Microbiota and the volatile profile of avian nests are associated with each other and with the intensity of parasitism

FEMS Microbiology Ecology , 2024, 100 , fiae106 DOI: 10.1093/femsec/fiae106 Ad v ance access publication date: 24 July 2024 Research Article Microbiota and the volatile profile of avian nests are associated with each other and with the intensity of parasitism Mónica Mazorr a-Alonso 1 , J uan Manuel Per alta-Sánc hez 2 ,* , Philipp Heeb 3 , Staffan Jacob 4 , Manuel Martin-Vivaldi 5 ,6 , Manuel Martínez-Bueno 6 ,7 , Rafael Núñez-Gómez 8 , Oriol Sacristán-Soriano 9 , Juan José Soler 1 ,6 ,* 1 Departamento de Ecología Funcional y Evolutiva, Estación Experimental de Zonas Áridas (CSIC), 04120 Almería, Spain 2 Departamento de Zoología, Universidad de Sevilla, 41012, Spain 3 Centre de Recherche en Biodiversité et Ecologie, UMR 5300 Bâtiment 4R1, Université Paul Sabatier, 118, Route de Narbonne, 31062 Toulouse Cedex 9, France 4 Station d’Ecologie Théorique et Expérimentale, CNRS, UMR 5321 Moulis, France 5 Departamento de Zoología, Universidad de Granada, 18071 Granada, Spain 6 Unidad asociada (CSIC): Coevolución: cucos, hospedadores y bacterias simbiontes. Universidad de Granada, Spain 7 Departamento de Micr obiología, Univ ersidad de Granada, 18071 Granada, Spain 8 Servicio de Instrumentación Científica, Estación Experimental del Zaidín (CSIC), 18008 Granada, Spain 9 Institut Català de Recerca de l’Aigua (ICRA), Emili Grahit 101, E-17003 Girona, Spain ∗Corresponding authors. Departamento de Ecología Funcional y Evolutiva, Estación Experimental de Zonas Áridas (CSIC), Carr eter a de Sacramento, s/n, 04120, Almería, Spain. E-mail: [email protected] ; Departamento de Zoología, Universidad de Sevilla, Avenida Reina Mercedes, 6, 41012, Seville, Spain. E-mail: [email protected] Editor: [Paolina Garbeva] Abstract Bacteria have been suggested as being partially responsible for avian nest odours and, thus, v olatiles fr om their meta bolism could influence the intensity of selection pr essur es due to parasites detecting olfactory cues of their hosts. Here , w e tested this hypothesis by exploring intraspecific and interspecific v aria bility in micr obial envir onments, v olatile pr ofiles and intensity of ectoparasitism by Carnus hemapterus in the nests of 10 avian species. As expected, we found that (i) alpha and beta di v ersity of micr obial and v olatile pr ofiles wer e associated with each other. Mor eov er, (ii) alpha di v ersity of bacteria and volatiles of the nest environment, as well as some particular bacteria and v olatiles, w as associated with the intensity of parasitism at early and late stages of the nestling period. Finally, (iii) alpha di v ersity of the nest microbiota, as well as some particular bacteria and v olatiles, w as corr elated with fledging success. When considering them together, the results support the expected links between the microbial environment and nest odours in different bird species, and between the microbial environment and both ectoparasitism intensity and fledging success. Relati v e a bundances of particular volatiles and bacteria predicted ectoparasitism and/or fledging success. Future resear c h should prioritise experimental appr oaches dir ected to determine the r ole of bacteria and v olatiles in the outcomes of host–ectopar asite inter actions. Ke yw ords: avian nest microbiota; avian nest odours; bacteria; ectoparasitism; nidobiome; volatiles Introduction Exploring how the microbiota influences animal behaviour in general and chemical communication in particular is no w adays a cutting edge line of r esearc h (Ezenwa and Williams 2014 , Carthey et al. 2018 , Maraci et al. 2018 , Mazorra-Alonso et al. 2021 ). The metabolism of symbiotic micr oor ganisms gener ates volatiles that influence the odour profile of individuals (Archie and Theis 2011 , Engl and Kaltenpoth 2018 ) and, thus, bacterial symbionts might play essential roles in animal signalling and chemical communication (Ezenwa and Williams 2014 , Carthey et al. 2018 ). It is known that c har acteristics of the bacterial comm unities r eflect the phenotypic and physiological conditions of their animal hosts (Theis et al. 2013 , Leclaire et al. 2017 , Bourne et al. 2023 ). T hus , specific volatiles from the metabolism of such bacterial symbionts would inform conspecifics and hetero-specifics (i.e. predators or parasites) of characteristics of their animal hosts (Archie and Theis 2011 , Mazorra-Alonso et al. 2021 ). Because, in general, parasites and predators use olfaction to seek and detect their victims (Bo w en 1991 , Reneerkens et al. 2005 , Poldy 2020 ), some of the eav esdr opped cues could be of bacterial origin, as occurs in mosquitoes seeking human hosts (Verhulst et al. 2009 , 2011a , 2011b ). Odours associated with the remains of animal physiological activities such as urine, faeces and decomposing residue of prey or food can act as attractants for ectoparasites and predators (Becker et al. 1995 , Hall 1995 , Vale et al. 2009 , Hassanali et al. 2011 ). For instance, the accumulation of faecal waste increases the attraction of mosquitoes to w ar ds cages with hamsters (Becker et al. 1995 ). Mor eov er, the occupation of avian nest-holes by mammals during winter affected the ectoparasitism of hoopoe ( Upupa epops ) nestlings in spring (García-Núñez et al. 2023 ), and the experimental addition of faeces in nests of spotless starlings ( Sturnus unicolor ) increases the probability of predation (Azcárate-García et al. 2019 ). Faeces and other debris of the biological activity of animals Recei v ed 15 February 2024; revised 22 May 2024; accepted 23 July 2024 ©The Author(s) 2024. Published by Oxford Uni v ersity Pr ess on behalf of FEMS. This is an Open Access article distributed under the terms of the Cr eati v e Commons Attribution-NonCommercial-NoDerivs licence ( https://creativecommons.org/licenses/bync- nd/4.0/ ), which permits non-commercial r e pr oduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commer cial re-use , please contact [email protected] Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 2 | FEMS Microbiology Ecology , 2024, Vol. 100, No. 10 harbour dense bacterial communities that may mediate the associations between debris and risk of parasitism and/or predation. T hus , to understand the possible role of volatiles of bacterial origin explaining the association between animal physiological activity and their detectability by parasites or predators, it is necessary to explore the association between symbiotic bacterial communities of animals and volatiles in the envir onment wher e the animals liv e (Mazorr a-Alonso et al. 2021 ). Mor eov er, finding associations between risk of infection or predation and the symbiotic bacterial communities of animals or their volatile profiles would further suggest a role of the symbiotic bacteria on the interactions of hosts with parasites and predators. A review of the literature sho w ed that those