Salivary proteome of a Neotropical primate: potential roles in host defense and oral food perception

Saliva samples

All research protocols reported here were reviewed and approved by the government of Mexico (SEMARNAT SGPA/DGVS/10426/14) and complied with the legal and ethical guidelines of the IUCN (1998), and of the Mexican authorities (Diario Oficial de la Federación, 1999). We used the saliva samples obtained by FCEG as part of a complementary research project to evaluate the relationship of dietary tannins and tannin-binding salivary proteins (FC Espinosa-Gómez, 2017, unpublished data).

Samples were obtained from 14 free-ranging black howler monkeys occupying four forest fragments near Balancán, Mexico (17°44′05″N; 91°30′17″W). This disturbed forest landscape lies within cattle pastures (Pozo-Montuy et al., 2013). Monkeys were darted and anaesthetized by a veterinarian with ketamine hydrochloride (8 mg/kg estimated body mass, Ketaset, Fort Dodge Animal Health, Iowa USA). Once monkeys were stabilized following sedation, the body weight was determined and the saliva flow was stimulated by an intra-muscular administration of the parasympathomimetic compound pilocarpine-hydrochloride (0.5 mg/ body mass) (Espinosa-Gómez et al., 2018; Da Costa et al., 2008). The whole saliva was collected from the mouth of each monkey using a micropipette, placed in a tube, and immediately frozen in liquid nitrogen. All saliva samples were transported from the field to the Proteomic Lab at INECOL, AC in Xalapa, Veracruz, México in a cryogenic container and then stored in an ultra freezer at −80 °C until analysis.

Saliva preparation and SDS-PAGE

At the lab, saliva aliquots were thawed, cells and debris were removed by centrifugation at 16,000 g for 10 min at 4 °C, and the supernatant was captured. We determined the salivary total protein concentration by the Bradford method (Bradford, 1976) using bovine serum albumin (BSA) as a standard. Absorbance was measured at 595 nm with a microtiter plate reader (SpectroMAX 340, Molecular Devices, Union City, CA, USA). We fractionated salivary proteins using 12% one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) following Laemmli (Laemmli, 1970). The 1D-SDS PAGE (8 × 7.3 cm × 1.5 mm) was run with 30 µg of salivary total protein with SDS loading buffer 4:1 (Biorad, CA, USA). Molecular mass markers (Precision Plus Protein Dual Color Standards, BioRad 1610374, CA, USA) were run in each gel to calibrate the molecular masses of the salivary proteins. Protein bands were fixed with a mixture of 26% ethanol, 14% formaldehyde, and 60% water for 3 hr, followed by 3 hr in a mixture of 50% methanol and 12% acetic acid (Steck, Leuthard, & Bürk, 1980). We followed the procedures suggested by Beeley et al. (1991) to detect PRPs, which allows PRPs stain pink or pinkviolet. Briefly, gels were stained overnight with a 0.25% Coomassie brilliant blue R-250 solution (Biorad 1610400) in 40% (v/v) methanol and 10% (v/v) acetic acid. We de-stained the protein bands with several changes of 10% acetic acid.

In-gel digestion proteins

The clear proteins bands observed in our protein gels provided sufficient clean samples for proteomic analysis using the Nano LC-MS/MS approach. Protein bands were manually removed from gels and cut into 13 different molecular weight ranges (bands a - m) by excising these regions with a sharp straight edge and then destained with 2.5 mM ammonium bicarbonate (NH4HCO3) in 50% acetonitrile (ACN), and then dehydrated with 100 µL of 100% ACN. Samples were then reduced with 20 µl of 10 mM DTT in 50 mM NH4HCO3 and incubated for 45 min at 56 °C. Subsequently, the samples were cooled to room temperature and proceeded with the alkylation by adding 20 µL of 100 mM iodoacetamide in 50 mM NH4HCO3, and incubating in the dark for 30 min. Then, the samples were washed with 100 µL of 100% ACN for 5 min, then with ¬100 µL of 5 mM NH4HCO3 for 5 min and then with 100 µL of 100% ACN for 5 min. Finally, samples were dried with CentriVap (Labconco Kansas, Missouri) for 5 min and rehydrated with 10 µL of digestion solution containing 12.5 ng/ µL mass spectrometry grade Trypsin Gold (Promega, Madison, WI, USA) in 5 mM NH4HCO3 and incubated in a water bath at 37 °C overnight. The reaction was stopped at −80 °C. The peptides were extracted with 30 µL of 50% acetonitrile with 5% formic acid by centrifugation at 1000× g for 30s and desalted with ZipTip- µC18 tips (Merck Millipore, Darmstadt, Germany) and dried using a CentriVap (Labconco Kansas, Missouri, USA).

