Relationship Between Altitude and Intensity of Malaria Transmission in the Usambara Mountains, Tanzania

Abstract

There is a consensus that malaria is a growing problem in African highlands. This is surprising because many parts of the highlands were considered too cold to support transmission. In this report, we examined how transmission of Plasmodium falciparum in six villages changed along an altitude transect in the Usambara Mountains, Tanzania, from 300 m to 1700 m. Routine entomological collections were made using spray catches and light traps for 15 mo. Direct estimates of entomological inoculation rates and indirect estimates of vectorial capacity suggested a >1000-fold reduction in transmission intensity between the holoendemic lowland and the hypoendemic highland plateau. Lowland transmission was perennial with a significant peak in the cool season after the long rains in May, when vectors densities were high. In the highlands, low temperatures prevented parasite development in mosquitoes during the cool season rains, and highland transmission was therefore limited to the warm dry season when vector densities were low. The primary effect of increasing altitude was a log-linear reduction in vector abundance and, to a lesser extent, a reduction in the proportion of infective mosquitoes. Highland malaria transmission was maintained at extraordinarily low vector densities. We discuss herein the implications of these findings for modeling malaria and suggest that process-based models of malaria transmission risk should be improved by considering the direct effect of temperature on vector densities. Our findings suggest that variation in the short rains in November and changes in agricultural practices are likely to be important generators of epidemics in the Usambaras.

Full text

Vector
/Pathogen
/Host
Interaction,
Transmission
Relationship
Between
Altitude
and
Intensity
of
Malaria
Transmission
in
the
Usambara
Mountains,
Tanzania
R.
B0DKER,12
J.
AKIDA,3
D.
SHAYO,1
W.
KISINZA,3
H.
A.
MSANGENI,3
E.
M.
PEDERSEN,1
and
S.
W.
LINDSAY1-"
J.
Med.
Entomol.
40(5):
706-717
(2003)
ABSTRACT
There
is
a
consensus
that
malaria
is
a
growing
problem
in
African
highlands.
This
is
surprising
because
many
parts
of
the
highlands
were
considered
too
cold
to
support
transmission.
In
this
report,
we
examined
how
transmission
of
Plasmodiumfalciparum
in
six
villages
changed
along
an
altitude
transect
in
the
Usambara
Mountains,
Tanzania,
from
300
m
to
1700
m.
Routine
entomological
collections
were
made
using
spray
catches
and
light
traps
for
15
mo.
Direct
estimates
of
entomological
inoculation
rates
and
indirect
estimates
of
vectorial
capacity
suggested
a
>
1000-fold
reduction
in
transmission
intensity
between
the
holoendemic
lowland
and
the
hypoendemic
highland
plateau.
Lowland
transmission
was
perennial
with
a
significant
peak
in
the
cool
season
after
the
long
rains
in
May,
when
vectors
densities
were
high.
In
the
highlands,
low
temperatures
prevented
parasite
development
in
mosquitoes
during
the
cool
season
rains,
and
highland
transmission
was
therefore
limited
to
the
warm
dry
season
when
vector
densities
were
low.
The
primary
effect
of
increasing
altitude
was
a
log-linear
reduction
in
vector
abundance
and,
to
a
lesser
extent,
a
reduction
in
the
proportion
of
infective
mosquitoes.
Highland
malaria
transmission
was
maintained
at
extraordinarily
low
vector
densities.
We
discuss
herein the
implications
of
these
findings
for
modeling
malaria
and
suggest
that
process-based
models
of
malaria
transmission
risk
should
be
improved
by
considering
the
direct
effect
of
temperature
on
vector
densities.
Our
findings
suggest
that
variation
in
the
short
rains
in
November
and
changes
in
agricultural
practices
are
likely
to
be
important
generators
of
epidemics
in
the
Usambaras.
KEY
WORDS
Anopheles
spp.,
highland
malaria,
malaria
transmission,
Tanzania,
vectorial
capacity
Malaria
in
the
African
highlands
appears
to
be
a
vectors,
their
feeding
frequency,
and
the
development
growing
problem,
with
areas
previously
considered
of
the
parasite
in
the
mosquito
are
summarized.
It
is
free
of
malaria
now
becoming
endemic
or
affected
by
well
known
that
vectors
are
relatively
scarce
in
the
severe
epidemics
(Lindsay
and
Martens
1998,
African
highlands,
but
mosquito
densities
have
been
Mouchet
et
al.
1998,
B0dker
et
al.
2000).
But
little
is
difficult
to
model
in
relation
to
environmental
param-
known
about
the
dynamics
of
malaria
transmission
in
eters
and
have
played
a
minor
role
in
the
construction
the
these
highlands
(Garnham
1948,
Heisch
and
of
transmission
models.
Most
process-based
models
of
Harper
1949,
Clyde
et
al.
1956,
Matola
et
al.
1987,
malaria
distribution,
therefore,
assume
that
the
pri-
Fontenille
et
al.
1990,
Khaemba
et
al.
1994,
Abose
et
al.
mary
impact
of
low
temperatures
on
malaria
trans-
1998,
Ellman
et
al.
1998,
Lindblade
et
al.
1999,
Lindsay
mission
is
exerted
by
prolonging
the
duration
of
par-
et
al.
2000).
asite
development,
thus
reducing
the
proportion
of
To
determine
where
highland
malaria
transmission
mosquitoes
surviving
long
enough
to
develop
infective
occurs
and
where
it
is
spreading,
we
need
to
have
an
parasites.
When
the
temperature
falls
below
the
min-
accurate
understanding
of
the
interactions
between
imum
for
parasite
development,
transmission
ceases
the
highland
environment
and
transmission.
Attempts
despite
the
presence
of
malaria
vectors,
to
model
malaria
transmission
have
been
based
largely
Scientists
have
attempted
to
map
the
risk
of
malaria
on
process-based
models.
In
this
report,
the
effect
of
transmission
across
large
geographical
areas
by
such
temperature,
and
less
often,
rainfall,
on
the
survival
of
process-based
models
(Sutherst
1993,
Martin
and
Lefebvre
1995,
Jetten
et
al.
1996,
Lindsay
and
Birley
'Danish
Bitharziasis
Laboratory,
Jaegersborg
AllelD,
DK-2920
1996;
Unds^
m*
Martens
1998,
Martens
1998,
Craig
Charlottenlund,
Denmark.
e*
*"•
1999)
and
predict
effects
of
climatic
change
2
Corresponding
author:
Danish
Veterinary
Laboratory,
BQlowsvej
(Martens
et
al.
1994,
Lindsay
and
Birley
1996).
At-
27
DK-1790
Copenhagen
v
Denmark.
tempts
have
been
made
to
model
highland
transmis-
T^r
°
^
"^
siontopredictthedistributionandintensityofmalaria
*
School
of
Biological
and
Biomedical
Sciences,
University
of
transmission,
to
forecast
epidemics,
and
to
predict
the
Durham,
South
Road,
Durham
DH1
3LE,
United
Kingdom.
impact
of
climatic
change
(Lindsay
and
Martens
1998,
0022-2585/03/0706-0717$04.00/0
©
2003
Entomological
Society
of
America