associations have rarely been explored in natural conditions (see the revision in Mazorra-Alonso et al. 2021 ). Here we use the avian nest environment of several bird species to explore them. T he a vian nest en vironments , particularly those located in cavities, are suitable habitats to explore such associations. Nests are central locations where bacterial communities associated with birds can be c har acterised, and cavities better than open nests allow capturing the volatile pr ofiles deriv ed fr om animal physiological activities . Most a vian species build nests where they lay and incubate the eggs, and where nestlings de v elop (Hansell 2000 ). Location and materials used for nest construction vary interspecifically and influence temperature and humidity (Windsor et al. 2013 , Deeming and Mainwaring 2015 ), risk of predation (Mainwaring et al. 2015 ), parasitism (López-Rull and Macías- García 2015 ) and also the bacterial environment where offspring de v elop (Per alta-Sánc hez et al. 2011 , Soler et al. 2015 , West et al. 2015 , Ruiz-Castellano et al. 2016 ). The nest microbial environment might include harmful bacteria that affect egg viability or feather degradation (Ramnani et al. 2005 , Peralta-Sánchez et al. 2018 ), or beneficial bacteria that produce antibiotics that pr e v ent the establishment of pathogens (Per alta-Sánc hez et al. 2010 , Soler et al. 2010 , Per alta-Sánc hez et al. 2011 , 2014 , Ruiz-Castellano et al. 2016 , 2019 ). T he bacterial en vironment of a vian nests would also depend on nest sanitation behaviour, which also varied interspecifically (Ibáñez-Álamo et al. 2017 ). T hus , because factors affecting the microbiota of avian nests vary interspecifically, it is expected that both the nest microbial environment and its effects on their hosts will be species specific (Soler et al. 2011 , 2012 , Peralta- Sánchez et al. 2012 ). Volatiles of the nest environment should also vary interspecifically because different bird species use different nest materials (i.e . grass , moss , feathers , flo w ers, gr een and ar omatic plants) to build their nests, and because volatiles, as by-products of metabolism of nest bacterial communities, also vary interspecifically. As nest materials would influence the nest bacterial environment, volatiles and bacterial profiles of the nest environment should be related to each other; a prediction that we test here in the nests of 10 species . Moreo ver, if parasites or predators use chemical cues to find the nests of their victims, interspecific variation in parasitism selection pressure should associate with that of volatile profiles and/or bacterial communities of avian nests (Mazorra-Alonso et al. 2021 ). Several results suggest that volatiles from bacterial metabolism were partiall y r esponsible for the incidence of par asitism and pr edation in different bird species . For instance , traps baited with bacteria and uropygial secretion of hoopoes ( Upupa epops ) repelled biting midges (Tomás et al. 2020 ), while autoclaving the nest material of hoopoe nests reduced the intensity of parasitism by Carnus hemapterus of nestlings (Mazorra-Alonso et al. 2020 ). Mor eov er, incr easing bacterial loads in spotless starling nests incr eased the pr obability of pr edation (Azcár ate-García et al. 2019 ). Whether the known effects of symbiotic bacterial communities on parasitism selection pressure are mediated by volatiles of bacterial origin is, ho w e v er, a matter of debate (Mazorra-Alonso et al. 2021 ). For instance, se v er al studies suggested that animal odours are associated with the probability of ectoparasitism and/or predation (Bo w en 1991 , Reneerkens et al. 2005 , Poldy 2020 ), but results indicating a link between the symbiotic bacterial community and animal odours are scarce (Leclaire et al. 2017 , Jacob et al. 2018 , Whittaker et al. 2019 ). More importantly, with few exceptions (e.g. Verhulst et al. 2009 , 2011a ), the effects of bacteria and odours on ectoparasitism have traditionally been explored separ atel y. T hus , exploring the expected associations between symbiotic bacteria, animal odours and par asitism r equir es testing the hypothesis that volatiles from bacterial metabolism mediated the potential effects of volatiles and microbial symbionts on the pr obability of par asitism. The importance of volatiles from bacterial metabolism on parasitism has r ar el y been examined (but see Bourne et al. 2023 ). In this study, we explore the association between bacteria and volatiles of avian nests and the hypothesis that bacteria in the avian nests emit volatiles that influence the ectoparasitism of avian nestlings by the hematophagous fly Carnus hemapterus , which is the most abundant and prevalent ectoparasite in our study ar ea. Mor eov er, her e we explor e interspecific v ariations in either the bacterial community of nest materials, nest odours or the parasitism of avian nestlings, as well as the associations among them. Because parasitism, by definition, should have negative effects on re producti ve success, we also explore the effect of bacterial community and of volatile profiles on fledging success. We collected information from the nest environments of 10 species that fr equentl y use installed nest boxes or ar e abundant in the study area. The hypothesis tested predicts two types of associations. On one hand, (i) volatile profiles and bacterial communities of the nest en vironment (i.e . nest material) should be related to each other, and, on the other hand, (ii) both bacterial and volatile pr ofiles should pr edict the pr obability and/or intensity of ectoparasitism and fledging success. We explored those associations at interspecific and intraspecific levels. Because ectoparasitism was centred on the same blood-sucking species, we predicted that the expected associations would a ppear, e v en after controlling for the species identity. Materials and methods Study area and species The study area was located in the Hoya the Guadix, (Granada, Southern Spain, 37 ◦18  N, 38 ◦11  W), a plateau at 1000 m a.s.l. with semiarid climate, where ∼400 cork-made nest-boxes were available for wild birds; most of them attached to tree trunks and walls, but also hidden in piled stones . T he dimensions of the nest-boxes were 35 ×18 ×21 cm (internal height ×width ×depth), 24 cm (bottom-to-hole height) and 5.5 cm (entrance diameter). Information about the study area is available elsewhere (Peralta-Sánchez et al. 2018 ). We sampled 10 species that fr equentl y br eed in this ar ea: hoopoes, scops owls ( Otus scops ), little owls ( Athene noctua ), Eur opean r ollers ( Coracias garrulus ), stoc k pigeons ( Columba oenas ), house sparrows ( Passer domesticus ), great tits ( Parus major ), jackdaws ( Coleous monedula ), magpies ( Pica pica ) and spotless starlings. All of these species use our nest-boxes for r epr oduction, with the Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 Mazorra-Alonso et al. | 