Results

We analysed a total of 138 dropping samples for Myotis escaleraiand 90 for Myotis crypticus from 49 locations, 26 of which were in the broad-scale allopatric regions and 23 in sympatric regions. Within the sympatric regions (La Rioja and southern Cantabria), 91 samples were classified as locally allopatric and 28 as locally sympatric (Fig. 1, Table 1; Supplementary Table S1). We recovered a total of 2,859,300 reads (Supplementary Table S3 for details) from the 228 dropping samples for the four combinations of PCR replicates and primers (1,403,636 from ANML1 and 1,455,664 from ZBJ). These reads were associated into 1461 different BINs. Based on the BINs present in sequencing blanks, we removed for the ANML primer 6 BINs from the first run of 56 dropping samples, and 2 BINs for the second run of 10 samples. For the ZBJ primer, we removed 5 BINs from the first run of 10 samples and 23 BINs from the second run of 88 samples. Based on BINs present in extraction blanks, we removed a total of 39 BINs from 16 dropping samples

Characterising the diet of M. escalerai and M. crypticus

A total 19 arthropod orders were obtained based on the 1461 BINs (Supplementary Data file S1 for list of prey items obtained for each bat species). The diets of M. escalerai and M. crypticus were characterised by high arthropod diversity, and were composed mostly of the orders Lepidoptera (M. escalerai = 26.6 %; M. crypticus = 23.7%), Diptera (24.8%; 33.2%), Araneae (20.7%; 17.2%), but also included Hemiptera (11.8%; 6.2%), Coleoptera (4.8%; 5.1%), and Orthoptera (4.3%; 4.8%), among others (<5%) (Fig. 2a-b; Supplementary Fig. S2 for diet composition based on POO and RRA measures). Diet composition at the prey order level was very similar between bat species (OJK = 0.98, above 1000 null models). However there were differences in the number of BINs per sample of Diptera, which was lower in M. escalerai (5.27 versus 6.75) (Negative binomial GLM: z1,226=-2.03, P = 0.042), and Hemiptera, which was higher in M. escalerai (2.09 versus 1.68) (Negative binomial GLM: z1,226=2.85, P = 0.004 Supplementary Fig. S3).
At the prey species (BIN) level, Levins’ niche breadth was similar for both species, BA = 0.17 for M. escalerai and BA = 0.19 for M. crypticus . Niche overlap between species was higher than expected by chance (OJK = 0.71, above 95% of 1000 null models). The samples from the two bat species showed some differences in prey item composition in NMDS ordination space (Fig. 3a, Stress: 0.25, k=3, non-metric fit R2=0.934, Linear fit, R2= 0.532). An analysis of similarity confirms that distance in prey item composition among samples is greater between species than within species (ANOSIM R statistic: 0.10, P= 0.001).

Trophic partitioning in sympatric versus allopatric locations

At the arthropod order level, there is no clear pattern of shift from high similarity in order composition between species to differential use in sympatry at any of both spatial scales (OJK regional allopatry =0.88, OJK regional sympatry = 0.96, >1000 null models, Fig 2c-d; OJK local allopatry =0.95, OJK local sympatry = 0.98, >1000 null models, Fig 2e-f). When examining the number of BINs of the main arthropod orders per sample, there were differences between bat species between the allopatric regions for Araneae and Hemiptera, which were both higher in M. escalerai (M. escalerai = 4.00, 1.65, M. crypticus = 2.16, 0.55 respectively), and for Lepidoptera, which was higher in M. crypticus (4.6, 11.94) (Negative binomial GLM: df=1,98, P<0.05). In the sympatric region, the higher number of Hemiptera in M. escaleraiholds (M. escalerai = 2.50, M. crypticus 1.50), and in Lepidoptera there is a shift whereby is M. escalerai the one that consumes a higher number (M. escalerai = 6.78, M. cryptius= 4.09, Negative binomial GLM: P<0.05). At the fine-scale, within the sympatric region, the only difference found between the bat species was the higher number of BINs per sample of Hemiptera (2.67, 1.49) (Negative binomial GLM: z1,89= -2.68, P=0.007) and Lepidoptera (6.70, 3.64) in M. escalerai in allopatric locations (Negative binomial GLM: z1,89= -2.92, P=0.004). There were no differences in arthropod orders consumed between the bat species in locally sympatric locations (Negative binomial GLM: P>0.05; Supplementary Fig. S4).
At the prey species (BIN) level, at the broad-scale, trophic niche similarity between species was lower in allopatric than in sympatric regions (OJK allopatric = 0.35, OJKsympatric= 0.62). Conversely, at the fine-scale, within the sympatric region, trophic niche overlap between species was higher in locally allopatric locations (OJK= 0.56) than in locally sympatric locations (OJK= 0.37). Despite the low values of overlap in regionally allopatric and locally sympatric locations, in all the four cases, observed niche overlap was higher than 1000 null models. When measuring trophic niche overlap between species using pairs of locations, we observed the same pattern. At the broad scale we found higher diet overlap in sympatric than allopatric locations (OJK sympatric= 0.107±0.056, OJKallopatric= 0.050±0.04; Gausian hurdle model: binomial GLM: z1,316=4.76, P<0.05; Gaussian GLM: t1,265=8.26, P<0.05). In contrast, at the fine-scale, niche overlap was lower among pairs of locally sympatric than among locally allopatric locations (OJK sympatric= 0.099±0.065, OJK allopatric= 0.126±0.057; Linear model: F1,73=6.34, P=0.014; Fig 3b).