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
707
2250m
2000m
1750m
1500m
1250m
1000m
750m
u
500m
J
10km
m
Fig.
1.
Map
of
the
south-eastern
part
of
the
West
Usambara
Mountains,
showing
the
position
of
the
six
transect
vil-
Sharp
and
Cox
1998,
Rogers
and
Randolph
2000).
Temperature
decreases
with
altitude
and
models
fo
cusing
on
the
impact
of
temperature
on
parasite
de
velopment
time
in
the
vectors,
assume
a
highland
zone
with
malaria
vectors,
but
without
malaria
transmis
sion.
Such
a
zone
would
be
prone
to
epidemics
if
temperatures
temporarily
increase
and
permit
sporo-
gony
in
the
existing
vector
population.
This
study
aimed
to
provide
a
detailed
description
of
the
effects
of
altitude
and
the
seasonality
of
tem
perature
and
rainfall
on
the
transmission
intensity
of
malaria
along
an
altitudinal
transect
in
the
West
Usambaras,
Tanzania,
from
the
holoendemic
foothills
at
300
m
to
the
hypoendemic
highland
plateau
at
1700
m.
The
specific
objectives
of
the
study
were
to:
(1)
quantify
the
impact
of
altitude
on
sporozoite
rates
and
vector
densities,
and
to
compare
the
relative
con
tribution
of
these
two
factors
on
transmission
intensity
by
direct
measurements
of
the
entomological
inocu
lation
rates
(EIR)
and
by
indirect
estimates
of
vecto-
rial
capacity
(Garrett-Jones
1964);
(2)
determine
whether
a
zone
of
anophelism
without
malaria
existed;
(3)
identify
potential
mechanisms
of
highland
malaria
outbreaks;
and
(4)
suggest
potential
impacts
of
envi
ronmental
and
climatic
change
in
the
Usambara
high
lands.
Materials
and
Methods
Study
Area.
Six
villages
were
selected
along
an
al
titude
transect
in
the
West
Usambara
Mountains
in
northeast
Tanzania,
«=100
km
inland
from
the
port
of
Tanga.
These
villages
were
situated
at
300
m
(Kwameta,
5°
06.6'
S,
38°
29.1'
E),
600
m
(Magundi
5°
05.3'
S,
38°
28.3'
E),
800
m
(Kwamhanya,
5°
03.5'
S,
38°
27.6'
E),
1,000
m
(Bagamoyo,
5°
04.0'
S,
38°
26.5'
E),
1,400
m
(Balangai,
4°
55.6'
S,
38°
27,7'
E),
and
1,700
m
(Milungui,
4°
45.3'
S,
38°
21.3'
E;
Fig.
1).
All
villages
grew
maize.
Rice
was
only
fanned
at
300
m,
beans
were
grown
up
to
1,400
m,
and
tea
was
cultivated
between
1,000
and
1,400
m.
The
highest
village
culti
vated
vegetables
supplemented
with
fruit
trees.
The
study
area
and
the
villages
have
been
described
pre
viously
in
greater
detail
(B0dker
et
al.
2000,
Bendixen
etal.
2001).

708
Journal
of
Medical
Entomology
Vol.
40,
no.
5
Meteorological
Recordings.
In
each
village
rainfall
was
measured
using
a
simple
rain
gauge
(model
CE/004,
Casella,
United
Kingdom)
and
outdoor
tem
peratures
recorded
at
1.25
m
in
a
Stevenson
screen
and
indoors
at
1.4
m
using
temperature
data
loggers
(Tinytalk-Temp,
Chichester,
West
Sussex,
United
Kingdom)
recording
20
times
per
24
h.
Mosquito
Collections.
A
geographically
defined
cluster
of
houses
containing
up
to
1,000
people
was
selected
in
each
village.
All
houses
in
each
cluster
were
numbered,
and
eight
houses
were
randomly
se
lected
for
mosquito
collections.
For
light
trap
catches
of
mosquitoes,
a
bedroom
was
selected
in
each
house
and
untreated
bednets
given
to
the
occupants.
For
four of
the
light-trap
houses
in
each
village,
a
bedroom
in
a
neighboring
house
was
selected
for
spray
catches.
The
number
of
people
sleeping
in
each
of the
12
collection
rooms
was
calculated
as
the
average
num
ber
of
people
recorded
sleeping
in
the
rooms
at
the
beginning
and
at
the
end
of the
study
period.
Female
mosquitoes
were
collected
weekly
in
each
village
from
October
1995
to
December
1996.
For
light-trap
catches, a
Centers
for
Disease
Control
(CDC)
light
trap
with
a
piece
of
glucose
soaked
cotton
wool was
suspended
with
the
trap
entrance
^LSm
from
the
floor
next
to
a
bed.
The
trap
was
operated
from
sunset
(==19.00
h)
until
sunrise
the
following
morning
(^06.00
h).
The
same
morning
the
floors
in
the
four
spray
rooms
were
covered
with
white
sheets
and
the
rooms
were
sprayed
with
a
pyrethroid
(2.5%
volume
Pycon
diluted
in
water).
Mosquitoes
were
collected
in
paper
cups
and
brought
to
the
laboratory
in
Muheza.
Mosquito
Processing.
Anopheline
mosquitoes
were
identified
according
to
Gillies
and
Coetzee
(1987).
A
sample
of
suspected
malaria
vectors
caught
in
light
traps
were
dried
and
the
head and
thorax
of
each
mosquito
later
tested
for
Plasmodium
falciparum
sporozoite
antigen
in
an
enzyme-linked
immunosor-
bent
assay
(ELJSA)
adapted
from
Wirtz
(1994).
Blood
meals
from
mosquitoes
caught
by
spray
catch
were
squashed
onto
filter
paper
(Whatman
no.
1).
Smears
were
eluted
in
0.5
ml
coating
buffer
and
tested
in
a
human
IgG
ELISA
adapted
from
B0gh
et
al.
(1998).
Data
Analysis.
Mean
daily
temperature
was
calcu
lated
as
the
mean
of
20
daily
recordings.
The
minimum
and
maximum
monthly
and
yearly
temperatures
were
calculated
as
means
of
the
minimum
and
maximum
value
of
daily
recordings.
Monthly
arithmetic
means
of
the
number
of
mos
quitoes
caught
were
calculated
in
each
of
the
light-
trap
houses.
Missing
catches
were
substituted
with
the
arithmetic
mean
of
the
number
of
mosquitoes
from
the
two
previous
and
the
two
following
weeks
for
that
particular
house,
and
the
substituted
data
used
in
the
calculation
of
monthly
means.
Evaluating
this
method
on
uninterrupted
data
series
by
removing
data
points
showed
this
method
to
give
a
more
realistic
estimate
of the
true
monthly
mean
than simple
omission of
data.
Monthly
vector
densities
in
each
village
were
calcu
lated
as
geometric
means
(GM)
logiO(n+l)
of
the
monthly
averages
of
the
eight
light
traps
in
each
vil
lage.
Monthly
GM
was
used
for
calculations
of
monthly
man
biting
rates
and
inoculation
rates,
and
annual
GM
was
used
for
annual
man
biting
rates.
Numbers
of
mosquitoes
in
each
room
were
divided
by
the
mean
of
the
number
of
people
sleeping
in
the
room,
to
give
the
number
of
host-seeking
mosquitoes
per
person,
the
man
biting
rate
(MBR),
assuming
that
the
light-trap
efficiency
relative
to
human
bait
catch
was
1:1.
Previous
studies
in
Tanzania
had
found
the
light-trap
efficiency
to
be
0.71
and
1.23
(Lines
et
al.
1991a,
Davis
et
al.
1995).
Differences
in
sporozoite
rates
and
parity
rates
between
seasons
and
between
villages
were
analyzed
by
•£
analysis
and
X1
for
trend,
respectively,
using
Epi
Info
6.02.
Remaining
calculations
were
done
using
SPSS
(1999)
version
9.0.0.
The
EIR
rate
was
calculated
as
the
product
of
the
MBR
and
the
sporozoite
rate.
In
villages
in
which
sporozoite
rates
differed
significantly
between
sea
sons,
EIR
was
calculated
by
season,
and
in
villages in
which
no
seasonal
differences
were
detected,
the
EIR
was
calculated
using
the
annual
mean
sporozoite
rate.
Duration
of
the
gonotrophic
cycle
was
calculated
as:
=
36.5°C/(T-9.9°C)
[1]
as
suggested
for
Anopheles
macidipennis
Meigen
(Diptera:
Culicidae)
by
Detinova
(1962),
where
Tis
the
mean
temperature
(°C)
assuming
a
relative
hu
midity
of
60-90%.
Indoor
temperature
was
calculated
as
the
mean
outdoor
temperature
+2.6°C, the
mean
difference
between
indoor
and
outdoor
temperatures.
The
man
biting
habit
a,
the
frequency
by
which
a
vector
will
feed
on man,
was
calculated
as
the
recip
rocal
of
the
duration
of
the
gonotrophic
cycle,
assum
ing
the
frequency
of
human
blood meals
(HBI)
to
be
100%
for
both
Anopheles
gambiae
Giles
sensu
lato
(Diptera:
Culicidae)
and
Anopheles
fimestus
Giles
(Diptera:
Culicidae)
in
all
villages.
Parasite
development
time
n was
calculated
as:
n(days)=lll/(T-fmin)
[2]
as
suggested
by
Detinova
(1962),
where
T,
the
mean
temperature,
is
the
mean
outdoor
temperature
(°C)
+
2.6°C,
and
tmin,
the
minimum
temperature
for
parasite
development,
is
16°C
(Detinova
1962)
orl9°C
(Molineaux
1988).
The
vectorial
capacity
C,
the
potential
number
of
inoculations
with
P.
falciparum
arising
from
one
in
fective
human
per
day,
assuming
that
all
vectors
feed
ing
on
humans
become
infective
(Garrett-Jones
1964),
was
calculated
monthly
for
An.
gambiae
s.
1.
and
An.
funestus
combined
as:
C
=
ma
*
a
*p"*l/
-In p
[3]
where
m
is
the
vector
abundance
relative
to
man
and
a
and
n
are
as
described
above.
The
values
of
the
daily
survival
rates
p
were
purposely
selected
at
fixed rates
of
0.8
or
0.9.
The
basic
reproductive
rate
Rq,
the
total
number
of
new
infections
originating
from
an
infective
human
within
the
entire
duration
of
infectivity,
was
calcu
lated
as:

