Soil erosion from hilltribe opium swiddens in the golden triangle, and the use of karren as an erosion yardstick
ENDINS, n.O 13.1987. Ciutat de Mallorca.
SOlL EROSION FROM HlLLTRlBE OPlUM
SWIDDENS IN THE GOLDEN TRIANGLE,
ANDTHEUSEOFKARRENASAN
EROSION YARDSTICK
by Kevin KIERNAN *
El tipus de cultiu itinerant realitzat per les tribus de les muntanyes així com la cremada regular
dels boscos en les arees productores d'opi del Triangle d'Or, són responsables de seriosos danys en
els recursos del sol. L'estudi de les formes exhumades de lapiaz, que presumiblement s'han desen-
volupades sota la superficie en lloc de fer-ho en condicions subaeries, es presenta com a promete-
dor mitja per documentar I'erosió del sol; no obstant aixo hi ha molts de problemes que romanen
cense resoldre, el que fa que encara no es puguin aconseguir resultats segurs.
Abstract
Shifting cultivation by hilltribes and regular burning of the forests in opium-producing areas of
the Golden Triangle is responsible for serious damage to the m i l resources. The study of exposed
karren forms that are assumed to have developed under subsurface rather than subaerial conditions
has come promise as a means of documenting soil erosion, although serious problems remain to
be resolved before reliable results can be achieved.
lntroduction
While the social costs of drug addiction and
ces. Many of these localities are underlain by Per-
, associated criminal activity are well known, the en-
mian limestones that have given rice in the wet
vironmental costs of the illicit drug trade gain far
tropical climate to deep terra rossa soils. This note
less publicity. A large proportion of the world's illi-
records an opportunistic reconnaissance of the
cit heroin supply originates in the opium swiddens
erosion of these soils in one such remote centre of
of subsistance hilltribe farmers near the common
opium production. It also explores the potential
borders of Burma, Thailand and Laos, an area po-
usefulness of certain karren forms as surrogate
pularly known as the Golden Triangle. The heroin
measures of soil erosion.
is refined in mobile laboratories under the control
of private armies. Shifting cultivation has led to se-
vere environmental deterioration in this region, in-
Ph ysical Environment
cluding serious soil erosion from swiddens (Hurni,
1982; Dunkley, 1985). Coupled with pressures from
The Red Lahu hilltribe communities of Pha
the western world to stem the flow of heroin this
Puek and Pha Daang lie close to the Burmese bor-
has stimulated efforts by the Thai government to
der in the limestone mountains of far north-west-
promote sedentary cultivation of alternative crops.
ern Thailand (figure 1). The area is located at about
Nonetheless, opium production continues, par-
19O 14'N at an elevation of 960-1020 m. asl. It is
ticularly in remote and insecure parts of the region
drained by the Huai Pong Saen Pik, a minor tribu-
that lie beyond the control of Thai government for-
tary of the Salween River. The climate is of Koep-
- -
-
pen Agw type and is characterised by a cool-dry
Department of Geography, University of Tasmania.
season from December to February, a hot-dry sea-

turalists may have been present in these mountains
since the close of the Pleistocene (Gorman, 1966)
the advent of increasing population pressures,
opium production and the environmental problems
associated with its cultivation, are comparatively
recent. Opiurn was probably introduced to China
by Arab traders in the seventh century, but imports
mushroomed from 1829-1839 when an average of
1841 tonnes of lndian opium was taken to China
each year by Britain to pay for silks and tea. At-
tempts by China to halt this influx led to the Opium
War of 1839-1842 between the two. With China's
defeat this drug traffic increased further and by the
1880s reached 65,000 tonnes each year. With ad-
diction now widespread, China began to encou-
A P h a P u e k
rage opium production by hilltribes in the South-
West in an effort to reduce the drain of money
from the country.
When China was subsequently divided between
rival warlords opium taxes came to provide a means
of financing private armies such as the Kuornin-
tang (KMT). The need to pay these taxes forced the
hilltribes to increase production. By the 1920s
A
groups such as the Lahu were escaping southwest-
wards into the Golden Triangle. Following the
cornrnunist takeover in China the nacionalist KMT,
with support from the USA and Taiwan, made three
attempts to recapture Yunnan. When they failed
they were abandonned by these allies, and were
left almost totally reliant upon opiurn for funding.
k l l o r n e t r e e
This cornpounded the pervasiveness of the indus-
try. In addition, anti-government forces in the Shan
Figure 1. Location of the Pha Puek-Pha Daang area.
