Urine samples were collected from nine (four males, five females) Microcebus murinus and from nine (three males, six females) M. lehilahytsara that were housed in the colony of the Institute of Zoology, University of Veterinary Medicine Hannover (Supplementary Table S1). Three males and one female were sampled twice. Samples were obtained ad libitum during the weekly handling routines when animals urinated spontaneously, or by collecting urine in modified sleeping boxes that contained a perforated metal floor that allowed collecting the urine with a metal funnel and test tube below. Although clean material was used, it was neither possible nor intended to collect sterile urine. We were looking for chemical signatures that potentially allow species or sex discrimination. Therefore, the chemical profiles carrying these information should be stable enough even during slight bacterial alteration potentially taking place during storage. These signatures should also be stable enough to be detected in urine marks in the environment with potentially much higher rates of bacterial contamination and therefore higher potential alteration of the urine marks chemical inventory. Sample collection for this study was non-invasive and was performed in accordance with the National Research Council (NRC) Guide for the Care and Use of Laboratory Animals, the European Directive 2010/63/EU on the protection of animals used for scientific purposes, and was approved by the relevant ethics committee of the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (reference number AZ 33.12-42502-04-14/1454). Both mouse lemur species were kept in the animal facility of the Institute of Zoology at the University of Veterinary Medicine Hannover, Germany (licensed by Ordnungsamt, Gewerbe- und Veterinärabteilung, Landeshauptstadt Hannover, AZ 42500/1H). All samples were frozen directly in inert sample tubes (KH-Flasche G 1, CS Chromatographie Service) at -20°C and stored until chemical analyses that took place shortly afterwards. Urine samples were collected between the 18th March and the 22nd April of the years 2013, 2014, or 2015, respectively, i.e., during the reproductive season of the colony (Wrogemann and Zimmermann, 2001; Wrogemann et al., 2001). All but one males (Guido) had fully developed testes on the day of sampling (Supplementary Table S1). All but two females were sampled during their interestrus, i.e., the time between two successive estrus periods (Wrogemann et al., 2001). Two females (Olympia, Irmi second sample), however, were sampled on the first day of their second estrus of the season.

Dichloromethane extracts of the mouse lemur urine were analyzed using gas chromatography coupled with mass spectrometry (GC/MS). Samples of 0.25–1.2 mL urine collected from each individual were extracted with 300 μL of dichloromethane by stirring at room temperature for 12 h. The two phases were separated and the organic phase was dried with NaCl. Extracts were analyzed by GC/MS (GC 7890B/MSD 5977A, Agilent Technologies, Santa Clara, CA, United States). The GC/MS system was equipped with a HP-5ms fused silica capillary column (30 m, 0.22 mm internal diameter, 0.25 μm film, Agilent Technologies). Conditions were as follows: inlet pressure: 67 kPa, He-flow: 1.2 ml/min, injector: 250°C, transfer line: 300°C, electron energy 70 eV. The GC oven temperature was kept at 50°C for 5 min, followed by a temperature gradient of 5°C/min to 320°C. Identification of compounds was performed by comparison of their mass spectra and retention indices (determined from a homologous series of n-alkanes, C₈–C₃₂) to those of commercial mass spectral libraries (Wiley 7, NIST 08) and an in-house data base. Water blank samples were prepared and extracted like the urine samples and analyzed by GC/MS. Compounds found in this sample were removed from the subsequent analyses. Only compounds eluting before retention index (RI) of <2000 (36 min) on the apolar column were included in the analysis, since volatile compounds elute before that time and later eluting compounds were overlaid by common fatty acid derivatives. In total, 22 urine samples of 18 different individuals were analyzed. A total of four individuals, two male M. murinus and one male and one female M. lehilahytsara, were sampled twice.

