CLIMATE CHANGE AND HIGHLAND MALARIA: FRESH AIR FOR AHOT DEBATE

Luis Fernando ChavesDepartment of Environmental Studies, Emory University, Atlanta, GA 30322 USA

e-mail: lfchave@emory.edu

Constantianus J. M. KoenraadtLaboratory of Entomology, Wageningen University, Wageningen, The Netherlands

e-mail: sander.koenraadt@wur.nl

keywordsmosquito, Anopheles, time series, population dynamics, disease emergence,

land use change

abstractIn recent decades, malaria has become established in zones at the margin of its previous distribu-

tion, especially in the highlands of East Africa. Studies in this region have sparked a heated debateover the importance of climate change in the territorial expansion of malaria, where positions rangefrom its neglect to the reification of correlations as causes. Here, we review studies supporting andrebutting the role of climatic change as a driving force for highland invasion by malaria. We assessedthe conclusions from both sides of the argument and found that evidence for the role of climate in thesedynamics is robust. However, we also argue that over-emphasizing the importance of climate ismisleading for setting a research agenda, even one which attempts to understand climate changeimpacts on emerging malaria patterns. We review alternative drivers for the emergence of this diseaseand highlight the problems still calling for research if the multidimensional nature of malaria is to beadequately tackled. We also contextualize highland malaria as an ongoing evolutionary process.Finally, we present Schmalhausen’s law, which explains the lack of resilience in stressed systems, asa biological principle that unifies the importance of climatic and other environmental factors indriving malaria patterns across different spatio-temporal scales.

T HE 1957 Cold Spring Harbor Sympo-sium on Quantitative Biology is widely

known among ecologists as the setting forone of the most heated debates in the historyof population ecology. This debate pittedproponents of exogenous factors as mecha-nisms behind population regulation against

those who emphasized endogenous factors.This argument can especially be seen in theopen discussions of the symposium seminarsled by Birch (1957) and Andrewartha(1957), in which they presented examplesillustrating the importance of exogenous,density-independent drivers (e.g., climate

The Quarterly Review of Biology, March 2010, Vol. 85, No. 1

Copyright © 2010 by The University of Chicago Press. All rights reserved.

0033-5770/2010/8501-0002$15.00

Volume 85, No. 1 March 2010THE QUARTERLY REVIEW OF BIOLOGY

27

variables such as rainfall or temperature) onthe distribution and abundance of insectpopulations. Some of these argumentswere questioned by Nicholson (1957) withhis laboratory results on blowflies. His worksupported the hypothesis that endogenous,density-dependent factors (e.g., number ofindividuals) can drive animal population dy-namics through changes in the life historytraits of individuals that have experienceddifferent levels of crowding. This debateprompted significant research efforts in pop-ulation ecology in the decades that followed.Today, the synthesis and abstraction of obser-vation acknowledges that populations have aself-regulatory component (negative feed-back), that the endogenous component ofNicholson is always present, and that exoge-nous forces also influence the fate of a pop-ulation; therefore, both factors are neededto explain population dynamics (Turchin2003). Infectious diseases, as phenomena in-volving organisms and their environments,raise the same questions that have been theobject of study of ecologists for a long time.For example, what determines the spatial dis-tribution and prevalence of certain diseases?What factors are responsible for their re-emergence, and how do these factors inter-act? Infectious diseases that are transmittedfrom person to person by vectors (e.g., mos-quitoes, sand flies, and ticks) are the mostchallenging, since both exogenous and en-dogenous factors could act on the vector andthe disease agent, as well as the humans ul-timately infected by the disease agent (Mac-donald 1953).

The proposition that climate is responsi-ble for the distribution and abundance ofinsects (Andrewartha 1957; Birch 1957) suchas mosquitoes, together with the realizationof climate change by the scientific commu-nity, has prompted some scientists to forecastan expansion of the current malaria distribu-tion and an increase in its burden followinga rise in temperature (e.g., Patz et al. 2002;Pascual and Bouma 2009). However, factorsregulating the abundance and distributionof malaria are multivariate and complex(e.g., Macdonald 1953; Lindsay and Birley1996). In fact, malaria has been absent forsome time from areas that harbor large pop-

ulations of competent mosquito vectors andwith environmental conditions conducive toparasite development and transmission. Ma-laria has also been endemically present inplaces with the most adverse conditions forparasite transmission, which shows that, atbest, the consideration of climate alone willlead to a rough approximation of malaria’sgeographical range and that other factorscan influence its distribution within orbeyond the boundaries that an optimallysuitable environment could suggest. For ex-ample, it is well-documented that the Ro-mans knew the relationship betweenmarshes and “malaria-like” fevers, and thisknowledge informed their selection of placesfor new settlements and military camps.Early human history is full of battles thatwere won with the aid of malaria as an addi-tional platoon attacking enemies (de Zulu-eta 1987; Najera 1994). Even during thecoldest years of the Little Ice Age (1560s to1730s), reports of malaria outbreaks in En-gland and Scotland were common. In En-gland, epidemics were associated with yearsof high famine (Reiter 2000), and, in Scot-land, the largest epidemics happened duringthe warmest and wettest summers of this pe-riod (Duncan 1993). Malaria was also com-mon in Scandinavian countries during the19th century (Reiter 2008). After World WarII, malaria endemic transmission was erasedfrom the USA (Humphreys 2001), and, since1973, Europe has been declared free of thedisease (de Zulueta 1987, 1994). In 1981,Australia—a large subtropical area—alsoachieved “malaria-free” status (Bryan et al.1996).

Unfortunately, the pattern of waningmalaria seen in the USA, Europe, and Aus-tralia is not globally widespread. Today, achild dies of malaria every 40 seconds, andbetween one and three million people dieeach year around the world, primarily insub-Saharan Africa (Sachs and Malaney2002). Although these figures depict spa-tially coarse patterns, they do not revealthe fact that spatially fine patterns havebeen changing. Such patterns are quite ev-ident in the geographical malaria hot spotthat is the focus of this paper: the sub-Saharan African highlands. These high-

28 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

lands are sparse land extensions in severalAfrican countries, with altitudes of at least1200 m (Lindsay and Martens 1998).

Historically, the African highlands havebeen used as a shelter against malaria. Inthis context, it is important to note thatevery 100 m increase in altitude is associ-ated with a 0.6 °C decrease in temperature(Strahler 1973). Also, malaria parasites donot develop inside their mosquito host iftemperatures fall below 15 °C (Patz andOlson 2006). Thus, malaria has been nat-urally excluded from these areas, mostlikely because they experience conditionsthat limit the biology of the parasite (Lind-say and Martens 1998).

In the past, malaria transmission in theEast African highlands was mainly sporadicand unstable (i.e., epidemic). Neverthe-less, since the second half of the 20th cen-tury, this pattern has changed, with anincrease in the frequency of epidemic out-breaks (Lindsay and Martens 1998; Pascualet al. 2008; Shanks et al. 2005). Malariainvasion of highlands is not restricted tosub-Saharan Africa, but is globally wide-spread. It has also invaded the highlands ofIndonesia (Anthony et al. 1992), PapuaNew Guinea (Mueller et al. 2005), Mada-gascar (Bouma 2003), and Afghanistan(Rab et al. 2003).

The absence of a common pattern leadingto the emergence or extinction of this dis-ease has created an intense debate over themain determinants of malaria resurgence inthe East African highlands. In this review, westart by presenting the effects of the differentelements of climate (temperature, rainfall,and humidity) and the natural environment(breeding sites, vegetation) on the entomo-logical parameters relevant for malaria trans-mission. We then re-analyze data thatsupport warming trends in the East Africanhighlands, where malaria is expanding itsgeographical range. In this respect, we alsoaddress the possible differences between theeffects of climatic trends and climatic vari-ability on malaria transmission. Because thecurrent global distribution shows that ma-laria has not expanded into areas as previ-ously forecasted by modelers, we also reviewother important factors that are considered

robust predictors of malaria transmission—primarily demographic drivers and land usechange. In the final two sections, we presentSchmalhausen’s law as an epistemologicaltool for understanding the contradictory ef-fects of climatic drivers at the local scale oftransmission, and to also better our under-standing of the effects of all the relevantfactors for malaria transmission at the differ-ent evolutionary and ecological spatio-temporal scales of this disease.

Entomological Parameters and thePlausibility of Climate Change

Driving Malaria Resurgence in theAfrican Highlands

Since Ronald Ross’s (1897) discovery ofmosquitoes as vectors of malarial parasites,many attempts have been made to under-stand the eco-epidemiology of the diseaseand the risk to which humans are exposed.The first major step was the introduction ofthe “basic reproduction rate” (Macdonald1953), which expresses the number of newcases originating from a single case of ma-laria in the absence of immunity (Table 1).By leaving out the parasitological aspects ofthe equation, Garrett-Jones (1964) intro-duced the concept of “vectorial capacity”(C), which provides a theoretical frameworkfor calculating the number of expected inoc-ulations of man per infective case per day.Most parameters of these equations are dif-ficult to measure under field conditions, andit is therefore more convenient to expressthe level of malaria transmission as the “en-tomological inoculation rate” (EIR), or thenumber of infective bites per person per day(Onori and Grab 1980). Public health sur-veys may also gather data related to the ac-tual outcome of the intensity of transmis-sion—i.e., the number of infected hosts(parasite prevalence), sick cases in the hostpopulation (morbidity), and deaths (mortal-ity). All these may be used to compare trans-mission dynamics between areas, such ashighland and lowland areas.

Various efforts have been made to studythe relationship between malaria risk pa-rameters. These have shown that EIR canbe used to predict levels of parasite preva-lence (Beier et al. 1999) and mortality

March 2010 29CLIMATE CHANGE AND HIGHLAND MALARIA

(Smith et al. 2001), although others haveargued that this relationship may not bevalid, as some epidemiological settings mayhave been under-represented in the meta-analyses (Trape et al. 2002). The use ofdifferent mosquito collection techniques,the estimation of yearly EIR from monthlyEIR, and the level of natural acquired im-munity in the population may all lead tothe variation observed in these studies.However, despite these variations, it canstill be asserted that studying the effects ofclimate on the entomological parametersof malaria transmission should be the ini-tial step to understanding possible impactsof changing environments on this disease.In the following sections, we present adetailed account of the effects of tempera-ture, rainfall, humidity, distance to breed-ing sites, and vegetation on malaria.

temperatureDevelopment of Plasmodium falciparum

in the mosquito host takes approximately9–10 days at a temperature of 28 °C, whiledevelopment ceases when temperaturesdrop below 16 °C (Macdonald 1953). Lar-val populations of Anopheles gambiae, themajor vector of P. falciparum in Africa, willnot develop into adults when ambient tem-peratures drop below �16 °C (Jepson et al.1947), and daily survival of adults reacheszero at around 40 °C (Lindsay and Martens

1998). A higher temperature leads to ahigher digestion rate of the bloodmeal,which results in more frequent vector-hostinteractions. Combining the survival of themosquito population (p) and the develop-ment time of the parasite (n) gives us theproportion of mosquitoes that survives theincubation period of the parasite (pn) (Ta-ble 1). The highest proportions survivingthe intrinsic incubation period can befound between 28 and 32 °C (Craig et al.1999).

In the field, both minimum and maxi-mum temperatures have been related to theabundance of vector populations: higherminimum temperatures led to increased vec-tor densities in the highlands of Uganda(Lindblade et al. 2000), while higher maxi-mum temperatures were correlated withlower biting rates in western Kenya (Patz etal. 1998). Interestingly, Minakawa et al.(2002b) found that two sibling species of theAn. gambiae complex, An. gambiae sensu strictoand An. arabiensis, were differentially sensi-tive towards ambient temperatures; highermaximum temperatures were associated withhigher densities of An. arabiensis, whereashigher minimum temperatures were relatedwith lower densities of An. gambiae s.s. This isin accordance with other findings on thisspecies complex, whereby An. gambiae s.s.performs better under cooler conditions(Petrarca et al. 1991; White 1972). Other

TABLE 1Overview of malaria risk parameters

Malaria risk parameter Definition Mathematical expression

Basic reproduction rate Number of secondary infections originating froma primary case in the absence of immunity

ma2 bpn/ � r (ln p)#

Vectorial capacity Number of secondary inoculations of man perinfective case per time unit

ma2 pn/ � ln p†

Entomologicalinoculation rate

Number of infective bites per person per timeunit

ma* s‡

Source: Macdonald (1953)#; Garrett-Jones (1964)†; Onori & Grab (1980)‡.Explanation of parameters:a � Man-biting habitm � Mosquito density relative to manb � Susceptibility of mosquito to parasite infectionp � Daily mosquito survivaln � Incubation period of parasite in mosquitor � Recovery rate from infections � Sporozoite rate (proportion of infected mosquitoes)

30 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

studies did not find a clear relationship be-tween temperature, on the one hand, andvectorial capacity or entomological inocula-tion rate on the other (Lindblade et al. 2000;Shililu et al. 1998), perhaps as a result of thepresence of other confounding factors.

Numerous efforts have been made to linktemperature directly to malaria incidencerates throughout the African continent(Bouma 2003; Craig et al. 2004; Klein-schmidt et al. 2001; Mabaso et al. 2007;Shanks et al. 2002; Ye et al. 2007; Zhou et al.2004). These have used various statisticalmodelling approaches and consideredmean, maximum, and minimum tempera-tures, as well as deviations from long-termaverages (anomalies) and lag times (see be-low for a discussion on the use of statisticaltools and inferring conclusions from data).Although there is no consensus on a single,most valuable temperature predictor formalaria risk, we argue that each ecologicalsetting requires its own validation of temper-ature and an assessment of how it can beincluded in predictive models.

rainfallMany African malaria vectors are depen-

dent on rainfall, since they prefer to breedin small temporary sites created by therains (Gillies and Coetzee 1987). It istherefore not surprising that in the earlydays of malaria research, the onset of rainyseasons was associated with increased vec-tor densities and malaria cases (Garnham1929; Gill 1920; Haddow 1942; Holstein1954). However, the lack of powerful tools(e.g., statistical tools and those used forgathering geographic information) madeit hard to quantify the relationship withrespect to spatial and temporal variation inrainfall patterns. Today, the amount ofweekly rainfall has been correlated withthe abundance of larval habitats with atime lag of one week and with vector den-sities inside local houses with a lag of twoweeks (Koenraadt et al. 2004a). The monthlyrainfall anomaly (i.e., the deviation in rain-fall from the month’s historical average)was correlated with vector density onemonth later (Lindblade et al. 1999).

