Stable Isotope Analysis Reveals Detrital Resource Base Sources of the Tree Hole Mosquito, Aedes triseriatus

Abstract

1. Detritus that forms the basis for mosquito production in tree hole ecosystems can vary in type and timing of input. We investigated the contributions of plant- and animal-derived detritus to the biomass of Aedes triseriatus (Say) pupae and adults by using stable isotope (¹⁵N and ¹³C) techniques in lab experiments and field collections.

2. Lab-reared mosquito isotope values reflected their detrital resource base, providing a clear distinction between mosquitoes reared on plant or animal detritus.

3. Isotope values from field-collected pupae were intermediate between what would be expected if a single (either plant or animal) detrital source dominated the resource base. However, mosquito isotope values clustered most closely with plant-derived values, and a mixed feeding model analysis indicated tree floral parts contributed approximately 80% of mosquito biomass. The mixed model also indicated that animal detritus contributed approximately 30% of mosquito tissue nitrogen.

4. Pupae collected later in the season generally had isotope values that were consistent with an increased contribution from animal detritus, suggesting this resource became more nutritionally important for mosquitoes as plant inputs declined over the summer.

Introduction

Determinants of adult mosquito production from larval habitats include abiotic factors such as temperature and rainfall, and biotic factors such as predation, parasitism, and competition (Blaustein & Chase 2007; Juliano 2009). Although such broad ecological factors affect mosquito production from small, discrete container habitats, resource inputs are often the primary and fundamental limits to larval growth and subsequent adult emergence (Carpenter 1983; Hard et al. 1989; Lounibos et al. 1993; Kitching 2000, 2001; Kaufman et al. 2002; Kneitel 2007). This resource limitation often manifests itself through severe intra- and inter-specific competition that affects numbers of adults, their size, and their vectorial capacity, and ultimately impacts disease transmission dynamics (Hawley 1985; Alto et al. 2005; Bevins 2007). Barrera et al.(2006), for example, concluded that larvae of Aedes aegypti, the primary vector of dengue world-wide, commonly are food limited and compete for resources, leading to reduced body sizes of adult females. This reduced size may in turn affect dispersal and biting rates of adult females (Maciel-de-Freitas et al. 2007).

Aedes triseriatus is a common container breeding mosquito in eastern North America and the primary vector of La Crosse encephalitis virus. Larvae develop in water filled tree holes and tires that are normally heterotrophic microbial habitats, driven largely by particulate inputs and subsequent microbial processing (Walker et al. 1991). Although tree holes are consistent recipients of plant detritus in the form of senescent leaf material (Carpenter 1983, Lounibos et al. 1992; Leonard & Juliano 1995), other inputs include flower parts, twigs, and terrestrial invertebrate carcasses (Lounibos et al. 1992; Yee et al. 2007a, b). Recent studies have emphasized the potential importance of animal (invertebrate) detritus inputs as they relate to container-breeding mosquito nutrition and to outcomes of larval competition (Daugherty & Juliano 2000; Yee & Juliano 2006; Harshaw et al. 2007; Yee et al. 2007a, b; Murrell and Juliano 2008). Insect carcasses appear to be roughly tenfold higher in food value for mosquito larvae compared to senescent leaf material (Yee & Juliano 2006; Yee et al. 2007b), potentially allowing co-existence of competing larval species in tree holes and increased production of Ae. triseriatus from these habitats (Harshaw et al. 2007; Yee et al. 2007b). Tree hole dwelling larvae have even been shown to alter their foraging (browsing) behaviors in response to different types of detritus (Kesavaraju et al. 2007).

Plant material inputs into larval container habitats in the field are typically 10 – 100X those of animal detritus, but invertebrate carcass inputs can periodically exceed those of plant derived material (Daugherty & Juliano 2000; Yee et al. 2007b), and invertebrate material can be the primary nutrient inputs in larval mosquito habitats such as pitcher plants (Gray et al. 2006; Hoekman et al. 2009). Previous experimental investigations of animal versus plant detritus effects on tree hole dwelling mosquitoes have been done in laboratory microcosms and one such study indicated that a 1:10 ratio of insect carcass to senescent leaf material is near optimal for Ae. triseriatus development (Yee et al. 2007b). Path analysis indicated that the insect detritus was largely responsible for the production of mosquito biomass in that study (Yee et al. 2007b). However, the contribution of animal-versus plant-derived detritus to mosquito production from natural tree holes or other container systems has not yet been determined. This represents an important unanswered question in our understanding of how organic inputs into tree hole ecosystems are translated into mosquito biomass and the related vectorial capacity of Ae. triseriatus and similar mosquitoes that breed in a wide variety of detritus-dependent habitats.

