Quantification of invertebrates on fungal fruit bodies by the use of time‐lapse cameras

Fungi and invertebrates comprise a major part of biodiversity in dead wood ecosystems and invertebrates depend on fungi to utilise the dead wood resource. Many invertebrates also visit the long‐lived fruit bodies of wood‐decay fungi to feed on spores, the hymenium or other invertebrates. However, as traditional sampling methods are labour‐intensive, we know little of these interactions. In this study, we use time‐lapse cameras to monitor invertebrates visiting the hymenium of a common wood‐decay fungus, Fomitopsis pinicola, and explain their activity in terms of temporal variation, temperature and presence of Gyrophaena boleti, a highly abundant fungivorous beetle living primarily in fruit bodies of F. pinicola. The most common invertebrates on F. pinicola fruit bodies were Coleoptera, Araneae, Diptera, Gastropoda and Chilopoda. The invertebrate activity exhibited strong temporal variation with higher abundance during night and, for Coleoptera, earlier in the season. We discuss how this might correlate with the sporulation period of F. pinicola. The presence of G. boleti had a positive impact on the predatory Lordithon lunulatus and Ipidia binotata, and a negative impact on the fungivorous Thymalus limbatus and Peltis ferruginea. Chilopoda and L. lunulatus were ephemeral visitors, while the fungivorous Coleoptera and Araneae stayed the longest. We estimated the invertebrates' visitation frequency and duration, which would be time‐consuming to obtain with traditional methods. We offer improvements to our method and urge future research on invertebrate–fungus interactions to quantify invertebrate visits to fungal fruit bodies.


INTRODUCTION
Wood makes up more than 90% of aboveground biomass in forests, and the carbon stored in dead wood has been estimated to include about 8% of global forest carbon stocks (Pan et al., 2011;Rayner & Boddy, 1988).
The wood carbon is embedded in a molecular complex of lignin, cellulose and hemicellulose, which, in boreal forests, becomes accessible as a broad food source after the initial decomposition by wood-decay fungi (Seibold et al., 2021;Stokland et al., 2012). The fungal biomass represents an important nutrient source for invertebrates (Boddy & Jones, 2008;Filipiak & Weiner, 2017), which, together with fungi, comprise the larger part of forest biodiversity (Stokland et al., 2012).
The invertebrates feeding on fungi are dominated by insects, particularly beetles (Coleoptera) and true flies (Diptera), mites, nematodes and slugs (Stokland et al., 2012). The fungus can be ingested as yeast, mycelium, spores or fruit bodies (Birkemoe et al., 2018). The fungal fruit body is a discrete and conspicuous fungal resource and therefore well suited for studies on interactions between invertebrates and fungi. Many fungivorous insects, for example, trogossitid beetles, undergo larval development inside the fruit body and emerge as adults (Schigel et al., 2006), while others visit the fruit body surface as adults to feed on spores, hymenium (i.e. the spore-producing layer) or other invertebrates. In wood-decay fungi, the fruit bodies may persist for many years and also host diverse invertebrate communities inside, both while alive (Lunde et al., 2022;O'Connell & Bolger, 1997) and dead (Hågvar & Steen, 2013;Pielou & Verma, 1968;Thunes et al., 2000).
Studies of invertebrate communities inside fruit bodies have shown niche differentiation by decay stages (Jonsell & Nordlander, 2004;Kadowaki, 2010;Thunes et al., 2000) and by different fruit body traits (Lacy, 1984;Lunde et al., 2022;Paviour-Smith, 1960;Thorn et al., 2015). Beetle communities feeding on spores or the hymenium of wood-decay fungi also seem to differ between fungal species, but more studies are needed in order to confirm this pattern (reviewed in Birkemoe et al., 2018). Any visit to the hymenium of a living fruit body might potentially lead to spore dispersal (Jacobsen et al., 2017;Lunde et al., 2023;Seibold et al., 2019). Thus, understanding how invertebrates use this fungal resource may have large implications on the fungal community itself. Spore production, which varies throughout the day and season (Haard & Kramer, 1970;Norros et al., 2023;Nuss, 1986), competition and predation might determine the realised niches of invertebrates visiting fungal fruit bodies.
Fomitopsis pinicola is a keystone wood-decay fungus in boreal forests, contributing both to the decay rate of dead wood (Gramss, 2020) and to shaping the biological communities within it (Pouska et al., 2013;Weslien et al., 2011). The rove beetle Gyrophaena boleti is a specialist fungivore, developing inside individual pores of the F. pinicola fruit body as a larva, while the adults can be seen on the surface in large clusters during spring and early summer (Hågvar, 2018;Staniec et al., 2016).
Many invertebrates, especially beetles, visit the fruit body surface at night in spring (Hågvar, 1999;Krasutskii, 2007), that is, when spore production is highest (Norros et al., 2023). A recent study showed how these beetles could serve as vectors for dispersing F. pinicola spores (Lunde et al., 2023), but more research is needed on their visitation patterns to understand the importance of this interaction.
The invertebrate community associated with F. pinicola fruit bodies has traditionally been investigated by rearing (e.g. Jonsell et al., 1999;Lawrence, 1973;Økland, 1995), use of trunk window traps installed in a cut fruit body (Hågvar & Økland, 1997;Kaila et al., 1994;Økland & Hågvar, 1994;Sverdrup-Thygeson, 2001), or by observations of the hymenium layer (Hågvar, 1999;Nikitsky & Schigel, 2004). The trunk window trap method is valuable but you cannot be sure that the trapped insects actually visited the fruit body. Hågvar (1999) manually observed 100 F. pinicola fruit bodies several times throughout the season over two years. This extensive field work only allowed for short periods of observations of each fruit body, making quantification of visits difficult. At present, the number and duration of visits as well as biotic and abiotic factors driving the activity are practically unknown. Furthermore, most studies so far have been on beetles (e.g. Hågvar, 1999;Krasutskii, 2007), thus giving a biased picture of the invertebrate community on fruit bodies as a whole.
Technical advances now enable camera monitoring of small insects in the field (Høye et al., 2021;Pegoraro et al., 2020). Camera monitoring has a big potential in supplementing human observations in the field as it is time-efficient and at the same time accumulates valuable ecological data (Bjerge et al., 2023;Hofmeester et al., 2020). Recently, insects visiting mushrooms on the forest floor have been monitored by the use of time-lapse cameras (Schmid et al., 2019) and similar techniques have also been used to observe insects visiting flowers (Alison et al., 2022;Bjerge et al., 2023;Suetsugu et al., 2017). In this study, we use time-lapse cameras to better understand the importance of fungal fruit bodies for forest invertebrates. By monitoring invertebrates on the hymenium of living F. pinicola, we intend to (1) observe and identify invertebrates on fruit bodies, (2) determine the importance of time of season, day, temperature and presence of G. boleti for the major invertebrate groups and (3) estimate the visitation frequency and duration of the most common taxa.
F I G U R E 1 Time-lapse camera mounted to face the hymenium of a Fomitopsis pinicola fruit body. See Figure S1.1 for more information about the camera set-up.

