The Hawaiian Islands are a tropical, oceanic archipelago consisting of eight current high islands (Hawai‘i, Kaho‘olawe, Kaua‘i, Lāna‘i, Maui, Moloka‘i, Ni‘ihau, and O‘ahu) in the northern Pacific Ocean. The islands are isolated from the closest land mass, North America, by more than 3,500 km, and have high elevation and rainfall heterogeneity between and within islands. For example, on Hawai‘i Island, average rainfall varies from <300 mm/year in Kawaihae to > 7,500 mm/year at the Makahanaloa rain gauge, and average annual temperature varies from 24°C in Kawaihae at sea level to 4°C on Maunakea at > 4,000 m. In this study, the islands of Ni‘ihau and Kaho‘olawe were excluded from all analyses due to lack of environmental data and historically few collections (e.g., one meeting our criteria from Ni‘ihau; zero from Kaho‘olawe for online abundance data). Additional information on the islands included in this study are available in Supplementary Table 1.

For our richness analysis, we first determined all Agaricomycetes species known to occur in the Hawaiian Islands based on authoritative sources. We then tabulated which month(s) basidiomata of each species were reported in, and additional environmental and life history data from these same sources. The species list of Hawaiian Agaricomycetes was tabulated from an unpublished list of Hawaiian Agaricomycotina compiled by Drs. Dennis Desjardin and Don Hemmes from ~1992 to 2007 and used by Mueller et al. (2007) for their analysis of Hawaiian and global macrofungal biodiversity. We checked primary literature, field guides, aggregated checklists, MyCoPortal.org, and GenBank to verify each species on this list occurred in the Hawaiian Islands, and added species documented since 2007, or those missing. To be included, a reference needed to explicitly state the species occurred in the islands; environmental sequencing studies were not considered. Species documented with incontrovertible DNA evidence and metadata were included, even if unpublished (e.g., Psilocybe cyanescens). Mycobank (Crous et al., 2004) and IndexFungorum (www.indexfungorum.org) were used to update fungal names and primary literature was consulted in case of disagreement. Names were compiled at the species level without consideration for infraspecific variation (e.g., subspecies or varieties).

For species on the list, months in which basidiomata were found, different islands they were found on, origin as native (indigenous to the Hawaiian Islands; not inferred to be established via human-mediated dispersal) or non-native (inferred to be established via human-mediated dispersal), and elevation and rainfall information were recorded. To determine a species origin as non-native or native we followed statements in the literature when available or inferred this information from its association with solely native vegetation or habitats. We used a conservative approach, not assigning species to native origin in ambiguous situations in which the species occurs in both native and non-native habitats, or mixed vegetation, unless additional evidence existed. Rainfall and elevation information for each island were compiled based on Hawaiian habitat zones by elevation (lowland < 1,000 m, montane 1,000–2,000 m, and subalpine/alpine > 2,000 m) and rainfall (wet > 2,500 mm, mesic 1,000–2,500 mm, and dry < 1,000 mm) similar to Gagné and Cuddihy (1999) using the Rainfall Atlas of Hawai‘i (Giambelluca et al., 2013) and web resources (https://www.freemaptools.com/elevation-finder.htm) based on reported collection locations. We used a liberal approach for environmental variables, allowing values approximately between two elevation or rainfall categories to be included in both as some location information is not specific (e.g., “Honolulu” or “slopes of Mauna Kea”) and could encompass multiple zones. Additionally, the Hawaiian Islands are known for steep environmental gradients where rainfall can quickly change. We note that temperature throughout the year in the islands, like other tropical locations, is relatively stable and the largest temperature swings generally come from daily cycles as opposed to annual cycles (Hartshorn, 2013). As elevation increases, average air temperatures decrease (Giambelluca et al., 2014).

