VECTORIAL CAPACITY ACROSS AN ENVIRONMENTAL
GRADIENT
NICHAR GREGORY
Department of Life Sciences
Imperial College London
A thesis submitted for the degree of Doctor of Philosophy
2019
1
“Mosquitoes are certainly our declared enemies, and very annoying enemies at that. But
they are enemies good for us to know. If we give them even the least bit of attention we find ourselves forced to admire them and to admire even the instrument they bleed us with; for this
it is only necessary to examine their structure. Furthermore, throughout their lives they offer
facts adequate to please any of those minds which are curious about the marvels of nature.
There are even times in the mosquitoes’ lives when, having made the observer forget that
they’ll persecute him some day, they make him feel actually, solicitous about their welfare”
- De Rèaumur’s (1740)
2 Declaration of Originality
I declare that the work described in this thesis is my own, except where stated, and that I have compiled it in my own words. Where others are included as authors on individual chapters, those individuals have provided comments on that chapter. Two anonymous reviewers provided a great deal of insight into the analyses in Chapter 3 when that chapter was submitted for publication in Environmental Research Letters.
Copyright declaration
The copyright of this thesis rests with the author. Unless otherwise indicated, its contents are licensed under a Creative Commons Attribution-Non Commercial 4.0 International Licence (CC
BY-NC). Under this licence, you may copy and redistribute the material in any medium or format. You may also create and distribute modified versions of the work. This is on the condition that: you credit the author and do not use it, or any derivative works, for a commercial purpose. When reusing or sharing this work, ensure you make the licence terms clear to others by naming the licence and linking to the licence text. Where a work has been adapted, you should indicate that the work has been changed and describe those changes. Please seek permission from the copyright holder for uses of this work that are not included in this licence or permitted under UK Copyright Law.
3 ABSTRACT
Disease transmitted by mosquitoes present some of the most pressing challenges facing human health today. Land-use change is a key driver of disease emergence, however the mechanisms linking environmental covariates of change, such as temperature, to the transmission potential of mosquitoes is poorly understood. Studies exploring these relationships have largely been correlative in nature, and thus have limited capacity to predict dynamics through space and time. Mechanistic approaches provide a valuable framework for understanding the processes underlying transmission, however they suffer from a dearth of field data on fundamental mosquito ecology. In both approaches, the environmental data used is typically coarse in scale and interpolated from weather stations located in open areas. In reality, local climatic conditions can vary considerably over fine spatial and temporal scales, particularly in dynamic working landscapes. Wild mosquitoes experience and respond to this highly dynamic environment, and failing to account for this variation may have significant implications for the accuracy of epidemiological models. This thesis uses an established epidemiological framework to explore the effects of tropical forest conversion to oil palm plantation on the potential for Ae. albopictus mosquitoes to transmit disease. Using field-derived microclimate data and published thermal responses of mosquito traits, I first examine how the scale of environmental data affects predictions of mosquito demography under land-use change. Next, I conduct field experiments to investigate whether microclimate heterogeneity across a land-use gradient drives variation in the rates of larval development. By pairing fine-scale microclimate data with temperature- dependent trait estimates, I find that forest conversion significantly increases the potential of
Ae. albopictus to transmit disease. Together, these findings advance our understanding of Ae. albopictus ecology, and highlight the importance of incorporating fine-scale environmental and mosquito data into epidemiological frameworks to better understand disease risk in changing landscapes.
4 Table of Contents List of Tables & Figures ...... 7 Acknowledgments ...... 8 CHAPTER 1 : BACKGROUND ...... 10 1.1 The emergence of mosquito-borne diseases ...... 10 1.2 Mosquito ecology: the linchpin of effective disease control ...... 14 1.3 Predicting disease transmission in dynamic environments ...... 20 1.4 Tropical forest conversion, oil palm plantations and vectorial capacity ...... 22 1.5 Aedes-transmitted arboviruses in Malaysia ...... 26 1.6 The Asian tiger mosquito (Aedes albopictus) ...... 27 CHAPTER 2: Aims and objectives ...... 28 2.1 Research aims and objectives ...... 28 2.2 Thesis outline ...... 29 2.3 Study area ...... 32 CHAPTER 3: Methodology ...... 36 3. 1 Data sources ...... 36 3.1.1 Temperature ...... 36 3.1.2 Thermal trait responses ...... 37 3. 2 Estimating mosquito population growth rates (rmax) ...... 38 3.3 Field data ...... 41 3.3.1 Larval experiments ...... 41 3.3.2 Human landing catches ...... 42 3.4 Derived parameters ...... 43 3.4.1 Estimating adult mosquito survival ...... 43 3.4.2 Vectorial capacity ...... 45 3.5 Statistical analysis ...... 46 3.6 Ethical statement ...... 49 CHAPTER 4: DISTURBANCE-MEDIATED MICROCLIMATE DETERMINES INTRINSIC POPULATION GROWTH OF MOSQUITOES ...... 50 4.1 ABSTRACT ...... 50 4.2 Introduction ...... 51 4.3 Results ...... 55 4.3.1 Effects of drought and land-use on diurnal temperature cycles ...... 55 4.3.2 Effects of land-use and drought on predicted performance of Ae. albopictus traits ...... 57 4.3.3 Temperature-sensitive rmax ...... 62 4.4 Discussion ...... 64 CHAPTER 5 ...... 71 EL NIÑO DROUGHT AND TROPICAL FOREST CONVERSION SYNERGISTICALLY DETERMINE MOSQUITO DEVELOPMENT RATE ...... 71 5.1 ABSTRACT ...... 71 5.2 Introduction ...... 72 5.3 Results ...... 75 5.3.1 Effect of ENSO drought and land-use type on microclimate ...... 75 5.3.2 Effect of microclimate on development time of Ae. albopictus mosquito larvae .... 77
5 5.4 Discussion ...... 80 5.4.1 Logged forests buffer the effects of ENSO drought on microclimate conditions ..... 80 5.4.2 Land-use and ENSO temperatures synergistically affect mosquito development .... 80 5.4.3 Beyond microclimate: potential effects of land-use and drought on larval development and survival ...... 84 CHAPTER 6: ...... 87 TROPICAL FOREST CONVERSION TO OIL PALM INCREASES MOSQUITO VECTORIAL CAPACITY ...... 87 6.1 ABSTRACT ...... 87 6.2 Introduction ...... 88 6.3 Results ...... 90 6.3.1 Mosquito reproduction ...... 90 6.3.2 Human landing catches ...... 91 6.3.3 Estimating adult survival and temperature-dependent transmission ...... 92 6.3.4 Effects of forest conversion on vectorial capacity ...... 93 6.3.5 Effects of adult mortality and land-use on vectorial capacity ...... 95 6.4 Discussion ...... 97 CHAPTER 7: ...... 104 CONCLUDING REMARKS & FUTURE DIRECTIONS ...... 104 7.1 Principal findings: ...... 104 7.2 Epidemiological implications ...... 105 7.3 Future directions ...... 107 7.3.1 Closing the gap between laboratory and field studies ...... 107 7.3.2 Incorporating mosquito-borne disease risk in natural area management ...... 110 7.3.3 The role of long-term ecological studies in progressing mosquito-borne disease research ...... 110 7.4 Conclusions ...... 111 Appendix ...... 112 REFERENCES ...... 116
6 List of Tables & Figures
Tables Table 2.1 Environmental parameters of SAFE Project sites
Table 3.1. Thermal response of Ae. albopictus life-history parameters required for rmax model. Table 3.2 Traits and data sources for vectorial capacity parameters Table 4.1. Average daily mean, minimum and maximum temperatures across a forest – oil palm plantation gradient. Table 5.1 Coefficients of Cox proportional hazards survival analysis Table 6.2 Summary of host-seeking mosquito collections and parity rates Table 4.3. Estimated density of Ae. albopictus mosquitoes
Figures Figure 1.1 Life-cycle of two key mosquito-borne diseases: malaria and dengue virus Figure 1.2. Life-cycle of Ae. albopictus mosquito Figure 1.3. Thermal performance curve Figure 2.1. Land use and land cover change in Borneo (1973–2015) Figure 2.2. Schematic diagram of thesis chapters Figure 2.3. Map of SAFE project sites and sampling locations in Malaysian Borneo Figure 3.1 Worked example of methods used to calculate temperature-dependent trait responses Figure 4.1. Diurnal temperature cycles along a forest conversion gradient during and after an ENSO event. Figure 4.2 Estimated trait performance across a forest conversion gradient Figure 4.3 Comparison of parameter estimation methods
Figure 4.4 Effects of drought and land-use on rmax
Figure 4.5 Temperature-dependent rmax estimates Figure 5.1 Daily diurnal temperature cycles across two land-use types Figure 5.2 Summary of land-use effects on microclimate and mosquito larval development Figure 6.1 Effects of land-use on mosquito traits that determine vectorial capacity Figure 6.2 Effects of land-use on vectorial capacity Figure 6.3 Effects of mortality and land-use on vectorial capacity
7
Acknowledgments
Writing this section of my thesis has been an exercise in compromise between my overwhelming desire to individually list the legion of people who have supported me, and the rather urgent need to finish. If I have left you out, please accept my apologies, my appreciation and my pints.
I owe tremendous thanks to my doctoral supervisors Lauren Cator and Rob Ewers. This work would not have been possible without their intellectual advice, contagious enthusiasm for science, and unfailing ability to send me out of a meeting more confident and less stressed than when I went in. I am undoubtedly a better scientist because of them. A number of people played a critical role in navigating me through my gradual and panicked realisation that there’s a reason people generally avoid conducting experiments in tropical rainforests. Ryan Gray coordinated the logistics for this fieldwork and brought a joie de vivre to the field camp. I thank him in particular for his morning karaoke and for pouring me that glass of Lemon’s Gin during the Great Colony Loss of 2017. Fieldwork would not have been possible without the help of the SAFE Project staff, the Royal Society’s South East Asia Rainforest Research Partnership, Maliau Basin Management Committee, Sabah Biodiversity Council and my local collaborator, Dr Arthur Chung, and not half as enjoyable without the dynamic SAFE research community. I am especially grateful to the Ewers Lab (Mike Boyle, Claire Wilkinson, Adam Sharp, Ross Gray, Phil Chapman, Sarab Sethi, Becky Heath, Terhi Ruitta and David Orme) for four years of superb scientific discussion and facetious hypothesis building; there is no one I would rather spend months in an isolated field camp with. Hayley Brant, the first SAFE mosquito researcher, patiently guided me through some confusing bureaucracy, and Olivia Daniels skilfully organised all of our lives, something I imagine being similar to herding cats. I thank the Cator lab (Marie Russell, Alima Qureshi, and Andy Aldersley) for providing endless intellectual and moral support, and a constant stream of baked goods. Being endowed with one great lab group is fortunate; having two means that I’ve probably used up all my good luck and that it’s all downhill for me from here.
The Wongsawad Parasitology group at Chiang Mai University taught me how to conduct mosquito dissections for parity assessments, for which I am immensely grateful and
8 considerably more near-sighted. I thank my assessors at Imperial College (Austin Burt and Kris Murray) for reviewing various permutations of my research plans, and to the Silwood Park community for providing a stimulating and welcoming environment from which to commence these plans. I am grateful to the National Environmental Research Council and Imperial College’s Grantham Institute for funding this work once I had been deemed British enough to conduct it. The process of proving my nationality was an arduous but insightful one, and I am forever indebted to the following individuals for their testimonials during those bizarre first few months: Rick Gregory, Diana Bell, Lourdes Gautier, Megan Cattau, Liz Nichols, Zen Makuch, Tim & Caterina Williams, Rachel Small, Gordon Conway, Andrew Kirkby, and Iain Craig. I quite literally could not have commenced this PhD without them.
Growing up as a third culture kid and moving every few years has meant that the concept of ‘home’ is an abstract one to me. I thank my international friends, particularly my Wholefools and my Bogart collective, for their unwavering long-distance love and support that created a sense of familiarity wherever I was. I thank Paul Kelly for beginning this journey with me, and Natalie Clark and Gen Finerty for seeing me through to the end. The Finerty family (Sharon, Terry, Gen, Josh, Jack, Nick, Trixie and Bella) welcomed me into their home in my final year, and I don’t quite have the words to thank them for this kindness. My relatives, Sandra Bell, John and Hazel Gregory, were a constant source of encouragement and curry, for which I am enormously grateful. I thank my favourite brother, Tawun, for keeping me humble and for leading an exemplar life of adventure, which I am not at all jealous of.
Finally, I dedicate this thesis to my parents, Rick and Nualchan Gregory, whose curiosity and enthusiasm for the world provided me with the most privileged of upbringings, and in loving memory of my aunt, Edd Sarisa Kajohnkittiyut.
9
CHAPTER 1 : BACKGROUND
1.1 The emergence of mosquito-borne diseases
Over half the global population is at risk of diseases transmitted by mosquito vectors (WHO
2017). In the past few decades, the incidence and geographical distribution of many mosquito- borne diseases has increased considerably, with diseases emerging in new areas or re-emerging in areas where they had previously been eradicated (Bhatt et al. 2013; Grobbelaar et al. 2016;
Paixão et al. 2018). The global re-emergence of dengue virus is perhaps the clearest example of this (Bhatt et al. 2013). Historically considered a disease of little priority due to low mortality rates and infrequent epidemics (Henderson & Morse 1993), dengue is now estimated to infect
390 million people annually (Bhatt et al. 2013) and is arguably the most important viral vector- borne disease infecting humans today (Gubler 2012). Whilst less severe in magnitude, similar patterns of re-emergence have been observed for other arboviruses, such as yellow fever (Gubler
2004; Grobbelaar et al. 2016), chikungunya (Weaver & Forrester 2015) and Zika viruses (Shuaib et al. 2016).
Land-use change, from natural to human dominated landscapes, is arguably the most important driver of mosquito-borne diseases emergence (Gubler 1998b; Patz et al. 2000; Kilpatrick &
Randolph 2012). Land-use change has the potential to affect disease transmission by altering the interactions between people, pathogens, vectors and vertebrate hosts. For example, in the
1960s, post-war economic recovery of some Southeast Asian countries resulted in rapid urbanization as people migrated to cities for work. The lack of planning, inadequate housing,
10 water and waste management that accompanied urbanization allowed the dengue vector, Ae. aegypti, to thrive (Gubler 1998a). Not long afterwards, the first epidemic of dengue hemorrhagic fever (DHF) was described (Halstead 1980; Gubler 1988), and within 20 years the disease had spread globally (Gubler 1998a). Land-use change, and particularly urbanisation, continues to play an important role in dengue emergence in Southeast Asia today (Vanwambeke et al. 2007; Sarfraz et al. 2012).
The extent to which land-use change exacerbates mosquito-borne disease depends largely on the nature of land modification, and the hosts, vectors and pathogens in question. For example, whilst urbanisation typically increases the transmission of dengue virus, it has been associated with a general reduction in malaria transmission globally (Tatem et al. 2013). In Nairobi, Kenya, whilst the human population increased from ~50 thousand in the 1930s to 350 thousand in
1960, the number of locally notified malaria cases decreased from 1,182 to 49 (Hay et al. 2005;
Mudhune et al. 2011). Similar rural-urban differences in malaria incidence have been observed in a number of other African countries, including The Gambia (Lindsay et al. 1990), Uganda
(Omumbo et al. 2005) and Ethiopia (Yohannes & Petros 1996). The reduction in malaria cases is primarily due to a combination of improved access to health service (Okeke & Okeibunor 2010), behavioural changes (e.g. increased use of insecticide-treated nets, Hay et al. 2005; Maxwell et al. 2006), and a reduction in the availability of larval breeding sites (Robert et al. 2003) associated with urban environments. Although malaria transmission generally decreases with increasing urbanisation, it has persisted in a number of African cities. In Accra, Ghana, malaria prevalence was found to be significantly higher in children residing in urban areas (Klinkenberg et al. 2008). Similarly, in a study of three neighbourhoods in urban Bouaké, Côte d’Ivoire,
Plasmodium falciparum malaria prevalence ranged from ~50-90% (Traoré et al. 2019). In both cities, urban agriculture (e.g. rice farming) plays a key role in malaria risk by creating breeding sites preferred by Anopheles mosquito vectors.
11 Agricultural development is often associated with the altered transmission of mosquito-borne diseases. Rice farming in particular appears to be important for a number of different mosquito- borne diseases. In East Asia and the Pacific, where lowland rice farming covers an estimated 197 million ha (Dixon et al. 2001), the abundance of Culex tritaeniorhynchus, an important vector of Japanese encephalitis, closely tracks rice-growing in space and time (Richards et al. 2010;
Thongsripong et al. 2013; Samuel et al. 2016). In California, the incidence of West Nile Virus
(WNV) increased with the percentage of the county growing rice (Kovach & Kilpatrick 2018). As with the case of urban malaria in Africa, the increase in disease incidence was due to the preference of Culex tarsalis, the main WNV vector, for breeding in rice paddies. Notably, in areas where Cu. quinquefasciatus was the dominant vector, rice cover was not correlated with
WNV incidence (Kovach & Kilpatrick 2018). Non-irrigated croplands are also associated with altered mosquito-borne disease dynamics. Rubber plantations are linked with an increased risk of malaria in Southeast Asia, due to a preference of the principal vectors, An. minimus and An. dirus, for shaded forest-like habitats (Satitvipawee et al. 2012; Tangena et al. 2016). However, in Africa, rubber plantations do not appear to be associated with increased malaria risk
(Tangena et al. 2016). Underlying the disparate responses are likely differences in the daily activities of the plantation workers; in Asia latex is collected at night, when both latex production and malaria vector biting times are at their peak (Bhumiratana et al. 2013).
Deforestation often precedes agricultural expansion, and is another dominant driver of mosquito-borne disease emergence. For example, an increase in the incidence of simian malaria, Plasmodium knowlesi, in the Southeast Asia is associated with proximity to forest and historical forest cover (Jongwutiwes et al. 2004; Fornace et al. 2016). In the Americas, 80% of malaria cases occur in Amazonia, and deforestation activities are associated with an increased prevalence of disease (Singer & de Castro 2001; de Castro et al. 2006). Similarly in South
America, yellow fever, a flavivirus which spills over into human populations from a sylvatic
12 (jungle) enzootic cycle, was typically associated with proximity of susceptible individuals to forested areas (Filippis et al. 2001). However, urban epidemics have become increasingly common in recent years (Van Der Stuyft et al. 1999; Gubler 2004). Whilst these cases share a common driver in deforestation, the mechanisms underlying altered disease transmission differ between them. The spillover of yellow fever from sylvatic into human cycles was largely due to a shift in feeding behaviour of Haemogogus mosquitoes from biting primates in the forest canopy to the humans on the forest floor (Monath 2001). In contrast, deforestation in the Amazon increases malaria prevalence by creating ideal breeding habitats for the dominant vector,
Anopheles darlingi (Singer & de Castro 2001). Importantly, land-use change does not always precede an increase in malaria transmission. Of 47 articles on forest cover and malaria in the
Amazon reviewed by Tucker Lima et al. (2017), only 11% found a positive association, 32% found a negative association and 26% found no clear relationship.
Despite the importance of land-use in driving disease emergence, fewer than 5% of studies investigating the effects of global change on transmission include these processes (Franklinos et al. 2019). Many of these studies are correlative in nature, linking broad categories of land-use to disease incidence, or some measure of risk (Gottdenker et al. 2014). However, because mosquito-borne disease burden is an emergent property of many interacting environmental and socioeconomic processes (Kilpatrick & Randolph 2012; Baeza et al. 2017), correlative approaches may not be sufficient for predicting dynamics through time and space. An alternative approach has been to incorporate the biological processes underlying transmission into a mechanistic, mathematical framework (Hopp & Foley 2001; Mordecai et al. 2017).
However, the accuracy of these models is contingent on a rigorous understanding of the ways in which mosquitoes use and are impacted by their environment.
13 1.2 Mosquito ecology: the linchpin of effective disease control
The transmission of mosquito-borne diseases is inextricably linked to the ecology of mosquito vectors. All medically important mosquitoes share the same general life cycle: gravid females deposit eggs on or near current or predicted water sources, these hatch into first larval instars which then go through four aquatic larval stages, pupate and emerge as adults (Reid 1968;
Nelson 1986). Both adult male and female mosquitoes feed on nectar, however after mating, females will take a blood meal from vertebrate hosts in order to produce eggs (Fig 1). The period from when females take and digest a blood meal, mature their ovaries, and then lay their eggs is denoted the gonotrophic cycle (Detinova 1962a), and the number of cycles is typically assumed to be directly proportional to blood feeding rate. The ecological requirements for each of these stages vary between species and across different environmental settings. For example, feeding behaviour varies considerably between mosquito species. The dengue vector, Ae. aegypti, is both highly anthrophilic, feeding almost exclusively on humans, and will take multiple blood meals in a gonotrophic cycle (Edman et al. 1992). In contrast, whilst being more susceptible to infection by dengue virus, Ae. albopictus mosquitoes are considered secondary to Ae. aegypti in transmitting dengue due in part to their generalist feeding behaviour (Gratz 2004).
Whilst disease transmission cycles vary depending on the pathogen in question (Fig 2), broadly, successful transmission depends on an infected mosquito blood feeding upon a susceptible host.
For this to occur, mosquitoes must survive until adulthood, blood feed upon and acquire an infection from an infected host, survive the time taken for a pathogen to replicate within the mosquito (the extrinsic incubation period, EIP) and then blood feed upon a susceptible host.
14
Fig 1.1 Lifecycle of an Ae. albopictus mosquito. Female mosquitoes oviposit their eggs along the water line of treeholes or artificial containers that are prone to flooding. When the eggs become submerged, first larval instars emerge. Three larval stages follow before the larvae molt into a non-feeding pupal stage, and adults emerge a few days later. Mating typically occurs 2-3 days after emergence after which females seek a bloodmeal to provide protein for egg production (https://us.biogents.com).
These measurable components of mosquito life history that affect their fitness are called
‘functional traits’ (McGill et al. 2006; Gibert et al. 2015), and will be referred to as ‘traits’ from hereon in. Variation in any of these key mosquito traits can profoundly alter disease transmission as well as the efficacy of control interventions. Mechanistic models that incorporate these traits provide a framework for understanding how mosquito traits interact and thus how changes in each trait will affect transmission. The formal integration of entomological concepts into a epidemiological models of mosquito-borne disease occurred half a century ago, when Macdonald (1952b) developed a mathematical model for the basic reproductive number for malaria (R0).
15 A B . .
Fig 1.2. Life-cycle of two key mosquito-borne diseases: malaria (A) and dengue virus (B). Malaria is caused by parasites in the genus Plasmodium and transmitted by Anopheles mosquitoes. Humans and other vertebrates act as intermediate hosts, within which parasite reproduction occurs. Transmission from humans to mosquitoes occurs once parasites have reached sexual maturity and are ingested by susceptible, blood feeding mosquitoes. In contrast, replication of dengue virus, an RNA virus of the genus Flavivirus, occurs in the Ae. mosquito. Transmission occurs when a mosquito ingests an infected bloodmeal, the virus infects the midgut epithelial cells, replicates and disseminates into the hemocoel, and then infects the salivary gland, at which point the mosquito is considered infectious. (Images from Centres for Disease Control and Prevention: https://phil.cdc.gov).
Building on Macdonald’s (1952b) work, Garrett-Jones (1964) developed the concept of vectorial capacity (VC), which describes the maximum potential amplification rate of a pathogen by a vector to a population of susceptible hosts, under conditions where all female mosquitoes become infected. Vectorial capacity is classically expressed as:
!"!! ! !" = [1] !!" (!)
16 Where m is the vector: human ratio, a is the vector biting frequency, µ is the daily probability of vector survival, and n is the extrinsic incubation period of the parasite (the interval in days from acquisition of an infectious agent by the vector to when that insect is capable of transmitting that agent to a vertebrate host). The equation highlighted the importance of adult mosquito survival to pathogen transmission, and explained the early success of DDT and indoor residual spraying (IRS) programs. However, the original vectorial capacity model proposed by Garrett-
Jones (1964) required a considerable list of conditions to be met (Dye 1986). These include:
i. That each parasite species has only one vertebrate and one invertebrate host.
ii. That daily survival of the vectors is constant
iii. That vectors randomly feed on hosts
iv. That the feeding rate of vectors is constant, regardless of host and vector abundance.
v. Susceptible hosts always become infected when bitten by an infected vector.
These assumptions are often violated, and the oversimplification of the mosquito-borne disease paradigm played a key role in limiting the early success of scaled-up control initiatives
(Ferguson et al. 2010; Nájera et al. 2011). A key reason for the simplicity of early models was a lack of data. Although the life cycles of medically important mosquitoes were relatively well understood, historically, the majority of studies focused on evaluating control initiatives leaving key gaps in our understanding of their fundamental ecology (Ferguson et al. 2010; Godfray
2013). A growing body of work seeks to better understand the extent to which the original assumptions are violated, and the consequences of variation in mosquito traits for transmission
(Cator et al. 2019).
A key source of variation in mosquito traits is the environment. As mosquitoes are small-bodied ectotherms, their physiology is closely linked to ambient environmental conditions (Huey &
17 Berrigan 2001). Temperature affects all components of vectorial capacity: mosquito density, by determining rates of mosquito demographic traits, such as growth, survival and reproduction
(Bayoh & Lindsay 2003; Delatte et al. 2009; Mohammed & Chadee 2011), biting rate (or proxies of; Afrane et al. 2005, Rùa et al. 2005, (Afrane et al. 2005), vector competence (Kramer et al.
1983; Turell 1993), and the EIP (Chamberlain & Sudia 1955; Rohani et al. 2009). Naturally the vast majority of studies have focused on this source of variation (Franklinos et al. 2019). ‘The temperature dependence of mosquito traits is typically assessed experimentally, where the response of a relevant trait is measured across a series of constant temperature treatments
(Tun Lin et al. 2000; Alto & Juliano 2001; Bayoh & Lindsay 2003). Fundamental thermal performance curves are then generated using the optimal temperature (T0pt; temperature at which the trait measured is maximised), and the operative range between the critical thermal minimum and maximum of thermal performance (CTmin and CTmax;, respectively; Huey &
Kingsolver, 1989, Fig 3). The resulting unimodal response curves are then scaled-up to estimate the effects of temperature on mosquito abundance and distribution (Tran et al. 2013; Ewing et al. 2016), and disease risk (Craig et al. 1999; Hopp & Foley 2001; Nsoesie et al. 2016).
Relative humidity (RH) can also alter mosquito fitness, particularly as their small body size results in a large surface area to volume ratio over which evaporative water loss can occur.
However, the effects of relative humidity on mosquito traits has received considerably less attention than temperature (Yamana & Eltahir 2013). Most studies focus on the malaria vector,
Anopheles gambiae, and find that low RH (30-45%) can drive physiological changes that allow the mosquito to cope with desiccation stress (Gray et al. 2009; Liu et al. 2011). Low humidity can significantly reduce adult survival (Gray & Bradley 2005) and short-range host-seeking behaviour (Mayne 1930).
18
Fig 1.3. A thermal performance curve. The reaction norm, performance as a function of body temperature, associated critical temperatures (CTmin, CTmax, Topt) and thermal breadth are shown. Thermal performance curves are typically unimodal in shape, with the peak being shifted to the right, such that performance increases slowly below the peak but decreases rapidly above it. Adapted from Dowd et al. (2015).
Whilst temperature drives the rate processes underlying population dynamics, the dynamics and distribution of mosquito populations is determined by the availability of aquatic breeding habitats, which in turn are dependent on precipitation (Fillinger et al. 2004; Koenraadt et al.
2004). A number of studies have identified a positive correlation between rainfall and the density of malaria vectors, such as An. gambiae (Koenraadt et al. 2003; Kelly-Hope et al. 2009) and An. darlingi (Adde et al. 2016), which are able to exploit a wide range of man-made and natural aquatic habitats. Associations between precipitation and the density of Ae. aegptyi and
Ae. albopictus mosquitoes, which transmit a suite of arboviruses, are less clear. Studies have found a positive effect of rainfall (Lourenço-de-Oliveira et al. 2004), a negative effect (Roiz et al.
19 2010), or no effect at all (Toma et al. 2003). The absence of a clear effect is likely due to the ecology of the two mosquito species, which prefer to breed in small, man-made containers.
Seasonal variations in human water storage practices may therefore impact the availability of breeding habitats, effectively decoupling mosquito abundance from rainfall (Trewin et al.
2013). Another reason may be that in some studies conducted in temperate regions (e.g. Toma et al. 2003), periods of high rainfall appear to occur during winter, when temperature may be less favourable for mosquito development and survival.
The characteristics of breeding habitats can have profound effects on adult mosquito traits. The size of aquatic habitats determines how closely water temperature tracks air temperature, with temperatures in small bodies of water corresponding better to ambient temperatures
(Paaijmans et al. 2008a; Paaijmans et al. 2010b). Temperature not only drives mosquito rate processes, such as larval development and survival, but can also determine outcomes of intraspecific (Reiskind & Lounibos 2009) and interspecific competition (Teng & Apperson
2000; Lounibos et al. 2002), nutrient availability (Hoekman 2010), predator-prey interactions
(Calliari et al. 2003), and adult vector competence (Carrington et al. 2013c). Density-dependent effects in mosquitoes are considered strongest during the larval stage, and are correlated with longer larval development times and smaller adult body sizes, but not always survival (Lyimo et al. 1992; Teng & Apperson 2000; Gimnig et al. 2002; Couret et al. 2014).
1.3 Predicting disease transmission in dynamic environments
In the past two decades, there has been a resurgence of interest in measuring and modelling the relationship between mosquitoes, the environment and disease transmission (Franklinos et al.