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
709
R0
=
C*
365
days
[4|
assuming
the
average
duration
of
human
infectivity
to
be
1
yr,
as
suggested
by
Martens
et
al.
(1994)
for
children
under
5
yr
of
age.
Although
the
number
of
mosquitoes
becoming
infected
by
feeding
on
an
in
fective
human
host
has
been
suggested
to
be
between
60%
and
83%
in
Kenya
(Githeko
et
al.
1992),
it
is
here
assumed
to
be
100%.
Ethics.
Ethical
approval
for
this
study
was
given
by
the
Medical
Coordinating
Committee
of
the
National
Institute
for
Medical
Research
in
Tanzania
and
reg
istered
at
the
Tanzania
Commission
for
Science
and
Technology
as
No
95-217-CC,
96-264-ER-95-121,
and
98-005-ER-95-21.
Results
Climate.
The
rainfall
pattern
showed
two
clearly
defined
peaks
in
1996:
a
large
peak
in
April-May,
the
long
rains,
and
a
much
smaller
peak
in
October-
November,
the
short
rains
(Fig.
2).
All
villages
re
ceived
between
1,000
and
1,300
mm
of
rainfall
in
1996,
except
for
the
village
at
1,000
m,
which
had
1,920
mm
of
rainfall
(Table
1).
There
was
a
seasonal
variation
in
weather
in
all
villages,
with
a
warm
season
peaking
in
January
with
mean
temperatures
5.3-7.1°C
higher
than
the
coldest
month,
July
(Fig. 2).
There
was
a
linear
decrease
in
mean
atmospheric
temperature
with
increasing
alti
tude
of
0.7°C
per
100
m
in
the
warm
season
(30.1-
0.0068
*
altitude;
r*
=
0.97,
P
<
0.001)
and
0.6°C
per
100
m
in
the
cold
season
(23.5-0.0063
*
altitude;
^
=
0.99,
P<
0.001).
The
annual
mean
indoor
temperatures
were
on
av
erage
2.6°C
higher
than
outdoor
temperatures
in
the
six
villages.
The
difference
was
largely
because
of
an
increase
in
annual
mean
minimum
temperatures,
which
were
on
average
3.9°C
higher
indoors
than
outdoors
(Table
1).
Man
Biting
Rate.
A
total
of
2,642
light-trap
catches
collected
18,227
An.
gambiae
s.l.
(93%
of
the
catch)
and
1,445
(7%)
resembling
An.
Junestus.
Anopheles
funestus
from
light
traps
were
difficult
to
distinguish
from
similar
species
because
many
were
damaged
by
the
fan
in
the
light
traps.
Of
the
1,302
An.
funestus,
94
were
identified
as
Anopheles
demeilloni
Evans
(Diptera:
Culicidae),
36
were
identified as
Anopheles
letabensis
Lambert
&
Coetzee
(Diptera:
Culicidae),
12
were
identified
as
Anopheles
rivulorum
Leeson
(Diptera:
Culicidae),
and
1
was
identified
as
Anopheles
parensis
Gillies
(Diptera:
Culicidae).
Anopheles
demeilloni
was
rare
in
the
three
lowest
vil
lages,
but
was
relatively
common
in
the
village
at
1,000
m,
constituting
23%
of
the
An.
funestus
complex
mos
quitoes.
At
1,400
m,
67%
of
the
An.
funestus
complex
mosquitoes
appeared
to
be
An.
deineilloni.
At
1,700
m,
five
An.
detneilloni
and
a
single
An.
parensis
appeared
to
constitute
the
six
An.
funestus
complex
mosquitoes
collected. In
the
analysis,
these
mosquitoes
were
con
sidered
as
An.
funestus
complex
when
collected
in
light
traps.
Geometric
mean
densities
for
An.
gambiae
and
An.
Junestus
complex
peaked
at
the
end
of
the
long
rains,
although
at
1,400
m
there
was
an
additional
peak
in
the
warm
season
(Fig.
2).
Moreover,
An.
Junestus
complex
in
the
village
at
1,000
m
remained
elevated
for
most
of
the
study
(Fig. 2).
There
was
a
strong
decrease
in
annual
MBR
with
increasing
altitude
(Table
2).
Linear
regression
of
log10
annual
MBR
for
1996
suggests
that
the
MBR
dropped
by
50%
for
every
126-m
increase
in
altitude
(Fig.
3;
Iogi0(annual
MBR)
=
4.339
-
0.0024*aItitude,
r*=
1.00;
P<
0.001).
Man
Biting
Habit.
Blood
meals
from
594
An.
gam
biae
s.l.
and
71
An.
funestus
showed
that
nearly
all
mosquitoes
fed
on
people
(Table
2).
Some
of the
negative
blood
meals
may
have
been
of
human
origin,
but
in
a
state
too
digested
to
give
a
positive
identifi
cation,
so
that
the
true
HBI
from
An.
gambiae
s.l.
may
be
very
close
to
100%.
Low
numbers
of An.
Junestus
were
analyzed,
but
similar
high
preferences
for
hu
mans
were
indicated
in
the
three
lowest
villages.
Because
most
blood
meals
were
of
human
origin,
the
man
biting
habit
a
is
simply
the
reciprocal
of
the
duration
of
the
gonotrophic
cycle.
Too
few mosqui
toes
were
caught
in
spray
catches
to
determine
the
duration
of
the
gonotrophic
cycle
directly
by
the
ratio
of
abdominal
stages.
Duration
of
gonotrophic
cycle,
therefore,
had
to
be
estimated
from
known
relation
ships
between
temperature
and
duration
of
egg
de
velopment
in
An.
maculipennis
(Detinova
1962).
Be
cause
the
egg
laying
and
blood
feeding
behavior
is
adjusted
to
a
nightly
activity
pattern,
the
duration
in
the
lowland
village
is
effectively
at
least
2
d
in
the
warmest
month
and
3
d
in
the
coldest
(Table
2).
Sporozoite
Rate.
A
total
of
5,379
An.
gambiae
and
An.
Junestus
complex
mosquitoes
were
examined
for
P.
falciparum
sporozoites
(Table
2).
Significant
sea
sonal
variation
in
sporozoite
rates
was
found
in
An.
gambiae
s.l.
at
300
m
(y2
=
28.5,
df
=
7,
P
<
0.001)
and
at
800
m
(/
=
11.05,
df
=
4,
P
=
0.03),
peaking
August-September,
but
not
at
600
m
(x2
=
5.36,
df
=
4,
P
=
0.25)
(Fig.
4).
No
significant
seasonal
variation
in
sporozoite
rates
was
found
with
the
An.
funestm
complex
in
any
village.
To
minimize
an
impact
of
seasonal
variation,
dif
ferences
in
the
sporozoite
rate
of
An.
gambiae
s.l.
between
villages
were
analyzed with
data
from
a3-mo
period,
around
the
peak
mosquito
density
(April to
June).
Although
sporozoite
positive
An.
gambiae
s.l.
were
found
only
in
the
three
lowest
villages,
there
was
a
significant
decline
in
sporozoite
rates
with
increasing
altitude
(x2
of
slope
=
8.59,
df
=
1,
P
=
0.003).
Com
pared
with the
lowland
village
the
odds
ratios
at
600
m
and
800
m
were
0.71
(95%
CI
=
0.37,
1.35)
and
0.22
(95%
CI
=
0.07,
0.73),
respectively.
Sporozoite
posi
tive
An.
Jkinestus
complex
were
only
found
in
the
villages
at
300
m
and
1,000
m,
and
no
significant
dif
ference
in
sporozoite
rate
was
detected
between
the
two
villages
(/
=
0.26,
df
=
1,
P
=
0.61).
Duration
of
Sporogony.
There
was
a
great
variation
in
the
estimated
sporozoite
development
time
along
the
transect
from
less
than
8
d
to
no
development
at