Stable isotopes, usually ¹³C and ¹⁵N, are now commonly employed to examine food webs in terrestrial and aquatic systems, and to determine the trophic status of components (Post 2002; Grey 2006; Herbert et al. 2006; Hood-Nowotny & Knols 2007; Pasquad et al. 2007; Layman et al. 2007). δ¹³C analyses can identify dietary sources of primary consumers because consumer tissue is typically close to the δ¹³C values of the food source (Goodkoep et al. 2006; Fry 2006). In contrast δ¹³N values typically increase with trophic level, making them useful for establishing trophic structure. Bi-plots of the isotope values usually help accentuate differences between consumer groups in a mixed food web (Phillips & Koch 2002; Pasquand et al. 2007). Based on food source and consumer isotope values, and elemental concentrations in source and consumer biomass, mass balance-concentration dependent mixing models allow estimation of dietary contributions to consumers (e.g., Phillips & Koch 2002; Fry 2006).

Because animal tissues are typically enriched in both ¹³C and ¹⁵N relative to plant material (Fry 2006), we sought to make use of this distinction to address the question of the detrital dietary resources for Ae. triseriatus in tree holes at our study site in Michigan. Based on the high nutritional content, rapid turnover of, and stimulation of mosquito growth by insect carcasses, we hypothesized that animal detritus would form a substantial portion of the resource base for mosquito biomass in tree holes.

Methods

We used a combination of laboratory studies and field collections to provide material for isotopic analysis. In lab studies, mosquitoes (Ae. triseriatus) were reared with single sources of detrital material in microcosms (e.g, Kaufman & Walker 2006). The microcosms included 300 ml of distilled water, a microbial inoculum from natural tree holes (Kaufman et al. 2002), and the detrital source. The detrital sources were: senescent oak (Quercus alba) leaves, beech (Fagus grandifolia) flower parts, lab-reared fruit flies (Drosophila melanogaster - Diptera) adults, or earthworms (unidentified taxa). Dry mass per microcosm of detritus was: oak leaves, 1 g; beech flowers, 0.6 g; earthworms, 0.4 g; and Drosophila, 0.3 g. Oak leaves, beech flower parts and earthworms were collected from the litter layer at our tree hole study field sites near the MSU campus (E. Lansing, MI). Plant material was added after drying (48 hr, 45° C), and animal material was lyophilized prior to microcosm introduction. Forty neonate mosquito larvae were added to each microcosm and adults were collected as they emerged over a period of several weeks. Adult mosquitoes from previous studies (Kaufman & Walker, 2006; Yee & Juliano 2006) that had been stored with desiccant were also assayed for isotope content. In one of the previous studies (Yee and Juliano 2006), cricket tissue was used as detrital source, and we subsequently obtained lab-reared Grylloides sigillatus from the same source colony (and fed the same diet) used in that study courtesy of Dr. Scott Sakaluk (Illinois State Univ., Normal IL).

Detritus (plant and animal) and Ae. triseriatus pupae were collected from tree holes at our E. Lansing study site periodically in late spring and summer of 2005 and 2006. We collected detritus samples primarily from the surface or near surface of the tree hole water column with the assumption that these reflected recent inputs into the system. In the cases of invertebrate carcasses and flower parts, this assumption was realistic, however; leaf detritus was problematic in that senescent leaves could have entered the system during the previous fall or been blown in anytime hence. We sought to obtain representative detrital inputs at the study site, but did not quantify the relative abundance of detritus categories. All samples were frozen (−80° C), lyophilized, and stored with desiccant before grinding and analysis. Invertebrate samples or mosquitoes collected from individual tree holes were pooled as separate subsamples. Quantities collected were sometimes of insufficient mass to allow for losses in sample preparation prior to isotope analysis, such as in the cases of 2 or fewer mosquito pupae, hence not all samples collected could be analyzed.