2019). This is driven largely by the concern that continued land-use change and global warming
20 will shift the distribution of mosquito vectors and alter the dynamics of disease transmission
(Githeko et al. 2000; Rogers & Randolph 2006; Parham & Michael 2009). Today, most modelling efforts incorporate environmental data, however, a common feature of these models is their use of data collected at coarse temporal and spatial scales (Liu-Helmersson et al. 2014;
Proestos et al. 2015; Ryan et al. 2015). For example, the oft-used WorldClim2 database provides monthly global climate grids at a spatial resolution of 30-arc seconds or approximately 1km
(Fick & Hijmans 2017). In contrast, because landscape structure shapes local climate, temperature and humidity can vary considerably over fine spatial scales (Paaijmans & Thomas
2011; Blonder et al. 2018; Jucker et al. 2018). Natural environments also vary over fine temporal scales, with daily temperature fluctuations of 10-20˚C common across many mosquito-borne disease transmission settings (Geerts 2003; Braganza et al. 2004; Paaijmans et al. 2010a). Given that the thermal response of mosquito traits is nonlinear (Fig 3), fluctuating temperatures can generate disproportionate effects of cool and warm events on performance that can interact with or be independent of mean conditions (Vasseur et al. 2014; Lawson et al.
2015). Indeed, there is a growing body of work demonstrating that rate processes predicted under constant temperatures may differ to those under fluctuating conditions over the same time period. Specifically, fluctuations at low mean temperatures tend to increase trait performance relative to mean conditions, with the opposite being true for fluctuations at high temperatures. This effect has been identified for mosquito life history traits (Paaijmans et al.
2009; Lambrechts et al. 2011; Carrington et al. 2013a; Carrington et al. 2013c) and vector competence (Carrington et al. 2013b), and modelling approaches that incorporate nonlinear temperature responses have been found to better fit empirical data on disease risk (Mordecai et al. 2013). Whilst these approaches are promising, a potential limitation is the use of laboratory- derived thermal performance curves to infer mosquito responses in the field.
21 To permit precise hypothesis testing, laboratory conditions are necessarily simplified. Colonised mosquitoes are typically reared under constant temperature, humidity, photoperiod, and density, and provided with abundant nutrition (Gerberg et al. 1994; Munstermann 1997), removing many of the selective pressures placed on their wild counterparts (Leftwich et al.
2016). In the absence of selection for resistance against environmental stresses (e.g. extreme temperature), mosquitoes may lose their capacity to resist such stresses (Hoffmann et al. 2001).
Studies from a range of ectotherm systems find that tolerance to cold and hot temperature extremes is associated with pre-exposure (Danks 2005; Overgaard & Sørensen 2008; Ravaux et al. 2012). The role of acclimation in mosquito systems is less known, however it would have important implications for the accuracy of thermal response curves derived using long- established mosquito colonies. Importantly, stress resistance may affect other life history traits.
Indeed, changes in fecundity and survival associated with rapid adaptation of mosquitoes to laboratory conditions are commonly observed (Watson et al. 2000; Oliva et al. 2011), although colonisation is not always associated with increased mosquito fitness (Huho et al. 2007;
Ponlawat & Harrington 2007). Given the importance of environmental drivers on vectorial capacity, there is a critical need to establish how well laboratory estimates act as proxies for field fitness. Ultimately, understanding the effects of environmental change on mosquito-borne disease transmission necessitates more complete pairing of field-validated mosquito phenology and life history with fine-scale data on the associated ecological changes.
1.4 Tropical forest conversion, oil palm plantations and vectorial capacity
Tropical forest conversion to agriculture is a dominant driver of environmental change globally (FAO 2018b) and can alter the prevalence of mosquito-borne pathogens by either affecting vector abundance or by affecting the components of mosquito life history that directly alter transmission (e.g. biting rates, mosquito competence). In Southeast Asia, deforestation
22 rates are amongst the highest of any tropical region (Sodhi et al. 2004), with the expansion of oil palm (Elaeis guineensis Jacq.) plantations playing a key role (McMorrow & Talip 2001). The conversion of tropical forest to plantation is associated with dramatic changes in habitat, with considerable habitat heterogeneity throughout the plantation lifecycle (Luskin & Potts 2011).
Establishing oil palm plantations in forested areas typically requires first clearing all vegetation, beginning with the selective removal of valuable timber trees. In these initial stages, roads are constructed into forested areas to enable the establishment of logging camps and to allow for the transportation of timber (Edwards et al. 2014). The remaining vegetation is then typically cleared mechanically and/or using fire (Simorangkir et al. 2002), before the land is terraced and oil palm seedlings planted (Butler 2011). Plantations typically have a 25-30 year lifecycle, producing fruit after approximately 3-5 years (Butler et al. 2009).
In the Malaysian Borneo, forest conversion to plantations has been found to generally decrease the abundance of most anopheline species, and increase the abundance of Ae. albopictus mosquitoes (Chang et al. 1997; Young et al. 2017; Brown et al. 2018). However, the effects of forest conversion to plantation on the abundance and vectorial capacity of mosquitoes will vary across each stage of the plantation lifecycle. Logging and deforestation activities dramatically alter the physical characteristics of the environment. First, by opening up the canopy and reducing vegetation structure, logging increases wind, evapotranspiration and solar penetration to the forest floor (Garratt 1994). This alters the microclimate, increasing mean temperatures and daily temperature range, and reducing humidity (Hardwick et al. 2015; Jucker et al. 2018).
Temperatures are highest following clear-felling, when all vegetation has been removed and the soil surface is exposed to direct solar radiation (Luskin & Potts 2011). In the late stages of the plantation lifecycle (~22 years old), mature palm trees provide a dense canopy, which may buffer microclimates. Ferns and other nitrogen-fixing plants are also grown in the plantation understory to prevent soil erosion, further buffering the microclimate close to the soil surface
23 (Luskin & Potts 2011; Arifin et al. 2015). Ultimately how altered microclimates affect vectorial capacity will depend on the range of microclimates available, whether temperatures experienced by mosquitoes exceed the thermal optimum for the underlying traits, by how much and for how long (Kingsolver et al. 2015). Tropical ectotherms are assumed to have relatively narrow thermal breadths and to live close to their optimum, suggesting that even marginal increases in temperature may have detrimental effects on fitness (Addo-Bediako et al. 2000; Sunday et al.
2014).
Logging and agricultural expansion can alter the availability of breeding habitats directly by changing the physical attributes of the landscape. For example, deforestation activities typically increase the availability of sunlit ground pools preferred by some anopheline and culecidine mosquitoes, with consequences for the risk of mosquito-borne diseases such as malaria and encephalitis (Walsh et al. 1993; Méndez et al. 2001; Vittor et al. 2009). The removal of trees during forest conversion may affect the persistence of mosquito species that rely on tree holes for breeding. However, in the Amazon, the relative number of colonisable plant-held waters, so- called phytotelmata, was higher in immediately disturbed forest than in intact forest, due to an increase in the number of fallen-plant-part phytotelmata (Yanoviak et al. 2006b). It is possible that palm trees themselves may harbour phytotelmata when the leaf axils fill with rainwater, however the potential for these to act as mosquito breeding sites has not been explored.
Regardless, most palm trees host an abundance of epiphytes, which are known to act as important breeding sites for mosquito larvae (Mocellin et al. 2009; Rosa et al. 2017). Oil palm plantations are typically terraced to reduce water run-off and soil erosion across the hill slopes
(Mohsen et al. 2014), resulting in the formation of shallow ground pools following rainfall. The availability of breeding habitats can also be altered indirectly through human activities associated with plantation expansion. For example, in logging and plantation camps, water storage for human use (e.g. waste systems, drinking; Mwangangi et al. 2009; Moglia et al.
24 2016;) (Mwangangi et al. 2009; Moglia et al. 2016) poorly managed rubbish, discarded car tyres from logging trucks (Harding et al. 2007), and ornamental pot plants (Rosa et al. 2017) can all provide suitable habitat for mosquito larvae.
Forest conversion to plantation may also affect mosquito population dynamics by altering resource availability. In larval breeding sites, resources are input into the system through fallen plant material (e.g. senescent leaves), which are then consumed by detritivorous invertebrates, which are in turn consumed by mosquito larvae (Muturi et al. 2012). The warmer temperatures in plantations, along with abundance of nitrogen-rich plants such as ferns (Bakker et al. 2011) may result in increased litter decomposition rates compared with forest. Additionally, the ubiquitous use of fertilisers (e.g. ammonium sulphate) in oil palm plantations (Ah et al. 2009) may leach into ground pools enriching the larval breeding sites. Indeed, fertilisers have been shown to significantly increase the survival of An. arabiensis and culicine mosquitoes in agricultural rice systems (Victor & Reuben 2000; Mutero et al. 2004). In contrast, breeding sites in man-made containers may be resource limited, either because they are isolated from vegetated areas, thereby limiting organic input, and/or because they are intentionally kept clean for human use (e.g. drinking water; Barrera et al. 2006). Container-breeding mosquitoes, such as Ae. aegypti and Ae. albopictus, reared under reduced nutrient conditions have delayed pupation (Puggioli et al. 2013), decreased adult size, produce fewer eggs (Mitchell Foster et al.
2012) and are more susceptible to viral infection (Zirbel et al. 2018). Resources do not typically limit adult mosquito populations (however, see Lafferty et al. 2018), but their spatial distribution may be determined by the availability of blood hosts. The establishment of oil palm plantations significantly reduces mammal richness (Loveridge et al. 2016; Wearn et al. 2016), but increase the abundance of some mammals (e.g. rats, domesticated dogs; Wearn et al. 2016).
Importantly, the density of human hosts is much greater in plantations than in forests, which may support greater mosquito abundance, particularly that of anthropophilic species such as Ae.
25 aegypti (Harrington et al. 2001). Mosquitoes may also shift their feeding behaviour and preference following forest disturbance to exploit changes in hosts (Brant et al. 2016).
Invertebrate species richness also declines when forests are converted to oil palm plantations
(Ewers et al. 2015), including that of key mosquito predators. For example, predators producing strong top-down predation of mosquito larvae in forested areas (e.g. Toxorhynchites spp. mosquitoes, diving beetles) have been found to be sensitive to land-use change (Leisnham et al.
2005; Kesavaraju et al. 2008). Similarly, the richness of spiders and ants, which predate on adult mosquitoes (Strickmann et al. 1997), declines in plantations (Fayle et al. 2010), and their persistence may be further impeded by intermittent pesticide application (Kuntom et al. 2007).
However, whilst mosquitoes may experience reduced predation pressure in plantations, their populations may be affected by vector control activities, such as fogging, larvicide application and use of insecticide-treated nets.
1.5 Aedes-transmitted arboviruses in Malaysia
Dengue virus has been endemic in Malaysia since the 1970s (Rudnick 1965), with all four dengue serotypes (DENV-1, DENV-2, DENV-3 and DENV-4) co-circulating in the country (Mia et al. 2013a). Outbreaks typically occur in 6-year cycles, with four years of high dengue cases and two subsequent years of low reported cases (Packierisamy et al. 2015). Over the past few decades, there has been a dramatic increase in the intensity and magnitude of outbreaks (Nani
2015). In 2019, there were 124,777 reported cases of dengue, with 174 deaths (World Health
Organization 2019), a significant increase from the previous year. The majority of cases were reported from Selangor state, the most populous state in peninsular Malaysia, however, an increasing number of sub-urban and rural cases have been reported in recent years (Murphy et al. 2019). Notably, in 2007 a fifth dengue virus serotype was isolated from an outbreak in the
26 Bornean state of Sarawak. The serotype was identified as a member of a sylvatic lineage of
DENV-2, representing the first identification of a sylvatic DENV circulating in Asia since 1975
(Normile 2013).
Other Aedes-transmitted diseases have also re-emerged in Malaysia in recent years. In 2019,
990 cases of chikungunya virus were reported (ECDC 2019). Sporadic cases of Zika virus
(ZIKV) have also been detected in Malaysia, however no known outbreaks have occurred (Dony et al. 2018).
1.6 The Asian tiger mosquito (Aedes albopictus)
In Malaysia’s Borneo states, Ae. albopictus is the dominant Aedes-borne disease vector (Murphy et al. 2019). The so-called Asian tiger mosquito, Ae. albopictus (Skuse 1894) is an aggressive day-biting species native to the forests of Southeast Asia. It is zoophilic, opportunistically feeding on a range of vertebrate hosts, but has been found to prefer humans when available
(Paupy et al. 2009; Delatte et al. 2010). In its natural habitat, the mosquito bred in tree holes, bamboo stumps and bromeliads, but has readily adapted to breed in artificial containers present in suburban and urban areas (Hawley 1988). In the past three decades, facilitated by the global trade in used tires (Medlock et al. 2012), Ae. albopictus has experienced a formidable global expansion and now occupies every continent except Antarctica (Benedict et al. 2007). A key trait enabling this invasive potential is the capacity for Ae. albopictus to undergo photoperiodic diapause, which permits the overwintering of eggs (Hanson & Craig Jr 1994). It is currently the fastest spreading mosquito species globally (Benedict et al. 2007). Historically, Ae. albopictus was largely considered a nuisance mosquito, but in recent years has been shown to be competent for 22 known arboviruses and now ranks second only to Ae. aegypti as a vector of dengue virus (Gratz 2004). Outbreaks of chikungunya virus (CHIKV) in La Reunion (Vazeille et
27 al. 2007), India (Yergolkar et al. 2006) and Italy (Rezza et al. 2007) were also caused by Ae. albopictus.
The Ae. albopictus mosquito is widely distributed across Malaysian Borneo, readily colonising a range of natural and human-dominated habitats (Rohani et al. 2008; Lau et al. 2017). Their abundance typically increases following disturbance, and has been observed to peak at the forest edge and in plantations (Chang et al.1997; Young et al. 2017; Brown et al. 2018). However, studies using oviposit traps to assess Ae. albopictus distribution have also found high larval density in urban areas (Rohani et al. 2008; Wan-Norafikah et al. 2012). The mosquito is considered to be the dominant vector of dengue virus in Borneo (Murphy et al. 2019).
CHAPTER 2: Aims and objectives
2.1 Research aims and objectives
This thesis investigates the effects of tropical forest conversion to agriculture on the vectorial capacity of Ae. albopictus mosquitoes. Specifically, the central objectives of this work are to: (1) evaluate how the scale of environmental data determines predictions for mosquito population dynamics, (2) characterise the relationship between fine-scale temperature and mosquito life history under field conditions, and (3) incorporate field data on mosquito life-history and microclimate into a vectorial capacity framework to estimate the effects of land-use on disease transmission potential. I conduct this work in Malaysian Borneo, a region experiencing amongst the highest rates of deforestation (Fig 4) and an increasing prevalence in arbovirus incidence in recent years (Mia et al. 2013b; Lee 2015).
28
Figure 2.1 Land use and land cover change in Borneo (1973–2015). (A) Total deforestation (18.7 Mha) and remaining old-growth and selectively logged forest in December 2015; and (B) The expansion of industrial oil-palm plantations (7.8 Mha; Gaveau et al. 2016).
2.2 Thesis outline
Specifically, Chapter 1 provided a general introduction and an overview of the literature on land-use change and mosquito-borne disease emergence. In Chapter 2, I outline the study design, as well as an overview of the materials, methodology and statistical analysis used. In
Chapter 3, I use established thermal responses of Ae. albopictus and fine-scale microclimate data from an agro-forest landscape to estimate the effects of temperature on intrinsic population growth (rmax). The microclimate data included temperatures from a short but severe El Niño southern oscillation (ENSO) drought event in 2016, which interacted with land-use to determine estimated rmax. By comparing rmax estimates derived using hourly temperatures to those using daily mean temperatures, I quantify the limitations to using coarse scale environmental data for understanding mosquito population dynamics. The results from this chapter sparked an interest in characterizing the effects of microclimate on a key Ae. albopictus life-history trait, which I conducted for Chapter 4. For this chapter, I reared Ae. albopictus larvae in logged forests and
29 oil palm plantations, and paired each treatment with a data logger that recorded air temperature at 30-minute intervals. The field experiments spanned three years, including the ENSO drought of 2016, which provided valuable insights into a surprising synergistic effect of micro- and macroclimate on mosquito development rates. Having identified the importance of fine-scale environmental data for driving a key mosquito life-history trait, I sought to collect further data on the traits that determine the transmission potential of Ae. albopictus across a land-use gradient. In Chapter 5, I combine empirical data with published thermal responses to parameterize a vectorial capacity model, and compare the transmission potential of mosquitoes in logged forest and oil palm plantations. Finally, in Chapters 6 and 7, I summarise the main findings of this thesis and discuss future research directions to further improve our understanding of how environmental change affects mosquito-borne disease transmission.
30
Fig 2.2. Schematic research framework and thesis outline.
31
2.3 Study area
All fieldwork for this thesis was conducted at sites within the Stability of Altered Forest
Ecosystems (SAFE) Project located within lowland dipterocarp forest regions of East Sabah in
Malaysian Borneo (4°33’N, 117°16’E), between January and April 2016-2018. Climate in the region is typically aseasonal (Walsh & Newbery 1999), with occasional droughts that are less frequent but more severe in Eastern Borneo than in other regions of the island, and are sometimes but not always associated with the positive phase of ENSO events. The 2015-2016 El
Niño event was amongst the strongest in historical records (Null 2016), and resulted in a severe drought in Borneo (Meijide et al. 2018). The SAFE Project is a large-scale deforestation and forest fragmentation experiment, comprising 11 blocks of sampling points across a gradient of forest disturbance from unlogged primary forest through to severely degraded forest and oil palm plantations (Ewers et al. 2011). Of the 11 blocks, three are “control” blocks, comprised of continuous unlogged primary forest sites located in the Maliau Basin Conservation Area
(4◦49’N, 116◦54’E, OG1-3) logged forest located within the Kalabakan Forest Reserve and continuous oil palm plantation (4°33’N, 117°16’E; Fig 2.3). The primary forest sites are characterised by mature forest with an open understory, and hilly topography. They have a mean aboveground biomass (ABG) of 350.4 t/ha. The logged forest sites have undergone two rounds of selective logging since 1978, and have a mean aboveground biomass (ABG) of 122.4 t/ha (Pfeifer et al. 2016). Logging intensity in this area varied considerably, however most stands are heavily degraded and are characterised by a paucity of commercial timber species, few emergent trees and the dominance of pioneer and invasive vegetation (Pfeifer et al. 2016).
Oil palm plantations were either established in 2000 (OG1) or in 2006 (OG1, OG2; Ewers et al.
2011), and are characterised by monocrop stands of closed or nearly-closed canopy oil palm. The
32 plantation sites have considerably lower biomass than the forest sites (ABG = 38.1 t/ha; Pfeifer et al. 2016). The SAFE project also encompasses six experimental blocks (Fig 2.3), comprised of twice-logged forest. Each block contains sampling points in circular forest fragments of 4 x 1 ha,
2 x 10 ha, and 1 x 100 ha, currently embedded in a clear-cut matrix that will eventually be converted to oil palm plantation (Ewers et al. 2011). Mean altitude across the sampling points is
450m (median = 460m, interquartile range = 72m). Within the SAFE Project sampling blocks, sampling points are organised in a nested structure, clustered at up to four spatial scales based on a fractal sampling pattern (Ewers et al. 2011; Marsh et al. 2013).
Table 2.1. Environmental parameters for SAFE Project sites. Parameter values for mean temperature and humidity were obtained from Hardwick et al. 2011; rainfall data were obtained from Nainar et al. 2017, although no data were available for plantation sites; above-ground biomass data were obtained from Pfeifer et al. 2016. Land-use Mean Relative Mean Mean annual Above-ground category temperature humidity (%) elevation rainfall (mm) biomass (t ha- (˚C) range (m) 1) Old growth forest 23.5 92.4 324, 351 3,774 350.4
Logged forest 24.1 91.7 404, 476 2,309 122.4
Oil palm 24.9 85.5 199-517 -- 38.1 plantation
For logistical reasons, fieldwork for this thesis focused only on logged forest and mature oil palm plantation. Larval mosquito experiments were conducted at randomly selected 2nd order sites within sampling block B (N = 9), and within oil palm site OP2 (N = 9; Fig 2.3). Larval experiment sites in each land-use type were at least 172m apart, and had an altitudinal range of approximately 400m.
33 Due to time constraints, a pilot survey of sites for adult mosquito collections was conducted at the four available human settlements in logged forest (A, B) and oil palm plantation (C, D), and at four randomly selected larval sampling sites in continuous forest and plantation over the course of two weeks in January each year. Based on these initial surveys, and the lack of Ae. albopictus collected at continuous habitat sites, the four human settlements were selected for adult mosquito collections (Fig 2.3). The logged forest and oil palm adult sampling sites were approximately 17km apart, and were located within 3km of the larval experimental sites within each land-use. Both oil palm sites were located in the Selangan Batu estate, a bench-terraced plantation with a dense network of access roads. The estate has a number of small housing clusters where plantation workers and their families reside, as well as administrative offices, a guesthouse, a school and supply shops. Site A was located next to the Selangan Batu guesthouse
(4.6460˚N, 117.4514˚E), which is an exposed site surrounded by mature oil palms (planted in
2006). Site B was located outside a small roadside supply shop (4.64005˚N, 117.4491˚E). The site was also relatively exposed and surrounded by mature oil palm. Site C was the SAFE project research camp (4.7225˚N, 117.6003˚E), which is a small cleared area with multiple bungalows, surrounded by logged forest. Site D was located at an active sawmill (4.6409˚N, 117.5827˚E); a large clearing separated into a sawmill area near the adjoining road, and worker residences closer to surrounding logged forest. Adult sampling occurred near the residences. No data on human density at these sites was available, however based on observation and the presence of housing, density was likely to be highest at Sites A and C.
34
Figure 2.3. Locations of sampling sites at the SAFE Project. Adult mosquito sampling occurred at points A-D, with A and B being logged forest sites and C and D the oil palm plantation sites.
35 CHAPTER 3: Methodology
3. 1 Data sources
3.1.1 Temperature To explore the impact of fine-scale temperature on estimated (Chapter 4) and actual (Chapter 5) mosquito demographic rates, and pathogen transmission potential (Chapter 6), temperature data were obtained for each of the SAFE project sites. Data was obtained from Hardwick et al.
(2018) who examined the impact of land-use on microclimate. Above-ground climate variables were monitored using microclimate sensors (Thermochron iButtons, Maxim Integrated Systems and Lascar EL-USB 2, Salisbury, temperature accuracy <±0.5˚C) placed at all SAFE project 2nd order sampling points. Sensors were suspended at a height of 1.5m and set to record air temperature at hourly intervals. Gaps in data coverage were supplemented with data collected from Gregory et al. (2019), who placed data loggers at randomly selected 2nd order sites within old growth sampling block 2, logged forest sampling block B, and oil palm block 2 (EL-USB-2,
LASCAR electronics, Salisbury, temperature accuracy <±0.5˚C). Data used in this thesis are constrained to 2015-2018 in order to include one ENSO drought and two pre- and post-ENSO periods, and to January through to April, as this was when the 2016 ENSO drought was most severe. Anomalous temperatures due to failed data loggers were identified as those exceeding
1.5x the interquartile range of daily temperatures for each site, and were excluded from the analysis. To characterise diurnal temperature cycles, mean hourly temperatures were calculated for each second order sampling site in old growth forest, logged forest and oil palm plantations
(N = 115), and then for each land-use type.
36 3.1.2 Thermal trait responses
To estimate Ae. albopictus life-history traits at a given mean temperature, thermal response curves were obtained from Mordecai et al. (2017). The response curves were assembled using estimates from published laboratory experiments that measured the following Ae. albopictus traits at a minimum of three constant temperatures: egg-to-adult development rate, survival probability, fecundity (eggs per female per day), biting rate, adult mosquito mortality rate, extrinsic incubation rate, and vector competence. Data on vector competence and the extrinsic incubation rate were obtained from one study in which Ae. albopictus (Shanghai strain, generation >F30) were experimentally infected with dengue virus serotype 2 (DENV2; New
Guinea C virus strain) at 18, 21, 26, 31 and 36˚C (Xiao et al. 2014). Mosquitoes were considered infected if the DENV2 antigen was detected by immunofluorescent assay in the head, salivary glands, thorax or abdomen, and considered infectious if dissemination of the virus to the head and salivary glands had occurred. Data on biting rate were calculated as the reciprocal of the oviposition cycle length (1/days), and were obtained from a study in which Ae. albopictus (La
Reunion strain, generation F2) were reared in temperature treatments at 5˚C increments between 5-40˚C degrees (Delatte et al. 2009). Data on fecundity, larval development rate and survival were also obtained from this study, as well as a number of additional sources (mean =
4.5, range = 2-9). All studies used colonised Ae. albopictus mosquito strains, and typically assessed trait performance at three temperatures across a range of 15-37˚C (Alto & Juliano
2001; Wiwatanaratanabutr & Kittayapong 2006; Westbrook et al. 2010; Muturi et al. 2011). All
Ae. albopictus traits assessed were found to respond strongly to temperature and peak between
23˚C and 34˚C. Following Johnson et al. (2015), Mordecai et al. (2017) fit a thermal response to the raw data for each trait with Bayesian models fit using Markov Chain Monte Carlo (MCMC) sampling. Trait values derived from the Mordecai et al. (2017) mean model fit were then combined with temperature data from the study sites (Chapter 2.3), and used to estimate
37 temperature-dependent trait responses. Trait responses were calculated using either the mean habitat temperature, or using rate summation (Chapter 3.2) to accumulate fitness at hourly temperatures.
3. 2 Estimating mosquito population growth rates (rmax)
Rate summation was used to estimate the effects of diurnal temperature variation across the field sites on mosquito life-history traits described above (Liu et al. 1995) was used. Rate summation is defined as:
! = !(! ! !" (3)
Where a given trait (x) has a rate (r) that is determined by the temperature (T) at each hour throughout the day (t; Murdock et al. 2016; 2017). The fitness estimates for mean hourly temperature over a 24-hour period at each sampling site are then aggregated to estimate mean trait performance over the range of temperatures experienced by mosquitoes in an average day.
If temperatures exceed the thermal maximum for the trait (Tmax,) trait performance was denoted zero. A worked example is provided in Fig 3.1.
The temperature dependence of rmax was determined using an approximation of the
Amarasekare & Savage (2012) model by Cator et al. (2019). The rmax equation (2) is developed from the Euler-Lotka equation for population growth (1) and is expanded to incorporate the effects of temperature on fecundity. The model is defined thus:
38
! (!!!)( !"# !" !! ! !!! ! ! ≈ (2) !"# ! !!! !!
where a is the age of maturation, bpk is the peak reproductive rate, k is the fecundity loss rate, z is the adult mortality rate and zj is the mortality rate across all juvenile stages.
Fig 3.1. Worked example of methods used to calculate temperature-dependent trait responses.
39 Thermal performance curves were obtained from Mordecai et al. (2017; Table 3.1), who fit either
1/2 symmetric (Quadratic, -c(T-T0)(T-Tm)) or asymmetric (Brière, cT(T-T0)(Tm-T) ) functions to
Ae. albopictus life-history data (for full list of references see Appendix: Table S1). In both functions, T0 and Tm are the minimum and maximum temperature for trait performance respectively, and c is a positive rate constant. The fecundity loss parameter, k, which describes the rate of decline in fecundity after its peak, was generated from a Sharpe-Schoolfield model
(Cator et al. 2019), which uses thermodynamic parameters (e.g. activation energy, normalization constant) to provide a mechanistic interpretation of a thermal performance curve.
The effects of land-use-mediated temperature on rmax were estimated by calculating the temperature-dependent performance of each underlying life-history parameter for each data logger sampling point. The estimated trait value for each point was then used as a replicate to compare the land-use types. To compare the effects of temporal scale on model estimates, hourly and daily aggregate measures of temperature were used. The latter incorporates the effects of diurnal temperature fluctuations on trait performance and rmax (Blanford et al. 2013;
Mordecai et al. 2017) by using rate summation (Chapter 3.2).
40 Table 3.1: Thermal response of Ae. albopictus life-history parameters required for rmax model. Data from Mordecai et al. 2017, with the following trait names differing to those in Amarasekare- Savage model: bpk = TFD, a = MDR, z = 1/lf, zj = pEA. The thermal response function (Quad = quadratic) are shown, along with mean values for the critical thermal minimum (T0), thermal maximum (Tm) and a rate constant (c). Further details on the methods used by Mordecai et al. 2017 to derive these parameters are provided in Chapter 3. *The fecundity loss parameter (k) was obtained from Cator et al. 2019 and uses a Sharpe-Schoolfield function (see Appendix S1 for parameter values).
Trait Units Definition Function T0 Tm c -02 bpk Eggs per Peak individual reproductive Brière 8.02 35.65 4.88 x 10 female- rate (TFD) 1/day-1 alpha Days Age of maturation (juvenile to Brière 8.60 39.66 6.38 x 10-05 adult development time
z Day-1 Adult mortality rate Quad 13.41 31.51 1.43 x 10+0
zj Day-1 Mortality rate across all Quad 9.04 39.33 3.61 x 10-03 juvenile stages k* Fecundity loss parameter Sharpe- * * * Schoolfield
3.3 Field data
3.3.1 Larval experiments
To investigate the effects of temperature on larval development (Chapter 5), a series of experiments were conducted in which local Ae. albopictus larvae were reared at logged forest and oil palm sites. Mosquito eggs were collected from a single logged forest site (4°33’N,
117°16’E) using modified oviposit traps. Traps consisted of a dark plastic cup, filled with approximately 300mL water and lined with paper towel, and placed in shaded areas around the site. Every two days, the paper towel in the traps were inspected for eggs, and removed and replaced where eggs were present. Eggs were stored on paper towels in Ziploc bags until a sufficient number to being the experiment had been amassed. Once a sufficient number had been reached, eggs were hatched at the field camp by submerging egg papers in water for 24
41 hours. The 1st instar larvae were distributed in rearing cups (N = 3 tanks per site with 50 larvae per tank), covered with a fine-mesh cloth to prevent immigration of mosquitoes and predation, and distributed to sites in oil palm plantation (N = 9) and in logged forest (N = 9) land-use types
(Fig 2.3). Rearing tanks were visited every 1-2 days and larvae were provisioned with approximately five Tetra Cichlid Colour™ fish pellets every two days. Pupae were removed and hatched at the field camp, and the number of development days (taken as days from egg hatching to emergence) and the sex of emerging adults were recorded. In 2017 and 2018, adult emerged mosquitoes were used for fecundity assays, as described below.