State of Burma, particularly the SUA, are also de-
pendant on opiurn taxes and heroin refining for fi-
son from March to June; and a hot-wet season from
nance. Western concern to see the maintenance of
July to November. Rainfall data are not available
an anti-communist buffer on Thailand's northern
but about 30 km. south-east of the study area at
frontier has further complicated eradication of the
Huai Thung Choa, downwind of the study area, it
illicit heroin industry. However, proposals by some
may considerably exceed 2000 mm. pa. Some 90 %
Shan groups to sell al1 their output to the west for
of the annual total falls between May and Novem-
medicinal purposes in exchange for political recog-
ber (Hurni and Nuntapong, 1983). The mountains
nition have been rejected and much of the world's
are aligned north-south and orographic effects are
medicinal supply is grown in economically more
probably considerable.
Origin and character
of the Agricultural System

The production of opium is important to the
economies of both Pha Puek and Pha Daang. These
opium fields probably lie within the catchment of
the notorious Burmese warlord Khun Sa, as his
Shan United Army (SUA) maintains a base camp
only a few kilometres to the north. Opium forms an
important cash crop of low bulk that can be easily
transported from remote areas and may be cultiva-
Figure 2. Erosion gully 3-4 m deep cut through colluvial depo-
ted on marginal lands. Although primitive agricul-
sits south-west of Doi Khu.

advantaged parts of the world, such as Tasmania,
were alternative crops are viable.
The opium poppies are cultivated on steep
swiddens cut from the forest and burnt prior to
planting. Under very favourable situations swid-
dens may be used for opium production for up to
ten years (McKinnon, 1983) but more usually they
are cultivated for less than five years. The fallow
period never spans less than ten years and may
have t o last decades. Corn is often first planted as
a food staple and cover crop for the young pop-
pies. These flower in January and the opium is la-
ter removed by incision of the seed head. The
stubble is commonly burnt at the end of the dry
Figure 3. Advanced soil erosion in a w i d d e n 5-8 years old,
season which leaves the ground bare with the on-
south of Pha Puek. Soil has been stripped frorn steep
set of the monsoon. Work by Hurni and Nuntapong
slopes and deposited in the adjacent sinkh~ole. Karren
(1983) suggest that these early rains of late May-
forms of subsurface origin are cornmon on rnany ex-
Early June may be the most intense of the year.
posed outcrops.
nings, 1971). However caution is required because
Evidence of soil erosion
some rounded forms may develop beneath a moss,
lichen or liverwort cover (Bogli, 1980).
The heavily sediment-laden rivers that flow
Estimates of cover loss from two swiddens near
from these mountains each wet season attest to
Pha Puek, based on this approach, are presented in
the extent of soil erosion. This sediment derives
table 1 (n=20 in al1 cases). These indicate loss of
not only from swiddens but also from forest areas
up to 80 cm. of soil and litter, although the propor-
repeatedly razed by humanly ignited wildfire each
tions of each cannot be differentiated. It is highly
dry season. Erosion gullies 4 m or more in depth
unlikely that this loss is attributable solely to swid-
have cut through colluviel accumulations near the
dening as karren forms also indicate considerable
base of mountain ridges (figure 2). Earlier darnage
loss elsewhere from steep forested areas subject
by erosion consequent upon forest burning corn-
to regular burning, and there is every reason to an-
plicates the assessment of erosion from swiddens
ticipate that the area near Pha Puek was burned
that rnay subsequently be cut. Erosion may also
repeatedly prior to clearing. Reconnaissance of
have been initiated during early logging operations
erosion from the forest floor in an area only now
by British and French interests that destroyed the
being converted to swidden severa1 kilometres
teak forests in the area, large parts of which are
south of Pha Puek on the southern slopes of Doi
now cloaked in secondary forest with widespread
Khu also revealed similar maximum losses in loca-
thickets of bamboo. Areas that have been subject
lised areas a!though the average figure was consi-
t o swiddening appear more seriously eroded than
derably less (19.6 cm) than at Pha Puek. If it is as-
the floors of the burnt forests. Extensive slopes of
sumed that a similar average depth may have been
bared limestone outcrops and thick accumulations
stripped at Pha Puek prior to forest clearing, only
of sediment on the floors of sinkholes and caves
20-25 cm of the cover loss there may be attributable
give the impression that.sheet erosion is rapid and
to erosion from the swiddens. However, the karren
continuing (figure 3).
method of estimating cover loss may significantly
over-estimate overall erosion due to preferential
lowering of the ground surface close to outcrops
Nleasuring the soil erosion
by runoff from them.
