Introduction
Psychedelic compounds are widespread in nature, occurring across plants [e.g., mescaline found in the Cactaceae (El‐Seedi et al. 2005)], animals (e.g., bufotenine found in anurans; Orsolini et al. 2018), and fungal kingdoms (ergot alkaloids in Claviceps; Matuschek et al. 2011). In humans, ingestion of serotonergic psychedelics induces profound alterations in perception, cognition, and self‐referential processing, characterised by visual and auditory hallucinations, altered sense of time, and ego‐dissolution (Vollenweider and Kometer 2010). This experience has recently been shown to have great potential for the treatment of post‐traumatic stress disorder, treatment‐resistant depression, and obsessive‐compulsive disorders (Ross et al. 2016; Carhart‐Harris et al. 2021; Barrett et al. 2020; Doss et al. 2021), triggering a resurgence of interest and excitement around psychedelic research. Despite this applied importance, cultural value, and widespread distribution, very little is understood about the evolution of psychedelic compounds in nature.
The most widespread and clinically significant natural psychedelic prodrug is psilocybin (Guzmán 2019), the compound found in ’magic’ mushrooms, most notably within the agaricomycete genus Psilocybe. In fact, distantly related but ecologically similar species across eight genera in the Agaricales (Psilocybe, Inocybe, Pluteus, Panaeolus, Conocybe, Pholiotina, Galerina and Gymnopilus) produce psilocybin (Guzmán 2019). Like other psychedelics, its psychotropic effects arise because it is an analog of 5‐hydroxytryptamine (serotonin) that interacts with 5‐HT2A receptors (D. E. Nichols 2016; Carhart‐Harris and Nutt 2017), a mechanism considered necessary for the induction of psychedelic effects (Xu et al. 2026). Moreover, several psychotropic but non‐psychedelic fungi produce neuroactive compounds that interact with animal nervous systems. For example, fly agaric (Amanita muscaria) and panther cap (A. pantherina) synthesise ibotenic acid and muscimol, which act as agonists at GABA receptors (Michelot and Melendez‐Howell 2003), and have demonstrable effects on invertebrate locomotion (Mustard et al. 2020), whereas muscarine, widespread across genera such as Clitocybe sensu lato, Omphalotus, Mycena, and Inocybe, targets the parasympathetic nervous system (Kosentka et al. 2013). However, why fungi should produce psychotropic compounds that interact with animal nervous systems is an intriguing and unresolved question.
Molecular evidence strongly supports the claim that psilocybin biosynthesis was shaped by natural selection. Phylogenomic analyses indicate that psilocybin biosynthesis originated in Psilocybe approximately 67mya and diversified around 56 mya, with 4–5 instances of horizontal gene transfer (HGT) to distantly related fungi (Bradshaw et al. 2024). The strong conservation of gene content and order across these lineages further indicates purifying selection acting on the biosynthetic gene cluster (BGC) (Bradshaw et al. 2024). Additionally, the diversification of psilocybin‐producing species across these lineages coincides with the expansion of dung and wood decay ecological niches (Reynolds et al. 2018), where invertebrates are also abundant (Rouland‐Lefèvre 2000), highlighting the potential for complex dynamics of mutual exploitation and antagonism between fungi and invertebrates. Moreover, recent genomic evidence demonstrates that psilocybin biosynthesis evolved independently in mycorrhizal Inocybe via distinct biosynthetic pathways (Schäfer et al. 2025), a process unlikely to occur and be maintained without selective pressure. The combination of horizontal transfer, conserved gene architecture, and repeated independent origins suggests that psilocybin biosynthesis is unlikely to represent a stochastic metabolic by‐product, but instead reflects recurrent positive selection. In line with this, and drawing on related research on plant secondary metabolites (Moreira et al. 2016; Richards et al. 2015; War et al. 2012), an obvious hypothesis for the evolution and maintenance of psilocybin biosynthesis by selection is defence and resistance from ingestion by invertebrates (Meyer and Slot 2023; Reynolds et al. 2018; Spiteller 2008). Indeed, fungal fruiting bodies are frequently attacked by mycophagous insects (e.g., sciarid and mycetophilid larvae) and gastropods (Santamaria et al. 2023). In both plants and ascomycetes, BGCs commonly produce secondary metabolites that deter insect herbivores (Dellapenna and O'Connor 2012; Lichman et al. 2020; Takos and Rook 2012). Furthermore, given that the serotonergic system is evolutionarily ancient and conserved across most bilaterian animals (Bacqué‐Cazenave et al. 2020), psilocybin's structural similarity to serotonin makes it plausible that interaction with invertebrate serotonergic receptors could impair physiological and behavioural processes in invertebrates including feeding, digestion, circadian rhythm, regulation of small muscle contraction and aggression (Bacqué‐Cazenave et al. 2020; Tierney 2020), providing a mechanical basis for a defensive role. Another possibility is that psilocybin and similar compounds arise through neutral processes. For example, the Firn‐Jones screening hypothesis (Firn and Jones 2003) suggests that rather than individual metabolites, selection favours the evolution of pathways for the synthesis of a broad array of secondary metabolites at low cost. However, once an active compound has evolved, evolutionary logic would predict its maintenance by selection, especially when it is costly to produce, as is true for psilocybin because of high nitrogen content (Fricke et al. 2017).
Important considerations arise when psilocybin‐producing fungi are viewed within an ecological framework. Firstly, psilocin, the active metabolite of psilocybin, exhibits a broad receptor binding profile, with affinity for multiple serotonergic subtypes, including those that are known to be conserved within invertebrates (5HT1A, 5HT1B, 5HT2A, 5HT2B and 5HT7) (Gasque et al. 2013; Johnson et al. 2009) as well as D1 and D3 dopaminergic receptors (Ray 2010), indicating the mechanistic action of psilocybin/psilocin is unlikely to be mediated exclusively through 5‐HT2A receptors alone. Secondly, psilocybin‐producing fungi also produce a suite of structurally related tryptophan derivatives. This includes norbaeocystin, baeocystin, norpsilocin, and aeruginascin (Fricke et al. 2017), several of which exhibit agonist activity at 5‐HT2A receptors (Glatfelter et al. 2022). In addition, β‐carbolines have been detected in some psilocybin‐producing fungi, which may slow the oxidative degradation of psilocin, thus impacting the onset and duration of effects (Blei et al. 2020). Supporting this, experimental studies have shown mice treated with crude mushroom extracts exhibit a stronger head‐twitch response, a behavioural proxy for 5‐HT2A activation, than those given equivalent doses of synthetic psilocybin (Zhuk et al. 2015). Complementing these findings, recent work has shown multiple mushroom‐derived compounds are involved in multi‐target interactions consistent with an “entourage effect” (Murray et al. 2026). These findings support the possibility of additive or synergistic interactions among these compounds, in which multiple “psiloids” may contribute to overall bioactivity. They also highlight a gap in current research, which has largely focused on the pharmacology of pure or synthetic psilocybin. This emphasises the need for ecologically relevant studies using whole organisms to better understand the evolutionary and ecological roles of psilocybin‐producing fungi.
