Artificial light at night (ALAN) is a prominent feature of most urban and semi-urban areas. The urbanization of the human population, i.e., the number of people living in urban areas, is projected to increase from the current 4 billion to 6.3 billion people by 2050 [1
]. This is accompanied by a yearly increase of artificially lit surface of Earth at night of 2–6%, both in radiance and extent [2
]. In addition, there is an increase of regularly illuminated urban underground infrastructures such as metro systems. While ALAN has been found to affect the physiology and behavior of individual organisms, the extent to which this affects fitness remains poorly understood [4
]. Studies of adult insects have focused on the attraction to artificial light [8
], the loss of orientation and exhaustion [10
], or dietary changes [13
]. Less studied are the effects ALAN may have on diel rhythms of exposed organisms. Light acts as a prominent environmental cue to set and synchronize circadian timekeeping with the ambient light cycles. This synchronization in turn regulates entire metabolic pathways. Disruption of the circadian clock can therefore lead to modifications in the timing of key behaviors such as foraging and mating [12
], which may ultimately lead to altered fitness due to changes in energy uptake or changes in lifetime mating success [16
Many mosquito species (Diptera, Culicidae) thrive in areas of human habitation [18
] and are consequently exposed to ALAN. Understanding the effects of ALAN on physiology, behavior and fecundity of mosquitoes may thus have important implications for mosquito populations, including their function in natural ecosystems [19
] and disease transmission dynamics [21
]. In Europe, the widespread species Culex pipiens
commonly occurs in urban and suburban areas [22
] and is comprised of two forms or “ecotypes”: Cx. pipiens
(hereafter molestus) and Cx. pipiens
(hereafter pipiens) [24
]. The molestus ecotype occurs in North America, Southern and Northern Europe. It was long believed to be restricted to underground habitats in its northern distribution range, e.g., New York and London [25
]; however, recent studies have found that both ecotypes occur sympatrically above-ground [28
]. Culex pipiens
readily mates in confined spaces and does not require a blood meal to produce eggs [24
]. These particular features make it well-suited for controlled laboratory experiments. Although pipiens mates primarily in open spaces, it requires a blood meal before oviposition, and undergoes winter diapause [24
]; there is demonstrated phenotypic plasticity in mating, as they are able to breed in the laboratory without swarm formation [23
]. The two forms are not easily distinguished and even genetic diagnostic tests fail to separate them in some parts of their range [30
]. Culex pipiens
activity is typically crepuscular-nocturnal, with the daily peak of host-seeking activity having been observed to shift from night in summer to evening in autumn [31
]. This is likely to expose urban populations of C
to ALAN because this coincides with the timing of their natural flying behavior [31
]. In many rural and urban areas, there is a temporal dividing line around midnight, after which full illumination ends as some lights are extinguished [33
]. Effectively, ALAN alters the photoperiod, providing less time for this nocturnal species to show behavior associated with the active phase of their circadian cycle.
The genetic basis of the mosquito circadian clock (Figure 1
) includes the genes Clock, cycle, period, timeless, cryptochrome1
]. These genes and their products comprise a central feedback-loop that rhythmically regulates transcription and repression of its components [39
]. The circadian clock of mosquitoes differs from the well-studied Drosophila
clock by the possession of another cryptochrome: cryptochrome2
. This closely resembles the hypothesized clock model of the Monarch butterfly (Dannaus plexippus
), which also possesses two cryptochromes. The possession of two cryptochromes is assumed to be the ancestral state of the clock in insects [41
] and this model is likely to be true for other insects with two cryptochromes [42
]. Therefore, it is assumed that the mosquito clock functions similar to that of the monarch butterfly (Figure 1
). In the monarch butterfly, and likely also in mosquitoes, the proteins CLOCK and CYCLE form a dimer in the nucleus that binds to the E-Box promoting the expression of period, timeless
). The proteins migrate into the cytoplasm where they form a complex. It remains unknown whether TIMELESS is part of this complex, stabilizing it, or whether CRYPTOCHROME2 interacts with PERIOD and TIMELESS in vivo [43
]. Environmental light cues are received by the blue-light receptor CRYPTOCHROME1, leading to its activation [42
]. This constitutes the pathway by which environmental light is perceived and affects the functioning of the circadian clock by entraining it to ambient light conditions. The activated CRYPTOCHROME1 then degrades TIMELESS while PERIOD is phosphorylated during the course of the night [44
]. It is thought that PERIOD assists CRYPTOCHROME2 in migrating back into the nucleus where it represses the CLOCK-CYCLE-dimer, thereby repressing the transcription of period
Artificial light at night presents an alteration of the light regime that differs from what has been investigated in most studied organisms. A number of studies have identified important links between light and activity in mosquitoes [31
], but we are not aware of any that examined how ALAN may affect physiology and activity. Because of the link of the circadian clock to downstream processes, it is reasonable to assume that light-mediated changes of the rhythm may lead to changes in activity patterns, which could then affect host-search, feeding, and mating. Altered feeding behavior could in turn influence the nutritional state of the individuals in several ways. Diurnal organisms may be able to use extra light to increase the time used for feeding, whereas nocturnal species may avoid light, thereby reducing available time for foraging.
