Larval superiority of Culex pipiens to Aedes albopictus in a replacement series experiment: prospects for coexistence in Germany

Background The Asian tiger mosquito Aedes albopictus is an extremely invasive, globally distributed and medically important vector of various human and veterinary pathogens. In Germany, where this species was recently introduced, its establishment may become modulated by interspecific competition from autochthonous mosquito species, especially Culex pipiens (s.l.). While competitive superiority of Ae. albopictus to Cx. pipiens (s.l.) has been described elsewhere, it has not been assessed in the epidemiological conditions of Germany. The present study aimed to determine if such superiority exists under the physicochemical and microclimatic conditions typical for container habitats in Germany. Methods In a replacement series experiment, the larval and pupal responses of Ae. albopictus and Cx. pipiens (s.l.) (mortality, development time, growth) to interspecific interaction (five larval ratios) at (sub-)optimal temperatures (15, 20 and 25 °C) and differing food supply (3 and 6 mg animal-based food larva-1) were investigated using a randomized split-plot design. In addition to physicochemical measurements of the test media, natural physicochemical conditions were determined for comparative analyses in mosquito breeding sites across the Rhine-Main metropolitan region of Germany. Results Under the physicochemical and microclimatic conditions similar to the breeding sites of the Rhine-Main region, competitive superiority of Cx. pipiens (s.l.) to Ae. albopictus in terms of larval survival was more frequently observed than balanced coexistence. Food regime and multifactorial interactions, but not temperature alone, were controlling factors for interspecific competition. Larval food regime and the larval ratio of Ae. albopictus influenced the physicochemistry and algal growth at 15 °C, with increased Ae. albopictus mortality linked to a decreasing number of Scenedesmus, Oocystis and Anabaena algae. Conclusions Under the present environmental conditions, the spread of Ae. albopictus from isolated foci in Germany may generally be slowed by biotic interactions with the ubiquitous Cx. pipiens (s.l.) (and potentially other container-breeding mosquito species) and by limnic microalgae in microhabitats with high resource levels. Detailed knowledge of the context dependency in temperate mosquito ecology, and interrelations of physicochemistry and phycology may help to achieve a better understanding of the upcoming Ae. albopictus colonization processes in central and northern Europe. Electronic supplementary material The online version of this article (10.1186/s13071-018-2665-3) contains supplementary material, which is available to authorized users.


Background
The extremely invasive mosquito Aedes (Stegomyia) albopictus Skuse 1894, a known vector of various human pathogens like dengue and chikungunya viruses, is gradually extending its distribution to colder climate regions [1]. At present, the northernmost breeding populations of this species in Europe occur in Freiburg [2], Heidelberg and Jena, Germany (Norbert Becker, personal communication, June 2016). A further spread of Ae. albopictus into the Rhine rift valley and parts of Bavaria and North Rhine-Westphalia, Germany, is very likely due to climatic habitat suitability and man-made features [3,4]. Apart from the frequently suboptimal low temperatures in these regions, the establishment of Ae. albopictus may be modulated by interactions with resident container-breeding mosquitoes [5][6][7].
Aedes albopictus was shown to be a strong competitor for resident mosquito species such as Aedes triseriatus as well as exotic species like Aedes aegypti in the United States of America [7,8]. However, environmental variation may change competitive balances [9] as recently demonstrated in Japan [10]: At Mount Konpira, Japan, the formerly predominant Ae. albopictus is currently becoming superseded by the formerly rare, but now superior competitor Aedes flavopictus, most probably due to changes in differential responses to environmental variability and altered species interactions [10]. As the competitive superiority of container-breeding mosquitoes strongly depends on context and is reversible, resident mosquitoes may have the potential to act as superior competitors in colder ecoregions due to their better adaptation to low temperatures than the invader Ae. albopictus which has a (sub-) tropical origin [11].
Negative interactions between different mosquito species frequently occur during their aquatic larval developmental stages and can be attributed to direct (chemical/ physical interference) or indirect interactions (resource competition), generally known as interspecific competition [7,9,12]. Numerous laboratory investigations have corroborated that the pattern of the heterospecific larval response of Ae. albopictus depends not only on the competitor but also on the environmental context [9], food quantity or quality [5,[13][14][15][16][17] and food-temperature interactions [18,19]. Larval resource competition has been considered as a probable reason for the decline of Ae. aegypti in concurrence with the arrival of Ae. albopictus in the United States [15,20]. Furthermore, species-specific differences in resource utilization are thought to shape microhabitat-specific mosquito assemblages [15,16,21]. Water chemistry, as a direct function of available food resources and resource utilization by mosquitoes, has also been considered to control niche differentiation [22][23][24]. For instance, concentrations of nitrite and nitrate and the pH shaped the local distribution of the exotic Aedes notocriptus and the endemic Culex pervigilans in New Zealand [24]. Furthermore, certain microalgal species have been shown to exclude Ae. albopictus from microhabitats in Hawaii [22]; consequently, microalgae have been recommended as agents of biological mosquito control [25].
The ecology of container-breeding mosquitoes in temperate regions is still poorly understood. To achieve a better mechanistic understanding of the upcoming Ae. albopictus colonization processes in Germany, we tested whether the described competitive superiority of Ae. albopictus over Cx. pipiens (s.l.) also exists under the physicochemical and microclimatic conditions that are typical for container habitats in the Rhine rift valley of Germany. We hypothesized that (i) competitive superiority depends on the speciesspecific thermal tolerance spectrum, and therefore Ae. albopictus might be advantaged at higher and Cx. pipiens (s.l.) at lower temperatures, respectively. We further hypothesized that (ii) food regime, as a major determinant for interspecific interactions, interacts with temperature and that (iii) resource competition between Ae. albopictus and Cx. pipiens (s.l.) is directly connected to physicochemical and phycological parameters of their microhabitat.