In general, the inference of selective benefits of fungal secondary metabolites is challenging. To truly demonstrate a selective effect for defence, reciprocal benefit for the producer, as well as impact on a hypothesised target, must be demonstrated (Rohlfs 2015; Biedermann 2019), and to do so, appropriate controls are necessary. One way that these could be developed is by using transgenic tools, as has been done in Ascomycota (Brakhage and Schroeckh 2011; Magan and Aldred 2007; Rohlfs et al. 2007). For example, disruption of the LaeA gene, a global regulator of secondary metabolites in filamentous fungi, restores Drosophila larval development in A. nidulans and simultaneously results in a significant loss in resistance to fungivory (Caballero Ortiz et al. 2013). Further, the application of methods like CRISPR/Cas9 in basidiomycetes remains limited, and the dominance of non‐homologous end joining (NHEJ) further impedes efficient targeted genome editing (Tu et al. 2021), so mutant fungi with secondary metabolite biosynthesis pathways knocked out are hard to produce. Finally, challenges remain in understanding the evolution of fungal chemical diversity because (1) Secondary Metabolite (SM) production is ecology driven, that is, dependent on multiple abiotic and biotic factors; (2) the difficulty in activating SM biosynthesis means the full metabolic repertoire often remains silent or undetectable in standard laboratory settings (Brakhage and Schroeckh 2011); and (3) genetic models are lacking. Consequently, secondary metabolite evolution remains an intriguing frontier in fungal biology and chemical ecology.
In summary, although a range of fungal factors influence invertebrate communities, including decay stage, habitat, fruit body size, toughness, surface area, and morphology (Lunde et al. 2022; Koskinen et al. 2022; Yamashita and Hijii 2003). There has been little exploration of soft‐bodied fungi, and current research fails to address the importance of SM profiles in driving invertebrate community composition. Drawing from this ecological framework, we hypothesise that fungal SMs, such as psilocybin, may exert similar selective pressures and lead to detectable shifts in the composition or diversity of arthropods associated with fruiting bodies. In this study, as a first step in advancing eco‐evolutionary research of fungal psychedelics, the effects of broad‐scale extracts from Psilocybe mushrooms are examined. We test the hypothesis that Psilocybe extracts will have a negative effect on (a) survival, (b) development, and/or (c) locomotion of Drosophila sp. and that 5‐HT2A mutant flies will be insensitive to treatment with psilocybin. We also predict that fruiting bodies of field‐collected Psilocybe sp. will have a depauperate and distinct arthropod community compared to other grassland fungi.
Materials and Methods
Extract Preparation
Fruiting bodies of Psilocybe cubensis were cultivated as described in the Supporting Information, ‘Psilocybe cubensis cultivation’. Psilocybin extracts for locomotion experiments (hereafter referred to as ‘P. cubensis extracts’) were prepared following a standard methanolic extraction (Gotvaldová et al. 2021), using 1 mL of 0.5% (v/v) acetic acid in methanol for every 10 mg of air‐dried Psilocybe cubensis powder. Extracts were then assessed for extraction efficiency and quantified for key tryptamines. Protocol details and results of the quantitative analysis and stability testing for psilocybin, psilocin, norbaeocystin, aeruginascin, baeocystin, and tryptophan are provided in the Supporting Information (‘Quantitative LC‐MS QQQ Analysis’, Table S1, ‘Stability testing’, Figure S2, and Figure S3).
Maintenance of Fly Stocks and Collection of Larvae
Fly stocks, D. melanogaster, D. affinis and D. melanogaster 5‐HT2A mutant (Häcker et al. 2003; C. D. Nichols 2007), were maintained on conventional agar, sugar, yeast, cornmeal and preservative (nipagin) at 21°C under a 12:12 L:D. To obtain experimental cohorts of flies, 10–14 day old adults were transferred on day ‘0’ to oviposition vials with yeast powder to encourage laying (equal ratio male: female). Adults were left in vials for 6 h in the dark and then transferred to a new set of oviposition vials or discarded. After transfer of the adult flies, vials were kept in the incubator for 24 or 96 h and used to collect first instar (L1) or 3rd instar (L3) for survival and locomotion assays, respectively. All larvae were washed in PBS before use in experiments.
Survival and Development of Flies Exposed to Psilocybe Extracts
Preliminary trials revealed experimental lines of D. affinis and 5‐HT2A mutants were challenging to cultivate under laboratory conditions and exhibited high mortality when transferred to non‐standard food sources containing methanolic extracts. Moreover, the use of methanolic P. cubensis extracts presented additional methodological challenges, including poor homogenisation with standard fly food, the heat sensitivity of psilocybin requiring addition post‐cooling (Gotvaldová et al. 2021), and potential adverse effects of the solvent (Mellerick and Liu 2004). Informed by these challenges, survival assays were performed for D. melanogaster only, using P. cubensis powder.
Experimental tubes were prepared with 250 μL cornmeal‐agar‐molasses medium and either 5 mg or 10 mg of P. cubensis mushroom powder per tube (equivalent to 0.1 μg/μL and 0.2 μg/μL, respectively) before storing at 4°C overnight. This dose was selected as a conservative, yet field‐realistic dose on the basis of the lower concentrations of psilocybin found in Psilocybe semilanceata in nature (0.17 μg/mg) (Christiansen et al. 1981). Two controls were used: 10 mg Agaricus bisporus powder (mushroom control) and a true control consisting of standard Drosophila food medium. On experimental day zero, five first instar D. melanogaster larvae were placed in an individual tube assigned to one of four treatments (0.1 μg/μL, 0.2 μg/μL, mushroom control, true control), and tubes were kept under controlled culturing conditions (see 2(b) Maintenance of fly stocks and collection of larvae). Survival was measured by daily scoring of the number of pupae and a number of adults. Survival data across 20 biological replicates (20 tubes of five larvae per tube) in each of the four treatments were collected over two experimental blocks (N = 160). Survival assays were conducted prior to locomotion experiments and were designed to approximate ecologically relevant exposure levels; therefore, lower‐dose survival treatments were not subsequently performed.