710
Journal
of
Medical
Entomology
Vol.
40,
no.
5
3
I
*
i
t
3
0.014
-i
0.012
t
70
[■^.f
v^i
rainfall
(cm)
-An.
gambiae
-
-
a-
-
An.
funestus
-temp
max
-
temp
min
Fig.
2.
Monthly
minimum
and
maximum
temperatures
(°C)
and
rainfall
(cm)
(right
axis),
and
monthly
CM
man
biting
rates
(left
axis,
note
that
the
scale
varies
between
villages)
in
the
six
villages.
all.
Because
of
a
disproportionately
large
impact
of
temperatures
near
the
tmin
(16-19°C)
the
duration
of
sporogony
is
greatly
extended
above
800
m
in
the
cold
season
(Fig.
5).
Entomological
Inoculation
Rate.
Because
sporozo-
ite-positive
An.
gambiae
s.l.
were
only
found
in
the
three
lowest
villages,
the
EIR
for
An.
gambiae
s.l.
in
the
three
highest
villages
was
estimated
using
the
average
sporozoite
rate
from
the
nearest
village
(800
m),
thus
ignoring
the
decreasing
trend
in
sporozoite
rate
with
increasing
altitude.
Likewise,
because
positive
An,
funestm
complex
were
only
found
at
300
m
and
1,000
m,
the
EIR
in
the
villages
at
600
m
and
at
800
m
were
estimated
using
the
mean
sporozoite
rates
in
the
villages
at
300
m
and
1,000
m,
and
the
EIR
for
the
An.
funestus
complex
at
1400
m
and
1,700
m
were
estimated
using
the
average
sporozoite
rate
from
the
nearest
village
(1,000
m)
(Table
2).
The
seemingly
high
contribution
of
An.
fitnestus
complex
to
the
EIR
in
the
village
at
1,700
m
may
be

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
711
Table
1.
Meteorological
data
for
the
six
villages
(1996)
Indoor
temperatures
are
given
as
deviations
from
outdoor
temperatures.
exaggerated
because
they
were
almost
all
An.
demei-
lloni
and
none
were
positive
for
malaria
parasites
at
these
altitudes.
The
EIR
at
1,700
m
is
thus
the
product
of
the density
of
An.
demeilloni
in
the
village
and
the
sporozoite
rate
of
the
An.
fimestus
complex
in
the
village
at
1,000
m,
which
mainly
consisted
of
An.
junestus.
There
was
a
highly
significant
decline
in
EIR
with
increasing
altitude
(EIR
log10
=
2.523-0,0025
*
alti
tude,
r2-
=
0.97,
P
<
0.001)
(Fig.
6).
Vectorial
Capacity.
Annual
mean
vectorial
capacity
was
estimated
from
monthly
calculations
of
the
actu
ally
measured
MBR
and
indoor temperatures
in
the
transect
villages.
The
duration
of the
gonotrophic
cy
cle
and
the
duration
of
sporozoite
development
were
estimated
from
relationships
with temperature
previ
ously
published
(Detinova
1962).
The
daily
survival
rates
were
selected
at
0.8
and
0.9
based
on
extreme
values
previously
estimated
for
the
Usambaras
(Lines
et
al.
1991b)
and
indicated
from
the
parity
rates
(B0dker
2000)
and
from
a
suggested
relationship
be
tween
temperature
and
survival
(Martens
1998).
The
vector
density
relative
to
man
m
(not
the
man
biting
rate
ma)
and
the
expression
a2
*
p"l
—
In
p,
the
product
of
which
gives
the
vectorial
capacity,
were
calculated
separately.
Assuming
a
constant
HBI
of
100%
and
fixed
values
of
p
at
all
altitudes,
the
expression
a2
*
p"l
—In
p
is
solely
dependent
on
the
impact
of
temperature
on
sporozoite
and
egg
development
times.
From
four
combinations
of
the
two
selected
daily
survival
rates
and
two
minimum
temperatures
for
parasite
develop
ment
of
16°C
or
19°C,
suggested
by
previous
authors
(Detinova
1962,
Molineaux
1988),
the
overall
reduc
tion
in
vector
density
contributed
more
to
the
reduc
tion
in
vectorial
capacity
than
did
the
combined
the
oretical
impact
of
mean
temperatures
on
the
duration
of
the
gonotrophic
cycle
and
the
sporozoite
develop
ment
(Fig.
7).
Only
after
1,400
m
did
the
rate
of
change
of
the
expression
o2*p"/-hp
clearly
exceed
the
rate
of
change
in
the
relative
vector
density.
Basic
Reproductive
Rate.
The
basic
reproductive
rate
Ro,
the
total
number
of
new
infections
arising
from
one
infective
person,
was
estimated
from
the
mean
of
the
four
estimated
values
of
vectorial
capacity
in
each
village.
We
assumed
the
duration
of
infectivity
to
be
1
yr
(Martens
et
al.
1994)
in
all
six
villages,
thus
disregarding
the
potential
impact
of
different
levels
of
immunity
in
the
villages
on
the
duration
of
infectivity.
The
mean
Rq
decreased
from
1,841
(ranging
from
168
to
4,257
in
the
four
separate
calculations)
new
infec
tive
bites
arising
per
infection
at
300
m
to
just
sus
tainable
transmission
at
1.1
(ranging
from
0.02
to
3.1)
new
infective
bites
at
1,400
m.
The
mean
Rq
at
1,700
m
was
below
1,
indicating
that
the
intensity
of
transmis
sion
was
insufficient
to
sustain
malaria,
with
just
0.005
(ranging
from
0.000
to
0.019)
new
infective
bites
being
generated
per
infection
(Fig. 7).
Discussion
We
measured
malaria
transmission
along
an
altitude
transect
from
holoendemic
malaria
to
hypoendemic
malaria
(Bendixen
et
al.
2001)
in
the
Usambara
Moun
tains.
It
became
progressively
colder
at
higher
alti
tudes,
with
a
difference
of
9°C
between
the
lowest
village
on
the
warm
plains
at
300
m
and
the
coldest
village
at
1,700
m.
These
temperatures
were
similar
to
those
reported
in
the
1930s
along
a
southwest
facing
transect
(Moreau
1934),
and
are
much
cooler
than
east
African
highlands
further
inland.
For
example,
mean
temperatures
in
the
Usambaras
at
900
m
are
similar
to
those
at
1400-1600
m
in
some
east
African
highlands
(Kenworthy
1966).
Such
climatic
differ
ences
make
it
impossible
to
directly
relate
our
findings
to
other
highlands
solely
by
altitude.
The
long
rains
(April
to
May)
marked
the
beginning
of
the
cool
season
(May
to
October),
and
thus
ef
fectively
separated
the
year
in
a
dry
warm
period
(December
to
March)
and
a
wet
cool
period.
The
short
rains in
November
are
generally
very
unreliable
(Lundgren
and
Lundgren
1979)
and
failed
in
the
study
period
in
1995
and
partly
in
1996,
resulting
in
an
almost
unimodal
rainfall
pattern.
Anopheles
gambiae
s.l.
was
the
dominant
vector
in
the
study
area
and
its
seasonal
abundance
was
strongly
related
to
rainfall,
with
greatest
numbers
biting
in
May
to
June,
approximately
lmo
into
the
long
rains
in
the
beginning
of
the
cold
season.
Anopheles
funestus
com
plex
contributed
relatively
little
to
the
MBR
in
the
lower
villages
and,
in
the
two
highest
villages,
this
group
was
dominated
by
An.
demeilloni,
which
is
prob
ably
not
a
malaria
vector
(Gillies
and
De
Meillon
1968).
However,
An.
funestus
complex
appeared
to
be
more
important
at
1,000
m.
Here,
the
high
rainfall
probably
provided
perennial
breeding
in
the
perma
nent
pools
formed
by
a
stream
running
through
the
middle
of
the
village.

