have any single dominant ethnic
group. Since the primary sources of settlers were
northern regions with large Amharic or Tigrayan
majorities (compare Figs. 2 and 3), it is clear that one
effect of the large southwesterly migration was to
increase the presence of Amharic and Tigrayan people
in areas where they were previously absent (or nearly
so), while diluting the demographic dominance of
minorities or the Oromo. The effect of such change
was magnified by the fact that before 1984, government
data on internal migration indicate that long-distance
migration was relatively rare in Ethiopia (Central
Statistical Authority (CSA), 1992). As the 1984 census
report for Illubabor states, for example, ‘‘streams of
migration into the region seems to be reflections [sic] of
the generally low level of inter-regional mobility’’
(Central Statistical Authority (CSA), 1989, p. 219),
and the statistics on in-migrants show that most
migrants to the area originated from the two neighboring provinces of Wellega and Keffa. Illubabor then
experienced very fast population growth between 1984
and 1994, and was the destination of many thousands of
migrants from much farther away. The same is true for
Gojjam, which received relatively little migration from
the northern provinces of Wello and Tigray before 1984
(Central Statistical Authority (CSA), 1990) but whose
Metekel region became a major resettlement area.
Because the settlers’ destination regions had lower
population density than their source regions (compare
Figs. 1 and 3), resettlement had a much larger effect on
the composition of the population in the receiving
regions compared to the regions of departure (Fig. 8).
9. Conclusion
Of the many countries in Africa affected by ENSO
events, Ethiopia is unique in that documented population movements can be explained partially as the result
of environmental crises (droughts/famines) caused by
climatic variations. In contrast to other African
countries affected by climatic crises related to El Niño,
Ethiopia has established a regular demographic enumeration schedule, and conducted censuses before and
after the 1982–83 and 1991–93 El Niño events. Yet the
country is not wealthy enough to respond to or mitigate
such crises through government aid intervention alone.
In Ethiopia, a climatic crisis accentuated a political
crisis, and one of the ways in which the government
responded to both problems was through imposed
migration policies. Because migration (forced and
voluntary) was mostly directed toward more sparsely
populated areas and led to changes in the ethnic
ARTICLE IN PRESS
126
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
balance, the demographic effect of El Niño can be
observed in Ethiopian demographic records and
through the type of cartographic analysis presented in
this paper. Though there are many catalysts promoting
population variability due to displacements in any
country, Ethiopia appears to be a good example of
climatic and political agents interacting to elicit significant changes in population distribution.
References
Aalen, L., 2002. Ethnic Federalism in a Dominant Party State: the
Ethiopian Experience 1991–2000. Chr. Michelsen Institute, Bergen,
Norway.
Adhana, A.H., 1991. Peasant response to famine in Ethiopia,
1975–1985. Ambio 20, 186–188.
Adugna, A., 1989. The 1984 drought and settler migration in Ethiopia.
In: Clarke, J.I., Curson, P., Kayastha, S.L., Nag, P. (Eds.),
Population and Disaster. Blackwell, Oxford, England, pp. 114–127.
Bariagaber, A., 1997. Political violence and the uprooted in the Horn
of Africa: a study of refugee flows from Ethiopia. Journal of Black
Studies 28, 26–43.
Berhanu, B., White, M., 2000. War, famine, and female migration in
Ethiopia, 1960–1989. Economic Development and Cultural
Change 49, 91–114.
Caviedes, C.N., 1982. Natural hazards in South America: in search of
a method and a theory. GeoJournal 6, 101–111.
Caviedes, C.N., 2001. El Niño in History. University Press of Florida,
Gainesville.
Center for International Earth Science Information Network (CIESIN), International Food Policy Research Institute, and World
Resources Institute, 2000. Gridded Population of the World,
Version 2. CIESIN, Columbia University, Palisades, NY. Online:
http://sedac.ciesin.columbia.edu/plue/gpw.
Clay, J.W., Holcomb, B.K., 1986. Politics and the Ethiopian Famine
1984–1985. Cultural Survival, Cambridge, MA.
Clay, J.W., Steingraber, S., Niggli, P., 1988. The Spoils of Famine.
Ethiopian Famine Policy and Peasant Agriculture. Cultural
Survival, Cambridge, MA.