Locomotion of Flies Exposed to Psilocybe Extracts
The concentration used to assess locomotory responses was lower than that used in survival assays since pilot locomotion trials revealed lethal effects at concentrations equivalent to those used in survival experiments. Although this concentration is 4–9 fold less than the lowest concentration observed in a study of P. semilanceata (Christiansen et al. 1981), here larvae were directly soaked in psilocybin. Although soaking for locomotion experiments does not mimic natural exposure routes, it provides a controlled means of delivering psilocybin to larvae and has been widely used in Drosophila pharmacological assays (Gasque et al. 2013), thus allowing short‐term behavioural effects and potential neuromodulatory responses to psilocybin to be assessed under controlled conditions.
Two species of Drosophila were used to test the locomotory response to Psilocybe extracts in a standard invertebrate model species: Drosophila melanogaster Dahomey (wild‐type), a generalist species, as well as D. affinis, a broadly saprophagous generalist that can utilise decaying plant and fungal substrates (Markow and O'Grady 2008). To compare the difference in physiological response and to understand the potential receptor target in larvae exposed to psilocybin‐containing extracts, assays were also performed on a Drosophila mutant line backcrossed with D. melanogaster Dahomey characterised by a 90% reduction in the expression of the 5‐HT2A receptor (Häcker et al. 2003; C. D. Nichols 2007).
Locomotion assays were performed for early 3rd instar larvae in a block design (eight blocks). Early 3rd instar larvae were chosen as they feed actively and are of sufficient size for reliable video tracking (Fernández‐Moreno et al. 2007). Experimental tubes contained 115 μL of 20% sucrose solution with either 10 μL undiluted P. cubensis extract, equivalent to a final psilocybin concentration of 0.04 μg/μL, or 10 μL of a 1:1 dilution of P. cubensis extract to achieve a concentration equivalent to 0.02 μg/μL. Larvae were randomly allocated to one of three treatments, specifically 0.02 μg/μL psilocybin, 0.04 μg/μL psilocybin, or to the extraction solvent without any addition of mushroom extract (control group). Solvents were supplemented with bromophenol blue, allowing visual confirmation that the larvae were feeding. The larvae were individually placed in experimental tubes, exposed to the relevant treatment for an incubation period of 1 h before being removed and placed on a 25 mm agar plate stained with bromophenol blue for recording. During each experimental block, 36 individual 3‐min videos of each treated larva were recorded under controlled conditions using the Imaging Source DFK 33UX226 USB 3.0 camera and IC Capture 2.5 software. Average speed, total distance and turn angle were calculated for each larva using EthoVision v17 (Noldus et al. 2001).
Fly Assay: Statistical Analysis
To explore the effect of psilocybin exposure on pupation rate, survival, thorax length, wing surface area, and locomotion, a series of linear mixed effects models (LMM or GLMM) were applied for each species/strain of Drosophila independently, where species/strain was the fixed effect and experimental block was a random effect to account for variability between blocks. Pupation rate, survival to adulthood, and locomotion were (log)transformed as necessary to meet linear modelling assumptions. Likelihood ratio tests were conducted, which compared two models in each case: a null model with a random effect for “Block” and an alternative model that additionally included “Treatment” as a fixed effect. These models quantified the effects of treatment while controlling for differences between sampling blocks.
Fly Wing Analysis
A macro, written in IJM (ImageJ Macro Language) and available in Dryad, was developed for analysis of wings of flies that successfully eclosed to adulthood in survival experiments (Figure S1). Using the processed images, vein point X,Y coordinates were manually determined on the basis of a reference image. To quantify fluctuating asymmetry, 36 pairwise distances between 10 landmarks were measured for each wing. All pairwise distances, including landmark 1, were excluded from analysis because of imprecise measurements (this landmark was frequently missing as a result of broken hinges upon dissections). Fluctuating asymmetry (FA) was calculated following a standard method (Carter et al. 2009). See Supporting Information, ‘Fly wing analysis’ for full details.
Sample Collection,Extraction, and Single‐Marker Sanger Sequencing DNA
Seven species of basidiomycetes associated with dung and grassland were sampled at three sites on Dartmoor, UK, between 21st and 27th October 2021 (Psilocybe semilanceata, Mycena epipterygia, Entoloma conferendum, Hygrocybe sp., Panaeolus semiovatus, Protostropharia spp., and Hypholoma ericaeum). Voucher specimens for all collections are retained in the laboratory at the University of Plymouth. All study sites were located within an approx. 28km2 area on the southwest of the National Park. Three individual sampling sites ranged from 1‐3 km2 and were separated from others by areas of intervening habitat unsuitable for Psilocybe (e.g., areas of blanket bog). Sites were located between 280 and 420 m asl. All sites comprised rough acid grassland grazed by cattle, ponies, and sheep. At each of the three sampling sites, 14–47 fruiting bodies per species, per site, were collected, depending on local abundance. Individual collections were made at least 5 m apart to minimise the likelihood of sampling clonally connected individuals from the same mycelium. Specimens were selected at maturity (defined by the expansion of the pileus to a plane or slightly convex shape, exposure of the lamellae, and visible spore deposition on the gills or adjacent caps) as secondary metabolite profiles are known to change with developmental age (Calvo et al. 2002; Li et al. 2022; Wu et al. 2005). Single individuals were placed in resealable plastic bags, labelled, kept on ice during transport, and stored at −20°C. Following retrieval of samples and transport to the laboratory, the samples underwent lyophilization for 36 h and were subsequently stored in a desiccator prior to DNA extraction. Collected fruiting bodies were manually homogenised in original resealable plastic bags, and DNA was extracted from 0.03 g of homogenised material using a modified salt extraction protocol (Aljanabi and Martinez 1997). DNA extracts were additionally purified using a 1:1 ratio of sample to MagBio SPRI beads following Koskinen et al. (2019). Samples were stored at −20°C in 96‐well plate format. Field identifications of fungal samples were made using macroscopic characters. Representative specimens from taxa with known morphological overlap among closely related species were validated using ITS Sanger sequencing. Species with highly distinctive diagnostic features (e.g., Psilocybe semilanceata, Hygrocybe spp.) were not sequenced, but voucher material is retained. See Supporting Information, ‘Molecular Identification of Fungi’ for full details.
