Battus philenor L. is found in warm climates throughout North America, mainly across the southern and eastern portions of the United States and in Mexico (Scott, 1986). This species is non-migratory (Hall, 2020) and females feed on a variety of flowering plants (e.g., thistles, phlox, ironweed) but only oviposit on pipevine species in the genus Aristolochia (Scott, 1986; Lotts and Naberhaus, 2020). In our study system, Battus larvae fed on the foliage of Aristolochia macrophylla Lam., a large liana with ropy stems that ascends into the forest canopy. Large vines can bear hundreds or even thousands of large heart-shaped leaves. Little is known about the nectar sources exploited by Battus in the Santeetlah Creek watershed, the amino acid concentrations of those nectars, or the many other protein sources exploited opportunistically. These unknowns generate potential uncertainty when interpreting results and are addressed in the discussion.

Specimens were collected throughout a narrow elevational band (872–1,191 m above sea level) from the watersheds of Big Santeetlah Creek (5350 ha) and Little Santeetlah Creek, Graham County, North Carolina, United States from 13 to 15 June 2013. These watersheds are heavily forested, dominated by C₃ plants, and situated on the eastern slope of the Unicoi Mountains, a subdivision of the southern Appalachians (Graves and Romanek, 2009). Caterpillars (n = 8) were collected directly from pipevine leaves, which were also collected, along with butterfly samples (n = 8); all insects were stored in cryovials in liquid nitrogen in the field. Specimens were subsequently freeze-dried and stored at room temperature in the lab until isotope analysis. Only a subset of the leaves collected were analyzed (n = 6), each from a different plant along a ∼4 km long transect parallel to Santeetlah Creek. For amino acid analysis, we sampled internal tissues, avoiding the gut. Water samples were also collected from the Santeetlah Creek Watershed and analyzed for bulk δ²H (average = –42‰; see Supporting Information—Text for details). Butterflies were sexed in the lab on the basis of dorsal hindwing iridescence, which is brighter in males (Supplementary Figure 1; Hall, 2020). All were males, except G-14. Variation in isotope data were unrelated to butterfly sex and elevation of sample collection (see Supporting Information—Text for details).

Approximately 10–15 mg of each dry sample was hydrolyzed for 20 h at 110°C in 1 mL of 6 N HCl, with a N₂ headspace to remove oxygen. Hydrolysates were evaporated to dryness under a gentle N₂ stream at room temperature then prepared as TFAA derivatives (Silfer et al., 1991). The total free amino acids in the hydrolysates were esterified with acetyl chloride/2-propanol (1:4, v/v) at 110°C for 1 h then acylated with trifluoroacetic anhydride/dichloromethane (1:1, v/v; DCM) at 110°C for 10 min. The resultant derivatives were completely dried then brought up in DCM and stored at –25°C until analysis. A standard mixture of 13 amino acids purchased from Sigma Aldrich (Saint Louis, MO) with known isotope values was derivatized alongside every batch of samples. See Supporting Information for details on bulk tissue analysis and findings.

All samples were analyzed at the EDGE Institute at the University of California Riverside. The δ¹³C and δ²H values of individual amino acids were determined by gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS) using a Trace 1310 gas chromatograph (GC) coupled with a GC-Isolink ll routed through a Conflo IV interface into a Delta V Plus isotope ratio mass spectrometer (Thermo Fisher Scientific, United States). Amino acids were either routed through a combustion reactor at 1,000°C for δ¹³C or an HTC reactor at 1,420°C for δ²H. Using a Triplus RSH autosampler, samples were injected splitless at 250°C onto a BPX-5 capillary column (60 m × 0.32 mm inner diameter, 1.0 um film thickness; SGE Analytical Science, United States). The GC oven temperature program was as follows: initial temperature 50°C for 2 min, followed by a ramp of 15°C/min to 125°C, a ramp of 3°C/min to 160°C, a ramp of 4°C/min to 190°C, a ramp of 6°C/min to 275°C, and a ramp of 15°C/min to 320°C. Carrier gas (He) flow was 2 mL min^–1.