Natural microhabitats of mosquito larvae Physicochemistry
In a radius of 50 km around the city of Frankfurt am Main (Rhine-Main Metropolitan region, northern border of the Rhine rift valley, Germany), intermittent and permanent waterbodies of natural and anthropogenic origin (natural: temporary and permanent puddles, flooded patches/pools in forests and meadows, tree holes, root and rock cavities; anthropogenic: agricultural plastic covers, troughs on paddocks, water storage basins, fountain basins, planters, vases, barrels, car tires, buckets, watering cans) were monitored for the occurrence of mosquito larvae from 27th August to 27th October 2010 (Additional file 1: Figure S1).

Replacement series experiment Experimental design
The interspecific larval interaction of Ae. albopictus and Cx. pipiens molestus under different food regimes and temperatures was investigated in a replacement series experiment using a randomized split-plot design. The replacement series experiment with 4 replications was designed to test the impact of the whole plot factors 'species' (2 levels) × 'larval ratio' (5 levels) × 'temperature' (3 levels) and the subplot factor 'larval food regime' (2 levels) on survival, development time and growth of the aquatic life stages of the 2 species.

Mosquito material
The mosquitoes used in the experiments were purchased from Biogents AG (Regensburg, Germany) and reared inhouse. We used 2 well-adapted long-term laboratory strains (Ae. albopictus strain with origin Singapore, Cx. pipiens biotype molestus with origin Regensburg, Germany) to minimize unknown impacts of the field histories of the mosquito strains on their responses and better determine the fundamental nature of the context-dependency of interspecific interactions.
The eggs of Ae. albopictus were collected on filter paper, dried at 20°C and 90% relative humidity, and exposed to a yeast solution stimulating larval hatching [33]. Firstinstars of Cx. pipiens molestus were released from egg rafts into 1:1 tap: deionized water (hereafter called larval medium). In total, 3600 larvae were investigated.
First-instar larvae of the two species, which had been synchronously released within 24-48 h, were allotted as pure or mixed cohorts of 30 larvae in 1 l test vessels (plastic cups) filled with 600 ml larval medium each.

Larval ratio
In total, 2 pure and 3 mixed larval groups were tested at 3 temperatures and under 2 food conditions. Following the recommendations of Oberg et al. [34], experimental units with only Ae. albopictus larvae (Ae: Cx 30:00 ) and only Cx.

Temperature
The experimental units were exposed to 15°C, 20°C and 25°C at a 16:8 h light:dark photocycle in environmental chambers (MKKL 1200, Flohr Instruments GmbH, Utrecht, The Netherlands). These experimental temperatures had been derived from in situ measurements in potential mosquito microhabitats (Fig. 1).

Larval food supply
The larvae were fed with animal-based fish food (Tetra Min®, Tetra, Melle, Germany) using a total amount of either 3 mg larva -1 or 6 mg larva -1 . Larval food was provided on days 0, 3, 6, 9, 12, 15, 18 and 21 in the 15°C treatments, on days 0, 2, 4, 6, 8, 10, 12 and 14 in the 20°C treatments, and on days 0, 2, 4, 5, 7, 8, 9 and 10 in the 25°C treatments. Food portions of either 0.25 or 0.5 mg larva -1 were offered during the first to fourth feeding and 0.5 or 1 mg larva -1 at later ones. A total food quantity of 3 mg larva -1 was considered as limited (at least at 20°C) and 6 mg larva -1 as adequate for Ae. albopictus larval development (at least at 20°C and 25°C; see Müller et al. [38]). The feeding protocol also allowed for an optimal development of Cx. pipiens molestus at 20°C and 25°C as shown by Kreß et al. [33].