Barcoding of Invertebrate Communities and Bioinformatic Processing DNA
Invertebrate communities associated with mushroom fruiting bodies were characterised using 250 PE NovaSeq 6000 SP amplicon sequencing of cytochrome oxidase I (COI). Library preparation followed an adjusted Illumina 16S metabarcoding protocol. In total, 621 CO1 samples were amplified, including 40 technical replicates, 21 negative controls (water), and five extraction controls. See, ‘Illumina metabarcoding protocol’, for full details. Supporting Information
Metabarcoding: Statistical Analysis
Richness, diversity, and evenness were quantified using the number of observed OTUs, the effective number of species [exp(H), where H is Shannon's index], and Simpson's evenness index, respectively (Morris et al. 2014). To meet linear modelling assumptions, richness values were modelled using negative binomial generalised linear models (GLMs), Shannon diversity was modelled using linear models, and Simpson's evenness index was logit‐transformed prior to modelling (qlogis) (Warton and Hui 2011). To control for differences in sequencing depth, log(library. size) was included as the first fixed covariate in all models. This approach to dealing with differences in sequencing depth is encouraged (Warton et al. 2015) as it avoids many of the unfavourable aspects of alternatives such as rarefaction (McMurdie and Holmes 2014). The site was modelled as a fixed effect; with only three sampling locations, there were insufficient levels to reliably estimate a random effect, and attempting to do so risked overfitting the model without meaningful contribution from the site term. One fungal species, E. conferendum, was not present at all sampling sites; therefore, all linear models were constructed additively (i.e., without interaction terms), allowing for the estimation of main effects despite some unbalanced group representation. The main explanatory variables were either fungal species identity or associated habitat. For linear models, model comparisons were performed using F‐tests via ANOVA; for generalised models, likelihood ratio tests were used. In all cases, the Akaike Information Criterion (AIC) was also used to assess the relative fit of competing models. Negative binomial models were implemented using the MASS package (Venables et al. 2002), and linear models were run using base R functions.
Differences in overall community composition were visualised using nonmetric multidimensional scaling (NMDS) on Bray–Curtis distances, calculated from a square‐root transformed relative read abundances (RRA) table (applied to reduce the influence of highly dominant OTUs and better emphasise community‐level structure). Ordination was performed using metaMDS() in the vegan package (Oksanen et al. 2013) (version 2.8–0). PERMANOVA analysis was performed on the square‐root transformed RRA table to determine variance in OTU community clustering by fungal species, habitat ’type’ (grassland or dung), and site. Models were run independently and additively, and included log‐transformed library size as a covariate to account for differences in sequencing depth. Significance was evaluated using 10,000 permutations.
To quantify significant co‐occurrences between the RRA of arthropod OTUs across fungal species, multipattern analysis was applied using INDISPECIES on the basis of 999 permutations in the package ’indispecies’ (De Cáceres and Jansen 2016). To visualise host‐arthropod interactions in ggplot2 (Wickham and Wickham 2016), ’igraph’ (Csardi and Nepusz 2013) and ’ggnetwork’ (Briatte 2016) were used to extract coordinates to represent associations (links) between nodes (COI OTUs and fungal species) on the basis of unweighted links.
Multivariate abundance analysis was performed to identify specific arthropod OTUs that increased/decreased between dung and grassland‐associated fungi (see, ‘Multivariate abundance analysis’ for details). Supporting Information
Results
Drosophila Exposure to Extracts Containing Psilocybin Impairs Development and Survival
We exposed D. melanogaster larvae to two controls (standard Drosophila medium and button mushroom) and psilocybin concentrations of 0.1 μg/μl and 0.2 μg/μl by spiking standard food with powder from laboratory‐cultivated Psilocybe cubensis (strain ’Amazon’) and investigated their pupation rate and survival. The pupation rate of wild‐type D. melanogaster exposed to extracts containing psilocybin was greatly reduced at both concentrations compared to the true control (χ2 = 113.13, d.f. = 3, p < 0.001), with reductions of 34.05% in the button mushroom control, 61.6% at 0.1 μg/μL, and 79.3% at 0.2 μg/μL (Figure 1A). Survival to adulthood was similarly affected (χ2 = 144.38, d.f. = 3, p < 0.001), with reductions of 12.06% in the button mushroom control, 57.24% at 0.1 μg/μL, and 74.34% at 0.2 μg/μL (Figure 1B).
Exposure of larvae to psilocybin also significantly affected thorax development (Figure 1C) for both females (χ23 = 100.1, p < 0.001) and males (χ23 = 85.92, p < 0.001). In females, thorax length was reduced by 3.2% in the button mushroom control, 14.1% at 0.1 μg/μL, and 16.5% at 0.2 μg/μL compared to the control group. In males, thorax length was reduced by 2.3% in the button mushroom control, 12.57% at 0.1 μg/μL, and 12.91% at 0.2 μg/μL. The total wing surface area (SA, Figure 1D) was also significantly reduced in larvae exposed to Psilocybe extracts and reared to adulthood (χ23 = 251.33, p < 0.001) but with no discernible dose‐dependent effect. Wing surface area significantly reduced by 3.96% for the button mushroom control, 19.78% for 0.1 μg/μL, and 21.66% for 0.2 μg/μL compared to the true control.