All samples were ground to fine powder using stainless steel ball bearings in microcentrifuge tubes on a multiple-sample, high speed shaker (Retsch MM300, Glen Mills, Clifton, NJ). Large bulk samples of plant material (e.g. leaves used in microcosm experiments) were first subsampled by taking small sections from several different leaves or flowers, followed by grinding in a mortar and pestle before being processed with the bead beater. Subsamples of the pulverized material were then weighed into tin cups and stored with desiccant until isotopic analysis.

Because animals partition ¹³C differently in lipid pathways compared to other tissues (Post et al. 2007), we compared isotope values from mosquito pools that were subjected to lipid extraction with untreated mosquito pools from the same study. The source of the mosquitoes was a microcosm experiment in which we had manipulated levels of added nitrate (Kaufman & Walker 2006). Lipid was extracted using a dichloromethane/methanol biphasic extraction procedure that we’ve used previously for microbial lipid analysis (Kaufman et al. 1999). This method is a modification of the standard Bligh & Dyer (1959) lipid extraction and differs primarily in the replacement of choloroform with dichlomethane as the non-polar lipid extractant (see also Peterson & Klug 1994 and references therein).

Carbon (¹³C, ¹²C) and nitrogen (¹⁵N, ¹⁴N) isotope content of the samples were determined with an elemental analyzer (EA3000, Eurovector, Milan, Italy) coupled to a stable isotope ratio mass spectrometer (Elementar, Mt. Laurel, NI) following procedures detailed in Ostrom et al. (1997) and Ghandi et al. (2004). Total C/N ratios were also determined for each sample. Stable isotope values are expressed in parts per mil (‰) according to the following equation:

$$\delta \text{X}=([{\text{R}}_{\text{sample}}/{\text{R}}_{\text{standard}}]-1)\times 1,000$$

where X is ¹³C or¹⁵N, and R is the corresponding ratio ¹³C/¹²C or ¹⁵N/¹⁴N. R_standard was V-PDB or atmospheric N, for δ¹³C and δ¹⁵N, respectively. For δ¹³C and δ¹⁵N, laboratory standards were analyzed after every 10 unknown samples, with an accuracy and precision of ≤0.2 ‰ for both δ¹³C and δ¹⁵N.

To avoid statistical problems with distribution and transformation of proportionate (ratio) data, we compared isotope values or C/N ratios using the non-parametric Wilcoxon/Kruskal-Wallis rank sum methods. We used JMP^® Statistical Discovery Software, V5.1 (www.jmpin.com, SAS Institute, Inc., Cary NC, USA) for theses analyses and for descriptive statistic calculations.

To estimate dietary contributions of detrital sources to mosquito tissue, we used the mixture model of Phillips & Koch (2002), available at www.epa.gov/wed/pages/models.htm. This model takes into account the total carbon and nitrogen concentrations of the diets (detrital inputs) and consumer (mosquito), the isotope values of the dietary sources and consumer, and the trophic fractionation (shifts in isotope ratio between organism and diet) of each isotope by the consumer. Fractionation of food resource isotope values was estimated from the results of lab-reared mosquitoes on single detritus sources. For the mixing model, we used the average isotope values for end members adjusted for fractionation factors determined in the lab studies, and the mixture was the average isotope values of field-collected mosquitoes (Phillips & Koch 2002).

Results

Lab studies

Plant detritus (oak leaves and beech flowers) used as initial sources in microcosm studies had lower δ¹³C andδ¹⁵N values relative to the invertebrate detrital sources (Fig. 1). Isotope values of earthworm tissues were roughly intermediate between the plants and insects examined. In general, adult mosquito tissue was enriched, relative to the detritus sources in ¹³C when reared on plant detritus, and in ¹⁵N when reared on animal sources, but mosquitoes reared on cricket carcasses were enriched in both ¹³C and ¹⁵N (Fig. 1) and there was little evidence of any isotope fractionation in mosquito tissue when Drosophila was the detrital source. Mosquitoes reared on oak leaves with additions of inorganic nitrogen (KNO₃) to microcosms were enriched in ¹⁵N compared to mosquitoes reared on oak leaves without KNO₃ addition (Fig. 1, Table 1).

Figure 1.