3 exception of the magpie. Sample sizes for each avian species and type of collected samples are shown in Table 1 . The most abundant ectoparasite in our study area is Carnus hemapterus , a common generalist hematophagous fly of ∼2 mm in length (Martín-Vivaldi et al. 2006 , Calero-Torralbo et al. 2013 ) that parasitises a large number of avian species (Capelle and Whitworth 1973 ). In spring, in sync hr onisation with their hosts, winged adults emer ge fr om pupae that de v eloped inside the avian nest materials fr om pr e vious r epr oductions (Valer a et al. 2003 , Caler o- Torralbo et al. 2013 ). Adult flies can stay in the same nest or disperse, and, once they arrive at the chosen host-nest for parasitism, they lose their wings and feed on nestlings or incubating adults, r eac hing the peak of abundance in avian nests just before the start of nestling feathering (Liker et al. 2001 , Avilés et al. 2009 ). F ield work The fieldwork was carried out during the breeding seasons (March–June) of 2017 and 2018. We checked all nest-boxes once per week and, once an active nest was detected, depending on the number of eggs in the nests and the length of the incubation period of each species, we estimated the expected hatching date of the first eggs (hereafter, day 1 of the sampling protocol). During the nestling stage, nests were visited four times to sample the bacterial communities of nest materials and the volatiles of the nest-box environment (see below), and to estimate the intensity of parasitism of nestlings and fledging success. For each nest visit, w e w or e ne w latex globes pr e viousl y washed with 96% ethanol to avoid contaminations among nests. We collected 1.5 ml of nest materials that were in contact with nestlings (i.e. nest cup) twice: at the beginning (i.e. first quarter of the nestling period; from days 4 to 6 depending on the length of nestling period of the bird species) and at the end of the nestling stage (i.e. last quarter of the nestling period; from days 17 to 23 depending on the bird species). These samples were stored in sterilised 1.5-ml microfuge tubes and k e pt cold in a portable fridge until the y arri v ed at the labor atory. Upon arriv al, the samples wer e stored at −20 ◦C until bacterial DNA extraction. We also sampled volatiles of nest-box environments twice, at the two nestling stages described abo ve . In the case of magpies that are non-hole nesters, we did not sample nest volatiles. Volatiles wer e ca ptur ed in Solid Phase Micr oextr action (SPME) fibres . T he fibre was installed on one of the walls ( ∼7 cm high from the nest material) with the sensitive end protected with a twosided opening glass pipette tip. The exposition time of capturing fibres in the nest environments was 24 h. Afterw ar ds, w e r emov ed the fibr e fr om the nest-box and the sensitive side was introduced into a closed glass vial that was k e pt cold in a portable fridge (0-4 ◦C) until we arrived at the laboratory. Upon arrival, the vial was stored at –20 ◦C until gas c hr omatogr a phy-mass spectr ometry (GC-MS) analysis . T he storage of samples under laboratory conditions ne v er exceeded 1 week. After the anal yses, the SPME fibr es wer e r econditioned (i.e. eliminating an y c hemical tr aces) and k e pt at –20 ◦C until they were reused in the field. The ectoparasitism of nestlings was also estimated twice, at the earl y (fr om day 5 to day 8 depending on the bird species) and the late sta ge (fr om da y 17 to da y 23 depending on the bird species) of the nestling period. As a proxy of parasitism intensity, we counted the number of blood marks and faeces r emains fr om C. hemapterus flies activity on the body (belly and wings) of each nestling following the protocol described by Tomás et al. ( 2018 ). Mean values per br ood wer e used in the anal yses. We estimated fledging success as the proportion of nestlings that were sampled at the beginning of the nestling period that survived until the end of the nestling period. Volatiles profile: GC-MS GC-MS anal yses wer e performed on a gas c hr omatogr a ph coupled to a mass spectrometer (model Varian 450GC) coupled to a variant 240MS mass spectrometry unit (Palo Alto, CA, USA) with an automatic injector Combi Pal (CTC Analytics) with SPME fibre option. The fibre used was a 50/30- μm D VB/CAR/PDMS , Stableflex 23Ga, for autosampler. Injector desorption was at 250 ◦C for 10 min in split (20:1) and helium flow at 2 ml/min. The capillary column was an Agilent HP-FFAP 30-m X 0.32 mm X 0.25 μm. T he o ven temper atur e was initiated at 50 ◦C for 1 min, and pr ogr ammed to increase 5 ◦C/min to 100 ◦C, then at 10 ◦C/min to 200 ◦C and 50 ◦C/min to 250 ◦C for 1 min. The mass spectrometer worked with EI ionisation (70 eV) and a scan range from 30 to 500 m/z in the TIC full scan mode . T he software used was MS Workstation v. 6.9.1. The identification of compounds was established by selected ion monitoring (SIM) analysis and by using the NIST 08 mass spectr al libr ary and pur e acids and aldehyde standards (FAMQ-004 Volatile Acid Standard Solution, AccuStandard; M-556-MIXA Carbonyl Compounds Mix A, AccuStandard). Because the MS Workstation (version 6.9.1) did not have the option to perform deconvolution or peak alignment, we initially explored the TIC full scan c hr omatogr ams and anal ysed peak by peak to identify potential compounds by comparing the mass spectra obtained (at different parts of the peak) and those in the libr ary (NIST 08). Subsequentl y, the identified peaks were displayed using the c har acteristic ion of that compound (SIM analysis). This strategy allows the identification and quantification of the compounds of interest, avoiding ov erla ps and some problems associated with shifts in retention times . T he analysis of the mass spectra was performed with NIST MS Searc h (v ersion 2.3). An identification is considered correct when the R.match value exceeds 800. The summary of volatile profiles of next-box en vironments , and the characteristic ions of the identified compounds, can be found in the electronic supplementary material ( Table S1 ). As a proxy of volatile abundance, we estimated the r elativ e importance (subarea) in percentage of each compound of the area delimited by all identified compounds. DN A extr action and high-throughput sequencing DNA from nest material was extracted using the MSOP protocol proposed by Martín-Platero et al. ( 2007 ). Briefly, ∼80 mg of solid nest material were suspended in 900 μL of buffer lysis and, after centrifugation, supernatant was collected and stored in 2-ml micr ofuge tubes. Extr acted DNA fr om nest material samples was cleaned using the kit One Step PCR Inhibitor Removal Kit (Zymo Researc h). We also pr ocessed labor atory blanks to detect possible contamination during the process. DNA sequences from nest materials were obtained by amplification of the V6-V8 hypervariable regions ( ∼400-bp fragment) of the 16S rRNA gene. In a first PCR, this fr a gment was amplified using the universal primers B969F (5  -ACGGGCRGTGWGTRCAA-3  ) and BA1406R (5  -ACGGGCRGTGWGTRCAA–3  ). In a second PCR, samples were amplified adding specific barcodes . Afterwards , the libr aries wer e pooled and sequenced in a single run of the Illumina MiSeq sequencer (2 ×300 bp output mode) at the Integr ated Micr obiome Resource (IMR) in the University of Dalhousie (Canada). Sequences ar e av ailable in the Sequence Read Arc hiv e (SRA) in the Genbank - NCBI webpage ( https://www.ncbi.nlm.nih. gov/ sra/ ), BioProject: PRJNA1147916. . Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 4 | FEMS Microbiology Ecology , 2024, Vol. 100, No. 10 Table 1. Number of nests of 10 avian species where we obtained information of the bacterial communities of nest material, the volatile profile of nest boxes and intensity of ectoparasites at the beginning and at the end of the nestling period. For species that were used to build sPLS models (shown in bold), we provide final sample sizes after discarding ASVs that r epr esented < 0.008% of the total ASV counts and a ppear ed in < 4 samples. Nest at early stage of nestlings Nest at late stage of nestlings Avian species Bacteria (nest mat) Volatiles (nest-box) Parasite Bacteria (nest mat) Volatiles (nest-box) Parasite Upupa epops 90 62 115 70 58 112 Restricted (ASVs) 45 (599) 40 (466) Otus scops 10 13 13 11 13 13 Restricted (ASVs) 11 (301) 12 (295) Athene noctua 11 12 11 10 10 10 Restricted (ASVs) 11 (236) 10 (268) Coracias garrulus 7 9 9 7 8 8 Columba oenas 5 4 2 3 4 4 Passer domesticus 8 6 6 6 6 6 Parus major 4 4 4 4 4 3 Coloeus monedula 10 10 12 12 10 13 Restricted (ASVs) 8 (509) 10 (373) Pica pica 0 0 0 13 0 0 Sturnus unicolor 16 16 15 9 11 12 Restricted (ASVs) 14 (525) 8 (217) Raw sequences were analysed using QIIME2 2019.10 (Bolyen et al. 2019 ). Briefly, primers were trimmed and, due to low quality of the r e v erse sequences, subsequent anal yses wer e based onl y on forw ar d sequences. Lo w-quality sequences (Phred < 20) w ere filtered and the Deblur algorithm was employed to produce the amplicon sequence variants (ASVs) table, establishing a sequence size of 220 bp. A phylogenetic tree was built using the fr a gment insertion algorithm (Janssen et al. 2018 ). Taxonomic assignation was performed against the Greengenes13_8 database at 97% similarity (DeSantis et al. 2006 , McDonald et al. 2012 ). Chloroplast, mitochondria and non-phylum assigned ASVs were removed, as well as positive control samples. ASV diversity reached a plateau phase at ∼6000 reads ( Fig. S1 ), and samples with < 6000 reads were excluded fr om downstr eam anal yses (11 samples of nest material and all control samples). The sequencing of nest material produced 22 476 907 sequences, and 7 112 413 wer e r etained in the ASV table after filtering [number of samples = 328; number of sequences per sample: mean: = 21 648.19, SD = 9955.19 (min–max) = (6165–56 018)]. With respect to ASVs, we obtained 14 303; 10 927 and 10 218 for nest samples collected at the beginning and at the end of the nestling period, r espectiv el y. For statistical analyses, we only arbitrarily considered ASVs that represented > 0.008% of the total ASV counts, and that a ppear ed in at least four samples . T hese r estrictions r educed the number of ASVs to 996 and 1111 in samples of nest material collected at the beginning (N = 110) and at the end (N = 96) of the nestling period, r espectiv el y. Statistical procedures Estimating alpha and beta diversity indexes Based on the abundances of ASVs, we calculated alpha (i.e. the micr obial div ersity within a particular sample) and beta diversity (i.e. the variability in community composition among different types of samples) for each of the collected bacterial samples (Whittaker 1972 ). For alpha diversity analyses, we calculated Shannon and Faith’s phylogenetic diversity (PD) indexes in QIIME2. The Shannon index is a quantitative measure of equity that combines the ric hness and e v enness of bacterial species (Shannon 1948 ), while the PD index is a qualitative measure of diversity that takes into account the length of all br anc hes on the bacterial phylogenetic tree within a sample (Faith and Baker 2006 ). For beta diversity analyses of bacterial samples, we calculated Aitchison’s and Phylogenetic Isometric Log Ratio transformed (PhILR) distance matrices for each sample. Aitchison distance matrix (Aitchison et al. 2000 ) was calculated after the centred log r atio (CLR) tr ansformation of ASV abundances in the microbiome 1.18.0 R pac ka ge (Shetty and Lathi 2019 ) . The CLR-tr ansformation controls for the compositional nature of the microbiome dataset and pr oduces v alues that ar e scale inv ariant (Gloor et al. 2017 ). We also estimated the PhILR distance matrix using ASV r elativ e abundances and accounting for the information of the phylogenetic relationship of ASVs (Silverman et al. 2017 ). The PhILR transformations were performed as implemented in the philr 1.22.0 R pac kage (Silv erman et al. 2017 ). For the volatile profiles, we estimated alpha and beta diversity indexes based on the r elativ e abundance of each volatile detected in the nest environment. We estimated the Shannon index for alpha diversity, and because r elativ e abundances of volatiles are also compositional, we estimated beta diversity (i.e. Aitchison distance matrix) after CLR-transformation of r elativ e abundances (for explanations of these indexes and their estimations, see above). Exploring interspecific differences in bacterial community and volatile profile diversity and their associations Interspecific differences in alpha diversity indexes and beta diversity matrices of the bacterial community of nest materials and of volatile profiles of avian nests wer e, r espectiv el y, explor ed in general linear mixed models (GLMMs) and mixed non-parametric m ultiv ariate anal ysis of v ariance (PERMANOVA) with 9999 perm utations. It is worth mentioning here that, for species with information collected during the two study years, we analysed the effect of study year in separate models. In all cases, the variance of values of differ ent div ersity indexes explained by the species identity was significantl y lar ger than that explained by the study year ( Fig. S2 ). T hus , the effect of study year was not included in subsequent statistical models. Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 Mazorra-Alonso et al. | 5 The associations between alpha diversity of volatile profiles of avian nests (dependent variable) and that of bacterial communities of nest material (continuous fixed factor) were explored in GLMMs that included bird species identity as a random factor. The interaction between species ID and bacterial diversity was studied in a separate model that also included main effects. Intraspecific associations between bacterial (predictors) and volatiles (explanatory) alpha diversities were also analysed in general linear models (GLMs), but only for species with > 7 samples and for the diversity indexes describing bacterial communities that r esulted, r elated to volatile diversity, in interaction with species identity. Residuals of all performed GLMMs and GLMs fit normal distributions. To analyse the association between beta diversities of bacterial communities of nest material and of volatile profiles of avian nests, we used m ultiv ariate Mantel tests with 9999 perm utations. These tests are equivalent to multiple regression tests but using matrices of differences among samples instead of linear information. The models included beta diversity of volatiles (based on Aitchison distances) as the dependent matrix, and beta diversity of bacterial communities (based on Aitchison or PhILR distances) and a matrix of bird species identity [whether each cell in the matrix belonged (cell value = 1) or not (cell value = 0) to the same bird species] as independent matrices. In this case, we also explored intraspecific associations in species with > 7 samples and for the index that resulted related to volatiles diversity. All GLMMs were performed in ST A TISTICA 12.0, PERMANOVAs in PRIMER-7.0.17 and Mantel tests in the R.3.6.1 (https// www.rpr oject.or g/) envir onment [function MRM in the ecodist pac ka ge (Goslee and Urban 2007 )]. To infer potential relationships between specific bacterial strains and specific v olatiles, w e integrated information of relative abundances (i.e. CLR-transformed) of the considered ASVs and v olatiles, using spar ce partial least squar es or pr ojection to latent structures (sPLS) models with a leave-one-out cross-validation method as implemented in the pac ka ge mixOmics v. 6.19.9 (Rohart et al. 2017 ). sPLS has been used as an alternative method to ordinary least squares for handling multicollinearity (Chun and Kele ¸s 2010 ). We used the regression mode of sPLS, which fits a linear relationship between multiple responses in Y and multiple predictors in X. The microbial data are being attempted to be used to explain the volatile profiles . T he primary difference between sPLS and other dimension-reduction techniques (i.e . PC A) is that it maximises the covariance between latent variables (also termed latent components) rather than correlation (Burguillo- Muñoz 2015 ). At the interspecific le v el, we performed separ ate sPLS anal yses for samples collected during the early stage and at the end of the nestling period. The sPLS models are quite sensitive to noise associated with low sample sizes (Arumugam et al. 2011 ) and, thus, we onl y arbitr aril y consider ed ASVs that r epr esented > 0. 8% of the total ASV counts, and that a ppear ed in at least four samples (see abov e). To gr a phicall y show the corr elation structur e (i.e. the network) among the latent components from sPLS models, we used the network function and the igraph pac ka ge v. 1.3.4 (Csárdi and Nepusz 2006 ). That function uses detected correlations between pr edictor and r esponse latent components of the sPLS model to draw a network among them. We only represented elements from the first latent component of the r esponse v ariables (volatiles) becaues no other v ariable sur passed the established minimum value of q 2 (0.095) for validation of latent components (Wold et al. 2001 ). Mor eov er, the Pearson corr elation v alue between the latent components was different for each sampling moment (i.e. at the early and at the late stage of the nestling period), and we used those within the range determined by the maximum and 6% less than that value [(Max–6 ∗Max/100)—(Max)]. Then we exported and visualised the resulting network in Cytoscape v. 3.9.9 (Shannon et al. 2003 ). At the intraspecific level, we performed separate sPLS models for each species with > 7 samples collected either at the beginning or at the end of the nestling period. In this case, we only used ASVs that r epr esented > 0.01% of the total counts and a ppear ed in at least four samples. Sample sizes and the number of considered ASVs after sieving information of each species are shown in Table 1 . Exploring predictors of intensity of parasitism and avian fledging success The relationships between intensity of par asitism (squar e-r oot tr ansformed v alues) or fledgling success and div ersities of volatiles and bacterial communities were explored in GLMMs and Mantel tests for alpha and beta div ersities, r espectiv el y. The GLMMs included bacterial and volatile alpha diversities as continuous fixed factors, and bird species identity as a random factor. Mor eov er, the inter actions between species identity and bacterial alpha diversities, and between species identity and v olatile alpha diversities, w ere also studied in two separate models that also included the main effects. Mantel tests included matrices of differences in the intensity of parasitism and fledging success between samples as dependent matrices and the beta diversity of volatiles and bacteria as independent factors . T he models also included a matrix with information of the species identity (binary matrix: 1 = equal, 0 = different species) as an additional independent factor. For exploring the associations between particular volatiles or bacteria and the intensity of parasitism or fledging success, we only used those elements from sPLS models with correlation coefficients within the considered range {[(Max–6 ∗Max/100)–(Max)]}. The associations were explored in GLMMs. To avoid problems of collinearity, we explored the effects of bacteria and volatiles in separate GLMMs . T hese models included species identity as a random factor and bacterial or volatile elements as continuous fixed factors . Moreo ver, in addition to the full models, we used Akaike information criterion (AIC) to look for the best model, choosing the one with the lo w est AIC value that included the lo w est number of independent factors. When species identity was included in the final reduced model, its effect was modelled as a random effect. The intraspecific associations between parasitism intensity or fledging success and particular bacteria or volatiles from sPLS models were also explored. Results Bacterial communities and volatile profiles of avian nests Alpha and beta diversities of nest bacterial communities and of volatile profiles differed among species during both the early and late nestling stages ( Table S2 ). Alpha diversity of the nest bacterial community explained the alpha diversity of volatiles of nest boxes during both early (Shannon diversity) and late (Faith’s PD) nestling stages, but only in interaction with species identity (Table 2 ). At the early stage, the interaction was mainly due to the positive and negative associations that were detected in Sturnus unicolor (Beta = 0.56, GLM: F 1,12 = 5.35, P = 0.039) and Athene noctua (Beta = –0.72, GLM: F 1,9 = 9.39, P = 0.013), r espectiv el y (see Fig. 1 and Table S3 for Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 6 | FEMS Microbiology Ecology , 