Presence or absence of all compounds was noted for each sample. Before statistical analyses, all compounds were removed that were present either in all samples or in one sample only. This approach yielded a total of 385 compounds for further analyses. The number of compounds in male and female samples of both species was first compared with a GLMM with the number of compounds as dependent variable, sex and species as fixed factors, and the individual identity as random factor. Since only four of 18 individuals were sampled twice, we used a Gauss-Hermite approximation (nAGQ = 10) to evaluate the log-likelihood and to reach model convergence. The model was calculated with the package lme4 in R (R Core Team, 2015).

The list with all presence and absence codes in all samples was subjected to a Principal Component Analysis in the program STATISTICA 12 (Statsoft, Inc.). A total of 21 principal components were derived with Eigenvalues ranging from 6.56 to 60.68, and single components explained between 1.71 and 15.76% of the variance contained in the dataset. The first 15 principal components (explaining 86.395% of the variance) were used in two discriminant function analyses (stepwise forward approach, p to enter = 0.05), one for the discrimination between the sexes and one for the discrimination between the species. All samples were classified by means of the calculated discriminant function according to sex (male, female) or species (M. murinus, M. lehilahytsara). The factor with the highest and significant contribution to the respective discriminant function was evaluated, and compounds with factor loadings of >0.6 or <−0.6 are identified and reported.

Pilot experiments showed that the highest number of compounds was observed by solvent extraction of urine with dichloromethane compared to solid phase microextraction (SPME) analysis. Therefore, the urine samples were extracted by the former procedure and analyzed by GC/MS. A total of 977 different compounds with a retention index of <2000 were detected in the 22 analyzed samples, of which 388 (40%) were identified (Supplementary Table S4). Despite this wealth of compounds, only 173 ± 49 (mean ± SD) compounds were present in a single urine extract. A wide range of different groups of compounds were present including alcohols, aldehydes, amides, ketones, (fatty) acids, esters, terpenes, steroids, and heterocycles including lactones, indoles, or quinolines. Of these, 42 components occurred in every sample or most samples (>82% occurrence) e.g., phenol, benzaldehyde, acetophenone, and nonanal, while 11 of them were unknown components (Table 1). The main compounds with the highest relative concentration in most samples were 2-methylpropyl acetate, dimethyl sulfone (1), 4-methylphenol (2), quinoline (3), nonanoic acid (4), 1-hexadecanol (5), hexadecenoic acid (6) and five unknown compounds (Figure 1).

Ten male urine and twelve female urine extracts of both species were analyzed with regard to sex differences in volatile compounds. There were noticeable differences in the number of compounds found in the urine extracts of males and females of both species (Figure 2). The overall mean number of compounds was 145 compounds in male urine samples (132 in M. lehilahytsara, 154 in M. murinus), but 195 compounds in female urine samples (201 in M. lehilahytsara, 188 in M. murinus) (Figure 2). The GLMM revealed these differences to be significant in the overall analyses [Estimate_(male) = −0.40388, SE = 0.10592, z-value = −3.813, p = 0.000137], while no influence of species on the numbers of compounds were detected [Estimate_(murinus) = 0.05986, SE = 0.10264, z-value = 0.583, p = 0.559732]. The difference in compound numbers between the sexes was also significant when analyzing both species separately [Estimate_{(male, lehilahytsara)} = −0.56280, p = 0.000233; Estimate_{(male, murinus)} = −0.26074, p = 0.045].

Generally, there was no single compound that was present in every sample of one sex and absent in the other one. However, the discriminant function analysis (DFA) based on the first 15 principal components provided a highly significant result [Wilks’ λ = 0.04290, F_{(10, 11)} = 24.539, P < 0.000001]. Ten of the 15 factors (1, 3, 5, 10, 12, 2, 13, 14, 4, 8) were included in the discriminant function with six factors (1, 3, 5, 10, 12, 2) contributing significantly (Supplementary Table S2). This discriminant function classified all urine extracts of males and females correctly to their appropriate sex. The factor loadings on the first and most influential factor (factor 1, Eigenvalue 60.68, 15.8% variance explained) revealed five compounds to contribute importantly to the specificity of the male urine profile (Table 2). These were benzyl acetate and propionate, and three unknown compounds. Four of these compounds were never detected in female urine samples. In contrast, 42 compounds were identified as important volatile signatures in the female urine. We identified 22 of these compounds that belonged to various compound classes including short alcohols, fatty acids, esters or sulfur compounds. The remaining 20 compounds remained unknown. Despite being present in most female samples, however, the 42 compounds were not exclusively present in female urine.