As with temperature, numerous studies

included rainfall as a direct predictor ofmalaria incidence (Binka et al. 1994; Craiget al. 2004; Drakeley et al. 2005; Githekoand Ndegwa 2001; Kazembe 2007; Klein-schmidt et al. 2001; Mabaso et al. 2007;Shililu et al. 2003; Ye et al. 2007; Zhou et al.2004), but the relationship between rain-fall and malaria risk is not always present(la Grange 1995; Lindblade et al. 1999;Shanks et al. 2002; Shililu et al. 1998). Thismay be caused by the fact that certain vec-tors of malaria, such as An. funestus, are lessdependent on rainfall, since they prefer tobreed in more permanent habitats (Gilliesand Coetzee 1987). For example, Mabasoet al. (2007) found that rainfall was animportant determinant of seasonality inEIR, but not so in areas characterized bytwo rainfall peaks or irrigation activities.The distribution of rains over the year, thecondition of the soil before the rains actu-ally start, and the soil water holding capac-ity may also obscure correlations (Kazembe2007).

humidityHumidity determines the life span of the

mosquito and, thus, its capability to transmitthe malaria parasite (Clements 1999). Hu-midity is a direct product of rainfall and tem-perature, since these govern the amount ofwater available and the amount of water theatmosphere can hold, respectively. Accord-ing to Minakawa et al. (2002b), a highermoisture index (expressed as the ratio ofrainfall over evapotranspiration) was associ-ated with a lower density of An. arabiensis, asconsistent with earlier experimental findingsthat this species performs better at lower hu-midity levels (Coz 1973). Soil moisture levelwas significantly correlated with EIR in west-ern Kenya (Patz et al. 1998), indicating thatthe water balance—i.e., rainfall � (evapo-transpiration � runoff)—plays an importantrole in determining the number of infectivebites people eventually receive. However, rel-ative humidity levels were not related to vec-tor density in a highland area of Uganda(Lindblade et al. 2000), but they were asso-ciated with malaria incidence in BurkinaFaso (Ye et al. 2007). All this suggests thatlocal variation exists.

March 2010 31CLIMATE CHANGE AND HIGHLAND MALARIA

distance to breeding sitesMalaria has been associated with the

presence and proximity of marsh areas—hence the name “mal aria” or “bad air”—since the earliest descriptions of the diseaseand its symptoms. Studies on the proximityof water bodies, either natural or man-made, showed that indoor vector densitieswere higher close to breeding sites (Lind-say et al. 1995; Minakawa et al. 2002a;Minakawa et al. 2004) and that malariarisk was, consequently, increased in theseareas (Bøgh et al. 2007; Clarke et al. 2002;Ghebreyesus et al. 1999; Kleinschmidt etal. 2001; Lautze et al. 2007; Oesterholt etal. 2006; Staedke et al. 2003; Trape et al.1992). Interestingly, Clarke et al. (2002)noted that, although risk in terms of expo-sure was decreased farther away from waterbodies, clinical illness was more common,probably as a result of reduced immunitydue to lower exposure to infective bites.These studies all suggest that control oflarval breeding sites may contribute signif-icantly to a decreased risk of malaria, andlarval control strategies should thereforeplay an integral role in fighting this disease(Killeen et al. 2002).

vegetationVegetation provides shelter and suitable

resting sites for mosquitoes, and it also cre-ates microhabitats in which temperatureand humidity conditions are more suitablethan in areas without vegetation (Clements1999). The presence of vegetation maytherefore enhance mosquito longevity. Theadvancement of satellite imagery has made itpossible to quantify the amount of photosyn-thesizing vegetation by means of the normal-ized difference vegetation index (NDVI),which is based on the absorption and reflec-tion of light waves of different lengths. Posi-tive correlations have been found betweenNDVI and the bite rate among humans forAn. gambiae (Patz et al. 1998), as well as be-tween NDVI and malaria prevalence/incidence (Gomez-Elipe et al. 2007; Hay etal. 1998; Thomson et al. 1999). These corre-lations most likely arise as a result of the closerelationship between temperature and rain-

fall, on the one hand, and vegetation growthon the other. However, another study didnot find any correlation between vectorabundance and NDVI (Shililu et al. 2003).

Time Series Analysis and theControversy about the Effect of

Climate on MalariaEvidence for global climate change greatly

relies on accurate observations of weathervariables. Thus, given the limited opportuni-ties for manipulation of observations, it iscritical for inferences to be robust. Robust-ness, or the confidence in the certainty ofone’s result, given that assumptions about suchphenomenon might be wrong or untestable,can be evaluated through the agreement ofresults obtained with different methods or withdifferent models used to analyze the data(Levins 1966, 2006).

Research that supports that the emer-gence of malaria in the African highlandsover the last fifty years is not related toclimate change relies on inferences fromtwo methods of data analysis. First, lowspectral power (variance) was found forperiods of time longer than one year in aspectral density analysis of temperatureand rainfall time series from weather sta-tions at Kericho, Kenya (Hay et al. 2000).Second, parametric trends for climate vari-ables from local weather stations and fromdata interpolated using geographical infor-mation systems (GIS) tools were not statis-tically significant (Hay et al. 2002a; Shankset al. 2002; Small et al. 2003).

Time series analysis can be performed inthe time domain, the frequency domain,and the time-frequency domain. In equa-tions of the time domain approach, thevalue of the variable under study is pre-dicted from its previous values (Brockwelland Davis 2002). The frequency domainlooks for cyclic regularities in time series,while the time-frequency domain showshow the cycles of a time series develop(Shumway and Stoffer 2000). In our sup-plementary material (available online atThe Quarterly Review of Biology homepage,www.journals.uchicago.edu/QRB), a glos-sary of technical terms related to time se-ries analysis is presented.

32 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

In the time domain, one regressionmodel was used to test for trends in climateand malaria suitability (Hay et al. 2002a;Shanks et al. 2002):

�yt � � � �t � �yt�1 � �i�1

p

�i�yt

� �j�1

12

�jdj � t , (1)

and a slightly modified version (Small et al.2003)

�yt � � � �t � �yt�1 � �i�1

p

�i�yt � t , (2)

where y is the variable of interest, �, �, �are regression parameters, � is the param-eter for a deterministic time trend, and t isthe time. Seasonality is included in themodel via the sum �j�1

12 �j�0, and dj , adummy variable that takes values of 1 or 0depending on whether y belongs to themonth that is being regressed. t repre-sents a random independently, identically,and normally distributed error, with vari-ance (

2) (i.e., t � N(0,2)).

This modeling approach tests whetherthere are unit roots in the autoregressiveprocess (� � 0; i.e., the process being non-stationary), as well as deterministic (� � 0)or random trends (� � 0, given � � 0).Depending on the hypothesis being evalu-ated, the statistics used to test it have anormal or an augmented Dickey-Fuller dis-tribution (for details, see Brockwell andDavies 2002). The aforementioned studiesdid not find statistically significant resultsfor deterministic or random trends; how-ever, error assumptions were not tested ad-equately. Only one portmanteau test forgeneral serial correlation was performed,whereas autocorrelation should be testedfor several time lags. Thus, the time do-main evidence against the existence of cli-mate change in the African highlandsneeds to be revised.

As we mentioned previously, one strategyfor achieving robustness in a result obtainedfrom observational data is to test several dis-

tinct models. Given that the gridded temper-ature time series (Hay et al. 2002a; Shanks etal. 2002) used for Kericho, Kenya (0.33°S,35.37°E, 1700-2200m) is available online(http://www.cru.uea.ac.uk/), we fitted threeadditional models for this series. For detailson the techniques that we used, please seethe supplementary material.

Figure 1 shows the results for the non-parametric reconstruction of the third or-der components of a Singular SpectrumAnalysis (SSA), the trends of a State SpaceBasic Structural Model (BSM), and a Sea-sonal Auto Regressive (SAR) model. Incontrast with the time series analyses de-scribed above, the methods we used agreeon the existence of positive trends in thetemperature of Kericho. Similar results—i.e., positive trends—not shown, were ob-tained for the other three places studied(Kabale, Muhanga, and Gikongoro) (Hayet al. 2002a), and, in all cases, the assump-tions of the parametric models BSM andSAR were not violated.

The frequency domain evidence againstexogenous drivers for malaria in the high-lands has methodological as well as inferen-tial problems (Hay et al. 2000). The reducedvariability (or power) at periods longer thana year is not enough to prove the lack ofimportance of climatic variables, since this isjust a descriptive measure of cycles in thedata, not their correlation to other variables.Data were also filtered with time averaging,which is known to erase variability peaks atlow frequencies, thus eliminating long pe-riod cycles (Shumway and Stoffer 2000). Tomake proper inferences about the lack ofimportance of climate in driving malaria dy-namics in humans, a bivariate or correla-tional analysis, such as a phase/coherencyanalysis, relating malaria incidence and cli-mate data in the frequency domain shouldhave been used (Shumway and Stoffer2000). However, because this type of analysiswas not used, an inference about the rela-tionship between both phenomena cannotbe made. In the supplementary material,Figure1S shows, using both frequency andtime frequency techniques, that cycles forperiods longer than a year are indeed signif-

March 2010 33CLIMATE CHANGE AND HIGHLAND MALARIA

icant, meaning that temperature has longercycles than the normal seasonality.

In summary, there are elements from allthe domains of time series analysis that

support an increasing trend in tempera-ture in the highlands of Kericho, as well asthe existence of changing interannual cy-cles in weather; therefore, there is evi-

Figure 1. Kericho Temperature Time Series Analysis in the Time Domain(a) CRU mean temperature time series for Kericho, Kenya, 1966–2002. (b) Non-parametric trend obtained

by reconstructing the third component of a 96 order Singular Spectrum Analysis (SSA). (c) Kalman smoothedBasic Structural Model (BSM) trend. For Kericho, the parameters are:

2 � 1.33 � 10 � 2; �2 � 8.43 � 10 � 2; �

2 � 1.39 � 10 � 5; 2 � 1.73 � 10 � 2.

(d) Linear trend from a Seasonal Auto Regressive (SAR) model. The model fitted was:

yt � �1 yt � 1 � �k�12

15

�kyt � k � �j�13

16

�1� j � 1yt � j � �time � wt, where wt�N(0, w2).

The parameters (� SE) are:�1 � 0.44 � 0.05;�12 � 0.20 � 0.05;�13 � 0.18 � 0.05;�14 � 0.11 � 0.05;�15 � 0.24 � 0.05;� � 0.0098 � 0.0001;w � 0.31.The three trends were normalized to have a mean of zero.

34 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

dence for the existence of climate changein this East African highland area.

The importance of climate is reinforcedby the globally widespread pattern of malariathat varies seasonally. Seasonal patterns havebeen described in India (Madhavan et al.2001), Pakistan (Bouma et al. 1996), SriLanka (van der Hoek et al. 1997), the Sahel(Delmont 1982; Ndiaye et al. 2001), Ethiopia(Abeku et al. 2004), Uganda (Killian et al.1999), Botswana (Thomson et al. 2005), andSouth Africa (Craig et al. 2004). Throughstatistically significant correlation measures,all the studies found an association betweenincreased malaria incidence and increasedvalues of temperature and rainfall; however,none of these studies addressed the issue ofweather variability at scales longer than an-nual.

Supporting the suggestion that “climatechange also applies to changes in variability”(Patz et al. 2002:627), several authors havefound evidence for changes in local annual,quarterly, or monthly malaria incidences as-sociated with the El Nino Southern Oscilla-tion, or ENSO (Bangs and Subianto 1999;Bouma 2003; Gagnon et al. 2002; Lindsay etal. 2000). ENSO is an interannual climaticphenomenon characterized by an increasein atmospheric pressure and sea surface tem-perature off the coast of Peru, with an oppo-site anomaly occurring in the western PacificOcean. This phenomenon is strongly associ-ated with several other weather anomalies—primarily, increased or decreased area-specific rainfall (Kovats et al. 2003). Usingmonthly data from the African highlands,Zhou et al. (2004) also found evidence ofclimate change by utilizing an entirely differ-ent approach. They examined the meanvalue and variance of annual data from sev-eral places grouped into two time intervals,1978–1988 and 1989–1998. They reportedsignificant differences in the variance of atleast one climatic variable in 6 of the 7 placesthey studied, although the analysis was basedon several t-tests. As pointed out by Hay et al.(2005), this approach is flawed, because itunderestimates the degrees of freedom formaking reliable comparisons. In response tothis criticism, Zhou et al. (2005) argued thatgiven their number of tests (a total of 21),

the possibility of getting a false rejection tothe null hypothesis of no climatic change isreduced, because of multiple testing, to 1 ofthe 21 tests they performed. However, whatmatters when making several comparisons isnot the significance and the probability ofmaking a type I error—that is, rejecting thenull hypothesis when true—but the power ofthe statistic being used (i.e., the probabilityof making a type II error, or the likelihood offailure to reject a null hypothesis when false).Also, the study by Zhou et al. (2004) had thelimitation of not taking into account the pos-sible spatial correlation between the differ-ent climate variables for each place, as well asthe possible temporal correlation for eachvariable.

As argued earlier, it is worth noting thatin all the studies we have discussed sup-porting linkages between climate changeand malaria, there is neither a unique cli-matic pattern nor a unique climate vari-able associated with the disease. In fact,both the mean value and the variability ofsuch variables appear to be associated withrecent changes seen in malaria in the Afri-can highlands. The diversity of these re-sults should not be taken as evidenceagainst the influence of exogenous cli-matic drivers, but instead as an indicationof the lack of knowledge on the mecha-nisms of such linkage.

To illustrate this lack of knowledge, wereview four models relating malaria dynam-ics to climatic variables. The first model wasproposed by Teklehaimanot et al. (2004). Itincluded a temporal trend and cited rainfalland temperature as major factors in predict-ing the number of malaria cases in Ethiopia.This study found mostly positive effects forboth temperature and rainfall. A secondmodel by Zhou et al. (2004) included theeffect of rainfall, temperature, and their non-additive interaction (i.e., “a synergistic ef-fect”). This model was fitted using inpatientdata—i.e., people whose infection was suffi-ciently severe to require hospitalization—and found both positive and negative effectsfor rainfall and temperature at different lags.Thus, parameters were obtained for the mostsevere clinical cases—a fraction of the totalnumber of cases (Malakooti et al. 1998). Ad-

March 2010 35CLIMATE CHANGE AND HIGHLAND MALARIA

ditionally, no adjustment for populationgrowth was carried out, thus implying thatboth population and health facilities hadgrown at similar rates during the studied pe-riod (Zhou et al. 2004). The third model,proposed by Abeku et al. (2004) for datafrom Ethiopia, developed a linear regressionmodel motivated by classical malaria trans-mission theory to give a biological interpre-tation to the statistical approach, in contrastto the two previous models. This model wasconstructed by assuming that malaria inci-dence (N) is mainly determined by vectorialcapacity (C) in the previous month:

Nt � aNt�1b Ct�1. (3)

In this basic version of the model, a and bare area specific constants, and b ranges from0 to 1 depending on the area’s endemicity (0for the highest endemicity). The model wasparameterized, using temperature and rain-fall data, by considering the spatial effects ofseveral counties and ignoring any demogra-phy in the populations (e.g., migration orsize). This study found positive effects forrainfall and a convex relationship with tem-perature (mostly postive).