Study site and fruit body selection
The study was carried out in Østmarka (59 84 0 N 11 02 0 E) and Nordre Pollen (59 75 0 N 10 76 0 E) nature reserves, consisting of spruce-dominated mature forests. At each study site, living fruit bodies of F. pinicola (Sw.) P.Karst. (Polyporales, Basidiomycota, Fungi), a common brown-rot fungus in Fennoscandia, were selected for monitoring invertebrates with time-lapse cameras. In 2019, 11 fruit bodies were selected in Østmarka, while in 2020, 6 fruit bodies were selected in Østmarka and 6 in Nordre Pollen. All selected fruit bodies were at least 100 m apart and their hymenial area ranged from 47 to 325 cm 2 .
Only fruit bodies from standing tree stumps of Norway spruce (Picea abies (L.) H.Karst.) were selected.

Camera set-up
Time-lapse cameras (Wingscapes ® Moultrie WCT-00126 Timelapse-Cam Pro) with 24-LED automatic flash were affixed underneath each F. pinicola fruit body and set to take pictures every 10 min. Temperatures were measured simultaneously by the camera's internal thermometer. The cameras were mounted around spruce stumps and faced upwards c. 20 cm from the hymenium ( Figure 1). As the fruit bodies differed in size, the photographed area (c. 110 cm 2 ) varied from the whole to c. half of the hymenium. The camera mount was adjustable to move the camera closer to the log or to tilt it at an angle directly facing the fruit body ( Figure S1.1). Technical issues of the setup were recognised and addressed as they were discovered: (1) to prevent dew, a rubber ring was placed around the lens and a transparent acrylic sheet pressed onto it, (2) to prevent glare, the flash was covered with dusted acrylic sheet, and (3) to reduce light reflection, metal plates were added adjacent to the lens ( Figure S1.1). Although this reduced some of these issues, moisture and light reflection remained somewhat problematic throughout the study.