2024, Vol. 100, No. 10 Ta ble 2. Relationships betw een alpha and beta diversities of v olatiles and bacterial comm unities of the nest envir onment [i.e. air inside the nest box (volatiles) and nest material (bacteria)] during the early and the late nestling sta ges. To c har acterise the alpha and beta diversity of v olatiles, w e used Shannon and Aitchison indexes, respectively, whereas Shannon or Faith’s Phylogenetic Diversity (PD) (F1), and Aitchison or Phylogenetic-Isometric-Log-Ratio-transformed (PhILR) distances, were used to characterise alpha and beta bacterial div ersity, r espectiv el y. In gener al linear mixed models (GLMMs), information of species identity (Sp ID) was included as a random effect (R) and as binary matrix (1 = equal, 0 = different species) in Mantel tests. In GLMMs, the interaction between bird-species identity and bacteria diversity was tested in separate models that also included main effects, while the main effects were explored in models that did not include interactions. We show partial beta values and associated P -values. Variables in bold are those with associated P -values < 0.05. ALPHA DIVERSITY Shannon PD F df P Beta F df P Beta Early nestling stage Sp ID (R) 9 .24 8,100 < 0 .001 10 .92 8,100 < 0 .001 Bacteria alpha diversity (F1) 0 .00 1,100 0 .952 0 .005 2 .91 1,100 0 .091 0 .143 F1 ∗Sp ID 2 .53 8,92 0 .016 0 .79 8,2 0 .616 Late nestling stage Sp ID (R) 8 .70 7,87 < 0 .001 8 .10 7,87 < 0 .001 Bacteria alpha diversity (F1) 0 .13 1,87 0 .719 0 .036 0 .29 1,87 0 .594 0 .063 F1 ∗Sp ID 1 .49 7,80 0 .181 3 .31 7,80 0 .004 BETA DIVERSITY Aitchison PhILR R R Early nestling stage Sp ID < 0 .001 0 .659 < 0 .001 0 .647 Bacteria beta diversity 0 .644 −0 .002 0 .730 −0 .003 Late nestling stage Sp ID < 0 .001 0 .931 < 0 .001 0 .709 Bacteria beta di v ersity 0 .094 −0 .009 0 .014 0 .032 r esults of intr aspecific associations). When considering samples from the late stage of the nestling period, the interaction was mainly due to the non-significant positive and negative associations detected in Coleus monedula (Beta = 0.23, GLM: F 1,8 = 0.43, P = 0.53) and Otus scops (Beta = -0.33, GLM: F 1,10 = 1.19, P = 0.30), respectiv el y (see Fig. 1 and the results of intraspecific associations in Table S3 ). Beta diversity of bacteria of nest material at the early nestling stage did not predict beta diversity of volatiles of nest environments after controlling for the significant effect of species identity (Table 2 ). The association did not r eac h statistical significance in any of the five species analysed, regardless of the diversity index used to build distance matrices (see results in Table S3 ). Ho w e v er, at the end of the nestling stage, beta diversity (i.e. PhILR matrix) of bacterial community of nest material predicted volatile profiles of avian nests e v en after contr olling for the statisticall y significant effect of species identity (Table 2 ). Within species, the positive association r eac hed statistical significance in two of the fiv e species analysed [i.e. Athene noctua (PhILR: Mantel tests, R 2 = 0.40, R = 0.30, F = 28.86, P < 0.002) and Otus scops (Aitchison: Mantel tests, R 2 = 0.16, R = 0.04 , F = 12.44, P < 0.001); see the intraspecific results in Table 3 ]. Optimisation of the sPLS models of samples from the early and late stages of the nestling period resulted in a single latent component in both cases . T he latent components from samples of the early nestling period explained 8.52% and 12.49% of the variance of the abundance of bacteria and volatiles, r espectiv el y. The r esulting latent components included five bacterial taxa [the taxonomic classification of detected ASVs is shown in Table S4 , and hereafter we will directly refer to the taxonomic classification (i.e. genus) of each ASV] and one volatile (heptanal). Similarly, the latent components of samples of the late nestling period explained 9.27% and 11.71% of the variance of the abundance of bacteria and volatiles, r espectiv el y, and included four bacterial taxa and one volatile (heptanal). The identity of the bacterial taxa and the strength of the associations with heptanal in the early and late nestling sta ges ar e shown in Fig. 2 . At the intr aspecific le v el, the identity of bacteria and volatiles that resulted as latent components (i.e. those related to each other) after the optimisation of the sPLS models depended on the species considered and the stage of nestling de v elopment ( Table S5 ). Interspecific and intraspecific covariation in parasitism and bacterial and volatile profiles of avian nests Alpha and beta diversities The intensity of parasitism that nestling suffered during the early sta ge of de v elopment was negativ el y r elated with the alpha div ersity of nest volatiles, after controlling for the statistically significant effects of species identity and the non-significant effects of the alpha diversity of the bacterial community of the nest materials (Table 3 ). Inter estingl y, that association did not depend on the species identity (see the interaction between the alpha diversity of volatiles and species identity in Table 3 ). When considering beta diversities, neither volatiles nor bacteria of avian nests predicted intensity of parasitism suffered by the nestlings during the early stage of the nestling period after controlling for the significant effect of species identity (Table 4 ). The intensity of parasitism of nestlings during the late stage of de v elopment was negativ el y r elated to the alpha diversity of the bacterial community of nest materials after controlling for the effect of species identity and the non-significant effect of alpha div ersity v alues of nest volatiles (Table 3 ). The detected association did not vary interspecifically (see the interaction between the alpha diversity of volatiles and species Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 Mazorra-Alonso et al. | 7 F igure 1. Associations betw een alpha diversity of v olatile pr ofiles and the bacterial comm unities of the nests of differ ent a vian species . Shannon indexes of volatile profiles were obtained at early and late stages of nestlings. Bacterial diversities were described at the early stage by the Shannon Index, and at the late stage by the Faith-PD Index. Bold and dotted lines are regression lines that, respectively, did or did not reach statistical significance. F igure 2. Netw ork r epr esentation of sPLS performed on bacteria and volatile pr ofiles data fr om the nest (A) at the earl y sta ge and (B) at the late sta ge of nestlings, from 10 bird species. Both networks are bipartite, where each edge links an ASV to a volatile node, according to a similarity matrix. The correlations shown are between the maximum of the Pearson coefficient correlation found and 6% less than that maximum. The degree of green colour of the edges r epr esents coefficients from the lo w est (light green colour) to the highest (dark green colour). No correlation is less than 0.60. The ASVs are named with the genus or family to which they belong to and the references are shown in the ASV table in the supplementary material. The asterisk ( ∗) indicates the family of bacteria for an unknown species, as identification at genus level was not possible. Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 8 | FEMS Microbiology Ecology , 2024, Vol. 100, No. 10 Ta ble 3. Relationships betw een intensity of parasitism or fledging success and alpha diversities of volatiles and bacterial communities of nest en vironments , both at the early and the late de v elopmental sta ge of nestlings. Diversity of volatiles was expressed with the Shannon index, while for bacteria we used the Shannon and Faith’s phylogenetic diversity (PD) index. Species identity (Sp ID) was included in a GLMM as a random (R) factor, and alpha diversities of bacterial communities and volatile profiles as first (F1) and second (F2) fixed factors, r espectiv el y. For continuous fixed factors, we show partial beta values . T he interactions betw een bir d-species identity and bacteria diversity or volatile profiles were tested in separate models that also included main effects, while the main effects were explored in models that did not include interactions. Variables in bold font are statistically significant at the 5% level. Shannon alpha di v ersity PD alpha di v ersity F df P Beta F df P Beta PARASITISM INTENSITY Early nestling stage Sp ID (R) 11 .72 7,93 < 0 .001 11 .54 7,93 < 0 .001 Bacteria alpha diversity (F1) 0 .20 1,93 0 .659 −0 .037 0 .31 1,93 0 .574 −0 .048 Volatile alpha di v ersity (F2) 4 .74 1,93 0 .032 −0 .221 4 .33 1,93 0 .040 −0 .213 F1 ∗Sp ID 0 .96 7,86 0 .461 0 .30 7,86 0 .949 F2 ∗Sp ID 1 .76 7,86 0 .106 1 .67 7,86 0 .127 Late nestling stage Species ID (R) 18 .90 6,79 < 0 .001 18 .40 6,79 < 0 .001 Bacteria alpha di v ersity (F1) 4 .18 1,79 0 .044 −0 .154 6 .25 1,79 0 .015 −0 .217 Volatile alpha diversity (F2) 2 .45 1,79 0 .121 −0 .139 2 .47 1,78 0 .120 −0 .138 F1 ∗Sp ID 0 .74 6,73 0 .619 0 .97 6,73 0 .451 F2 ∗Sp ID 1 .64 6,73 0 .148 1 .58 6,73 0 .165 FLEDGING SUCCESS Early nestling stage Sp ID (R) 0 .45 6,81 0 .842 0 .71 6,81 0 .642 Bacteria alpha di v ersity (F1) 0 .03 1,81 0 .867 −0 .019 5 .68 1,81 0 .019 −0 .268 Volatile alpha diversity (F2 ) 0 .70 1,81 0 .791 −0 .038 0 .00 1,81 0 .994 0 .001 F1 ∗Sp ID 0 .98 6,75 0 .445 1 .02 6,75 0 .419 F2 ∗Sp ID 0 .64 6,75 0 .638 0 .69 6,75 0 .657 Late nestling stage Species ID (R) 0 .94 6,79 0 .470 0 .97 6,79 0 .449 Bacteria alpha diversity (F1) 0 .25 1,79 0 .615 0 .061 0 .18 1,79 0 .676 −0 .059 Volatile alpha diversity (F2) 0 .83 1,79 0 .365 0 .131 0 .89 1,79 0 .350 0 .135 F1 ∗Sp ID 0 .82 6,73 0 .561 1 .75 6,73 0 .121 F2 ∗Sp ID 1 .14 6,72 0 .221 1 .69 6,73 0 .135 Ta ble 4. Relationships betw een matrices of differences in intensity of parasitism and fledging success and beta diversity of volatile profiles and bacterial communities of nest en vironments , both during the early and the late de v elopmental sta ge of nestlings. Beta diversity of volatiles was estimated as Aitchison’s distances, while that of bacterial beta diversity was estimated by Aitchison or Phylogenetic- Isometric-Log-Ratio-transformed (PhILR) distances. In multivariate Mantel tests, the partial effects of volatile beta diversity were estimated in models that included bacteria beta diversity estimated by either Aitchison or PhILR distances . T he models also included a matrix with information of the species identity (Sp ID) [binary matrix (1 = equal, 0 = different species)]. For all independent matrices [species identity (Sp ID) and bacterial and volatile beta diversities], we show partial correlation coefficients and associated P -values after 9999 permutations. Variables in bold are statistically significant at the 5% level. Intensity of parasitism Fledging success Aitchison PhILR Aitchison PhILR R P R P R P R P Early nestling stage Sp ID 0 .570 < 0 .001 0 .589 < 0 .001 −2 .771 0 .249 −1491 0 .527 Bacteria beta diversity 0 .004 0 .625 0 .010 0 .611 0 .177 0 .176 0 .144 0 .626 Volatile beta diversity −0 .701 0 .437 −0 .071 0 .421 −0 .583 0 .648 −0 .587 0 .653 Late nestling stage Sp ID 1 .805 < 0 .001 1 .622 < 0 .001 −1 .632 0 .519 −1 .795 0 .475 Bacteria beta diversity −0 .010 0 .342 0 .007 0 .761 0 .006 0 .959 0 .105 0 .706 Volatile beta di v ersity 0 .261 0 .006 0 .265 0 .008 0 .407 0 .683 0 .361 0 .716 identity in Table 3 ). Beta diversity of volatiles, but not that of bacteria from the nest environment, was associated positively with the intensity of parasitism that nestlings suffer ed, e v en after controlling for the significant effect of species identity (Table 4 ). Particular bacteria and volatiles from latent components from sPLS models Final models explaining the intensity of parasitism of nestlings at the beginning and at the end of the nestling period included several bacterial taxa, but heptanal, the only volatile in latent compo- Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 Mazorra-Alonso et al. | 9 nents of sPLS models, did not enter into any of those final models (Table 5 ). The abundance of the class Thermoleophilia was associated positiv el y with intensity of par asitism at the beginning of the nestling period, while P orph yromonas sp . w as associated negativ el y (Table 5 , Fig. 3 ). Intensity of parasitism during the late nestling period was positiv el y associated with the abundance of Polaromonas sp. (Table 5 , Fig. 3 ; see full models in Table S6 ). These associations were in all cases found after controlling for the significant effect of species identity. The subset of bacteria in the final models explained a significant proportion of the residual variance of the intensity of parasitism (see statistics of bacterial effects in Table 5 ). At the intraspecific level, we also found evidence of significant associations between parasitism intensity and the r elativ e abundance of particular bacteria or volatiles (i.e. little owls and hoopoes; Table S7 ). Ho w e v er, except for hoopoes, sample sizes were quite small for reliable statistical inferences . In hoopoes , the r elativ e abundance of Polaromonas sp.1 (Beta = 0.35, F 1,43 = 6.09, P = 0.018; Fig. 4 ) and the concentration of acetic acid (Beta = 0.33, F 1,43 = 5.23, P = 0.027; Fig. 4 ) at the early nestling stage were positiv el y r elated with the intensity of par asitism of younger nestlings. The intensity of parasitism of older nestlings was positiv el y r elated with the abundance of decanoic acid (Beta = 0.44, F 1,37 = 9.02, P = 0.005; Fig. 4 ) and sho w ed a non-significant tendenc y with the abundance of Sporosarcina sp.1 (Beta = 0.31, F 1,35 = 3.79, P = 0.059; Fig. 4 ). Interspecific and intraspecific variation in fledging success and in bacterial and volatile profiles of avian nests Alpha and beta diversities Fledging success was negativ el y r elated with alpha diversity of the bacterial community of the nest materials during the early, but not during