h
to
Table
2.
Entomological
data
for
the
study
villages
(1996)
Average
biting
rate
per
man
per
night
are
given
as
geometric
means
(GM)
with
95%
confidence
intervals
of
annual
catches
in
eight
light
traps
per
village
(Kwamcta,
Magundi,
Kwamhanya,
Bagomoyo,
Balangai,
and
Milungui,
respectively).
The
human
blood
index
(HBI)
was
based
on
n
bloodmeals
from
spray catch
collected
mosquitoes.
Annual
PlasmodiumfaltHjutrum
sporozoitc
rates
are
here
adjusted
for
seasonal
variation
in
vector
densities,
by
weighting
monthly
sporozoite
rates
by
the
monthly
man
biting
rates.
The
estimated
numbers
of
infective
bites
received
per
man
per
year
(ELK)
were
based
on
monthly
GM
man
biting
rates.
£
3
» 3
g
la.®
3 $
?
3-3 S-3
*<
«*£ S-Z
S3
—
J
P
I3-(S
«i
—
Q
5-<5>
to
n.
—
S
C
2-w?
^*
rTi
r^
-^
^
**
•
III.
Sih
§511
S
p
O-
CD
§.
s-1-
i"
8
*
a.;-!
<
2
?
3
W
<
^.
WEHIi
:
8
q-1
3"*§
«
BrS.
Annual
MBR
(log
scale)
000
£
3
O

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
713
—An.
gambiae
300
m—■—An.
gamblaa
600
m
-An.
funestus
300
m
- -
X-
-
-An.
funestus
1000
m
-An.
gam
bias
800
m
Fig.
4.
Seasonal
variation
in
sporozoite
rate
(%)
in
the
four
villages
up
to
1,000
m.
Sporozoite
rates
in
the
three
lowest
villages
increased
after
the
main
rainy
season
in
April-May
and
peaked
in
the
following
cool
season.
sities
probably
reflected
the
decreasing
temperature
along
the
transect.
It
is
possible
that
temperature
sim
ply
governs
vector
densities
through
a
direct
physio
logical
effect
on
larval
development
time.
Alterna
tively,
it
may
be
that
temperature
has
an
indirect
effect
on
anopheline
ecology.
We
frequently
observed
pud
dles
packed
with
culicine
larvae
in
the
highlands,
and
it
is
possible
that
An.
gambiae
cannot
compete
with
the
culicine
larvae
in
the
cool
water
in
the
highlands,
rather
than
it
is
temperature per
se
that
limits
the
breeding.
That
complex
ecological
interactions
may
be
involved
in
the
Usambaras
is
supported
by
the
finding
of
high
densities
of An.
arabiensis
Patton
(Diptera:
Culicidae)
in
Ethiopian
highlands
at
similar
temperatures
(Abose
et
al.
1998).
The
indoor
vector
densities
in
the
highland
transect
villages
were
con
siderably
lower
than
those
reported
from
other
East
African
highlands
at
similar
altitudes,
Uganda
(Garnham
et
al.
1948,
Lindblade
et
al.
2000),
Kenya
(Garnham
1948,
Heisch
and
Harper
1949,
Roberts
700
1200
1700
Altitude
(m)
--O---Jan.
Tmln=16
—O—JulyTmln=16
-
-
x-
-
-
Jan.
Tmin=19
—x—
July
Tmln=19
Fig.
5.
The
duration
of
sporogony
was
estimated
for
the
peak
warm
and
the
peak
cold
season
(January
and
July,
respectively)
for
the
six
villages.
Two
estimates
of
duration
were
calculated
for
each
season
assuming
the
minimum
tem
perature
for
parasite
development
to
be
either
16°C
or
19°C,
respectively.
1964,
White
1972),
and
Ethiopia
(Fontaine
et
al.
1961,
Abose
et
al.
1998).
The
main
cause
of
the
reduced
vector
densities
is
almost
certainly
the
much
cooler
climate
in
the
Usambara
highlands
compared
with
other
East
African
highlands.
Even
when
the
lower
temperature
are
taken
into
account,
the vector
den
sities
appeared
to
be
lower
in
the
Usambaras
than
elsewhere.
Almost
all
bloodfed
An.
gambiae
s.l.
and
An.
funestus
caught
had
fed
on
people,
confirming
the
very
anthropophilic
feeding
behavior of
these
vectors
in
East Africa
(White
1971,
White
et
al.
1972).
Both
An.
funestus
and
An.
gambiae
s.l.
are
likely
to
be
as
endophilic
in
the
highlands
as
they
are
in
the
lowlands
(Draper
and
Garnham
1959,
Lines
et
al.
1986).
Therefore,
the
duration
of the
gonotrophic
cycle
and
the
sporogonic
cycle,
will
predominantly
be
determined
by
indoor
temperatures.
Indoor
temper
atures
were
only
recorded
in
a
single
house
in
each
village,
but
are
likely
to
differ
between
houses
de
pending on
the
construction.
Indoor
temperatures
were
on
average
2.6°C
warmer
than
outdoors,
mainly
because
of
an
increase
in
minimum
temperature.
Sim
ilar
increases
have
been
reported
in
the
Kenyan
high
lands
(Haddow
1942,
Garnham
1948).
No
systematic
100
10
0.1
0.01
200
600
1000
1400
Altitude
(m)
1800
Fig.
6.
The
estimated
total
annual
EIR
from
An.
gambiae
s.l.
and
An.
funestus
complex
in
six
villages.
The
EIR
is
pre
sented
on
a
log10
scale.
The
villages
are
connected
with
a
linear
regression
line.

714
Journal
of
Medical
Entomology
Vol.
40,
no.
5
100
T
0.001
0.001
300 600 800
1000
1400 1700
altitude
(m)
—O—(pO.8;16C)
--O--{p0.9;16C)
—X—
(p0.8;
18C)
-
-
-X-
-
■
(p0.9;
1BC)
Fig.
7.
Reductions
in
annual
mean
daily
vectorial
capac
ity.
The
vectorial
capacity
is
presented
by
its
two
main
com
ponents
(left
axis),
the
relative
density
m
and
the
tempera
ture
and
survival
rate
sensitive
component
(a2
*
p"l—
In
p)
the
latter
calculated
in
four
ways.
Combination
of
two
ex
treme
daily
survival
rates
p
(0.8
and
0.9)
and
two
extreme
values
for
sporozoite
development
(16°C
or
19°C)
were
cal
culated
from
monthly
mean
temperatures
giving
the
four
graphs.
The
values
for
the
two main
components
of
vectorial
capacity
are
shown
in
percent
of
the
value
at
300
m.
The
calculation
of
Rq
was
based
on
the
mean
value
of
the
four
estimates
of
vectorial
capacity
assuming
a
human
infectivity
of
1
yr
(right
axis).
relationship
between
increased
temperature
and
alti
tude
was
detected
in
the
selected
houses,
although
the
greatest
increase
was
recorded
at
the
highest
elevation
where
the
impact
of
cold
nights
were
dramatically
reduced
by
6.5°C
higher
indoor
temperatures.
At
this
elevation,
houses
were
generally
more
draft-proof
and
often
with
ceilings
as
an
adaptation
to
the
cool
climate
(29%
had
ceilings
at
1,700
m
compared
with
0-3%
in
the
other
villages).
We
only detected
sporozoites
in
An.
gambiae
s.l.
and
An.
funestus
complex
indicating
that
the
major
high
land
vectors
were
the
same
as
those
in
the
plains.
Sporozoite
rates
in
An.
gambiae
s.l.
declined
from
300
m
to
800
m
(April-June)
suggesting
that
decreas
ing
temperatures
increased
parasite
development
time,
the
first
time
a
systematic
decrease
has
been
documented
along
a
temperature
transect
in
Africa.
However,
we
did
not
detect
a
decrease
in
sporozoite
rates
in
An.
funestus
complex
between
300
m
and
1,000
m.
The
lowland
sporozoite
rates
of
An.
gambiae
s.l.
followed
a
cyclic
trend,
with
a
minimum
during
the
rainy
season,
where
there
was
a
great
influx
of
young
mosquitoes
to
the
population.
The
sporozoite
rate
then
rapidly
built
up
and
peaked
during
the
cool
season
(July
to
September),
contrary
to
what
is
pre
dicted
from
experimentally
derived
relationships
be
tween
decreasing
temperature
and
increasing
sporo
zoite
development
time
(Detinova
1962).
This
is
possibly
because
vector
survival
rates
increased
after
the
rains.
A
similar
pattern
was
observed
with
An.
funestus
complex
mosquitoes
in
the
lowland.
But
at
1,000
m,
sporozoite-positive
An.
funestus
complex
were
found
just
before
and
in
the
beginning
of
the
long
rains.
From
1,000
m
to 1,700
m
the
low
temper
atures
during
the
cool
season
(17-13°C)
would
sub
stantially
prolong
the
duration of
sporogony
(>30
d).
None
of
the
31
vectors
collected
in
the
two
highest
villages
were
infective
during
the
survey
period.
But
with
estimated
sporozoite
development
times
longer
than
1
mo
at
indoor
temperatures
during
July,
suc
cessful
development
of
sporozoites
appears
unlikely
during
the
cold
season
at
1,000
m
and
above.
The
apparent
shift
of
high
sporozoite
rates
from
the
cool
to
warm
season
at
1,000
m
may
indicate
the
beginning
of
such
a
shift
in
the
cyclic
pattern
in
the
middle
of
the
transect,
with
sporogony
being
restricted
to
the
warm
dry
season
at
higher
altitudes,
albeit
the
data
are
too
few
to
be
conclusive.
But
during
the
warm
season
temperatures
were
high
enough
to
allow
sporozoite
development
up
to
1,700
m.
This
is
because
infective
mosquitoes
were
collected
during
the
cold
season
at
800
m
(19.0°C),
at
a
temperature
similar
to
the
warm
est
monthly
temperature
at
1,700
m
(18.8°C),
we
therefore
conclude
that
the
warm
season
up
to
1,700
m
was
warm
enough
to
sustain
sporozoite
development.
The
notion
that
transmission
at
high
altitudes
is
limited
to
the
warm
season,
from
December
to
March,
is
supported
by
evidence
that
highland
epidemics
occur
at
this
time
of
the
year
(unpublished
data).
High
sporozoite
rates
of
5
and
6%
have
been
re
ported
in
An.
funestus
and
An.
gambiae
s.l.
in
the
Nandi
highlands
in
Kenya,
higher
than
the
rates
in
the
much
warmer
Nandi
lowlands
(White
1972).
The
seasonal
variations
in
sporozoite
rates
in
the
Usambaras
also
demonstrated
that
the
relationship
between
temper
ature
and
sporozoite
rates
in
the
field
may
be
more
complicated
than
previously
thought.
Furthermore,
because
the
human
infectivity
is
likely
to
decrease
with
the
decreasing
endemicity
levels
along
an
alti
tude
transect,
we
conclude
that
there
may
not
be
a
simple
relationship
between
altitude/temperature
and
sporozoite
rate.
Reductions
in
EIR
with
altitude
in
the
four
lowest
villages,
where
sporozoite
rates
were
measured
di
rectly,
was
primarily
caused
by
the
rapid
reduction
in
vector
density,
and
only
secondarily,
and
with
a
much
smaller
impact,
by
a
reduction
in
sporozoite
rates.
Sporozoite
rates
in
An.
gambiae
s.l.
decreased
only
4-fold
from
300
m
to
800
m,
whereas
the
density
of
An.
gambiae
s.l.
dropped
>10-fold.
Anopheles
funestus
complex
showed
no
reduction
in
mean
annual
sporo
zoite
rate
from
300
m
to
1,000
m,
whereas
the
density
of
these
mosquitoes
decreased
more
than
five
times.
Sporozoite
rates
were
impossible
to
measure
in
the
two
highest
villages
because
of
the
low
vector
densi
ties.
An
alternative
measure
of
exposure
risk
is
vec
torial
capacity,
which,
unlike
EIR,
has
the
advantage
that
it
can
be
estimated
without
measuring
sporozoite
rates.
Because
precise
values
for
daily
survival
rates