Eleven urine extracts of M. murinus and M. lehilahytsara, respectively, were analyzed to detect species-specific signatures in the urine constituents. Generally, there was no single compound that was systematically present in every sample of one species and absent in the other one. However, the DFA of the 15 first principal components provided a highly significant discriminant function [Wilks’ λ = 0.11607, F_{(7, 14)} = 15.230, P < 0.000001]. Seven factors (2, 6, 5, 3, 10, 1, 15) were included in the function with four factors (2, 6, 5, 3) contributing significantly to the separation of the species (Supplementary Table S3). All urine extracts of M. murinus and M. lehilahytsara were assigned to the correct species by the DFA. The DFA revealed 15 compounds loading strongly onto the most influential factor 2 (Eigenvalue 43.52, 11.3% variance explained) which were characteristic for the M. murinus profile (Table 3). Ten of these compounds were specific for M. murinus, while the remainder occurred in M. lehilahytsara but in very low concentration. Nine of the fifteen compounds were identified mainly as ketones and aromatic amides, such as 3-octanone or 2-aminobenzamide. The DFA analysis further revealed seven compounds that were characteristic for the M. lehilahytsara profiles. However, all of these were unknown. Two of them were specific to M. lehilahytsara, while the remainder was only more abundant in this species than in M. murinus (Table 3).

The large number of compounds detected by our method, over 900, suggests a very high potential of solvent extraction for compound identification and assessing signature mixtures. Even compounds not commonly detected by headspace methods, e.g., larger compounds with low vapor pressure such as cholesterol, or strongly H-bonding compounds such as amines or phenols could be successfully extracted. These compounds may potentially serve as olfactory signal components, and may be perceived e.g., by the vomeronasal organ or the main olfactory epithelium of the mouse lemurs. The effectiveness of solvent extraction is particularly evident when comparing with the previous study on urine volatiles of lemurs using solid phase dynamic extraction (SPDE) (delBarco-Trillo et al., 2011).

The large number of compounds present in most samples indicates the potential of urine to be used as scent mark by Microcebus spp. and might differentiate them from other genera, although there are so far no comparative results available from other lemur species obtained by the method used by us. In the study of delBarco-Trillo et al. (2011), SPDE-GC/MS was used to investigate the headspace volatiles of the urine of 12 lemur species. They reported 33 compounds as the highest number of urine compounds from one species. Benzaldehyde, nonanal, and decanal were present in almost every sample of the 12 species investigated.

These compounds were also present in almost all Microcebus samples investigated by us. Because of their common occurrence, it is not surprising that they were not diagnostic for sex or species in the DFA. In the headspace of the urine of Cheirogaleus medius, the only member of the Cheirogaleidae investigated so far, delBarco-Trillo et al. (2011) detected only 15 compounds. Some lemur-typical compounds like phenol, 4-methylphenol, or acetophenone were not detected. In contrast, these compounds were detected in every urine sample of the mouse lemurs investigated here. In general, the six compounds discussed in this paragraph are typical compounds of urine, not only in strepsirrhine primates but also in other mammals like rats, mice, cheetah or elephants (Burger et al., 2006; Röck et al., 2006; Osada et al., 2009). 4-Methylphenol and another widespread urinary compound also present in mouse lemur urine, cyclohexanone, are released from temporal gland secretions of elephants in larger amounts during musth, indicating a putative role as semiochemicals (Rasmussen and Perrin, 1999).

The comparison of the urine extracts of male and female mouse lemurs showed a significant difference between the sexes with females excreting significantly more compounds than males, although in males a wide variation in the number of compounds occurred. The total ion chromatograms (TIC) of males and females (Figure 1) also indicated that the concentration of the volatiles was higher in female samples. The statistical analyses further revealed that not only the number of compounds, but also the composition differed significantly between the sexes. The distinction of male and female urinary compounds is likely due to physiological and metabolic differences between the sexes, but it also suggests that females potentially convey more information via urine compared to males.