Evidence of the effects of ALAN on fitness is scarce. ALAN is able to alter the timing of reproductive physiology and lay date in songbirds [47
] and to suppress different hormones along the reproductive axis in common freshwater fish species [5
]. In insects, ALAN was found to reduce sex pheromone production and mating in moths [49
]. We are aware of only one study investigating how extra light in otherwise dark phases affects fecundity in Diptera, which reports a decrease in the number of eggs produced in Drosophila melanogaster
]. A reduction in fecundity could provide evidence that ALAN may affect population growth, which could further have an effect on disease epidemiology in vector species.
We used three laboratory experiments to test whether (1) clock gene expression, (2) behavior, and (3) fecundity in Cx. pipiens
were altered by 3.5 h of additional ALAN (100–300 lux) during the otherwise dark phase of the diurnal cycle. The aim was to mimic a lighting scheme typically found in urban environments whereby nocturnal individuals experience an extended dusk which could disrupt the normal cue for the onset of daily activities, e.g., foraging, during the night [31
]. We expect that ALAN could affect all three tested responses, by providing brighter than natural light at dusk. In the first experiment, changes in expression were measured for five genes of the circadian clock relative to a constitutive gene over 10 time points from late afternoon until after midnight. In the second experiment, activity was continuously measured in light:dark cycles followed by measurements in constant darkness. Finally, population fecundity was assessed as mean egg size, numbers of eggs produced, and numbers of egg rafts.
We found ALAN to simultaneously influence clock-gene expression, activity, and reproductive output in a laboratory study of the mosquito Cx. pipiens
, a widespread temperate species occurring in urban and suburban areas [22
]. In addition, alterations of gene expression and activity were sex-specific. The consequences of a sex-specific effect of ALAN may have important consequences on the population level because of its potential effect on mating and the differing roles of sexes in ecosystems [62
]. Sex-specific changes in gene expression might potentially shift the timing of activity, leading to only little overlap in active phase negatively influencing mating behavior, leading to reduced population sizes. We also found a reduced number and size of eggs when ALAN was present which would be in line with this hypothesis. Future studies are needed to specifically test these links.
4.1. Gene Expression
Genes of the central circadian clock have been well-studied in several mosquito species, including the closely related Cx. quinquefasciatus
]. Past research focused on changes in day length (i.e., changes in L:D ratios) and interactions with other factors (e.g., temperature [34
]). We were interested in artificial light generated by outdoor lighting, which does not simply modify L:D ratios, but consists of an abrupt switching on of lower-intensity light source and a constant brightness for some time in a period that would otherwise be dark. An important overall finding was that the patterns we observed were consistent across days of exposure to the ALAN treatment (1, 2, or 4 days), indicating re-occurring expression patterns and consistent treatment-induced changes over time. The only exception was the cycle
gene, where expression decreased with days of exposure in both treatments. This could indicate age-related changes in cycle
expression patterns, although individuals were of mixed ages (5–10 days old), or a different response of cycle
compared to the other genes that were studied.