Fluctuating asymmetry is the non‐directional variation between the left and right sides of a bilateral trait, arising from genetic or environmental stress during development and is commonly used as an indicator of developmental stability and overall organismal fitness (Bourguet 2000; Lewandowska‐Wosik and Chudzińska 2024). When exposed to Psilocybe extracts, D. melanogaster showed significant differences in fluctuating asymmetry (FA) among treatment groups (LMM; F3,105 = 16.98, p < 0.001). Tukey‐adjusted pairwise comparisons indicated that flies exposed to 0.2 μg/μl showed significantly greater FA than all other treatment groups (estimate = 0.004–0.006, all p < 0.001). In addition, FA in the mushroom control group was significantly higher than in the true control (estimate = 0.001, p < 0.05). No significant difference was detected between the true control and the 0.1 μg/μL treatment. Of 36 pairwise distances between wing vein landmarks (Figure 2), excluding vein point 1 because of frequent damage, a significant treatment × trait interaction was detected (LMM; F105, 12,775 = 1.59, p < 0.001). Follow‐up trait specific comparisons revealed that eight distance traits (2–3, 2–4, 2–5, 2–6, 2–8, 2–9, 2–10, and 3–8) showed significant treatment differences after Benjamini–Hochberg correction (all adjusted p < 0.05, Figure 3). In each case, differences were driven by elevated FA in the 0.2 μg/μL treatment relative to the other groups. All other treatment contrasts were non‐significant, except for trait 4–9, where FA in the mushroom control was significantly higher than both the true control (estimate = −0.00444) and the 0.1 μg/μL (estimate = 0.00572) treatment (p < 0.05 for both contrasts).
Effects ofsupplementation on development and morphology in. Pupation rate (A), survival to adult stage (B), thorax length (C), and wing surface area (D) of wild‐typereared on a standard food medium (True Control), standard food supplemented with 10 mg of dried button mushroom powder (Mushroom Control), or with 5 mg or 10 mg ofpowder, corresponding to psilocybin concentrations of 0.1 μg/μL and 0.2 μg/μL, respectively. Each box plot displays the median (horizontal line), mean (solid black square), and interquartile range (25th to 75th percentiles), with whiskers representing data variability. Different letters denote statistically significant differences among treatments (post hoc comparisons,< 0.05). Psilocybe cubensis Psilocybe cubensis p Drosophila melanogaster Drosophila melanogaster
Vein landmarks used for pairwise distance and fluctuating asymmetry (FA) calculations. Annotated image of awing showing the locations of vein landmarks used to measure pairwise distances and assess fluctuating asymmetry (FA). Red circles indicate specific landmark points along the wing veins, which serve as reference points for FA analysis. Drosophila
Trait‐specific fluctuating asymmetry across treatments. Measurement‐error‐corrected mean absolute fluctuating asymmetry (FA) for each wing vein landmark pair across treatments (True control, Mushroom control, 0.1 μg/μl, and 0.2 μg/μl). FA was calculated as the absolute left–right difference following correction for measurement error, averaged across individuals within each treatment. Colours represent mean absolute FA, with lighter colours indicating greater asymmetry. Asterisks indicate landmark pairs showing significant treatment effects following model‐based analyses (see‘Fly Wing Analysis’ for details). Supporting Information
Psilocybe Exposure to Extracts Shows Species‐Specific Reductions in Locomotion
We exposed wild‐type D. melanogaster, 5‐HT2A mutant D. melanogaster (Häcker et al. 2003; C. D. Nichols 2007), and a fungal feeder (D. affinis), to one control (sucrose + extractant solvent, i.e., methanol), and extracts containing psilocybin concentrations of 0.02 μg/μl and 0.04 μg/μl extracted from laboratory‐cultivated Psilocybe cubensis. The 5‐HT2A mutant was included in the study because its reduced receptor expression is expected to confer insensitivity to psilocybin, given its known high affinity for the 5‐HT2A receptor. In wild‐type D. melanogaster, exposure to extract with psilocybin concentrations of 0.02 μg/μl and 0.04 μg/μl compared to the methanol control resulted in a significant reduction in distance moved (χ2 = 24.44, df = 2, p < 0.001) and time spent moving (χ2 = 15.34, df = 2, p < 0.001; AIC = 209.36), whereas turn angle significantly increased (χ2 = 22.24, df = 2, p < 0.001). For D. affinis, exposure to extracts of both concentrations resulted in a significant reduction in distance moved compared to the control (χ2 = 24.08, df = 2, p < 0.05). There were no significant dose‐dependent responses in wt D. melanogaster or D. affinis. 5‐HT2A mutants showed a dose‐dependent response with a greater reduction in distance moved (χ2 = 24.08, df = 2, p < 0.001) and increased turn angle (χ2 = 25.07, df = 2, p < 0.001) when treated with the higher dose, whereas time spent moving reduced significantly between the control and both dose treatments (χ2 = 11.54, df = 2, p < 0.01), but not between doses (Figure 4, Table S1).
Locomotory behaviour ofandlarvae followingextract exposure. Distance crawled (A), time spent moving (B), turn angle (C) on an agar surface by(dah), 5‐HT2A mutant (5ht), and(aff) larvae treated with either sucrose + extraction solvent (Control) or sucrose +extract, corresponding to psilocybin concentrations of 0.02 μg/μL and 0.04 μg/μL, respectively. Mixed‐effects linear models were used for statistical comparisons, with drug‐fed larvae compared to control larvae of the same strain. Each box plot displays the median (horizontal line), mean (solid black square), and interquartile range (25th to 75th percentiles), with whiskers representing data variability. Different letters denote statistically significant differences among treatments within species (post hoc comparisons,< 0.05). Drosophila Psilocybe cubensis Drosophila affinis Psilocybe cubensis p D. affinis Drosophila melanogaster
Psilocybe Mushrooms Have Reduced Diversity and Distinct Communities of Invertebrates
Indicator species analysis (Figure 5), revealed 13 OTUs (operational taxonomic units) host‐specific to Psilocybe (Table S3). Of these, 11 were assigned to specialist fungal gnats in the genus Exechia (Mycetophilidae; one OTU confirmed confidently to E. frigida), one to Suillia (Heleomyzidae), and one to Sminthurididae (Sminthurididae sp. BIOUG24231‐A02). In contrast, M. epipterygia, which shared a similar community composition to P. semilanceata, held only one host‐specific OTU from the genus Exechia (Table S4) and 19 OTUs that were shared with P. semilanceata (Table S5). Thirteen of these were assigned to Exechia, one to an unassigned dipteran, one to Phortica (Drosophilidae), one to Coleoptera, and three to Sminthurididae (Collembola). Overall, a network consisting of 133 out of 309 arthropod OTUs that exhibited significant associations with more than one fungal host was revealed. Eighty‐three OTUs were associated with one fungal host (62%), 38 OTUs were associated with two groups (28.6%), 10 OTUs were associated with three groups (7.5%), and 2 OTUs were associated with > four groups.