Camera monitoring and image processing
The cameras were active during two periods in 2019 (15 May-19 Jun and 31 Jul-23 Sep) and one longer period in 2020 (18 May-19 Sep). In 2019, the cameras were not checked (for defects, battery change, etc.) during the monitoring periods, which resulted in loss of data. Image data from four cameras were not used for further analyses; one camera took less than 500 images (2019), one was stolen and two were malfunctioning (2020), leaving image data from a total of 19 cameras (Table S1.1).
In total, 222,713 images (2019: 84254, 2020: 132590) were taken of the hymenium of F. pinicola. It was not possible to discern invertebrates in many of these images due to blurriness, bad lighting conditions or dew on the lens ( Figure S2.1). In order to sort images of 'good' quality (defined as an image in which invertebrates could be discerned), a variance of Laplacian (VL) value was calculated for each photo, estimating image sharpness based on an edge detection algorithm (Bansal et al., 2016;Pech-Pacheco et al., 2000). In 2019, a VL threshold value was determined as the mean VL -SD VL of 123 randomly selected images that had been identified manually as of 'good' quality. The edge detection algorithm's performance in discerning images of 'good' quality was compared with that of three human observers by using a random subset of 2000 images (Table S2.1). Classifications done by edge detection and human observers were weakly correlated, but the former was more conservative, removing 43,197 images (51%). In 2020, in an attempt to account for the variation in light and moisture conditions between sites, the VL threshold value was calculated based on 100 manually identified images of 'good' quality for each camera, removing 46,405 images (35%).

Image annotation
The VGG Image Annotator (VIA) software (Dutta & Zisserman, 2019) was used to manually classify invertebrates into the lowest possible taxonomical rank (only beetles were annotated to species). Most invertebrates larger than 2 mm were identified, but the lower size limit was determined by image quality. The identification of Ipidia binotata was confirmed by an expert from a few images, but the identifications from more blurry images are uncertain as these beetles are very alike Glischrochilus species in both morphology and behaviour.
Invertebrates that could not be identified to order (or class for noninsects) were marked as "unknown". Larvae were not annotated; fungus gnat larvae occurred on successive images over large time periods on most fruit bodies, but they were not annotated due to a lack of time. During annotation, additional 39,514 images (2019: 11041, 2020: 28473) were removed manually as the quality was too low to detect invertebrates despite passing the edge detection filter as described above. In the end, the total number of images available for analysis were 30,016 in 2019 and 57,712 in 2020. A flow chart of the image sorting process is shown in Figure S2.2.
The percentage of images of 'good' quality varied strongly between fruit bodies (6%-57% in 2019, 21%-77% in 2020), which might be caused by differences in local conditions such as Sun exposure and moisture caught by the lens or technical differences between individual cameras. 'Good' quality images varied throughout the season (Table S2.2), possibly due to large variation in rainfall and fewer images taken in July, and throughout the day (Table S2.3), most likely because more blurry images were taken in the morning (moisture on the lens) and day (direct sunlight).

Statistical analyses
All analyses and data preparation were done in R v 4.1.2 (Team, 2021). Figures were generated with the packages ggplot2 (Wickham et al., 2016) and sjPlot (Lüdecke, 2021). Binomial distribution if not. Fruit bodies (i.e. cameras) with less than 10 overall observations of the respective taxon were removed prior to analyses, which meant that the models for the different invertebrates taxa were based on different subsets of the same dataset (Table S3.1).
Model selection was done with the MuMIn package (Barton, 2020), and model averaging was used with cut-off at the 95% confidence interval of the cumulative sums of weights (Burnham & Anderson, 2002). Based on this criterion, only one model was selected to explain observations of T. limbatus and P. ferruginea (Table S3.1).
Because a large proportion of the original images were removed, the time interval between successive images was not always 10 min ( Figure S3.1). To model the responses as 'observations per 10 min', the natural logarithm of the time interval between successive images was used as an offset variable (Reitan & Nielsen, 2016). An observation-level random effect was included to deal with overdispersion (Harrison, 2014). Fruit body identity was set as random effect to control for repeated measures among fruit bodies, even though the number of random levels was low in some models (Gomes, 2022). In two cases, camera identity could not be used as random effect: (1) models of P. ferruginea observations because they were only from one fruit body, and (2) models of L. lunulatus observations because they were from two fruit bodies in which the observations were, more or less, completely separated in time.
Time of season, time of day, temperature and presence of G. boleti clusters were used as predictors to explain invertebrate observations ( Table 1). Time of season was the day of year, scaled and centred so that 0 equalled 27 June (i.e. day 178). Time of day was represented by four harmonic functions, sine and cosine, that oscillate between À1 and 1 throughout the day ( Figure S3.2). The relative contribution of each predictor was quantified with squared standardised regression coefficients (Afifi et al., 2003).