Stable isotope values are expressed in per mil (‰) using delta notation: δX = [(R_sample/R_standard) − 1] × 10³ where X is ¹³C or ²H, R is the corresponding ratio of ¹³C:¹²C or ²H:¹H, and R_standard is Vienna Pee-Dee Belemnite or VSMOW for δ¹³C and δ²H, respectively. We analyzed the δ¹³C and δ²H values of 12 individual amino acids, 6 essential—valine (Val), threonine (Thr), leucine (Leu), isoleucine (Ile), phenylalanine (Phe), lysine (Lys)—and 6 non-essential—alanine (Ala), glycine (Gly), serine (Ser), proline (Pro), aspartic acid (Asp), glutamic acid (Glu)—listed in chromatographical order. Because acid hydrolysis converts glutamine (Gln) and asparagine (Asn) into Glu and Asp, respectively, by cleaving the terminal amine group, the measurements we report are combined Gln + Glu and Asn + Asp, hereby referred to as just Glu and Asp, to be consistent with previous studies. Tyr and Arg were excluded from analyses because of missing measurements caused by low concentrations near or below detection limits.

Amino acid standards of known isotopic composition were derivatized alongside and analyzed between unknown samples that were analyzed in triplicate. Reproducibility of replicate sample injections ranged from 0.1 to 0.6‰ (average < 0.3‰) for δ¹³C and 1.2–5.2‰ (average 2.5‰) for δ²H. Reproducibility of the amino acid standards used for sample corrections averaged < 0.6‰ for δ¹³C and 2.6‰ for δ²H. Because derivatization reagents add carbon atoms to the amino acids, the standards prepared alongside the samples are used to correct for the reagent carbon added. Similarly, exchangeable hydrogen atoms are removed from the carboxyl and amine groups of the amino acids during derivatization, and an additional hydrogen atom from the amine group is likely exchanged during hydrolysis. So, our measured δ²H data are corrected using amino acid-specific mass balance equations that adjust for the percentage of hydrogen atoms removed during derivatization (Fogel et al., 2016). Thus, the amino acid δ²H values we report reflect the nonexchangeable (intrinsic) hydrogen atoms only, and not hydrogen contributed by reagents during derivatization.

We used Shapiro-Wilks and Levene’s tests to confirm normality and homoscedasticity, respectively, and considered nonparametric equivalents before selecting ANOVA with Tukey HSD post-hoc tests to assess differences between tissue types in each individual amino acid. We further explored patterns in our amino acid δ²H and δ¹³C datasets with principal component analysis (PCA). AA_Ess and AA_NEss were evaluated separately. These statistical analyses were performed in R (version 4.0.3) with RStudio interface (version 1.3.1093; R Development Core Team, 2020) using the stats package.

The AA_Ess δ¹³C of the four tissue types were distinct from one another, a finding very different from O’Brien et al. (2002, 2003, 2005)’s laboratory studies (Figures 2A,B). PCA of AA_Ess δ¹³C clustered the samples according to their tissue type fairly well, with caterpillars the least variable and butterfly bodies the most variable (Figure 3A). Caterpillars separated from pipevines and butterfly wings along the first (PC1) and second (PC2) principal component axes. Differences between pipevines and caterpillars were driven by Thr, Leu, and Ile, while differences between caterpillars and adult butterflies (especially wings) were driven by Val, Phe, and Lys. Higher Thr and Leu also distinguished butterfly bodies from pipevines. In the ANOVA, all AA_Ess showed significant differences between some of the tissue types (Supplementary Table 5). Thr differed significantly between butterfly bodies and each of the other three tissues. Val differed for every pair except pipevines and butterfly bodies. Leu differed between pipevines and each of the other tissues. Ile was significantly different for half of the pairings (pipevines-caterpillars, pipevines-butterfly bodies, caterpillars-butterfly wings). Phe and Lys differed between butterfly wings and every other tissue type.

With PCA for AA_NEss δ¹³C, we found very tight clustering and excellent separation among pipevines, caterpillars, and butterfly wings. However, the two ellipses for butterfly tissues overlap (Figure 3B) because bodies show a high degree of variation, producing an ellipse that stretches across the entire PC1 axis. Consequently, half of the butterfly body samples plot near the wings, while the other half plot below pipevines and caterpillars. PC1, and the separation between butterfly wings on one end and caterpillars and pipevines on the other end, is strongly influenced by Ala, Pro, and Asp. PC2 is strongly influenced by Gly and Ser, and accounts for even more of the separation among pipevines, caterpillars, and butterfly wings. Interestingly, the half of the butterfly body samples that plotted most closely with pipevines and caterpillars along PC1 plot the farthest from them along PC2. ANOVA indicated that every AA_NEss except Glu showed significant differences among some of the tissue types (Supplementary Table 5). Ala and Gly differed between caterpillars and each of the butterfly tissues. Ser differed between butterfly bodies and the other tissues. Pro and Asp behaved similarly and only differed significantly between two pairings—-pipevines-caterpillars, caterpillars-butterfly bodies.