Mass spectrometry (Nano LC-MS/MS analysis)

Suspended samples (5 µl of 0.1% formin acid) were injected into a nanoviper C18 trap column (3 µm, 75 µm X two cm, Dionex) at 3 µl min-1 flow rate and separated on an EASY spray C18 RSLC column (2 µm, 75 µm × 25 cm) with a flow rate of 300 nl min-1 connect to an UltiMate 3000 RSLC system (Dionex, Sunnyvale, CA) and interfaced with an Orbitrap FusoinTM TribidTM (Thermo-Fisher Scientific, San Jose, CA) mass spectrometer equipped with an “EASY Spray” nano ion source (Thermo-Fisher Scientific, San Jose, CA). For peptide separation, a chromatographic gradient using MS grade water (solvent A) and 0.1% formic acid in 90% acetonitrile (solvent B) for 30 min was set as followed: 10 min solvent A, 7–20% solvent B within 25 min, 20% solvent B for 15 min, 20–25% solvent B for 15 min, 25–95% solvent B for 20 min, and 8 min solvent A. The mass spectrometer was operated in positive ion mode with nanospray voltage set at 3.5 kV and source temperature at 280 °C. External calibrant included caffeine, Met-Arg-Phe-Ala (MRFA), and Ultramark 1621. The mass spectrometer was operated in a data-dependent mode to automatically switch between MS and MS/MS. The survey full-scan MS spectra were acquired in the Orbitrap analyzer, scanning of mass range was set to 350–1,500 m/z at resolution of 120,000 (FWHM) using an automatic gain control (AGC) setting to 4.0e5 ions, maximum injection time to 50 ms, dynamic exclusion 1 at 90S and 10 ppm mass tolerance. A top speed survey scan for 3s were selected for subsequent decision tree-based Orbitrap collision-induced dissociation (CID) or higher-energy collisional dissociation (HCD) fragmentation (Swaney, McAlister & Coon, 2008; Frese et al., 2011). The signal threshold for triggering an MS/MS event was set to 1.0e4 and the normalized collision energy was set to 35 and 30% for CID and HCD, respectively. The AGC of 3.0e4 and isolation window of 1.6 m/z was set for both fragmentations. Additional parameter for CID included activation Q was set to 0.25 ms and injection time to 50 ms. For HCD, first mass was set to 120 m/z and injection time to 100 ms. The settings for decision tree were as follows: for HCD fragmentation charge states 2 or 3 were scan in a range of 650–1,200 m/z, charge states 4 were scan in a range of 900–1,200 m/z, and charge states 5 were scan in a range of 950–1,200 m/z; for CID fragmentation charge states 3 were scan in a range of 650–1,200 m/z, charge state 4 were scan in a range of 300–900 m/z, and charge state 5 in scan range of 300–950 m/z. All data were acquired with Xcalibur 4.0.27.10 software (Thermo-Fisher Scientific).

Database search and protein/peptide identification

Raw data were analyzed with Proteome Discoverer 2.1 (PD, Thermo Fisher Scientific Inc.) and subsequent searches were carried out using Mascot server (version 2.4.1, Matrix Science, Boston, MA) and SQUEST HT (Eng, McCormack & Yates, 1994). The search with both engines was conducted against Homo sapiens, Macaca fascicularis, Macaca mulatta, and the complete UniProt reference proteome (http://www.uniprot.org/). We included as parameters in the search: full-tryptic protease specificity, two missed cleavage allowed, static modifications covered carbamidomethylation of cysteine (+57.021 Da). Furthermore, dynamic modifications included methionine oxidation (+15.995 Da) and deamidation in asparagine/glutamine (+0.984 Da). For the MS2 method, in which identification was performed at high resolution in the Orbitrap, precursor and fragment ion tolerances of ±10 ppm and ± 0.2Da were applied. Resulting peptide hits were filtered for maximum 1% FDR using the Target Decoy PSM validator. We considered a MASCOT score >20 for proteins identified with two or more peptides and MASCOT score >34 for proteins identified with one single peptide.