In all five genes examined, GLMs clearly indicated that expression was reduced in individuals exposed to ALAN. This was as a single factor in period
and as an interaction with sampling time in cryptochrome2
where significant reduction in expression occurred in a subset of time points. Interestingly, there was no clear pattern among time points that would indicate a direct and short-term response to light. Changes were also not restricted to the time of “extra-light”, indicating that the response to altered light regimes may be complex and subject to the different loops of the circadian clock machinery (see Figure 1
) rather than consistent short-term responses. Additionally, the temporal variation was ‘flattened’ in period, timeless
in extra-light, indicating less variation in transcript numbers. While this was not within the scope of our study it seems a promising field for future studies on the effect of ALAN. Male and female mosquitoes exhibit many sex-specific characteristics (e.g., blood-feeding in females [63
]) and thus ALAN may affect the sexes differently. Genome-wide studies of gene expression have reported pronounced differences between males and females in molestus [65
] and in Anopheles gambiae
]. Here, we found sex-specific differences in gene expression in response to ALAN in all genes except in period.
In the case of cycle
, males and females differed in the direction of response (up-or down-regulation of gene expression). These differences may result in changes in clock regulation.
The differences we observed in gene expression are not easily compared to published reports because the only related study was carried out on the sister species Cx. quinquefasciatus
or distantly related Aedes aegypti
]. In these earlier studies, individuals were sampled every two hours over a 24-h period in 12:12 L:D, in contrast to the higher frequency sampling over a 7-h period here. This makes it difficult to estimate whether our data reflect intrinsic (i.e., species-specific) differences in clock gene expression patterns. Although closely related, Cx. quinquefasciatus
has a tropical distribution and may thus differ in its clock-gene expression patterns because of daily and seasonal differences in light regime, temperature, and other factors relevant to the functioning of the circadian clock (reviewed in [38
]). Based on previous studies in other species, we anticipated Clock
to peak when the levels of period, timeless
were low [36
]. In contrast, overall temporal changes were minimal in our control treatment, with Clock
increasing as timeless
decreased. This might result in protein levels that impede CLOCK-CYCLE-complex formation possibly disrupting the feedback loop. A notable finding was a lack of linkage between cryptochrome2
expression and activity in Cx. pipiens
. Gentile et al. [36
] speculated cryptochrome2
might be involved in controlling activity patterns based on differences in expression patterns between a diurnal (Ae. aegypti
) and a nocturnal (Cx. quinquefasciatus
) species. In our study of Cx. pipiens
was not found to cycle over time and varied little across treatments, despite pronounced differences in behavior. This may suggest that, at least in this species, activity is controlled by different genes or via post-translational regulation [66
In order to obtain sufficient numbers of samples, we sampled individuals 5-10 days after emergence. Although this is a relatively small window of the total lifespan, individuals may have been in different gonotrophic states. Gene expression has been reported to vary in different gonotrophic states (Anopheles gambiae
]; Aedes aegypti
]). However, other processes such as digestion (of a bloodmeal) and egg formation also influence gene expression profiles [67
]; it is therefore not straightforward to determine which process exerts more influence on gene expression profiles. By pooling samples, we reduced the influence of individuals’ gene expression profiles on the overall outcome. On the other hand, pooling may have reduced our ability to detect differences in response to ALAN. Our data are therefore likely to underestimate its effect. In the laboratory, Cx. pipiens
females have been reported to live an average of 10 days longer than males [69
], thus males and females could have been in different states of senescence, contributing to the sex-specific differences we observed. Nonetheless, the gene expression and activity experiments were completed when the individuals were within the first half of their expected average lifespan (females: mean = 42.3 days, max. = 75 days; males: mean = 32.7, max. = 52.5 days [69
]). Sex-specific stages of senescence are therefore expected to have only had marginal effects.
Individuals exposed to ALAN were consistently less active, except for females which were more active at night. The effect of ALAN was more pronounced in the Light:Dark (LD) experiment compared to constant darkness (DD). Females were also more active than males in nearly all phases and regardless of treatment. Sex-specific activity patterns are known from An. gambiae
] where virgin females commence flight activity 5–12 min later than males. Our data suggest that there are also light-induced differences in sex-specific activity, which might add to inherent differences between male and female activity. Activity varied consistently throughout all phases in DD, suggesting that internal timekeeping was not disrupted by the extra-light treatment that both groups received prior to the experiment in constant darkness. However, the onset of activity shifted to dusk in DD (as opposed to the trial phase in LD). This might reflect earlier anticipation of darkness, and may indicate that changing light intensity, whether increasing or decreasing, was an important trigger for activity. We found very different levels of activity in all phases except for mid-day, when activity was always lowest.