Solutional karren fluting on exposed limestone
surfaces provides one possible means of estimat-
ing the extent of soil loss. Because the rounded
site
a
b
morphology of karren that develops beneath a soil
aspect
ESE
S
or litter cover (eg. rundkarren and solution pipes)
slope (%)
75
40
contrasts with the sharper crested forms that develop
cover loss: range (cm) 2-80
0-80
subaerially (eg. rillenkarren and rinnenkarren) the
mean (cm) 44.4
40.9
extent to which rounded forms are exposed rnight
provide a measure of the depth of soil lost (Jen-
Table 1. Estimates of cover loss, Pha Puek.

Any estimation of the rate at which erosion
between the base of the burnt wood and the present
proceeds demands some measure of the time pe-
ground surface is assumed t o define the amount of
riod over which the cover has been lost. Rounded
cover lost, burning is assumed to have penetrated
karren forms should undergo progressive sharpen-
t o the base of the litter layer, and fresh and well
ing once exposed to subaerial processes. Where
defined burnlines are assumed to date from the
exposed outcrops occur in deforested areas this
preceeding dry season, then the rate of erosion of
probably occurs fairly rapidly under a monsoonal
the soil may be estimated (table 2).
regime. Precise rates are unknown, but some guide
t o the order of magnitude may be offerred by Indo-
aspect
NNE
nesian data (Balazs, 1968) that indicates an overall
slope (%)
60-70
limestone surface denudation rate of 83 mm. 1000
cover loss:
range (cm)
1-10
yrs. Even though flutes represent sites where rock
mean (cm)
3.5
solution occurs more rapidly than the average it
seems unlikely from this figure that sharpening
Table 2. Soil loss estimates. Pha Daang, May 1985 - May,1986.
would become evident in less than 10-15 years.
This means it is unlikely to occur during the life-
This suggests a mean surface lowering rate of
time of any swidden. Sharpening will probably be
3.5 cm.year which would imply that soil l o s amounts
considerably delayed where a forest canopy inter-
t o 350 tonnes.ha.yr which is consistent with the
cepts much of the rainfall. In addition, because so-
data from Pha Puek. Once again, this figure may
lution processes that give rise to karren proceed
well be too high since runoff from the stumps
most rapidly under warm conditions (Bogli, 1960)
themselves may erode the adjacent ground surface
the shading of outcrops from the sun by a forest
to a greater extent than is typical for most of the
canopy may compound the delay.
swidden. Howing close to the stumps may also be
Two main problems therefore exist. Firstly, the
a factor (cf. Hurni, 1983). Nonetheless, erosion at
exposure of rounded karren in this area seems to
broadly this rate would achieve the surface lower-
offer a useful guide to the total depth of rhizosphe-
ing suggested at Pha Phuek in 5.7-7.1 years, which
re lost, but provides no means by which to fix pre-
is consistent with the reported age of the swiddens.
cisely the proportion represented by soil. Secondly,
Given that these rates are founded on only a
no time control is available. However, local infor-
very limited date base gathered on the run, and in-
mation suggested that the swiddens were cleared
volve untested assumptions, the coincident nature
5-8 years ago. If it is assumed that the soil surface
of the results may be merely fortuitous, and certain-
has been lowered by 22 cm during this time, a sur-
ly they do not represent a realistic vindication of
face lowering rate of between 2.8 and 4.4 cm.year
the use of karren as erosion measures. However,
is implied. If this were al1 soil, which is likely to be
these results do suggest there may be promise in
substantially the case given annual burning, it would
the approach.
ammount to a loss of hetween 280 and 440 tonnes.
ha.yr.
In a brief attempt to verify this rate, soil loss
Conclusion
from a swidden north of Pha Daang was estimated
using basal burnlines on stumps left standing in
The principal problems in the use of karren to
the swidden (figure 4). If the difference in elevation
measure soil erosion include the impracticability of
determining the relative proportions of soil, litter,
and perhaps moss and lichen within the lost cover;
the lack of any presently available time control on
the subaerial modification of exposed subsurface
karren; and the probability of disproportionately
great erosion close to the outcrops themselves.