Bioinformatic analysis

Proteins were screened for the predicted presence of N-terminal endoplasmic reticulum (ER) targeting signal peptide (SP) using the Signal P 4.1 program (http://www.cbs.dtu.dk/services/SignalP/, Petersen et al., 2011). In addition, we used the server Secretome P 2.0 to determine non-classical and leaderless protein secretion in proteins identified in the saliva of monkeys (http://www.cbs.dtu.dk/services/SecretomeP/, Bendtsen et al., 2004). The program MHMM server v. 2.0 were used for the prediction of transmembrane helices in salivary proteins (http://www.cbs.dtu.dk/services/TMHMM/). Proteins were classified base on GO ontology enrichment of biological processes using David ontology tool (Sherman & Lempicki, 2009) (https://david.ncifcrf.gov/). We used REVIGO web server (http://revigo.irb.hr/) with a median similarity for the visual representation of the clustering of biological processes.

Search for proteins/peptides related with host-defense and taste sensitivity

To distinguish the salivary proteins related with taste sensitivity (beside with astringent detection in mouth), host defense, and antimicrobial properties (anti-bacterial, antiviral and anti-fungal), we carried out detailed scrutiny of the UniProt functional annotation (http://www.uniprot.org/) and also reviewed papers on salivary proteomics/peptidomics from humans and other animals that have identified proteins with specific functions on immunity and taste sensitivity of food. Most of the salivary proteins related with a specific function in an animal specie, has been identified in several others, which suggest that their function is conserved across species.

Salivary protein separation by SDS-PAGE

We observed similar salivary protein patterns on 1-D electrophoresis gels in all individuals. There were multiple bands (a–m) ranging from 10 to 250 kDa (Fig. 1), with the most intense protein bands being located at low molecular weight from 10–15 kDa (k, l, m). However, the intensity of the bands did vary, with the j band being more apparent in individuals P-M1 and P-F1, the band k was more intense in B-F2, and bands l and m displayed a darker and more significant area of staining in B-F2 and P-M1. We visualized a main protein band (j) with an apparent molecular mass between 22–30 kDa that displayed a pink staining, which might be PRP according to Beeley et al. (1991) and described in Espinosa-Gómez et al. (2018).

Identification of salivary proteins by Nano LC-MS/MS

We use proteomics to evaluate all 13-protein bands fractionated on SDS-PAGE-1D; we digested and subjected to LC MS-MS pools of the same protein band from all individuals, and 156 proteins were identified (Table S1). Among these, 55 were predicted with both signal peptide (SP) and transmembrane helices domains (TMHMM), including well-known secreted proteins, such as the Lactoperoxidase (P22079), Lactotransferrin (P02788), Serotransferrin (A5A6I6), the glycosylated Prosaposin (P07602), and the Histidine-rich glycoprotein (P04196). Besides, we were able to predict five non-secreted proteins with TMHMM (Fig. 2A). Using Secretome P 2.0 (http://www.cbs.dtu.dk/services/SecretomeP/, Bendtsen et al., 2004) we predicted ten proteins with a non-classical and leaderless secretion that include for example, Galectin-7 (P47929), Putative ubiquitin-conjugating enzyme E2 N-like (Q5JXB2), Putative ubiquitin-conjugating enzyme E2 N-like (Q5JXB2).

After gene ontology enrichment by David Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/, Huang, Sherman & Lempicki, 2009) and clustering by REVIGO web server (http://revigo.irb.hr/, Supek et al., 2011), we obtained a tree map displaying key biological processes associated with howler monkey saliva, including negative regulation of endopeptidase activity, defense response to fungus, gluconeogenesis, protein folding, cytoskeleton organization, platelet degranulation, and epidermis development. Each of these major groups included several gene ontology (GO) groups (Fig. 2B). The most representative group corresponded to negative regulation of endopeptidase activity that clustered gene ontology, such as proteolysis (GO:0006508), protein stabilization (GO:0050821), retina homeostasis (GO:0001895), retinoic acid metabolism (GO:0042573), negative regulation of endopeptidase activity (GO:0010951), and negative regulation of endothelial cell chemotaxis (GO:2001027).