Host-seeking (i.e., female) flight activity in Cx. pipiens
was reported to be induced when light intensity dropped below 5 lux, resulting in a strong peak two to three hours after sunset with low activity until sunrise [31
]. The authors also reported a shift in the timing of activity after the autumnal equinox to before sunset and suggested that this was a light-induced change in behavior because temperatures were still favorable at the time [31
]. This coincides with the phases of peak activity in our study. The extra-light clearly inhibited activity although light dropped below 5 lux prior to the addition of ALAN (see Figure 2
), usually a signal to commence flying. The reduced activity under the extra-light condition (i.e., prolonged resting periods) may be followed by a reduction in foraging time and thus decreased nutrient uptake. Additionally, it may lead to fewer mating encounters.
Owing to the malfunction of some lights in the extra-light treatment, the light regimes in the activity experiment differed in their daily maximum. Although the proportion of change in the extra-light treatment was the same as in the other experiments, the absolute intensity was lower throughout the day compared to control. We cannot be certain if this led to the observed differences between treatments, or how it generally influenced activity (i.e., overall activity may be affected by daytime light intensity). However, given that the absolute light intensities in our treatment group were lower than in controls, it is remarkable that the extra light produced such a strong effect (i.e., greatly reduced activity compared to dark phases within extra-light treatment). The general pattern (active in the dark vs. resting with light) in constant darkness was very similar between the treatments, suggesting that the differences in the absolute light intensity between treatments did not have a strong effect on the results.
We observed changes in the size and number of eggs per raft, indicating that ALAN may affect larval fitness (reduced egg size) and female fecundity (fewer eggs). Adult densities were maintained at comparable levels and the number of egg rafts did not differ significantly between treatments. We therefore presume that females laid eggs at a similar rate and the amount of energy used for finding oviposition sites was comparable. Smaller eggs may have been produced because adult females were smaller under extra-light, because energy allocation to egg production and provisioning was reduced, or both. The significant changes in egg diameter over time suggest that adult female size distribution was not skewed in the overall data set. The smaller egg size observed is therefore most likely a result of reduced energy allocation. Because food was available ad libitum
, reduced feeding is one explanation for reduced resource allocation. Another possibility is that fewer females were laying eggs in the ALAN treatment. The fact that the number of rafts was equal in treatments means that fewer ovipositing females would have had to produce more rafts, which would likely mean that they allocate fewer resources to each. However, because ovary competence is stimulated through food intake [71
] and food was available, we assume all females were capable of producing eggs.
McLay et al. [51
] observed a reduction in the probability of females to commence oviposition in ALAN treatments in D. melanogaster.
However, females that did oviposit showed, similar to our study, a reduction in the number of eggs laid. Although this did not translate into differences at the juvenile phase, there was an ALAN-induced reduction in adult survival. In the mosquito Wyeomyia smithii
, Emerson et al. [6
] found that, of all fitness components measured (pupal survivorship, embryonic viability, adult longevity, fecundity as mean number of eggs per eclosing female), it was the number of eggs that was significantly reduced in a non-resonant (L:D 10:25) light environment. The mechanisms behind these changes, however, may be different in our study because Cx. pipiens
is a nocturnal species, we used different L:D ratios, added ALAN, and we measured overall population fecundity rather than individual female fecundity. Taken together, the results of these very different studies suggest that a reduction in egg production is a common response to altered light conditions during day and night cycles.
Food limitation can influence the number of eggs per clutch; e.g., adults reared under low food conditions led to daughters laying more eggs in An. stephensi
]. Larval diet influenced the number of eggs in Ochlerotatus atropalpus
]. In the present study, however, no developmental stage was subjected to food limitation; we can therefore exclude the possibility that larval diet influenced the outcome. Interestingly, the difference in diameter over time (205 d) (eggs were smallest in March, largest in August, and intermediate in October) occurred in both treatments but was more pronounced in the control. Our experimental set-up lacked obvious cues of seasonal change (e.g., day length, temperature). The lake water used for oviposition and rearing of larvae may have introduced a seasonal signal, although it was filtered and autoclaved to remove bacterial and other planktonic cells that might provide potential cues from the lake community. Innate seasonality might have given rise to this pattern. It is well known that there is a complex genetic basis for seasonal rhythms [74
], but it is not clear how reliable seasonality will be expressed in a constant environment, and we are unaware of any studies evaluating the egg size of Culex
over the course of a year. We cannot fully exclude that some selective processes occurred in the colony, but we believe that the fact that this pattern is detectable in both treatments and that the time since establishment of the colony is comparatively short, suggesting that the role of selective processes is negligible in producing our results.
