Integrating omics to characterize eco-physiological adaptations: How moose diet and metabolism differ across biogeographic zones

Abstract

With accelerated land conversion and global heating at northern latitudes, it becomes crucial to understand, how life histories of animals in extreme environments adapt to these changes. Animals may either adapt by adjusting foraging behavior or through physiological responses, including adjusting their energy metabolism or both. Until now, it has been difficult to study such adaptations in free-ranging animals due to methodological constraints that prevent extensive spatiotemporal coverage of ecological and physiological data.Through a novel approach of combining DNA-metabarcoding and nuclear magnetic resonance (NMR)-based metabolomics, we aim to elucidate the links between diets and metabolism in Scandinavian moose Alces alces over three biogeographic zones using a unique dataset of 265 marked individuals.Based on 17 diet items, we identified four different classes of diet types that match browse species availability in respective ecoregions in northern Sweden. Individuals in the boreal zone consumed predominantly pine and had the least diverse diets, while individuals with highest diet diversity occurred in the coastal areas. Males exhibited lower average diet diversity than females.We identified several molecular markers indicating metabolic constraints linked to diet constraints in terms of food availability during winter. While animals consuming pine had higher lipid, phospocholine, and glycerophosphocholine concentrations in their serum than other diet types, birch- and willow/aspen-rich diets exhibit elevated concentrations of several amino acids. The individuals with highest diet diversity had increased levels of ketone bodies, indicating extensive periods of starvation for these individuals.Our results show how the adaptive capacity of moose at the eco-physiological level varies over a large eco-geographic scale and how it responds to land use pressures. In light of extensive ongoing climate and land use changes, these findings pave the way for future scenario building for animal adaptive capacity.

Keywords: DNA‐metabarcoding; biomarker; energy metabolism; metabolomics; nutritional ecology; starvation; ungulate.

Figure 1

Moose winter diet types per capture location of 264 moose captured across all major ecoregions (gray—montane, green—boreal, orange—coastal) in northern Sweden. Diet types are categorized based on individuals consuming more (“specialists”: Betula, Pinus, and Saliceae) or less (“generalist”) than 60% of a single diet item

Figure 2

Diet diversity of 264 moose winter diets categorized by diet types based on 17 MOTUs. Diet types are categorized based on individuals consuming more (“specialists”: Pinus—turquoise, Betula—yellow, and Saliceae—burgundy) or less (“generalist”—gray) than 60% of a single diet item

Figure 3

Serum metabolomics multivariate analysis of two‐way comparison of diet types. Cross‐validated score plots of OPLS‐DA models for each comparison. (a) Pinus versus Generalist, (b) Pinus versus Saliceae, (c) Pinus versus Betula, (d) Generalist versus Saliceae, (e) Generalist versus Betula, (f) Saliceae versus Betula. Ecoregions were marked by shape symbols: squares—montane, circles—boreal, triangles—coastal

Figure 4

Simplified representation of metabolic pathways showing significantly altered metabolites in four categories based on moose diet types: Pinus (turquoise bars), Generalist (gray bars), Saliceae (burgundy bars), Betula (yellow bars)

Figure A1

Maps representing the study area. A) The three ecoregions where moose were sampled (black dots). B) Map showing the human modification of the landscape (Source: Kennedy et al., 2019). The green areas are claimed as unmodified lands, representing boreal forests and the less productive and remote areas in high latitudes, inaccessible permanent rock and ice, or within tundra, and to lesser extent montane grasslands. The areas are claimed to have a low degree of human modification (0 < HMc ≤0.1), and largely reside ≥ 10 km away from more modified edges. Nevertheless, the yellow to red gradient demonstrates high modifications. In our study area, these represent mostly settlements and mines

Figure A2

Diversity of 264 moose winter diets based on the ecoregion (corresponding to boxplot shades) that individuals were captured in

Figure A3

Diet diversity of 264 moose winter diets based on 17 MOTUs and distinguished by sex (A; 186 females, 78 males). Neither pregnancy status (B) nor the number of calves (C) is driving the significant (p = .004) difference between sexes

Figure A4

Representative 600‐MHz 1H CPMG NMR spectrum of moose serum. The regions of δ 5.1–8.7 (above) were magnified compared with the corresponding regions δ 0.6–4.5 (below) for purposes of clarity. Ino: inosine, Hyp: hypoxanthine, His: histidine, 1‐ MH: 1‐methylhistidine, Phe: phenylalanine, Tyr: tyrosine, Fum: fumarate, Acetylcar: acetylcarnitine, L5: lipid –HC = CH, α‐Glc: α‐glucose, Lac: lactate, Cre: creatine, Cn: creatinine, α‐&‐ß‐Glc: α‐&‐ß‐glucose, PC: phosphocholine, GPC: glycerophosphocholine, Chol: choline, DS: dimethyl sulfone, L4: lipid C = CCH2C=C, Cit: citrate, Gln: glutamine, Glu: glutathione, NAG: N‐acetylglycoprotein, L3: lipid CH2C = C, Ace: acetate, Lys: lysine, Ala: alanine, L2: lipid (CH2)n, Val: valine, Leu: leucine, Ile: isoleucine, L1: lipid –CH3

Figure A5

Serum metabolomics multivariate analysis of two‐way comparison of sexes (gray circles = females; pink circles = males). PCA (above) and OPLS‐DA (below) score plots for comparison of sex differences. Hostelling's ellipse delineates the confidence at 95% of the distance of the scores to the mean values

Figure A6

Serum metabolomics multivariate analysis of two‐way comparison of diet types. PCA score plots for each comparison, the lined ellipse represents 95% confidence intervals. A. Pinus vs. Generalist, B. Pinus vs. Saliceae, C. Pinus vs. Betula, D. Generalist vs. Saliceae, E. Generalist vs. Betula, F. Saliceae vs. Betula

Figure A7

Plots obtained after performing a random permutation test with 200 permutations on OPLS‐DA models of 1H NMR data. R2 is the explained variance, and Q2 is the predictive ability of the model. Low value of Q2‐intercept depicts the high predictability of the model. A. Pinus vs. Generalist, B. Pinus vs. Saliceae, C. Pinus vs. Betula, D. Generalist vs. Saliceae, E. Generalist vs. Betula, F. Saliceae vs. Betula

Figure A8

Serum metabolomics multivariate analysis of two‐way comparison of diet types only in montane ecoregion. Cross‐validated score plots of OPLS‐DA models for each comparison. A. Pinus vs. Generalist; (1 + 1 components), R2X(cum) = 0.583, R2Y(cum) = 0.646, Q2(cum) = 0.389, B. Pinus vs. Saliceae; (1 + 1 components), R2X(cum) = 0.549, R2Y(cum) = 0.941, Q2(cum) = 0.527, C. Generalist vs. Saliceae; (1 + 1 components), R2X(cum) = 0.530, R2Y(cum) = 0.345, Q2(cum) = 0.0454, D. Generalist vs. Betula; (1 + 1 components), R2X(cum) = 0.472, R2Y(cum) = 0.208, Q2(cum) = 0.0833, F. Saliceae vs. Betula; (1 + 1 components), R2X(cum) = 0.234, R2Y(cum) = 0.698, Q2(cum) = 0.433