3.3.2 Human landing catches
In order to determine the human to mosquito ratio (m), and to estimate adult survival (µ), host- seeking mosquitoes were collected from four human settlements across the SAFE project landscape (Fig 2.3). Details regarding the collection sites are provided in Chapter 2.3. Host- seeking mosquitoes were collected using the human landing catch (HLC) method over two seasons (February – April 2017 and 2018). Collectors sat on stools with their limbs exposed and aspirated off mosquitoes as they attempted to feed*. Despite the availability of more passive trapping methods (e.g. CDC light traps, BG sentinel traps), the HLC remains a ‘gold standard’ and is the closest means of approximating of human-vector contact rates (Tangena et al. 2015).
To minimise risks associated with the HLC method, all collectors were trained prior to involvement in the project and their health monitored for three weeks following. Mosquito collections were carried out for three consecutive days when possible, twice per land-use type, between January and April in 2017 and in 2018. We endeavoured for sampling to coincide with the crepuscular dawn (08:30 to 11:00) and dusk (15:00-18:00), however logistical constraints resulted in a later start on some days (although never later than 10:00), and infrequent dusk
* In 2017, HLCs conducted by NG only. In 2018, HLCs conducted by NG and research assistant.
42 sampling. As such, data obtained from evening sampling was excluded from this study. Ideally, collections were carried out with two collectors, with the identity of the second collector swapped after every sampling day to account for variation in host attractiveness (Takken &
Verhulst 2013). However, for logistical reasons, the number of collectors was not consistent across sampling days. As such, only HLC data collected on days where the author was the sole collector have been included in this analysis.
3.4 Derived parameters
3.4.1 Estimating adult mosquito survival
Direct ageing of field mosquitoes remains a logistical challenge. Adult survival was therefore estimated from the proportion of host-seeking parous mosquitoes (those that have laid one or more batches of eggs) collected using the HLCs (Chapter 3.3.2; Detinova 1962a). The population of adult females is thus split into two age groups representing young and old individuals. Female mosquitoes collected during HLCs were immobilised in a freezer for fifteen minutes before having their wings removed and measured using a stereoscope with an ocular eyepiece. Female mosquitoes were then dissected and their ovaries removed for observation of the tracheole coils, following Detinova (1962b). Nulliparous mosquitoes (those that have not yet laid eggs) have tightly coiled ovarioles that become irreversibly distended during the passage of eggs, creating bead-like dilatations (Appendix Fig 1). Adult survival (s) can then be estimated using the equation:
! ! = ! !! (4)
43 where M is the parous rate of the population sampled and i0 is the length of the first gonotrophic cycle (Macdonald 1952c; Davidson 1954).
Initial attempts were made to characterise the fecundity of field mosquitoes in order to obtain data on gonotrophic cycle length. Mosquitoes that emerged from the larval experiments in 2017 and in 2018 (Chapter 3.3.1) were used. Upon emergence, adult females were transferred into mating cages with 1-2 males for 3 days to mate. On the third day, females were offered a bloodmeal on a live chick for 30 minutes. Fully engorged females were transferred to transparent plastic vials containing damp filter paper and provisioned with 10% sucrose ad libitum for 20 days. After this period, all mosquitoes were dissected to identify any retained eggs, and their wings measured to determine body size (Van Handel & Day 1989).
Field attempts to determine gonotrophic cycle length were unsuccessful, due to almost 100% egg retention by female mosquitoes. Consequently, a separate laboratory experiment was conducted at Imperial College’s Silwood Park campus in August 2018 to determine gonotrophic cycle length. The Ae. albopictus mosquitoes used were obtained from a colony of mosquitoes collected in New York State, USA. Adult females were provisioned with defibrinated horse blood every 3-4 days using an artificial feeder (Hemotek Discovery Workshops, Accrington, UK) and provided with substrate for egg laying. Eggs between 2-4 weeks old were vacuum hatched in dechlorinated water over a period of 36 hours. First larval instars were reared following the same methodology as the field larval development experiments (Chapter 2.7). Larval rearing cups were placed in one of two incubators (Panasonic MLR-350H, temperature accuracy
±0.5˚C, RH accuracy ±0.5%) running fluctuating temperature treatments simulating mean hourly diurnal temperatures in either logged forest or oil palm plantation sampling sites (see
Appendix T2 for temperature schedules). Relative humidity was kept constant at 85%. Larvae were provisioned with 5 pellets of fish food (Tetra Cichlid Colour™) daily. Upon emergence,
44 adults were provisioned with 10% sugar solution ad libitum and females offered defibrinated horse blood via a Hemotek feeding system three days after emergence. Mosquitoes that did not fully imbibe were removed from the fecundity experiments. Mosquitoes that failed to feed were offered a bloodmeal each day until a successful feed was achieved or until four additional days had passed. Blood-fed females were transferred into individual laying tubes (2.9cm x 11.7cm) containing damp filter paper as an egg laying substrate. Papers were observed every day for the presence of eggs and the length of the pre-bloodmeal period and first gonotrophic cycle recorded.
3.4.2 Vectorial capacity
Using data from a combination of field, laboratory and published temperature-dependent trait responses (Table 3.2), we estimated vectorial capacity at every larval sampling site for 2017 and
2018 (N = 18). Vectorial capacity was estimated using a dengue specific framework:
!! !. !!. !. !. !!"# !" = (5) −!
where m denotes vector to human ratio, a is the daily probability of a female mosquito taking a bloodmeal, µ is the daily probability of adult mortality, b is the probability of transmission from an infectious mosquito to a susceptible human, c is the probability of transmission from an infectious human to susceptible mosquito and REI is the rate of extrinsic incubation of the pathogen (denoted EIR Mordecai et al. 2017 and Murdock et al. 2017). Parameters that were not measured in this study (REI, b, c) were estimated using temperature-dependent trait response curves from Mordecai et al. (2017; Table 3.1) and field temperature data collected at each larval
45 site. Rate summation was then used to incorporate the effects of diurnal temperature fluctuation by estimating trait responses at each hour throughout the day and summing the proportional hourly changes (Equation 3, Murdock et al. 2016; 2017).
Table 3.2 Traits and data sources for vectorial capacity parameters.
Trait Symbol Source Vector density m Measured directly in field collections
Daily probability of a female mosquito taking a bloodmeal a Calculated from laboratory experiments
Probability of transmission from an infectious mosquito to a Mordecai et al. 2017 temperature-dependent b susceptible human estimates
Probability of transmission from an infectious human to susceptible Mordecai et al. 2017 temperature-dependent c mosquito estimates
Extrinsic incubation rate of the Mordecai et al. 2017 temperature-dependent REI pathogen. estimates Estimated from parity in field collections and Daily probability of adult mortality µ gonotrophic cycle length from laboratory experiments
3.5 Statistical analysis
All analyses for this thesis were carried out in R Version 3.5 (R Core Team 2014, http://www.R- project.org). To assess the effect of land-use and ENSO on microclimate, temperature data were filtered to remove outliers due to instrument failure (N2016 = 8; N2017 = 3). Outliers were defined as temperatures readings greater than the highest ever recorded temperature in Malaysia
(40°C). Daily mean, maximum, and minimum temperature and daily temperature range were aggregated for each sampling site, and generalised linear models (GLM) fit with Gamma errors
(package ‘lme4’; link = log; Bates et al. 2012) were used to investigate the effects of land-use
46 type on each microclimate variable. In Chapter 3, ENSO drought was included as a variable along with land-use type. Models were examined for overdispersion and model selection was carried out using stepwise selection and Chi-squared statistic.
Linear and beta-logistic regressions (package: betareg; Cribari-Neto & Zeileis 2010) were used to investigate the effects of land-use and drought on trait performance. Separate models were constructed for juvenile to adult development time (alpha), peak reproductive rate (bpk) and fecundity loss parameter (k), and beta regressions were used for adult mortality rate (z) and juvenile mortality rate (zj). Land-use and drought were included as predictors in both models, with an interaction term fit between them (land use x enso), and time-aggregated trait performance fit as the response variable. Differences between the trait estimates derived using the two methods were compared with Mann-Whitney U tests. Linear models were used to determine the effect of temperature on the difference between estimates derived using the two estimation methods. Mean temperature and daily standard deviation of mean temperature were set as predictors, and the difference in trait estimates derived using rate summation and mean daily temperature fit as the response. Hourly thermal responses were then substituted into the rmax equation (2) to obtain time-averaged intrinsic population growth for each site. Linear models were fit with rmax as the response and land-use and ENSO drought as predictors, with an interaction between the terms. The rmax values estimated using the two methods were compared using a Wilcox test, and linear models used to investigate the effects of temperature on the difference between them. Posthoc significance tests using the ghlt function in the multcomp package (Hothorn et al. 2008) were used to compare treatment effects and their interactions for all linear models.
To investigate the effects of microclimate and land-use type on mosquito larvae development, mixed effects time-to-event models (Cox proportional hazards; coxme, kinship package; Terry et
47 al. 2003) were used. Proportional hazards models are a semi-parametric regression method that analyse the effect of explanatory variables on a hazard rate, defined as the instantaneous risk of an event occurring, given that it has not occurred up until that time. Here, event was defined as day of adult mosquito emergence, taken as the number of days from hatching of eggs. For this study, higher hazard rates thus denote earlier mosquito emergence times. Due to high correlation between ENSO and the microclimate variables, we retained ENSO and the interaction with land-use (ENSO x land-use) alone in the model, and mosquito sex, as fixed effects. Larval rearing tank was set as a random effect. Including sampling year as an additional random effect in the model did not improve model fit so was excluded from the final model.
Model selection was achieved using Akaike Information Criterion (AIC) and Chi-squared tests were used to ensure that hazard functions were proportional over time for each treatment.
To estimate the effects of land-use mediated temperature on vectorial capacity, parameter estimates derived from field and laboratory experiments were combined with temperature- dependent estimates from the literature (Table 4.1). Traits that were estimated from Mordecai et al. (2017) were compared between land-use types using Mann-Whitney U tests. A generalised linear negative binomial model (link = log; R package MASS; Venables 2002), with land-use type set as a predictor was used to determine whether differences in the number of host-seeking mosquitoes between oil palm plantations and logged forest sites were significant. Sampling site and year were initially nested as random effects, but neither was significant in the model. To calculate survival, parous rates were compared between land-use type and sampling years using a generalised linear model (family = Gamma, link = “log”). Laboratory data on gonotrophic cycle length were compared using the same model structure with the length of the first gonotrophic cycle in days set as the response variable and experimental treatment (forest or plantation) set as the predictor. Adult survival was then calculated from mean parity rate and gonotrophic cycle using Equation 4, with the previous model results determining whether values were aggregated
48 for each land-use or across all sites. Mean trait estimates for each land-use type were then used to calculate vectorial capacity for each sampling site from the larval development experiments
(Chapter 3.31). As data on adult survival (µ; derived from parity and gonotrophic cycle length) and mosquito density (m) were collected at different sampling sites to the larval data, the mean values per land-use were used. If the point estimates did not differ significantly between land- use types, the average trait value across both land-use types was used. The effect of land-use on vectorial capacity estimates was then explored using linear models, with land-use and sampling year as predictors.
3.6 Ethical statement
The use of HLCs was approved by the ethical committee of the Ministry of Health in Malaysia
(Approval Number: NMRR-17-3242-39250, Issued: 13 March 2018), and the Imperial College
Research Ethics Committee (ICREC Reference: 17IC3799, Issued: 22/02/17). Prior to participation in the HLCs, research assistants were provided with a participant information sheet outlining the study, and required to sign a consent form (both in Malay). Permission to use live chickens for blood provisioning was provided by the Imperial College London’s Animal
Welfare and Ethics Research Board (Approval number: 20170206A, issued 07/02/17). All documents and approved protocols provided in the Appendix (S3-6).
49
CHAPTER 4:
DISTURBANCE-MEDIATED MICROCLIMATE DETERMINES
INTRINSIC POPULATION GROWTH OF MOSQUITOES
4.1 ABSTRACT
Characterising the structure and dynamics of mosquito populations is critical to understanding mosquito-borne disease risk. Temperature is an important driver of mosquito population dynamics, due to the effects on mosquito physiology, and features frequently in statistical and mechanistic models of disease transmission. In both approaches, the climate data used is typically coarse in scale, aggregated over monthly and even yearly timescales. This is unlikely to accurately reflect the realized thermal experience of mosquitoes, particularly in heterogenous landscapes where temperature can vary over fine spatial scales. To explore how the temporal scale of temperature data used affects model predictions, we incorporated fine-scale temperature collected over three years, including one severe drought period, with temperature- dependent mosquito life-history data, into a mathematical framework of intrinsic population growth (rmax). We used a rate summation method with hourly temperature data from across an agro-forest gradient to incorporate diurnal temperature fluctuations, and compared these to estimates using mean daily temperatures. We found that estimates of mosquito traits and rmax derived from mean temperatures diverged significantly from those using rate summation, particularly when daily temperature fluctuations were high. Our results highlight the
50 importance of incorporating fine-scale environmental data for predicting the population dynamics of disease vectors, and ultimately mosquito-borne disease risk.
4.2 Introduction
The transmission of mosquito-borne diseases is inextricably linked to the presence of mosquito vectors (Ross 1911), making mosquito population dynamics an important driver of disease risk
(Garrett-Jones & Grab 1964). Classical approaches to incorporating mosquito demography into epidemiological models have historically focused on only the adult stages, and have assumed a constant population capable of transmitting disease (Macdonald 1952a). Juvenile stages have typically been ignored as they do not directly contribute to transmission, however the fact that environmental conditions experienced by larvae affect transmission-relevant adult life-history traits is now well recognized (Lyimo et al. 1992; Alto et al. 2008; Westbrook et al. 2010;
Christiansen-Jucht et al. 2014). As small-bodied ectotherms, mosquitoes have survival, growth and reproduction that are sensitive to environmental temperature (Lyimo et al. 1992; Alto &
Juliano 2001; Bayoh & Lindsay 2003; Couret et al. 2014). Mosquito physiology respond to temperature in relatively predictable ways as explained by the thermodynamics of enzyme reaction rates (Van der Have & De Jong 1996; Kingsolver 2009). At low and high temperatures, enzyme denaturation limits the catalytic reaction, which otherwise increases exponentially with increasing temperature. Combined, the two processes produce a unimodel and often left-skewed temperature response for reaction rates (Ratkowsky et al. 2005). Thermal performance curves for life-history traits (e.g. growth, survival) are typically generated from experiments measuring trait values across a range of constant temperatures and fit using either mechanistic or statistical functions (Mordecai et al. 2013; Shapiro et al. 2017). As mosquito survival, growth and reproduction determine population dynamics, mechanistic models that incorporate the thermal
51 responses of these traits can be used to predict the effects of temperature on disease risk
(Amarasekare & Coutinho 2013; Mordecai et al. 2013; Mordecai et al. 2017).
For example, the potential for transmission of dengue virus by Ae. albopictus is estimated to peak at 26.4˚C (95% CI: 25.2-27.4˚C), based on the thermal performance of the underlying mosquito life-history traits, which typically peak between 23˚C and 34˚C (Mordecai et al. 2013).
In a study using a strain of Ae. albopictus from La Réunion Island, Delatte et al. (2009) found that the rate and proportion of eggs hatching occurred at 20˚C, and declined significantly at temperatures above 30˚C. At temperatures below 10˚C, mosquito larvae were not able to survive beyond the first instars, and at 35˚C only 25% of first instars reached the adult stage
(Delatte et al. 2009). Temperatures between 20 and 35˚C do not appear to affect the pre- bloodmeal period, but was approximately seven times longer at 15˚C and 30˚C (Delatte et al.
2009). The thermal optimum for gonotrophic cycle length was relatively high compared to the other Ae. albopictus traits, and was shortest at 30˚C and 35˚C (Delatte et al. 2009). The capacity of Ae. albopictus to acquire and transmit infections (its vector competence) is also sensitive to temperature. In a study using a laboratory colony of Ae. albopictus (Shanghai strain; generation > F30), the dissemination rate of DENV-2 (New Guinea strain) increased significantly over a temperature range of 18˚C to 36˚C (Xiao et al. 2014). At 18˚C, although viral antigens were detected in the thorax-abdomen, none were detected in the mosquito salivary glands. Considerable spatial variation in the thermal response of life-history traits has been observed, due largely to the capacity of Ae. albopictus to adapt to a range of thermal environments. For example, a study conducted using an Ae. albopictus strain from North
Carolina, US, found that the thermal minimum for development was 8.8˚C (Teng & Apperson
2000). In a study conducted using a Brazilian population of Ae. albopictus, no complete development was observed at 35˚C (Monteiro et al. 2007), compared with an emergence success of 2.5% in the La Réunion study. The extent to which the thermal response of Ae. albopictus
52 varies geographically remains poorly characterised, and as such is largely excluded from population and risk models.
Models that aim to predict the effects of temperature on mosquito populations and vectorial capacity have typically ignored another important source of variation: diurnal temperature dynamics. Often the temperature data used is aggregated at monthly or even yearly scales (Craig et al. 1999; Killeen et al. 2000; Rogers & Randolph 2000). Given that mosquitoes respond physiologically to marginal shifts in temperature, coarse scale climate data is unlikely to accurately reflect their thermal environment (Paaijmans et al. 2010b). Additionally, because the thermal response of many mosquito life-history traits is non-linear, fine-scale variation in temperature may shift where peak performance occurs (Vasseur et al. 2014). The disparity between time-averaged performance in a variable environment and performance at a constant mean temperature has been demonstrated in a number of studies (Lambrechts et al. 2011;
Carrington et al. 2013c; Paaijmans et al. 2013; Beck-Johnson et al. 2017). This can be explained mathematically by ‘Jensen’s inequality’ rule, which states that where performance, !, is a non- linear function of temperature, !, then performance in a variable environment will exceed that in mean conditions if !(!) is accelerating, and vice versa where !(!) is decelerating (Ruel &
Ayres 1999; Drake 2005; Bernhardt et al. 2018).
At local scales, temperature is determined by vegetation structure (Hardwick et al. 2015;
Blonder et al. 2018). Plant canopies play a critical role in determining microclimate by moderating the amount of solar radiation, wind and water vapour that interact to cool or heat below-canopy air (Garratt 1994; Hardwick et al. 2015). Processes that alter vegetation structure
(e.g. deforestation) can therefore alter the temperatures to which mosquito vectors are exposed.
Importantly, because temperature data used to drive epidemiological models are typically
53 interpolated from weather stations in open areas, they fail to capture this heterogeneity. The effect of altered vegetation structure on microclimate varies depending on the type of habitat preceding, and the new land-use proceeding, change. For example, the conversion of closed- canopy forest to agricultural monocrop can increase mean daily temperatures by over 2˚C
(Luskin & Potts 2011; Hardwick et al. 2015); a magnitude on par with that projected under global climate change (Suggitt et al. 2011). Land-use changes that reduce canopy cover also hinder the capacity of forest to withstand changes in broader-scale climate (Aussenac 2000;
Ewers & Banks-Leite 2013). For example, during El Niño southern oscillation (ENSO) droughts in the Amazon rainforest, the greatest predictor of forest fire risk was the presence of historical logging and disturbance (Cochrane & Laurance 2002; Alencar et al. 2004). Similarly, in
Indonesian Borneo, intense logging of Dipterocarpaceae trees for timber has increased the intensity and frequency of droughts and wildfires during ENSO events (Curran et al. 1999).
Understanding how mosquitoes respond to changes in temperature requires first characterizing temperature at a scale relevant to these small-bodied ectotherms (Paaijmans et al. 2010b).
Second, because mosquito life-history traits exhibit different thermal responses and interact non-linearly to determine population dynamics, a temperature-dependent trait-based approach is critical for predicting future dynamics. Using fine-scale temperature data from across a land- use gradient and established thermal response curves, we estimate the effects that disturbance- mediated changes in microclimate have on the intrinsic population growth of Ae. albopictus, an epidemiologically important mosquito species. We compare estimates derived from mean daily temperatures to those using a time-aggregated method that accounts for diurnal temperature cycles.
54 4.3 Results
4.3.1 Effects of drought and land-use on diurnal temperature cycles
Diurnal temperature cycles differed significantly between land-use types (GLM, F(2, 120) = 56.58, p < 0.0001) in both ENSO and non-ENSO years (F(2, 117) = 45, p < 0.0001; Fig 4.1; Table 4.1). In non-ENSO periods, average daily temperatures increased significantly between old growth forest sites and oil palm plantations (β = -1.52, p < 0.0001), but were not significantly different between old growth and logged forest sites (β = -0.32, p = 0.25). The conversion of old growth forest to oil palm plantation was associated with an increase in mean temperature of ~1.4˚C, a difference consistent with, but smaller than in previous work carried out at the site (Hardwick et al. 2015). Maximum temperatures were reached between noon and 13:00 in all sites, and were significantly higher in oil palm plantations than in either logged forest or old growth forest.
Forest disturbance marginally increased maximum temperatures, but this effect was not statistically significant. Land-use significantly affected the daily minimum temperature GLM,
F(2, 117) = 14.60, p = 0.001), with the lowest temperatures observed in oil palm plantations.
However, the difference to logged forest and old growth forest was small (0.4˚C and 0.9˚C, respectively).
The occurrence of an ENSO drought in 2016 interacted with land-use to significantly increased temperatures across the land-use gradient (GLM, F(2,117) = 208.55, p < 0.0001). Average daily temperatures increased by 1.5˚C in logged forest, 1.8˚C in old growth forest and 2.1˚C in oil palm plantations. However, although maximum temperatures remained significantly higher in the plantation sites than in either forest site, the effect of ENSO was greatest in logged forest which saw a 18% increase in maximum daily temperature (Table 2).
55
Table 4.1: Ambient temperatures along a forest conversion gradient. Daily mean, minimum and maximum temperatures (mean ± standard error) in an ENSO and non-ENSO years are shown.
Climatic period Land-use Mean (˚C) Minimum (˚C) Maximum (˚C)
ENSO Old growth 25.20 ± 0.33 21.20 ± 0.05 31.60 ± 0.12
Logged forest 25.50 ± 0.60 21.20 ± 0.09 38.10 ± 0.23
Oil palm 26.90 ± 0.90 21.40 ± 0.09 40.5 ± 0.44
Non-ENSO
Old growth 23.40 ± 0.39 19.30 ± 0.04 29.10 ± 0.12
Logged forest 24.00 ± 0.50 19.50 ± 0.06 31.40 ± 0.36
Oil palm 24.80 ± 0.73 18.70 ± 0.50 35.00 ± 0.55
33
30 ENSO
27 )
C 24
° (
21 33 Non 30 Temperature − 27 ENSO
24
21 00 04 08 12 16 20 Time (hours)
Fig 4.1. Diurnal temperature cycles along a forest conversion gradient during and after an ENSO event. Long dashed lines are oil palm plantation sites (OP), short dashed lines are logged forest sites (LF), and solid lines are old growth (OG) forest sites. Shaded areas denote 95% confidence intervals.
56
4.3.2 Effects of land-use and drought on predicted performance of Ae. albopictus traits
Juvenile to adult development time
Estimates of larval development time (alpha) were significantly affected by land-use and ENSO
2 (F(3,119) = 50.22, R = 0.68, p < 0.0001, Fig 4.2). In both ENSO and non-ENSO years, predicted development time was negatively correlated with disturbance, however these differences were small. Logged forest temperatures were associated with a 10-hour decrease in development time
(β = -0.52, p = 0.0003), and oil palm plantation temperatures were associated with a 14-hour decrease in development time (β =-0.77, p < 0.001) compared to temperatures in old growth forest. However, estimates derived from logged forest and plantation temperatures were not significantly different from each other (β =-0.33, p = 0.06). Temperatures associated with the
ENSO drought significantly decreased development time in old growth forest (β= -1.55, p <
0.0001), but increased development time in logged forest sites (β = 0.58, p = 0.007) and did not significantly impact development time in oil palm plantations (β= 0.25, p = 0.32).
Estimates derived from mean daily temperatures significantly underestimated predicted larval development time compared to the time-averaged method (Mann-Whitney U test, W = 5447, Z
= -13.6, p = 0.0001). Differences in estimates were predicted by both mean temperature and daily temperature variation (taken as the standard deviation of the daily mean; F(5,117) = 1863,
R2=0.97, p < 0.0001, Fig 4.3). Increasing mean temperature was associated with a greater difference between trait values estimated using rate summation and mean daily temperature
(β=0.05, p < 0.0001), and standard deviation of the mean daily temperature decreased this difference in estimates (β= -0.32, p < 0.0001).
57
Mosquito survival
Beta regression models of daily mosquito mortality rate indicate a significant effect of land-use and drought temperatures for both larval (zj, X2 =70.4, p < 0.0001) and adult mortality (z, X2 =
39.43, p < 0.0001). In non-ENSO years, temperatures in logged forest did not significantly affect predicted larval survival (mean= 0.19 ±0.0008, SE, β= 0.0.17, p = 0.522) relative to old growth forest (mean = 0.187 ±0.0008, SE). However, plantation temperatures significantly were associated with a significant decrease in estimated larval survival probability (mean = 0.21
±0.009, SE, β= 0.114, p = 0.002). ENSO drought temperatures interacted with forest disturbance to increase predicted larval mortality rates: oil palm plantations saw an increase in predicted larval mortality of 14.5% (mean =0.237 ±0.013, SE, β= -0.14, p = 0.17) compared to
12.6% in logged forest (mean = 0.216 ±0.007, β= -0.12, p = 0.01). Larval mortality in old growth forest was not significantly affected by drought temperatures (mean = 0.193 ±0.003, SE, β=
0.037, p = 0.033). Temperature-dependent larval mortality estimates differed significantly between the two estimation methods (Mann-Whitney U, W = 13986, Z = -13.5, p < 0.0001), with mean daily temperature underestimating larval mortality compared to the rate summation method (Fig 4.3). A linear model identified a significant effect of temperature on the difference
2 between trait estimates (F(3,119) = 2204 , R = 0.98, p < 0.0001), strongly correlated with mean temperature (β = 0.0006, p < 0.0001), temperature variation (β = 0.05, p < 0.0001) and their interaction (β = -0.002, p < 0.0001, interaction = mean x sd).
Adult mosquitoes were predicted to be less sensitive to temperatures associated with land-use change and drought than juvenile stages (beta regression, X2 = 34.5, p < 0.0001). In non-ENSO years, predicted adult mortality (z) was lowest in old growth forest, but was not significantly different between logged forest (mean = 0.01 ±0.0004, SE; β = 0.07, p = 0.36) and oil palm
58 plantations (mean = 0.012 ±0.001, SE; β = 0.22, p = 0.06). Drought temperatures increased predicted adult mortality across all sites, but was only significantly greater in logged forest
(mean = 0.015 ±0.017, SE; β = 0.33, p = 0.02; Fig 4.2). Overall, estimates of adult mortality derived from time-averaged and mean temperatures differed significantly (Mann-Whitney U, W
=12246 , Z = -13.55, p < 0.0001), but were comparable at low temperatures. Temperatures above 25˚C, mean temperatures underestimated mortality rate by up to 80 days (Fig 4.3). As with larval mortality, differences in estimates were determined by temperature (F(3,119) = 8.56 ,
R2 = 0.15, p < 0.0001), with mean temperature (β = -0.004, p = 0.003) and temperature variation (β = -0.03, p =0.03), and a significant interaction between the terms (β = 0.001, p =
0.03) significant in the model.
Mosquito fecundity
Predictions of peak reproductive rate (bpk) were significantly affected by both land-use and
2 drought temperatures (LM, F(5,17) = 18.97, R = 0.42, p < 0.0001). During non-ENSO years, estimates of peak reproduction increased along the disturbance gradient. Reproductive rate was predicted to be highest in oil palm plantations (mean = 63.8 ±0.59, SE, β = 2.26, p = 0.0007) and lowest in old growth forest (mean = 60.7, ±0.43, SE, β = 3.1, p =0.001). Drought temperatures significantly decreased fecundity estimates in logged forest (β = -3.83, p = 0.003) and marginally decreased fecundity in plantations (β = -2.84, p = 0.05). However, predicted peak fecundity increased significantly under drought and old growth forest temperature conditions (mean = 67.6 ±0.679, SE; β = 6.89 , p < 0.001, Fig 6). Peak fecundity estimates derived from mean and rate summation methods differed significantly (Wilcox, W = 9508, Z = -
13.56, p = 0.0005), and differences between them were predicted by an interaction between
2 mean temperature and daily temperature variation (F(3,119) = 1058, R = 0.96, p < 0.0001). At low temperatures and small daily variations in temperature, using mean daily temperature and hourly temperature produced similar estimates, but as temperature and temperature variation
59 increased using mean daily temperature significantly overestimated peak reproductive rate (Fig
4.3).