Similar to the δ¹³C data, AA_Ess δ²H were distinct from one another, supporting the above finding that essential amino acids were not simply routed from plants to caterpillars to adult butterflies (Figures 2C,D, 4). PCA of AA_Ess δ²H shows the tightest clustering of samples by their tissue types, with four non-overlapping ellipses (Figure 3C). The two butterfly tissues show the greatest variability and the greatest similarity to each other. PC1 provides excellent separation of pipevines and caterpillars from both butterfly tissues. PC2 similarly provides substantial separation between pipevines and caterpillars. Differences between pipevines and adult butterfly tissues were mainly due to Lys, Leu, and Phe; Thr was most responsible for separating adult butterfly tissues from the caterpillars and plant samples, and Val and Ile drove the separation between caterpillars and butterfly bodies. ANOVA revealed that Thr differed significantly between all but two pairs—pipevines-caterpillars and the two butterfly tissues (Supplementary Table 5). Leu was significantly different between all pipevine pairs and between caterpillars and butterfly wings. For Ile, caterpillars differed from both butterfly tissues. Phe was significantly different for all pairings involving butterfly tissues. Lys was significantly different for all pipevine and butterfly wing pairs.

Like AA_Ess δ²H, all PCA ellipses for AA_NEss δ²H are non-overlapping (Figure 3D). In contrast to δ¹³C, AA_NEss δ²H varies most among pipevines and least among butterfly wings. PC1 mainly separates butterfly wings from pipevines and caterpillars and is controlled by Pro, Gly, and Glu. PC2 is governed primarily by Asp and Ser, and to a lesser extent, Ala. These three amino acids create the separation of caterpillars from the other three tissue types. Asp and Ser are also responsible for the separation between the two butterfly tissues. The pipevine ellipse is oblong, such that it stretches across most of PC2. Within this ellipse space, the pipevine samples form two clusters—four samples group at the top of PC2 to the left of the butterfly body samples while the other two samples are centrally located between caterpillars and butterfly wings. This suggests high variability among pipevine Asp and Ser. In ANOVA, Ala was significantly different in half of the pairings—pipevines-butterfly bodies, caterpillars-butterfly bodies, and caterpillars-butterfly wings (Supplementary Table 5). Gly differed for every butterfly wing pairing. Ser was significantly different for all six pairs. Pro differed for every pairing except that of the two butterfly tissues. Like Ala, Asp differed significantly in half of the pairs—pipevines-caterpillars, caterpillars-butterfly bodies, and the two butterfly tissues. Glu only differed between caterpillars-butterfly wings, and the two butterfly tissues.

Because AA_Ess cannot be synthesized by B. philenor, they should be obtained directly from dietary amino acids. Direct routing from dietary protein is assumed to result in minimal isotopic fractionation such that isotope values of AA_Ess in caterpillars/butterflies and pipevines would be similar. In contrast, the use of AA_Ess synthesized by microbes from hydrogen or carbon in dietary biochemicals would yield a mismatch in the isotopic composition of AA_Ess between consumer and dietary protein. Adult B. philenor consume very little—if any—protein, which means their AA_Ess primarily derive from larval resources (Scott, 1986; O’Brien et al., 2005; Lotts and Naberhaus, 2020). Thus, we expected the AA_Ess δ¹³C values of B. philenor tissues to closely resemble each other and that of pipevines. However, our data show significant variation in AA_Ess δ¹³C values among pipevines, caterpillars, and butterfly tissues (Figure 2A and Supplementary Table 3), which must be driven by foraging behavior and/or gut microbial contributions. One such foraging behavior is compensatory feeding, wherein insects increase food consumption—and frass production—to compensate for a low-protein diet (Lavoie and Oberhauser, 2004; Le Gall and Behmer, 2014). This foraging strategy could potentially lessen the severity of protein-limitation in B. philenor caterpillars but the extent to which it might have a carryover effect on adult fitness is unknown.