Central Statistical Authority (CSA), 1989. Population and Housing
Census 1984: Analytical Report on Illubabor Region. Central
Statistical Authority, Addis Ababa.
Central Statistical Authority (CSA), 1990. Population and Housing
Census 1984: Analytical Report on Gojjam Region. Central
Statistical Authority, Addis Ababa.
Central Statistical Authority (CSA), 1991. Report on the census of
population and housing and on survey of demographic changes in
the resettlement areas, Central Statistical Authority, Addis Ababa.
Central Statistical Authority (CSA), 1992. Report on Experimental
Vital Registration System (ESVRS) in Rural Ethiopia, Vol. 2.
Central Statistical Authority, Addis Ababa.
Central Statistical Authority (CSA), 1995. The 1994 Population and
Housing Census of Ethiopia Results for Gambella Region, Vol. 1:
Statistical Report. Central Statistical Authority, Addis Ababa.
Central Statistical Authority (CSA), 1998. The 1994 Population
and Housing Census of Ethiopia Results at Country Level,
Vol. 1: Statistical Report. Central Statistical Authority, Addis
Ababa.
Central Statistical Office (CSO), 1971. Population of Ethiopia: Results
from the National Sample Survey 1st Round, 1964–1967. Central
Statistical Office, Addis Ababa.
Colchester, M., Luling, V. (Eds.), 1986. Ethiopia’s Bitter Medicine:
Settling for Disaster. Survival International, London.
Dejene, A., 1990. Environment, Famine, and Politics in Ethiopia:
a view from the village. Lynne Rienner, Boulder, CO.
Ethiopian Mapping Authority (EMA), 1988 National Atlas of
Ethiopia. Ethiopian Mapping Authority, Addis Ababa.
Ezra, M., 2001. Demographic responses to environmental stress in the
drought- and famine-prone areas of northern Ethiopia. International Journal of Population Geography 7, 259–281.
Gebre, Y.D., 2001. Population displacement and food insecurity in
Ethiopia: resettlement, settlers, and hosts. Doctoral Dissertation
(Anthropology), University of Florida, Gainesville.
Gebre, Y.D., 2002. Contextual determination of migration behaviours:
the Ethiopian resettlement in light of conceptual constructs.
Journal of Refugee Studies 15, 265–282.
Graham, J., ca. 1997 (no date). The Southern Region at a Glance.
Agricultural Bureau, Southern Nations Nationalities Peoples
Region, Awassa, Ethiopia.
Grepperud, S., 1996. Population pressure and land degradation: the
case of Ethiopia. Journal of Environmental Economics and
Management 30, 18–33.
Hancock, G., 1985. Ethiopia: the Challenge of Hunger. Victor
Gollancz, London.
Hastenrath, S., 1990. Decadal-scale changes of the circulation in the
tropical Atlantic sector associated with Sahel drought. International Journal of Climatology 10, 459–472.
Henze, P.B., 1995. Ethiopia and Eritrea in Transition: the impact of
ethnicity on politics and development. RAND, Santa Monica, CA.
Kaplan, I., Farber, M., Marvin, B., McLaughlin, J., Nelson, H.D.,
Whitaker, D., 1971. Area Handbook for Ethiopia [includes
Eritrea]. Government Printing Office, Washington, DC.
Kates, R.W., 1981. Drought Impact in the Sahelian–Sudanic Zone of
West Africa: a comparative analysis of 1910–1915 and 1968–1974.
Department of Geography, Clark University, Worcester, MA.
Kebbede, K., 1988. Cycles of famine in a country of plenty: the case of
Ethiopia. GeoJournal 17, 125–132.
Keller, E.J., 1992. Drought, war, and the politics of famine in Ethiopia
and Eritrea. Journal of Modern African Studies 30, 609–624.
Keller, E.J., 1995. The ethnogenesis of the Oromo nation and its
implications for politics in Ethiopia. Journal of Modern African
Studies 33, 621–634.
Khalif, M.H., Doornbos, M., 2002. The Somali Region in Ethiopia: a
neglected human rights tragedy. Review of African Political
Economy 91, 73–94.
Kibreab, G., 1996. Eritrean and Ethiopian urban refugees in
Khartoum: what the eye refuses to see. African Studies Review
39, 131–178.