For our abundance analysis, we first obtained online observation and/or collection records of Hawaiian fungi identified to the class Agaricomycetes (whether to species, or a higher level), then used their associated geospatial information to extract environmental metadata. We searched for collections and/or observations of Agaricomycetes species in the Hawaiian Islands through May 25, 2021 from iNaturalist.org, MushroomObserver.org, and Mycoportal.org. Observation and collection data were cleaned considering the shortcomings of fungal repository data outlined in Hao et al. (2021): observations or collections that were duplicates, lacked geospatial or temporal data, or fell outside the Agaricomycetes were removed. To be included, associated geospatial coordinates needed to fall on land within the Hawaiian Islands. ESRI ArcGIS (version 10.7.1) was used to visualize observations and collections throughout the Hawaiian Islands and bin these into rainfall and elevation categories. Rainfall data was again obtained from the Rainfall Atlas of Hawai‘i, elevation data was obtained from the National Centers for Coastal Ocean Science Observations (NCCOS, 2022), and temperature data from the Climate of Hawai‘i (Giambelluca et al., 2014). Observations and/or collections that matched species identified as native from the list used for our richness analysis were also indicated as native in our abundance analysis.

Because the habitat in the Hawaiian Islands is highly heterogeneous, correlating increased basidiomata production with environmental variables across the entire archipelago, or even a single island, is difficult. To better investigate whether basidiomata production correlated with average monthly elevation or rainfall, we subset our abundance data into categories by elevation, annual rainfall, and island, as in our richness analysis. Maui, Moloka‘i, and Lāna‘i were considered as a single entity, Maui Nui, as is often done in biogeographical studies due to the connectivity of these islands in the recent past (Price and Elliott-Fisk, 2004). Bins with < 200 data points were excluded. This left wet, lowland habitats on Hawai‘i and O‘ahu; mesic, lowland habitats on Hawai‘i, O‘ahu, and Kaua‘i; and dry, lowland habitats on O‘ahu. Point interpolations were performed in ArcGIS using default settings to determine where the highest density of observations and collections occurred. Average monthly rainfall and temperature from the interpolated highest density region were then obtained from Giambelluca et al. (2013, 2014).

For both our richness and abundance datasets, the following species were excluded from our analyses as detection may occur year-round due to long-lasting, persistent structures: species characterized as polyporoid (e.g., Polyporales; Hymenochaetales spp.), corticioid (e.g., Corticiaceae spp.), gelatinous (e.g., Auriculariales spp.), or species with otherwise persistent basidiomata (e.g., Nidulariaceae).

We used chi-square tests to determine if Agaricomycetes reproductive phenology varied by month throughout the entire Hawaiian archipelago. The null hypothesis was that no significant difference existed between months in richness or abundance counts across all Agaricomycetes species, and native species only. Pearson's correlation coefficient tests were performed between monthly basidiomata counts of richness and abundance in binned environmental categories with ≥ 200 abundance data points on Hawai‘i Island, Kaua‘i Island, and O‘ahu Island, and the average monthly rainfall and temperature corresponding to the densest location of their abundance. The null hypothesis was that no significant correlation existed between months of increase in basidiomata richness or abundance in a binned habitat (e.g., wet and low elevation) on an island and the monthly temperature or rainfall at the densest location of their abundance. Analyses were conducted using R (R Core Team, 2021).

Based on our literature review, we found 643 known Agaricomycetes occur in the Hawaiian Islands (Supplementary Table 2). Of the 643 species, 321 had associated seasonality data and lacked long-lasting or persistent reproductive structures to be included in our analysis. Of these, 33 species were considered native in our study and included in the native-only richness analyses.

We downloaded 11,292 observations or collections from iNaturalist.org, MushroomObserver.org, and Mycoportal.org through May 25, 2021. After cleaning, we found 3,285 unique observations or collections meeting our study criteria (Supplementary Table 3). The 3,285 observations or collections contain 478 unique names, although due to the high number of records identified above the species level, the number of unique species these records constitute is unknown. Of these, 169 observations and/or collections of 22 unique species were considered native and included in the native-only abundance analyses.