Invertebrate observations and taxonomic distribution on fruit bodies
Of the total 87,728 images (taken every 10 min) with adequate quality for identification of invertebrates (see Materials and Methods for details), 6248 invertebrate observations were made ( Table 2). The most observed taxon was by far Coleoptera (beetles, 56.3%), followed by Araneae (spiders, 11.3%), Diptera (true flies, 11.3%), Gastropoda («slugs», 6.4%) and Chilopoda (centipedes, 4.8%). All Gastropoda observations were of slugs and all Hymenoptera were ants or parasitic wasps. Eight Coleoptera species were identified, and three additional genera (Table S4.1). With the exception of G. boleti, which can be recognised based on its unique ecology and behaviour, only beetles larger than 4.75 mm were identified to species (Figure 2, Table S4.1). The most common ones were T. limbatus, P. ferruginea, L. lunulatus and I. binotata (Table 2). There was considerable variation in invertebrate observations between fruit bodies-ranging from 0.4% to 27.6% observations per image-between 1 June and 20 June ( Figure S4.1, Table S4.2).

Factors that control invertebrate observations
The four beetle species were more abundant in spring and early summer, while Diptera and Gastropoda were more common later in the season ( Figure 3, Table 3  Sine/cosine * (2*π*(minute of day/1440)). whose peak activity was during the day and evening, respectively ( Figure 4, Table 3). Time of day was the most important variable explaining Gastropoda observations (Table S5.1). Temperature was included in the top models explaining observations of all invertebrates (Table 3). Gyrophaena boleti clusters were present on 24% of all images on average, ranging from less than 1% to 72% between fruit bodies ( Figure S4.1, Table S4.2). Their presence was strongly driven by time of season, with peaks from April to June (Table S5.1, Figure S5.2). The presence of G. boleti clusters on fruit bodies had a positive effect on L. lunulatus and I. binotata (i.e. incidence rate ratio >1), and a negative effect on T. limbatus and P. ferruginea (i.e. incidence rate ratio <1) (Figure 4, Table 3).

Visitation frequency and duration
The most frequent visitor was Thymalus limbatus (466 visits during the study period) (Table 4)  decreased. The duration of visits was longest for the two fungivorous beetles and shortest for the predatory centipedes and L. lunulatus.