The second most prominent cluster was the defense response to fungus conglomerating GO like protein kinase A signaling (GO:0010737), complement activation classical pathway GO:0006958 (GO:0006958), defense response to fungus (GO:0050832), response to ethanol (GO:0045471), and zinc ion (GO:0010043). The third most representative cluster named gluconeogenesis gathered the GO oxidation–reduction process (GO:0055114), cellular aldehyde metabolism (GO:0006081), and gluconeogenesis (GO:0006094).

Howler monkey salivary proteins associated with host-defense in mammals

It is widely recognized that salivary proteins have many functional properties, and some have more than one function. According to data available on UniProt functional annotation (http://www.uniprot.org/) and review papers on salivary proteomics/peptidomics from humans and other mammals, we identified 10 proteins with dual function, including oral food perception and host-defense (6.4% of total identified proteins). We also identified proteins related with taste sensitivity or innate/acquired immunity (Fig. 3). We identified 28 salivary proteins/peptides (17.9% of total identified proteins) associated with functions, such as host defense, innate immunity, and antimicrobial properties (anti-bacterial, antiviral and anti-fungal). There were identified cationic peptides, and defense proteins (such as immunoglobulins) that have been reported as effective against parasites, fungi and cancer cells. Table 1 presents the complete list of proteins/peptides identified in saliva of howler monkeys related with host-defense and anti-microbial properties, and the references where the link between these proteins and that immune function has been reported.

Howler monkey salivary proteins associated with oral food perception

We detected 16 proteins in saliva of howler monkeys (10.25% of total identified proteins) related with oral food perception; the complete list is shown in Table 2. There were identified six proteins associated with gustatory sensitivity of sweet, salty, umami, fatty-acids, and pungent flavors. For instance, carbonic anhydrase VI or “gustin” was identified and plays an important role in human taste perception of bitterness or fatty acids (Morzel et al., 2014). Likewise we identified four types of cystatins, histidine-rich glycoprotein, and IgA, which are associated with a major inhibition of the feeling of astringency and bitter taste (Nayak & Carpenter, 2008; Canon & Neyraud, 2017; Shimada, 2006).

Salivary proteins linked with host-defense in mammals

A major finding of our research with howler monkeys is the identification salivary proteins and cationic molecules belonging to the two major antimicrobial peptides families: cathelicidins and defensins that rapidly inactivate infectious agents (Wiesner & Vilcinskas, 2010; Zanetti, 2005). Cathelicidins have been identified in cattle, sheep, rat, and dogs, but not in humans (De Sousa-Pereira et al., 2015). We also identified the antimicrobial peptide dermcidin that is recognized as a first line of skin defense in primates and has been identified in eccrine sweat glands of humans. Some argue that dermcidin is not found in other body fluids, such as nasal secretions, tears, saliva, semen, milk, and urine (Schittek, 2012); however, we identified this peptide in saliva of howler monkeys and it has been found in tears and cervicovaginal fluid in humans (Shaw, Smith & Diamandis, 2007). Dermcidin-homologous genes exist only in apes (Pan troglodytes, Gorilla gorilla, Pongo abelii) and Old and New World monkeys (Schittek, 2012).

Our proteomic analysis identified four members of the cystatin family (A, B, C, and D) in saliva of howlers that may inhibit the action of endogenous, bacterial, and parasitic protozoan proteases (Fábián et al., 2012). Similarly, the GO analysis of the salivary proteins indicates the most representative group corresponded to negative regulation of endopeptidase activity (Fig. 2B). Cystatins comprise a large superfamily of related proteins with diverse biological activities found in variable tissues, but salivary cystatins are important due their functions in immunimodulation, antimicrobial, and antiviral (Dickinson, 2002). A number of members of this protein family have been identified in saliva of humans (Carneiro et al., 2012) and in different mammals (e.g., cystatin D has been found in rat, cystatin S in dogs, cystatin C is present in Artiodactyla, Rodentia, Lagomorpha, Carnivora, and Primates (De Sousa-Pereira et al., 2015).