Alternatively, some nectarivorous butterflies combat protein limitation by supplementing their adult diet with amino acids, which can increase their fecundity or even lifespan (O’Brien et al., 2005; Otis et al., 2006; Cahenzli and Erhardt, 2013). These supplements can potentially be acquired from nectar (Cahenzli and Erhardt, 2013; Levin et al., 2017), pollen (Erhardt and Baker, 1990; O’Brien et al., 2003), or mud-puddling, a behavior of aggregating at moist soil, tears, animal scat, or carcasses that is typically employed by males (Otis et al., 2006). However, amino acids are only present in trace amounts in nectar (Baker and Baker, 1973; Baker, 1975), B. philenor does not consume pollen (Konagaya et al., 2015), and mud-puddling is likely used to acquire sodium rather than protein (Otis et al., 2006). Moreover, even if amino acid supplements were contributing to the AA_Ess pool in butterfly body tissues, this would not explain differences in δ¹³C values among pipevines, caterpillars, and butterfly wings. In four out of six AA_Ess, it is either the caterpillar or the wing tissue that deviates most strongly from the δ¹³C fingerprint of pipevine leaves (Figure 2A). Thus, while B. philenor butterflies may opportunistically consume some amino acids from nectar, our data suggest that amino acid supplements do not substantially contribute to their AA_Ess budgets.

A more likely explanation for observed differences in the δ¹³C of AA_Ess between plant and consumer is subsidization by the gut microbiome. Although it remains unknown whether or not most Lepidopteran species host a gut microbiome, our data suggest that it contributes to the AA_Ess pool used by both caterpillars and butterflies for tissue synthesis. How the microbial signature deviates from that of dietary AA_Ess depends on whether gut microbes use carbohydrates or lipids—which are ¹³C-enriched or depleted, respectively, relative to dietary proteins—as a carbon source to synthesize AA_Ess. For example, gut microbes appear to synthesize Val, Leu, and Ile in caterpillars and Thr in butterflies from carbohydrates as the δ¹³C of these particular amino acids is substantially more positive (∼10‰) than those in leaf tissue (Figure 2A). Conversely, Val, Phe, and Lys in wing tissues seem to reflect larval gut microbial contributions from lipid stores during metamorphosis because these amino acids have more negative δ¹³C (∼7‰) than those in leaf tissue. We also found evidence of direct routing of AA_Ess between pipevines and caterpillars and during metamorphosis. δ¹³C values of Leu are identical among the insect tissues and only ∼5‰ lower in pipevines (Supplementary Table 3).

AA_NEss δ¹³C values are known to reflect a combination of direct routing from dietary protein and de novo synthesis by consumers from dietary carbohydrates and lipids (Jim et al., 2006; Newsome et al., 2014). Because pipevine leaves and nectar have less protein content than animal tissue, we expected that caterpillars and butterflies would use other dietary macromolecules (e.g., nectar carbohydrates) and limited endogenous nitrogen stores to construct some of the AA_NEss they need for metabolism and reproduction. In support of this prediction, we did not find a common AA_NEss δ¹³C fingerprint among tissue types, indicating extensive metabolic remodeling of amino acids during each insect life history stage (Figure 2B and Supplementary Figure 4B). Glu was the only AA_NEss that had a similar δ¹³C composition among tissue types (Supplementary Table 3); we hypothesize that this pattern results from Glu being the most abundant amino acid in animals, so its isotopic composition is relatively stable in comparison to amino acids that have smaller pool sizes. In contrast, Pro and Asp in wings are ¹³C-depleted relative to caterpillars, suggesting that pupae, or their gut microbes, draw from lipid stores to synthesize these AA_NEss for wing synthesis. Lipids are the most abundant energy reserve available during metamorphosis (Arrese and Soulages, 2010), and lipid carbon is easily introduced as acetyl CoA into the TCA cycle, from which Pro and Asp are derived (Horton et al., 2002).

Because adult B. philenor consume carbohydrates almost exclusively, it follows that amino acids synthesized from intermediaries in glycolysis (Ala, Gly, Ser) should reflect the strongest carbohydrate δ¹³C signature of all the tissue types. In support of this prediction, Ser δ¹³C is on average ∼20‰ higher in butterfly bodies than in the other three tissues. Of all the AA_NEss measured, Ser has the most variable δ¹³C values (Figure 2B and Supplementary Table 3), which range > 50‰ among pipevines and butterflies and almost 40‰ in butterfly bodies alone (Supplementary Figure 3B). This is a substantial amount of δ¹³C variation for a single plant-herbivore system and indicates extensive de novo synthesis, particularly by the nectarivorous adult butterflies. Ala δ¹³C values are significantly higher by ∼9‰ in caterpillars relative to pipevine leaves (Supplementary Table 5), also indicative of synthesis from carbohydrates. Interestingly, caterpillars have uniquely low Gly δ¹³C values, suggesting either a mass balance effect associated with metamorphosis or synthesis of this AA_NEss with lipid carbon. Overall, isotope data for glycolytic amino acids demonstrate that the magnitude of ¹³C-enrichment imparted by carbohydrate carbon is not universal. Rather, it likely depends on the interaction between the pool sizes and fluxes of each amino acid dictated by supply and demand of these metabolites, as well as the individual fractionation factors of the enzymes used to synthesize them.