The invertebrate community on fruit bodies
Beetles were the most dominant group (56% of observations), but spiders, true flies, slugs and centipedes still made up almost 35% of all recordings, making these groups important components of the invertebrate community on the hymenium of F. pinicola. We are not aware of any study summarising invertebrate observations other than beetles on living fruit bodies of wood-decay fungi, but true flies such as gall midges (Cecidomyiidae) and dark-winged fungus gnats (Sciaridae) have been reared from living F. pinicola before (Økland & Hågvar, 1994). Further, keroplatid fungus gnats whose larvae produce nets on the fruit body hymenium to feed on spores (Jakovlev, 2012) were frequently visible in the images, albeit not included in this study. True flies may rival beetles as the most numerous and diverse order of insects in dead wood (Ulyshen, 2018) and, although far less frequent than beetles in the present study, they came out as the third most common order.
In our study, P. ferruginea and T. limbatus were the most active invertebrates. These are well known as grazers of the hymenium of wood-decay fungi , but have formerly been recorded in low numbers from trunk window traps installed in cut fruit bodies of F. pinicola (Hågvar & Økland, 1997;Sverdrup-Thygeson, 2001) as well as by direct observations (Hågvar, 1999). The discrepancies between these studies and our results might be due to between-year variation in population densities or the grazing behaviour of these beetles, which could inflate the number of observations on images compared with the number of individuals sampled manually or in traps.
The observational study carried out in Østmarka, which partly overlaps geographically with our study, found 27 beetle species visiting the fruit bodies of F. pinicola (Hågvar, 1999). Despite this high diversity, the time-lapse cameras from our study detected four beetle species not observed by Hågvar (1999): Triplax russica, Melanotus castanipes, I. binotata and L. lunulatus. The latter two, I. binotata and L. lunulatus, also lacked in the extensive survey by Schigel (2011) on beetles associated with wood-decay fungi in Finland. We found that these beetles visited the fruit bodies frequently, but for short time periods only, which might reduce detectability by human observations and cause the discrepancies with our results. Both L. lunulatus and Ipidia binotata were caught in trunk window traps on F. pinicola in the same area, the first being far more numerous than the latter (Hågvar & Økland, 1997). These findings match our observations. Several species that have been recorded in earlier studies were missing from our observations; Epuraea variegata, Henoticus serratus and Cis glabratus were among the most common species observed on F. pinicola by Hågvar (1999), of which E. variegata and C. glabratus have also been trapped in high numbers in trunk window traps (Hågvar & Økland, 1997, Sverdrup-Thygeson, 2001 before July (Burner et al., 2022), indicating that the flight period of most forest beetles is indeed early in the season. For fungusvisiting beetles, this seasonal pattern could also be related to the spore production of F. pinicola, which is largest in spring (Norros et al., 2023;Nuss, 1986). True flies, slugs and centipedes exhibited seasonal variation as well, but appeared later in the season,  (Table S2.2). Note that the scale on the y-axis of Thymalus limbatus is larger than the others. Icons by Henrik Aubert.
indicating that they are less dependent of sporulation rates. As these groups were not identified to species, the patterns are more difficult to explain, but late summer activity may reduce competition from beetles.
For five of the invertebrate taxa, observations varied strongly throughout the day. Slugs, T. limbatus and L. lunulatus were most abundant late at night and true flies in the evening. Although not measured in this study, spore production of F. pinicola is highest during night (Haard & Kramer, 1970;Hågvar, 1999;Norros et al., 2023), probably to minimise mortality factors like sunlight (Lagomarsino Oneto et al., 2020). Night activity due to spore feeding has been suggested earlier for adult fungivorous beetles (Hågvar, 1999;Paviour-Smith, 1965), and this might explain the observations of invertebrates at night in our study. Further, invertebrates might better avoid predators, in particular birds, by limiting their activity to the night.
From our results, it is obvious that the high abundance of G. boleti is important for shaping the invertebrate community on F. pinicola fruit bodies. When a cluster of G. boleti was present, the predatory beetles I. binotata and L. lunulatus were more common, while the fungivorous beetles T. limbatus and P. ferruginea were less common. The latter may be competitively excluded by G. boleti that, despite their small size, is present in large numbers and may compete for the same resources-spores or other fungal tissues on the hymenial surface (Hågvar, 1999;Hågvar, 2018;Staniec et al., 2016).