Kidane, A., 1989. Demographic consequences of the 1984–1985
Ethiopian famine. Demography 26, 515–522.
Kiros, P.G., 1991. Economic consequences of drought, crop failure
and famine in Ethiopia, 1973–1986. Ambio 20, 183–185.
Kloos, H., Adugna, A., 1989a. The Ethiopian population: growth and
distribution. The Geographical Journal 155, 33–51.
Kloos, H., Adugna, A., 1989b. Settler migration during the 1984/85
resettlement programme in Ethiopia. GeoJournal 19, 113–127.
Mattsson, J.O., Rapp, A., 1991. The recent droughts in western
Ethiopia and Sudan in a climatic context. Ambio 20, 172–175.
McNeill, J.R., 2000. An Environmental History of the TwentiethCentury. Norton, New York.
Nicholson, S.E., 1997. An analysis of the ENSO signal in the tropical
Atlantic and western Indian Oceans. International Journal of
Climate 17, 345–375.
Oak Ridge National Laboratory (ORNL), 2001. Landscan Global
Population Database, version 3. Oak Ridge National Laboratory,
Oak Ridge, TN. Online: http://www.ornl.gov/gist/landscan.
Pankhurst, R., 1990. The History of Famines and Epidemics in
Ethiopia prior to the Twentieth Century. Relief and Rehabilitation
Commission, Addis Ababa.
ARTICLE IN PRESS
J. Comenetz, C. Caviedes / Environmental Hazards 4 (2002) 113–127
Pankhurst, A., 1991. People on the move: settlers leaving Ethiopian
resettlement villages. Disasters 15, 61–67.
Pankhurst, A., 1992. Resettlement and Famine in Ethiopia: the
Villagers’ Experience. Manchester University Press, Manchester,
England.
Rubenson, S., 1991. Environmental stress and conflict in Ethiopian
history: looking for correlations. Ambio 20, 179–182.
Simpson, L., 2002. Geography conversion tables: a framework for
conversion of data between geographical units. International
Journal of Population Geography 8, 69–82.
Sirocko, F., 1996. Past and present subtropical summer monsoons.
Science 274, 937–938.
Tareke, G., 1991. Ethiopia: Power and Protest. Cambridge University
Press, Cambridge, England.
Tareke, G., 2002. From lash to red star: the pitfalls of counterinsurgency in Ethiopia, 1980–82. Journal of Modern African
Studies 40, 465–498.
United Nations Development Program (UNDP), 1999. United
Nations Development Program Emergencies Unit for Ethiopia:
Databases Found at UNDP-EUE. Online: http://www.africa.
upenn.edu/eue web/map des.htm. Last update September 1, 1999.
US Census Bureau, 2001. Census Dates for Countries and Areas of the
World: 1945–2004. Online: http://www.census.gov/ipc/www/
cendates/. Last update August 15, 2001.
127
US Agency for International Development Famine Early Warning
System Network (USAID FEWS NET), 2002. Africa Data
Dissemination Service. Online: http://edcintl.cr.usgs.gov/adds/
adds.html. Accessed February 1, 2002.
Webb, P., Habtu, Y., 1994. Famine in Africa. National Geographic
Research and Exploration 10, 158–171.
Webb, P., von Braun, J., 1994. Famine and Security in Ethiopia.
Wiley, Chichester, England.
Whetton, P.H., Rutherford, I., 1994. Historical ENSO teleconnections
in the Eastern Hemisphere. Climatic Change 28, 221–253.
World Meteorological Organization (WMO), 1985. The global climate
system. A critical review of the climate system during 1982–1984.
World Meteorological Organization/World Climate Program,
Geneva.
Wolde-Georgis, T., Aweke, D., Hagos, Y., 2001. The case of Ethiopia:
reducing the impacts of environmental emergencies through early
warning and preparedness: the case of the 1997–98 El Niño. In:
Glantz, M.H. (Ed.), Once Burned, Twice Shy? Lessons Learned
from the 1997–98 El Niño. United Nations University Press,
Tokyo.
Wolde-Mariam, M., 1986. Rural Vulnerability to Famine in Ethiopia
1958–1977. Intermediate Technology Publications, London.
Woldemeskel, G., 1989. The consequences of resettlement in Ethiopia.
African Affairs 88, 359–374.