Functional diet analysis

Both species had a similar high percentage of non-volant (M. escalerai =21.4 %, M. crypticus =19.5%) and not actively-volant (44.6%, 45.8%) prey items in the diet. Only 34.0% and 34.7% of weighted percent of occurrence (wPOO) was composed of arthropods classified as nocturnally volant (Fig. 4a). There were no differences in the overall percentage of not nocturnally volant prey taxa (BINs) per sample between bat species (66% ±20%, 66% ±21%, Linear model: F1,217 <0.001, P= 0.990; Fig. 4b). When analysing functional diet differences separately in allopatric versus sympatric regions, we found differences between species in allopatric regions, whereby M. crypticus consumed lower percentage of prey that were not nocturnally volant (allopatric regions:M. escalerai = 66% ±19%, M. crypticus =48% ±25%; F1,98=11.72, P<0.05; Fig. 4c; sympatric regions: 65% ±22%, 71% ±17%; F1,117=2.3, P=0.13; Fig. 4d). At the fine-scale, there were no differences among bats in locally allopatric locations (M. escalerai = 67% ±22%, M. crypticus =72% ±16%, F1,89=1.325, P=0.250; Fig. 4e) while in locally sympatric locations the percent of prey that were not nocturnally volant was borderline lower in the diet of M. escalerai (52% ±17%) than M. crypticus (68% ±18%; F1,26=4.03, P=0.055; Fig. 4f).

Prey consumption relative to availability

In nearly all cases, we could not detect over- or under-selection of arthropod orders by the bats relative to their availability in sweeping samples. The distribution of prey order selection values between the 1st and 3rd quartiles overlapped with zero in all cases, except in the case of M. escalerai and Lepidoptera, where positive selection values could indicate over-selection (1st-3rd quartile: +0.43 — +18.7; Supplementary Fig. S5).

Metabarcoding and primer performance

There were compositional differences in the prey orders that each primer recovered. A large proportion of the BINs identified in dropping samples were only recovered by one of the primers (Supplementary Fig. S6). Neuroptera, Orthoptera, Coleoptera were more frequently recovered by ZBJ, while Plecoptera and Thysanoptera, Dermaptera Mantodea where more frequently recovered by ANML (Supplementary Fig. S6). Supplementary Figure S7a-b shows composition for a subset of dropping samples comparing each primer.
In sweeping samples, 7065 insects were identified morphologically to order level, with an average of 174.3 individuals per sample (range 7-624). Using molecular tools, we recovered 899,853 reads (Supplementary Table S2), and identified 813 different BIN items. Some of the rarer orders were under-represented in the molecular analysis. Specifically, Opiliones, Dermaptera, and Archaeognatha, appeared in more than 10 sweeping samples each identified morphologically, but were rarely recovered in the molecular approach, despite being present in the reference databases (Supplementary Fig. S8).