A fourth model was developed by Pas-cual et al. (2006) in which the life historystages of Anopheles gambiae are parameter-ized as a function of temperature. Thismodel showed that a small change in tem-perature can amplify mosquito populationabundance by several folds. This work alsoshowed a key aspect of the potential effectsof climate change on the transmission ofmalaria and, more generally, on the fate ofpopulations: even if changes in the envi-ronment are very small or not even statis-tically significant, they can be amplifiedbecause of the biology of the organisms(Shaman et al. 2002).

These four examples illustrate how differ-ent models for different places or aspects ofthe same phenomenon can find differentpatterns of association between malaria andclimate. As shown in this section, malaria canbe correlated, linearly and non-additivelythrough synergistic effects, with several cli-matic variables at intra- and inter-annualscales. Thus, although the evidence of thelinkage between changes in climate and ma-

laria incidence in the African highlands isrobust, there is still a lack of knowledgeabout the ecological mechanisms behind themore phenomenological models. There isan urgent need to investigate if and how gen-eralizations can be made among the diverseresults found on the relationship between cli-matic variables and malaria. For example, ma-laria in relatively stable climates may bedriven more strongly by mean temperature,whereas minimum temperature is a strongerdriver for cooler climates, such as highlands.Also, researchers are just uncovering the roleof diversity in vector species, of sibling spe-cies within species complexes, and of thevarious chromosomal (based on inversionpolymorphisms) and molecular forms of themain vector, Anopheles gambiae s.s., in malariatransmission (della Torre et al. 2005). Theconsensus on this complexity seems to bethat the various forms are incipient speciesand, thus, in the process of occupying theirown niches. Because they are differentiallysensitive towards climatic conditions, under-standing their biology in relation to malariatransmission and environmental changeshould hold high priority in future researchprograms. Table 2 presents further direc-tions for future research in ecological ento-mology.

Malaria’s Neglected Phenomena: ThePanclimatic Paradigm

The panclimatic paradigm is the reduc-tion of global vector-borne disease emer-gence or resurgence, as linked to changes inclimate over recent years. This paradigm isthe product of reifying results from statisticalmodels applied to study the patterns of associ-ation between climate and infectious diseases,or from the successful fitting of mathematicalmodels that include climate variables. De-spite evidence, or suggestive observations,for alternative causes behind the emergenceand resurgence of malaria, the panclimaticparadigm has been a cliche in the resur-gence research agenda for some time now(e.g., Hoshen and Morse 2004; Rogers andRandolph 2000; Tanser et al. 2003; Thomaset al. 2004; Pascual and Bouma 2009). Reiter(2000, 2008) and Reiter et al. (2004) havecriticized this argument, emphasizing that

36 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

the abuse of “climate change” arguments toexplain malaria trends in the African high-lands is inaccurate. They argue that climate,if associated with this resurgence, is not nec-essarily the ultimate and only cause of suchchange. The search for ultimate causes andthe focus on the isolated “main drivers” ofmalaria resurgence in the African highlandsis a necessary step in the research process.However, a belief in a unique, isolated, andautonomous main driver is at best an expres-sion of a reductionist approach to under-standing nature. This approach is based onthe assumption that the understanding ofisolated parts will lead to a full comprehen-sion of the entire system that embeds them(Lewontin and Levins 2000). This way ofthinking about phenomena is reinforced bythe fact that mathematical tools, especiallythose of statistical data analysis, tend to bereductionistic.

For example, Teklehaimanot et al. (2004)pointed out the instability in parameter esti-mation by linear models. This instabilityarises when predictors have a very high lin-ear correlation, close to � 1. The problem iscaused by the unidentifiability of the predic-tor matrix (Faraway 2005), which does notallow for some necessary manipulation tocompute parameters using linear models.Basically, the predictor matrix is near singu-lar, and this produces numerical errors dur-ing matrix inversion, ultimately reflecting aconstraint of linear algebra. In addition, amodel fitted using the framework of linearor generalized linear models will always carrythe uncertainty of predictors that can covarywith the “causal” predictor, without havingany direct effect on malaria transmission.However, there are statistical strategies thatmay avoid the problems of strongly corre-lated variables, such as the use of multivari-ate tools like principal components analysisand multidimensional scaling (e.g., Chaveset al. 2008a). Thus, although the limitationsof statistical tools are always present, ways tohandle them have been devised. Yet, in anycomplex system with relatively simple feed-back pathways, some perturbations canchange the signs of correlations, leading tothe absence of correlation or the finding ofparadoxical or unexpected relationships

(Lewontin and Levins 2000). In fact, it hasbeen widely documented that interactionsbetween organisms can change in sign andmagnitude through the dynamic develop-ment of their association (Hernandez 1998).Therefore, problems with parameter estima-tion need to be understood as products of asystem that will likely become unstable givenany perturbation. The parameters estimatedwith linear models, which are correct anduseful for predictions of the data used to fitthe model and to gain insights into the dy-namics of disease transmission, need to bepresented with caution based upon the sta-bility of the system. By contrast, forecasts orpredictions about the future can be very im-precise because of the sensitivity of nonlinearmultidimensional systems to all of their un-derpinnings, especially those that are not ac-counted for by the studied models.

As a result of the above constraints, thepanclimatic paradigm is epistemologicallyweak, especially since it attempts to makepredictions outside its limits of validity whenforecasting future scenarios for malaria dis-tribution and incidence based on climatealone. Consequently, the search for alterna-tive malaria drivers has become one of themajor fields of inquiry over recent years. Thebiggest problem with alternative drivers formalaria resurgence is that they have notbeen quantified and, in general, there is nota good understanding of their dynamicmechanisms (Jones and Williams 2004). Aspresented by Wilson (1994), diseases in hu-mans show patterns of emergence and resur-gence resulting from new cultural practices,new patterns of human migration, biologicalinvasions of pathogens and vectors, and en-vironmental change (widely defined and notrestricted to climate). The plausibility ofother drivers behind the global resurgenceof malaria has been widely acknowledged.The pioneering work of MacDonald (1953)already recognized the cyclical nature of ma-laria as a product of the interaction of factorsintrinsic to human hosts (e.g., immunitybuildup) and exogenous to human beings,such as climatic seasonality, as well as theeffects of changes in the environment thatbring about the modification of the ecologyof mosquitoes and their interaction with hu-

March 2010 37CLIMATE CHANGE AND HIGHLAND MALARIA

TABLE 2Setting a research agenda: topics for malaria research in a changing environment

Field Current observations Tools/Concepts Key questions/Goals

Ecological entomology Rainfall and temperature● Pulses in rainfall can

sometimes favorvector populations(e.g., Koenraadt etal. 2006), but this isnot always the case(e.g., Chase andKnight 2003).

● Temperatureincrease reduces thedevelopmental timeof mosquitoes (e.g.,Bayoh and Lindsay2003)

● Differences in adultheat tolerance bymosquito species(e.g., Kirby andLindsay 2004)

● Density-dependenceand mosquitopopulationregulation is notfully understood(Yang et al. 2008).

● Classical population andcommunity ecology: laboratoryand field studies on oviposition/larval habitat selection (e.g.,Huang et al. 2005), predation(e.g., Fillinger et al. 2009), andintra- (e.g., Gimnig et al. 2002)and interspecific competition(e.g., Kirby & Lindsay 2009;Paaijmans et al. 2009) usingweather stations/sensors

● Rainfall simulators to studyaquatic vector ecology(Koenraadt and Harrington2008)

● Models for density dependenceon adult and pre-imaginal stages(Rodriguez 1988)

● What types of landscapeare conducive forincreased mosquitoproductivity with changesin rainfall seasonality?

● Is mosquito habitatselection influenced bythe potential forautonomization from achanging environment(Janisch 1932)?

● How important arechanges in mosquitophenology for malariatransmission? What is therole of densitydependence versusclimatic factors herein?

Landscape and vegetation● Mosquitoes feed on

different sources ofnectar, in flowersfrom different plantspecies (Yuval 1992).

● Landscape affectspresence of habitatsand abundance ofmosquito larvae(Fillinger et al. 2004,2009; Minakawa etal. 2002b).

● Gas chromatography for nectarsource identification (e.g., Mandaet al. 2007b)

● Field and laboratory sugarfeeding studies (e.g., Manda et al.2007a)

● Topographic modeling oflandscape to understandmicroclimatic differences(e.g., Cohen et al. 2008)

● Remote sensing and GeographicInformation Systems (e.g., Hayet al. 1998)

● Are mosquito andparasite fitness affectedby nectar source?

● Can the vegetationcomposition around ahousehold bemanipulated to controlmalaria?

● Can remote sensing andgeographical informationsystems be used topredict the presence of(productive) larvalhabitats?

continued

38 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

TABLE 2Continued

Field Current observations Tools/Concepts Key questions/Goals

Mosquito surveillance● Mosquitoes are

difficult to capturein highland areasdue to extremely lowdensities (Koenraadtet al. 2006).

● Mosquito larvalhabitats are dynamic(Minakawa et al.2002b; Fillinger et al.2004).

● Standardized collection methodsacross environmental gradientswith more sensitive traps, tomaximize diversity andprobability of capture (e.g., Njiruet al. 2006; Qiu et al. 2007;Odiere et al. 2007)

● Metapopulation and source-sinkecology (Pulliam 1988)

● Can control efforts betargeted to key habitats?

● Can temporary habitatsbe managed?

● Can we use mathematicalmodels to explainchanges in theentomologicalinoculation rateassociated with mosquitometapopulationdynamics driven byhabitat abundance?

Neglected phenomena Biological interventions● Biological control

remains a largelyunderexploredalternative to usinginsecticides for larvaland adult control. Insome instances, ithas been successfulwhen utilized(Yasuoka and Levins2007a; Scholte et al.2005).

● Misuse ofinsecticides anddrugs can renderthem useless (Shankset al. 2005).

● The use andimplementation ofsuccessful controlstrategies (e.g.,bednets and drugs)is often limited bystrategies that followindustrialized modelsof trading. Thepoorest of the poorretain a high risk ofmalaria infectionbecause of their lackof access to basicresources to stopmalaria transmission(e.g., Mathanga andBowie 2007).

● Community-based, ecologicallysound, and biodiversity-friendlycontrol strategies (Yasuoka et al.2006a,b)

● Health impact assessments (Birley1985): benefits of developmentalstrategies (e.g., dams, agriculture)can be outweighed by theexacerbation or emergence ofdisease transmission.

● Surveys on knowledge, attitudes,and practices to assess theunderstanding of malaria biologyby affected communities, and tounderstand how decisions aremade in regard to choices fortreatment or bednet use (e.g.,Yasuoka et al. 2006a,b)

● More research on the use of socialmarketing and traditional tradingstrategies for drug and bednetdistribution (e.g., Mathanga andBowie 2007)

● Goal: Sustainable, long-term control of malariausing biological controlstrategies. Need for morefunding and research

● Knowledge intensiveagriculture: vector andpest management arebased on a robustknowledge of biologicalinteractions, as opposedto the indiscriminate useof chemical inputs whoselong-term environmentalcosts may be muchhigher than currentbenefits.

● Why do people misusedrugs and insecticides?How does (in)accurateknowledge modifypractices to protect theindividual/communityfrom malariatransmission?

continued

March 2010 39CLIMATE CHANGE AND HIGHLAND MALARIA

TABLE 2Continued

Field Current observations Tools/Concepts Key questions/Goals

Social factors● All malaria transmission

is local (Spielman 2006).Large scale patterns forother vector-bornediseases are determinedby levels of socialexclusion, with largedifferences acrosspopulations determinedby the heterogeneity ofthe natural environment(Chaves et al. 2008b).

● Principles of social epidemiology(Cohen et al. 2007): populationsstratified by income or access toresources; inclusion ofsocioeconomic variables in riskstudies.

● The unequal ecological exchange(Jorgenson 2006): environmentalcosts and disease risks arecanalized into less developedcountries.

● At what spatial scale (i.e.,local, regional, global)are social factors relevantin determining malariarisk?

● How to incorporatehealth andenvironmental costs andecosystem services intoglobal trade?

Evolutionary ecology Drug resistance● Drug resistance could

have evolved during theshift toward endemicmalaria transmission inthe African highlands(e.g., Carter and Mendis2002).

● Quantitative genetics modelingapproach for convergingevolutionary and ecologicaldynamics (Hairston et al. 2005;Khibnik and Kondrashov 1997)

● Is drug resistance adriver for malaria rangeexpansion? How can therisk of selection for drugresistance be minimized?

Vector Fitness● Larger mosquitoes are

more fecund, andsmaller mosquitoesrequire morebloodmeals to completea gonotrophic cycle(Lyimo and Takken1993).

● When food deprived,larger non malariavectors emergefollowing theapplication of controlagents (Wilson et al.1990)

● Intermediate mosquitobody size maximizesparasite fitness undercontrolledenvironmentalconditions (Lyimo andKoella 1992).

● Oviposition choices canimpact non-malariamosquito fitness (e.g.,Ellis 2008).

● Life history theory for trade-offs(Stearns 2000): fasterdevelopment-shorter size, longerdevelopment-larger size

● Metabolic theory (Brown et al.2004) and physiological ecology(Briegel 2003): gonotrophicdiscordance as an evolutionarystrategy; mosquito populationdynamics and vectorial capacityaffected by metabolic processes

● Mosquito oviposition studies atmultiple spatial scales (e.g.,Chaves et al, 2009). Ideal freedistribution as a null hypothesis:no average fitness differencesacross heterogeneous landscapes(Ellis 2008)

● What are theevolutionary effects ofwarming trends on thelife history of parasitesand vectors?

● Does larviciding selectfor more efficientvectors?

● What are the effects ofgonotrophic discordanceon mosquito fitness andvectorial capacity? Whatare the trade-offs?