Stable isotope composition of detrital source and mosquitoes from laboratory microcosm experiments. Values are means from 3 – 4 analytical reps of pooled material. Arrows connect diet source with mosquitoes reared on that source. No maple leaf detrital (diet) material was available for analysis.

Table 1.

Nitrate addition during larval growth and lipid extraction effects on isotopic content and carbon/nitrogen (C/N) ratios of Ae. triseriatus adult tissue. Values are mean ± one S.E., n = 3. Non-parametric comparisons are results of Wilcoxon/Kruskal-Wallis rank sums tests.

Nitrate	Lipid extraction	δ 15N	δ 13C	C/N
0	0	−2.2 + 0.2	−26.4 + 0.4	4.6 + 0.2
0	+	−2.1 + 0.1	−25.9 + 0.2	4.0 + 0.1
+	0	0.4 + 0.3	−25.4 + 0.3	4.7 + 0.1
+	+	0.6 + 0.2	−26.3 + 0.4	4.1 + 0.1
Non-parametric test		p-value
Nitrate		0.005*	0.298	0.575
Lipid Extraction		0.810	0.689	0.005*

^* = significant with sequential Bonferroni adjustment

Lipid extraction did not affect δ¹³C or δ¹⁵N values from mosquito tissue, however; extraction significantly lowered the C/N ratio (Table 1).

Field collections

Leaf detritus from tree holes consisted of maple, oak, beech and unidentified leaves, and flower components were from oak and beech trees. Animal detritus from tree holes consisted mainly of small Diptera, other insects (Coleoptera, Lepidoptera), unidentified arthropods, and earthworms. As was seen in the laboratory studies, the two main categories of detrital material (plants and invertebrates) from field collections were distinct, but differences were not as pronounced as seen in the lab studies (Fig. 2). Plant detritus collected directly from tree holes had isotope values very similar to those seen in the lab studies (Fig. 1), and animal detritus isotope values were higher than plants. Field-collected mosquitoes had isotope values (Fig. 2) that were intermediate between those that would be expected if either plant or animal material were the sole resource base. Stable isotope values in field mosquitoes were most similar to those seen in mosquitoes reared in the lab on leaf material with an external source of nitrogen, or with tree flowers, as the detritus base.

Figure 2.

Stable isotope composition of detrital material and mosquitoes collected from field sites. Values are mean ± one SE, n = 16 – 40. Mean values of lab-reared mosquitoes are illustrated for reference with small open circles: A – reared with oak leaves, B – reared with beech flowers, C – reared with oak leaves plus nitrate, D – reared with insect (crickets, Drosophila) carcasses, E – reared with earthworm carcasses.

Carbon and nitrogen isotope values in mosquitoes collected from natural tree holes tended to increase during the course of a season in 2006 (Fig. 3). In 2006, the monthly trend was significant for both C and N isotope values (rank sum test, p = 0.012 for δ¹³C and p = 0.002 for δ¹⁵N). Mosquitoes were collected only in June and July in 2005, and there was no difference between months for either isotope (rank sum test, p = 0.647 for δ¹³C and p = 0.160 for δ¹⁵N).

Figure 3.

Stable isotope composition of mosquitoes collected from individual tree holes in 2006. Values are single analytical replicates from pooled pupae collected from an individual tree hole.

Because leaf material, tree floral parts, and invertebrates were ubiquitous inputs into tree holes at our study sites, we used isotope values and C and N concentrations for these three resource types in the mixing model. We also used the overall mean of isotope values from all field-collected mosquitoes from both years for the mixture (i.e., consumer) parameters. Estimates of diet contributions to mosquito biomass (Table 2) indicated that 82% of pupal biomass could be attributed to beech flower detritus, but that 31% of the nitrogen in pupal tissue was derived from invertebrate detritus. Surprisingly, the model indicated that leaf material contributed relatively little to mosquito growth. We should point out that the source parameters in Table 2 did not adequately circumscribe (i.e., enclose the mixture values in a triangle formed by the 3 sources - Phillips & Koch 2002) all subsets of the field mosquitoes, and therefore these source estimates do not model the late season (July and August) 2006 mosquitoes. A single isotope (¹⁵N), dual source (plant material and invertebrates) version of the model indicated that invertebrate material would have contributed almost nothing to mosquito biomass in June 2006, but approximately 10% in July and 20% in August.