AA_Ess δ²H data further show that the hydrogen isotope composition of individual amino acids is systematically modified according to the biochemical pathways used for synthesis and that eukaryotes and many prokaryotes utilize the same basic metabolic machinery to do so (Rodriguez Curras et al., 2018; Newsome et al., 2020). As was the case with E. coli (Fogel et al., 2016) and house mice (Newsome et al., 2020) reared in controlled feeding experiments, Ile has the lowest δ²H values of all amino acids in our dataset. Ile also has the most uniform δ²H values, suggesting it was mainly routed directly from the larval diet into the insect tissues. Furthermore, the extreme variation we found in AA_Ess δ²H within individuals and among tissue types (Figure 2C and Supplementary Table 4) mirrors that seen previously in E. coli and mice. AA_Ess δ²H varied over 300‰ in a single caterpillar (Figure 4C) and variation was similar in the other tissues measured, even within a single pipevine leaf (∼230‰).

In every AA_Ess other than Ile, δ²H was highest in either butterfly wings (Phe, Lys) or both wings and butterfly bodies (Thr, Val, Leu) relative to pipevines and caterpillars. This ²H-enrichment in adult insect tissues suggests that gut microbes, rather than direct incorporation from nectar, are contributing AA_Ess to their hosts using some combination of hydrogen sourced from the metabolic breakdown of carbohydrates, and possibly from drinking water. Environmental water typically has higher δ²H values than all dietary hydrogen sources because of the large negative discrimination during the synthesis of organic compounds in photosynthesis (Estep and Hoering, 1981). The fact that AA_Ess in wings are similar to or more enriched than the butterfly bodies points to body water as the predominant hydrogen source over carbohydrates because although stored glycogen can be accessed for glycolytic fuel during metamorphosis, without frequent feeding these carbohydrates are far less abundant than fat reserves and are quickly depleted (Arrese and Soulages, 2010). Although the adult gut microbiome may also play a role in butterfly body δ²H, carryover effects from the larval stage appear to persist throughout the butterfly’s lifetime and are explored further below.

In agreement with the analogous δ¹³C data, AA_NEss δ²H data suggest considerable de novo synthesis in B. philenor, particularly in Ser that ranges > 300‰ (–187‰–137‰) among tissue types (Figure 2D and Supplementary Table 4). Variation in Ser δ²H is not surprising given the essential function of serine proteases in insect metabolism—such as aiding digestion and development (Cao and Jiang, 2018)—which involves serine donating a hydrogen then gaining one from water (Radisky et al., 2006). Generally, structurally simple amino acids like Ser are more likely to exchange hydrogen with body water because a larger proportion of their hydrogen is bound to oxygen and nitrogen relative to more complex amino acids (Newsome et al., 2020). In controlled feeding studies on E. coli (Fogel et al., 2016) and mice (Newsome et al., 2020), drinking water appeared to have a particularly strong effect on the δ²H of glycolytic amino acids—this was especially true when organisms were fed diets containing no or low protein contents. Our data indicate that the same is true of our protein-limited herbivores.

The influence of body water on tissue δ²H is likely amplified in a holometabolous insect like B. philenor. Inside its chrysalis, which the caterpillar must construct from protein, the pupa undergoes a complete morphological remodel. While wings are an iconic part of this transformation, the pupa must also shorten the gut, grow limbs, expand the tracheal system, reshape the mouthparts, and develop a hard exoskeleton (Lowe et al., 2013). This complex process requires energy, expends water, and generates waste, and it is unclear how differential routing of amino acids might affect fractionation of carbon and hydrogen isotopes among newly synthesized tissues. What is clear, though, is that there are myriad opportunities during metamorphosis for isotopic discrimination to occur.