Effect variables
Mortality, pupal development time and pupal growth parameters (size, weight) were examined. During the daily census, pupae were collected and preserved in 70% ethanol. Pupal abdominal length (AL, 3 rd to 8 th abdominal segment) was measured using a stereo microscope and the software DISKUS (Carl H. Hilgers, Königswinter, Germany) within 14 days after collection. The gender of the pupae was determined by examination of their genital lobes. Species identity was determined using 4 criteria because overlapping morphological traits were occasionally observed: (i) the colour of the prothoracic trumpets (respiratory tubes); (ii) general pigmentation; (iii) paddle marginal spicule hairs; and (iv) compaction of the habitus. After gender and species identification, individual pupae were dried at 60°C for 14 ± 2 h, and their dry weight (DW) was measured using a microbalance (± 0.01 μg, Sartorius model 708501, Göttingen, Germany).

Physicochemistry in test vessels at 15°C
Physicochemical effects on larval ratio, temperature and food regime were investigated in the medium of test vessels kept at 15°C due to long algal growth. After the last pupation had occurred in a given replicate, the following parameters were quantified (in parentheses: test used): standing time of vessel (days until last pupation occurred), concentrations of chlorophyll a (algal cells > 0. 45  In every environmental chamber used in this study, there had previously been some vessels with Scenedesmus obliquus cultures and outdoor mesocosm water which probably led to an aerial proliferation of algal cells. Therefore, 10 ml aliquots of test media kept at 15°C were diluted (1:10) with tap water, and phytoplankton were collected for 3 h in Utermöhl chambers. The number of algal cells per ml and replicate was estimated based on the microscopic counting of 2 random fields of 116.28 mm 2 (Axiovert 40 C microscope, ZEISS; DISKUS software, Carl H. Hilgers, Königswinter). Algae were identified to family or genus level according to Streble et al. [40] and Linne von Berg [41].

Data analysis
Data processing, graphing and statistics were performed using Microsoft Excel, GraphPad Prism version 7 and Statistica version 8 software. Effect variables and physicochemical measures are reported as mean ± standard deviation unless stated otherwise.

Data processing
Means of species-and gender-specific DWs and ALs were tabulated per replicate. Also, species-specific mortalities (%), gender-specific averages of the mean pupation time (PT 50 ) and gender-specific daily biomass accumulation , pH, conductivity (μS cm -1 ) and dissolved oxygen concentration (mg l -1 ) (mean ± SD) and respective Cx. pipiens abundance [# larvae: number of larvae (n = number of microhabitats)] are shown. Grey bars denote overall 95% confidence intervals. Key: 1, car tires; 2, vases; 3, rain barrels; 4, small reservoirs (bucket, plastic tarpaulin, pot, steel girder); 5, large reservoirs (metal and plastic tub, excavator shovel, water bowl); 6, dung discharge; 7, puddles; 8, ponds. b Spline curve of daily water temperature [mean ± SD] in 1 litre plastic cups logged from mid-May to end of July 2011. The cups representing potential mosquito habitats had been set up in shaded versus sunny localities in Frankfurt am Main rates (BA) were calculated for every replicate. The PT 50 indicated the time when 50% of male or female larvae had metamorphosed to pupae (for detailed calculations see Müller et al. [38]). The BA [mg larva -1 day -1 ] was calculated as the average for each gender and replicate by the division of the root of the mean DW and the root of the PT 50 . It should be noted that the BA neglected the initial DW of first-instar larvae.