4.4. Linkages among the Components of the Study
We combined multiple experimental methods to gain insight into the complex interaction of organismal responses to an important environmental cue. While our study was not designed to mechanistically link the three sets of observations, individuals subject to ALAN had reduced clock-gene expression that was more pronounced in females, with lower activity levels, and fewer eggs per raft that were smaller in size. We propose the hypothesis that individuals subjected to ALAN rested more often and fed less frequently and that these three sets of observations are therefore related. We did not measure individual fecundity, total feeding time, sugar consumption, or growth rate, so we cannot make this link with certainty. Nonetheless, it seems unlikely that the altered clock gene expression, reduced activity and decreased reproductive output are unrelated. Our observations provide insights into important processes that are influenced by ALAN. This adds to the currently sketchy body of knowledge while at the same time highlighting interesting avenues of research.
4.5. Considerations of the Design
The laboratory setup allowed us to control critical environmental parameters (temperature, humidity) and simulate daily light cycles. Seasonal cues were absent in order to minimize confounding factors such as shifts in peak expression of clock genes depending on day length [75
]. Regularly attending the colony to monitor densities, supply food, change water, and collect egg rafts, may have introduced olfactory cues to the females. This may have lead to different results than would occur in an environment without hosts or with hosts always present. Importantly, potential cues were introduced to both experimental groups and therefore results are comparable, albeit not directly applicable to natural populations. The climate chamber was separated into two compartments, limiting the possibilities of replication and perhaps leading to divergence between the two groups. Individuals used for the gene expression experiments were taken from a single source population (control), and largely support the differences obtained in the other two experiments. Activity in LD was similarly low during the day phase in both treatments and in constant darkness. This indicates that the groups did not differ systematically in their activity and that behaviors were triggered by the light-environment. We chose not to translocate individuals from the control group for the activity experiment because we wanted to exclude the possibility of aberrant behavior caused by a new environment and the differences of the two treatments. We therefore believe that the experimental design was adequate given the logistical constraints on any such experiment. However, additional experimentation is needed to gain a better mechanistic understanding and to establish the consistency of responses across populations.
The molestus-ecotype is thought to be restricted to warmer climates due to the fact that it does not undergo diapause. This is why in urban settings they are often, but not exclusively, found in below-ground structures like tunnels [22
]. This may result in higher sensitivity to artificial light at night. On the other hand, most below-ground structures used by mosquitoes in urban areas are equipped with artificial light sources, thus populations may be adapted to it. Our study showed that artificial light at night does impact a number of relevant processes in this mosquito and that this may have negative effects on the individual and even the population as a whole. However, long-term studies are needed to clarify whether the influence we detected is indeed detrimental or whether the mosquito can avoid this, for example by adjusting behavior such that the impact is minimized.
A recent study estimating genetic divergence between pipiens and molestus revealed that processes related to the different habitats differed between ecotypes [75
]. However, processes involved with body/cell maintenance (e.g., signal transducers and transcription regulators) were highly conserved between ecotypes suggesting similar effects on both [75
]. This question could be addressed by directly comparing the response of the two ecotypes to ALAN. We hypothesize that, given the conserved nature of most of the examined traits, we would find similar patterns in Cx. pipiens
. This raises important questions relevant to pest control. Changes in activity in response to ALAN, either directly or potentially as downstream effects of altered circadian regulation, might influence host-seeking behavior in mosquitoes thus altering disease transmission dynamics. It has been reported for Aedes aegypti
that a diurnal mosquito may make use of ALAN by prolonging its host-seeking phase, thereby potentially increasing the transmission rate [76
]. Nocturnal species may shift the timing of host-seeking and resting behavior, and could potentially avoid certain control measures with negative effects for nuisance control. In summary, our results highlight the importance of further investigations into the impact of ALAN on mosquitoes with respect to mosquito control and vector capacity.