The predicted rate of loss of fecundity (k) did not differ between old growth (mean = 0.243
±0.0006, SE) and logged forest (mean = 0.243 ± 0.0005, SE) during the non-ENSO period (β =
0.0002, p = 0.92), but both were significantly greater than in oil palm plantations (mean =
0.235 ±0.005; β = -0.008, p = 0.01). Drought conditions increased the predicted rate of fecundity loss in old growth forest (mean = 0.246 ±0.002, SE), but decreased the rate in logged forest (mean = 0.232 ±0.004, SE) and oil palm plantations (mean = 0.225 ±0.005, SE; Fig 4.2).
Mean temperature estimates significantly overestimated the fecundity loss rate and was strongly correlated with temperature variation but not mean temperature.
60
Fig 4.2. Temperature-dependant trait performance. Results are shown for estimates of juvenile to adult development time (alpha), juvenile mortality rate (zj), adult mortality rate (z), peak reproductive rate (bpk), and the fecundity loss rate (k) derived from temperatures across a land-use gradient during an ENSO and non-ENSO periods, using the rate summation method.
61
Fig 4.3. Comparison of parameter estimation methods. The effect of daily temperature variation (standard deviation of the daily mean) on the differences between mean and hourly temperature estimates are shown for A) juvenile to adult development time (alpha), B) juvenile mortality rate (zj), C) adult mortality rate (z), D) peak individual fecundity (bpk), and E) fecundity loss rate (k). Points above the dashed horizontal line indicate that using mean daily temperature over-estimates trait performance relative to time-averaged method, and vice versa for points below the line.
4.3.3 Temperature-sensitive rmax
Land-use and ENSO drought temperatures synergistically affected predicted rmax (F(5,117) = 9.19,
R2 = 0.25, p < 0.0001). Estimates based on non-ENSO temperatures were significantly greater under logged forest (mean = 0.251 ±0.002, SE, β = 0.015, p = 0.005) and oil palm plantation
(mean = 0.267 ±0.005, SE, β = 0.012, p = 0.03) temperatures compared to those in old growth forest (mean = 0.23 ±0.004, SE). However, estimates of rmax did not differ between those derived from plantations and logged forest temperature regimes (β = 0.001, p = 0.83). The
62 OG LF OP
ENSO drought temperatures increased rmax estimates across all land-use types (β = 0.014, p =
0.03), however this was only significant for old growth forest temperatures (mean = 0.282
±0.004, SE), for which rmax increased by 22%. The rmax estimates based on old growth temperatures during the drought were also predicted to be the highest overall, however the difference in estimates across the land-use types was not significantly different (p = 0.44) during this time. Estimates of rmax derived from mean daily temperature were significantly higher than those calculated using time-averaging (Mann-Whitney U, W = 3822, Z = -13.6 , p < 0.0001), and were predicted by an interaction between mean temperature and daily temperature variation
2 (F(3,119), R = 0.96, p < 0.0001; Fig 4.5). Larger daily temperature variation was associated with greater disparity between estimates derived using daily mean temperature and those from hourly temperatures.
Fig 4.4. Variation in temperature-dependent rmax estimates. Estimates were calculated using an approximation of the Amarasekare & Savage (2012) model by Cator et al. (2019). rmax was determined on an hourly basis for each sampling site (see Chapter 2.3), and averaged for each land-use type. Violin plots show the distribution of rmax estimates based on each land-use type during non-ENSO (A) and ENSO (B) periods. Horizontal bars of the boxplots denote 1.5 x the interquartile range, and the whiskers highlight outliers. Black triangles denote mean rmax derived from hourly estimations and red circles are mean rmax estimated from daily mean temperatures. Significant differences are indicated with asterisks (**).
63
Fig 4.5. Temperature-dependent rmax estimates. Estimates derived from mean daily temperatures are shown in grey, and estimates using rate summation are shown in a gradient from yellow (low standard error of mean) to red (high standard error of the mean).
4.4 Discussion
Efforts to directly infer the relationship between mosquito abundance and temperature under field conditions have had mixed success. Similarly, mechanistic approaches to estimate the effects of temperature on mosquito populations have been limited by their assumption that mosquito fitness under fluctuating temperatures approximates that under constant temperatures. By incorporating established thermal performance curves and fine-scale temperature data into a model of intrinsic population growth, we provide a mechanistic prediction of Ae. albopictus persistence under environmental change. For all models, increasing variation in daily temperatures resulted in greater differences between trait estimates derived
64 from hourly and daily mean temperatures. However, whether the models underestimated or overestimated trait performance, and the magnitude of this difference, was determined by the shape and skew of the corresponding thermal performance curves.
Our results are consistent with Jensen’s inequality where at low temperatures with small daily temperature fluctuations, the performance of a trait will move along the linear accelerating part of the curve and may be comparable to that at mean temperatures (Smallwood 1996).
However, as temperatures approach or pass the vertex of the curve, trait response becomes non- linear, diverging from the response predicted by daily mean temperatures. The effects of temperature fluctuations on trait performance will differ for traits thermal responses fit with symmetrical (e.g. quadratic) and asymmetrical (e.g. Brière) functions due to the relative slopes of the accelerating and decelerating phases of the curve. This is reflected in the estimates for adult mosquito mortality, which exhibit a symmetrical unimodal response to temperature, and for which mean daily temperature overestimates trait performance relative to hourly aggregates.
The opposite is observed for peak fecundity and the fecundity loss rate, which are defined by asymmetrical curves, where the slope of the decelerating phase of the curve is considerably steeper than the accelerating phase. The effect of temperature fluctuations is also apparent in calculations of rmax, where high temperatures and large temperature fluctuations significantly increase the difference between daily and hourly-derived estimates.
The large difference between rmax estimates using mean hourly and daily temperature in logged forest and oil palm plantations during the drought may be explained by the greater sensitivity of the model to two parameters: juvenile mortality rate (zj) and adult survival probability (z). During the ENSO drought, estimates of zj and z increased significantly in oil palm plantations due to temperatures that exceeded the thermal optimum for both traits, sometimes by 3.5˚C and 9˚C degrees, respectively. Consequently, despite an increase in the
65 performance of the other life-history traits, estimates of rmax in the plantations remained relatively unaffected by the drought.
This study builds on a growing body of work that highlights the importance of characterizing temperature-dependent trait responses at finer spatiotemporal scales than is typically done
(Paaijmans et al. 2013; Murdock et al. 2017; Gregory et al. 2019). This is particularly important for composite metrics, such as rmax, that build on the non-linear, interacting thermal responses of a suite of life-history traits. Thermal performance curves provide a critical framework for qualitatively understanding how trait performance will fare under fluctuating temperatures, as explained by Jensen’s inequality. However, improved quantitative prediction necessitates approaches that use time-averaging of traits at temporal scales relevant to the species in question (Bernhardt et al. 2018).
This study has sought to provide more accurate estimates of how changes in fine-scale temperature driven by tropical forest conversion might affect the intrinsic growth rate of a dominant mosquito vector. However, these predictions have yet to be empirically tested, and a number of factors may determine how accurately they represent mosquito responses under field conditions. Temperature responses were not available for all Ae. albopictus traits, so data from
Ae. aegypti were used. For traits with sufficient data on both species, Ae. albopictus tends to have lower thermal tolerance than Ae. aegypti (Mordecai et al. 2017). For example, the critical maximum for adult survival in Ae. aegypti is 37.73˚C, compared to 31.51˚C in Ae. albopictus
(95% CIs: 35.68-39.89; 29.14, 33.57). Given the sensitivity of rmax to adult survival, using the thermal response for Ae. aegypti would have led to an overestimation of adult survival at high temperatures, such as those experienced during the ENSO drought. Additionally, even temperature-trait relationships that have been determined using Ae. albopictus mosquitoes may vary depending on the mosquito strain used, particularly given the species capacity for cold
66 adaptation (Armbruster & Conn 2006; Leisnham et al. 2014). Characterising geographic variation in and plasticity of Ae. albopictus thermal responses will be critical to improving the predictive capacity of population models.
Under natural field settings, interactions may exist between the thermal responses of life- history traits. For example, the effects of extreme temperatures on larval stages may further decrease rmax through carry-over effects on adult traits (Evans et al. 2018). For example, larval development rate is strongly correlated with adult body size, which in turn affects adult fecundity and survival (Nasci 1986). Characterising the links between various life-history traits will be paramount to understanding the net effect of temperature on mosquito population dynamics (Chaves et al. 2014). Third, although time-aggregated temperature responses are an improvement on those derived from coarse temperature data, they may still not reflect actual temperatures to which mosquitoes are exposed. Water temperature, which is more important for larval stages, has been shown to be up to 10˚C warmer than ambient temperature under natural field settings (Paaijmans et al. 2008b). Ambient temperatures will be more relevant to adult mosquitoes, however adult stages have greater dispersal capacity than larvae and may thermoregulate via behavioural avoidance of unsuitable temperatures.
This study sought to determine the effects of temperatures associated with land-use change and drought on rmax, and did not account for additional exogenous factors that may have affected predictions. Rainfall is known to be a significant driver of mosquito population dynamics, as it determines the availability of aquatic larval sites. However, whilst the effect of rainfall on
Anopheles (Russell et al. 2011) and Culex (Yang et al. 2008) mosquitoes has been relatively well described, the effects on Ae. albopictus populations are less certain. A number of field studies have identified a positive correlation between rainfall and the abundance of Ae. albopictus mosquitoes (Ho et al. 1971; Lourenço-de-Oliveira et al. 2004). In a study conducted over five
67 years in France, Tran et al. 2013 found that mean daily temperature and precipitation most accurately predicted the abundance of Ae. albopictus eggs collected from ovitraps. However, it is worth noting that the population model was partitioned such that egg hatching was driven by rainfall in the spring, but not in the summer periods, based on the assumption that regular artificial flooding events would trigger hatching in the drier months. Indeed, model simulations where the egg hatching function is driven by rainfall throughout the year shows that Ae. albopictus populations would be largely absent during the summer, with possible local extinctions, supporting the role of artificial environments in maintaining Ae. albopictus populations during dry periods. In contrast, in a study conducted in northern Italy, accumulated precipitation was negatively correlated with the abundance of host-seeking Ae. albopictus (Roiz et al. 2010). Another study conducted in Italy found no effect of rainfall on Ae. albopictus hatching rate (Toma et al. 2003). The variation in rainfall-responses may be due to the ecology of Ae. albopictus, which breeds in artificial habitats located around human dwellings. As such, water management by humans during dry periods (e.g. increased water storage) may act to decouple Ae. albopictus population dynamics from rainfall.
Rainfall dynamics may also determine density-dependent effects as the shrinking of larval habitats during dry periods increases the density of mosquito larvae, enhancing intraspecific competition and resource limitation. This study did not incorporate density-dependent effects, or carry-over effects of life-history traits. A large body of work has demonstrated that at high larval densities, Ae. albopictus experiences increased longer development times and decreased adult body size (Mitchell Foster et al. 2012; Couret et al. 2014), which in turn is associated with decreased fecundity (Blackmore & Lord 2000) and adult survival (Nasci 1986).
Density-dependence is considered a major source of larval mortality for container-breeding mosquitoes such as Ae. albopictus and Ae. aegypti, and is frequently incorporated into population models (Focks et al. 1993; Atkinson et al. 2007; Phuc et al. 2007). In reality little is
68 understood about their density-dependent larval mortality under field conditions. Indeed, the source of density-dependent parameter values for most models comes from a single study by
Southwood et al. in 1972 (Legros et al. 2009), and attempts to characterise density-dependence in the field have been limited by confounding varying factors such as nutrition and temperature
(Seawright et al. 1977; Romero-Vivas & Falconar 2005). The breeding behaviour of Ae. albopictus may also limit density-dependent larval mortality. Females typically lay eggs above the water line, and these remain dormant until flooded with water, thus maximizing the time for larval development (Day 2016). Additionally, females perform skip oviposition, where they distribute their eggs in multiple breeding sites per gonotrophic cycle, and the degree of egg distribution has been found to correlate strongly with larval habitat quality, particularly with larval density (Davis et al. 2015).
Nonetheless, as the rmax model is particularly sensitive to juvenile and adult survival, incorporating rainfall and density-dependent mortality may produce different predictions of how land-use and drought temperatures affect estimated population growth. How these predictions change will depend on the how density-dependence is characterized, particularly whether the initial density of larvae increases (overcompensatory density dependence), decreases (undercompensatory density dependence) or does not change the final number of survivors (compensatory density dependence; Legros et al. 2009).
Finally, the predictions made in this study have not been externally validated. One way to evaluate the effects of realistic temperature fluctuations on the mosquito demographic traits underlying rmax is to conduct semi-field experiments. For example, Afrane et al. (2008) assessed the effects of microclimate on adult Anopheles gonotrophic cycle length by hanging cages of mosquitoes at sites in deforested and forested sites in the Kenya. The cages were paired with data loggers set to record hourly temperatures and humidity throughout the experimental period. Forested sites were found to be cooler, and mosquitoes housed there took almost 60%
69 longer to complete their gonotrophic cycles than those in deforested sites. In a semi-field study of Ae. albopictus larvae reared in enclosed containers across an urbanization gradient, Murdock et al. (2017) found that urban areas were generally warmer and less humid than suburban and rural sites, and that this decreased larval survival, resulted in smaller adults and lower per capita growth rates of mosquitoes.
Further insight into the relative importance of temperature and rainfall on population dynamics could be achieved through longitudinal studies of each Ae. albopictus life stage, at fine spatio-temporal scales across the forest conversion gradient, paired with fine-scale environmental data. The ability to assess the effects of drought conditions on mosquito population would be largely opportunistic; however having a long-term sampling effort in place would increase the chance of capturing the impacts of an ENSO event. Finally, data on human demography and behaviour throughout the land-use change process could provide valuable insights into drivers of mosquito population dynamics that may exacerbate or counteract the effects of environmental factors.
The caveats discussed above likely limit the predictive capacity of this work. However the aim of this study was not to provide exact estimates per se, but rather to compare the effects of temperatures associated with land-use and drought on predictions of rmax using an existing mechanistic framework. This study adds to a growing body of work highlighting the limitations of using coarse scale environmental data to predict non-linear temperature responses (Mordecai et al. 2013; Paaijmans et al. 2013; Mordecai et al. 2017; Gregory et al. 2019), particularly for composite metrics, such as rmax, for which the thermal performance curves of underlying parameters may differ considerably. However, characterising the overall impact of land-use change and drought on Ae. albopictus mosquito populations requires a better understanding of the extent to which other exogenous and endogenous factors regulate the response of key demographic traits.
70 CHAPTER 5
EL NIÑO DROUGHT AND TROPICAL FOREST CONVERSION
SYNERGISTICALLY DETERMINE MOSQUITO DEVELOPMENT
RATE†
5.1 ABSTRACT
Extreme warming events can profoundly alter the transmission dynamics of mosquito–borne diseases by affecting the physiology of mosquito vectors. At local scales, temperatures are determined largely by vegetation structure and can be dramatically altered by drivers of land- use change (e.g. forest conversion). Disturbance activities can also hinder the buffering capacity of natural habitats, making them more susceptible to seasonal climate variation and extreme weather events (e.g. droughts). Using experiments spanning three years, we demonstrate that variation in microclimates due to forest conversion dramatically increases development rates in
Ae. albopictus mosquitoes. However, this effect was mediated by an El Niño Southern
Oscillation (ENSO) drought event. In normal years, mean temperatures did not differ between land-use types, however mosquitoes reared in oil palm plantations typically emerged 2-3 days faster than in logged forests. During an ENSO drought, mean temperatures did differ between land-use types, but surprisingly this did not result in different mosquito development rates.
Driving this idiosyncratic response may be the differences in daily temperature fluctuations between the land-use types that either push mosquito larvae towards optimal development, or
† Gregory N., Ewers R.M., Chung A.Y. & Cator L.J. (2019). El Niño drought and tropical forest conversion synergistically determine mosquito development rate. Environmental Research Letters, 14, 035003. Reprinted here with permission of the publisher.
71 over the thermal optimum, thereby reducing fitness. This work highlights the importance of considering the synergistic effects of land-use and seasonal climate variations for predicting a key disease transmission-relevant mosquito trait.
5.2 Introduction
Extreme warming events can influence the distribution and dynamics of vector-borne disease transmission (Messina et al. 2015; Vincenti-Gonzalez et al. 2018). Inter-annual climate variations, such as the El Niño southern oscillation (ENSO), have been linked to outbreaks of dengue (Vincenti-Gonzalez et al. 2018) and malaria (Hashizume et al. 2009; Chaves et al. 2012) across the globe. Dramatic changes to local climates driven by land conversion (e.g. deforestation) are also associated with altered transmission dynamics of mosquito-borne diseases (Conn et al. 2002; Fornace et al. 2016). However, at both scales there is an absence of strong, consistent relationships between extrinsic drivers and patterns of disease (e.g.
Johansson et al. 2009), highlighting the need for better understanding of the interactions between the environment, mosquitoes, and pathogen transmission.
Composite metrics of mosquito-borne disease transmission (e.g. vectorial capacity, which defines the number of potentially infectious bites arising from all mosquitoes biting a single human on a single day) are typically derived using mosquito life-history traits. Small changes in key parameters (e.g. adult survival and adult feeding behaviour) can profoundly alter transmission dynamics (Garrett-Jones 1964). As mosquitoes are small-bodied ectotherms, these traits are all sensitive to ambient temperature (Lyimo et al. 1992; Delatte et al. 2009). The thermal sensitivity of a trait is often summarised as a nonlinear asymmetric curve, comprising a thermal optimum (Topt) at which the performance rate is maximised, and critical minimum
(CTmin) and maximum (CTmax) temperatures, at which performance is zero (Huey & Kingsolver
72 1989; Angilletta Jr & Angilletta 2009). Typically, the thermal sensitivity of mosquito traits is assessed experimentally, under constant temperature conditions (Alto & Juliano 2001; Bayoh &
Lindsay 2003).
In contrast, wild mosquitoes encounter temperature fluctuations as great as 20oC in many normal transmission settings (Paaijmans et al. 2009). Drivers of extreme warming events, such as ENSO and land-use change increase daily temperature fluctuations even further (Luskin &
Potts 2011). Importantly, regional climate and local weather interact non-linearly to determine the temperatures to which mosquitoes are exposed (Stenseth et al. 2003). At local scales, vegetation cover modifies solar radiation, air and soil temperature, rainfall, air humidity and wind, to create a microclimate (Aussenac 2000). In forests, dense canopies absorb relatively high amounts of solar radiation producing cool, less variable microclimates (Hardwick et al.
2015; Kovács et al. 2017). Disturbance activities that reduce canopy cover (e.g. selective logging or forest conversion) profoundly alter microclimates, increasing mean and maximum temperatures and decreasing humidity (Luskin & Potts 2011; Meijide et al. 2018). Importantly, the effects of vegetation cover on microclimate are mediated largely by the general climate
(Aussenac 2000) and different land-use types vary in their buffering capacity against extreme climate events (e.g. ENSO).
Identifying the effects of microclimate on mosquito population dynamics and pathogen transmission is important. A growing body of work has found that fluctuating temperatures differ considerably in their effects on mosquito performance when compared to equivalent constant mean temperatures (Lambrechts et al. 2011; Carrington et al. 2013a; Carrington et al.
2013c). In general, fluctuations at lower mean temperatures act to speed up rate processes, whereas the opposite occurs at high mean temperatures (Paaijmans et al. 2010a). As a consequence, models that use averaged temperatures and fluctuations collected over coarse
73 spatiotemporal scales may fail to accurately predict the effects on mosquito life-history, and consequently on the dynamics of mosquito-borne disease transmission. This has been demonstrated for larval development under laboratory conditions, where large fluctuations around a low mean temperature reduced development time by approximately five days compared to constant temperature at the same mean (Carrington et al. 2013a). Given that mosquito population dynamics are determined in part by the rate at which new adult mosquitoes are produced (Garrett-Jones 1964), and that larval development time is linked to a suite of adult traits relevant to vectorial capacity (Alto et al. 2008; Araújo & Gil 2012; Zirbel et al. 2018), understanding how this trait responds to ecologically realistic temperature fluctuations is critical to effective vector and disease control.
The immature stages (an egg stage, four larval stages and a pupal stage) of mosquitoes are particularly sensitive to the effects of temperature fluctuations as, unlike the adult stages, they are confined to their habitats. For Ae. albopictus, these habitats are typically small bodies of water, such as in tree holes in more rural areas, and in artificial containers in more human- dominated sites (Cox et al. 2007; Yee 2008). Gravid female mosquitoes lay eggs above the water line along the inside of water-holding containers, and these lay dormant until submerged; a behaviour that maximises larval development time and survival in variable environments
(Hawley 1988). Upon hatching, the larvae feed primarily on heterotrophic microorganisms and nutrients and develop through four larval stages before pupating. Pupation typically lasts for two days, during which time the mosquitoes do not feed but are still active (e.g. diving in response to stimulus; (Romoser & Lucas 1999). Each of the larval stages is strongly temperature dependent, and these effects carry over into the adult stages. For example, warmer temperatures generally increase larval development rates, but result in smaller adult mosquitoes (Rueda et al.
1990). Laboratory studies have found that larval development rate is maximised at approximately 30˚C, and that no larval development occurs below 8˚C and above 40˚C
74 (Wiwatanaratanabutr & Kittayapong 2006; Delatte et al. 2009; Muturi et al. 2011). Larval survival has a similar temperature response, with an estimated thermal optimum of between 20 and 30˚C, and bound between 9˚C and 40˚C (Teng & Apperson 2000; Wiwatanaratanabutr &
Kittayapong 2006; Delatte et al. 2009; Muturi et al. 2011). Temperature fluctuations can shift the temperature at which mean trait performance occurs (Martin & Huey 2008), however the effects of diurnal temperature fluctuations on Ae. albopictus larval rate processes remains poorly characterised.
In a series of experiments conducted across a tropical agro-forest landscape, we investigated the effects of temperature variation on mosquito life-history. Specifically, we ask: 1) how does the conversion of tropical forest to cropland alter microclimates to which mosquito larvae are exposed; 2) how does an ENSO drought affect these microclimates, and 3) how do land-use change and climate interact to affect mosquito larval development? We explore these questions using local Asian tiger mosquitoes, Ae. albopictus; a species native to Southeast Asia that has spread rapidly throughout the globe, and which is a vector for over 22 known arboviruses, including dengue, chikungunya and Zika viruses.
5.3 Results
5.3.1 Effect of ENSO drought and land-use type on microclimate
The effect of land-use type and ENSO drought on microclimate variables was explored with generalised linear models (GLMs) fit with a gamma distribution. Land-use type and ENSO drought both significantly affected the microclimate to which mosquito larvae were exposed. In non-ENSO periods, average daily temperatures in logged forest and oil palm plantation sites did
75 not differ significantly (Fig 5.3; F(1, 16) = 5.13, p = 0.78). While the average daily temperature did not differ during this period, the daily temperature ranges at oil palm sites were greater due to higher temperature fluctuations (Fig 5.3; F(1, 16) = 9.05, p = 0.008). Maximum temperatures were higher in oil palm plantation than in logged forest sites (F(1,16) = 6.41, p = 0.02), however minimum temperatures did not differ significantly between land-use types during the non-
ENSO period (F(1,16) = 0.97, p = 0.34).
Figure 5.1. Daily diurnal temperature cycles across the two land-use types. Mean hourly temperature in logged forest (solid line) and oil palm plantations (dashed line) during a normal and an ENSO period. Shaded areas denote 95% confidence intervals of the mean.
During the ENSO drought, average daily temperatures increased significantly across all land-use types (F(1,16) = 45.90, p < 0.001). However, temperatures in oil palm plantations were higher and increased by more than twice as much than in logged forest, resulting in a higher average daily temperature (Fig 10). Again, daily temperature range was significantly greater in oil palm
76 plantations than in logged forests (F(1,16) = 9.05, p = 0.008). Average maximum temperatures were significantly greater in oil palm plantation than in logged forest, with oil palm plantations
o experiencing peak temperatures of 4 C warmer on average than in logged forest (F(1,16) = 6.41, p
= 0.02). Daily minimum temperatures did not differ by land-use type during the ENSO drought
(F(1,16) = 0.06, p = 0.82). In both land-use types, average maximum temperatures were significantly greater during the ENSO drought than during non-drought periods (F(1,16) = 31.88, p < 0.001).
5.3.2 Effect of microclimate on development time of Ae. albopictus mosquito larvae
Mosquito larval development was significantly affected by microclimate and by ENSO (Fig 5.3B;
Table 3). During non-drought periods, the duration of larval development was longer in logged forests (Mann-Whitney U, W = 119450, p < 0.001). In logged forests development was 16.8
±0.29 (SE) days for females and 14.2 ±0.25 (SE) days for males, while in oil palm average development was 14.1 ±0.19 (SE) days for females and 11.3 ±0.19 (SE) days for males. The regional climatic shift associated with the ENSO drought decreased the average emergence time for mosquitoes reared in both land use types. During the ENSO drought there was no difference in the development time of larvae reared in different land use types (Mann-Whitney U, W =
7282, p = 0.15). In logged forests females took 10.9 ± 0.21 (SE) days to emerge and males took
9.29 ± 0.20 (SE) days. In oil palm plantations females took 10.3 ± 0.25 (SE) days and males took 9.09 ± 0.25 (SE) days to emerge.
Results of the Cox proportional hazards model supported the interacting effects of land-use type and ENSO drought on mosquito emergence times (Table 3). Mosquito larvae developing during an ENSO drought, that survive to adulthood, experienced a 13-fold increase in the hazard rate,
77 leading to earlier emergence times compared to those in a non-drought period (Cox Regression, z = 7.86, p < 0.001). Developing in an oil palm plantation also significantly increased the hazard rate of mosquito emergence, with mosquitoes 2.73 ±0.25 times more likely to emerge per day than in logged forest (Cox Regression, z = 3.95, p < 0.001). Emergence times were also strongly influenced by sex, with male mosquitoes emerging earlier with a daily emergence probability of
2.34 ± 0.07 times that of females (Cox Regression, z = 12.85, p < 0.001). There was also a significant interaction between ENSO drought and land-use, such that the daily probability of emergence of mosquitoes in oil palm plantations was 0.8 ±0.52 times less during the drought
(Cox Regression, z = -0.44, p = 0.0001).
Table 5.1 Effects of land-use and ENSO drought on larval development rate. Coefficients of Cox proportional hazards survival analysis estimating effects of ENSO, sex and land-use type on the timing of adult mosquito emergence, with larval rearing tank fit as a random effect (factor with X levels). Regression coefficients and hazard ratios (exponentiated coefficients) are shown. Positive coefficients imply higher risk; negative coefficients imply the opposite. P values indicate the significance of the coefficients using a Wald z statistic. Maximal model: coxme(Surv(days) ~ ENSO * landuse type + sex +(1|tank)
Model terms β e(β) se(β) z p ENSO 2.62 13.8 0.33 7.86 <0.001** Land-use type (Oil palm) 1.00 2.73 0.25 3.95 <0.001** Sex (Male) 0.85 2.34 0.07 12.85 <0.001** ENSO * Land-use type (Oil palm) -0.23 0.80 0.52 -0.44 <0.001**
78 ENSO Non−ENSO
A 27 ) C
° ( 26
25 Mean temperature Mean temperature 24
Logged forestENSOOil palm Logged forestNon−ENSOOil palm Land−use type
B 25
20
15
10 Mean development time (days) Mean development
Logged forest Oil palm Logged forest Oil palm Land−use type ENSO Non−ENSO C 1) −
0.12 Topt
Topt
0.09
0.06 Mean development rate (days rate Mean development
CTmin Ctmax CTmin Ctmax 20 30 40 20 30 40 Mean temperature (° C) Figure 5.2. Summary of land-use effects on microclimate and on mosquito larval development. (A) Mean temperatures in logged forest and oil palm plantations; (B) Mean development time of Ae. albopictus larvae (measured as number of days from eggs hatching to adults emerging), vertical boxplot lines represent values within 1.5x the interquartile range of the upper and lower quartiles, horizontal boxplot lines indicate the median and quartile values, large points denote means for logged forest (grey circles) and oil palm (black diamonds), small points are outliers; (C) Conceptual thermal performance curve demonstrating how temperature fluctuations in logged forest (grey circles) and oil palm plantations (black diamonds) may affect larval development rate (CTmin = critical minimum, CTmax = critical maximum and Topt = thermal optimum). 79 5.4 Discussion
5.4.1 Logged forests buffer the effects of ENSO drought on microclimate conditions
The results of this study indicate that oil palm plantations experience more severe changes in temperature than logged forests during an ENSO drought event. The greater buffering capacity of logged forests has been identified in a number of other studies (Frey et al. 2016; Meijide et al.
2018), and is related to canopy closure (Renaud et al. 2011) and leaf area index (Hardwick et al.
2015). However, in contrast with a previous study, mean daily temperature in the logged forest and oil palm plantation sites did not differ significantly in non-drought years (Meijide et al.
2018). The similarity in temperature observed in this study is likely due to the maturity of the oil palm plantations sampled, as older plantations have more closed canopies and more complex understory vegetation structure (Luskin & Potts 2011), and to the history of selective logging which has reduced canopy cover in the forest sites (Ewers et al. 2011).
5.4.2 Land-use and ENSO temperatures synergistically affect mosquito development
The effects of land-use type on mosquito emergence times are surprising, given the difference in their microclimates. In normal years, logged forests and oil palm plantations did not differ significantly in mean temperature, however mosquitoes had considerably higher hazard rates in oil palm plantations compared to logged forest. During an ENSO drought, when mean
80 temperatures and daily temperature ranges did differ significantly, mosquito hazard rates were almost identical between the sites. It is worth reiterating that for this study, the hazard rate refers only to adult emergence times and not to mosquito survival. It is possible that other unmeasured sources of variation may influence mosquito larval development. For example, relative humidity is expected to be greater in logged forest than in oil palm (Hardwick et al.