● Are mosquito ovipositionchoices finely or coarselygrained? Does larvalhabitat heterogeneitymap into phenotypicallydifferent vectors?

continued

40 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

mans. Demographic drivers, housing, andland use changes seem to be the most robustlylinked to patterns of malaria transmission.

demographic driversHuman migration in the context of ma-

laria transmission has been widely studied(Longstreth and Kondrachine 2002; Mar-tens and Hall 2000; Prothero 1965; Sevilla-Casas 1993), and, in the case of the Africanhighlands, it is well-known that the inva-sion of these regions by malarial parasiteshas been associated with the migration ofpeople from the lowlands to the highlands

(Lindsay and Martens 1998; Shanks et al.2000; Shanks et al. 2005). The phenome-non of human migration is tightly linkedto socioeconomic inequity and poverty(Martens and Hall 2000), the developmentof market-based agriculture, and land ten-ure inequity (Celli 1977; Humphreys 2001;Prothero 1965). Although malaria is be-coming a major problem in urban areastoday (Matthys et al. 2006a,b; Sattler et al.2005), it is, as it has traditionally been, mostlya rural disease affecting people living andworking in agricultural areas (Celli 1977;Ernst et al. 2006; Kitron 1987; Wilson 2003;

TABLE 2Continued

Field Current observations Tools/Concepts Key questions/Goals

Vector Longevity● Mortality is age

dependent in non-malaria mosquitoes(Harrington et al. 2008;Styer et al. 2007).

● Vectorial capacity conceptassumes that mosquito mortalityis constant (Garret-Jones 1964)

● How does senescence(i.e., the increase ofmortality with age) affectvectorial capacity?

Ecological interactions● Competition among

sibling species andmolecular chromosomalforms of the mainmalaria vector (An.gambiae) (Diabate et al.2005; Koenraadt et al.2004b; Koenraadt andTakken 2003)

● Oviposition and larvalhabitat selection by An.gambiae is influencedby presence ofconspecifics (Munga etal. 2006)

● Indirect effects ofpredator presence inlarval habitat selectionby molecular forms ofAn. gambiae (Diabate etal. 2008)

● Predation perceptionand resource abundancecan modify phenotypicand life cycle traits ofnon-malaria mosquitoes(Beketov and Liess2007).

● Chromosome banding; nuclear,ribosomal, and microsatellitemolecular markers (Krzywinskiand Besansky 2003)

● Transplantation experiments(Diabate et al. 2008): removinglarvae from original habitats tothose that they do not colonize

● Feedback loops (Levins andSchultz 1996): mosquito densitycan increase if their predators aremore affected by warmertemperatures than their foragingresources.

● Theory for indirect effects andtrait-mediated interactions onfoodweb, disruption of ecologicalinteractions (Werner and Peacor2003): resource rich larvalhabitats may produce less andsmaller mosquitoes because ofpredator presence.

● How does competitionduring the aquatic lifestage affect vectorialcapacity?

● Which vectors willinvade/establish in areasthat undergoenvironmental change?

● How will climate changeimpact mosquitoresources and/or theirpredators?

● Can biological controlreduce vectorial capacity?

● What is the impact ofclimate change on thebiological interactionsinvolving mosquitoes?

March 2010 41CLIMATE CHANGE AND HIGHLAND MALARIA

Yasuoka and Levins 2007b; Yasuoka et al.2006a; Yasuoka et al. 2006b). As clearly pre-sented by Prothero (1965), the commodifi-cation of agriculture in Africa led to a pat-tern of seasonal movement among landlessfarmers, migrating in order to work in com-mercially oriented estates, and this, subse-quently, allowed for the invasion of parasitesin new ecosystems where most of the nativepopulation were not immune to malaria in-fections. The latter is probably a global pat-tern, since the same patterns have been seenin Colombia (Sevilla-Casas 1993) and Af-ghanistan (Rab et al. 2003) as well. However,migrations can have the opposite effect, asexemplified by the cases of Venezuela andthe USA, where urbanization and improvedconditions for living have been associatedwith major declines in malaria (Chaves 2007;Humphreys 2001). Other demographic fac-tors include population growth, the collapseof health services and vector control mea-sures (Carter and Mendis 2002; Hay et al.2002b; Lindsay and Martens 1998), and thedeficient nutritional status of economicallyimpoverished populations (Kiszewski andTeklehaimanot 2004).

housingHaddow (1942) and Garnham (1948)

have already noted that the greater the num-ber of people occupying a house, the moremosquitoes will be attracted into that house.By contrast, the design of houses may reducethe degree of exposure to blood-seekingmosquitoes, as simple construction may actas a physical barrier to prevent mosquitoesfrom entering (Lindsay et al. 2002). Houseswith open eaves, mud rather than stonewalls, and thatched roofing, as well as thoseoccupied by a greater number of people andthose without ceilings or cooking fires, wereall associated with increased vector densities(Adiamah et al. 1993; Lindblade et al. 2000;Lindsay et al. 1995; Palsson et al. 2004; Zhouet al. 2007). The apparent increase in thenumber of mosquito bites also had an effecton morbidity levels, since such housing con-ditions were associated with increased mor-bidity (Adiamah et al. 1993; Ghebreyesus etal. 2000; Koram et al. 1995; Ong’echa et al.2006; Somi et al. 2007; Ye et al. 2006). In

contrast, Mbogo et al. (1999) found no dif-ference in vector abundance between houseswhere children suffered from severe malariaand their control houses, while some hous-ing factors, such as more than six occupantsand the absence of a second bedroom, wereassociated with an increased risk. Other stud-ies did not document any connection be-tween house construction or household sizeand malaria morbidity (Luckner et al. 1998;Snow et al. 1998).

land use change (agriculture anddeforestation)

Landscape transformations can lead tomajor changes in the functioning of eco-systems and their resilience or ability tocope with changes (Holling 1973). Thishas been shown in several cases for ma-laria. For example, Lindblade et al. (2000)showed that differences in temperaturebetween forested and transformed land-scapes (primarily for agriculture) at thesame altitude in East Africa had major ef-fects on mean temperature values, as wellas temperature variability. Such differencesare well-known to affect the ecology of ma-laria vectors in these regions; more specif-ically, these differences have been provento affect the survival and gonotrophic cycleof vectors (Afrane et al. 2005; Afrane et al.2006). In some cases, the difference intemperature can far surpass the increase of0.5 °C attributable to climate change in thelast sixty years (Pascual et al. 2006).

Growing and harvesting crops, either forself-subsistance or for monetary income, isthe main economic activity in rural Africa.Besides their importance as a food source,crops may provide suitable micro-habitats foradult mosquitoes. Some crops also need reg-ular irrigation, thereby creating aquatic hab-itats for the larval stages of malaria vectors.Ever since their introduction, rice irrigationschemes have been associated with a highmalaria incidence in these agricultural com-munities (Grainger 1947). In many areasacross Africa, irrigation seems to increasevector densities; however, the level of trans-mission may increase, remain unchanged, oreven decrease (Ghebreyesus et al. 2000;Githeko et al. 1993; Ijumba et al. 2002;

42 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

Klinkenberg et al. 2005; Lindblade et al.2000; Muturi et al. 2008; Okoye et al. 2005).In their review, Ijumba and Lindsay (2001)concluded that irrigation schemes in Africado not seem to increase malaria risk in areasof stable transmission. Moreover, malariarisk may be reduced in communities withirrigation schemes as a result of species dis-placement (i.e., the opportunistic An. ara-biensis thrives better in rice fields than thehighly anthropophilic An. funestus) andgreater wealth in these communities—a phe-nomenon known as the “paddies paradox.”By contrast, in areas of unstable transmis-sion, irrigation may aggravate the malariasituation. De Plaen et al. (2003) argued that,next to biological changes resulting fromirrigation practices, socioeconomic transfor-mations and gender repositioning are im-portant mediators of change in malaria risk.The practice of intermittent irrigation signif-icantly reduces vector populations, althoughfields should be drained completely, sincethe remaining pools and puddles may pro-vide ideal sites for mosquito breeding(Klinkenberg et al. 2003). The effects of in-termittent irrigation on clinical manifestationsof malaria remain largely unknown (Keiser etal. 2005; Keiser et al. 2002) and, therefore,should be investigated in the future.

The invasion of malaria in the Bure high-lands of Ethiopia (Kebede et al. 2005) waslikely mediated by an increase in the fitnessof Anopheles arabiensis—the mosquito vectorpresent in the area—that resulted from itsfeeding on maize pollen after this crop wasintroduced to the area (Ye-Ebiyo et al. 2003;Ye-Ebiyo et al. 2000). However, this patternwas very localized, which led to the questionof why it was not a problem in other placeswhere maize had been introduced. But cropsare not the only factors that can affect mos-quito ecology and malaria transmission;other farming practices can lead to the exac-erbation of malaria transmission as well. Forexample, the introduction of fish ponds hasbeen associated with the establishment ofAn. funestus—a species whose ecology is asso-ciated with large ponds that can ultimatelylengthen the transmission season—in areaswhere the species was never present before(Lockhart et al. 1969). The effects of agricul-

ture on mosquito ecology and the transmissionof malaria seem to be wider than commonlyacknowledged. In a recent review, Yasuokaand Levins (2007b) presented the variety ofeffects that both deforestation and shiftingagricultural practices have had on the ecol-ogy of mosquitoes vectoring malaria, findinga positive association between practices thatpromote the presence of water pools withaccess to light. Their work also showed howmajor landscape changes for the cultivationof agricultural crops—from coffee and cacaoto rice, maize, and cassava—have alwaysbeen associated with increased densities ofmosquitoes and malaria outbreaks. There-fore, the links between agriculture and ma-laria are probably the most robust and theones deserving more detailed research.

Agricultural practices related to the herdingand farming of animals also impact malariathrough the alteration of the community ofhosts that could potentially serve as blood-feeding sources for mosquitoes. The pres-ence of alternative hosts on the abundanceof malaria vectors, and hence the risk of ma-laria, has received much attention, since theuse of animals to divert host-seeking mosqui-toes away from humans has been suggestedas a measure to control malaria (zooprophy-laxis) (WHO 1982). In many African set-tings, the presence of animals did not affectvector densities, entomological inoculationrates, or morbidity levels (Bøgh et al. 2001;Lindblade et al. 2000; Minakawa et al. 2002a;Snow et al. 1998). Moreover, in Gambia andEthiopia, the malaria risk was greater inhouseholds that had many animals in oraround their house (Adiamah et al. 1993;Deressa et al. 2007; Ghebreyesus et al. 2000).In a study on the coast of Kenya, the pres-ence of more than two sheep was associatedwith an increased risk of severe malaria,while the presence of more than one dogwas associated with a decreased risk (Mbogoet al. 1999). It can be concluded that, in theworst case scenario, the presence of animalsprovides an additional risk of malaria andsurely does not divert host-seeking mosquitoesaway from people vulnerable to the disease,although this may depend on the compositionof the local sibling species (Mahande et al.2007). However, it should also be mentioned

March 2010 43CLIMATE CHANGE AND HIGHLAND MALARIA

that in some settings outside Africa, the pres-ence of domestic animals does have a protec-tive effect (Charlwood 2001).

The interaction between agriculture andmalaria has also produced other patternsworthy of detailed study, specifically con-cerning how the disease can interfere withstrategies for social development. The bestdocumented study on this issue is historicaland deals with inequity in land tenure in theAgro Romano, Italy (Celli 1977). Endemicmalaria in this region led to the continuousabandonment or selling of land by smallfarmers, promoting the development of lati-fundia, or concentration of land with fewowners. The latter occurred following nu-merous land redistribution reforms by differ-ent governments throughout the history ofthe Agro Romano, from classical times to the20th century. Only after malaria was eradicateddid latifundia and chronic poverty became athing of the past in the Roman countryside(Celli 1977; Najera 1994). This pattern hasbeen found in other places, as illustrated bya positive association between latifundia andmalaria endemism in Spain (Beauchamp1988) and the southeastern United States(Humphreys 2001) in the 1930s. This pat-tern also seems to be a global one, sincemalaria became endemic in several regionsof the world only after these places experi-enced major deforestation in the nameof economic development. Among theseregions are the archipelago of the Masca-reignes (modern day Mauritius and Re-union) (Julvez et al. 1990), Sao Tome andPrincipe (Baptista 1996), and the Amazonicregion of Peru (Vittor et al. 2006).

Thus, studies of land tenure dynamics andthe effects of the historical disturbances thatpromoted its inequity (such as colonialism[Phombeah 2005]) can inform efforts tocontrol malaria, and can also encourage itscontrol as a way of promoting social changein zones where the disease is a burden (Celli1977). This fact is further reinforced if theproblem is viewed from a broader perspec-tive and attention is given to conservation offorests and their associated biodiversity. Aspreviously mentioned, deforestation alterslandscape quality, which is of fundamentalimportance for biodiversity conservation. De-

forestation diminishes the dispersal ability ofspecies, thus increasing their extinction risk(Perfecto and Vandermeer 2008). It hasbeen shown that large scale latifundia anddisparity in land property size (Fearnside1993), large external debts (Bawa and Day-anandan 1997), and precarious conditionsfor human social development (Jha andBawa 2006), as well as unequal ecologicalexchange where more developed countriesexternalize their consumption-based envi-ronmental costs to less developed countries,thereby resulting in the environmental deg-radation of the latter (Jorgenson 2006), aresome of the important underlying causes ofdeforestation. In general, the primary out-come of deforestation in terms of malariatransmission is an increase in its risk (deCastro et al. 2006; Guerra et al. 2006; Vittoret al. 2006). Therefore, understanding theeffects of deforestation on malaria transmis-sion will necessarily reveal that this is a prob-lem whose roots are not in the nature ofbiological interactions across species, butrather in the implementation of models forsocial development and economic growth, asalready shown for other vector-borne dis-eases (Chaves et al. 2008a). Table 2 presentsdirections for future research into the ne-glected aspects of malaria transmission.

Evolutionary Change at DifferentTime Scales

Ivan I. Schmalhausen (1949), in his fascinat-ing but largely unnoticed work on evolution,clearly implied that organisms coping withchanges in their environment become morevulnerable to small changes when pushed to-wards the limits of tolerance in any dimensionof their existence (Schmalhausen 1949). Thissimple realization, which has been calledSchmalhausen’s law (Awerbuch et al. 2002;Chaves et al. 2008a,b), has major implica-tions for all of ecology since it embodies theunderlying evolutionary nature of the majorecological patterns seen in nature. It predictsthat with a changing climate, changes in thedynamics of malaria (or any other biologicalphenomena) are to be expected in geo-graphical regions at the edge of the distribu-tion of the disease (or, more generally, aspecies). This fact is so universal that even

44 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

early malaria studies by Macdonald (1953)realized that climatic variation has its largesteffects on “the margins of the distribution ofmalaria whether in tropical or cooler zones”(p. 882). Today, major changes are beingseen in the dynamics of malaria in the high-lands of East Africa—a zone within the mar-gins of what used to be the distribution ofmalaria (Lindsay and Birley 2004; Lindsayand Birley 1996; Lindsay and Martens 1998).As we discussed earlier, a multidimensionalarray of underlying factors is likely to be atplay here, most of which may be sensitive toclimatic change, whereas others may be un-dergoing evolutionary change. The three bi-ological entities involved in malaria—humans, mosquitoes, and parasites—are verydifferent, as are their effects on ecosystems,and possibly the way in which they can beaffected by global climate change as well.The pressures imposed by such change candisplay evolutionary outcomes at differenttime scales and can have effects at levelsranging from the genetic to the cultural andsocial.