Statistics
The mosquito and physicochemical datasets were tested for Gaussian distribution using the Kolmogorov-Smirnov test and the D' Agostino Pearson omnibus normality test (P ≤ 0.01). Variance homogeneity of the data was tested using the Cochran's test at the level of P ≤ 0.01. After confirmation of variance homogeneity for arcsin and logtransformed mortality [%], log-transformed PT 50 data (gender-specific) and parameter-specific RCC values, fourfactorial ANOVAs (independent factors: species, larval, temperature, food regime) and subsequent Tukey HSD tests were computed for these response variables. The heteroscedastic, but normally distributed dependent variables AL, BA and DW were analyzed with general linearized normal-log models with backward factor removal during model building and subsequent Wald statistics. The total average mortality of the two species (irrespective of experimental treatment) was compared with a t-test.
For reasons of straightforward whole-plot interpretation, polynomic regressions of second order (least squares fit) were computed with mortality data against larval ratio. Also, first order regressions (least squares fit) were computed with AL, DW and BA data against the number of larvae of the other species, respectively. Extra-sum-of-squares-tests were applied to compare independent fits with global fits for mosquito responses to food regime within a temperature block (mortality, corrected by Benjamini-Hochberg (BH) procedure with false discovery rate (FDR) of 0.05) and independent fits with global fits for each species and gender and within each temperature (AL, BA, BH corrected with FDR of 0.05), respectively.
Water quality parameters examined at 15°C were tested by 2-way ANOVA (P < 0.05) for their dependence on initial larval ratio and food. Water quality parameters were further tested for correlation with algal growth (Spearman). The dependence of algal growth at 15°C from initial larval ratio or larval mortality and food regime and their interactions were tested with 2-way ANOVAs. The algal growth data at 15°C were tested for Spearman's correlation with Ae. albopictus mortality at 15°C and Cx. pipiens mortality at 15°C, respectively. Also, non-parametric Spearman's correlation analyses with the initial larval quantity of Ae. albopictus or Cx. pipiens molestus and the cell number of specific algal groups were performed for both food-related subplots.

Outdoor versus experimental temperature
From mid-May to the end of July 2011, the water temperature in 1 l plastic cups was on average 19.2 ± 1.8°C at the shaded site and 20.6 ± 4.8°C at the sunny site (Fig. 1b). Minimum daily average temperatures were 14.2°C and 14.1°C in the shaded and sun-exposed test vessel, and maximum daily average temperatures, 25.3°C and 27.9°C, respectively. Thus, the temperatures chosen for the replacement series experiment (15, 20 and 25°C) mirrored the minimum, mean and maximum daily average water temperatures in potential mosquito microhabitats during spring and summer in Frankfurt am Main, Germany.

Multifactorial impact on mosquito response General factorial response
The single factors species and temperature influenced every tested effect variable significantly (ANOVA, Table 1). This was analogous to the factor larval food regime except its non-significance for PT 50 and male BA. In contrast, the single factor larval ratio had a significant influence on mortality and PT 50 only. At the level of two-factorial Table 1 Results for four-factorial significance tests (df, degree of freedom) with the independent factors species (Aedes albopictus, Culex pipiens), food regime (3 and 6 mg larva -1 food supply), larval ratio (0, 10, 15 and 20 larvae of the other species) and temperature (15, 20 and 25°C interactions, food × temperature was always significant and species × food and species × temperature almost always significant for mosquito response variables (16 out of 18 sub-models). In contrast, the larval ratio in combination with species, food or temperature was significant in only 8 out of 27 sub-models. Three-factorial interactions of species × food × larval ratio strongly influenced the data (except AL and male DW), as did species × food × temperature (except BA). The triple interaction species × larval ratio × temperature had a significant influence on mortality, male PT 50 and AL, whereas food × larval ratio × temperature had no significant influence on any effect variable. The four-factorial interaction was significant for female BA, female PT 50 and male DW only (Table 1).

Mean pupation time (PT 50 )
Species and temperature were the major factors affecting the PT 50 of male and female larvae (Table 1). Also, a larval ratio significantly influenced the PT 50 of female larvae while the single factor food had no significant effect (P > 0.08). The PT 50 of male and female larvae was additionally influenced by the 2-factorial interactions species × temperature, food × temperature and species × food and the 3-factorial interactions species × food × larval ratio and species × food × temperature. The triple interaction species × larval ratio × temperature had a significant effect on the PT 50 of male, and the 4-factorial combination on that of female larvae only .
The PT 50 of both genders and species decreased as a function of temperature, with maxima at 15°C and minima at 25°C (Fig. 3). At 25°C, a food supply of 3 mg larva -1 decreased the mean PT 50 of females by 2.1 days (Ae. albopictus) and 0.7 days (Cx. pipiens molestus) and the mean PT 50 of males by 2.1 days (Ae. albopictus) and 0.5 days (Cx. pipiens molestus) if compared to the cohorts fed 6 mg larva -1 at 25°C. At lower temperatures, the PT 50 had an opposite trend with a higher average PT 50 of females of 1.9 days (Ae. albopictus) and 0.8 days (Cx. pipiens molestus) at 15°C and 1.2 days (Ae. albopictus) at 20°C if larvae had been supplied with 3 mg food larva -1 , although the average PT 50 of Cx. pipiens molestus was similar in both food treatments at 20°C. This general pattern of 2-and 3-factorial interactions of species × food × temperature equally applied to the average PT 50 of male larvae that had received 6 mg food larva -1 , which was 0.3 days (Ae. albopictus) and 1.6 days (Cx. pipiens molestus) shorter at 15°C and 0.7 days (Ae. albopictus) and 1.1 days (Cx. pipiens molestus) shorter at 20°C than the average PT 50 of males that had been fed with 3 mg larva -1 . With regard to larval ratio, the PT 50 of females decreased with an increasing larval ratio of the competing species, but this was only observed in the cohorts of Ae. albopictus provided with 3 mg food larva -1 and those of Cx. pipiens molestus given 6 mg larva -1 .

