Phylogeny, Taxonomy and Evolutionary Trade-Offs in Reproductive Traits of Gomphoid Fungi (Gomphaceae, Gomphales)

Although functional ecology is a well-established field, our understanding of the evolutionary and ecological significance of the reproductive traits in macrofungi is still limited. Here, we reconstructed a phylogeny tree of gomphoid fungi in the narrower sense, including the species of the genera Gomphus and Turbinellus and used it to uncover the evolution of reproductive traits. Our analyses indicated that fungal fruit bodies and spores did not enlarge at a steady rate over time. Early gomphoid fungi essentially maintained their fruit body size, spore size and spore shape through the Mesozoic. In the Cenozoic, gomphoid fungi acquired significantly larger and more spherical spores by simultaneously expanding in length and width, with the fruit body size first decreasing and then enlarging. We argue that these trade-offs were driven by the effect of biological extinction and the dramatic climate changes of the Cenozoic. Gomphoid fungi initially increased in spore size and fruit body number as extinction survivors filled vacant niches. Both fruit bodies and spores eventually became larger as ecosystems saturated and competition intensified. One new species of Gomphus and nine new species of Turbinellus are described.


Introduction
Gomphoid fungi are conspicuous groups of the family Gomphaceae [1]. Systematically, the position of this group has been unclear, and the monophyly of this group has not been statistically supported [2][3][4][5]. Earlier molecular phylogenetic studies demonstrated that Gomphus sensu lato (Gloeocantharellus Singer, Gomphus Pers., Phaeoclavulina Brinkmann and Turbinellus Earle) and Ramaria sensu lato were paraphyletic within the Gomphaceae, and, consequently, a number of species published in Gomphus were transferred into Gloeocantharellus, Phaeoclavulina and Turbinellus. Turbinellus, once regarded as a synonym for Gomphus [2,6,7] and Gomphus s. str., have been recognized as independent genera [1,3]. However, a number of species have been published in Gomphus but do not have the current diagnostic morphological features of Gomphus s. str., causing confusion when the names of gomphoid fungi are published.
Gomphoid fungi show conspicuous diversity in the size and shape of their fruit bodies and spores. Although there is no solid evidence to demonstrate that this variation in reproductive traits is the result of an evolutionary adaptive response, previous studies [8][9][10][11] have shed light on some selective pressure from the environment on reproductive traits. Gomphoid fungi produce spores in the fruit bodies for reproduction. However, the production of these propagules is often costly; therefore, trade-offs among traits related to reproduction (e.g., between size and number of fruit bodies/spores, between reproduction 2 of 35 and growth) become common, allowing effective reproduction at the minimal costs in a specific environment [8]. As a consequence, the reproductive syndromes that co-occur in species should be related to the trade-offs of reproductive traits in specific environments. There are commonly held beliefs that fungal species with larger fruit bodies can produce larger spores and larger spores are more spherical [12][13][14]. However, our understanding of possible correlations of fruit body size with other reproductive traits in macrofungi has been limited to current resources and environments, regardless of the evolutionary history.
Most species of gomphoid fungi are ecologically important ectomycorrhizal partners of Fagaceae, Myrtaceae and Pinaceae plants [1, 15,16]. A mutualistic (ectomycorrhizal) lifestyle allows gomphoid fungi to receive carbon from their host plants and therefore adapt to predictable resources, which provides degrees of freedom for reproduction [8,17]. Large fruit bodies, which generally offer advantages by producing more spores and dispersing them farther [12], have a longer life expectancy and more sporulation events [18][19][20] and reduce the chance of desiccation and pathogens [8]. Conversely, among the advantages of small fruit bodies, are their faster maturation; thus, they have greater flexibility in adjusting their reproductive investment to fluctuating resources and climatic variation [8,9]. Likewise, large spores tend to have higher fitness and allow prolonged dormancy but are more costly to produce and more difficult to disperse than small spores [13,21,22]. If the resources a species can allocate to reproduction are limited, each species faces the challenge of either investing in a few large or more numerous small fruit bodies and likewise investing in a few large or more numerous small spores. However, how reproductive trade-offs of fungi evolved and which driving forces have shaped these trade-offs are not clear.
In this study, we sampled extensively from known gomphoid fungi and carried out a phylogenetic trait analysis in a phylogenetic context. Our aims are (1) to establish the phylogeny of gomphoid fungi and reveal their phylogenetic diversity; (2) to explore the relationships among the fruit body size, the spore size and the spore shape by incorporating information on the phylogenetic relationships of gomphoid fungi; (3) to test whether gomphoid fungi have reproductive trade-offs and to elucidate factors that might have driven these trade-offs in evolutionary history.