The effects of malaria on the populationgenetics of humans are well-known, datingback to the founders of population genetics(Haldane 1949) and biological anthropology(Livingstone 1958). These effects are mainlyshown by the maintenance of deleterious al-leles associated with advantages in copingwith malaria, with the classical example be-ing the heterozygote advantage for the sickle-cell anemia allele (Lewontin 1974). Sinceevolutionary change is known to occur as afunction of the number of generations un-der natural selection or neutral evolution, itis likely that climate change will have littleimmediate effect on the ways in which hu-mans evolve genetic means of defense, as thehuman pattern of evolution is generally as-sociated with long-term agricultural practices(Odling-Smee et al. 2003). However, thismay not be the case for mosquitoes and par-asites, where developmental—and thereforegenerational—times are known to be re-duced with rising temperatures (Pascual etal. 2006; Patz and Olson 2006), thereby in-creasing the likelihood of evolutionarychanges in these two biological componentsof malaria as compared to the possibility of

these changes in humans (using a commontime scale). Such effects of natural selectionhave already been seen in the developmentof insecticide and drug resistance in mosqui-toes and parasites, respectively (Carter andMendis 2002; Shanks et al. 2005). The oddsfor neutral evolution (i.e., random geneticdrift) are also likely to be increased, becausean increased environmental variability isknown to reduce the effective populationsize (Mueller and Joshi 2000), thus resultingin a greater level of uncertainty about thedirection of any evolutionary change.

Schmalhausen’s law also emphasizes theinherent multidimensional nature of all bio-logical phenomena, but little attention hasbeen given to other aspects of the biology ofmalaria. For example, predators and com-petitors can impact the distribution andabundance of mosquitoes (Blaustein andChase 2007; Chase and Knight 2003; Knightet al. 2004), and, as argued before, theseorganisms can also evolve or even co-evolve,as evolutionary changes can happen atshorter temporal scales than commonly rec-ognized among interacting populations. Ta-ble 2 presents directions for future researchin the evolutionary ecology of malaria.

Schmalhausen’s Law and Diseases inChanging Environments

The multidimensionality of malaria alsocalls for the realization that human biologyis a socialized biology (Levins 1995). As wediscussed earlier, a major underlying forcein the spread of malaria, and even in thegenetic structure of human populations,has been agriculture—a unique feature ofthe human species, deeply entangled withmalaria historical dynamics in societies.For malaria, examples supporting its social-ized nature are abundant: children of thepoorest households are more likely to suf-fer malaria (Clarke et al. 2001); malariatransmission risk is reduced by a factor of25 in countries with good health services(e.g., those able to provide prompt treat-ment) (Bouma 2003); educated communi-ties are better at managing malaria risk factors,especially by reducing mosquito sources with-out compromising agricultural productivity(Yasuoka et al. 2006b); and only when robust

March 2010 45CLIMATE CHANGE AND HIGHLAND MALARIA

structures of socioeconomic development havebeen present has malaria control been effec-tive in the long run (Celli 1977; Kitron 1987;Lindsay and Birley 2004). Thus, only the in-tegration of knowledge from the variousfields discussed in this paper will provide newinsights into the biology of malaria. To thatend, Schmalhausen’s law provides a concep-tual framework within which the importanceof different factors for malaria risk is evalu-ated at different spatial and temporal scales.The sensitivity to each factor at each scalecan then be assessed, and interventions canbe planned and coordinated in such a waythat global, regional, and local scale actionsare in concert, and opposing effects from agiven driver can be understood as part of thesame whole. With this framework, expecta-tions regarding the time scales at which re-sults are to be evaluated will be well–defined,and the goals of malaria eradication or sus-tained suppression will be more likely to beachieved (Feachem and Sabot 2008; Spiel-man et al. 1993).

Finally, global warming plays a major rolein the collapse of ecosystems as functioningwholes (Collier and Webb 2002; Scheffer etal. 2001). However, areas where non-climaticfactors have eliminated or controlled ma-

laria are likely, ceteris paribus (i.e., everythingelse being equal), to be insensitive to theeffects of global warming on disease trans-mission by vectors, while other areas wheremalaria is not present because of climaticconditions (e.g., cities in the highlands of thedeveloping world, especially in Africa) are atpotential risk of having an increased burdenof the disease if no concerted action is putforward to stop global warming. In addition,the socialized biology of humans reveals thatefforts to cope with malaria in a changingenvironment that is defined beyond scenar-ios of global warming should be aimed atpromoting socioeconomic changes in the al-ready disease-stricken populations aroundthe world. Schmalhausen’s law predicts thatbetter conditions in people’s lives can com-pensate for the effects of changes in otherelements of the environment that may causedisease transmission to occur.

acknowledgments

Partial funding for this paper was provided by a Gor-gas Research Award from the American Society ofTropical Medicine and Hygiene to LFC. The authorsthank Mark L. Wilson, Mercedes Pascual, and WillemTakken for insightful discussions. Constructive com-ments were made by the editor and anonymous re-viewers. Thank you.

REFERENCES

Abeku T. A., De Vlas S. J., Boorsboom G. J. J. M., TadegeA., Gebreyesus Y., Gebreyohannes H., Alamirew D.,Seifu A., Nagelkerke N. J. D., Habbema J. D. F.2004. Effects of meteorological factors on epi-demic malaria in Ethiopia: a statistical modellingapproach based on theoretical reasoning. Parasi-tology 128:585–593.

Adiamah J. H., Koram K. A., Thomson M. C., LindsayS. W., Todd J., Greenwood B. M. 1993. Entomolog-ical risk factors for severe malaria in a peri-urbanarea of The Gambia. Annals of Tropical Medicine andParasitology 87(5):491–500.

Afrane Y. A., Lawson B. W., Githeko A. K., Yan G. Y.2005. Effects of microclimatic changes caused byland use and land cover on duration of gonotro-phic cycles of Anopheles gambiae (Diptera: Culici-dae) in western Kenya highlands. Journal ofMedical Entomology 42(6):974–980.

Afrane Y. A., Zhou G. F., Lawson B. W., Githeko A. K.,Yan G. Y. 2006. Effects of microclimatic changescaused by deforestation on the survivorship and

reproductive fitness of Anopheles gambiae in West-ern Kenya highlands. American Journal of TropicalMedicine and Hygiene 74(5):772–778.

Andrewartha H. G. 1957. The use of conceptual mod-els in population ecology. Cold Spring Harbor Sym-posium in Quantitative Biology 22:219–236.

Anthony R. L., Bangs M. J., Hamzah N., PurnomoH. B., Subianto B. 1992. Heightened transmissionof stable malaria in an isolated population in thehighlands of Irian Jaya, Indonesia. American Jour-nal of Tropical Medicine and Hygiene 47(3):346–356.

Awerbuch T., Kiszewski A. E., Levins R. 2002. Sur-prise, nonlinearity and complex behaviour. Pages96–119 in Environmental Change, Climate andHealth: Issues and Research Methods, edited by P.Martens and A. J. McMichael. Cambridge (UK):Cambridge University Press.

Bangs M. J., Subianto D. B. 1999. El Nino and associ-ated outbreaks of severe malaria in highland pop-ulations in Irian Jaya, Indonesia: a review andepidemiological perspective. Southeast Asian Jour-

46 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

nal of Tropical Medicine and Public Health 30(4):608–619.

Baptista J. L. 1996. Subsidio para a historia do palud-ismo em S. Tome. Acta Medica Portuguesa 9:259–265.

Bawa K. S., Dayanandan S. 1997. Socioeconomic fac-tors and tropical deforestation. Nature 386:562–563.

Bayoh M. N., Lindsay S. W. 2003. Effect of tempera-ture on the development of the aquatic stages ofAnopheles gambiae sensu stricto (Diptera: Culici-dae). Bulletin of Entomological Research 93:375–381.

Beauchamp C. 1988. Fievres d’hier, paludismed’aujourd’hui: vie et mort d’une maladie. AnnalesESC 43:249–275.

Beier J. C., Killeen G. F., Githure J. I. 1999. Shortreport: entomologic inoculation rates and Plasmo-dium falciparum malaria prevalence in Africa. Amer-ican Journal of Tropical Medicine and Hygiene 61(1):109–113.

Beketov M. A., Liess M. 2007. Predation risk percep-tion and food scarcity induce alterations of life-cycle traits of the mosquito Culex pipiens. EcologicalEntomology 32:405–410.

Binka F. N., Morris S. S., Ross D. A., Arthur P., Ary-eetey M. E. 1994. Patterns of malaria morbidityand mortality in children in northern Ghana.Transactions of the Royal Society of Tropical Medicineand Hygiene 88(4):381–385.

Birch L. C. 1957. The role of weather in determiningthe distribution and abundance of animals. ColdSpring Harbor Symposium in Quantitative Biology 22:203–218.

Birley M. H. 1985. Forecasting the vector-borne dis-ease implications of water development. Parasitol-ogy Today 1:34–36.

Blaustein L., Chase J. M. 2007. Interactions betweenmosquito larvae and species that share the sametrophic level. Annual Review Entomology 52:489–507.

Bøgh C., Clarke S. E., Pinder M., Sanyang F., LindsayS. W. 2001. Effect of passive zooprophylaxis onmalaria transmission in the Gambia. Journal ofMedical Entomology 38(6):822–828.

Bøgh C., Lindsay S. W., Clarke S. E., Dean A., JawaraM., Pinder M., Thomas C. J. 2007. High spatialresolution mapping of malaria transmission risk inThe Gambia, West Africa, using LANDSAT TMsatellite imagery. American Journal of Tropical Med-icine and Hygiene 76(5):875–881.

Bouma M. J. 2003. Methodological problems andamendments to demonstrate effects of tempera-ture on the epidemiology of malaria. A newperspective on the highland epidemics in Mada-gascar, 1972–1989. Transactions of the Royal Societyof Tropical Medicine and Hygiene 97(2):133–139.

Bouma M. J., Dye C., van der Kaay H. J. 1996. Falcip-

arum malaria and climate change in the northwestfrontier province of Pakistan. American Journal ofTropical Medicine and Hygiene 55(2):131–137.

Briegel H. 2003. Physiological bases of mosquito ecol-ogy. Journal of Vector Ecology 28(1):1–11.

Brockwell P. J., Davis R. A. 2002. Introduction to TimeSeries and Forecasting. Second Edition. New York:Springer.

Brown J. H., Gillooly J. F., Allen A. P., Savage V. M.,West G. B. 2004. Toward a metabolic theory ofecology. Ecology 85(7):1771–1789.

Bryan J. H., Foley D. H., Sutherst R. W. 1996. Malariatransmission and climate change in Autralia. Med-ical Journal of Australia 164(6):345–347.

Carter R., Mendis K. N. 2002. Evolutionary and his-torical aspects of the burden of malaria. ClinicalMicrobiology Reviews 15(4):564–594.

Celli A. 1977. The History of Malaria in the RomanCampagna from Ancient Times. New York: AMSPress.

Charlwood D. 2001. Zooprophylaxis: are we in Plato’scave? Trends in Parasitology 17(11):517.

Chase J. M., Knight T. M. 2003. Drought-inducedmosquito outbreaks in wetlands. Ecology Letters6(11):1017–1024.

Chaves L. F. 2007. Casas Muertas and Oficina No. 1:internal migrations and malaria trends in Venezu-ela 1905–1945. Parasitology Research 101(1):19–23.

Chaves L. F., Cohen J. M., Pascual M., Wilson M. L.2008a. Social exclusion modifies climate and de-forestation impacts on a vector-borne disease. Pub-lic Library of Science Neglected Tropical Diseases 2(2):e176.

Chaves L. F., Kaneko A., Taleo G., Pascual M., WilsonM. L. 2008b. Malaria transmission pattern resilienceto climatic variability is mediated by insecticide-treated nets. Malaria Journal 7:100.

Chaves L. F., Keogh C. L., Vazquez-Prokopec G. M.,Kitron U. D. 2009. Combined sewage overflowenhances oviposition of Culex quinquefasciatus Say(Diptera: Culicidae) in urban areas. Journal of Med-ical Entomology 46(2):220–226.

Clarke S. E., Bøgh C., Brown R. C., Pinder M., Wal-raven G. E. L., Lindsay S. W. 2001. Do untreatedbed nets protect against malaria? Transactions of theRoyal Society of Tropical Medicine and Hygiene 95(5):457–462.

Clarke S. E., Bøgh C., Brown R. C., Walraven G. E. L.,Thomas C. J., Lindsay S. W. 2002. Risk of malariaattacks in Gambian children is greater away frommalaria vector breeding sites. Transactions of theRoyal Society of Tropical Medicine and Hygiene 96(5):499–506.

Clements A. N. 1999. The Biology of Mosquitoes, Volume2: Sensory Reception and Behaviour. Wallingford(UK): CABI Publishing.

Cohen J. M., Ernst K. C., Lindblade K. A., Vulule J. M.,

March 2010 47CLIMATE CHANGE AND HIGHLAND MALARIA

John C. C., Wilson M. L. 2008. Topography-derivedwetness indices are associated with household-levelmalaria risk in two communities in the westernKenyan highlands. Malaria Journal 7:40.

Cohen J. M., Wilson M. L., Aiello A. E. 2007. Analysisof social epidemiology research on infectious dis-eases: historical patterns and future opportunities.Journal of Epidemiology and Community Health61(12):1021–1027.

Collier M., Webb R. H. 2002. Floods, Droughts, andClimate Change. Tucson (AZ): University of Ari-zona Press.

Coz J. 1973. Les mecanismes d’isolement genetiquedans le complexe Anopheles gambiae Giles. CahORSTOM ser Entomol med Parasitol 11:41–56.

Craig M. H., Kleinschmidt I., Nawn J. B., le Sueur D.,Sharp B. L. 2004. Exploring 30 years of malariacase data in KwaZulu-Natal, South Africa: Part I.The impact of climatic factors. Tropical Medicineand International Health 9(12):1247–1257.