Metabarcoding of cytochrome oxidase I (COI) revealed that P. semilanceata and M. epipterygia hosted a distinct arthropod community compared to all other grassland and dung fungi sampled, and all other species overlapped in their community composition (Figure 6). PERMANOVA analyses revealed that fungal species identity was the strongest predictor of arthropod community composition (R2 = 0.598, F6,514 = 149.37, p < 0.001). Habitat type (dung vs. grassland species) also explained a substantial but smaller proportion of the variation (R2 = 0.13, F1, 519 = 84.49, p < 0.001), followed by sampling site (R2 = 0.02–0.03, F2, 512–518 = 9.45–20.21, p < 0.001), Table S6. In all models, log‐transformed library size was included as a covariate and found to be significant (R2 = 0.06, p < 0.001), though it accounted for a relatively small proportion of the variation compared to biological factors. After excluding Mycetophilidae (fungus gnats, which account for 98% of the dataset), fungal species still explained most of the variation in arthropod communities (R2 = 0.32, p < 0.001), with smaller but significant contributions from habitat type (R2 = 0.06, p < 0.001), and site (R2 = 0.018–0.02, p < 0.001) (Figure S4).
As well as showing unique OTUs and a community distinct from most other grassland fungi (see Table S7 for a complete taxonomic list of invertebrates identified), P. semilanceata shared the lowest invertebrate diversity along with two other grassland species, H. ericaceum and M. epipterygia (Figure 7). All invertebrate diversity indices significantly varied across fungal species (richness LR512,518 = 121.21; diversity F512,518 = 107.01; evenness F512,518 = 98.69. p < 0.001) and between dung vs. grassland habitats (richness LR517,518 = 7.73; diversity F517,518 = 17.75; evenness F517,518 = 38.81. p < 0.001p < 0.05) and all three measures of diversity across invertebrate communities were lowest in grassland associated fungi (Figure S5, Table S8).
Species indicator analysis of arthropod families associated with different fungal hosts. Indicator species analysis showing arthropod families associated with different fungal hosts. Black circles represent fungal hosts, with arthropod families positioned according to their association with fungal hosts.
Influence of fungal species and habitat type on arthropod community composition in natural grasslands. Nonmetric multidimensional scaling (NMDS) ordination illustrating the effects of fungal species and associated habitat (Type) on arthropod communities (310 OTUs) across three natural grassland sites in southwest England, UK (stress: 0.17). Ordinations were generated using Bray–Curtis distance matrices on the basis of square‐root‐transformed relative abundances of the total fungal community. The NMDS plot highlights a distinct separation of arthropod communities associated withandcompared to all other sampled fungal species, indicating potential host‐specific interactions. Psilocybe semilanceata Mycena epipterygia
![Arthropod diversity metrics across fungal species in natural grasslands. Arthropod (A) OTU richness, (B) effective number of species [exp(Shannon's Diversity Index)], and (C) Simpson's Diversity Index across fungal species at three natural grassland sites in southwest England, UK. Each box plot displays the median (horizontal line), mean (solid black square), and interquartile range (25th to 75th percentiles), with whiskers representing data variability. Fungi associated with grasslands are represented by grey box plots, whereas dung‐associated fungi are also shown in grey box plots. Different letters denote statistically significant differences among treatments within species (post hoc comparisons,< 0.05).’ to the Figurelegend, and all other subsequent figures that now include group lettering. P 3](https://europepmc.org/articles/PMC13103472/bin/ECE3-16-e73522-g006.jpg.jpg "Click to view full size")
Arthropod diversity metrics across fungal species in natural grasslands. Arthropod (A) OTU richness, (B) effective number of species [exp(Shannon's Diversity Index)], and (C) Simpson's Diversity Index across fungal species at three natural grassland sites in southwest England, UK. Each box plot displays the median (horizontal line), mean (solid black square), and interquartile range (25th to 75th percentiles), with whiskers representing data variability. Fungi associated with grasslands are represented by grey box plots, whereas dung‐associated fungi are also shown in grey box plots. Different letters denote statistically significant differences among treatments within species (post hoc comparisons,< 0.05).’ to the Figurelegend, and all other subsequent figures that now include group lettering. P 3
Discussion
Understanding why fungi have evolved psychedelic compounds that mimic neurotransmitters is a fascinating but unresolved question in evolutionary chemical ecology. Here, we provide the first empirical evidence that extracts from Psilocybe species affect insect survival, development, and behaviour, consistent with a possible defensive role. Drosophila larvae exposed to extracts from Psilocybe species show reduced pupation rates, decreased survival to adulthood, decreased thorax size and wing size, and increased fluctuating asymmetry. Locomotion was also compromised, with larvae spending less time moving, moving more slowly, and showing increased turning behaviour. These results lend support to the hypothesis that these compounds have evolved for defence purposes. High‐throughput NGS metabarcoding revealed distinct invertebrate communities in P. semilanceata compared with most other species sampled. However, its overlap with a non‐psychedelic species, Mycena epipterygia, and the results observed for 5‐HT2A‐deficient Drosophila suggest certain species may have adaptations that allow them to be resistant to the potential negative effects of psilocybin, and receptor subtypes other than 5HT2A are likely involved in signalling in invertebrates. Alternatively, other factors might drive the ecology and evolution of psilocybin and related compounds.