2015), which could decrease the surface tension of water in the experimental pots (Pérez-Díaz et al. 2012) and thereby impact mosquito pupation success (Murdock et al. 2017). Insecticide spraying may also have occurred near the plantation sites, although this occurs infrequently in plantations and was not observed during the experimental period. Additionally, larval survival was greater in plantation sites, suggesting that the effects of reduced humidity or any chemical spraying in plantations were not significant.
An alternative explanation for the absence of a temperature effect on development during the
ENSO drought is that the elevated temperatures were high enough to cause stress and reduce the fitness of developing mosquitoes (Feder & Hofmann 1999). Thermal performance curves, which describe the effects of changes in body temperature on physiological sensitivity and fitness, are non-linear, such that mean performance of a trait under fluctuating conditions may differ to the performance of that trait under mean temperature (Fig 1.3; Martin & Huey 2008).
Fluctuations that raise temperatures towards a mosquito’s optimum temperature are expected to increase performance relative to constant temperature around the same mean. However, fluctuations that push temperatures beyond the thermal optimum may result in a decline in performance (Huey & Stevenson 1979). As the slope of the performance curve is steeper above the thermal optimum, increases in body temperature beyond this are associated with a relatively rapid reduction in fitness (Angilletta Jr & Angilletta 2009). This would explain the similarity of mosquito emergence rates during the ENSO drought despite microclimates differing significantly between the land-use types. Under drought conditions, mosquitoes in both logged
81 forest and oil palm sites would have experienced temperatures closer to their thermal optimum, resulting in the observed faster development rates compared to normal years. However, temperature fluctuations in oil palm were approximately 5°C greater than in logged forest, potentially exposing mosquito larvae to deleterious temperatures for long enough to reduce growth rates. The temperature fluctuations could also explain the difference in larval development in non-ENSO years, when mean temperatures are similar in the land-use types, but greater daily temperature fluctuations in oil palm plantations result in faster larval development times.
Differences in mean larval development under constant and fluctuating temperatures have been observed for Ae. aegypti mosquitoes under laboratory conditions. Carrington et al. (2013) found that temperature fluctuations around a low temperature (16˚C) significantly increased larval development rates relative to at that constant temperature, and that the reduction was greater for large (18.6˚C) than (8.6˚C) small fluctuations. At high temperatures (36-37˚C), larval development time was lowest for the constant temperature treatment, and increased with small and large fluctuations. Larval survival also decreased significantly in high temperatures - large fluctuation treatments compared to constant and low temperature fluctuations. Similar results have been found in a study by Tun-Lin et al. (2000), where Ae. aegypti larvae were reared under constant conditions, and in a variety of artificial containers with different volumes of water in the field. Larvae reared under field conditions had greater variability in development time than their lab-reared counterparts, and generally took longer to pupate. This effect was particularly pronounced for larvae reared in small containers, which experienced greatest temperatures fluctuations. In a similar semi-field study of Ae. albopictus, Murdock et al. (2017) found that warmer temperatures associated with urbanisation decreased larval survival and faster development rates. Elevated temperatures during the summer increased larval mortality in urban sites relative to non-urban sites, which the authors posit my have been due to
82 unmeasured sources of variation across the urbanisation gradient. However, it is possible that this was due to prolonged exposure of larvae to temperatures above their thermal optimum. The authors also found that variation in relative humidity was correlated with decreased likelihood of larval survival and emergence. Although humidity was not accounted for in this study, larvae reared in plantations, which have lower humidity than logged forest (Hardwick et al. 2011) did not have significantly reduced survival.
A considerable body of work has been dedicated to elucidating the effects of temperature on the physiology of mosquitoes (Parker 1952; Lyimo et al. 1992; Alto & Juliano 2001; Delatte et al.
2009; Paaijmans et al. 2011; Couret et al. 2014), however key knowledge gaps remain in our understanding of the thermal environment experienced by these small-bodied ectotherms under natural conditions. For example, the development rates observed in this study differed considerably to those predicted from published thermal responses in Chapter 4, but only during non-ENSO years. Based on temperatures during non-ENSO years, the time to adult emergence was estimated to take 11 days in logged forest and 11.7 days in oil palm, which is significantly shorter than that observed in this study (mean = 16.8 and 14.1 days, respectively). However, predictions of larval development time based on ENSO temperatures were much closer to those observed in this study. In the ENSO year, predicted development time was 10.1 days in logged forest and 9.42 days in oil palm, compared to the observed 10.9 and 10.3 days in logged forest and oil palm, respectively. There are a number of reasons why estimates may differ between predicted and observed larval development. First, the thermal response curves used in Mordecai et al. (2017) are largely based on laboratory-colonised mosquitoes, which may have adapted to homogenous, resource-rich environments. For example, in a study of laboratory adaptation, high generation Ae. aegypti were found to have significantly faster development rates than their ancestral counterparts, but only when held under high nutrient conditions. However, that study did not assess the impact of temperature on performance (Ross et al. 2019). Ectotherms living
83 in moderately variable environments have been found to live farther away from their thermal limits than species in either stable or extremely variable environments (Tomanek 2010).
Consequently, it is possible that mosquitoes reared under constant laboratory conditions have shifted their thermal response to match rearing conditions. The insectary rearing temperatures for Ae. aegypti and albopictus is recommended to be approximately 26˚C, which is close to the mean temperatures observed in logged forest and plantations during a non-ENSO year. An additional source of uncertainty arises from the fact that the rate summation method used in
Chapter 4 assumes that mosquito body temperature accurately tracks environmental temperature. However, given that only ambient temperature data was used, it is possible that the temperatures in the larval rearing tanks were slightly cooler, particularly as they were placed in shaded areas.
5.4.3 Beyond microclimate: potential effects of land-use and drought on larval development and survival
The aim of this study was to investigate the effects of diurnal temperature fluctuations associated with land-use and drought on the development rate of a key mosquito vector.
However, in the field, a range of exogenous and endogenous factors associated with land-use change and drought will mediate larval development times. For example, the quality of breeding sites may vary between forest and plantation sites due to differences in the plant and microbial communities that determine resource availability (Yanoviak et al. 2006; Singh et al. 2015).
Nutrient-poor conditions reduce the development rate and survival of Ae. albopictus larvae
(Pugglioli et al. 2013), as well as their capacity to compete with other container-dwelling mosquitoes such as Ae. aegypti (Murrell & Juliano 2014). Altered vegetation structure in plantations may also reduce the availability of plant phytotelmata, limiting the capacity for
84 females to mediate interspecific larval competition through skip-ovipositing of eggs. High larval densities have been found to significantly increase development time (Couret et al. 2014).
However the reduction in natural breeding habitats may be compensated for by an increase of artificial containers in human-dominated plantation sites. Finally, the presence of predators can also significantly delay larval development rates (van Uitregt et al. 2012), and predation pressure may vary across a disturbance gradient. For example, the dominant predators of mosquitoes in phytotelmata in the Amazon shifted from damselfly and cranefly larvae in forest sites to Toxorhychites larvae spp. in the plantation sites (Yanoviak et al. 2006a).
Drought conditions are most likely to affect larval development and survival through the drying out of aquatic habitats. Ae. albopictus mosquitoes preferentially breed in small bodies of water
(Hawley 1988), which are particularly prone to evaporation during dry periods. However, increased water storage by humans during drought periods may counteract this, allowing Ae. albopictus larvae to persist in areas independent of rainfall (Mwangangi et al. 2009). Droughts may also reduce the abundance of predators and competitors of Ae. albopictus by exposing them to lethal temperatures and reducing the availability of habitats (Chase & Knight 2003).
This study identifies synergistic effects of land-use and climate on the development rates of small-bodied mosquitoes, highlighting the need to characterise environmental conditions at finer scales than is typically done (e.g. using regional weather stations) to fully understand how environmental change may drive mosquito population dynamics. Importantly, because vectorial capacity is a composite measure of multiple life-history parameters, each with potentially different thermal sensitivities, the effect of temperature on disease transmission will be integrated across these traits (Martin & Huey 2008). For example, mosquito larval development rate is negatively correlated with adult body size (Mohammed & Chadee 2011), which in turn can have downstream effects on adult survival and feeding behaviour (Nasci & Mitchell 1994;
85 Farjana & Nobuko 2013). Thus, although mosquitoes developing during the ENSO drought emerged earlier than those in a non-ENSO year, they are likely to be smaller, have reduced adult survival (Nasci 1986), and therefore fewer opportunities for host contact over a lifetime.
However, due to their limited teneral reserves smaller mosquitoes may take more frequent bloodmeals (Takken et al. 1998), which would increase host contacts (Farjana & Nobuko 2013).
Understanding how key environmental drivers (e.g. temperature) mediate these trade-offs in transmission ecology will be critical to prediction of disease transmission and effective disease control.
86 CHAPTER 6:
TROPICAL FOREST CONVERSION TO OIL PALM INCREASES MOSQUITO VECTORIAL CAPACITY
6.1 ABSTRACT
Land-use change is linked to the emergence and altered transmission dynamics of mosquito- borne diseases, however how the relationships between environment, mosquitoes and their pathogens drive transmission remains poorly understood. Process-based epidemiological models have played a key role in elucidating the underlying mechanisms, but the environmental data driving these models tends to be coarse in scale. Here, we used a combination of empirical and theoretical approaches derived at fine spatial scales to estimate the effects of tropical forest conversion to oil palm plantation on mosquito transmission potential. We found that larval and adult mosquito survival was greater in oil palm plantations than in logged forests, and that this drove significantly higher vectorial capacity estimates in plantation sites. The prediction that oil palm plantations would support mosquito populations with higher vectorial capacity was robust to uncertainties in our adult survival estimates. We found that unless adult lifespan in logged forest habitats exceeded 1.5 times that in oil palm plantations, vectorial capacity was always predicted to be greater in the latter. In addition, published thermal responses of key life history traits were inaccurate predictors of trait performance measured under field conditions, highlighting the need to better characterise variation in environment–mosquito interactions that determine transmission.
87
6.2 Introduction
More than 80% of the global population is at risk of infection by one or more mosquito- borne diseases (WHO 2017). Dengue virus, the most prevalent of the mosquito-borne diseases, is estimated to infect 390 million people annually (Bhatt et al. 2013), and two previously obscure arboviruses, Zika and chikungunya virus, infected over 1 million people between 2016 and 2017 (WHO 2017). Compounding this growing health burden is the almost complete dependence of disease control efforts on the management of mosquito vectors and their breeding sites, which has had little success in curbing the expansion of arboviruses such as dengue (Gubler 1998a).
Climate and environmental change (e.g. land-use) have been implicated as drivers of mosquito-borne disease emergence (Patz et al. 2000; Parham et al. 2015), however the mechanisms for how environment influences transmission dynamics remain poorly understood.
Efforts to identify environmental drivers of disease have historically relied on statistical approaches that link environmental data to disease prevalence, or on occurrence mapping of the mosquito vectors (Medley 2010; Fischer et al. 2011). Mechanistic models that incorporate mosquito and parasite life-history into a mathematical framework for transmission (e.g. vectorial capacity; Garett-Jones 1964) have shown promise (e.g. Mordecai et al. 2013), however these have often been limited by a lack of data on the key parameters (Johnson et al. 2015). For both statistical and mechanistic approaches, the environmental data used to drive models is often coarse in scale, often using average monthly temperature values or broad categories of land-use. These data may not represent the realised environments of mosquitoes or their pathogens. This mismatch between the underlying biological processes driving transmission and
88 predictor variables can have profound effects on predictions for transmission (Lambrechts et al.
2011; Paaijmans et al. 2011; Mordecai et al. 2013; Murdock et al. 2017).
Fine-scale heterogeneity in landscape structure can significantly increase the disparity between available environmental data and actual environmental conditions. For example, the temperature data underpinning many disease transmission models is obtained from the
WorldClim2 database, which uses data interpolated from weather stations located in open areas, and has a resolution of approximately 1km. In contrast, vegetation structure mediates microclimates such that temperatures under dense canopy can be ~2-3˚C cooler than in open areas (Hardwick et al. 2015; Jucker et al. 2018). Similarly, land cover data from satellite-based remote sensing (e.g. Landsat) is typically available at a resolution of 15-60m, and suffers from high cloud cover in the tropics (Dong et al. 2014). Given that landscape structure determines a suite of factors that affect mosquitoes (e.g. microclimate, availability of breeding sites), capturing this heterogeneity will be a critical component of understanding how mosquito-borne disease transmission responds to environmental change.
Amongst anthropogenic drivers of environmental change, the conversion of tropical forest to agriculture is perhaps the most ubiquitous (Laurane 2007). Over the past few decades, an increase in demand for African oil palm (Elaeis guineensis) products has resulted in the dramatic expansion of industrial plantations (Fitzherbert et al. 2008). More than 80% of global oil palm production is produced in Malaysia and Indonesia (Koh & Wilcove 2007), and has been at the expense of either selectively logged or old growth forest (Gaveau et al. 2016). The effects of land-use change on arbovirus transmission in this region remain largely unexplored, despite observations that the abundance of Ae. albopictus, an important arbovirus vector, increases following forest conversion (Chang et al. 1997; Young et al. 2017).
89
To investigate how tropical forest conversion may affect the transmission potential of mosquitoes, we measured life history traits of mosquitoes in different land use types. We then combined these data with temperature-dependent literature estimates to parameterise a vectorial capacity model. We expect life-history traits to vary between forest and oil palm due to their fundamental environmental differences. For example, reduced vegetation structure in oil palm plantations may create warmer microclimates that increase mosquito rate processes relative to logged forest habitats. Alternatively, should temperatures in plantations exceed the thermal optimum for traits, warmer conditions may reduce performance, resulting in reduced vectorial capacity. By accounting for non-linear thermal responses and interactions between different life-history traits, we provide a mechanistic basis for predicting the effects of forest conversion on disease risk, and a framework for interpreting emergent relationships between land-use and disease transmission. Finally, we compare our field data to published thermal curves to evaluate the generalizability of laboratory-derived responses.
6.3 Results
6.3.1 Mosquito reproduction
Studies of mosquito reproduction were conducted in order to obtain data on the gonotrophic cycle length, required for estimating adult survival. Of the 840 females emerging from the field larval experiments, a total of 228 female mosquitoes were offered a bloodmeal. The average pre- bloodmeal period was 2.65 ±1.23 (SD) days, although two individuals took a bloodmeal on the day of emergence. Wing length was weakly correlated with a longer pre-bloodmeal period, but this was not statistically significant (GLM, F1,27 = 3.13, p = 0.09). Feeding success was low with only 34 females fully engorging, all from the logged forest sites, and despite being left in laying
90 tubes for >20 days, not a single individual deposited eggs. To determine whether mating had successfully occurred and whether eggs were being retained, all females >20 days post- bloodmeal were dissected (Steinwascher 1982; Lounibos et al. 1990; Armbruster & Hutchinson
2002). Based on observations of the spermatheca, all blood-fed females in the fecundity experiment had successfully mated and 85% had retained eggs. The average number of eggs per female was 48.97 ±17.9 (SD), and fecundity was positively correlated with wing size (β = 96.35, df = 25, R2 = 0.38, p < 0.001).
In the complimentary laboratory experiment, the pre-bloodmeal period did not differ significantly between logged forest and oil palm plantation treatments (Mann-Whitney U, Z = -
1.90, n = 83, p = 0.06). The length of the first gonotrophic cycle also did not differ between logged forest and oil palm plantation treatments (Mann-Whitney U, Z = -0.99, n = 83, p = 0.33), averaging 8.24 ± 3.2 (±SD) days. Wing length did not have a significant effect on the total number of eggs produced (GLM, F1,92 = 3.72, p = 0.06). Gonotrophic cycle length was taken as the mean number of days of the first gonotrophic cycle length from laboratory experiments.
6.3.2 Human landing catches
A total of 276 Ae. albopictus mosquitoes collected from 46 HLC surveys in 2017, and 22 surveys in 2018. Male mosquitoes typically arrived first, and made up between 19-58% of the collections at each site (Table 6.2). However, this is likely an underestimate of the number of males swarming as only those that landed on the collectors were collected. The number of female mosquitoes collected per day ranged from 1 –6 in logged forest, and 1 – 15 in plantations, and were not significantly different between sampling years (GLM, RR = 0.01, df = 64, 95% CI: -
0.29,0.31 , p = 0.9). Overall, human landing catches conducted in plantations yielded on average
1.5 times more mosquitoes per day than those at logged forest sites (GLM, RR = 0.42, df = 64,
91
95% CI: 1.17, 2.0, p = 0.002). The mean number of mosquitoes collected per day was 3.59 ± 1.62
(SD) in logged forest, and 5.48 ±3.50 (SD) in plantation sites.
The proportion of parous mosquitoes was not significantly different between land-use types
(GLM, F1,6 = 0.2, p = 0.69) or sampling years (GLM, F1,5 = 0.98, p = 0.37). Parity rates were highest in the oil palm site OPK1 in 2017, and lowest at this site in 2018 (0.70 and 0.98, respectively). For logistical reasons (ants), the number of mosquitoes for which parity could be assessed varied between sites.
Table 6.2 Human landing catch data. Sampling sites are listed along with the number of sampling days per site (n). For the proportion of parous females, n indicates the total number of females for which parity status could be determined. Year Land-use Sampling site Total number of Mean number of Proportion type (n) females/males females per day parous (n) (95% CI) 2017 Oil palm OPK1 (11) 76/22 6.91 (4.1, 9.7) 0.98 (44) OPSB (8) 32/25 4.57 (2.5, 5.5) 0.75 (18) Logged forest SAFE (18) 58/14 3.22 (2.1, 7.0) 0.80 (41) SWML (5) 28/13 4 (2.7, 5.3) 0.82 (11) 2018 Oil palm OPK1 (6) 28/38 4.67 (2.3, 7.0) 0.70 (20) OPSB (5) 23/9 5.75 (0.03, 12) 0.72 (18) Logged forest SAFE (6) 38/7 4.75 (3.5, 7.0) 0.76 (37) SWML (5) 9/10 2.25 (0.7,3.8) 0.89 (9)
6.3.3 Estimating adult survival and temperature-dependent transmission
Survival was estimated using Equation 1. As neither gonotrophic cycle length nor parity rate differed significantly between land-use types, the mean value across all sites was used in the equation. Daily probability of adult mortality (µ) was then taken as 1 – S, and estimated to be
0.03 for all sites, corresponding to an estimated adult lifespan of 37 days.
92
None of the parameters estimated using the temperature-dependent responses from Mordecai et al. (2017) differed significantly between the sites for either sampling year (Wilcox, Z=0.43, p
= 0.663; Fig 6.1). Biting rate (a) was estimated to be 0.239 ±0.002 (mean ±SE) per day in logged forest and 0.240 ±0.004 oil palm, which translates to a mosquito feeding every four days. The probability of dengue transmission from an infectious mosquito to a susceptible human (b) was estimated to be 0.501 ±0.007 in logged forest and 0.503 ±0.0123 (mean ±SE) in oil palm. Human to mosquito to transmission probability (c) was estimated to be 0.762 in both logged forest and oil palm, with marginally greater variation in the latter (logged forest SE
= 0.005, oil palm SE = 0.008). The rate extrinsic incubation (REI) was also the same in logged forest and oil palm, and estimated to be 0.155 ±0.002.
6.3.4 Effects of forest conversion on vectorial capacity
Vectorial capacity was estimated for each sampling site from the larval experiments using field, laboratory and published data (Table 3.1). Vectorial capacity was found to be 1.2 times greater in oil palm plantations than in logged forest (F(1) = 21.7 , β = 1.18, p = 0.006) and did not differ significantly between sampling years (β= -0.20, p = 0.47). The higher vectorial capacity in oil palm plantations is driven by greater mosquito density (m) at plantation sites. To investigate the extent of this effect, vectorial capacity was also calculated without m to determine the VC relative to the vector-to-human population ratio (rVC). Removing mosquito density resulted in the rVC estimates that were not significantly different between land-use types (Wilcox, W = 6.5,
Z = 0.43, p = 0.66).
93
Figure 6.1 Effects of land-use on mosquito traits that determine vectorial capacity. Traits for which data were measured in the field are shown in Panel A, and traits estimated using published thermal performance curves are shown in Panel B. Dark grey bars are logged forest sites, and light grey are oil palm sites. Error bars represent standard error of the mean.
94
Fig 6.2 Effects of land-use on vectorial capacity. Temperature dependent estimates and field data are shown for each parameter underlying mosquito density (m) and vectorial capacity (VC) for 2017. Dark grey bars are logged forest sites, and light grey are oil palm sites. All parameter descriptions can be found in Appendix Table 1.
6.3.5 Effects of adult mortality and land-use on vectorial capacity
We found that holding all other parameters constant, if adult mortality was equal in logged forest and oil palm plantations, vectorial capacity would always be higher in plantations (Fig
6.3), with the difference increasing exponentially with decreasing mortality rate. However, below a mortality rate of 0.02, corresponding to an adult lifespan of 50 days, the difference in vectorial capacity was less than 2. Given field estimates of adult mortality in oil palm plantations (μ = 0.03), vectorial capacity in logged forest could only exceed that in plantations if
95 mortality in forests was lower than 0.02. This would translate to an adult lifespan of 50 days, compared to 33 days in oil palm.
Fig 6.3 Effects of mortality and land-use on vectorial capacity. Contour plot of differences in vectorial capacity estimates between oil palm plantations and logged forest, with reds denoting higher vectorial capacity in logged forest and blues denoting higher vectorial capacity in oil palm plantations. As you move along the x-axis, adult mosquito lifespan in logged forest increases, and the same applies for oil palm along the y-axis. The white symbol indicates adult mortality observed from the field and the dashed black line indicates estimated vectorial capacity when mortality is equal in forest and plantations. The pale pink (-0.001, 0.001] identifies where vectorial capacity switches from being higher in oil palm to higher in logged forest.
Calculating rVC with equal mortality between the sites demonstrates that rVC will not differ significantly between the land-use types within the range of most mosquito lifespans (μ < 0.007, or a lifespan < 140 days; Wilcox, W = 42180, Z = -21.2, p = 0.72).
96
6.4 Discussion
The non-linear interactions of the parameters underlying transmission have complicated efforts to disentangle the effects of land-use change on mosquito-borne disease dynamics. This study seeks to address this challenge by integrating fine-scale field data and temperature- dependent estimates from the literature into a mathematical framework for transmission potential. Based on field–derived temperature data, estimates for biting rate (a), dengue virus extrinsic incubation rate (REI) and transmission capacity of humans and mosquitoes (b and c) were not predicted to differ between the land-use types. Underlying this is the small temperature ranges of our sites relative to the temperatures defining the operational range of
Ae. albopictus. For example, using the temperature responses from Mordecai et al. (2017), the thermal optimum for biting rate is approximately 33˚C. Daily temperatures in oil palm and logged forest are all below the thermal optimum, where the response of trait performance to temperature approximates a linear response. Thus, small differences in daily temperature fluctuation may not be sufficient to produce significant differences in estimated biting rate. As trait performance declines above the thermal optimum, and typically at a faster rate than below the optimum, accurate characterisation of thermal responses will be critical to predicting the effects of temperature on traits and disease transmission.
Thermal performance curves have typically been derived under conditions that sought to minimise variation, such as in laboratories, using established, typically laboratory adapted, mosquito strains and under constant temperatures (Paaijmans et al. 2009). However, often variation in temperatures begets variation in mosquito behaviour and ecology (Cator et al.
2019), and failing to incorporate this may limit the predictive capacity of our models. For example, when we compare our field data on larval development to estimates derived from the
97 temperature-responses in Mordecai et al. (2017), we find that the published curves underestimated development by up to 8 days. This is equivalent to the time taken from egg hatching to adult emergence at some oil palm plantation sites. Additionally, the estimates of larval development derived from published curves do not differ between logged forest and oil palm sites (Chapter 4, Fig 5.4). In contrast, field experiments suggest that on average mosquitoes in oil palm plantations emerged 2 days earlier than in logged forest (Chapter 5, Fig
5.2).
Given the discordance between field estimates and estimates derived from published thermal response curves observed in this study, it is reasonable to consider whether similar differences may exist for the remaining parameters. Under laboratory conditions, large temperature fluctuations (26˚C ± 10˚C) have been shown to reduce the probability that female
Ae. aegypti become infected with dengue virus, but not the extrinsic incubation rate of the parasite (Lambrechts et al. 2011). The latter is surprising, given the temperature sensitivity of the extrinsic incubation rate (Watts et al. 1987), but may be explained by the sine function that determined the temperature regimes in the experiment. Where temperature dynamics are symmetrical about the mean, a decrease in trait performance below the mean will be offset by increased performance above the mean (Lambrechts et al. 2011). However, in the field, daytime temperatures increase sinusoidally, before decreasing exponentially in the evening, and this asymmetry may result in shorter extrinsic incubation rates when temperature fluctuations are large. As the extrinsic incubation rate appears as an exponent in the vectorial capacity equation, small changes in rates can have large effects on predicted transmission potential. However, very few field experiments have been conducted that explore the effects of temperature on extrinsic the incubation rate. A study conducted in Kenya found that the extrinsic incubation rate of
Plasmodium falciparum in Anopheles gambiae mosquitoes was greatest in deforested areas
98 where daily temperatures were highest (Afrane et al. 2008). However, we know of no similar field studies for arboviruses.
Due to the challenges of measuring biting rate of individual mosquitoes (a) in the field, studies often use the inverse of the gonotrophic cycle length as a proxy parameter (e.g. Muturi et al. 2011), based on the assumption that females only blood feed once per cycle. The thermal response curve used to calculate biting rate in this study was fit to such studies (Mordecai et al.
2017), introducing additional uncertainty around our estimates. Aedes mosquitoes are known to take multiple blood meals per gonotrophic cycle (Scott et al. 2000), sometimes foregoing sugar feeding completely (Edman et al. 1992). As this behaviour dramatically increases vectorial capacity, by assuming single blood feeding we may have significantly underestimated it in this study.
The vectorial capacity of a mosquito population is especially sensitive to changes in adult lifespan (WHO 1957; Garrett-Jones 1964). Using parity assessments, we calculated that adult lifespan was not significantly greater in oil palm plantations than logged forests. However, these estimates were limited by very modest sample sizes of host-seeking mosquitoes. Additionally, the method of using parous rates and gonotrophic cycle to assess mosquito population structure suffers from a number of limitations, including the inability to distinguish age classes beyond
“young” and “old”, and the assumption of age-independent survival (Davidson 1954). Early canonical understanding of mosquito survivorship, first articulated by MacDonald (1952), was that “environmental insults, disease and predation would kill most mosquitoes before they had an opportunity to die of old age”. However, a growing body of work has found that mosquito survivorship typically follows complex age-dependent patterns (Harrington et al.2001; Styer et al. 2007; Christiansen-Jucht et al. 2015), and that senescence is likely to be important in determining vectorial capacity. In particular, when accounting for age-dependent mortality, a
99 population composed of only young individuals will have a higher vectorial capacity relative to an older population of the same population size, due to an increased probability of surviving the extrinsic incubation period, and having a longer period over which to transmit pathogens once infectious (Styer et al. 2007).
An additional limitation of the mosquito abundance data was the limited number of sites sampled. For logistical reasons, it was not feasible to conduct simultaneous HLCs at multiple logged forest and plantation sites. Additionally, although a number of plantation villages existed, only two human settlements were available for sampling that could be classified as being surrounded by logged forest. The two sites differed considerably in area and human population size, with the sawmill site being larger and home to more people than the SAFE research camp. Evening sampling was often not feasible, so the data presented likely represent a conservative estimate Ae. albopictus, which exhibits diurnal feeding behaviour.
This study focused primarily on the effects of temperature on the parameters underlying vectorial capacity, however in reality a number of factors could affect the transmission potential of Ae. albopictus across a forest conversion gradient. As outlined in previous chapters, variation in the quality and quantity of larval habitats, adult resting habitats, access to hosts as well as vector control activities can all affect the parameters underlying vectorial capacity. Studies investigating the effects of logging and forest conversion on mosquito-borne disease have found considerable variation of risk through space and time, highlighting the complexity of the underlying mechanisms driving transmission. A key determinant of how disease transmission is altered under land-use change is the response of the main disease vectors, with the malaria system providing the best-studied examples of this. In the Amazon, malaria incidence has been found to increase in the early stages of deforestation, where logging and slash-and-burn clearing of vegetation create favourable breeding habitats for An. darlingi (Sawyer et al. 1993; Guerra et
100 al. 2006). However, once agriculture and urbanization have become established, malaria transmission tends to decrease due to improved access to healthcare facilities and reduced vector abundance (Hay et al. 2005). The effect of deforestation on malaria risk in central Africa is similar, with An. gambiae and An. funestus mosquitoes thriving in the disturbed habitats following deforestation (Gillies & de Meillon 1968). In Southeast Asia, however, malaria risk has typically been associated with dense forest cover. For example, in India in 1987, although tribal communities living in forested areas represented only 7% of the country’s population, they contributed 30% of the country’s malaria cases (Narasimham 1991). This geographic difference in risk is likely driven by the presence of many species of highly efficient vectors adapted to forest habitats in Asia, compared to few forest-interior vectors in the Amazon and Central Africa
(Guerra et al. 2006).