Craig M. H., Snow R. W., le Sueur D. 1999. A climate-based distribution model of malaria transmission insub-Saharan Africa. Parasitology Today 15(3):105–111.

de Castro M. C., Monte-Mor R. L., Sawyer D. O.,Singer B. H. 2006. Malaria risk on the Amazonfrontier. Proceedings of the National Academy of Sci-ences USA 103(7):2452–2457.

della Torre A., Tu Z. J., Petrarca V. 2005. On thedistribution and genetic differentiation of Anoph-eles gambiae s.s. molecular forms. Insect Biochemistryand Molecular Biology 35(7):755–769.

Delmont J. 1982. Paludisme et variations climatiquessaisonnieres en savane soudanienne d’Afrique del’Ouest. Cahiers d’Etudes Africaines 22(85–86):117–133.

De Plaen R., Geneau R., Teuscher T., Koutoua A.,Seka M. L. 2003. Living in the paddies: a socialscience perspective on how inland valley irrigatedrice cultivation affects malaria in Northern Coted’Ivoire. Tropical Medicine & International Health8:459–470.

Deressa W., Ali A., Berhane Y. 2007. Household andsocioeconomic factors associated with childhoodfebrile illnesses and treatment seeking behaviourin an area of epidemic malaria in rural Ethiopia.Transactions of the Royal Society of Tropical Medicineand Hygiene 101(9):939–947.

de Zulueta J. 1987. Changes in the geographical dis-tribution of malaria throughout history. Parassito-logia 29:193–205.

de Zulueta J. 1994. Malaria and ecosystems: fromprehistory to posteradication. Parassitologia 36(1–2):7–15.

Diabate A., Dabire R., Heidenberger K., Crawford J.,Lamp W. O., Culler L. E., Lehmann T. 2008. Evi-dence for divergent selection between the molec-

ular forms of Anopheles gambiae: role of predation.BMC Evolutionary Biology 8:5.

Diabate A., Dabire R. K., Kim E. H., Dalton R., Mill-ogo N., Baldet T., Simard F., Gimnig J. E., HawleyW. A., Lehmann T. 2005. Larval development ofthe molecular forms of Anopheles gambiae (Diptera:Culicidae) in different habitats: a transplantationexperiment. Journal of Medical Entomology 42:548–553.

Drakeley C. J., Carneiro I., Reyburn H., Malima R.,Lusingu J. P. A., Cox J., Theander T. G., NkyaW. M. M. M., Lemnge M. M., Riley E. M. 2005.Altitude-dependent and -independent variationsin Plasmodium falciparum prevalence in northeast-ern Tanzania. Journal of Infectious Diseases 191(10):1589–1598.

Duncan K. 1993. The possible influence of climate onhistorical outbreaks of malaria in Scotland. Pro-ceedings of the Royal College of Physicians Edinburgh23(1):55–62.

Ellis A. M. 2008. Incorporating density dependenceinto the oviposition preference-offspring perfor-mance hypothesis. Journal of Animal Ecology 77(2):247–256.

Ernst K. C., Adoka S. O., Kowuor D. O., Wilson M. L.,John C. C. 2006. Malaria hotspot areas in a high-land Kenya site are consistent in epidemic andnon-epidemic years and are associated with eco-logical factors. Malaria Journal 5:78.

Faraway J. J. 2005. Linear Models with R. Boca Raton(FL): CRC Press.

Feachem R., Sabot O. 2008. A new global malariaeradication strategy. Lancet 371(9624):1633–1635.

Fearnside P. M. 1993. Deforestation in Brazilian Ama-zonia: the effect of population and land tenure.Ambio 22:537–545.

Fillinger U., Sombroek H., Majambere S., van LoonE., Takken W., Lindsay S. W. 2009. Identifying themost productive breeding sites for malaria mos-quitoes in The Gambia. Malaria Journal 8:62.

Fillinger U., Sonye G., Killeen G. F., Knols B. G. J.,Becker N. 2004. The practical importance of per-manent and semipermanent habitats for control-ling aquatic stages of Anopheles gambiae sensu latomosquitoes: operational observations from a ruraltown in western Kenya. Tropical Medicine & Inter-national Health 9(12):1274–1289.

Gagnon A. S., Smoyer-Tomic K. E., Bush A. B. G.2002. The El Nino southern oscillation and ma-laria epidemics in South America. InternationalJournal of Biometeorology 46(2):81–89.

Garnham P. C. C. 1929. Malaria in Kisumu, Kenyacolony. Journal of Tropical Medicine and Hygiene 32:207–216.

Garnham P. C. C. 1948. The incidence of malaria athigh altitudes. Journal of the National Malaria Society7(4):275–284.

48 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

Garrett-Jones C. 1964. The human blood index ofmalaria vectors in relation to epidemiological as-sessment. Bulletin of the World Health Organization30(2):241–261.

Ghebreyesus T. A., Haile M., Witten K. H., GetachewA., Yohannes M., Lindsay S. W., Byass P. 2000.Household risk factors for malaria among chil-dren in the Ethiopian highlands. Transactions of theRoyal Society of Tropical Medicine and Hygiene 94(1):17–21.

Ghebreyesus T. A., Haile M., Witten K. H., GetachewA., Yohannes A. M., Yohannes M., TeklehaimanotH. D., Lindsay S. W., Byass P. 1999. Incidence ofmalaria among children living near dams innorthern Ethiopia: community based incidencesurvey. British Medical Journal 319(7211):663–666.

Gill C. A. 1920. The relationship between malaria andrainfall. Indian Journal of Medical Research 37:618–632.

Gillies M. T., Coetzee M. 1987. A supplement to theanophelinae of Africa south of the Sahara. Publi-cations of the South African Institute for Medical Re-search 55:79–86.

Gimnig J. E., Ombok M., Otieno S., Kaufman M. G.,Vulule J. M., Walker E. D. 2002. Density-dependentdevelopment of Anopheles gambiae (Diptera: Culici-dae) larvae in artificial habitats. Journal of MedicalEntomology 39(1):162–172.

Githeko A. K., Ndegwa W. 2001. Predicting malariaepidemics in the Kenyan highlands using climatedata: a tool for decision makers. Global Change andHuman Health 2:54–63.

Githeko A. K., Service M. W., Mbogo C. M., AtieliF. K., Juma F. O. 1993. Plasmodium falciparumsporozoite and entomological inoculation rates atthe Ahero rice irrigation scheme and the Miwanisugar-belt in western Kenya. Annals of Tropical Med-icine and Parasitology 87:379–391.

Gomez-Elipe A., Otero A., van Herp M., Aguirre-Jaime A. 2007. Forecasting malaria incidencebased on monthly case reports and environmentalfactors in Karuzi, Burundi, 1997-2003. MalariaJournal 6:129.

Grainger W. E. 1947. The experimental control ofmosquito breeding in rice fields in Nyanza Prov-ince, Kenya, by intermittent irrigation and othermethods. East African Medical Journal 24:16–22.

Guerra C. A., Snow R. W., Hay S. I. 2006. A globalassessment of closed forests, deforestation and ma-laria risk. Annals of Tropical Medicine and Parasitol-ogy 100(3):189–204.

Haddow A. J. 1942. The mosquito fauna and climateof native huts at Kisumu, Kenya. Bulletin of Ento-mological Research 33:91–142.

Hairston N. G., Jr., Ellner S. P., Geber M. A., YoshidaT., Fox J. A. 2005. Rapid evolution and the con-

vergence of ecological and evolutionary time. Ecol-ogy Letters 8:1114–1127.

Haldane J. B. S. 1949. Disease and evolution. LaRicerca Scientifica 19(supplement):68–76.

Harrington L. C., Vermeylen F., Jones J. J., KitthaweeS., Sithiprasasna R., Edman J. D., Scott T. W. 2008.Age-dependent survival of the dengue vector Aedesaegypti (Diptera: Culicidae) demonstrated by si-multaneous release-recapture of different age co-horts. Journal of Medical Entomology 45(2):307–313.

Hay S. I., Cox J., Rogers D. J., Randolph S. E., SternD. I., Shanks G. D., Myers M. F., Snow R. W. 2002a.Climate change and the resurgence of malaria inthe East African highlands. Nature 415(6874):905–909.

Hay S. I., Myers M. F., Burke D. S., Vaughn D. W.,Endy T., Ananda N., Shanks G. D., Snow R. W.,Rogers D. J. 2000. Etiology of interepidemic peri-ods of mosquito-borne disease. Proceedings of theNational Academy of Sciences USA 97(16):9335–9339.

Hay S. I., Rogers D. J., Randolph S. E., Stern D. I., CoxJ., Shanks G. D., Snow R. W. 2002b. Hot topic orhot air? Climate change and malaria resurgencein East African highlands. Trends in Parasitology18(12):530–534.

Hay S. I., Shanks G. D., Stern D. I., Snow R. W.,Randolph S. E., Rogers D. J. 2005. Climate vari-ability and malaria epidemics in the highlands ofEast Africa. Trends in Parasitology 21(2):52–53.

Hay S. I., Snow R. W., Rogers D. J. 1998. Predictingmalaria seasons in Kenya using multitemporal me-teorological satellite sensor data. Transactions of theRoyal Society of Tropical Medicine and Hygiene 92(1):12–20.

Hernandez M. J. 1998. Dynamics of transition be-tween population interactions: a nonlinear inter-action �-function defined. Proceedings of the RoyalSociety of London, Series B: Biological Sciences 265(1404):1433–1440.

Holling C. S. 1973. Resilience and stability of ecolog-ical systems. Annual Review of Ecology and Systematics4:1–23.

Holstein M. H. 1954. Biology of Anopheles gambiae:Research in French West Africa. Geneva (Switzer-land): World Health Organization.

Hoshen M. B., Morse A. P. 2004. A weather-drivenmodel of malaria transmission. Malaria Journal3:32.

Huang J., Walker E. D., Giroux P. Y., Vulule J., MillerJ. R. 2005. Ovipositional site selection by Anophelesgambiae: influences of substrate moisture and tex-ture. Medical and Veterinary Entomology 19:442–450.

Humphreys M. 2001. Malaria: Poverty, Race, and PublicHealth in the United States. Baltimore (MD): JohnsHopkins University Press.

Ijumba J. N., Lindsay S. W. 2001. Impact of irrigation

March 2010 49CLIMATE CHANGE AND HIGHLAND MALARIA

on malaria in Africa: paddies paradox. Medical andVeterinary Entomology 15(1):1–11.

Ijumba J. N., Mosha F. W., Lindsay S. W. 2002. Malariatransmission risk variations derived from differentagricultural practices in an irrigated area of north-ern Tanzania. Medical and Veterinary Entomology16(1):28–38.

Janisch E. 1932. The influence of temperature on thelife history of insects. Transactions of the Entomolog-ical Society of London 80:137–168.

Jepson W. F., Moutia A., Courtois C. 1947. The ma-laria problem in Mauritius: the bionomics of Mau-ritian anophelines. Bulletin of Entomological Research38:177–208.

Jha S., Bawa K. S. 2006. Population growth, humandevelopment, and deforestation in biodiversityhotspots. Conservation Biology 20(3):906–912.

Jones C. O. H., Williams H. A. 2004. The social bur-den of malaria: what are we measuring? AmericanJournal of Tropical Medicine and Hygiene 71(2 sup-plement):156–161.

Jorgenson A. K. 2006. Unequal ecological exchangeand environmental degradation: a theoreticalproposition and cross-national study of deforesta-tion, 1990-2000. Rural Sociology 71:685–712.

Julvez J., Mouchet J., Ragavoodoo C. 1990. Epidemi-ologie historique du paludisme dans l’archipel desMascareignes (Ocean Indien). Annales Societe Belgede Medecine Tropicale 70:249–262.

Kazembe L. N. 2007. Spatial modelling and risk fac-tors of malaria incidence in northern Malawi. ActaTropica 102(2):126–137.

Kebede A., McCann J. C., Ye-Ebiyo Y., Kiszewski A. E.2005. New evidence of the effects of agro-ecologicchange on malaria transmission. American Journalof Tropical Medicine and Hygiene 73(4):676–680.

Keiser J., De Castro M. C., Maltese M. F., Bos R.,Tanner M., Singer B. H., Utzinger J. 2005. Effectof irrigation and large dams on the burden ofmalaria on a global and regional scale. AmericanJournal of Tropical Medicine and Hygiene 72(4):392–406.

Keiser J., Utzinger J., Singer B. H. 2002. The potentialof intermittent irrigation for increasing rice yields,lowering water consumption, reducing methaneemissions, and controlling malaria in African ricefields. Journal of the American Mosquito Control Asso-ciation 18(4):329–340.

Khibnik A. I., Kondrashov A. S. 1997. Three mecha-nisms of Red Queen dynamics. Proceedings of theRoyal Society of London, Series B: Biological Sciences264(1384):1049–1056.

Killeen G. F., Fillinger U., Kiche I., Gouagna L. C.,Knols B. G. J. 2002. Eradication of Anopheles gam-biae from Brazil: lessons for malaria control inAfrica? Lancet Infectious Diseases 2(10):618–627.

Killian A. H. D., Langi P., Talisuna A., Kabagambe G.

1999. Rainfall pattern, El Nino and malaria inUganda. Transactions of the Royal Society of TropicalMedicine and Hygiene 93:22–23.

Kirby M. J., Lindsay S. W. 2004. Responses of adultmosquitoes of two sibling species, Anopheles ara-biensis and A. gambiae s.s. (Diptera: Culicidae), tohigh temperatures. Bulletin of Entomological Re-search 94(5):441–448.

Kirby M. J., Lindsay S. W. 2009. Effect of temperatureand inter-specific competition on the develop-ment and survival of Anopheles gambiae sensustricto and An. arabiensis larvae. Acta Tropica109(2):118–123.

Kiszewski A. E., Teklehaimanot A. 2004. A review ofthe clinical and epidemiologic burdens of epi-demic malaria. American Journal of Tropical Medicineand Hygiene 71(supplement 2):128–135.

Kitron U. 1987. Malaria, agriculture, and develop-ment: lessons from past campaigns. InternationalJournal of Health Services 17(2):295–326.

Kleinschmidt I., Sharp B. L., Clarke G. P. Y., Curtis B.,Fraser C. 2001. Use of generalized linear mixedmodels in the spatial analysis of small-area malariaincidence rates in KwaZulu Natal, South Africa.American Journal of Epidemiology 153(12):1213–1221.

Klinkenberg E., McCall P. J., Hastings I. M., WilsonM. D., Amerasinghe F. P., Donnelly M. J. 2005.Malaria and irrigated crops, Accra, Ghana. Emerg-ing Infectious Diseases 11(8):1290–1293.