Reducing the survival of dipteran larvae, likely leading to a smaller population within fruiting bodies, could confer a fitness advantage to the psilocybin‐producing organism through bottom‐up control. Comparable patterns have been observed in both plant and fungal kingdoms, for example, in Aspergillus nidulans, the production of SMs reduces Drosophila larval survival (Rohlfs et al. 2007), and Brassica plants, which produce certain glucosinolates (sinigrin), show significantly smaller aphid colony sizes (Newton et al. 2009), decreasing insect pressure and altering antagonistic insect interactions. Additionally, larvae with impaired activity, as demonstrated in this study, will likely feed less as mobility is closely tied to their ability to locate and access food (Troncoso et al. 1987). Serotonergic signalling may also contribute to this effect as 5‐HT agonism is a known appetite suppressant (Halford and Harrold 2012; Dacks et al. 2003). In Drosophila larvae, receptor subtypes 5HT1B and 5HT2A are involved in feeding behaviour, whereas 5HT2A and 5HT7 influence locomotion (Majeed et al. 2016), providing a plausible mechanistic basis for psilocybin‐mediated defence. Such reductions in mobility and feeding could, in turn, decrease fungivory pressure on the fruiting body, thereby providing a potential benefit to the mushroom. Although evidence remains mixed regarding the correlation between fluctuating asymmetry and fitness (Bourguet 2000; Lewandowska‐Wosik and Chudzińska 2024; Carter et al. 2009), increased developmental stress and the resulting morphological impairments indicate possible deleterious effects on fungivores. Our results suggest that developmental buffering remains effective at lower exposure levels but becomes compromised at higher concentrations, with effects disproportionately affecting vein distances associated with landmark 2. Because these distances span multiple longitudinal veins and much of the wing length, the resulting asymmetry may alter wing stiffness, aerodynamic efficiency, and vibrational properties during flight (Combes and Daniel 2003; Wootton et al. 2003). Although FA was slightly elevated in the mushroom control relative to the true control, this increase was small and largely trait‐specific, indicating that the pronounced asymmetry observed at 0.2 μg/μL is unlikely to be driven solely by the mushroom matrix. These results show that Psilocybe extracts have negative impacts on multiple aspects of invertebrate fitness and behaviour, supporting chemical defence as a driver of secondary metabolite diversification. However, it is important to consider that reduced survival and/or developmental impairment may arise indirectly through reduced locomotion and suppressed feeding; therefore, the smaller adult thorax/wing surface area could reflect decreased food intake rather than direct toxicity.
In this study, broad‐scale extracts from Psilocybe were tested. Although ecologically realistic, because in nature larvae are also exposed to entourage compounds such as norbaeocystin and baeocystin, this extract will also contain numerous other polar compounds (Fricke et al. 2017), making it difficult to directly infer a causal role of psilocybin. Similarly, Agaricus bisporus is not an optimal control as exposure also resulted in reduced pupation/survival and impaired development in D. melanogaster. This effect may reflect additional fungal SMs responsible for reductions in invertebrate fitness and development, or it may reflect nutritional differences between mushroom tissue and the standard laboratory Drosophila diet, which has been optimised for Drosophila cultivation. Closely related and cultivable non‐psychedelic basidiomycetes (containing similar compounds except psychedelic indole alkaloids) or targeted gene knockouts (e.g., CRISPR/Cas9 disruption of the psilocybin biosynthetic cluster) would provide stronger inference. Although CRISPR/Cas9 methods have been established for some basidiomycetes (Liu et al. 2020), efficiency remains limited by dominant non‐homologous end joining (NHEJ). The authors have since made some progress producing basidiomycete CRISPR/Cas9 mutants; however, such models were not available for experimental application at the time of this study. Future mutant‐based approaches could also determine whether psilocybin biosynthesis confers reciprocal fitness benefits for the producer, which is essential for the inference of defence (Rohlfs 2015; Biedermann 2019). Furthermore, P. cubensis extracts were used in Drosophila assays because of their ease of cultivation, whereas invertebrate communities in P. semilanceata were investigated in the field. Unfortunately, P. semilanceata cannot be cultivated in the laboratory. Although both fungi produce psilocybin, differences in the relative abundance of psilocybin and related compounds are well documented (Cohen et al. 2025), and such variation may influence synergistic interactions, alter physiological responses, that is, ‘the entourage effect’ (Murray et al. 2026), and potentially reflect distinct ecological roles. To fully understand the potentially diverse ecological functions of distantly related psilocybin‐producing fungi, comparable assays using whole extracts from species with contrasting ecologies will be required. Additionally, Drosophila was used as the experimental model rather than a species from the Sciaroidea (the superfamily containing most fungus gnats). Although Drosophila provides a well‐established, genetically tractable system for testing defensive effects in a naïve insect lineage and for examining mutant responses to determine 5HT2A interaction, it may not fully represent natural mycophagous interactions. Indeed, larval‐rearing experiments have indicated that insect life‐history (mycophagous vs. frugivorous species) shapes tolerance to ibotenic acid from fly agaric (Tuno et al. 2007), stressing the need to test on ecologically relevant taxa. Conversely, using more ecologically relevant taxa such as mycetophilid gnats, specialists that have evolved to develop within fungal tissues, could obscure defensive effects because of potential tolerance or coadaptation. Employing fly models with contrasting ecological traits and life histories will be valuable in future work.
In contrast to the significant effects on survival, pupation rate, locomotion, and fluctuating asymmetry, two of our findings challenge a strictly 5HT2A‐mediated defensive role for psilocybin‐containing polar extracts from Psilocybe. Firstly, 5HT2A mutants (with a 90% reduction in 5‐HT2A receptors (Häcker et al. 2003, C. D. Nichols 2007)) exhibited heightened sensitivity to extracts. Potentially, larvae were responding to other co‐extracted compounds from the mushrooms, or psilocybin may interact with additional receptor pathways. The latter could arise, for example, because of 5‐HT2A receptor saturation (because the relative concentration of psilocin to receptors is increased), interference with 5‐HT kinetics, or interaction with alternative targets such as serotonin subtypes, adrenergic, dopaminergic, or histaminergic receptors. Drosophila possess five known 5‐HT receptor subtypes (Gasque et al. 2013; Johnson et al. 2009), and cross‐reactivity with 5‐HT1A or dopaminergic targets is possible, as observed for LSD (Nichols et al. 2002). To fully understand the involvement of the central nervous system in invertebrate responses to psilocybin, pharmacological approaches employing receptor antagonists will be essential. Additionally, further research on single compounds is necessary to confirm whether the effects observed result directly from psilocybin or from synergistic actions with related metabolites. Secondly, although metabarcoding revealed a distinct separation among invertebrate communities and P. semilanceata harboured 13 unique OTUs, its community composition overlapped substantially with that of the non‐psychedelic M. epipterygia. This overlap suggests that psilocybin is not the only factor structuring arthropod associations. Microhabitat preference of ovipositing flies, fruit body morphology, or shared volatile cues may also be drivers of these patterns. In relation to insect‐fungal interactions more broadly, the high host specificity (62% of OTUs restricted to a single fungal species) underscores the potential for coevolutionary interactions to be mediated by fungal traits. More research is necessary to investigate fungal host selection by fungivorous invertebrates.