As one would expect, we identified carbonic anhydrase VI (CA-VI), which is an active mammalian isozyme specifically secreted by salivary glands that have multiple functions (Kivelä et al., 1999). The CA-family are zinc metalloenzymes responsible for the conversion of carbon dioxide to bicarbonate (CO2 + H2O ↔ HCO3), which buffers saliva. CA-VI also has the ability to bind enamel and act in pH homeostasis of oral cavity and prevention of dental caries (Kimoto et al., 2006). Adding strength to the host-defense capacity of Alouatta pigra, we identified lactoperoxidase LPO, bactericidal permeability-increasing protein BPI, and histidine-rich glycoprotein, that are primarily responsible for innate immunity (Bingle & Craven, 2004; Marra et al., 1990; Shin et al., 2011; Wiesner & Vilcinskas, 2010; Wijkstrom-Frei et al., 2003). The microglobulin we identified is critical for immune modulation in vertebrate animals and has been identified as a biomarker for cancer cells malignancies (Li, Dong & Wang, 2016). Salivary heat shock 70 kDa protein may represent an important immune defense mechanism in saliva of howlers, as this protein has been identified in humans to bind bacteria and increases the release of proinflammatory cytokines from immune cells (Fábián et al., 2012; Fábián et al., 2009). It is important to emphasize the presence of three salivary secretory immunoglobulins in saliva of howlers as IgA, IgG, IgM and other five isoforms (Table 1). IgA is known to induce an antigen-unspecific manner by commensal microbiota; therefore, these secretory antibodies may bind multiple antigens and are thought to eliminate commensal bacteria and self-antigens to avoid systemic recognition (Schroeder & Cavacini, 2010; Teeuw et al., 2004).

Several salivary proteins related with innate immunity of mammals were not identified in black howler monkeys (e.g., mucins De Sousa-Pereira et al., 2015). The failure to detect mucins may be due to the difficulty of assessing these proteins because of their large molecular mass, high viscosity, and poor solubility in aqueous solvents (Herzberg et al., 1979). However, we recognized that our preparation procedure of saliva samples, using 16,000 g × 10 min to separate the supernatant could result in loss of mucins in the precipitate. Other important proteins highly related to oral homeostasis, such as sthaterins and PRPs were also not identified.

We actually observed pink-staining bands in our SDS-PAGE gels, following the procedures suggested by Beeley et al. (1991) to detect PRPs, suggesting the presence of these proteins; however, some factors in our method could have interfered to detect PRPs such as the centrifugation process, and the use of trypsin for the protein digestion. Moreover, it is known that the identification of PRPs by mass spectrometry is unusually difficult (Leymarie et al., 2002); it could be possible that multiple PTM generated specific mass spectrum of modified peptides, which mass/charge (m/z) values could match with public databases (Kim, Zhong & Pandey, 2016). It is also possible that the sequences of PRPs are highly specific in Alouatta pigra.

Many of the proteins identified in howler monkey saliva are likely components of the early mammalian host defense against infection (De Sousa-Pereira et al., 2015). However, howler monkey saliva may have evolved a specific set of protein families to help them cope with infection risk and permit them to deal with habitat loss, fragmentation, and nutritional stress. (Chapman, Gillespie & Goldberg, 2005; Chapman et al., 2013). Zoonotic protozoa infection is related to degree of human contact with wild howler monkeys (Kowalewski et al., 2011).

Salivary proteins linked with taste perception and food preference

We identified several important proteins in saliva of howler monkeys that might allow them to be selective and discerning while feeding, likely facilitate their feeding selectivity. Salivary proteins to perceive beneficial traits of food were found (e.g., CA-VI, lactotransferrin, ER-Golgi intermediate compartment 53 kDa protein, microbulin, defensing, cystatin D, fatty acid-binding protein, salivary heat shock 70 kDa protein, and IgA). We also identified salivary proteins that have been related to acceptance/detection of bitter and astringent solutions in humans, which may help howler monkeys to perceive and cope with negative characteristics of food, such as bitterness and astringency (related to plant secondary metabolites and toxic compounds) including cystatins (Dsamou et al., 2012; Morzel et al., 2014; Mounayar et al., 2014; Quintana et al., 2009), histidine-rich glycoprotein (Dinnella et al., 2010), albumin and IgA (Dsamou et al., 2012). The feeding flexibility of howler monkeys enables them to thrive in small and disturbed habitat patches, where food scarcity is common (Chaves & Bicca-Marques, 2016). To our knowledge, physiological studies on taste in howler monkeys have not been conducted and there are no data of taste detection thresholds or on the ability to discriminate between different qualities of tastants (Salazar, Dominy & Laska, 2015). Consequently, for now we can only assume that the salivary proteins that we identified help these primates to choose the right diet.