The distribution of hosts following forest conversion also affects disease transmission. In
Borneo, the risk of P. knowlesi malaria is positively associated with the proportion of forest loss surrounding villages during the 5-year period before reported cases (Fornace et al. 2016). This appears to be driven by increased spatial overlap between susceptible humans and macaques, the primate reservoir (Imai et al. 2014). Human migration to newly deforested areas for work can also influence the spatial patterns of transmission, particularly where migrants lack local immunity to pathogens. For example, in Tak province in western Thailand, approximately 57% of the 36,536 malaria cases detected between 2012 and 2015 were foreign migrants (Mercado et al. 2019). In South America, the risk of yellow fever (YF), a viral disease caused by a flavivirus of the Flaviviridae family (Monath & Vasconcelos 2015) is often associated with spatial overlap between humans, the dominant mosquito vectors (Haemogogus spp.) and monkey hosts (de
Almeida et al. 2019), although human-to-human transmission has become increasingly common in urban areas (Monath & Vasconcelos 2015). Disease risk at the forest frontier may be
101 further exacerbated by socioeconomic factors, such as lack of investment in public health infrastructure and poor water and waste management (Baeza et al. 2017).
Despite the dominant role of oil palm plantation expansion on deforestation in Malaysia’s
Borneo states, few studies have explicitly investigated the impacts on mosquito-borne disease risk. Using human-landing catches conducted over three years, Chang et al. (1996) found that forest conversion to oil palm plantation in Sarawak, Borneo, decreased the number of Anopheles spp. by 82%, but did not affect Ae. albopictus abundance. However, because only night- sampling was conducted, this may not have captured the magnitude of change in Ae. albopictus abundance. This is the only known study where mosquito communities were sampled throughout the forest conversion process. However, mosquito collections conducted along other forest disturbance gradients in Borneo have also found that Ae. albopictus dominates disturbed habitats (Young et al. 2017; Brown et al. 2018). In Côte d’Ivoire, where 410,000 tonnes of palm oil were produced in 2018 (FAO 2018a), plantations are associated with a significant increase in the abundance of Ae. aegypti. In particular, polyculture plantations of oil palm, rubber and banana increased the biting rate of Ae. aegypti by ~35 times compared to forest sites (Zahouli et al. 2017). It should be noted that Zahouli et al. (2017) defined biting rate as the number of mosquitoes per person/day, however in this study, biting rate is taken as the expected number of times that an individual mosquito will feed per day. Nonetheless, these studies support the observation made in this study that the density of Ae. albopictus mosquitoes is greater in plantations than in logged forest.
Much of the research conducted on forest conversion to agriculture in Borneo has focused on malaria risk, however the risk of dengue virus has also increased in the past few. Between 2010 and 2015, 11,882 dengue cases were reported in Sabah alone (Murphy et al. 2019). Surprisingly, from 2015-2016 disease incidence was higher in rural (47/100,000 people) compared to and
102 urban areas (44/100,000 people). Rural outbreaks of dengue in Malaysia have also been observed in other studies, (Chew et al. 2016)(Dhoana et al. 2018). A large volume of human movement between rural and urban areas, associated with and assisted by the plantation industry, has been implicated as a reason for the increased rural incidence. However, this may also be due to the role of the partially sylvatic Ae. albopictus mosquito as the dominant dengue vector in Borneo. In light of continued oil palm expansion (Cheng et al. 2019) and increasing population size (Department of Statistics Malaysia, 2019) in the region, understanding the mechanisms underlying dengue transmission will be critical to the development of effective control initiatives. By characterizing the Ae. albopictus traits that comprise their vectorial capacity mosquitoes, this study provides valuable insights into some of these mechanisms.
103
CHAPTER 7:
CONCLUDING REMARKS & FUTURE DIRECTIONS
The overarching goal of this thesis was to estimate the effects of tropical forest conversion on the potential for Ae. albopictus mosquitoes to transmit infection by pairing fine-scale environmental data with data on fundamental mosquito traits. In this final chapter, I highlight the main research findings and describe the scientific contributions and limitations of this work.
I conclude by discussing future research avenues to address remaining knowledge gaps.
7.1 Principal findings:
There is a growing recognition that fine-scale environmental data is necessary for understanding how the environment drives variation in mosquito traits. By complementing an existing SAFE
Project dataset with data collected during a strong ENSO drought event, this thesis identifies a strong interactive effect of forest conversion and macroclimate on local climatic conditions
(Chapter 4, Chapter 5). In contrast to previous studies, mean daily temperatures did not differ between the logged forest and oil palm plantation sites sampled during non-ENSO years, highlighting the importance of characterising habitat types beyond broad categories of land use
(e.g. “forest” or “plantation”). We observed that oil palm plantations had a significantly greater increase in mean temperature than either logged forest or old growth during the ENSO drought, indicating a negative effect of disturbance on the buffering capacity of habitats against extreme weather events.
104
The effect of land-use mediated microclimate on mosquito traits is determined by their thermal biology. That mosquito abundance and vectorial capacity are linked to temperature is not novel.
However, several important knowledge gaps have limited our understanding of how temperature affects mosquito-borne disease, and therefore of how drivers of warming (e.g. land- use) will be expected to alter disease prevalence. By estimating the thermal response of each trait underlying population growth (Chapter 4) and transmission potential (Chapter 6), this thesis provides a mechanistic interpretation of how forest conversion may alter disease risk. I demonstrate that estimates of mosquito population growth derived from mean temperature deviate considerably from those derived using methods that incorporate daily temperature fluctuations, and that this increases with the magnitude of the fluctuations (Chapter 4). These predictions are supported through empirical work that identifies an effect of daily temperature fluctuations on the development rate of Ae. albopictus larvae (Chapter 5). Interestingly, increased larval development rate in oil palm plantations did not appear to trade-off with larval survival, suggesting that the thermal optimum for the two traits are either similar or higher for the latter. Finally, I estimated that the conversion of tropical forest to oil palm plantations would significantly increase vectorial capacity; an effect driven by increased survival and mosquito density in plantations (Chapter 6). Notably, given the other parameter values, adult survival in logged forest would need to exceed 1.5% that in oil palm for vectorial capacity to be greater in forests.
7.2 Epidemiological implications
Vectorial capacity captures how local mosquito populations determine the intensity of pathogen transmission, and is an important metric for informing the development and deployment of mosquito control initiatives (Brady et al. 2016). Mechanistic approaches that realistically
105 incorporate mosquito traits will be critical to understanding baseline transmission dynamics from which the effectiveness of different intervention strategies is measured. This thesis predicts that tropical forest conversion to oil palm plantation increases Ae. albopictus vectorial capacity, driven largely by increasing larval development rates, and larval and adult mosquito survival in plantations. This has several implications for understanding the emergence and transmission of arboviruses in Malaysian Borneo. First, Ae. albopictus is a mosquito that thrives on disturbance (Burkett-Cadena & Vittor 2017). I chose to focus on this mosquito species, in part because it was the most abundant and tractable species for collection and experimentation at the field sites, but largely because of this capacity to thrive in modified landscapes. Borneo is a region that continues to experience rapid urbanisation, with many of these urban centers surrounded by forest or strongly connected to forest via logging and palm oil transport routes. The region is endemic for all four dengue serotypes, and has experienced an increasing number of dengue cases in recent years.. In 2018, a total of 3,423 dengue cases were reported from Sabah, a substantial increase from the 2,560 cases in the year previous (Murphy et al. 2019). Chikungunya has also recently re-emerged in the region (AbuBakar et al. 2007).
Ae. albopictus is suspected to be the dominant vector of both arboviruses in the region (Murphy et al. 2019). The conversion of forest to oil palm has been linked to an increased abundance of
Ae. albopictus (Chang et al. 1997), however this thesis provides a mechanistic interpretation for the increased transmission in plantations. Specifically, I find that because both larval and adult survival is much higher in oil palm plantations than in logged forests, interventions that simply target one life stage may not be sufficient to reduce vectorial capacity in oil palm plantations.
An additional concern of increased Ae. albopictus abundance and vectorial capacity in Borneo is the risk of disease spillover. Given the propensity of Ae. albopictus for zoophily, their competence for a number of arboviruses, and their capacity to thrive in both converted and forested areas, Ae. albopictus is an obvious candidate for facilitating disease spillover. This has
106 already occurred with the simian malaria, Plasmodium knowlesi, for which disease risk is closely linked to proximity to forest (Grigg et al. 2014; Fornace et al. 2016). Activities associated with the oil palm economy mean that strong transport connections exist between frontier regions (where selective logging and clear-felling occurs), plantations and urban centers. This could increase contacts between low-density, remote human populations and densely populated urban areas, where explosive outbreaks can occur.
More broadly, understanding how Ae. albopictus responds to environmental change is important given the species considerable invasive capacity and its primary role in recent outbreaks of chikungunya (CHIKV) and dengue (DENV; (Paupy et al. 2009). Studying the ecology of Ae. albopictus in its native range may provide valuable insights into the extent of the ecological plasticity of this species, which in turn may permit more accurate prediction of how environmental change might affect their abundance and transmission potential.
7.3 Future directions
7.3.1 Closing the gap between laboratory and field studies
Laboratory experiments have been an invaluable and practical tool for identifying causal relationships in mosquito ecology. However, laboratory-derived trait estimates may not accurately capture responses under field conditions. This may be due to aspects of mosquito biology (e.g. use of laboratory mosquito strains that experience reduced selective pressure, increased inbreeding depression and population bottlenecks resulting from an initial small parental population size, and loss of genetic variability and fitness compared to wild mosquitoes; Aguilar et al. 2010) or .to oversimplification of environmental conditions (e.g. the
107 use of constant temperatures). Addressing the former requires identifying which life history components are most affected by mosquito colonization in order to maintain mosquito colonies that more accurately reflect wild mosquitoes. Small field experiments, such as the larval experiments conducted in Chapter 2 of this thesis, can help to address the latter, by evaluating mosquito responses under more ecologically realistic conditions. Scaling up from these are
Semi-Field Systems (SFS), defined as enclosed environments exposed to ambient environmental conditions, which can accommodate the full lifecycle of the target vector (Knols et al. 2002;
Ferguson et al. 2008). As the number of mosquitoes in these enclosures is known or can be manipulated, SFSs can provide more accurate data on mosquito life history and demography than from either the laboratory or field (Ferguson et al. 2008). Importantly, these studies could allow for the effects of carry-over effects and trade-offs between traits (e.g. fecundity and longevity) on population dynamics to be explored.
Although estimates of mosquito abundance can be derived from models that incorporate life history (e.g. in Chapter 4), ultimately they are poor proxies for field data. Sampling of immature and adult mosquito stages over fine spatiotemporal scales will be necessary for capturing population dynamics in working landscapes. Sampling should cover each stage of land-use change, as the associated ecological effects can vary dramatically through conversion (Luskin &
Potts 2011). Additionally, studies that span long time-scales are also more likely to capture the effects of periodic, extreme weather (e.g. El Niño and La Niña events) on mosquito ecology.
This study considers plantations and forests as separate, closed systems, whereas in reality the former is typically embedded in the latter and adult mosquitoes may move freely between them
(Reinbold-Wasson et al. 2012). Forests can act as microrefugia for mosquitoes during extreme warming events (e.g. following clear-felling of trees or during drought), and if mosquitoes are partitioning their habitat use to optimize body temperature (Martin & Huey 2008), this will have profound effects on their vectorial capacity. Understanding how mosquitoes disperse
108 across working landscapes, and the effects of climate on their distribution, will be critical to understanding and predicting disease risk.
Vectorial capacity provides a useful framework for understanding the entomological components of transmission in isolation from the parasitological components (Dye 1986).
However, local models of risk would benefit from data on the prevalence of target pathogens in host-seeking mosquitoes. Sampling forest-dwelling mosquitoes for a broad range of pathogens may be important, given the emergence of a number of zoonotic infectious diseases at this interface (Diallo et al. 2003; Lee et al. 2011; Rezza & Weaver 2019). In particular, as Ae. albopictus mosquitoes exhibit both zoophilly and a strong preferences for humans (Richards et al. 2006), they have great potential to act as bridge vectors for arboviruses. In Borneo, sylvatic cycles of dengue virus, Japanese encephalitis and sindbis viruses are known to circulate amongst primates (Wolfe 2001), however the eco-epidemiology of these arbovirus strains remains largely unknown.
The gap between available climate data and local-scale microclimate conditions has significant implications for the accuracy of up-scaled vectorial capacity models. This is particularly important at forest frontiers, where conditions under the canopy can differ significantly to those in open areas. Considerable advancements have been made in our capacity to estimate environmental data over spatial scales more relevant to mosquitoes. Drones have enabled the identification of mosquito breeding habitats (Fornace et al. 2014), and models have been developed to estimate temperature and vapour pressure deficit at increasingly high resolution from vegetation structure and topography (Jucker et al. 2018). Pairing fine-scale environmental data to mosquito ecology will significantly improve inference about the effects of land-use change on disease risk.
109
7.3.2 Incorporating mosquito-borne disease risk in natural area management
Globally, agricultural expansion into forests continues to be a dominant driver of land-use change (FAO 2018b), however a concerted and positive effort has been made to protect remaining forests (Sodhi et al. 2010). The result is that a vast majority of these forests exist in a matrix of human activity (e.g. agriculture, logging; Bryan et al. 2013). Vector control should not occur at the expense of economic development, particularly as health infrastructures tend to improve as areas become wealthier. The focus of policy and public health should instead be to develop land-use and land management strategies that trade-off the benefits of economic development with the costs of disease risk. Developing models that pair field-validated, process- based estimates of vectorial capacity with socioeconomic data on human behaviour would be an important step in informing the development of such strategies. Such models would also provide insights into variation in disease risk through space and time, and could highlight where investment in vector control policies and health infrastructure could be most beneficial to mitigating the risks of mosquito-borne disease.
7.3.3 The role of long-term ecological studies in progressing mosquito-borne disease research
Large ecological studies, such as the SAFE project, have the potential to help address fundamental knowledge gaps in mosquito-borne disease research (Faust et al. 2017). A key barrier to understanding how land-use change affects disease emergence has been the spatiotemporal scale of environmental, vector and host data used. Many of these fine-scale baseline datasets already exist for the SAFE project sites as parts of separate research efforts to understand land-use effects on ecological processes. For example, eight datasets on vertebrate
110 diversity or movement across the SAFE project land-use gradient are currently available online
(https://www.safeproject.net/datasets/view_datasets?keywords=mammal
), and fine-scale, 50 x 50m resolution microclimate maps of the entire landscape are also now available (Jucker et al. 2018). Pairing these data with field studies on mosquito ecology and disease prevalence have the potential to greatly improve our understanding of the complex processes underlying mosquito-borne disease emergence under land-use change.
7.4 Conclusions
Considerable advances have been made in the global effort to control mosquito-borne disease, however the emergence and re-emergence of diseases (e.g. dengue) highlight significant remaining challenges. Closing the gap between currently achievable levels of control and eradication requires integrated approaches that account for the complexities inherent in disease transmission. This is particularly important given that land-use change continues to be a dominant force for environmental change globally, and that climate change expected to increase the frequency of extreme weather events. This thesis provides a starting point from which to make broad predictions about the effects of forest conversion to agriculture on the vectorial capacity of an epidemiologically important mosquito species. Incorporating data and tools from other fields (e.g. economics, anthropology) will be critical to disentangling the complex processes underlying transmission, however as the eradication of mosquito-borne diseases re- emergences as a goal on the global agenda, it will be pertinent to remember that “at its best, mosquito control is applied ecology” (Patterson 2016).
111
Appendix
T1. Parameter values for estimating the fecundity shape schedule. The Sharpe-Schoolfield equation was used, which the equation below. The values for activation energy (EA), deactivation energy
(ED), and the reaction rate (k) were kept consistent for for all traits. B0 corresponds to T0, and Tpk corresponds to Tm in Table 3.1.
!!! ! ! ! ! ∙(! ! !"#.!") !" = !0 ∙ ! ! ! ! ( !∙( ! ) 1 + ! ∙ ! ! !!" ! !! − !!
Parameter Description Value (units) B0 Thermal minimum 283.15 (Kelvins)
EA Activation energy 0.55 (constant) k Reaction rate 8.617-5 (constant)
ED Deactivation energy 3 (constant)
Tpk Thermal maximum 298.45 (Kelvins)
112
S2. Temperature schedules of incubators used for larval rearing. Temperatures were set to change every hour. Minimum (*) and maximum (∆) temperatures for each treatment are indicated once in each column.
Time (hours) Logged forest incubator (°C) Oil palm incubator (°C) 0 23 23 1 23 23 2 22* 22* 3 22 22 4 22 22 5 22 22 6 22 22 7 22 23 8 23 25 9 24 27 10 25 27 11 26 28
12 27 ∆ 29∆ 13 27 29 14 27 29 15 27 28 16 27 27 17 26 26 18 25 25 19 24 24 20 24 24 21 23 23 22 23 23 23 23 23
113
F1. Technique for age-grading wild mosquitoes. Images show examples of a nulliparous (A) and parous (B) female, the latter characterized by distended ovarian tracheoles and lack of skeins.
A. B,
114
REFERENCES
. Department of Statistics Malaysia (2019). Accessed: https://http://www.dosm.gov.my/. In. (2019a). Chikungunya worldwide overview. In. European Centre for DIsease Prevention and Control (2019b). Dengue Situation Update Number 584. In. World Health Organization AbuBakar S., Sam I.-C., Wong P.-F., MatRahim N., Hooi P.-S. & Roslan N. (2007). Reemergence of endemic chikungunya, Malaysia. Emerging infectious diseases, 13, 147. Adde A., Roux E., Mangeas M., Dessay N., Nacher M., Dusfour I., Girod R. & Briolant S. (2016). Dynamical mapping of Anopheles darlingi densities in a residual malaria transmission area of French Guiana by using remote sensing and meteorological data. PLoS One, 11. Addo-Bediako A., Chown S.L. & Gaston K.J. (2000). Thermal tolerance, climatic variability and latitude. Proceedings of the Royal Society of London. Series B: Biological Sciences, 267, 739- 745. Afrane Y.A., Lawson B.W., Githeko A.K. & Yan G. (2005). Effects of microclimatic changes caused by land use and land cover on duration of gonotrophic cycles of Anopheles gambiae (Diptera: Culicidae) in western Kenya highlands. Journal of medical entomology, 42, 974-980. Afrane Y.A., Little T.J., Lawson B.W., Githeko A.K. & Yan G. (2008). Deforestation and vectorial capacity of Anopheles gambiae Giles mosquitoes in malaria transmission, Kenya. Emerging Infectious Diseases, 14, 1533. Aguilar R., Simard F., Kamdem C., Shields T., Glass G.E., Garver L.S. & Dimopoulos G. (2010). Genome wide analysis of transcriptomic divergence between laboratory colony and field Anopheles gambiae mosquitoes of the M and S molecular forms. Insect molecular biology, 19, 695-705. Ah T., Mohd K.Y., Nik M.M., Joo G. & Huang G. (2009). Effect of N and K fertilizers on nutrient leaching and groundwater quality under mature oil palm in Sabah during the monsoon period. American Journal of Applied Sciences, 6, 1788-1799. Alencar A.A., Solórzano L.A. & Nepstad D.C. (2004). Modeling forest understory fires in an eastern Amazonian landscape. Ecological Applications, 14, 139-149. Alto B.W. & Juliano S.A. (2001). Temperature effects on the dynamics of Aedes albopictus (Diptera: Culicidae) populations in the laboratory. Journal of medical entomology, 38, 548-556. Alto B.W., Lounibos L.P., Mores C.N. & Reiskind M.H. (2008). Larval competition alters susceptibility of adult Aedes mosquitoes to dengue infection. Proceedings of the Royal Society of London B: Biological Sciences, 275, 463-471. Amarasekare P. & Coutinho R.M. (2013). The intrinsic growth rate as a predictor of population viability under climate warming. Journal of Animal Ecology, 82, 1240-1253. Angilletta Jr M.J. & Angilletta M.J. (2009). Thermal adaptation: a theoretical and empirical synthesis. Oxford University Press. Araújo M.d.-S. & Gil L.H.S. (2012). Larval food quantity affects development time, survival and adult biological traits that influence the vectorial capacity of Anopheles darlingi under laboratory conditions. Malaria journal, 11, 261. Arifin A., Setyono P. & Dewi W.S. (2015). Temperature effect investigation toward peat surface CO2 emissions by planting leguminous cover crops in oil palm plantations in West Kalimantan. J. Agric. Sci. Technol. B, 5, 170-183.
116
Armbruster P. & Conn J.E. (2006). Geographic variation of larval growth in North American Aedes albopictus (Diptera: Culicidae). Annals of the Entomological Society of America, 99, 1234- 1243. Armbruster P. & Hutchinson R.A. (2002). Pupal mass and wing length as indicators of fecundity in Aedes albopictus and Aedes geniculatus (Diptera: Culicidae). Journal of medical entomology, 39, 699-704. Atkinson M.P., Su Z., Alphey N., Alphey L.S., Coleman P.G. & Wein L.M. (2007). Analyzing the control of mosquito-borne diseases by a dominant lethal genetic system. Proceedings of the National Academy of Sciences, 104, 9540-9545. Aussenac G. (2000). Interactions between forest stands and microclimate: ecophysiological aspects and consequences for silviculture. Annals of Forest Science, 57, 287-301. Baeza A., Santos-Vega M., Dobson A.P. & Pascual M. (2017). The rise and fall of malaria under land- use change in frontier regions. Nature Ecology & Evolution, 1, 0108. Bakker M.A., Carreño Rocabado G. & Poorter L. (2011). Leaf economics traits predict litter decomposition of tropical plants and differ among land use types. Functional Ecology, 25, 473-483. Bates D., Maechler M. & Bolker B. (2012). lme4: linear mixed-effects models using S4 classes. R package version 0.999375-42. 2011. Google Scholar. Bayoh M. & Lindsay S. (2003). Effect of temperature on the development of the aquatic stages of Anopheles gambiae sensu stricto (Diptera: Culicidae). Bulletin of entomological research, 93, 375-381. Beck-Johnson L.M., Nelson W.A., Paaijmans K.P., Read A.F., Thomas M.B. & Bjørnstad O.N. (2017). The importance of temperature fluctuations in understanding mosquito population dynamics and malaria risk. Royal Society open science, 4, 160969. Benedict M.Q., Levine R.S., Hawley W.A. & Lounibos L.P. (2007). Spread of the tiger: global risk of invasion by the mosquito Aedes albopictus. Vector-borne and zoonotic diseases, 7, 76-85. Bernhardt J.R., Sunday J.M., Thompson P.L. & O'Connor M.I. (2018). Nonlinear averaging of thermal experience predicts population growth rates in a thermally variable environment. Proceedings of the Royal Society B: Biological Sciences, 285, 20181076. Bhatt S., Gething P.W., Brady O.J., Messina J.P., Farlow A.W., Moyes C.L., Drake J.M., Brownstein J.S., Hoen A.G. & Sankoh O. (2013). The global distribution and burden of dengue. Nature, 496, 504-507. Bhumiratana A., Sorosjinda-Nunthawarasilp P., Kaewwaen W., Maneekan P. & Pimnon S. (2013). Malaria-associated rubber plantations in Thailand. Travel Medicine and Infectious Disease, 11, 37-50. Blackmore M.S. & Lord C.C. (2000). The relationship between size and fecundity in Aedes albopictus. Journal of Vector Ecology, 25, 212-217. Blanford J.I., Blanford S., Crane R.G., Mann M.E., Paaijmans K.P., Schreiber K.V. & Thomas M.B. (2013). Implications of temperature variation for malaria parasite development across Africa. Scientific reports, 3, 1300. Blonder B., Both S., Coomes D., Elias D., Jucker T., Kvasnica J., Majalap N., Malhi Y.S., Milodowski D. & Riutta T. (2018). Extreme and highly heterogeneous microclimates in selectively logged tropical forests. Frontiers in Forests and Global Change, 1, 5. Brady O.J., Godfray H.C.J., Tatem A.J., Gething P.W., Cohen J.M., McKenzie F.E., Perkins T.A., Reiner R.C., Tusting L.S. & Sinka M.E. (2016). Vectorial capacity and vector control: reconsidering sensitivity to parameters for malaria elimination. Transactions of the Royal Society of Tropical Medicine and Hygiene, 110, 107-117. Braganza K., Karoly D.J. & Arblaster J.M. (2004). Diurnal temperature range as an index of global climate change during the twentieth century. Geophysical research letters, 31.
117
Brant H.L., Ewers R.M., Vythilingam I., Drakeley C., Benedick S. & Mumford J.D. (2016). Vertical stratification of adult mosquitoes (Diptera: Culicidae) within a tropical rainforest in Sabah, Malaysia. Malaria Journal, 15, 1. Brown R., Hing C.T., Fornace K. & Ferguson H.M. (2018). Evaluation of resting traps to examine the behaviour and ecology of mosquito vectors in an area of rapidly changing land use in Sabah, Malaysian Borneo. Parasites & vectors, 11, 346. Burkett-Cadena N.D. & Vittor A.Y. (2017). Deforestation and vector-borne disease: forest conversion favors important mosquito vectors of human pathogens. Basic and Applied Ecology. Butler R.A. (2011). Greening the world with palm oil. Online at http://news. mongabay. com/2011/0126-palm_ oil. html. Butler R.A., Koh L.P. & Ghazoul J. (2009). REDD in the red: palm oil could undermine carbon payment schemes. Conservation letters, 2, 67-73. Calliari D., Sanz K., Martinez M., Cervetto G., Gomez M. & Basso C. (2003). Comparison of the predation rate of freshwater cyclopoid copepod species on larvae of the mosquito Culex pipiens. Medical and veterinary entomology, 17, 339-342. Carrington L.B., Armijos M.V., Lambrechts L., Barker C.M. & Scott T.W. (2013a). Effects of fluctuating daily temperatures at critical thermal extremes on Aedes aegypti life-history traits. PLoS One, 8, e58824. Carrington L.B., Seifert S.N., Armijos M.V., Lambrechts L. & Scott T.W. (2013b). Reduction of Aedes aegypti vector competence for dengue virus under large temperature fluctuations. The American journal of tropical medicine and hygiene, 88, 689-697. Carrington L.B., Seifert S.N., Willits N.H., Lambrechts L. & Scott T.W. (2013c). Large diurnal temperature fluctuations negatively influence Aedes aegypti (Diptera: Culicidae) life-history traits. Journal of medical entomology, 50, 43-51. Cator L., Johnson L.R., Mordecai E.A., El Moustaid F., Smallwood T., La Deau S., Johansson M., Hudson P.J., Boots M. & Thomas M.B. (2019). More than a flying syringe: Using functional traits in vector borne disease research. bioRxiv, 501320. Chamberlain R. & Sudia W.D. (1955). The effects of temperature upon the extrinsic incubation of eastern equine encephalitis in mosquitoes. American journal of hygiene, 62. Chang M., Hii J., Buttner P. & Mansoor F. (1997). Changes in abundance and behaviour of vector mosquitoes induced by land use during the development of an oil palm plantation in Sarawak. Transactions of the Royal Society of Tropical Medicine and Hygiene, 91, 382-386. Chase J.M. & Knight T.M. (2003). Drought induced mosquito outbreaks in wetlands. Ecology Letters, 6, 1017-1024. Chaves L.F., Hashizume M., Satake A. & Minakawa N. (2012). Regime shifts and heterogeneous trends in malaria time series from Western Kenya Highlands. Parasitology, 139, 14-25. Chaves L.F., Scott T.W., Morrison A.C. & Takada T. (2014). Hot temperatures can force delayed mosquito outbreaks via sequential changes in Aedes aegypti demographic parameters in autocorrelated environments. Acta tropica, 129, 15-24. Cheng Y., Yu L., Xu Y., Lu H., Cracknell A.P., Kanniah K. & Gong P. (2019). Mapping oil palm plantation expansion in Malaysia over the past decade (2007–2016) using ALOS-1/2 PALSAR-1/2 data. International Journal of Remote Sensing, 40, 7389-7408. Chew C.H., Woon Y.L., Amin F., Adnan T.H., Wahab A.H.A., Ahmad Z.E., Bujang M.A., Hamid A.M.A., Jamal R. & Chen W.S. (2016). Rural-urban comparisons of dengue seroprevalence in Malaysia. BMC Public Health, 16, 824. Christiansen-Jucht C., Parham P.E., Saddler A., Koella J.C. & Basáñez M.-G. (2014). Temperature during larval development and adult maintenance influences the survival of Anopheles gambiae ss. Parasites & vectors, 7, 1.