Klinkenberg E., Takken W., Huibers F., Toure Y. T.2003. The phenology of malaria mosquitoes in irri-gated rice fields in Mali. Acta Tropica 85(1):71–82.

Knight T. M., Chase J. M., Goss C. W., Knight J. J. 2004.Effects of interspecific competition, predation, andtheir interaction on survival and development timeof immature Anopheles quadrimaculatus. Journal of Vec-tor Ecology 29(2):277–284.

Koenraadt C. J. M., Githeko A. K., Takken W. 2004a.The effects of rainfall and evapotranspiration onthe temporal dynamics of Anopheles gambiae s.s.and Anopheles arabiensis in a Kenyan village. ActaTropica 90(2):141–153.

Koenraadt C. J. M., Harrington L. C. 2008. Flushingeffect of rain on container-inhabiting mosquitoesAedes aegypti and Culex pipiens (Diptera: Culicidae).Journal of Medical Entomology 45(1):28–35.

Koenraadt C. J. M., Majambere S., Hemerik L.,Takken W. 2004b. The effects of food and spaceon the occurrence of cannibalism and predationamong larvae of Anopheles gambiae sl. EntomologiaExperimentalis et Applicata 112(2):125–134.

Koenraadt C. J. M., Paaijmans K. P., Schneider P.,Githeko A. K., Takken W. 2006. Low larval vectorsurvival explains unstable malaria in the westernKenya highlands. Tropical Medicine & InternationalHealth 11(8):1195–1205.

Koenraadt C. J. M., Takken W. 2003. Cannibalism and

50 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

predation among larvae of the Anopheles gambiaecomplex. Medical and Veterinary Entomology 17(1):61–66.

Koram K. A., Bennett S., Adiamah J. H., GreenwoodB. M. 1995. Socio-economic risk factors for ma-laria in a peri-urban area of The Gambia. Trans-actions of the Royal Society of Tropical Medicine andHygiene 89(2):146–150.

Kovats R. S., Bouma M. J., Hajat S., Worrall E., HainesA. 2003. El Nino and health. Lancet 362(9394):1481–1489.

Krzywinski J., Besansky N. J. 2003. Molecular system-atics of Anopheles: from subgenera to subpopula-tions. Annual Review of Entomology 48:111–139.

La Grange J. J. P. 1995. Survey of anopheline mosqui-toes (Diptera: Culicidae) in a malarious area ofSwaziland. African Entomology 3:217–219.

Lautze J., McCartney M., Kirshen P., Olana D., Jayas-inghe G., Spielman A. 2007. Effect of a large damon malaria risk: the Koka reservoir in Ethiopia.Tropical Medicine & International Health 12(8):982–989.

Levins R. 1966. The strategy of model building inpopulation biology. American Scientist 54:421–431.

Levins R. 1995. Toward an integrated epidemiology.Trends in Ecology and Evolution 10(7):304.

Levins R. 2006. Strategies of abstraction. Biology &Philosophy 21(5):741–755.

Levins R., Schultz B. B. 1996. Effects of density de-pendence, feedback and environmental sensitivityon correlations among predators, prey and plantresources: models and practical implications. Jour-nal of Animal Ecology 65(6):802–812.

Lewontin R. C. 1974. The Genetic Basis of EvolutionaryChange. New York: Columbia University Press.

Lewontin R., Levins R. 2000. Let the numbers speak.International Journal of Health Services 30(4):873–877.

Lindblade K. A., Walker E. D., Onapa A. W., KatunguJ., Wilson M. L. 1999. Highland malaria in Ugan-da: prospective analysis of an epidemic associatedwith El Nino. Transactions of the Royal Society ofTropical Medicine and Hygiene 93(5):480–487.

Lindblade K. A., Walker E. D., Onapa A. W., KatunguJ., Wilson M. L. 2000. Land use change altersmalaria transmission parameters by modifyingtemperature in a highland area of Uganda. Trop-ical Medicine & International Health 5(4):263–274.

Lindsay S. W., Armstrong Schellenberg J. R., ZeilerH. A., Daly R. J., Salum F. M., Wilkins H. A. 1995.Exposure of Gambian children to Anopheles gam-biae malaria vectors in an irrigated rice productionarea. Medical and Veterinary Entomology 9(1):50–58.

Lindsay S. W., Birley M. H. 1996. Climate change andmalaria transmission. Annals of Tropical Medicineand Parasitology 90(6):573–588.

Lindsay S. W., Birley M. 2004. Rural development and

malaria control in Sub-Saharan Africa. Ecohealth1(2):129–137.

Lindsay S. W., Bødker R., Malima R., Msangeni H. A.,Kisinza W. 2000. Effect of 1997–1998 El Nino onhighland malaria in Tanzania. Lancet 355(9208):989–990.

Lindsay S. W., Emerson P. M., Charlwood J. D. 2002.Reducing malaria by mosquito-proofing houses.Trends in Parasitology 18(11):510–514.

Lindsay S. W., Martens W. J. 1998. Malaria in theAfrican highlands: past, present and future. Bulle-tin of the World Health Organization 76:33–45.

Livingstone F. B. 1958. Anthropological implicationsof sickle-cell gene distribution in West Africa.American Anthropologist 60:533–562.

Lockhart J. D. F., Highton R. B., McMahon J. P. 1969.Public health problems arising out of man-madefish ponds in the Western Province of Kenya. EastAfrican Medical Journal 46(8):471–480.

Longstreth J., Kondrachine A. 2002. Population mi-gration and malaria. Pages 270–280 in The Contex-tual Determinants of Malaria, edited by E. Casmanand H. Dowlatabadi. Washington (D.C.): Re-sources for the Future.

Luckner D., Lell B., Greve B., Lehman L. G., Schmidt-Ott R. J., Matousek P., Herbich K., Schmid D.,Mba R., Kremsner P. G. 1998. No influence ofsocioeconomic factors on severe malarial anae-mia, hyperparasitaemia or reinfection. Transac-tions of the Royal Society of Tropical Medicine andHygiene 92(5):478–481.

Lyimo E. O., Koella J. C. 1992. Relationship betweenbody size of adult Anopheles gambiae s.l. and infec-tion with the malaria parasite Plasmodium falcipa-rum. Parasitology 104:233–237.

Lyimo E. O., Takken W. 1993. Effects of adult bodysize on fecundity and the pre-gravid rate of Anoph-eles gambiae females in Tanzania. Medical and Veter-inary Entomology 7(4):328–332.

Mabaso M. L. H., Kleinschmidt I., Sharp B. L., SmithT. 2007. El Nino Southern Oscillation (ENSO)and annual malaria incidence in Southern Africa.Transactions of the Royal Society of Tropical Medicineand Hygiene 101(4):326–330.

Macdonald G. 1953. The analysis of malaria epidem-ics. Tropical Diseases Bulletin 50(10):871–889.

Madhavan K. T., Jajoo U. N., Bhalla A. 2001. Seasonalvariations in incidence of severe and complicatedmalaria in central India. Indian Journal of MedicalSciences 55(1):43–46.

Mahande A., Mosha F., Mahande J., Kweka E. 2007.Feeding and resting behaviour of malaria vector,Anopheles arabiensis with reference to zooprophy-laxis. Malaria Journal 6:100.

Malakooti M. A., Biomndo K., Shanks G. D. 1998.Reemergence of epidemic malaria in the high-

March 2010 51CLIMATE CHANGE AND HIGHLAND MALARIA

lands of western Kenya. Emerging Infectious Diseases4:671–676.

Manda H., Gouagna L. C., Foster W. A., Jackson R. R.,Beier J. C., Githure J. I., Hassanali A. 2007a. Effectof discriminative plant-sugar feeding on the sur-vival and fecundity of Anopheles gambiae. MalariaJournal 6:113.

Manda H., Gouagna L. C., Nyandat E., Kabir E. W.,Jackson R. R., Foster W. A., Githure J. I., BeierJ. C., Hassanali A. 2007b. Discriminative feedingbehaviour of Anopheles gambiae s.s. on endemicplants in western Kenya. Medical and Veterinary En-tomology 21(1):103–111.

Martens P., Hall L. 2000. Malaria on the move: hu-man population movement and malaria transmis-sion. Emerging Infectious Diseases 6(2):103–109.

Mathanga D. P., Bowie C. 2007. Malaria control inMalawi: are the poor being served? InternationalJournal for Equity in Health 6:22.

Matthys B., N’Goran E. K., Kone M., Koudou B. G.,Vounatsou P., Cisse G., Tschannen A. B., TannerM., Utzinger J. 2006a. Urban agricultural land useand characterization of mosquito larval habitats ina medium-sized town of Cote d’Ivoire. Journal ofVector Ecology 31(2):319–333.

Matthys B., Vounatsou P., Raso G., Tschannen A. B.,Becket E. G., Gosoniu L., Cisse G., Tanner M.,N’Goran E. K., Utzinger J. 2006b. Urban farmingand malaria risk factors in a medium-sized town inCote d’Ivoire. American Journal of Tropical Medicineand Hygiene 75(6):1223–1231.

Mbogo C. N. M., Kabiru E. W., Glass G. E., Forster D.,Snow R. W., Khamala C. P. M., Ouma J. H.,Githure J. I., Marsh K., Beier J. C. 1999. Vector-related case-control study of severe malaria inKilifi district, Kenya. American Journal of TropicalMedicine and Hygiene 60(5):781–785.

Minakawa N., Seda P., Yan G. 2002a. Influence of hostand larval habitat distribution on the abundanceof African malaria vectors in western Kenya. Amer-ican Journal of Tropical Medicine and Hygiene 67:32–38.

Minakawa N., Sonye G., Mogi M., Githeko A., Yan G.2002b. The effects of climatic factors on the dis-tribution and abundance of malaria vectors inKenya. Journal of Medical Entomology 39(6):833–841.

Minakawa N., Sonye G., Mogi M., Yan G. 2004. Hab-itat characteristics of Anopheles gambiae s.s. larvae ina Kenyan highland. Medical and Veterinary Entomol-ogy 18(3):301–305.

Mueller I., Namuigi P., Kundi J., Ivivi R., TandrapahT., Bjorge S., Reeder J. C. 2005. Epidemic malariain the highlands of Papua New Guinea. AmericanJournal of Tropical Medicine and Hygiene 72(5):554–560.

Mueller L. D., Joshi A. 2000. Stability in Model Popula-tions. Princeton (NJ): Princeton University Press.

Munga S., Minakawa N., Zhou G. F., Barrack O. A. J.,Githeko A. K., Yan G. Y. 2006. Effects of larvalcompetitors and predators on oviposition site se-lection of Anopheles gambiae sensu stricto. Journal ofMedical Entomology 43(2):221–224.

Muturi E. J., Muriu S., Shililu J., Mwangangi J., JacobB. G., Mbogo C., Githure J., Novak R. J. 2008.Effect of rice cultivation on malaria transmissionin central Kenya. American Journal of Tropical Med-icine and Hygiene 78(2):270–275.

Najera J. A. 1994. The control of tropical diseases andsocioeconomic development (with special refer-ence to malaria and its control). Parassitologia 36:17–33.

Ndiaye O., Le Hesran J. Y., Etard J. F., Diallo A.,Simondon F., Ward M. N., Robert V. 2001. Varia-tions climatiques et mortalite attribuee au palud-isme dans la zone de Niakhar, Senegal, de 1984 a1996. Cahiers d’etudes et de recherche francophones/Sante 11:25–33.

Nicholson A. J. 1957. The self-adjustment of popula-tions to change. Cold Spring Harbor Symposium inQuantitative Biology 22:153–173.

Njiru B. N., Mukabana W. R., Takken W., KnolsB. G. J. 2006. Trapping of the malaria vectorAnopheles gambiae with odour-baited MM-X traps insemi-field conditions in western Kenya. MalariaJournal 5:39.

Odiere M., Bayoh M. N., Gimnig J., Vulule J., IrunguL., Walker E. 2007. Sampling outdoor, restingAnopheles gambiae and other mosquitoes (Diptera:Culicidae) in western Kenya with clay pots. Journalof Medical Entomology 44:14–22.

Odling-Smee F. J., Laland K. N., Feldman M. W. 2003.Niche Construction: The Neglected Process in Evolution.Princeton (NJ): Princeton University Press.

Oesterholt M. J. A. M., Bousema J. T., MwerindeO. K., Harris C., Lushino P., Masokoto A., Mwer-inde H., Mosha F. W., Drakeley C. J. 2006. Spatialand temporal variation in malaria transmission ina low endemicity area in northern Tanzania. Ma-laria Journal 5:98.

Okoye P. N., Wilson M. D., Boakye D. A., Brown C. A.2005. Impact of the Okyereko irrigation project inGhana on the risk of human malaria infection byAnopheles species (Diptera: Culicidae). AfricanEntomology 13(2):249–253.

Ong’echa J. M., Keller C. C., Were T., Ouma C.,Otieno R. O., Landis-Lewis Z., Ochiel D., SlingluffJ. L., Mogere S., Ogonji G. A., Orago A. S., VululeJ. M., Kaplan S. S., Day R. D., Perkins D. 2006.Parasitemia, anemia, and malarial anemia in in-fants and young children in a rural holoendemicPlasmodium falciparum transmission area. American

52 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

Journal of Tropical Medicine and Hygiene 74(3):376–385.

Onori E., Grab B. 1980. Indicators for the forecastingof malaria epidemics. Bulletin of the World HealthOrganization 58:91–98.

Paaijmans K. P., Huijben S., Githeko A. K, Takken W.2009. Competitive interactions between larvae ofthe malaria mosquitoes Anopheles arabiensis andAnopheles gambiae under semi-field conditions inwestern Kenya. Acta Tropica 109(2):124–130.

Palsson K., Jaenson T. G. T., Dias F., Laugen A. T.,Bjorkman A. 2004. Endophilic Anopheles mosqui-toes in Guinea Bissau, West Africa, in relation tohuman housing conditions. Journal of Medical En-tomology 41(4):746–752.

Pascual M., Ahumada J. A., Chaves L. F., Rodo X.,Bouma M. 2006. Malaria resurgence in the eastAfrican highlands: temperature trends revisited.Proceedings of the National Academy of Sciences USA103(15):5829–5834.

Pascual M., Bouma M. J. 2009. Do rising temperaturesmatter? Ecology 90(4):906–912.

Pascual M., Cazelles B., Bouma M. J., Chaves L. F.,Koelle K. 2008. Shifting patterns: malaria dynam-ics and rainfall variability in an East African high-land. Proceedings of the Royal Society of London, SeriesB: Biological Sciences 275(1631):123–132.