Overall Agaricomycetes richness shows a peak in January with lower diversity from February to June. Richness increases in July and continues at an elevated rate through December (Figure 2A). A significant difference in richness between months was found (X² = 96.7, p < 0.01, df = 11). Native-only richness show that February to April is a period of lower mushroom diversity with June and October also having low diversity (Figure 2B). For richness counts of native species only, the null hypothesis that there is no difference in richness between the 12 months cannot be rejected (X-squared = 19.2, p = 0.06, df = 11).

Overall abundance data shows peaks in January, March, and October to December (Figure 2C). A significant difference in species abundance between months was found (X² = 309.1, p < 0.01, df = 11). Native-only abundance data show that January has very high abundance, February to April have low abundance, and May to December have a low baseline of consistent abundance counts (Figure 2D). A significant difference in species abundance between months was found considering native species only (X² = 140.3, p < 0.01, df = 11).

Richness data by rainfall and elevation categories generally show a trend of high diversity in January, low diversity from February to April, May, or June, then high diversity for the remainder of the year with outliers being dry, and subalpine/alpine locations that also having high diversity in March (Supplementary Figure 1). Abundance data by rainfall amount and elevation do not follow an obvious pattern, although abundance is generally highest from October through March (Supplementary Figure 2).

Pearson's correlation coefficient tests show significant positive correlations between basidiomata abundance and average monthly rainfall in lowland, mesic environments on Kaua‘i (r = 0.88; p-value < 0.01) and O‘ahu (r = 0.66; p = 0.02), and lowland, dry environments on O‘ahu (r = 0.83; p-value < 0.01). Correlations between abundance of basidiomata and average monthly rainfall or temperature in lowland, wet environments on Hawai‘i Island and O‘ahu, and lowland, mesic environments on Hawai‘i Island were not significant. Correlations between species richness and average monthly rainfall or temperature in either elevation or annual rainfall category were not significant (Figure 3; Supplementary Table 4).

Despite a warm climate and areas with consistent year-round rainfall, we found Hawaiian mushrooms generally exhibit seasonality patterns in richness and abundance, except in native-only species richness where we could not reject our null hypothesis (Figure 2). Our results incorporating greater taxonomic breadth agree with the prior findings of Stallman (2019) for richness in the family Agaricaceae, which is primarily represented by non-native species in the islands. Comparing our results of native abundance data to that of Hemmes and Desjardin (2002), we found the month of January to be by far the most abundant, which was one of their least abundant. Additionally, we found native-only basidiomata production is as abundant from May–June as July–December, but May and June showed low abundance in their dataset. These differences could be due to overall low detection of native species, uneven collecting efforts in different locations, or species selection. For example, Hemmes and Desjardin (2002) found Rhodocollybia laulaha to be the most abundant native species in their study, but its origin is now uncertain after it was found in the American tropics Keirle et al. (2010). Its origin was assigned as unknown in this study, and therefore not included in our native-only analysis.

Our hypothesis that abundance and richness of mushroom species would not correlate to monthly average rainfall or temperature is supported when considering species richness and elevation, but not for all instances when considering abundance and rainfall. In lowland, dry and mesic environments on O‘ahu, and lowland mesic environments on Kaua‘i, basidiomata abundance is correlated to increased rainfall. Rainfall as a limiting factor to increased abundance in areas lacking consistently high year-round precipitation is intuitive, and follows anecdotal observations and prior studies (e.g., Chacón and Guzmán, 1995). In wet areas with the highest concentration of observations and collections on Hawai‘i and O‘ahu, monthly rainfall is a minimum of 200–300 mm, and increasing rainfall above this baseline does not lead to increased monthly abundance or richness (Figures 3A,C). Increased rainfall in lowland mesic areas on Hawai‘i Island is not positively correlated with increased abundance in basidiomata production despite O‘ahu and Kaua‘i seeing increased production during wetter months in mesic locations (and O‘ahu in lowland dry areas). The reason for this is unclear, but Hawai‘i is a large island (10,430 Km²) and by only examining rainfall at the location interpolated to have the highest number of mesic observations and/or collections, accuracy may be lower than for the smaller islands of Kaua‘i (1,456 Km²) and O‘ahu (1,545 Km²) (Supplementary Table 1).