118
Cochrane M.A. & Laurance W.F. (2002). Fire as a large-scale edge effect in Amazonian forests. Journal of Tropical Ecology, 18, 311-325. Conn J.E., Wilkerson R.C., Segura M.N.O., de Souza R.T., Schlichting C.D., Wirtz R.A. & Póvoa M.M. (2002). Emergence of a new neotropical malaria vector facilitated by human migration and changes in land use. The American journal of tropical medicine and hygiene, 66, 18-22. Couret J., Dotson E. & Benedict M.Q. (2014). Temperature, larval diet, and density effects on development rate and survival of Aedes aegypti (Diptera: Culicidae). PloS one, 9. Cox J., Grillet M.E., Ramos O.M., Amador M. & Barrera R. (2007). Habitat segregation of dengue vectors along an urban environmental gradient. The American journal of tropical medicine and hygiene, 76, 820-826. Craig M.H., Snow R. & le Sueur D. (1999). A climate-based distribution model of malaria transmission in sub-Saharan Africa. Parasitology today, 15, 105-111. Cribari-Neto F. & Zeileis A. (2009). Beta regression in R. Curran L.M., Caniago I., Paoli G., Astianti D., Kusneti M., Leighton M., Nirarita C. & Haeruman H. (1999). Impact of El Nino and logging on canopy tree recruitment in Borneo. Science, 286, 2184-2188. Danks H.V. (2005). Key themes in the study of seasonal adaptations in insects I. Patterns of cold hardiness. Applied Entomology and Zoology, 40, 199-211. Davidson G. (1954). Estimation of the survival-rate of anopheline mosquitoes in nature. Nature, 174, 792. Davis T.J., Kaufman P.E., Hogsette J.A. & Kline D.L. (2015). The effects of larval habitat quality on Aedes albopictus skip oviposition. Journal of the American Mosquito Control Association, 31, 321-328. Day J.F. (2016). Mosquito oviposition behavior and vector control. Insects, 7, 65. de Castro M.C., Monte-Mór R.L., Sawyer D.O. & Singer B.H. (2006). Malaria risk on the Amazon frontier. Proceedings of the National Academy of Sciences of the United States of America, 103, 2452-2457. De Réaumur R.A.F. (1740). Mémoires pour servir à l'histoire des insectes. de l'Imprimerie royale. Delatte H., Desvars A., Bouétard A., Bord S., Gimonneau G., Vourc'h G. & Fontenille D. (2010). Blood- feeding behavior of Aedes albopictus, a vector of Chikungunya on La Réunion. Vector-Borne and Zoonotic Diseases, 10, 249-258. Delatte H., Gimonneau G., Triboire A. & Fontenille D. (2009). Influence of temperature on immature development, survival, longevity, fecundity, and gonotrophic cycles of Aedes albopictus, vector of chikungunya and dengue in the Indian Ocean. Journal of medical entomology, 46, 33-41. Detinova T.S. (1962a). Age grouping methods in Diptera of medical importance with special reference to some vectors of malaria. Age Grouping Methods in Diptera of Medical Importance with special reference to some Vectors of Malaria. Detinova T.S. (1962b). Age grouping methods in Diptera of medical importance with special reference to some vectors of malaria. Age Grouping Methods in Díptera of Medical Importance with special reference to some Vectors of Malaria. Diallo M., Ba Y., Sall A.A., Diop O.M., Ndione J.A., Mondo M., Girault L. & Mathiot C. (2003). Amplification of the sylvatic cycle of dengue virus type 2, Senegal, 1999–2000: entomologic findings and epidemiologic considerations. Emerging infectious diseases, 9, 362. Dixon J.A., Gibbon D.P. & Gulliver A. (2001). Farming systems and poverty: improving farmers' livelihoods in a changing world. Food & Agriculture Org. Dong J., Xiao X., Sheldon S., Biradar C., Zhang G., Duong N.D., Hazarika M., Wikantika K., Takeuhci W. & Moore III B. (2014). A 50-m forest cover map in Southeast Asia from ALOS/PALSAR and its application on forest fragmentation assessment. PloS one, 9, e85801.
119
Dony J.J.F., Ibrahim M.Y., Jeffree M.S., Yahiro T., Taib N.M., Dominic A., Jaimin J., Giloi N., Mustapha Z.A. & Chua T.H. (2018). The first outbreak of autochthonous Zika virus in Sabah, Malaysian Borneo. PLOS Currents Outbreaks. Dowd W.W., King F.A. & Denny M.W. (2015). Thermal variation, thermal extremes and the physiological performance of individuals. Journal of Experimental Biology, 218, 1956-1967. Drake J.M. (2005). Population effects of increased climate variation. Proceedings of the Royal Society B: Biological Sciences, 272, 1823-1827. Dye C. (1986). Vectorial capacity: must we measure all its components? Parasitology Today, 2, 203- 209. Edman J.D., Strickman D., Kittayapong P. & Scott T.W. (1992). Female Aedes aegypti (Diptera: Culicidae) in Thailand Rarely Feed on Sugar. Journal of Medical Entomology, 29, 1035-1038. Edwards D.P., Magrach A., Woodcock P., Ji Y., Lim N.T.-L., Edwards F.A., Larsen T.H., Hsu W.W., Benedick S. & Khen C.V. (2014). Selective logging and oil palm: Multitaxon impacts, biodiversity indicators, and trade offs for conservation planning. Ecological Applications, 24, 2029-2049. Evans M.V., Shiau J.C., Solano N., Brindley M.A., Drake J.M. & Murdock C.C. (2018). Carry-over effects of urban larval environments on the transmission potential of dengue-2 virus. Parasites & vectors, 11, 426. Ewers R.M. & Banks-Leite C. (2013). Fragmentation impairs the microclimate buffering effect of tropical forests. PLOS one, 8, e58093. Ewers R.M., Boyle M.J., Gleave R.A., Plowman N.S., Benedick S., Bernard H., Bishop T.R., Bakhtiar E.Y., Chey V.K. & Chung A.Y. (2015). Logging cuts the functional importance of invertebrates in tropical rainforest. Nature communications, 6. Ewers R.M., Didham R.K., Fahrig L., Ferraz G., Hector A., Holt R.D., Kapos V., Reynolds G., Sinun W. & Snaddon J.L. (2011). A large-scale forest fragmentation experiment: the Stability of Altered Forest Ecosystems Project. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 366, 3292-3302. Ewing D.A., Cobbold C.A., Purse B., Nunn M. & White S.M. (2016). Modelling the effect of temperature on the seasonal population dynamics of temperate mosquitoes. Journal of theoretical biology, 400, 65-79. FAO (2018a). FAO statistics: Production and crops. In. The Food and Agriculture Organization of the United Nations. Available at: http://www.fao.org/faostat/en/ - data/QC. FAO F. (2018b). The State of the World s Forests 2018 Forest pathways to sustainable development. Farjana T. & Nobuko T. (2013). Multiple blood feeding and host-seeking behaviour in Aedes aegypti and Aedes albopictus (Diptera: Culicidae). Journal of Medical Entomology, 50, 838-846. Faust C.L., Dobson A.P., Gottdenker N., Bloomfield L.S., McCallum H.I., Gillespie T.R., Diuk-Wasser M. & Plowright R.K. (2017). Null expectations for disease dynamics in shrinking habitat: dilution or amplification? Philosophical Transactions of the Royal Society B: Biological Sciences, 372, 20160173. Feder M.E. & Hofmann G.E. (1999). Heat-shock proteins, molecular chaperones, and the stress response: evolutionary and ecological physiology. Annual review of physiology, 61, 243-282. Ferguson H.M., Dornhaus A., Beeche A., Borgemeister C., Gottlieb M., Mulla M.S., Gimnig J.E., Fish D. & Killeen G.F. (2010). Ecology: a prerequisite for malaria elimination and eradication. Ferguson H.M., Ng'habi K.R., Walder T., Kadungula D., Moore S.J., Lyimo I., Russell T.L., Urassa H., Mshinda H. & Killeen G.F. (2008). Establishment of a large semi-field system for experimental study of African malaria vector ecology and control in Tanzania. Malaria Journal, 7, 158.
120
Fick S.E. & Hijmans R.J. (2017). WorldClim 2: new 1 km spatial resolution climate surfaces for global land areas. International journal of climatology, 37, 4302-4315. Filippis A., Schatzmayr H.G., Nicolai C., Baran M., Miagostovich M.P., Sequeira P.C. & Nogueira R. (2001). Jungle yellow fever, Rio de Janeiro. Emerging infectious diseases, 7, 484. Fillinger U., Sonye G., Killeen G.F., Knols B.G. & Becker N. (2004). The practical importance of permanent and semipermanent habitats for controlling aquatic stages of Anopheles gambiae sensu lato mosquitoes: operational observations from a rural town in western Kenya. Tropical medicine & international health, 9, 1274-1289. Fischer D., Thomas S.M., Niemitz F., Reineking B. & Beierkuhnlein C. (2011). Projection of climatic suitability for Aedes albopictus Skuse (Culicidae) in Europe under climate change conditions. Global and Planetary Change, 78, 54-64. Fitzherbert E.B., Struebig M.J., Morel A., Danielsen F., Brühl C.A., Donald P.F. & Phalan B. (2008). How will oil palm expansion affect biodiversity? Trends in ecology & evolution, 23, 538-545. Focks D.A., Haile D., Daniels E. & Mount G.A. (1993). Dynamic life table model for Aedes aegypti (Diptera: Culicidae): analysis of the literature and model development. Journal of medical entomology, 30, 1003-1017. Fornace K., Abidin T.R., Alexander N., Brock P., Grigg M.J., Murphy A., William T., Menon J., Drakeley C.J. & Cox J. (2016). Association between landscape factors and spatial patterns of Plasmodium knowlesi infections in Sabah, Malaysia. Emerging Infectious Diseases, 22, 1-8. Fornace K.M., Drakeley C.J., William T., Espino F. & Cox J. (2014). Mapping infectious disease landscapes: unmanned aerial vehicles and epidemiology. Trends in parasitology, 30, 514- 519. Franklinos L.H., Jones K.E., Redding D.W. & Abubakar I. (2019). The effect of global change on mosquito-borne disease. The Lancet Infectious Diseases. Frey S.J., Hadley A.S., Johnson S.L., Schulze M., Jones J.A. & Betts M.G. (2016). Spatial models reveal the microclimatic buffering capacity of old-growth forests. Science Advances, 2, e1501392. Garratt J.R. (1994). The atmospheric boundary layer. Earth-Science Reviews, 37, 89-134. Garrett-Jones C. (1964). The prognosis for interruption of malaria transmission through assessment of the mosquito's vectorial capacity. Garrett-Jones C. & Grab B. (1964). The assessment of insecticidal impact on the malaria mosquito's vectorial capacity, from data on the proportion of parous females. Bulletin of the World Health Organization, 31, 71. Gaveau D.L., Sheil D., Husnayaen M.A.S., Arjasakusuma S., Ancrenaz M., Pacheco P. & Meijaard E. (2016). Rapid conversions and avoided deforestation: examining four decades of industrial plantation expansion in Borneo. Scientific Reports, 6. Geerts B. (2003). Empirical estimation of the monthly-mean daily temperature range. Theoretical and Applied Climatology, 74, 145-165. Gerberg E.J., Barnard D.R. & Ward R.A. (1994). Manual for mosquito rearing and experimental techniques. American Mosquito Control Association, Inc. Gibert J.P., Dell A.I., DeLong J.P. & Pawar S. (2015). Scaling-up trait variation from individuals to ecosystems. In: Advances in Ecological Research. Elsevier, pp. 1-17. Gimnig J.E., Ombok M., Otieno S., Kaufman M.G., Vulule J.M. & Walker E.D. (2002). Density- dependent development of Anopheles gambiae (Diptera: Culicidae) larvae in artificial habitats. Journal of Medical Entomology, 39, 162-172. Githeko A.K., Lindsay S.W., Confalonieri U.E. & Patz J.A. (2000). Climate change and vector-borne diseases: a regional analysis. Bulletin of the World Health Organization, 78, 1136-1147. Godfray H.C.J. (2013). Mosquito ecology and control of malaria. Journal of Animal Ecology, 82, 15-25. Gottdenker N.L., Streicker D.G., Faust C.L. & Carroll C. (2014). Anthropogenic land use change and infectious diseases: a review of the evidence. Ecohealth, 11, 619-632.
121
Gratz N. (2004). Critical review of the vector status of Aedes albopictus. Medical and veterinary entomology, 18, 215-227. Gray E.M. & Bradley T.J. (2005). Physiology of desiccation resistance in Anopheles gambiae and Anopheles arabiensis. The American journal of tropical medicine and hygiene, 73, 553-559. Gray E.M., Rocca K.A., Costantini C. & Besansky N.J. (2009). Inversion 2La is associated with enhanced desiccation resistance in Anopheles gambiae. Malaria journal, 8, 1. Gregory N., Ewers R.M., Chung A.Y. & Cator L.J. (2019). El Niño drought and tropical forest conversion synergistically determine mosquito development rate. Environmental Research Letters, 14, 035003. Grigg M., William T., Drakeley C., Jelip J., von Seidlein L., Barber B., Fornace K., Anstey N., Yeo T. & Cox J. (2014). Factors that are associated with the risk of acquiring Plasmodium knowlesi malaria in Sabah, Malaysia: a case-control study protocol. BMJ open, 4, e006004. Grobbelaar A.A., Weyer J., Moolla N., van Vuren P.J., Moises F. & Paweska J.T. (2016). Resurgence of yellow fever in Angola, 2015–2016. Emerging infectious diseases, 22, 1854. Gubler D. (1988). Epidemiology of arthropod-borne viral diseases. In. Boca Raton, USA: CRC Press Inc. Gubler D.J. (1998a). Dengue and dengue hemorrhagic fever. Clinical microbiology reviews, 11, 480- 496. Gubler D.J. (1998b). Resurgent vector-borne diseases as a global health problem. Emerging infectious diseases, 4, 442. Gubler D.J. (2004). The changing epidemiology of yellow fever and dengue, 1900 to 2003: full circle? Comparative immunology, microbiology and infectious diseases, 27, 319-330. Gubler D.J. (2012). The economic burden of dengue. The American journal of tropical medicine and hygiene, 86, 743. Halstead S.B. (1980). Dengue haemorrhagic fever—a public health problem and a field for research. Bulletin of the World Health Organization, 58, 1. Hanson S.M. & Craig Jr G.B. (1994). Cold acclimation, diapause, and geographic origin affect cold hardiness in eggs of Aedes albopictus (Diptera: Culicidae). Journal of medical entomology, 31, 192-201. Harding J.S., Brown C., Jones F. & Taylor R. (2007). Distribution and habitats of mosquito larvae in the Kingdom of Tonga. Australian Journal of Entomology, 46, 332-338. Hardwick S.R., Toumi R., Pfeifer M., Turner E.C., Nilus R. & Ewers R.M. (2015). The relationship between leaf area index and microclimate in tropical forest and oil palm plantation: forest disturbance drives changes in microclimate. Agricultural and Forest Meteorology, 201, 187- 195. Harrington L.C., Edman J.D. & Scott T.W. (2001). Why do female Aedes aegypti (Diptera: Culicidae) feed preferentially and frequently on human blood? Journal of Medical Entomology, 38, 411- 422. Hashizume M., Terao T. & Minakawa N. (2009). The Indian Ocean Dipole and malaria risk in the highlands of western Kenya. Proceedings of the National Academy of Sciences, 106, 1857- 1862. Hawley W.A. (1988). The biology of Aedes albopictus. Journal of the American Mosquito Control Association. Supplement, 1, 1. Hay S.I., Guerra C.A., Tatem A.J., Atkinson P.M. & Snow R.W. (2005). Urbanization, malaria transmission and disease burden in Africa. Nature Reviews Microbiology, 3, 81-90. Henderson D.A. & Morse S.S. (1993). Emerging Viruses. In. Oxford University Press Oxford, UK. Ho B., Chan K. & Chan Y. (1971). Aedes aegypti (L.) and Aedes albopictus (Skuse) in Singapore city: 3. Population fluctuations. Bulletin of the World Health Organization, 44, 635. Hoekman D. (2010). Turning up the heat: temperature influences the relative importance of top down and bottom up effects. Ecology, 91, 2819-2825.
122
Hoffmann A.A., Hallas R., Sinclair C. & Partridge L. (2001). Rapid loss of stress resistance in Drosophila melanogaster under adaptation to laboratory culture. Evolution, 55, 436-438. Hopp M.J. & Foley J.A. (2001). Global-scale relationships between climate and the dengue fever vector, Aedes aegypti. Climatic change, 48, 441-463. Hothorn T., Bretz F. & Westfall P. (2008). Simultaneous inference in general parametric models. Biometrical Journal: Journal of Mathematical Methods in Biosciences, 50, 346-363. Huey R.B. & Berrigan D. (2001). Temperature, demography, and ectotherm fitness. The American Naturalist, 158, 204-210. Huey R.B. & Kingsolver J.G. (1989). Evolution of thermal sensitivity of ectotherm performance. Trends in Ecology & Evolution, 4, 131-135. Huey R.B. & Stevenson R. (1979). Integrating thermal physiology and ecology of ectotherms: a discussion of approaches. American Zoologist, 19, 357-366. Huho B., Ng'habi K., Killeen G., Nkwengulila G., Knols B. & Ferguson H. (2007). Nature beats nurture: a case study of the physiological fitness of free-living and laboratory-reared male Anopheles gambiae sl. Journal of Experimental Biology, 210, 2939-2947. Imai N., White M.T., Ghani A.C. & Drakeley C.J. (2014). Transmission and control of Plasmodium knowlesi: a mathematical modelling study. Johansson M.A., Cummings D.A. & Glass G.E. (2009). Multiyear climate variability and dengue—El Nino southern oscillation, weather, and dengue incidence in Puerto Rico, Mexico, and Thailand: a longitudinal data analysis. PLoS medicine, 6, e1000168. Johnson L.R., Ben-Horin T., Lafferty K.D., McNally A., Mordecai E., Paaijmans K.P., Pawar S. & Ryan S.J. (2015). Understanding uncertainty in temperature effects on vector borne disease: a Bayesian approach. Ecology, 96, 203-213. Jongwutiwes S., Putaporntip C., Iwasaki T., Sata T. & Kanbara H. (2004). Naturally acquired Plasmodium knowlesi malaria in human, Thailand. Emerging infectious diseases, 10, 2211. Jucker T., Hardwick S.R., Both S., Elias D.M., Ewers R.M., Milodowski D.T., Swinfield T. & Coomes D.A. (2018). Canopy structure and topography jointly constrain the microclimate of human modified tropical landscapes. Global change biology, 24, 5243-5258. Kelly-Hope L.A., Hemingway J. & McKenzie F.E. (2009). Environmental factors associated with the malaria vectors Anopheles gambiae and Anopheles funestus in Kenya. Malaria journal, 8, 268. Kesavaraju B., Damal K. & Juliano S.A. (2008). Do natural container habitats impede invader dominance? Predator-mediated coexistence of invasive and native container-dwelling mosquitoes. Oecologia, 155, 631-639. Killeen G.F., McKenzie F.E., Foy B.D., Schieffelin C., Billingsley P.F. & Beier J.C. (2000). A simplified model for predicting malaria entomologic inoculation rates based on entomologic and parasitologic parameters relevant to control. The American journal of tropical medicine and hygiene, 62, 535-544. Kilpatrick A.M. & Randolph S.E. (2012). Drivers, dynamics, and control of emerging vector-borne zoonotic diseases. The Lancet, 380, 1946-1955. Kingsolver J.G. (2009). The Well-Temperatured Biologist: (American Society of Naturalists Presidential Address). The American Naturalist, 174, 755-768. Kingsolver J.G., Higgins J.K. & Augustine K.E. (2015). Fluctuating temperatures and ectotherm growth: distinguishing non-linear and time-dependent effects. Journal of Experimental Biology, 218, 2218-2225. Klinkenberg E., McCall P., Wilson M.D., Amerasinghe F.P. & Donnelly M.J. (2008). Impact of urban agriculture on malaria vectors in Accra, Ghana. Malaria Journal, 7, 151. Knols B.G., Njiru B.N., Mathenge E.M., Mukabana W.R., Beier J.C. & Killeen G.F. (2002). MalariaSphere: A greenhouse-enclosed simulation of a natural Anopheles gambiae (Diptera: Culicidae) ecosystem in western Kenya. Malaria journal, 1, 19.
123
Koenraadt C., Githeko A. & Takken W. (2004). The effects of rainfall and evapotranspiration on the temporal dynamics of Anopheles gambiae ss and Anopheles arabiensis in a Kenyan village. Acta tropica, 90, 141-153. Koenraadt C.J., Paaijmans K.P., Githeko A.K., Knols B.G. & Takken W. (2003). Egg hatching, larval movement and larval survival of the malaria vector Anopheles gambiae in desiccating habitats. Malaria journal, 2, 20. Koh L.P. & Wilcove D.S. (2007). Cashing in palm oil for conservation. Nature, 448, 993-994. Kovach T.J. & Kilpatrick A.M. (2018). Increased human incidence of West Nile virus disease near rice fields in California but not in Southern United States. The American journal of tropical medicine and hygiene, 99, 222-228. Kovács B., Tinya F. & Ódor P. (2017). Stand structural drivers of microclimate in mature temperate mixed forests. Agricultural and Forest Meteorology, 234, 11-21. Kramer L.D., Hardy J.L. & Presser S.B. (1983). Effect of temperature of extrinsic incubation on the vector competence of Culex tarsalis for western equine encephalomyelitis virus. The American journal of tropical medicine and hygiene, 32, 1130-1139. Kuntom A., Ai T.Y., Kamaruddin N. & Beng Y.C. (2007). Pesticide application in the oil palm plantation. Oil palm bulletin, 54, 52. Lafferty K.D., McLaughlin J.P., Gruner D.S., Bogar T.A., Bui A., Childress J.N., Espinoza M., Forbes E.S., Johnston C.A. & Klope M. (2018). Local extinction of the Asian tiger mosquito (Aedes albopictus) following rat eradication on Palmyra Atoll. Biology letters, 14, 20170743. Lambrechts L., Paaijmans K.P., Fansiri T., Carrington L.B., Kramer L.D., Thomas M.B. & Scott T.W. (2011). Impact of daily temperature fluctuations on dengue virus transmission by Aedes aegypti. Proceedings of the National Academy of Sciences, 108, 7460-7465. Lau K., Chen C., Lee H., Low V., Moh H. & Sofian-Azirun M. (2017). Ovitrap surveillance in Sarawak, Malaysia: A comprehensive study. Tropical Biomedicine, 34, 795-803. Lawson C.R., Vindenes Y., Bailey L. & van de Pol M. (2015). Environmental variation and population responses to global change. Ecology Letters, 18, 724-736. Lee H.L., Rohani, A., Khadri, M.S., Nazni, W.A., Rozilawati, H., Nurulhusna, A.H., Nor Afizah, A.H., Roziah, A., Rosilawati, R. & Teh, C.H. (2015). Dengue vector control in Malaysia - challenges and recent advances. International Medical Journal Malaysia, 14, 6. Lee K.-S., Divis P., Zakaria S.K., Matusop A., Julin R.A., Conway D.J., Cox-Singh J. & Singh B. (2011). Plasmodium knowlesi: reservoir hosts and tracking the emergence in humans and macaques. PLoS Pathog, 7, e1002015. Leftwich P.T., Bolton M. & Chapman T. (2016). Evolutionary biology and genetic techniques for insect control. Evolutionary Applications, 9, 212-230. Legros M., Lloyd A.L., Huang Y. & Gould F. (2009). Density-dependent intraspecific competition in the larval stage of Aedes aegypti (Diptera: Culicidae): revisiting the current paradigm. Journal of medical entomology, 46, 409-419. Leisnham P.T., Sala L. & Juliano S.A. (2014). Geographic variation in adult survival and reproductive tactics of the mosquito Aedes albopictus. Journal of medical entomology, 45, 210-221. Leisnham P.T., Slaney D.P., Lester P.J. & Weinstein P. (2005). Increased larval mosquito densities from modified landuses in the Kapiti region, New Zealand: vegetation, water quality, and predators as associated environmental factors. EcoHealth, 2, 313-322. Lindsay S., Campbell H., Adiamah J., Greenwood A., Bangali J. & Greenwood B. (1990). Malaria in a peri-urban area of The Gambia. Annals of Tropical Medicine & Parasitology, 84, 553-562. Liu K., Tsujimoto H., Cha S.-J., Agre P. & Rasgon J.L. (2011). Aquaporin water channel AgAQP1 in the malaria vector mosquito Anopheles gambiae during blood feeding and humidity adaptation. Proceedings of the National Academy of Sciences, 108, 6062-6066.
124
Liu S.-S., Zhang G.-M. & Zhu J. (1995). Influence of temperature variations on rate of development in insects: analysis of case studies from entomological literature. Annals of the Entomological Society of America, 88, 107-119. Liu-Helmersson J., Stenlund H., Wilder-Smith A. & Rocklöv J. (2014). Vectorial capacity of Aedes aegypti: effects of temperature and implications for global dengue epidemic potential. PloS one, 9, e89783. Lounibos L., Larson V. & Morris C. (1990). Parity, fecundity and body size of Mansonia dyari in Florida. Journal of the American Mosquito Control Association, 6, 121-126. Lounibos L., Suárez S., Menéndez Z., Nishimura N., Escher R., O Connell S. & Rey J. (2002). Does temperature affect the outcome of larval competition between Aedes aegypti and Aedes albopictus? Journal of Vector Ecology, 27, 86-95. Lourenço-de-Oliveira R., Castro M.G., Braks M.A. & Lounibos L.P. (2004). The invasion of urban forest by dengue vectors in Rio de Janeiro. Journal of Vector Ecology, 29, 94-100. Loveridge R., Wearn O.R., Vieira M., Bernard H. & Ewers R.M. (2016). Movement behavior of native and invasive small mammals shows logging may facilitate invasion in a tropical rain forest. Biotropica, 48, 373-380. Luskin M.S. & Potts M.D. (2011). Microclimate and habitat heterogeneity through the oil palm lifecycle. Basic and Applied Ecology, 12, 540-551. Lyimo E., Takken W. & Koella J. (1992). Effect of rearing temperature and larval density on larval survival, age at pupation and adult size of Anopheles gambiae. Entomologia experimentalis et applicata, 63, 265-271. Macdonald G. (1952a). The analysis of equilibrium in malaria. Tropical diseases bulletin, 49, 813- 829. Macdonald G. (1952b). The analysis of the sporozoite rate. Tropical diseases bulletin, 49. Macdonald G. (1952c). Symposium on Insecticides. II. The Objectives of Residual Insecticide Campaigns. Transactions of the Royal Society of Tropical Medicine and Hygiene, 46, 227-34. Martin T.L. & Huey R.B. (2008). Why “suboptimal” is optimal: Jensen’s inequality and ectotherm thermal preferences. The American Naturalist, 171, E102-E118. Maxwell C.A., Rwegoshora R., Magesa S.M. & Curtis C.F. (2006). Comparison of coverage with insecticide-treated nets in a Tanzanian town and villages where nets and insecticide are either marketed or provided free of charge. Malaria journal, 5, 44. Mayne B. (1930). A Study of the Influence of Relative Humidity on the Life and Infectibility of the Mosquito. Indian Journal of Medical Research, 17. McGill B.J., Enquist B.J., Weiher E. & Westoby M. (2006). Rebuilding community ecology from functional traits. Trends in ecology & evolution, 21, 178-185. McMorrow J. & Talip M.A. (2001). Decline of forest area in Sabah, Malaysia: relationship to state policies, land code and land capability. Global Environmental Change, 11, 217-230. Medley K.A. (2010). Niche shifts during the global invasion of the Asian tiger mosquito, Aedes albopictus Skuse (Culicidae), revealed by reciprocal distribution models. Global Ecology and Biogeography, 19, 122-133. Medlock J.M., Hansford K.M., Schaffner F., Versteirt V., Hendrickx G., Zeller H. & Bortel W.V. (2012). A review of the invasive mosquitoes in Europe: ecology, public health risks, and control options. Vector-borne and zoonotic diseases, 12, 435-447. Meijide A., Badu C.S., Moyano F., Tiralla N., Gunawan D. & Knohl A. (2018). Impact of forest conversion to oil palm and rubber plantations on microclimate and the role of the 2015 ENSO event. Agricultural and Forest Meteorology, 252, 208-219. Méndez W., Liria J., Navarro J.-C., García C.Z., Freier J.E., Salas R., Weaver S.C. & Barrera R. (2001). Spatial dispersion of adult mosquitoes (Diptera: Culicidae) in a sylvatic focus of Venezuelan equine encephalitis virus. Journal of Medical Entomology, 38, 813-821.
125
Messina J.P., Brady O.J., Pigott D.M., Golding N., Kraemer M.U., Scott T.W., Wint G.W., Smith D.L. & Hay S.I. (2015). The many projected futures of dengue. Nature Reviews Microbiology, 13, 230. Mia M.S., Begum R.A., Er A.C., Abidin R. & Pereira J.J. (2013a). Trends of dengue infections in Malaysia, 2000-2010. Asian Pac J Trop Med, 6, 462-466. Mia M.S., Begum R.A., Er A.C., Abidin R.D.Z.R.Z. & Pereira J.J. (2013b). Trends of dengue infections in Malaysia, 2000-2010. Asian Pacific journal of tropical medicine, 6, 462-466. Mitchell Foster K., Ma B., Warsame Ali S., Logan C., Rau M. & Lowenberger C. (2012). The influence of larval density, food stress, and parasitism on the bionomics of the dengue vector Aedes aegypti (Diptera: Culicidae): implications for integrated vector management. Journal of Vector Ecology, 37, 221-229. Mocellin M.G., Simões T.C., Nascimento T.F.S.d., Teixeira M.L.F., Lounibos L.P. & Oliveira R.L.d. (2009). Bromeliad-inhabiting mosquitoes in an urban botanical garden of dengue endemic Rio de Janeiro-Are bromeliads productive habitats for the invasive vectors Aedes aegypti and Aedes albopictus? Memórias do Instituto Oswaldo Cruz, 104, 1171-1176. Moglia M., Gan K. & Delbridge N. (2016). Exploring methods to minimize the risk of mosquitoes in rainwater harvesting systems. Journal of Hydrology, 543, 324-329. Mohammed A. & Chadee D.D. (2011). Effects of different temperature regimens on the development of Aedes aegypti (L.)(Diptera: Culicidae) mosquitoes. Acta tropica, 119, 38-43. Mohsen B., TBS C., Husni M. & AR Z. (2014). Soil, Nutrients And Water Conservation Practices In Oil Palm Plantations On Sloping And Steep Lands In Malaysia. In: Inetrnational Agriculture Congress 2014, p. 23. Monath T.P. (2001). Yellow fever: an update. The Lancet infectious diseases, 1, 11-20. Monteiro L.C., de Souza J.R. & de Albuquerque C.M. (2007). Eclosion rate, development and survivorship of Aedes albopictus (Skuse)(Diptera: Culicidae) under different water temperatures. Neotropical entomology, 36, 966-971. Mordecai E.A., Cohen J.M., Evans M.V., Gudapati P., Johnson L.R., Lippi C.A., Miazgowicz K., Murdock C.C., Rohr J.R. & Ryan S.J. (2017). Detecting the impact of temperature on transmission of Zika, dengue, and chikungunya using mechanistic models. PLoS neglected tropical diseases, 11, e0005568. Mordecai E.A., Paaijmans K.P., Johnson L.R., Balzer C., Ben Horin T., Moor E., McNally A., Pawar S., Ryan S.J. & Smith T.C. (2013). Optimal temperature for malaria transmission is dramatically lower than previously predicted. Ecology letters, 16, 22-30. Mudhune S.A., Okiro E.A., Noor A.M., Zurovac D., Juma E., Ochola S.A. & Snow R.W. (2011). The clinical burden of malaria in Nairobi: a historical review and contemporary audit. Malaria journal, 10, 138. Munstermann L.E. (1997). Care and maintenance of Aedes mosquito colonies. In: The molecular biology of insect disease vectors. Springer, pp. 13-20. Murdock C., Sternberg E. & Thomas M. (2016). Malaria transmission potential could be reduced with current and future climate change. Scientific Reports, 6. Murdock C.C., Evans M.V., McClanahan T.D., Miazgowicz K.L. & Tesla B. (2017). Fine-scale variation in microclimate across an urban landscape shapes variation in mosquito population dynamics and the potential of Aedes albopictus to transmit arboviral disease. PLoS neglected tropical diseases, 11, e0005640. Murphy A., Rajahram G.S., Jilip J., Maluda M., William T., Hu W., Reid S., Devine G.J. & Frentiu F.D. (2019). Spatial and epidemiologic features of dengue in Sabah, Malaysia. BioRxiv, 657031. Mutero C.M., Wekoyela P., Githure J. & Konradsen F. (2004). Ammonium sulphate fertiliser increases larval populations of Anopheles arabiensis and culicine mosquitoes in rice fields. Acta tropica, 89, 187-192.