the late, stage of nestlings (Table 3 ). The detected association did not v ary interspecificall y (see the interaction between alpha diversity of volatiles and species identity in Table 3 ). Beta diversities of bacterial communities and volatile profiles estimated during both nestling stages did not explain fledging success (Tables 3 and 4 ). All the results were controlled for the non-significant effects of species identity (Tables 3 and 4 ). Particular bacteria and volatiles from latent components of sPLS models Final models explaining fledging success included both volatile and bacterial components (Table 5 ). The abundance of Corynebacterium sp. (negativ el y) and P eptoniphilus sp . (positiv el y) during the earl y sta ge of the nestling period and that of Polaromonas sp. (negativ el y) during the late stage of the nestling period were associated with fledging success (Table 5 , Fig. 5 ; see full models in Table S6 ). The abundance of heptanal during both stages of the nestling period tended to be negativ el y associated with fledging success (Table 5 , Fig. 5 ; see full models in Table S6 ). At the intraspecific level, neither the volatiles nor bacteria that were selected in sPLS models were associated significantly with fledging success ( Table S7 ). Discussion The hypothesis tested here is 2-fold: first, volatiles from bacterial symbionts are partially responsible for the nest odours; and second, these micr oor ganisms partiall y determine the pr obability of par asitism and r epr oductiv e success of their bird hosts. To explore these hypotheses, we took adv anta ge of the avian nest environments, which allo w ed characterising the bacterial environment, the volatile profiles, the intensity of parasitism and the r epr oductive success in the same focal location. Moreover, we used nests of differ ent species, and explor ed the expected associations both interspecifically and intraspecifically. Our main findings were that (i) alpha and beta diversity of bacteria predicted the volatile profile of the nest, although the strength and the sign of the association varied interspecifically and depended on the nestling stage (i.e. early vs late) and the index used to estimate the diversity. Although the strength of the associations depended on the diversity index used, (ii) alpha and beta diversities of volatiles and/or nest bacterial comm unity pr edicted the par asitism intensity of nestlings either at the beginning or at the end of the nestling period. Inter estingl y, (iii) these associations did not depend on the species identity. Furthermore, (iv) particular bacteria and volatiles responsible for the associations found between bacterial and volatile en vironments , and between them and parasitism and fledging success, varied interspecifically and depended on the nestling stage. Belo w, w e discuss particularities of these results and their importance for the hypotheses tested. Bacterial communities and volatile profiles of avian nests Interspecific differences in nest c har acteristics may explain the detected species-specific differences in the bacterial communities of nest materials, in the volatile profiles of the nest environment and in the associations between these two nest components. Avian nests vary not only in the le v el of isolation from the external environment (e.g. hole vs open nest), location (e.g. on the soil, trees or cliffs) and orientation, but also in the nest structure and the material used for lining the nests (Hassel 2007 ). These c har acteristics will dir ectl y affect temper atur e and humidity (Mertens 1977 ), which could also determine the bacterial communities of nest environments (Godard et al. 2007 , Peralta- Sánchez et al. 2012 , Soler et al. 2015 , Goodenough et al. 2017 , Martínez-Renau et al. 2022 ). Mor eov er, some of the studied species do not build nests (owls, hoopoes and rollers), while others use aromatic plants with antimicrobial properties (Clark and Mason 1985 , Mennerat et al. 2009 ) and/or feathers that favour the growth of particular antibiotic-producing bacteria (Peralta-Sánchez et al. 2014 , Ruiz-Castellano et al. 2019 ), which could also modulate bacterial communities of the nest environment. Our results, ho w ever, ar e corr elational and, thus, do not allow distinguishing particular causes explaining detected interspecific differences or covariations. Experimental manipulations of some of the nest materials or of the bacterial communities of the nest environment would help to r eac h further conclusions on the r easons determining the associations detected. The species-specific nest c har acteristics will also determine the volatile profiles of avian nests, either because of the direct emission of volatiles of some nest-lining components (i.e. aromatic plants), or because of volatiles from the metabolism of birds or the species-specific bacterial communities of avian nests. Thus, the expected associations between c har acteristics of the bacterial communities and volatile profiles should vary interspecifically; a prediction supported by the detected interaction between bacterial community and species identity explaining the volatile profiles of avian nests. That interaction is exemplified by the positive association between alpha diversities of volatiles and bacteria detected in the spotless starling, a species that uses nest-lining materials with antimicrobial properties (Ruiz-Castellano et al. 2016 , 2018 , 2019 ). This correlation became negative in the little owl, a species that does not use nest-lining material in their nests. In Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025 16 | FEMS Microbiology Ecology , 2024, Vol. 100, No. 10 West A , Cassey P, Thomas CM. Microbiology of nests and eggs. In: Deeming DC, Reynolds SE (eds), Nest, Eggs, & Incubation New Ideas about avian Reproduction. Oxfor d, UK: Oxfor d University Press, 2015, pp. 75–81. Whittaker DJ , Slowinski SP, Gr eenber g JM et al. Experimental evidence that symbiotic bacteria produce chemical cues in a songbird. J Exp Biol 2019; 222 :jeb202978. Whittaker RH . Evolution and measurement of species diversity. Taxon 1972; 21 :213–51. Windsor RL , Fegely JL, Ardia DR. The effects of nest size and insulation on thermal properties of tree sw allo w nests. J Avian Biol 2013; 44 :305–10. Wold S , Sjöström M, Eriksson L. PLS-r egr ession: a basic tool of chemometrics. Chemom Intell Lab Syst 2001; 58 :109–30. Recei v ed 15 February 2024; revised 22 May 2024; accepted 23 July 2024 ©The Author(s) 2024. Published by Oxford Uni v ersity Pr ess on behalf of FEMS. This is an Open Access article distributed under the terms of the Cr eati v e Commons Attribution-NonCommercial-NoDerivs licence ( https://creativecommons.org/licenses/bync- nd/4.0/ ), which permits non-commercial r e pr oduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact [email protected] Downloaded from https://academic.oup.com/femsec/article/100/10/fiae106/7720530 by guest on 10 March 2025