Several limitations exist in our study. First, our data incorporates several broad categories. Temporally we examined monthly instead of weekly; elevation by 1,000 m, and rainfall by 1,000–1,500 mm. Additionally, we have not considered how seasonality of Agaricomycetes may differ between substrates (e.g., wood or soil) or ecological roles (e.g., saprotroph or symbiotroph), which have been important explanatory variables in a Japanese oak forest (Sato et al., 2012) and across Europe (Andrew et al., 2018).

Second, the majority of Hawaiian Agaricomycetes lack a DNA barcode sequence to help verify species identity and allow quick biogeographical comparisons with other sequence data in GenBank. Our understanding of which species are native may change as more data become available. Additionally, in our abundance analysis, collections and/or observations not identified to the specific level were deemed non-native, although a small portion of these may constitute native species.

Third, our abundance data aggregated from MyCoPortal.org, iNaturalist.org, and MushroomObserver.org may hold biases such as focus on or near population centers, and increased documentation of conspicuous (large, colorful) over inconspicuous species (Gonçalves et al., 2021). Non-native mushrooms often occur near population centers and may have additional water inputs from irrigation. While we believe that most Hawaiian Agaricomycetes collections in herbaria have been digitized (at SFSU, BISH, and ARIZ), some collections at these locations and other herbaria may not be digitized, representing an untapped data source in our abundance analysis that was not captured via MyCoPortal.org searches.

Finally, our data is biased toward the recent past: only 14% of collections or observations in our abundance analysis occurred before 1990, and 29% were before 2000. As climate change has been shown to change macrofungal reproductive phenology (Andrew et al., 2018), some signal in seasonality patterns may be lost due to changes over time in a changing climate. With our data's strong bias toward the present, we are likely seeing less of this effect.

The six Hawaiian fungi that have so far been assessed for the IUCN Red List and were found to be threatened are the endangered Hygrocybe noelokelani and Hy. pakelo (Vellinga, 2017b, 2019d), and the vulnerable Callistosporium vinosobrunneum, Humidicutis peleae, Humidicutis poilena, and Hy. lamalama (Vellinga, 2017a, 2019a,b,c). Regarding richness of these species, some show strong seasonality (or rareness in general) with C. vinosobrunneum only being found November to January and Hu. poilena only in November, while other species, such as Hu. peleae, occur throughout much of the year. These threatened species are most abundant in January, November, and December (Figure 4). Although current funds are not allocated to Agaricomycetes conservation by the state or federal government, understanding seasonal basidiomata production could help managers plan field efforts to monitor these species or find additional populations, as is done for imperiled Hawaiian plants (e.g., Kawelo et al., 2012). Although morphological identification of fungi is not always possible (Taylor et al., 2006) or may require experts or DNA data, some threatened species in the Hawaiian Islands such as Hy. noelokelani are easily recognizable without molecular identification (Vellinga, 2017b).

Molecular tools can also be used to monitor populations of threatened fungi (Gordon and Norman, 2015). The first barrier to detection of threatened Hawaiian fungal species via these methods is generating reference sequences for the threatened species; currently none have an associated DNA barcode (Schoch et al., 2012). Additionally, substrate would need to be carefully considered as many threatened Hawaiian species form basidiomata on mosses, making it unclear if high throughput sequencing of soil would lead to detection if nearby populations were present.

We note that all threatened Hawaiian Agaricomycetes species are most often found in montane, wet, and mesic habitats which are dominated by the native ‘ōhi‘a tree (Metrosideros polymorpha). ‘Ōhi‘a are being killed by the introduced pathogens Ceratocystis lukuohia and C. huliohia, together known as Rapid ‘Ōhi‘a Death (ROD). The disease has killed over one million trees in the islands (ROD working group, personal communication) and will likely have a detrimental impact on associated native plant communities (Fortini et al., 2019). While direct evidence of native fungal population declines from this disease is lacking, monitoring of threatened fungi is needed to determine demographic and distribution changes as Hawaiian forests change due to the effects of ROD.