In general, several ecological and physiological factors suggest that psilocybin may serve functions other than defence. Many basidiomycetes release tens of millions of spores per hour (Rockett and Kramer 1974; Wójcik and Kasprzyk 2023) and soft‐bodied fungi are short‐lived, making long‐term investment in chemical defence potentially unnecessary. Psilocybin is also present in vegetative mycelia (Blei et al. 2020), suggesting potential roles beyond fruiting body protection. Moreover, the delayed onset of psychoactive effects, at least in vertebrates (Passie et al. 2002), seems inconsistent with a purely defensive role (Bradshaw et al. 2024). Manipulation of animal behaviour for spore dispersal is an obvious alternative hypothesis (Meyer and Slot 2023). Some fungal spores can survive transit in the insect gut and remain viable (Lunde et al. 2023), raising the possibility of dissemination via mycophagous insects. Supporting this, Massosopora species, which can produce either psilocybin or cathinone (an amphetamine), induce hyperactive behaviour in infected hosts, increasing spore transmission (Meyer and Slot 2023; Slot and Hoffmeister 2025). Further, fungal manipulation of invertebrate behaviour is widespread. For example, entomophthoralean fungi, which infect ants, grasshoppers, and flies, alter their behaviour prior to sporulation, where infected individuals climb to high points on vegetation prior to the emergence of conidiophores (“summit disease”; Hughes et al. (2016)). In other cases, the insect is kept alive during the spore dispersal stage e.g., in Hylemya flies where spores are released during flight (Hughes et al. 2016). Similarly, the hypocrealean ‘zombie‐ant’ fungi (Ophiocordyceps) also manipulate host behaviour for the adaptive benefit of the fungus (Andersen et al. 2009; Loreto et al. 2014) through the production of neuromodulatory metabolites when in contact with ant brain tissue (de Bekker et al. 2014). It is also intriguing that genes within the psilocybin BGC, PsiH and PST, show extensive gene family duplication in Fibulorhizoctonia sp. (Reynolds et al. 2018), a fungus known to produce termite egg‐mimicking sclerotia in an ancient mutualistic relationship with Reticulitermes termites (Matsuura, 2006). Finally, the effects of psilocybin on animals, like its psychedelic action in humans, could simply be fortuitous, with selection acting on other ecological functions such as microbial interactions. Consistent with this hypothesis, Psilocybe cubensis extracts have been reported to suppress certain pathogenic bacteria, indicating potential antimicrobial properties (Abdul‐Hadi et al. 2025; Karthiyayini et al. 2025); ergot alkaloids show inhibition of bacteria such as Escherichia coli (Panaccione 2005), and early research on P. semilanceata established that this species can inhibit other fungal colonists of the plant root (Keay and Brown 1989).
An additional, non‐exclusive hypothesis worth consideration is the polymer hypothesis, which proposes that psilocin oxidation products may polymerise following tissue damage via an enzymatically controlled pathway, potentially serving an inducible defensive role (Lenz et al. 2020). Future studies should investigate whether physical injury, in combination with invertebrate enzymatic cues (e.g., salivary secretions), triggers elevated psilocybin, polymer formation, or up‐regulated gene expression, and whether such responses reduce fungivore damage. The observed heightened sensitivity of 5‐HT2A receptor mutants to Psilocybe extracts further suggests that the ecological effects of this compound may extend beyond direct 5‐HT2A receptor‐mediated behavioural modulation. Psilocybin may act similarly to cyanide‐releasing systems in plants (Zagrobelny et al. 2004), acting as a stable precursor that, upon tissue damage, is enzymatically transformed into a compound with toxic or deterrent effects.
This is the first empirical study to investigate the origin of psilocybin in nature. We present ecological evidence that extracts from Psilocybe cubensis, including ecologically relevant doses of psilocybin among other compounds, significantly alter insect development and behaviour, consistent with a potential defensive role. Despite the complexity of the extract and the resulting caveat that a direct causal relationship cannot be confirmed, the presence of psilocybin and its oligomers remains a parsimonious explanation for the observed effects on invertebrate fitness and behaviour. However, the persistence of these effects in 5‐HT2A mutants, community overlaps with non‐psychedelic fungi, and the lack of suitable mushroom controls challenge an interpretation strictly limited to defence. These findings highlight the untapped potential of wild Basidiomycota as systems for studying the ecology and evolution of neuroactive natural products. Future studies integrating transgenic tools, comparative metabolomics, receptor antagonists, single‐compound assays, and ecologically relevant insect models offer exciting avenues to resolve the evolutionary drivers of the origin of psilocybin in nature. This future work should also address additional hypotheses regarding mutualism or microbial interactions, as well as extend assays to gastropod models. Beyond its ecological significance, these findings also have important pharmacological implications. The inducible defence model aligns with prodrug activation strategies in pharmacology and highlights wild basidiomycetes as a rich yet underexplored source of neuroactive compounds. Activation of fungal metabolites by invertebrate‐associated molecular patterns is also an exciting and underexplored area to discover novel silent metabolites.
Author Contributions
K. J. Matthews Nicholass: data curation (lead), formal analysis (lead), investigation (lead), methodology (lead), visualization (lead), writing – original draft (lead), writing – review and editing (lead). I. Flis: investigation (equal), methodology (equal), writing – review and editing (equal). M. E. Hanley: conceptualization (equal), funding acquisition (equal), methodology (equal), supervision (equal), writing – review and editing (equal). M. E. Knight: conceptualization (equal), methodology (equal), supervision (equal), writing – review and editing (equal). S. M. Lane: methodology (equal), supervision (equal), writing – review and editing (equal). G. Littlejohn: conceptualization (equal), funding acquisition (equal), methodology (equal), supervision (equal), writing – review and editing (equal). R. A. Billington: methodology (equal), writing – review and editing (equal). R. Boden: methodology (equal), writing – review and editing (equal). R. Cummins: investigation (supporting). B. J. Green: methodology (supporting). C. Griffin: methodology (equal), visualization (equal), writing – review and editing (equal). S. Jones: investigation (supporting). D. Salmon: investigation (supporting). I. Sleep: investigation (supporting). N. Smirnoff: methodology (supporting), writing – review and editing (equal). J. S. Ellis: conceptualization (equal), funding acquisition (lead), project administration (lead), supervision (lead), writing – original draft (equal), writing – review and editing (lead).
Funding
This work was supported by the Leverhulme Trust grant RPG‐2020‐023 (JSE, MEH, GL, and MT).
Conflicts of Interest
The authors declare no conflicts of interest.
Psychedelic Science