Table 2.

Concentration-dependent mixing model estimates of dietary contributions to field-collected mosquito biomass and components. Isotope values are based on means for each diet type and adjusted for fractionation by mosquitoes determined in lab studies. Concentrations of carbon and nitrogen are mean % C and %N for each diet.

	Mosquito tissue (%)			Detritus values
	Biomass	Carbon	Nitrogen	δ13C	δ15N	[C]	[N]
Leaf	8.6	8.8	3.4	−24.3	−2.8	46	1
Beech flower	81.8	82.5	65.6	−23.4	−2.1	45	2
Invertebrate	9.6	8.7	31.0	−24.1	4.7	40	8

Discussion

To our knowledge, this study is the first to use stable isotopes for examining the natural food resource base for mosquito larvae. Ae. triseriatus reared on single sources of detritus had tissue δ¹³C and δ¹⁵N values that reflected the type of detritus, allowing back-calculated estimation of detrital resource bases for mosquitoes emerging from natural tree holes. Field-collected mosquitoes apparently grew mainly on compounds derived from plant detritus with animal detritus supplements. This is consistent with current conceptions about tree hole ecosystems (Kitching 2001) and helps to validate microcosm studies that utilize senescent plant material as the resource base (Carpenter 1983).

The form of plant material resource base, however, might need re-evaluation. Tree floral parts, mainly beech tree flowers, appeared to drive tree hole mosquito production in this study. In 2006 in particular, some tree holes in our study area in May seemed filled to capacity with flower parts. This material did not persist through the summer and was clearly much less refractory than senescent leaf material entering the system. Lounibos et al. (1992) also found tree flower inputs into tree holes in Florida to be substantial, but strongly seasonal and ephemeral. Nitrogen content of beech flowers was double that of typical leaf material (Table 2), indicating a much higher quality microbial substrate. In preliminary growth studies (Kaufman & Pelz-Stelinski, unpub. data), we found that mosquito production from microcosms with beech tree flowers was comparable to that found when Drosophila carcasses are used as the detrital source. The latter source is a very high quality growth substrate for mosquito larvae (Yee & Juliano 2006). Lounibos et al. (1993) also showed that Ae. triseriatus reared on flowers from live oak trees developed significantly faster than larvae reared on similar amounts of live oak leaves, indicating that this form of plant detritus is nutritionally superior for tree hole mosquitoes. Pulse inputs of these higher quality but inconsistent resources may be critical for Ae. triseriatus emergence in the face of intra- and interspecific larval competition (Kaufman and Walker 2006; Yee et al. 2007b).

Field mosquito isotope values were also similar to those of mosquitoes lab-reared on leaf material with an external nitrogen addition. The potassium nitrate used in that experiment was enriched (δ¹⁵N = 3.65, Kaufman, unpub. data) compared to the oak leaf material (δ¹⁵N = −4.15) and was likely incorporated into microbial biomass harvested by larvae. Nitrogen entering the system via stemflow (Carpenter 1982; Kaufman et al. 1999; Verdonschot et al. 2008) is likely to have very different isotope values than those found in plant detritus, and it has been shown that microbial and larval transformations of N compounds in tree holes are also dynamic (Walker et al. 1991; Kaufman et al. 1999; Kaufman & Walker 2006; Verdonschot et al. 2008). Therefore, the external sources of nitrogen incorporated by the leaf microbial community and subsequently assimilated by larvae could greatly alter the δ¹⁵N values in field mosquito tissues compared to original leaf δ¹⁵N values.

Another source of nitrogen entering the system would be invertebrate carcasses. Our results indicate that invertebrate detritus may contribute proportionately more to nitrogen-containing compounds in pupal biomass than plant material, consistent with the observations that mosquito isotope values changed during the season and that invertebrate carcass influence on growth of Ae. triseriatus in tree holes is most pronounced when plant material resources are limiting (Harshaw et al. 2007). That carbon may come primarily from one diet source while nitrogen comes from another is not surprising ((Stenroth et al. 2006)), but the ecological consequences of this have not often been addressed. For Ae. triseriatus larvae in tree holes, this may mean that emergence is delayed while waiting for nutritional input provided by a particular detritus category. In the case of nitrogen limitations, this might even be in the form of conspecific larval mortality.