Come of the more extreme subsurface karren expo-
sures recorded may be the result of rounded forms
having developed well above the ground surface
beneath moss or lichen prior to the advent of regu-
lar anthropogenic forest fires. The circumstances
of the visit and the very limited amount of time
that was available to gather data while moving
quickly through the area also meant that only a
Figure 4. A basal burnline on a sturnp rernaining in a swidden
small number of measurements could be made.
north of Pha Daang. The soil surface has been lowered
by several centirnetres since the swidden was burnt at

The correspondence between the burnline and
the end of the previous dry season.
karren data suggests that any disproportionate

lowering due t o runoff from the outcrops and
Acknowledgements
stumps themselves may have approximately ba-
lanced out. The suggestion that the determined ra-
I am grateful to John Dunkley for kindling my
tes may be excessive is perhaps supported by ero-
interest in the karsts of northern Thailand and to
sion data from granite soils at Huai Thung Choa
John Spies without whose logistical efforts this re-
where Hurni (1983) has documented an average soil
connaissance would not have been possible. This
loss of 120 tonnes.ha.yr. Although it might be anti-
paper has benefitted from earlier comments by
cipated that the granite soils would be even more
Steve Harris, Greg Middleton and Albert Goede to
highly erodible than the limestone soils, the rates
whom my thanks are also extended.
calculated for the Pha Puek-Pha Daang area are ap-
proximately three times those applicable to the
granite swiddens, and exceed even the maximum
Bibliography
figure of 300 tones.ha.yr obtained by Hurni. It
seems improbable that rainfall differences alone
BALAZS, D. (1968): Karst regions in Indonesia. Karszt-es Barlang-
could account for this, nor even the fact that the
kutatas 1963- 1967: 3-67.
Lisu people at Huai Thung Choa appear to go to
BOGLI, A. (1960): Kalklosung und Karrenbildung Z Geomorph.
Supp. 2: 4-21.
greater lengths to minimise soil erosion than do
BOGLI, A. (1980): Karst Hydrology and Physical Speleology. Berlin:
the Red Lahu people in the limestone mountains
Springer-Verlag.
(Hurni, 1983).
DUNKLEY, J.R. (1985): Karst and caves of the Nam Lang-Nam Khong
Despite these difíiculties it is clear that swidden
region, North Thailand. Helictite 23(1): 3-22.
agriculture, abetted by rampant forest incendiarism,
GORMAN, C.F. (1970): Excavations at Spirit Cave, North Thailand:
Some interim explanations. Asian Perspectives 13: 80-107.
is exerting profound effects upon the limestone
HURNI, H. (1982): Soil erosion in Hual Thung Choa, Northern
soils: destruction of the humus layer; volatilisation
Thailand: Causes and constraints. Mountain Research and
of many potential new nutrient inputs; erosional
Development 2(2): 141-156.
loss of the remaining ash fraction that contains a
HURNI, H. & NUNTAPONG, S. (1983): Agro-forestry improvements
for shifting cultivation systems: Soil consewation research in
high proportion of the surviving nutrients; modifi-
Northern Thailand. Mountain Research and Devdopment 3(4):
cation of the vegetation cover; and serious progres-
338-345.
sive reduction of the mineral capital of the soils.
McKINNON, J. (1983): Introductory essay: A Highlandeas Geogra-
Although forest burning causes serious erosion, that
phy of the Highlands: Mythology, Process and Fact. Mountain
caused by swidden agriculture appears to be about
Research and Development 3(4): 313-317.
3-4 orders of magnitude greater if the forest floor
erosion at Doi Khu is assumed to have occurred
over about 40 years and is at al1 representative.
But however serious their effects, these gentle
hilltribe people in their remote mountain commu-
nities are merely responding to historical and eco-
nomic realities that confront them but which were
not of their own making. Those who would change
the situation must be sensitive to this. The occasio-
nal official burning of hilltribe opium crops in res-
ponse t o political pressures from the west may be
more likely to create resentment and hardship than
changed crop preferences or wiser land manage-
ment. And, despite the commendable efforts to the
Thai government to develop sedentary agro-fores-
try alternatives to shifting cultivation of opium, re-
moteness, political instability, and the lack of simi-
lar initiatives north of the border remain a stum-
bling block. With perhaps 30 % of the population
of the southern Shan States economically depen-
dant upon the illicit heroin industry, it is likely that
unless equitable political and economic solutions
can be found to regional problems, physical devas-
tation of the mountains of the Golden Triangle and
social devastation in the cities of the western world
will continue to go hand in hand for some time to
come.