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
715
and
sporozoite
development
time
were unknown,
we
calculated
four
combinations
of
two
extreme
daily
vector
survival
rates
and
two
extreme
values
for
the
minimum
temperature
required
for
parasite
develop
ment.
Despite
vector
density
only
affecting
vectorial
capacity
in
a
linear
manner,
all
four
calculations
in
dicated
that
the
observed
reduction
in
density
with
altitude
contributed
relatively
more
to
the
decrease
in
the
vectorial
capacity,
than
the
combined
impact
of
temperature
on
the
duration
of
sporogony
and
blood
meal
digestion.
The
theoretical
calculations
of
the
vectorial
capacity
thus
supported
the
more
direct
find
ing
from
the
calculation
of
EIR
that
reduced
vector
densities
were
more
important
than
prolonged
para
site
development
times
in
reducing
malaria
transmis
sion
in
the
Usambara
highlands.
Peak
transmission
occurred
in
the
lower
parts
of
the
transect
during
the
relatively
cool
months
of
May
to
July
after
the
long
rains,
when
both
vector
densities
and
sporozoite
rates
were
at
their
highest.
In
contrast,
the
very
low
temperatures
in
the
highlands
during
the
cold
season
must have
confined
sporogony
to
the
warm
dry
months
of
December
to
March
when
vector
densities
were
low.
Therefore,
the
highest
and
lowest
villages
along
the
steep
slopes
of the
Usambaras
may
have
had
almost
opposite
seasonal
transmission
peaks
during
the
survey
period.
We
suggest
that
there
may
be
two
epidemiological
systems
operating
in
the
East
African
highlands.
One
system
in
which
malaria
transmission
is
primarily
re
duced
by
very
low
vector
densities
and
another
system
in
which
transmission
is
reduced
despite
high
vector
densities
because
cool
highland
temperatures
limit
sporogony.
We
did
not
find
indications
of
the
latter
in
the
Usambara
Mountains;
there
was
not
a
general
highland
zone
of
Anophelism
without
malaria.
On
the
contrary,
it
appeared
that
transmission
in
the
high
lands
was
maintained
at
very
low
vector
densities.
Only
during
the
cool
season
at
high
altitudes
is
there
likely
to
be
a
temporal
zone
in
which
low
tempera
tures
prevent
transmission
despite
the
presence
of
vectors.
In
a highland
zone
of
anophelism
without
malaria,
created
by
low
temperatures,
an
outbreak
may
be
initiated
by
an
increase
in
temperature
speeding
up
parasite
development
in
already
existing
large
vector
populations.
Because
temperature
fluctuations
are
re
gional
rather
than
local,
one
might
expect
such
tem
perature driven
epidemics
to
affect
a
substantial
por
tion
of
the
zone
of
anophelism
without
malaria,
and
thus
initiate
large-scale
epidemics.
However,
in
high
land
areas
where
vector
densities
are
extremely
low,
epidemics
maybe
driven
by
increased
vector
breeding
and
may
thus
be
very
local
in
nature,
because
artificial
changes,
topographical
features,
and
even
rainfall
in
tensity
may
vary
greatly
between
East
African
high
land
valleys
(Hamilton
1989).
Indeed,
epidemics
in
the
Usambaras
appear
to
affect
only
a
few
villages
at
a time
(unpublished
Lushoto
Hospital
records,
Lind
say
et
al.
2000).
This
pattern
is
not
consistent
with
the
idea
that
localized
epidemics
are
generated
by
high
temperatures.
We
suggest
that
epidemics
in
highland
Usambaras
may
be
generated
by
an
increase
in
vector
density,
resulting
either
from
high
rainfall
in
the
beginning
of
warm
season
or
from
creation
of
artificial
breeding
sites.
Although
we
cannot
rule
out
that
inter-annual
variations
in
temperature
may
initiate
outbreaks,
we
suggest
that
the
unreliable
short
rains
may
be
a
more
important
generator
of
highland
epidemics
by
allow
ing
vector
breeding
just
before
the
warm
season
in
wet
years
and
preventing
epidemics
in
dry
years,
because
there
is
a
lack
of
vector
breeding
sites.
An
outbreak
in
three
neighboring
highland
villages
in
the
West
Usambaras
was
recently
associated
with
the
construc
tion
of
fish
farms,
where
the
empty
basins
may
have
supported
vector
breeding
(unpublished
Ministry
of
Health
Report,
1997).
It
is
likely
that
the
increase
in
malaria
in
the
Usambara
highlands
over
the
past
50
yr
is
largely
a
result
of
a
rise
in
chloroquine-resistant
strains
of par
asites,
land
use
changes
increasing
mosquito
abun
dance,
although
there
is
weak
evidence
of
a
relation
ship
with
climate
change
(B0dker
et
al.
2000).
A
global
warming
trend
can
potentially
increase
malaria
trans
mission
in
the
Usambaras
and
extend
the
malaria
150
m
higher
up
for
every
1°C
increase.
However,
long-term
changes
in
the
warm
season
rainfall
pattern
may
be
of
greater
significance
in
the
Usambaras.
The
finding
that
highland
transmission
was
limited
by
low
numbers
of
vectors
at
altitudes
where
temper
atures
permitted
sporozoite
development
is
important
for
modeling
malaria
transmission
in
Africa.
Most
ef
forts
so
far
have
focused
on
the
impact
of
temperature
on
parasite
development
time,
and
have
made
limited
attempts
to
model
vector
densities
(Martin
and
Lefebvre
1995,
Jetten
et
al.
1996,
Lindsay
and
Birley
1996,
Lindsay
and
Martens
1998,
Martens
1998,
Craig
et
al.
1999).
To
predict
malaria
distribution
accurately
and
more
importantly
to
predict
transmission
inten
sity,
epidemics
and
potential
impacts
of
environmen
tal
changes,
our
study
strongly
suggest
that
we
need
to
develop
models
that
incorporate
estimates
of vector
densities.
High
sporozoite
rates
up
to
14%
reported
at
very
high
altitudes
in
the
Kenyan
and
Ugandan
high
lands
(White
1972,
Mouchet
et
al.1998,
Lindblade
et
al.
1999)
further
emphasize
the
need
to
rely
less
on
the
sporozoite
development
time
for
predicting
malaria.
Similar
simple
relationships
between
altitude
and
vector
density
may
exist
in
other
East
African
high
lands,
and
it
should
be
a
priority
in
future
research
to
identify
and
understand
these
relationships
to
model
malaria
transmission
intensity
on
a
larger
scale.
Such
altitude-vector
density
algorithms
may
need
to
be
adjusted
to
local
conditions
in
East
Africa
and
it
may
not
be
possible
to
develop
one
simple
model
that
is
valid for
all
African
highlands.
But
the
importance
of
developing
descriptive
and
predictive
models
of
high
land
malaria
transmission
is
underlined
by
the
fact
that
direct
entomological
parameters
like
sporozoite
rates
and
EIR,
for
all
practical
purposes
are
close to
impos
sible
to
measure
routinely
in
the
African
highlands.