Isotopic evidence here (Fig. 1) would indicate direct incorporation of Drosophila detritus because of the lack of fractionation between diet and mosquito. However, decomposer microorganisms associated with detritus have isotope values that are usually indistinguishable from the substrate (see discussion below). Yee et al. (2007a) suggest that both direct incorporation of animal detritus and harvesting of associated decay microorganisms is important for larval growth. Interestingly, mosquitoes grown on earthworm and cricket carcasses did show isotopic fractionation, possibly because particle size and decay rates differed enough from Drosophila to prevent direct ingestion of tissue. Additionally, gut contents were not removed from any invertebrates tested in this study, and this would influence not only carcass decay rates and associated microbial communities, but also digestion and assimilation processes in larval mosquitoes and resultant isotope values (Fry 2006).

Although our results point to plant derived material as being the dominant resource base for mosquitoes at the study site, animal detritus sources might be expected to increase in importance in other habitats. Invertebrate carcass inputs into tires, a habitat commonly exploited by Ae. triseriatus, occur at rates than can support mosquito production independent of other inputs (Daugherty & Juliano 2000), and animal detritus supports the mosquito, Wyeomyia smithii, in the pitcher plant ecosystem (Gray et al. 2006; Hoekman et al. 2009). It would be expected that many larval habitats would vary greatly in placement and proximity to sources of plant detritus. Additionally, as our results indicate (Fig. 3), the relative importance of animal detritus inputs varies seasonally and year to year, as documented for detrital inputs into tire habitats (Kling et al. 2007; Yee et al. 2010). Tires located in forested areas showed decreasing inputs of plant material during a season, while animal inputs remained constant (Kling et al. 2007), indicating an increased relative importance of animal inputs over time.

Plant detritus samples in this study appear to be atypical from two perspectives. First, the values seen for leaf detritus are relatively low in ¹³C compared to what has been reported for most C3 plants and are more similar to what might be expected from some algal groups (Post 2002). We found δ¹³C levels of −28 to −37 for leaf material, which are notably lower than ranges of −29.5 to −26 in leaf litter from similar tree taxa reported by Balesdent et al. (1993). However, Collier et al. (2002) measured δ¹³C a range of −32 to −30.3 in riparian vegetation at a New Zealand site and the range of values found in C3 plants extends to −34 and lighter (O’Leary 1988). Location and growth conditions can further influence isotope composition of plants (O’Leary 1988; Fry 2006). Second, the ¹³C fractionation of the plant material by mosquito larvae was much higher (more ¹³C) than expected, even with the assumption that larvae are consuming microbially transformed material. The isotopic composition of microbial heterotrophs on decaying material is thought to mirror the plant material substrate (Balesdent et al. 1993, Fry 2006) and mosquito larvae harvest this microbial biomass directly (Kaufman et al. 2001). Given the accepted range of fractionation of ¹³C into the next higher trophic levels (0.5 – 1%), we would need to account for at least 3 trophic levels between tree flower parts and mosquito consumption (difference of + 3.6 δ¹³C between flowers and mosquito), and 9 trophic levels between leaf and mosquito larvae (difference of + 8.6 δ¹³C between leaves and mosquito). Although it is clear that the larvae of many mosquito species feed primarily on microorganisms and not directly on leaf material (Kaufman et al. 2001, 2002), and that they also feed upon intermediate microbial grazers such as protozoans and rotifers (Kaufman et al. 2002; Knietel 2007), it’s difficult to conceptualize a food web with that many links in the tree hole system or that mosquitoes harvest only the higher trophic levels. The apparent lack of ¹⁵N fractionation between leaves and flowers and the mosquito was also unexpected, but within the range of insects developing in plant-based systems (Spence & Rosenheim 2005).