Pupal growth
Pupal growth (AL, DW) was mainly influenced by the single factors species, temperature and food and their 2-and 3-factorial interactions ( Table 1). The larval ratio had a significant effect only if it interacted with the factors species (DW), food (AL of females, DW of males) and temperature (AL, DW of males) or in concert with species × food (DW of females) and species × temperature (AL). In the following, the detailed response patterns are described exclusively for the pupal growth of females due to their biological significance for size/weight-fecundity-relationships [21]. However, male response patterns were similar.
The mean DW of female Cx. pipiens molestus (0.80 mg) was higher than that of female Ae. albopictus (0.58 mg). In both species, the DW of females was always largest in the cohorts with 6 mg larva -1 food supply (Additional file 3: Figure S3). In the cohorts fed 6 mg larva -1 , the DW of Ae. albopictus increased with increasing temperature, whereas the DW of Cx. pipiens molestus decreased with increasing temperature. In the cohorts provided with 3 mg food larva -1 , there was no temperature-dependent change of DW. By overall comparison, species-specific differences in DW were minor at 25°C under both food regimes. In the Cx. pipiens molestus cohorts given 6 mg food larva -1 at 15°C and 20°C, a decreased DW became apparent as heterospecific larval numbers increased, but this was not the case at 25°C nor for Ae. albopictus (Additional file 3: Figure S3).

Daily increase in weight (BA)
The daily increase of weight (BA) was mainly influenced by the factors species (mean BA Ae = 0.053 mg larva -1 day, mean BA Cx = 0.058 mg larva -1 day) and temperature (Table 1). Minimum and maximum values of BA were always found at 15°C and 25°C (cohorts fed 3 mg larva -1 ), demonstrating the strong impact of temperature on BA. The single factor food had a significant influence on females only, but food × species or food × temperature and the 3-factorial interaction species × food × temperature had significant effects on both males and females. The BA of both species was similar to 3 mg larva -1 food supply except Ae. albopictus at 25°C, but the BA was lower in Ae. albopictus than Cx. pipiens molestus cohorts fed with 6 mg larva -1 .
Similar to AL, the BA of Ae. albopictus fed 6 mg larva -1 at 25°C slightly decreased with an increasing number of Cx. pipiens molestus larvae, whereas the BA of those fed 3 mg larva -1 increased with an increasing number of Cx. pipiens molestus larvae (Fig. 4). In contrast, the BA of Cx. pipiens molestus did not change with different heterospecific larval ratios. This complex pattern led to larval ratio having a significant influence only within the combination species × food × larval ratio or within the 4-factorial interaction (Table 1).