716
Journal
of
Medical
Entomology
Vol.
40,
no.
5
Acknowledgments
We
thank
A.
Telaki,,B.
Chambika,
F.
Msuya,
and
J.
Nyon-
gole
at
Ubwari
Field
Station
for
their
hard
work
in
the
field;
.
and
M.
Malimi
and
J.
Kivugo
at
Bombo
Hospital
Field
Station
and
F.
Shenton
for
helping
with
the
ELISAs.
We
are
grateful
.
for
the
patient
collaboration
from
all
the
owners
of
the
houses
where
we
collected
mosquitoes
and
for
the
excellent
work
done
by
the
six
village
assistants.
Reagents
used
in
the
ELISA
procedure
to
test
for
P.
faicipanim
sporozoites
were
kindly
provided
by
Robert
A.
Wirtz,
Centers
for
Disease
Control
and
Prevention,
and
were
produced
by
SmithKline
Beecham
(Pf-positive
controls)
and
New
York
University
(Pf
Mab
cell
line).
These
reagents
were
developed
with
support
from
the
UNDP/World
Bank/WHO
Special
Program
for
Research
and
Training
in
Tropical
Diseases.
The
project
was
funded
by
the
Danish
Council
for
Development
Research
and
the
Danish
Bilharziasis
Laboratory.
References
Cited
Abose,
T.,
Y.
Yeebiyo,
D.
Olana,
D.
Alamirew,
Y.
Beyene,
L.
Regassa,
and
A.
Mengesha.
1998.
Re-orientation
and
:
definition
of
the
role
of
malaria
vector-control
in
Ethio-
. '
pia.
WHO,
WHO/MAL/98.1085.
B0dker,
R.
2000.
Variation
in
malaria
risk
in
the
Usambara
Mountains,
Tanzania.
Ph.D.
thesis,
University
of
Copen
hagen,
Copenhagen,
Denmark.
Bfidker,
R.,
W.
Kisinza,
R.
Mali
ma,
H.
A.
Msangeni,
and
S.
W.
Lindsay.
2000.
Resurgence
of
malaria
in
the
Usambara
Mountains,
Tanzania,
an
epidemic
of
drug-resistant
par
asites.
Global
Change
&
Human
Health
1:
134-153.
B0gh,
G,
E.
M.
Pedersen,
D.
A.
Mukoko,
and
J.
Ouma.
1998.
Permethrin-impregnated
bednets:
effects
on
resting
and
feeding
behaviour
of
lymphatic
filariasis
vector
mosqui
toes
in
Kenya.
Med.
Vet.
Entomol.
12:
52-59.
Bendixen,
M.,
H.
A.
Msangeni,
B.
V.
Pedersen,
D.
Shayo,
and
R.
B0dker.
2001.
Diversity
of
Plastnodium
falciparum
populations
and
complexity
of
infections
in
relation
to
transmission
intensity
and
host
age:
a
study
from
the
Usambara
Mountains,
Tanzania.
Trans.
R.
Soc.
Trop.
Med.
Hyg.
95:
143-148.
Clyde,
D.
F.,
G.
Webbe,
and
G.
T.
Shute.
1956.
Single
dose
pyrimethamine
treatment
of
Africans
during
a
malaria
epidemic
in
Tanganyika.
East
Afr.
Med.
J.
35:
23-29.
Clyde,
D.
F.
1967.
Malaria
in
Tanzania.
Oxford
University
Press,
London,
England.
Craig,
M.
H.,
R.
W.
Snow,
and
D.
Le
Sueur.
1999.
A
climate-
based
distribution
model
of
malaria
transmission
in
sub-
Saharan
Africa.
Parasitol.
Today
15:
105-111.
Davis,
J.
R.,
T.
Hall,
E.
M.
Chee,
A.
Majala,
J.
N.
Minjas,
and
C.
J.
Shiff.
1995.
Comparison
of
sampling
anopheline
mosquitoes
by
light-trap
and
human-bait
collections
in
doors
at
Bagamoyo,
Tanzania.
Med.
Vet.
Entomol.
9:249-
255.
Detinova,
T.
S.
1962.
Age-grouping
methods
in
Diptera
of
medical
importance.
WHO,
Geneva,
Switzerland.
Draper,
C.
C,
and
P.C.C.
Garnham.
1959.
Effect
of
residual
house
spraying
in
the
plains
on
anopheline
densities
in
huts
on
the
Para
Mountains.
Nature
(Lond.)
183:
198-
199.
Ellman,
R.,
C. A.
Maxwell,
R.
Finch,
and
D.
Shayo.
1998.
Malaria
and
anaemia
at
different
altitudes
in
the
Muheza
district
of
Tanzania:
childhood
morbidity
in
relation
to
level
of
exposure
to
infection.
Ann.
Trop.
Med.
Parasitol.
92:
741-753.
Fontaine,
R.
B.,
A.
E.
Najjar
and
J.
S.
Prince.
1961.
The
1958
malaria
epidemic
in
Ethiopia.
Am.
J.
Trop.
Med.
Hyg.
10:
795-803.
Fontenille,
D.,
J.
P.
Lepers,
G.
H. Campbell,
M.
Coluzzi,
I.
Rakotoarivony,
and
P.
Coulanges.
1990.
Malaria
trans
mission
and
vector
biology
in
Manarintsoa,
high
plateau
of
Madagascar.
Am.
J.
Trop.
Med.
Hyg.
43:
107-115.
Garnham,
P.C.C.
1948.
The
incidence
of
malaria
at
high
altitudes.
J.
Natl.
Malaria
Soc.
7:
275-284.
Garnham,
P.C.C.,
D.
B.
Wilson,
and
M.
E.
Wilson.
1948.
Malaria
in
Kigezi,
Uganda.
J.
Trop.
Med.
Hyg.
5l:
156-159.
Garrett-Jones,
C.
1964.
Prognosis
for
interruption
of
malaria
transmission
through
assessment
of
the
mosquito's
vec-
torial
capacity.
Nature
(Lond.)
204:
1173-1175.
Gillies,
M.
T.,
and
M.
Coetzee.
1987.
A
supplement
to
the
Anophelinae
of
Africa
south
of
the
Sahara.
The
South
African
Institute
for
Medical
Research,
Johannesburg,
South
Africa.
Gillies,
M.T.,
and
B.DeMeillon.
1968.
The
Anophelmae
of
Africa
south
of
the
Sahara.
The
South
African
Institute
for
Medical
Research,
Johannesburg,
South
Africa
Githeko,
A.
K.,
A.
D.
Brandling-Bennet,
M.
Beier,
F.
Afleli,
M.
Owaga,
and
F.
H.
Collins.
1992.
The
reservoir
of
Plasmoditim
faicipanim
malaria
in
a
holoendemie
area
of
western
Kenya
Trans.
Roy.
Soc.
Trop.
Med.
Hyg.
86:
355-358.
'
,
Haddow,
A.
J.
1942.
The
mosquito
fauna
and
the
climate
of
native
huts
at
Kisumu,
Kenya.
Bull.
Entomol.
Res.
33:
91-142.
•
.
;•
'.
Hamilton,
A.
C.
1989.
The
climate
of
the
East
Usambaras,
pp.
97-102.
In
A.
C.
Hamilton
and
R.
Bensted-Smith
[eds.],
Forest
conservation,in
the
East
Usambara,"Tan
zania.
IUCN,
Tanzania.
-•
■
•
'
Heisch,
R.
B.,
and
J.
O.
Harper.
1949.
An
epidemic
of
ma
laria
in
the
Kenyan
highlands
transmitted
by
Anopheles
fimestus.
J.
Trop.
Med.-Hyg.
52:
187-190.
Jetten,
T.
H.,
W.J.M.
Martens,
and
W.
Takken.
1996.
Model
simulations
to
estimate
malaria
risk
under
climate
change.
J.
Med.
Entomol.
33:
361-371.
Kenworthy,
J.
M.
1966.
Temperature
conditions
in
the
tropical
highland
climates
of
East
Africa.
East
Afr.
Geo.
Rev.
4:
1-11.
Khaemba,
B.
M.,
A.
Mutani,
and
M.
K.
Bett.
1994.
Studies
of
anopheline
mosquitoes
transmitting
malaria
in
a
newly
developed
highland
urban
area:
a
case
study
of
Moi
Uni
versity
and
its
environs.
East
Afr.
Med.
J.
71:
159-164.
Lindblade,
K.,
E.
D.
Walker,
A.
W.
Onapa,
J.
Katunga,
and
M.
L.
Wilson.
1999.
Highland
malaria
in
Uganda:
pro
spective
analysis
of
an
epidemic
associated
with
El
Nino.
Trans.
Roy.
Soc.
Trop.
Med.
Hyg.
93:
480-487.
Lindblade,
K.,
E.
D.
Walker,
and
M.
L.
Wilson.
2000.
Early
warning
of
malaria
epidemics
in
African
highlands
using
Anopheles
(Diptera:
Culicidae)
indoor
resting
density.
J.
Med.
Entomol.
37:
664-674.
Lindsay,
S.
W.,
and
M.
H.
Birley.
1996.
Climate
change
and
malaria
transmission.
Ann.
Trop.
Med.
Parasitol.
90:
573-
588.
Lindsay,
S.
W.,
and
W.J.M.
Martens.
1998.
Malaria
in
the
African
highlands:
past,
present
and
future.
Bull.
World
Health
Org.
76:
33-45.
Lindsay,
S.
W.,
R.
B0dker,
R.
Malima,
H.
A.
Msangeni,
and
W.
Kisinza.
2000.
Effect
of
the
1997-98
El
Nino
on
highland
malaria
in
Tanzania.
Lancet
355:989-990.
Lines,
J.
D., E.
Lyimo,
and
C.
F.
Curtis.
1986.
Mixing
of
indoor-
and
outdoor-resting
adults
of
Anoplteles
gambiae
Giles
s.
I.
and
A.
jimesttis
Giles
(Diptera:
Culicidae)
in
Coastal
Tanzania.
Bull.
Entomol.
Res.
76:
171-178.