Visitation frequency and duration
There was a clear difference among invertebrate taxa in how long they visited the fruit bodies of F. pinicola. Spiders and the two fungivorous beetles T. limbatus and P. ferruginea stayed the longest, which could probably be explained by their foraging strategies. Spiders are sit-and-wait predators, typically hiding behind a silk web before ambushing their prey. The beetles spend their whole life cycle feeding at, or close to, fungal fruit bodies; the larvae feed on brown-rot mycelium or wood, and the adults feed on the hymenial surface or dwell underneath bark adjacent to fruit bodies Schigel et al., 2006). Indeed, one individual of T. limbatus was even observed in successive images over more than 20 h.
Slugs, true flies and I. binotata beetles stayed on the hymenium for half an hour on average. True flies are an incredibly diverse group, but we have not identified them to species or families. Some of them may be fungivores, while other may be predators, parasites or visiting the fruit body for courtship or oviposition (Jakovlev, 2012;Ševčík, 2010;Ulyshen, 2018). Some of the slugs left clear grazing marks on the hymenium when tracking them over successive images, suggesting that many of them are fungivores. Ipidia binotata is classified both as a fungivore and a predator from different sources (Seibold et al., 2015;SLU, 2022), perhaps reflecting an opportunistic or omnivorous foraging strategy.
The fastest visitors were the centipedes and L. lunulatus beetles, which makes sense as they are pursuit predators. However, as most of their visits were recorded only once, and the cameras only took pictures every 10 min, these estimates may be inflated and the true duration of their visits could be even shorter. To gain a better understanding of their activity patterns, we therefore suggest future image-based monitoring studies to increase time-lapse frequency.
Using image data from time-lapse cameras, we managed to get an estimate of the visitation frequency of different invertebrate groups.
Over the course of the study period, 700 individual visits by beetles were made to F. pinicola fruit bodies and more than a 100 visits by each of the other invertebrate groups. As the activity patterns of the T A B L E 3 Incidence rate ratios (± SE) from generalised linear mixed models of predictors (see Table 1) explaining variation in invertebrate observations on Fomitopsis pinicola fruit bodies. Araneae ----- Note: Incidence rate ratios show the ratio of a one unit change in the value of a predictor. A ratio higher than 1 means expected observations increase with the predictor, whereas ratios lower than 1 means that they decrease. For example, for Diptera, 1.03 is the mean ratio when G. boleti clusters are present to when they are not, that is, the model predicts a 3% increase in Diptera observations when G. boleti clusters appear. Model-averaged coefficients (natural averaging method) are used in all but three models and distribution was either Poisson or or Binomial (Table S3. invertebrates seemed to correlate with the fungal sporulation period, that is, in spring and at night (Norros et al., 2023;Nuss, 1986), these visitors could potentially be dispersing the spores of F. pinicola. In the same study area, Lunde et al. (2023) found that all beetles visiting  F I G U R E 4 Relative variance contribution of predictors (a, c, e and g) and predicted observations at different times of day (b and d) and in the presence/absence of G. boleti clusters (i.e. >10 individuals in an image; f, h) for three beetle species and Diptera on fruit bodies of Fomitopsis pinicola. Based on generalised linear mixed models (Table 3), which are detailed in Materials and Methods. Icons by Henrik Aubert.
F. pinicola fruit bodies could disperse viable spores of that species.
Although they did not test the vectoring capacity of other invertebrates, it is likely that some spores would survive transport by them, as well. Indeed, a wide variety of organisms, ranging from microfauna to propagules of plant and fungi, could be dispersed by slugs (Turchetti & Chelazzi, 1984;Türke et al., 2018). A recent study even saw improved germinability of spores from three genera of wooddecay fungi that had passed through slugs' digestive tracts compared with spore print controls (Kitabayashi et al., 2022). To understand the relative importance of potential vectors, however, we need to estimate the number and duration of visits, known as the quantitative component in the dispersal effectiveness framework commonly used for seed dispersers (Birkemoe et al., 2018;Schupp, 1993;Schupp et al., 2017). Here, we show that the use of camera monitoring is a feasible and time-efficient way to obtain large amounts of image data that can be used to estimate these parameters.

Concluding remarks
In this study, we have used time-lapse cameras to identify and quantify the invertebrate community on the hymenium of F. pinicola fruit bodies. Beetles were the most common group, in particular the fungivore T. limbatus, followed by spiders, true flies, slugs and centipedes. The invertebrate activity exhibited strong temporal variation (seasonal and diurnal), but was also determined by temperature and the presence of competitors/prey items (G. boleti). Most observations were at night and early in the season, and our results suggest that G. boleti have a positive impact on predatory beetles, and a negative impact on fungivorous beetles. The duration of visits was longest for the two fungivorous beetles and shortest for the predatory L. lunulatus and centipedes.
To our knowledge, this is the first study using camera technology to monitor invertebrates on fruit bodies of wood-decay fungi. The technology is a time-efficient way of obtaining quantitative estimates of activity and visitation compared with traditional methods. However, our method recorded a lower diversity than one study using manual observations in the same area (Hågvar, 1999). This could be because smaller invertebrates were not possible to identify on the images or because only 19 fruit bodies were monitored. Invertebrate observations varied a lot between fruit bodies, suggesting that future studies should include more fruit bodies. Further, if the aim is to obtain proper visitation estimates of faster invertebrates such as predators, the cameras should take pictures at shorter time intervals, for example, 1-min intervals, which are used in pollination studies (e.g. Alison et al., 2022). However, more precise estimates will come at the cost of more image data. As image processing can be very timeconsuming, careful evaluation should be undertaken to balance the trade-off between precision and processing efforts. One way to reduce the number of images could be to monitor for shorter time periods, for example, early in the season when activity is at its highest.
Deep learning could also be used to automatically detect and identify insects in the images for future studies (Høye et al., 2021), and the annotations produced in the current study could serve as data to train a generic invertebrates-on-fungi detector. Although the existing image quality will probably not have sufficient accuracy to identify smaller species, alternative cameras with higher image resolution could be used. A benefit of this study system is that the background of the image frame (i.e. the F. pinicola hymenium) is quite flat, pale and homogeneous. We believe that further advances into the use of timelapse cameras on fruit bodies could be valuable to understand both the identity and ecology of wood-associated invertebrates and fungi.

SUPPORTING INFORMATION
Additional supporting information can be found online in the Supporting Information section at the end of this article.