Relative crowding coefficient (RCC)
Both species exhibited a weak to intermediate interspecific competitiveness with regard to growth and development as indicated by RCC values between 0.91 and 2.31 (Table 2;  Additional file 4: Table S1; [18,43]). Relative crowding coefficient values < 1.39 were observed in 38 out of 96 experimental groups indicating no competition between species in 39.6% of treatments (Additional file 4: Table S1). Values   Table 2 Relative crowding coefficients (RCCs). Integrated relative crowding coefficient values for female and male Aedes albopictus and Culex pipiens in different experimental treatments indicate a generally low competition between species regarding growth and development due to equilibration of the traits of different species. To evaluate the predomination of one species trait over the other in more detail, the differences between RCC Ae and RCC Cx that are based on biomass accumulation rate (BA), pupal dry weight (DW), pupal abdominal length (AL) and mean pupation time (PT 50 ) are additionally reported, whereas the respective parameter-specific RCCs can be found in the Additional file 4: The pattern of interspecific competitiveness described by parameter-specific RCCs was similar between genders but altered by larval exposure conditions and the parameter. The number of groups with the superiority of Ae. albopictus or Cx. pipiens molestus (groups with negative versus positive parameter-specific RCC differences) was balanced ( Table 2). Disregarding temperature, PT 50 and BA produced a mirrored RCC difference pattern with the dominance of the PT 50 -RCC and a parallel lesser BA-RCC in the Ae. albopictus cohorts fed 6 mg larva -1 and the Cx. pipiens molestus cohorts given 3 mg larva -1 , respectively. The DW-RCC of Ae. albopictus was larger in the Ae. albopictus cohorts fed 6 mg larva -1 and those fed 3 mg larva -1 at 20°C. The AL-RCC was higher in Ae. albopictus than in Cx. pipiens molestus cohorts at 15°C, but the AL-RCC of Cx. pipiens molestus became higher than in Ae. albopictus at a higher temperature.
The integration of RCCs revealed that differences from growth and development specific RCCs were almost balanced out given that the single factors species, gender, food and temperature and almost all interactions had no significant influence. Notably, the integrated RCC Ae was higher in the 6 mg larva -1 food treatment and the integrated RCC Cx in the 3 mg larva -1 food treatment (species × food interaction, P < 0.001).

Water parameters in test media of 15°C treatments
The total standing time of test vessels in parallel with the maximum pupation time was altered by larval ratio and larval food regime: The longest standing time of 40 ± 0 days was observed in the Ae:Cx 00:30 cohort provided with 6 food mg larva -1 and the shortest standing time of 33 ± 2 days in the Ae:Cx 00:30 cohort fed with 3 mg larva -1 (Table 3). Physicochemical water parameters depended significantly on the larval food regime (ammonium, nitrate, nitrite, silicon dioxide, oxygen, pH), the larval ratio (nitrate) and their interaction (silicon dioxide). At the end of the half-life-cycle tests where larvae had been fed with 6 mg larva -1 , the concentrations of nitrate (2.4 to 6-fold) and potentially toxic nitrite (2-fold) were lower than in the water of 3 mg larva -1 cohorts. In contrast, concentrations of ammonium (1.7 to 18-fold), silicon dioxide (1.8 to 5.6-fold; apart from 0.45-fold in the Ae:Cx 30:00 treatment), oxygen (1.1 to 1.3-fold) and pH (1.2 to 1.3-fold) were higher than in the water of cohorts fed with 3 mg larva -1 (Table 3).

Discussion
In the present study, we show that the physicochemical and microclimatic conditions of potential and actual mosquito breeding sites in Germany offer suitable conditions for a balanced coexistence and frequent competitive superiority of Cx. pipiens molestus over Ae. albopictus with regard to larval and pupal survival. We demonstrate that the interspecific competition pattern of these two mosquito species does not directly depend on the thermal regime (rejection of hypothesis i), but is rather controlled by food regime and species-food and species-temperature interactions (support of hypothesis ii). Furthermore, we discuss how the resource competition between Ae. albopictus and Cx. pipiens molestus is directly connected to species-specific foraging behaviour and physicochemical as well as phycological microhabitat parameters (support of hypothesis iii).