One of the very few compounds that were characteristic for males in our study were the two esters benzyl acetate and benzyl propionate (Table 2). Benzyl acetate is an attractant and pheromone for many Coleoptera and Hymenoptera, for example male euglossine bees (Willams and Whitten, 1983; Schiestl and Roubik, 2003). It seems not unlikely that these compounds may also play a role in the communication of mouse lemurs.

The analysis of the female mouse lemur urine profiles revealed a wide range of chemically different compounds to contribute to the specificity of their urine composition (Table 2). Some of these compounds are known from other mammals. Hexanoic acid was found in the gland secretions of L. catta (Knapp et al., 2006) and Old World monkeys (Mandrillus sphinx) (Setchell et al., 2010), in the urine of male rats (Osada et al., 2009) and male Bengal tigers (Burger et al., 2008; Osada et al., 2009), African wild dogs (Apps et al., 2012) and Iberian wolves (Martín et al., 2010; Apps et al., 2012). Heptanoic acid, dodecanoic acid, and 1-tetradecanol have also been identified in the urine of male Bengal tigers (Burger et al., 2008) and feces of the African Wild Dogs (Burger et al., 2008; Apps et al., 2012). Dodecanoic acid was reported from the labial glandular secretions of L. catta (Scordato et al., 2007) and the scent marks of female marmoset monkeys, Callithrix jacchus (Smith et al., 2001).

Our results are in contrast to those of delBarco-Trillo et al. (2011), who found no difference in the total number or relative abundance of compounds between male and female strepsirrhines. However, this may be due to the small number of samples they analyzed and the low overall number of compounds detected. In addition, the closest relative of the mouse lemurs, Cheirogaleus medius, was not analyzed under this aspect. Future comparative studies are needed to clarify if there is indeed a suite of compounds that is typical for all or most female lemur species or strepsirrhines and may then represent an ancestral communication signals, or whether they were largely derived during the rather recent radiation of the subclades, such as the genus Microcebus (Hotaling et al., 2016; Yoder et al., 2016). Future tests will be needed to evaluate the biological relevance of the discussed compounds for olfactory communication.

Microcebus murinus and Microcebus lehilahytsara can be clearly differentiated by the composition of the urine extracts. Both species showed a distinct urinary profile of compounds. Unfortunately, the structure of the compounds identified as being specific for M. lehilahytsara remain unknown (Table 2). Two of these compounds were specific for this species, while the other five compounds occurred occasionally in M. murinus as well (Table 3).

Fifteen compounds differentiated M. murinus from M. lehilahytsara. Within these two distinct compound classes prevailed, ketones and aromatic amides, and 10 compounds were specific for M. murinus (Table 3). A few of the identified compounds are known as potential semiochemicals in other mammals. For example, 3-octanone was also detected in the urine of the strepsirrhine Daubentonia madagascariensis (delBarco-Trillo et al., 2011) and the scent marks of Giant pandas, Ailuropoda melanoleuca (Hagey and MacDonald, 2003). 5-Methylhexan-3-one has been identified in the preorbital gland secretion of klipspringer, Oreotragus oreotragus (Burger et al., 1997) and the blue duiker, Cephalophus monicola (Burger and Pretorius, 1987; Burger et al., 1997). Trimethylpyrazine was detected in the scent marks of female marmoset monkeys, Callithrix jacchus (Smith et al., 2001) and in the urine of bobcats, Lynx rufus (Mattina et al., 1991) as well as maned wolves, Chrysocyon brachyurus (Mattina et al., 1991; Smith et al., 2001; Goodwin et al., 2013). 2,4-Dimethylquinazoline was reported from dorsal patches of males of the Curaçaoan long-nosed bat, Leptonycteris curasoae (Muñoz-Romo et al., 2012).