Humans can differentiate among five flavors: sweet, sour, salty, bitter and umami (Van Dongen et al., 2012); although recently it has been proposed that humans can taste fatty acids (Mattes, 2011). Generally, it is accepted that each taste quality in food is related to its nutritional content (e.g., sweetness is associated with sugar, mono, and disaccharides; saltiness with sodium and protein content (Van Dongen et al., 2012); and umami with sodium and protein (Van Langeveld et al., 2017). Also, gustatory stimuli categorized as bitter and sour are associated with compounds that are potentially harmful (e.g., free protons or organic acid; bitter taste is related some toxins, Lamy et al., 2016).

Howler monkeys are herbivorous energy-maximizers and their diet is mainly leaves and ripe/unripe fruits (Chapman, 1987; Chapman, 1988; Righini, Garber & Rothman, 2017), but they can feed only on leaves for extended periods (Behie & Pavelka, 2005). A fruit-based diet is linked with a low protein intake and a decrease in mineral concentration (Silver et al., 2000), which may require selecting protein-rich and mineral-rich food items. For this purpose, howlers may benefit by secreting salivary proteins associated with gustatory sensitivity of salty and umami flavors (e.g., beta-defensin, CA-VI; cystatin D, and fatty acid-binding protein, IgA, salivary heat shock 70 kDa protein, Table 2). A leaf-diet is a diet poor in energy and fatty-acids, but high in fiber and often tannins (Righini, Garber & Rothman, 2017; Espinosa-Gómez et al., 2018), which makes selecting food difficult (Silver et al., 2000). Under these conditions, monkeys should select food items high in energy (carbohydrates, fatty-acids), but low in PSMs (tannins). Some studies in humans have found a relationship between sweet taste sensitivity and salivary proteins as cystatins (Rodrigues et al., 2019) and CA-VI (Rodrigues et al., 2017). We found in howler monkeys saliva four varieties of cystatins (Tables 1 and 2), which may help them to increase their sensitivity for sweet foods, although it remains to be investigated. For fatty-acids or lipids, it has been shown that these nutrients are important in the diet of howler monkeys (Righini, Garber & Rothman, 2017) and free fatty acids are one of the most abundant classes of nutrient metabolites in black howler monkeys foods (Amato et al., 2017). CA-VI or “gustin” plays principal role in taste sensitivity of fatty acids and sweet, salty, and sour flavors (Feeney & Hayes, 2014).

Corresponding to howler monkeys’ ability to feed on tannin-rich diet (Espinosa-Gómez et al., 2015; Espinosa-Gómez et al., 2018), we identified several salivary proteins that have been related with the capacity to accept astringent and bitter foods e.g., cystatins (Dsamou et al., 2012; Dinnella et al., 2010; Quintana et al., 2009), glyceraldehyde-3-phosphate (Quintana et al., 2009), lactoperoxidase (Morzel et al., 2014), histidine-rich glycoproteins (Dinnella et al., 2010), and albumin (Dsamou et al., 2012) (Table 2). PRPs, histatins, statherins, cystatins, and amylase are salivary proteins with considerable affinity for tannins and are involved in astringency and bitter taste (Lamy et al., 2016; Torregrossa et al., 2014). We did not identify the well-known salivary PRPs and statherins identified as first line of defense against tannins (Shimada, 2006). However, we observed in our electrophoresis gels strong bands with pink-staining that may indicate the presence of PRPs (Beeley et al., 1991). Similarly, mucins also seem to have a role in astringency, but we did not identify mucins. This may be linked to their high molecular mass, high viscosity, and poor solubility in aqueous solvents (Lamy et al., 2010).

This study supports the suggestion that α-amylase is not a component of saliva of animals feeding only on plants due their low ingestion of starch (Boehlke, Zierau & Hannig, 2015), as this enzyme was not identified in saliva of howlers. Also, chitinase was not found in our proteomic analysis, which is consistent with howlers’ feeding behavior as this protein has been identified in insectivorous-omnivorous non-human primates (Tabata et al., 2019).