Larval coexistence under present climatic conditions in Germany
In general agreement with our data, Ae. albopictus and Cx. pipiens molestus had an almost identical population growth at 20°C but a divergent one at a higher temperature [33]. A simple extrapolation of the laboratory optimum to mosquito co-occurrence in the field suggests that microhabitats in Germany with a typical median temperature of 20°C in spring and summer offer favourable conditions for the coexistence of these two species, even if the food regime may ultimately restrict the success of Ae. albopictus in mixed larval habitats. The longer development time of Ae. albopictus at 15°C and Cx. pipiens molestus at 20°C and 25°C will directly translate to the population level [44] suggesting a shift from Cx. pipiens (s.l.) with advantaged population activity at 15°C to Ae. albopictus at a higher temperature. During warmer seasons, Ae. albopictus may make up its population delay by faster development time, faster growth and superior biomass accumulation. This experimentally based expectation has support from field observations elsewhere [18,27,28]: In Italy, for example, the population growth of Cx. pipiens (s.l.) starts in mid-April and peaks in July, and that of Ae. albopictus in mid-May and September, respectively [18]. Marini et al. [28] calculated for Ae. albopictus an average delayed population growth by 29 days when compared with Cx. pipiens (s.l.) from the same location and year.
Our results indicate a general adaptation of Cx. pipiens to low temperatures although the thermal optimum of Cx. pipiens (s.l.) is between 20°C and 25/27°C (present study, [45,46]). In contrast, the higher total average mortality of Ae. albopictus and the especially high mortality at 15°C suggest that Ae. albopictus is less adapted to colder ecoregions than its autochthonous competitor. However, despite these differences in thermal biology, the outcomes from available competition studies for Ae. albopictus and Cx. pipiens (s.l.) might be transferrable to colder ecoregions [18,27], because the temperature did not influence the interspecific pattern at other than optimal test conditions. We, therefore, reject hypothesis i, that a competitive superiority depends on the species-specific thermal tolerance spectrum and therefore Ae. albopictus might be advantaged at higher and Cx. pipiens (s.l.) at lower temperatures. Similarly, temperatures ranging from 24 to 30°C did not change the larval competition between Ae. albopictus and A. aegypti [6], and a range of 15°C to 31°C did not alter the larval competition between Ae. albopictus and O. triseriatus [5].
The competitive superiority of Culex pipiens over Aedes albopictus -or vice versa?
All available evidence from this study suggests a competitive superiority of Cx. pipiens molestus over Ae. albopictus with regard to survival at every other temperature and food treatment than 20°C and 6 mg larva -1 . Agonistic growth and development strategies levelled out the parameterspecific sublethal crowding effects. Mortalities comparable to the ones registered in the present study have been Fig. 5 Algal growth in test vessels at 15°C. Algal cells (mean and SD) grown in test media with pure and mixed cohorts of Aedes albopictus and Culex pipiens larvae in dependence of larval ratio (Ae:Cx -number of Ae. albopictus: Cx. pipiens larvae) and larval food regime with (a) 3 mg larva -1 larval food supply versus (b) 6 mg larva -1 larval food supply at 15°C. Algal growth was not observed in test vessels at 20°C and 25°C observed in Cx. pipiens (s.l.) cohorts regardless of the presence or absence of larvae of a second mosquito species, whereas the presence of Cx. pipiens larvae especially affected Ae. albopictus survival and biomass accumulation at 25°C and 3 mg larva -1 food supply. Carrieri et al. [18] observed balanced survival and development in their Ae. albopictus and Cx. pipiens (s.l.) cohorts exposed to 25°C with 2.85 mg larva -1 food supply, but growth differed significantly between species indicating a slightly suboptimal food supply especially for Cx. pipiens (s.l.). As an additional contrast to our study, the powerful competitive superiority of Ae. albopictus over Cx. pipiens (s.l.) became apparent at poorer larval food conditions at 20°C and 25°C [18]. Likewise, Costanzo et al. [27] reported competitive superiority of Ae. albopictus over Cx. pipiens (s.l.).
The contrary outcomes from the replacement series experiments in this study and those by Carrieri et al. [18] and Costanzo et al. [27] are not expected to result from the use of different Ae. albopictus mosquito strains [47,48] but more likely from the investigation of different members of the Cx. pipiens complex with differential autecological characteristics [26]. The autogenic and stenogamic biotype Cx. pipiens molestus shows a homodynamic development, strong synanthropic behaviour, mammal feeding type and urban to suburban distribution [30,49,50]. This ecotype often occupies biotopes different from those of the ornithophilic Cx. pipiens pipiens in regions with cold temperate climate (hypogeous, [30]), although Cx. pipiens molestus is not excluded from epigeous breeding sites in the Rhine rift valley and warm temperate European regions [49,51,52].
However, resource competition may give a more reliable explanation for the opposite competition pattern observed between the present study and those by Carrieri et al. [18] and Costanzo et al. [27], because different types of food were supplied (animal-based food versus plant-based detritus), species-specific food quantity responses can be assumed [17,18] and differential food quantitytemperature interactions may have taken place [5,33,38].