September
2003
B0DKER
ET
AL.:
MALARIA
TRANSMISSION
ALONG
AN
ALTITUDE
TRANSECT
717
Lines,
J.
D.,
C.
F.
Curtis,
T.
J.
Wilkes,
and
K.
J.
Njunwa.
1991a.
Monitoring
human-biting
mosquitoes
(Diptera:
Culicidae)
in
Tanzania
with
light-traps
hung
beside
mos
quito
nets.
Bull.
Entomol.
Res.
81:
77-84.
Lines,
J.,T.J.
Wilkes,
and
E.
Lyimo.
1991b.
Human
malaria
infectiousness
measured
by
age-specific
sporozoile
rates
in
Anopheles
gambiae
in
Tanzania
Parasitology
102:167-
177.
Lundgren,
L*,
and
B.
Lundgren.
1979.
Rainfall,
interception
and
evaporation
in
the
Mazumbai
forest
reserve,
West
Usambara
Mts.,
Tanzania
and
their
importance
in
assess
ment
of
land
potential.
Geografiska
Annaler.
61:157-178.
Manga,
L_,
E.
Fondjo,
P.
Carnevale,
and
V.
Robert.
1993.
Importance
of
low
Dispersion
of
Anopheles
gambiae
(Diptera:Culicidae)
on
Malaria
Transmission
in
Hilly
Towns
in
South
Cameroon.
J.
Med.
EntomoL
30:936-938.
Martens,
W.J.M.,
J.
Rotmans,
and
L.
W.
Niessen.
1994.
Cli
mate
change
and
malaria
risk.
GLOBO
Report
Series
no
3.
RIVM,
Bilthoven,
The
Netherlands.
Martens,
P.
1998.
Health
and
climatic
change.
Earthscan
Publications,
London,
England.
Martin,
P.
H.,
and
M.
G.
Lefebvre.
1995.
Malaria
and
cli
mate:
sensitivity
of
malaria
potential
transmission
to
cli
mate.
Ambio
24:
200-207.
Matola,
Y.
G.,
G.
B.
White,
and
S.
A.
Magayuka.
1987.
The
changed
pattern
of
malaria
endemicity
and
transmission
at
Amani
in
the
eastern
Usambara
Mountains,
north
eastern
Tanzania.
J.
Trop.
Med.
Hyg.
90:
127-134.
Molineaux,
L.
1988.
The
epidemiology
of
human
malaria
as
an
explanation
of
its
distribution,
including
some
impli
cations
for
its
control,
pp.
913-998.
In
W.
H.
Wernsdorfer
and
I.
McGregor
|eds.],
Malaria.
Churchill
Livingstone,
Edmbourgh,
United
Kingdom.
Moreau,
R. E.
1934.
A
contribution
to
tropical
African
bird-
ecology.
J.
Animal
Ecol.
3:
41-69.
Mouchet,
J.,
S.
Manguin,
J.
Sircoulon,
S.
Laventure,
O.
Faye,
A.
W.
Onapa,
P.
Carnevale,
J.
Julves,
and
D.
Fontenille.
1998.
Evolution
of
malaria
in
Africa
for
the
past
40
years:
impact
of
climatic
and
human
factors.
J.
Am.
Mosq.
Con.
Ass.
14:
121-130.
Rogers,
D.J.,
and
S.E.
Randolph
2000.
The
global
spread
of
malaria
in
a
future,
warmer
world.
Science
289:1763-1766.
Roberts,
J.M.D.
1964.
The
control
of
epidemic
malaria
in
the
highlands
of
westerm
Kenya.
Part
II.
The
campaign.
J.
Trop.
Med.
Hyg.
67:191-199.
Sharp,
B.,
and
J.
Cox.
1998.
Highland
malaria,
pp.
22-24.
In
Anonymous.
Towards
an
atlas
of
malaria
risk
in
Africa.
MARA/ARMA.
Durban,
South
Africa.
SPSS.
1999.
SPSS
Base
user's
guide,
release
9.0.
SPSS,
Chi
cago,
EL.
Sutherst,
R.
W.
1993.
Arthropods
as
disease
vectors
in
a
changing
environment.
In
Environmental
Change
and
Human
Health.
Ciba
Found.
Symp.
175:
124-145.
White,
G.
B.
1971.
Blood
feeding
habits
of
mosquitoes
in
the
South
Para
District
of
Tanzania
ten
years
after
cessation
of
a
Dieldrin
residual
spraying
campaign.
East
Afr.
Med.
J.
48:
122-134.
White, G.
B.
1972.
The
Anopheles
gambiae complex
and
malaria
transmission
around
Kisumu,
Kenya.
Trans.
R.
Soc.
Trop.
Med.
Hyg.
66:
572-581.
Wirtz,
R.
A.
1994.
Sporozoite
ELISA
directions.
Walte:
Reed
Army
Institute,
Washington,
DC.
Received
for
publication
2
October
2002;
accepted
9
May
2003.