It seems more likely that the difference in δ¹³C values between leaf and mosquito reflects an unrecognized fractionation by the microorganisms. It’s been shown, for example, that the relatively low δ¹³C values from methanotrophic bacteria are detectable in midge larvae and other aquatic organisms that consume benthic detritus (Doi et al., 2006; Deines et al. 2007), but this would not help explain ¹³C enrichment in the mosquito larvae food web. Nadon & Himmelman (2006) have noted higher than expected enrichment of ¹³C in primary consumers of marine benthic detritus (+ 4 δ¹³C). They suggested selective feeding by the macroinvertebrate consumers, but this was not verified. We examined eubacterial and fungal community structures associated with leaf detritus in tree holes and noted differences in relative abundances of microbial groups when larval feeding ceased (Kaufman et al. 2008), but have not yet targeted the Archaea – a group that would be highly active in carbon isotope fractionation (Fry 2006). Because mosquito larvae feed on many microbial groups (bacteria, fungi, protists) associated with detritus, future studies will need to determine how these components alter detritus δ¹³C values before they reach mosquito tissue. While previous studies indicate that δ¹³C values of consumers are marginally higher than their diet, such isotope shifts are not necessarily typical of arthropod consumers. For example, trophic fractionation of δ¹³C by insects ranges from −2.7 to 5.5 % (Ostrom et al., 1997; McCutchan et al. 2003; Scheu & Folger 2004) and amphipods feeding on live or decaying seagrass had δ¹³C values that differed from the source by 9–10 % (Crawley et al. 2007). While the reasons for this variation are uncertain, it is clear that additional estimates of trophic fractionation for arthropods are needed.

Our results indicate that removal of lipids did not affect δ¹³C values and adjustment of δ¹³C values for lipid content of aquatic invertebrates seems to be problematic in general (Kiljunen et al. 2006). Removal of lipids is more of a concern in larger animal samples, where fat tissue can be a considerable portion of biomass (Post et al. 2007). In addition to the impracticality of removing lipids from small samples, extraction of lipids in mosquitoes adds additional steps to sample processing while reducing the mass of the minimal material available for analysis. Because storage lipids in mosquitoes are usually less than 20% of dry weight (Timmermann & Briegel 1996), lipid extraction may be inadvisable forδ¹³C measurements in these insects (Post et al. 2007). Using C:N normalization models in lieu of lipid extraction for aquatic invertebrates is also tenuous (Kiljunen et al. 2006; Logan et al. 2008). However, additional studies should be conducted to determine if this step is generally unwarranted for mosquitoes.

This study illustrates the utility of stable isotopes in studies of larval mosquito feeding ecology, but also points out some of the limitations. Mixing models for determining diet sources have many caveats (Fry 2006) and the one employed here may not have adequately addressed the particulars of the tree hole system. Specifically, the isotope values of the three observed primary inputs into tree holes, (leaves, flower parts, and invertebrate carcasses) did not adequately circumscribed the targeted consumer (mosquito) values in all cases, with late season 2006 samples falling out of the mixing triangle primarily due to δ¹³C values. This may have been due, in part, to our estimates of fractionation of ¹³C from plant material (see above), but may also reflect differences in lab vs. field conditions. δ¹⁵N values proved more useful in this study in estimating dietary source contributions from a mixture, dovetailing with our findings that nitrogen dynamics are an important driving force for mosquito production in these habitats (Kaufman & Walker 2006). Additionally, since most of our detrital inputs representatives were collected in the late spring (when larvae had hatched and were starting development), we did not account for possible changes in the isotope values of leaves, flowers, or invertebrates over the course of an entire mosquito season. It’s unlikely that the plant material isotope δ values would change substantially, even if exposed to long periods of decay (e.g., Osono et al. 2008; Lau et al. 2009), but the invertebrate category in our study already showed high variability in spring collection isotope values and later season values are unknown. Finally, this study did not account for all potential inputs into the system as we only collected the larger particulate fractions. Fine particulate inputs into tree holes and similar habitats may influence larval abundance and species composition in container habitats (Kling et al. 2007; Yee et al. 2010), and may have different isotope values from their large particulate counterparts. Inclusion of fine particulates in mixing models might help to explain more of the variation in isotope values we saw in field-collected mosquitoes over the season.

Our observations that ephemeral resources added in pulse inputs could be driving Ae. triseriatus production in these habitats also point out the need to re-consider established perceptions. In the case of mosquitoes and tree holes, the fact that leaf material is most often observed as the most abundant detrital source could be misleading because the important sources fueling mosquito development have disappeared from view due to more rapid decay rates and incorporation into microbial and mosquito biomass.