Larval food supply is decisive for competitive superiority
Hypothesis ii stating food regime is a major determinant for interspecific interactions with temperature becomes supported by the comparison of our data with literature. In an overall comparison of three competition studies (present study, [18,27]), the food supply in our experiment was perhaps always ad libitum as implied by evenly balanced biomass accumulation rates. On the other hand, pupal sizes and weights were always larger in the cohorts fed with 6 mg larva -1 . This surplus of food may especially have favoured Cx. pipiens molestus in our study, whereas the progressively limited food resources in the studies by Carrieri et al. [18] and Costanzo et al. [27] may have resulted in a competitive superiority of Ae. albopictus. Similarly, O. triseriatus had a better survival than Ae. albopictus in water sampled from tree-holes with significant detritus whereas probably limited resource levels in tire water favoured Ae. albopictus [21]. In another replacement series experiment, the larval superiority of Ae. albopictus over O. triseriatus was observed under 0.5 mg but diminished under two mg larval food supply [5].
Food quality could have alternatively triggered the different competitive interactions between Ae. albopictus and Cx. pipiens molestus given that spatiotemporal natural variation in detritus quality (in terms of the animal to plant-based ratio and associated microbial growth) can shape mosquito assemblages [53]. It has been shown that the detritus type, its decay rate and associated microbial growth influence the interspecific competition between Ae. albopictus and O. triseriatus and Ae. aegypti, respectively [17,53]. It has also been shown that Ae. aegypti is weaker in competition than Ae. albopictus if larvae feed on oak, pine, or insect detritus (related to low to intermediate microbial growth) but face competition if feeding on grass which is related to high microbial growth [17]. Aedes albopictus has been described as an opportunistic feeder [54] with great resistance to starvation [14], but animalbased detritus and associated microorganisms always yield higher performance of Ae. albopictus [20,38,53,55,56].
Under resource-limiting conditions (limiting at least for Cx. pipiens molestus), Ae. albopictus with its continuously foraging behaviour may usurp detritus, and specifically essential nutrients, more quickly than the selectively feeding larvae of Cx. pipiens (s.l.) which rest between meals [22,56,57]. Culex pipiens molestus larvae are known to react to phagostimulants released from high-value food which may allow the species to control foraging and filtering activity to usurp mostly food particles with high nutritional value [57,58]. However, the opportunistic features of Ae. albopictus might not be successful in microhabitats with eutrophic conditions, as observed at 25°C and 6 mg larva -1 food supply where the BA of Ae. albopictus even decreased, and the PT 50 increased if compared to 3 mg larva -1 cohorts.
Phycology is a crucial factor for Ae. albopictus survival Our study demonstrates that resource competition between Ae. albopictus and Cx. pipiens molestus is directly connected to physicochemical and phycological parameters of their microhabitat (hypothesis iii). The competitive disadvantage of Ae. albopictus with lower survival under eutrophic conditions became very substantial in the cohorts fed 6 mg larva -1 at 15°C. Here, the algal growthmortality relationship directly pointed towards the importance of phycology in temperate mosquito ecology. Notably, Ae. albopictus ingests every available particle very quickly, especially green algal cells with an average of 22.4 ± 0.33 cells per second [59]. The green and blue-green algal growth in our study (Oocystis, Scenedesmus, Anabaena) were in fact negatively correlated to the initial Ae. albopictus larval ratio when 6 mg larva -1 food was supplied. Marten [22,25] demonstrated a starvation effect of a variety of algal species, and in particular the lethal effect on first-and second-instars of Ae. albopictus if feeding on indigestible algal cells. For instance, larval feeding on Scenedesmus obliquus (the likely Scenedesmus species in our study) almost always killed Ae. albopictus first-and second-instar larvae, whereas Ae. albopictus first-and second-instar larvae survived feeding on some (but not every) species of Scenedesmus, Oocystis and Anabaena [25].

Water physicochemistry suits both mosquito species
Water physicochemistry in the test vessels well resembled that of Cx. pipiens (s.l.) microhabitats at the northern border of the Rhine rift valley in Germany. The mosquito larvae had a direct impact on nutrient cycling (e.g. nitrate, silicon dioxide concentration) and algal growth, at least in our test vessels. According to physicochemistry, the ecology of Cx. pipiens (s.l.) is described as euryoecious [26] (βmesosaprob microhabitats, [60]). For Ae. albopictus, however, the physicochemical parameters in the replacement series experiment were sometimes higher than reported in the few published studies in their natural microhabitats (pH = 7.97, dissolved oxygen ≤ 9.7 mg l -1 , [23,61]). A pH between 6.8 and 7.6 seems optimal, although Ae. albopictus larvae were observed within a pH range of 5.2 to 8.4 and at dissolved oxygen levels of 1.3 mg l -1 [60]. Wu & Chang [62] reported the greatest rate of food displacement of Ae. albopictus in water with a pH of 5.5. However, given that physicochemical conditions similar to our experiment were interference-free for Ae. albopictus larval/pupal performance at 20°C to 30°C [38], the physicochemistry of natural Culex spp. microhabitats in warmer regions of Germany, specifically in the Rhine-Main Metropolitan region, would not prevent an establishment of Ae. albopictus.