Evaluation of over 50 Lepidopteran species revealed that pupae typically lose a large percentage (≤80%) of their weight during metamorphosis (Molleman et al., 2011). Most of the weight lost is water (Wappner and Quesada-Allué, 1996; Molleman et al., 2011). This substantial water loss likely results in ²H-enrichment of the remaining body water pool, which is then used to synthesize ²H-enriched amino acids for adult butterfly tissues. δ²H patterns in Ala, Pro, and many microbially synthesized AA_Ess follow this prediction (Figures 2C,D). If wings are synthesized later in metamorphosis than adult body tissues, they may acquire higher δ²H values as a consequence of this gradual process, particularly if they are strongly influenced by hydrogen in body water. For instance, δ²H values of Gly, Ser, and Asp in butterfly wings are 40–50‰ more positive than in butterfly bodies (Supplementary Figure 4A and Supplementary Table 4). Furthermore, prior studies have identified bulk tissue ¹⁵N- and ²H-enrichments following metamorphosis in multiple species, including Lepidoptera, and primarily ascribed them to the excretion of ¹⁵N-depleted waste products and the inclusion of ²H-enriched body water for tissue synthesis (Doi et al., 2007; Tibbets et al., 2008; Peters et al., 2012; see Supporting Information—Text for additional discussion of Pro).

We focus this discussion on δ²H, though we lean on δ¹³C for context, because understanding inter-individual δ²H variation among primary producers and herbivores is critical for isoscape development and further elucidates the influence of physiology on tissue δ²H values. Also see Supporting Information—Text for a brief discussion of mycoheterotrophy as a potential explanation for inter-individual variation in plant isotope values. We found that bulk tissue δ²H values of caterpillars are often less variable than those of pipevines (Figures 4C,D). Relative to all other amino acids, Asp has the most variable δ²H values in both pipevine and caterpillar samples; Asp δ¹³C values are also variable in caterpillars (Supplementary Figure 3C). In plants, aspartate is a precursor for asparagine, which is used to transport and store nitrogen, and a precursor for Thr, Ile, Lys, and methionine synthesis (Azevedo et al., 2006). In caterpillars, Asp is abundant in a storage protein (Telfer and Kunkel, 1991) that is catabolized in adult Lepidoptera and used as a critical substrate for reproduction (O’Brien et al., 2002). These key metabolic functions may explain the unusually high inter-individual variability observed in Asp. Despite these exceptions, the similarity in AA_Ess δ¹³C and δ²H fingerprints among individual caterpillars implies relatively consistent fractionation during the plant-herbivore trophic transfer (Figure 4C and Supplementary Figure 3C). This uniformity strongly contrasts with the high degree of inter-individual variation among adult butterflies (Figures 4A,B and Supplementary Figures 3A,B). It is not surprising that caterpillars would be less variable than butterfly body tissue, given the generalist diet of the adults and the obligate host plant relationship maintained by the larvae. However, the fact that caterpillars also tend to be less variable than wings—which are largely metabolically inert—emphasizes the dominant role played by physiology during metamorphosis in determining δ²H values of adult tissues.

Concurrently assessing both δ¹³C and δ²H fingerprints for individual butterfly bodies reveals an additional interesting feature of our dataset that is best illustrated by the highly variable Ser and Thr: the relative rankings of individuals’ ¹³C-enrichment are fairly consistent independent of amino acid (Supplementary Figure 3B). That is, the three butterflies with the highest Ser δ¹³C also have the highest Thr δ¹³C. The same is not true for δ²H (Figure 4B). This is likely a reflection of the increased complexity of hydrogen cycling relative to that of carbon, but perhaps it also signifies a promising opportunity. δ²H analysis of specific subsets of amino acids may be best suited to answering different kinds of ecological or physiological questions. For example, glycolytic amino acids have previously been proposed as the most faithful proxies for δ²H of drinking water (Newsome et al., 2020) and so may be most beneficial for assigning geographic origin to migratory animals, particularly when bulk tissue δ²H is insufficient. Because caterpillars are an important food source for many higher trophic level organisms, their AA_Ess δ²H (and δ¹³C) values could be useful dietary tracers. Moreover, AA_Ess δ²H and δ¹³C provide a way to assess the presence and role of more complex metabolic processes like gut microbial input in systems where the isotopic composition of diet is well characterized. Several microbially synthesized AA_Ess as well as Ala, Pro, and Asp may be particularly valuable for studying physiology and biochemistry during metamorphosis, including the pupa’s use of a ²H-enriched body water pool to synthesize AA_NEss or AA_Ess (via its microbiome), amino acid routing, and the order in which adult tissues are synthesized.