126
Muturi E.J., Allan B.F. & Ricci J. (2012). Influence of leaf detritus type on production and longevity of container-breeding mosquitoes. Environmental entomology, 41, 1062-1068. Muturi E.J., Lampman R., Costanzo K. & Alto B.W. (2011). Effect of temperature and insecticide stress on life-history traits of Culex restuans and Aedes albopictus (Diptera: Culicidae). Journal of medical entomology, 48, 243-250. Mwangangi J.M., Muturi E.J. & Mbogo C.M. (2009). Seasonal mosquito larval abundance and composition in Kibwezi, lower eastern Kenya. Journal of Vector Borne Diseases, 46, 65. Nájera J.A., González-Silva M. & Alonso P.L. (2011). Some lessons for the future from the Global Malaria Eradication Programme (1955–1969). PLoS Med, 8, e1000412. Nani R. (2015). Dengue incidence and the prevention and control program in Malaysia. Nasci R.S. (1986). The size of emerging and host-seeking Aedes aegypti and the relation of size to blood-feeding success in the field. J Am Mosq Control Assoc, 2, 61-62. Nasci R.S. & Mitchell C.J. (1994). Larval diet, adult size, and susceptibility of Aedes aegypti (Diptera: Culicidae) to infection with Ross River virus. Journal of Medical Entomology, 31, 123-126. Nelson M.J. (1986). Aedes aegypti: Biology and ecology. Nsoesie E.O., Kraemer M., Golding N., Pigott D.M., Brady O.J., Moyes C.L., Johansson M.A., Gething P.W., Velayudhan R. & Khan K. (2016). Global distribution and environmental suitability for chikungunya virus, 1952 to 2015. Euro surveillance: bulletin Europeen sur les maladies transmissibles= European communicable disease bulletin, 21. Null J. (2016). El Nino and La Nina years and intensities based on Oceanic Nino index (ONI). Accessed January. Okeke T.A. & Okeibunor J.C. (2010). Rural–urban differences in health-seeking for the treatment of childhood malaria in south-east Nigeria. Health policy, 95, 62-68. Oliva C.F., Benedict M.Q., Lempérière G. & Gilles J. (2011). Laboratory selection for an accelerated mosquito sexual development rate. Malaria journal, 10, 135. Omumbo J.A., Guerra C.A., Hay S.I. & Snow R. (2005). The influence of urbanisation on measures of Plasmodium falciparum infection prevalence in East Africa. Acta Tropica, 93, 11-21. Overgaard J. & Sørensen J.G. (2008). Rapid thermal adaptation during field temperature variations in Drosophila melanogaster. Cryobiology, 56, 159-162. Paaijmans K., Jacobs A., Takken W., Heusinkveld B., Githeko A., Dicke M. & Holtslag A. (2008a). Observations and model estimates of diurnal water temperature dynamics in mosquito breeding sites in western Kenya. Hydrological Processes: An International Journal, 22, 4789- 4801. Paaijmans K.P., Blanford S., Bell A.S., Blanford J.I., Read A.F. & Thomas M.B. (2010a). Influence of climate on malaria transmission depends on daily temperature variation. Proceedings of the National Academy of Sciences, 107, 15135-15139. Paaijmans K.P., Blanford S., Chan B.H. & Thomas M.B. (2011). Warmer temperatures reduce the vectorial capacity of malaria mosquitoes. Biology letters, rsbl20111075. Paaijmans K.P., Heinig R.L., Seliga R.A., Blanford J.I., Blanford S., Murdock C.C. & Thomas M.B. (2013). Temperature variation makes ectotherms more sensitive to climate change. Global change biology, 19, 2373-2380. Paaijmans K.P., Imbahale S.S., Thomas M.B. & Takken W. (2010b). Relevant microclimate for determining the development rate of malaria mosquitoes and possible implications of climate change. Malaria J, 9, 196. Paaijmans K.P., Jacobs A., Takken W., Heusinkveld B., Githeko A., Dicke M. & Holtslag A. (2008b). Observations and model estimates of diurnal water temperature dynamics in mosquito breeding sites in western Kenya. Hydrological Processes: An International Journal, 22, 4789- 4801. Paaijmans K.P., Read A.F. & Thomas M.B. (2009). Understanding the link between malaria risk and climate. Proceedings of the National Academy of Sciences, 106, 13844-13849.
127
Paaijmans K.P. & Thomas M.B. (2011). The influence of mosquito resting behaviour and associated microclimate for malaria risk. Malaria journal, 10, 183. Packierisamy P.R., Ng C.-W., Dahlui M., Inbaraj J., Balan V.K., Halasa Y.A. & Shepard D.S. (2015). Cost of dengue vector control activities in Malaysia. The American journal of tropical medicine and hygiene, 93, 1020-1027. Paixão E.S., Teixeira M.G. & Rodrigues L.C. (2018). Zika, chikungunya and dengue: the causes and threats of new and re-emerging arboviral diseases. BMJ global health, 3, e000530. Parham P.E. & Michael E. (2009). Modeling the effects of weather and climate change on malaria transmission. Environmental health perspectives, 118, 620-626. Parham P.E., Waldock J., Christophides G.K., Hemming D., Agusto F., Evans K.J., Fefferman N., Gaff H., Gumel A. & LaDeau S. (2015). Climate, environmental and socio-economic change: weighing up the balance in vector-borne disease transmission. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 370, 20130551. Parker A.H. (1952). The effect of a difference in temperature and humidity on certain reactions of female Aedes aegypti (L.). Bulletin of Entomological Research, 43, 221-229. Patterson G. (2016). Looking backward, looking forward: The long, torturous struggle with mosquitoes. Insects, 7, 56. Patz J.A., Graczyk T.K., Geller N. & Vittor A.Y. (2000). Effects of environmental change on emerging parasitic diseases. International journal for parasitology, 30, 1395-1405. Paupy C., Delatte H., Bagny L., Corbel V. & Fontenille D. (2009). Aedes albopictus, an arbovirus vector: from the darkness to the light. Microbes and Infection, 11, 1177-1185. Pérez-Díaz J.L., Álvarez-Valenzuela M.A. & García-Prada J.C. (2012). The effect of the partial pressure of water vapor on the surface tension of the liquid water–air interface. Journal of colloid and interface science, 381, 180-182. Pfeifer M., Kor L., Nilus R., Turner E., Cusack J., Lysenko I., Khoo M., Chey V., Chung A. & Ewers R. (2016). Mapping the structure of Borneo's tropical forests across a degradation gradient. Remote Sensing of Environment, 176, 84-97. Phuc H.K., Andreasen M.H., Burton R.S., Vass C., Epton M.J., Pape G., Fu G., Condon K.C., Scaife S. & Donnelly C.A. (2007). Late-acting dominant lethal genetic systems and mosquito control. BMC biology, 5, 11. Ponlawat A. & Harrington L.C. (2007). Age and body size influence male sperm capacity of the dengue vector Aedes aegypti (Diptera: Culicidae). Journal of medical entomology, 44, 422- 426. Proestos Y., Christophides G., Ergüler K., Tanarhte M., Waldock J. & Lelieveld J. (2015). Present and future projections of habitat suitability of the Asian tiger mosquito, a vector of viral pathogens, from global climate simulation. Philosophical Transactions of the Royal Society B: Biological Sciences, 370, 20130554. Puggioli A., Balestrino F., Damiens D., Lees R., Soliban S., Madakacherry O., Dindo M., Bellini R. & Gilles J. (2013). Efficiency of three diets for larval development in mass rearing Aedes albopictus (Diptera: Culicidae). Journal of Medical Entomology, 50, 819-825. R Core Team (2014). R: A language and environment for statistical computing. Ratkowsky D.A., Olley J. & Ross T. (2005). Unifying temperature effects on the growth rate of bacteria and the stability of globular proteins. Journal of theoretical biology, 233, 351-362. Ravaux J., Léger N., Rabet N., Morini M., Zbinden M., Thatje S. & Shillito B. (2012). Adaptation to thermally variable environments: capacity for acclimation of thermal limit and heat shock response in the shrimp Palaemonetes varians. Journal of Comparative Physiology B, 182, 899-907. Reid J. (1968). Anopheline mosquitoes of Malaya and Bornea. Anopheline mosquitoes of Malaya and Bornea.
128
Reinbold-Wasson D.D., Sardelis M.R., Jones J.W., Watts D.M., Fernandez R., Carbajal F., Pecor J.E., Calampa C., Klein T.A. & Turell M.J. (2012). Determinants of Anopheles seasonal distribution patterns across a forest to periurban gradient near Iquitos, Peru. The American journal of tropical medicine and hygiene, 86, 459-463. Reiskind M. & Lounibos L. (2009). Effects of intraspecific larval competition on adult longevity in the mosquitoes Aedes aegypti and Aedes albopictus. Medical and veterinary entomology, 23, 62-68. Renaud V., Innes J., Dobbertin M. & Rebetez M. (2011). Comparison between open-site and below- canopy climatic conditions in Switzerland for different types of forests over 10 years (1998− 2007). Theoretical and Applied Climatology, 105, 119-127. Rezza G., Nicoletti L., Angelini R., Romi R., Finarelli A., Panning M., Cordioli P., Fortuna C., Boros S. & Magurano F. (2007). Infection with chikungunya virus in Italy: an outbreak in a temperate region. The Lancet, 370, 1840-1846. Rezza G. & Weaver S.C. (2019). Chikungunya as a paradigm for emerging viral diseases: Evaluating disease impact and hurdles to vaccine development. PLoS neglected tropical diseases, 13, e0006919. Richards E.E., Masuoka P., Brett-Major D., Smith M., Klein T.A., Kim H.C., Anyamba A. & Grieco J. (2010). The relationship between mosquito abundance and rice field density in the Republic of Korea. International journal of health geographics, 9, 32. Richards S.L., Ponnusamy L., Unnasch T.R., Hassan H.K. & Apperson C.S. (2006). Host-feeding patterns of Aedes albopictus (Diptera: Culicidae) in relation to availability of human and domestic animals in suburban landscapes of central North Carolina. Journal of medical entomology, 43, 543-551. Robert V., Macintyre K., Keating J., Trape J.-F., Duchemin J.-B., Warren M. & Beier J.C. (2003). Malaria transmission in urban sub-Saharan Africa. The American journal of tropical medicine and hygiene, 68, 169-176. Rogers D. & Randolph S. (2006). Climate change and vector-borne diseases. Advances in parasitology, 62, 345-381. Rogers D.J. & Randolph S.E. (2000). The global spread of malaria in a future, warmer world. Science, 289, 1763-1766. Rohani A., Chan S., Abdullah A., Tanrang H. & Lee H. (2008). Species composition of mosquito fauna in Ranau, Sabah, Malaysia. Trop Biomed, 25, 232-236. Rohani A., Wong Y., Zamre I., Lee H. & Zurainee M. (2009). The effect of extrinsic incubation temperature on development of dengue serotype 2 and 4 viruses in Aedes aegypti (L.). Southeast Asian Journal of Tropical Medicine and Public Health, 40, 942. Roiz D., Rosa R., Arnoldi D. & Rizzoli A. (2010). Effects of temperature and rainfall on the activity and dynamics of host-seeking Aedes albopictus females in northern Italy. Vector-Borne and Zoonotic Diseases, 10, 811-816. Romero-Vivas C.M. & Falconar A.K. (2005). Investigation of relationships between Aedes aegypti egg, larvae, pupae, and adult density indices where their main breeding sites were located indoors. Journal of the American Mosquito Control Association, 21, 15-21. Romoser W.S. & Lucas E.A. (1999). Buoyancy and diving behavior in mosquito pupae. Journal of the American Mosquito Control Association-Mosquito News, 15, 194-199. Rosa E., Kanedi M., Oktaviani P.M. & Ningsih W.N. (2017). Phytotelmata Might Account for the High Prevalence ofDengue Hemorrhagic Fever in Lampung, Indonesia. Advances in Life Sciences, 7, 15-20. Ross P.A., Endersby Harshman N.M. & Hoffmann A.A. (2019). A comprehensive assessment of inbreeding and laboratory adaptation in Aedes aegypti mosquitoes. Evolutionary applications, 12, 572-586. Ross R. (1911). The Prevention of Malaria. John Murray.
129
Rúa G.L., Quiñones M.L., Vélez I.D., Zuluaga J.S., Rojas W., Poveda G. & Ruiz D. (2005). Laboratory estimation of the effects of increasing temperatures on the duration of gonotrophic cycle of Anopheles albimanus (Diptera: Culicidae). Memorias Do Instituto Oswaldo Cruz, 100, 515- 520. Rudnick A. (1965). Studies of the ecology of dengue in Malaysia: a preliminary report. Journal of medical entomology, 2, 203-208. Ruel J.J. & Ayres M.P. (1999). Jensen’s inequality predicts effects of environmental variation. Trends in Ecology & Evolution, 14, 361-366. Russell T.L., Lwetoijera D.W., Knols B.G., Takken W., Killeen G.F. & Ferguson H.M. (2011). Linking individual phenotype to density-dependent population growth: the influence of body size on the population dynamics of malaria vectors. Proceedings of the Royal Society of London B: Biological Sciences, rspb20110153. Ryan S.J., McNally A., Johnson L.R., Mordecai E.A., Ben-Horin T., Paaijmans K. & Lafferty K.D. (2015). Mapping physiological suitability limits for malaria in Africa under climate change. Vector- Borne and Zoonotic Diseases, 15, 718-725. Samuel P.P., Ramesh D., Thenmozhi V., Nagaraj J., Muniaraj M. & Arunachalam N. (2016). Japanese Encephalitis vector abundance and infection frequency in Cuddalore District, Tamil Nadu, India: a five-year longitudinal study. Journal of Entomological and Acarological Research, 48, 366-371. Sarfraz M.S., Tripathi N.K., Tipdecho T., Thongbu T., Kerdthong P. & Souris M. (2012). Analyzing the spatio-temporal relationship between dengue vector larval density and land-use using factor analysis and spatial ring mapping. BMC public health, 12, 853. Satitvipawee P., Wongkhang W., Pattanasin S., Hoithong P. & Bhumiratana A. (2012). Predictors of malaria-association with rubber plantations in Thailand. BMC Public Health, 12, 1115. Scott T.W., Amerasinghe P.H., Morrison A.C., Lorenz L.H., Clark G.G., Strickman D., Kittayapong P. & Edman J.D. (2000). Longitudinal studies of Aedes aegypti (Diptera: Culicidae) in Thailand and Puerto Rico: blood feeding frequency. Journal of medical entomology, 37, 89-101. Seawright J., Dame D. & Weidhaas D. (1977). Field survival and ovipositional characteristics of Aedes aegypti and their relation to population dynamics and control. Mosquito News, 37, 62- 70. Shapiro L.L., Whitehead S.A. & Thomas M.B. (2017). Quantifying the effects of temperature on mosquito and parasite traits that determine the transmission potential of human malaria. PLoS biology, 15, e2003489. Shuaib W., Stanazai H., Abazid A.G. & Mattar A.A. (2016). Re-emergence of Zika virus: a review on pathogenesis, clinical manifestations, diagnosis, treatment, and prevention. The American journal of medicine, 129, 879. e7-879. e12. Simorangkir D., Moore P., Haase N. & Ng G. (2002). Land clearing on degraded lands for plantation development. In: A Workshop on Economics of Fire Use in Agriculture and Forest Plantations, pp. 24-25. Singer B.H. & de Castro M.C. (2001). Agricultural colonization and malaria on the Amazon frontier. Annals of the New York Academy of Sciences, 954, 184-222. Smallwood P.D. (1996). An introduction to risk sensitivity: The use of Jensen's inequality to clarify evolutionary arguments of adaptation and constraint. American Zoologist, 36, 392-401. Sodhi N.S., Koh L.P., Brook B.W. & Ng P.K. (2004). Southeast Asian biodiversity: an impending disaster. Trends in ecology & evolution, 19, 654-660. Sodhi N.S., Koh L.P., Clements R., Wanger T.C., Hill J.K., Hamer K.C., Clough Y., Tscharntke T., Posa M.R.C. & Lee T.M. (2010). Conserving Southeast Asian forest biodiversity in human-modified landscapes. Biological Conservation, 143, 2375-2384. Steinwascher K. (1982). Relationship between pupal mass and adult survivorship and fecundity for Aedes aegypti. Environmental Entomology, 11, 150-153.
130
Stenseth N.C., Ottersen G., Hurrell J.W., Mysterud A., Lima M., Chan K.S., Yoccoz N.G. & Ådlandsvik B. (2003). Studying climate effects on ecology through the use of climate indices: the North Atlantic Oscillation, El Nino Southern Oscillation and beyond. Proceedings of the Royal Society of London B: Biological Sciences, 270, 2087-2096. Suggitt A.J., Gillingham P.K., Hill J.K., Huntley B., Kunin W.E., Roy D.B. & Thomas C.D. (2011). Habitat microclimates drive fine scale variation in extreme temperatures. Oikos, 120, 1-8. Sunday J.M., Bates A.E., Kearney M.R., Colwell R.K., Dulvy N.K., Longino J.T. & Huey R.B. (2014). Thermal-safety margins and the necessity of thermoregulatory behavior across latitude and elevation. Proceedings of the National Academy of Sciences, 111, 5610-5615. Takken W., Klowden M. & Chambers G. (1998). Effect of body size on host seeking and blood meal utilization in Anopheles gambiae sensu stricto (Diptera: Culicidae): the disadvantage of being small. Journal of medical entomology, 35, 639-645. Takken W. & Verhulst N.O. (2013). Host preferences of blood-feeding mosquitoes. Annual review of entomology, 58, 433-453. Tangena J.-A.A., Thammavong P., Hiscox A., Lindsay S.W. & Brey P.T. (2015). The human-baited double net trap: an alternative to human landing catches for collecting outdoor biting mosquitoes in Lao PDR. PLoS One, 10, e0138735. Tangena J.-A.A., Thammavong P., Wilson A.L., Brey P.T. & Lindsay S.W. (2016). Risk and control of mosquito-borne diseases in Southeast Asian rubber plantations. Trends in parasitology, 32, 402-415. Tatem A.J., Gething P.W., Smith D.L. & Hay S.I. (2013). Urbanization and the global malaria recession. Malaria journal, 12, 133. Teng H.-J. & Apperson C.S. (2000). Development and survival of immature Aedes albopictus and Aedes triseriatus (Diptera: Culicidae) in the laboratory: effects of density, food, and competition on response to temperature. Journal of medical entomology, 37, 40-52. Therneau T.M., Grambsch P.M. & Pankratz V.S. (2003). Penalized survival models and frailty. Journal of computational and graphical statistics, 12, 156-175. Thongsripong P., Green A., Kittayapong P., Kapan D., Wilcox B. & Bennett S. (2013). Mosquito vector diversity across habitats in central Thailand endemic for dengue and other arthropod-borne diseases. PLoS neglected tropical diseases, 7. Toma L., Severini F., Di M.L., Bella A. & Romi R. (2003). Seasonal patterns of oviposition and egg hatching rate of Aedes albopictus in Rome. Journal of the American Mosquito Control Association, 19, 19-22. Tomanek L. (2010). Variation in the heat shock response and its implication for predicting the effect of global climate change on species' biogeographical distribution ranges and metabolic costs. Journal of Experimental Biology, 213, 971-979. Tran A., L'Ambert G., Lacour G., Benoît R., Demarchi M., Cros M., Cailly P., Aubry-Kientz M., Balenghien T. & Ezanno P. (2013). A rainfall-and temperature-driven abundance model for Aedes albopictus populations. International journal of environmental research and public health, 10, 1698-1719. Traoré D.F., Sagna A.B., Adja A.M., Zoh D.D., Adou K.A., Lingué K.N., Coulibaly I., Assi S.B., Poinsignon A. & Dagnogo M. (2019). Exploring the heterogeneity of human exposure to malaria vectors in an urban setting, Bouaké, Côte d’Ivoire, using an immuno-epidemiological biomarker. Malaria journal, 18, 68. Tucker Lima J.M., Vittor A., Rifai S. & Valle D. (2017). Does deforestation promote or inhibit malaria transmission in the Amazon? A systematic literature review and critical appraisal of current evidence. Philosophical Transactions of the Royal Society B: Biological Sciences, 372, 20160125.
131
Tun Lin W., Burkot T. & Kay B. (2000). Effects of temperature and larval diet on development rates and survival of the dengue vector Aedes aegypti in north Queensland, Australia. Medical and veterinary entomology, 14, 31-37. Turell M.J. (1993). Effect of environmental temperature on the vector competence of Aedes taeniorhynchus for Rift Valley fever and Venezuelan equine encephalitis viruses. The American journal of tropical medicine and hygiene, 49, 672-676. Van der Have T. & De Jong G. (1996). Adult size in ectotherms: temperature effects on growth and differentiation. Journal of theoretical biology, 183, 329-340. Van Der Stuyft P., Gianella A., Pirard M., Cespedes J., Lora J., Peredo C., Pelegrino J., Vorndam V. & Boelaert M. (1999). Urbanisation of yellow fever in Santa Cr uz, Bolivia. The Lancet, 353, 1558-1562. Van Handel E. & Day J.F. (1989). Correlation between wing length and protein content of mosquitoes. J Am Mosq Control Assoc, 5, 180-182. van Uitregt V.O., Hurst T.P. & Wilson R.S. (2012). Reduced size and starvation resistance in adult mosquitoes, Aedes notoscriptus, exposed to predation cues as larvae. Journal of Animal Ecology, 81, 108-115. Vanwambeke S.O., Lambin E.F., Eichhorn M.P., Flasse S.P., Harbach R.E., Oskam L., Somboon P., Van Beers S., Van Benthem B.H. & Walton C. (2007). Impact of land-use change on dengue and malaria in northern Thailand. EcoHealth, 4, 37-51. Vasseur D.A., DeLong J.P., Gilbert B., Greig H.S., Harley C.D., McCann K.S., Savage V., Tunney T.D. & O'Connor M.I. (2014). Increased temperature variation poses a greater risk to species than climate warming. Proceedings of the Royal Society B: Biological Sciences, 281, 20132612. Vazeille M., Moutailler S., Coudrier D., Rousseaux C., Khun H., Huerre M., Thiria J., Dehecq J.-S., Fontenille D. & Schuffenecker I. (2007). Two Chikungunya isolates from the outbreak of La Reunion (Indian Ocean) exhibit different patterns of infection in the mosquito, Aedes albopictus. PloS one, 2, e1168. Victor T. & Reuben R. (2000). Effects of organic and inorganic fertilisers on mosquito populations in rice fields of southern India. Medical and veterinary entomology, 14, 361-368. Vincenti-Gonzalez M., Tami A., Lizarazo E. & Grillet M. (2018). ENSO-driven climate variability promotes periodic major outbreaks of dengue in Venezuela. Scientific reports, 8, 5727. Vittor A.Y., Pan W., Gilman R.H., Tielsch J., Glass G., Shields T., Sánchez-Lozano W., Pinedo V.V., Salas- Cobos E. & Flores S. (2009). Linking deforestation to malaria in the Amazon: characterization of the breeding habitat of the principal malaria vector, Anopheles darlingi. The American journal of tropical medicine and hygiene, 81, 5. Walsh J., Molyneux D. & Birley M. (1993). Deforestation: effects on vector-borne disease. Parasitology, 106, S55-S75. Wan-Norafikah O., Nazni W., Noramiza S., Shafa’ar-Ko’Ohar S., Heah S., Nor-Azlina A., Khairuh-Asuad M. & Lee H. (2012). Distribution of Aedes mosquitoes in three selected localities in Malaysia. Sains Malaysiana, 41, 1309-1313. Watson T.M., Marshall K. & Kay B.H. (2000). Colonization and laboratory biology of Aedes notoscriptus from Brisbane, Australia. Journal of the American Mosquito Control Association- Mosquito News, 16, 138-142. Watts D.M., Burke D.S., Harrison B.A., Whitmire R.E. & Nisalak A. (1987). Effect of temperature on the vector efficiency of Aedes aegypti for dengue 2 virus. The American journal of tropical medicine and hygiene, 36, 143-152. Wearn O.R., Carbone C., Rowcliffe J.M., Bernard H. & Ewers R.M. (2016). Grain dependent responses of mammalian diversity to land use and the implications for conservation set aside. Ecological Applications, 26, 1409-1420.
132
Weaver S.C. & Forrester N.L. (2015). Chikungunya: Evolutionary history and recent epidemic spread. Antiviral research, 120, 32-39. Westbrook C.J., Reiskind M.H., Pesko K.N., Greene K.E. & Lounibos L.P. (2010). Larval environmental temperature and the susceptibility of Aedes albopictus Skuse (Diptera: Culicidae) to Chikungunya virus. Vector-Borne and Zoonotic Diseases, 10, 241-247. WHO (1957). Expert Committee on Malaria, Sixth Report. In. World Health Organization Geneva. WHO (2017). Global vector control response. 2017-2030. In. World Health Organisation Geneva. Wiwatanaratanabutr S. & Kittayapong P. (2006). Effects of temephos and temperature on Wolbachia load and life history traits of Aedes albopictus. Medical and veterinary entomology, 20, 300-307. Wolfe N.D., Kilbourn, A. M., Karesh, W.B., Rahman, H.A., Bosi, E.J., Cropp, B.C., Andau, M., Spielman, A., Gubler, D.J. (2001). Sylvatic transmission of arboviruses among Bornean orangutans. American Journal of Tropical Medicine and Hygiene 64, 310-316. Xiao F.-Z., Zhang Y., Deng Y.-Q., He S., Xie H.-G., Zhou X.-N. & Yan Y.-S. (2014). The effect of temperature on the extrinsic incubation period and infection rate of dengue virus serotype 2 infection in Aedes albopictus. Archives of virology, 159, 3053-3057. Yamana T.K. & Eltahir E.A. (2013). Incorporating the effects of humidity in a mechanistic model of Anopheles gambiae mosquito population dynamics in the Sahel region of Africa. Parasites & vectors, 6, 235. Yang G.-J., Bradshaw C.J., Whelan P.I. & Brook B.W. (2008). Importance of endogenous feedback controlling the long-term abundance of tropical mosquito species. Population Ecology, 50, 293-305. Yanoviak n., Lounibos n. & Weaver n. (2006a). Land use affects macroinvertebrate community composition in phytotelmata in the Peruvian Amazon. Annals of the Entomological Society of America, 99, 1172-1181. Yanoviak S.P., Paredes J.E., Lounibos L.P. & Weaver S.C. (2006b). Deforestation alters phytotelm habitat availability and mosquito production in the Peruvian Amazon. Ecological Applications, 16, 1854-1864. Yee D.A. (2008). Tires as habitats for mosquitoes: a review of studies within the eastern United States. Journal of Medical Entomology, 45, 581-593. Yergolkar P.N., Tandale B.V., Arankalle V.A., Sathe P.S., AB S., Gandhe S.S., Gokhle M.D., Jacob G.P., Hundekar S.L. & Mishra A.C. (2006). Chikungunya outbreaks caused by African genotype, India. Emerging infectious diseases, 12, 1580. Yohannes M. & Petros B. (1996). Urban malaria in Nazareth, Ethiopia: parasitological studies. Ethiopian medical journal, 34, 83-91. Young K.I., Mundis S., Widen S.G., Wood T.G., Tesh R.B., Cardosa J., Vasilakis N., Perera D. & Hanley K.A. (2017). Abundance and distribution of sylvatic dengue virus vectors in three different land cover types in Sarawak, Malaysian Borneo. Parasites & Vectors, 10, 406. Zahouli J.B., Koudou B.G., Müller P., Malone D., Tano Y. & Utzinger J. (2017). Effect of land-use changes on the abundance, distribution, and host-seeking behavior of Aedes arbovirus vectors in oil palm-dominated landscapes, southeastern Côte d’Ivoire. PloS one, 12. Zirbel K., Eastmond B. & Alto B.W. (2018). Parental and offspring larval diets interact to influence life-history traits and infection with dengue virus in Aedes aegypti. Royal Society open science, 5, 180539.
133