Patz J. A., Hulme M., Rosenzweig C., Mitchell T. D.,Goldberg R. A., Githeko A. K., Lele S., McMichaelA. J., LeSueur D. 2002. Regional warming andmalaria resurgence. Nature 420:627–628.

Patz J. A., Olson S. H. 2006. Malaria risk and temper-ature: influences from global climate change andlocal land use practices. Proceedings of the NationalAcademy of Sciences USA 103(15):5635–5636.

Patz J. A., Strzepek K., Lele S., Hedden M., Greene S.,Noden B., Hay S. I., Kalkstein L., Beier J. C. 1998.Predicting key malaria transmission factors, bitingand entomological inoculation rates, using mod-eled soil moisture in Kenya. Tropical Medicine &International Health 3(10):818–827.

Perfecto I., Vandermeer J. 2008. Biodiversity conser-vation in tropical agroecosystems: a new conserva-tion paradigm. Annals of the New York Academy ofSciences 1134:173–200.

Petrarca V., Beier J. C., Onyango F., Koros J., Asiago C.,Koech D. K., Roberts C. R. 1991. Species compositionof the Anopheles gambiae complex (Diptera: Culicidae)at two sites in western Kenya. Journal of Medical Entomol-ogy 28(3):307–313.

Phombeah G. 2005. Kenya: overthrowing coloniallandlords. BBC Focus on Africa 16:30–33.

Prothero R. M. 1965. Migrants and Malaria. London(UK): Longmans.

Pulliam H. R. 1988. Sources, sinks, and populationregulation. American Naturalist 132:652–661.

Qiu Y. T., Smallegange R. C., Ter Braak C. J. F.,

Spitzen J., Van Loon J. J. A., Jawara M., Milligan P.,Galimard A. M., Van Beek T. A., Knols B. G. J.,Takken W. 2007. Attractiveness of MM-X trapsbaited with human or synthetic odor to mosqui-toes (Diptera: Culicidae) in the Gambia. Journal ofMedical Entomology 44(6):970–983.

Rab M. A., Freeman T. W., Rahim S., Durrani N.,Simon-Taha A., Rowland M. 2003. High altitudeepidemic malaria in Bamian province, Central Af-ghanistan. La Revue de Sante de la Mediterranee Ori-entale 9:232–239.

Reiter P. 2000. From Shakespeare to Defoe: malariain England in the Little Ice Age. Emerging InfectiousDiseases 6(1):1–11.

Reiter P. 2008. Global warming and malaria: knowingthe horse before hitching the cart. Malaria Journal7:S3.

Reiter P., Thomas C. J., Atkinson P. M., Hay S. I.,Randolph S. E., Rogers D. J., Shanks G. D., SnowR. W., Spielman A. 2004. Global warming andmalaria: a call for accuracy. Lancet Infectious Dis-eases 4(6):323–324.

Rodriguez D. J. 1988. Models of growth with densityregulation in more than one life stage. TheoreticalPopulation Biology 34:93–117.

Rogers D. J., Randolph S. E. 2000. The global spreadof malaria in a future, warmer world. Science289(5485):1763–1766.

Sachs J., Malaney P. 2002. The economic and socialburden of malaria. Nature 415(6872):680–685.

Sattler M. A., Mtasiwa D., Kiama M., Premji Z., TannerM., Killeen G. F., Lengeler C. 2005. Habitat charac-terization and spatial distribution of Anopheles sp.mosquito larvae in Dar es Salaam (Tanzania) duringan extended dry period. Malaria Journal 4:4.

Scheffer M., Carpenter S., Foley J. A., Folke C., WalkerB. 2001. Catastrophic shifts in ecosystems. Nature413:591–596.

Schmalhausen I. I. 1949. Factors of Evolution: The The-ory of Stabilizing Selection. Trans. From Russian by I.Dordick. Philadelphia (PA): Blakiston Co.

Scholte E.-J., Ng’habi K., Kihonda J., Takken W., Paaij-mans K., Abdulla S., Killeen G. F., Knols B. G. 2005.An entomopathogenic fungus for control of adultAfrican malaria mosquitoes. Science 308(5728):1641–1642.

Sevilla-Casas E. 1993. Human mobility and malariarisk in the Naya river basin of Colombia. SocialScience and Medicine 37(9):1155–1167.

Shaman J., Day J. F., Stieglitz M. 2002. Drought-induced amplification of Saint Louis encephalitisvirus, Florida. Emerging Infectious Diseases 8(6):575–580.

Shanks G. D., Biomndo K., Hay S. I., Snow R. W. 2000.Changing patterns of clinical malaria since 1965among a tea state population located in the Ken-

March 2010 53CLIMATE CHANGE AND HIGHLAND MALARIA

yan highlands. Transactions of the Royal Society ofTropical Medicine and Hygiene 94(3):253–255.

Shanks G. D., Hay S. I., Omumbo J. A., Snow R. W.2005. Malaria in Kenya’s western highlands. Emerg-ing Infectious Diseases 11(9):1425–1432.

Shanks G. D., Hay S. I., Stern D. I., Biomndo K., SnowR. W. 2002. Meteorologic influences on Plasmo-dium falciparum malaria in the highland tea estatesof Kericho, western Kenya. Emerging Infectious Dis-eases 8:1404–1408.

Shililu J., Ghebremeskel T., Mengistu S., Fekadu H.,Zerom M., Mbogo C., Githure J., Gu W. D., NovakR., Beier J. C. 2003. Distribution of anophelinemosquitoes in Eritrea. American Journal of TropicalMedicine and Hygiene 69(3):295–302.

Shililu J. I., Maier W. A., Seitz H. M., Orago A. S. 1998.Seasonal density, sporozoite rates and entomolog-ical inoculation rates of Anopheles gambiae andAnopheles funestus in a high-altitude sugarcanegrowing zone in Western Kenya. Tropical Medicine& International Health 3(9):706–710.

Shumway R. H., Stoffer D. S. 2000. Time Series Analysisand Its Applications. New York: Springer.

Small J., Goetz S. J., Hay S. I. 2003. Climatic suitabilityfor malaria transmission in Africa, 1911-1995. Pro-ceedings of the National Academy of Sciences USA100(26):15341–15345.

Smith T. A., Leuenberger R., Lengeler C. 2001. Childmortality and malaria transmission intensity in Af-rica. Trends in Parasitology 17(3):145–149.

Snow R. W., Peshu N., Forster D., Bomu G., MitsanzeE., Ngumbao E., Chisengwa R., Schellenberg J. R.,Hayes R. J., Newbold C. I., Marsh K. 1998. Envi-ronmental and entomological risk factors for thedevelopment of clinical malaria among childrenon the Kenyan coast. Transactions of the Royal Societyof Tropical Medicine and Hygiene 92(4):381–385.

Somi M. F., Butler J. R. G., Vahid F., Njau J., KachurS. P., Abdulla S. 2007. Is there evidence for dualcausation between malaria and socioeconomic sta-tus? Findings from rural Tanzania. American Jour-nal of Tropical Medicine and Hygiene 77(6):1020–1027.

Spielman A. 2006. Ethical dilemmas in malaria con-trol. Journal of Vector Ecology 31(1):1–8.

Spielman A., Kitron U., Pollack R. J. 1993. Time lim-itation and the role of research in the worldwideattempt to eradicate malaria. Journal of MedicalEntomology 30(1):6–19.

Staedke S. G., Nottingham E. W., Cox J., Kamya M. R.,Rosenthal P. J., Dorsey G. 2003. Short report:proximity to mosquito breeding sites as a risk fac-tor for clinical malaria episodes in an urban co-hort of Ugandan children. American Journal ofTropical Medicine and Hygiene 69(3):244–246.

Stearns S. C. 2000. Life history evolution: successes,

limitations, and prospects. Naturwissenschaften 87:476–486.

Strahler A. N. 1973. Introduction to Physical Geography.New York: Wiley.

Styer L. M., Carey J. R., Wang J. L., Scott T. W. 2007.Mosquitoes do senesce: departure from the para-digm of constant mortality. American Journal ofTropical Medicine and Hygiene 76(1):111–117.

Tanser F. C., Sharp B., le Sueur D. 2003. Potentialeffect of climate change on malaria transmissionin Africa. Lancet 362(9398):1792–1798.

Teklehaimanot H. D., Lipsitch M., Teklehaimanot A.,Schwartz J. 2004. Weather-based prediction ofPlasmodium falciparum malaria in epidemic-proneregions of Ethiopia. I. Patterns of lagged weathereffects reflect biological mechanisms. Malaria Jour-nal 3:41.

Thomas C. J., Davies G., Dunn C. E. 2004. Mixedpicture for changes in stable malaria distributionwith future climate in Africa. Trends in Parasitology20(5):216–220.

Thomson M. C., Connor S. J., D’Alessandro U., Rowl-ingson B., Diggle P., Cresswell M., Greenwood B.1999. Predicting malaria infection in Gambianchildren from satellite data and bed net surveys:the importance of spatial correlation in the inter-pretation of results. American Journal of TropicalMedicine and Hygiene 61(1):2–8.

Thomson M. C., Mason S. J., Phindela T., Connor S. J.2005. Use of rainfall and sea surface temperaturemonitoring for malaria early warning in Botswana.American Journal of Tropical Medicine and Hygiene73:214–221.

Trape J. F., Lefebvre-Zante E., Legros F., Ndiaye G.,Bouganali H., Druilhe P., Salem G. 1992. Vectordensity gradients and the epidemiology of urbanmalaria in Dakar, Senegal. American Journal of Trop-ical Medicine and Hygiene 47(2):181–189.

Trape J. F., Pison G., Spiegel A., Enel C., Rogier C.2002. Combating malaria in Africa. Trends in Par-asitology 18(5):224–230.

Turchin P. 2003. Complex Population Dynamics: A The-oretical/Empirical Synthesis. Princeton (NJ): Prince-ton University Press.

van der Hoek W., Konradsen F., Perera D., Ameras-inghe P. H., Amerasinghe F. P. 1997. Correlationbetween rainfall and malaria in the dry zone of SriLanka. Annals of Tropical Medicine and Parasitology91(8):945–949.

Vittor A. Y., Gilman R. H., Tielsch J., Glass G., ShieldsT., Lozano W. S., Pinedo-Cancino V., Patz J. A.2006. The effect of deforestation on the human-biting rate of Anopheles darlingi, the primary vectorof falciparum malaria in the Peruvian Amazon.American Journal of Tropical Medicine and Hygiene74(1):3–11.

Werner E. E., Peacor S. D. 2003. A review of trait-

54 Volume 85THE QUARTERLY REVIEW OF BIOLOGY

mediated indirect interactions in ecological com-munities. Ecology 84(5):1083–1100.

White G. B. 1972. The Anopheles gambiae complex andmalaria transmission around Kisumu, Kenya.Transactions of the Royal Society of Tropical Medicineand Hygiene 66(4):572–581.

Wilson M. L. 1994. Developing paradigms to antici-pate emerging diseases: transmission cycles and asearch for a pattern. Annals of the New York Academyof Sciences 740:418–422.

Wilson M. L. 2003. Tropical agriculture and humandisease: ecological complexities pose researchchallenges. Pages 245–262 in Tropical Agroecosys-tems, edited by J. H. Vandermeer. Boca Raton(FL): CRC Press.

Wilson M. L., Agudelo-Silva F., Spielman A. 1990.Increased abundance, size, and longevity of food-deprived mosquito populations exposed to a fun-gal larvicide. American Journal of Tropical Medicineand Hygiene 43(5):551–556.

World Health Organization [WHO]. 1982. Manual onenvironmental management for mosquito control, withspecial emphasis on malaria vectors. Geneva (Switzer-land): WHO.

Yang G. J., Bradshaw C. J. A., Whelan P. I., Brook B. W.2008. Importance of endogenous feedback control-ling the long-term abundance of tropical mosquitospecies. Population Ecology 50(3):293–305.

Yasuoka J., Levins R. 2007a. Ecology of vector mosqui-toes in Sri Lanka—suggestions for future mos-quito control in rice ecosystems. Southeast AsianJournal of Tropical Medicine and Public Health 38(4):646–657.

Yasuoka J., Levins R. 2007b. Impact of deforestationand agricultural development on anophelineecology and malaria epidemiology. AmericanJournal of Tropical Medicine and Hygiene 76(3):450 – 460.

Yasuoka J., Levins R., Mangione T. W., Spielman A.2006a. Community-based rice ecosystem manage-ment for suppressing vector anophelines in SriLanka. Transactions of the Royal Society of TropicalMedicine and Hygiene 100:995–1006.

Yasuoka J., Mangione T. W., Spielman A., Levins R.2006b. Impact of education on knowledge, agri-

cultural practices, and community actions formosquito control and mosquito-borne disease pre-vention in rice ecosystems in Sri Lanka. AmericanJournal of Tropical Medicine and Hygiene 74:1034–1042.

Ye-Ebiyo Y., Pollack R. J., Kiszewski A. E., Spielman A.2003. Enhancement of development of larvalAnopheles arabiensis by proximity to floweringmaize (Zea mays) in turbid water and whencrowded. American Journal of Tropical Medicine andHygiene 68(6):748–752.

Ye-Ebiyo Y., Pollack R. J., Spielman A. 2000. En-hanced development in nature of larval Anoph-eles arabiensis mosquitoes feeding on maizepollen. American Journal of Tropical Medicine andHygiene 63(1):90 –93.

Ye Y., Hoshen M., Louis V., Seraphin S., Traore I.,Sauerborn R. 2006. Housing conditions and Plas-modium falciparum infection: protective effect ofiron-sheet roofed houses. Malaria Journal 5:8.

Ye Y., Louis V. R., Simboro S., Sauerborn R. 2007.Effect of meteorological factors on clinical malariarisk among children: an assessment using village-based meteorological stations and community-based parasitological survey. BMC Public Health7:101.

Yuval B. 1992. The other habit: sugar feeding by mos-quitoes. Bulletin of the Society for Vector Ecology 17:150–156.

Zhou G., Minakawa N., Githeko A. K., Yan G. 2004.Association between climate variability and ma-laria epidemics in the East African highlands. Pro-ceedings of the National Academy of Sciences USA101(8):2375–2380.

Zhou G., Minakawa N., Githeko A. K., Yan G. 2005.Climate variability and malaria epidemics in thehighlands of East Africa. Trends in Parasitology21(2):54–56.

Zhou G., Munga S., Minakawa N., Githeko A. K., YanG. 2007. Spatial relationship between adult ma-laria vector abundance and environmental factorsin western Kenya highlands. American Journal ofTropical Medicine and Hygiene 77(1):29–35.

Editor: Daniel Dykhuizen

March 2010 55CLIMATE CHANGE AND HIGHLAND MALARIA