Fungal Data
Samples of the gomphoid fungi in the broader sense consisted of 101 collections from the tropical, subtropical and temperate regions of different parts of the world. In total, 108 collections representing 33 species of gomphoid fungi in the broader sense and six species of the other taxa were included in this study. Among them, 184 sequences from 71 collections were newly generated, and the remaining 58 sequences from 30 collections were retrieved from the database of GenBank (accessed on 9 March 2023). Voucher information and GenBank accession numbers for each collection are provided in Table S1. Gomphus is abbreviated as G. and Turbinellus as T.
Macroscopic characteristics were described based on field notes, photographs and literature studies. Most dried specimens were deposited in the Cryptogamic Herbarium of the Kunming Institute of Botany, Chinese Academy of Sciences (HKAS). Macromorphological measurements were conducted with a ruler in the field when the basidiome is mature and fresh. In the descriptions of pileus, some species of Gomphus are irregularly fan-shaped; the values of the widest part of the mature basidioma are regarded as their pileus diameter. Color codes designated in the descriptions are from Kornerup and Wanscher [23]. Microscopic features were observed and measured on dried material with light microscopy and a scanning electron microscope [24]. In the descriptions of basidiospores, the abbreviation (n/m/p) means n basidiospores measured from m basidiomata of p collections; the sizes for basidiospores are given using a range notation of the form (a-) b-c (-d): The range b-c contains a minimum of 90% of the measured values. Extreme values (a, d) are given in parentheses. Q is used to present "length/width ratio" of a spore in side view, and Qm is the mean Q of all basidiospores ± sample standard deviation.
Following our study and previous ones [1, 9,[12][13][14]25], the character data, including the maximum cap diameter as an approximation for mushroom size, the volume of basidiospores (4π/3 × (length/2) × (width/2) 2 ) as an approximation for basidiospore size, and the Qm (mean of the spore length divided by width) as an approximation for basidiospore shape, were extracted for each record. Since the maximum size of the fruit body is evolutionarily conserved [5,9,12], the maximum cap diameter was used in our analyses following previous studies [9,26]. In order to obtain the maximum cap diameter of each species, 3-5 mature basidiomata per collection were measured in the field, and the maximum value was extracted. Given the skew of spore measurements within species, a 90% interval range was used in our analyses [27]. Therefore, the mean of the minimum and maximum is a reliable measure for our cross-species analysis. The mean of spore size was used for all further analyses. Some of the included taxa lacked the character data mentioned above, which were replaced with the characteristic values of the species reported in the literatures.
The characteristic values are included in Table S2.
Furthermore, correlations between biological traits such as fruit body size and spore volume might be influenced by common ancestry. The lambda model in the package phytools ver. 1.2 [28] was used to detect the phylogenetic signals. Both spore volume (λ = 0.998) and fruit body size (λ = 0.869) showed strong phylogenetic signals. Then, the function "pic" [29] in the package "Ape" [30] was used to detect phylogenetic correlations between traits. The analysis showed that there was a significant phylogenetic correlation between the spore volume and the fruit body size (pr = 0.00476**). Thus, in this study, the phylogenetic spore quotient (PSQ), a measure of relative spore size compared with its value predicted by allometry and phylogeny, was used to convey how small or large a species spore is as compared to that of other species of similar fruit body size. The PSQ was calculated as Si/Sc, which corresponds to the ratio between the actual spore size (Si) and the expected spore size (Sc). To create the PSQ equation for our sample, we used the PGLS regression [31] of the log10(spore volume) vs. log10(fruit body size) to obtain Sc (the expected spore size for a fungus of its size), which is equal to 0.6307 (fruit body size) 0.7415 for our topology.

DNA Extraction, Amplification and Sequencing
Total genomic DNA was extracted from silica-dried material or occasionally from herbarium fragments using the Ezup Column Fungi Genomic DNA Purification Kit (Sangon Biotech, Shanghai, China) following the manufacturer's protocol. Three nuclear markers, including the internal transcribed spacer (ITS), the translation elongation factor 1-α gene (tef1-α) and the large subunit of nuclear ribosomal RNA gene (nrLSU), were selected for the phylogenetic study. The ITS regions were amplified with primers ITS1F and ITS4 [32,33]. The nrLSU sequences were amplified with primers LROR and LR5 [34]. The tef1-α sequences were amplified with primers 983F and Efgr [35]. For some herbarium specimens with degraded DNA, the internal primers of the ITS gene 5.8S and 5.8SR and the internal primers of the tef1-α gene 1567R were employed when amplification of the larger region was unsuccessful [34,35]. All PCR conditions followed Li et al. [24]. PCR products were purified and sequenced by TSINGKE Biological Technology (Kunming, China). Sequencher ver. 4.1 (Gene Codes Corp., Ann Arbor, MI, USA) was used to assemble and edit contiguous sequences.

Sequence Alignment and Phylogenetic Analysis
Sequences obtained for each marker were initially aligned using MAFFT ver. 7 [36] and then manually adjusted in BioEdit ver. 2.0 [37]. The concatenated alignment is available at TreeBASE (http://purl.org/phylo/treebase/phylows/study/TB2:S30369, accessed on 11 May 2023). jModelTest ver. 0.1.1 was used to select the best-fitting likelihood model for maximum likelihood (ML) and Bayesian analyses [38]. The Akaike information criterion (AIC) was used to select among models instead of the hierarchical likelihood ratio test [39]. Model GTR+G+I was finally selected for all phylogenetic analyses.
For each marker and the simultaneous analysis of all nucleotide characters, maximum likelihood tree searches and ML bootstrapping (BS) were conducted using the web server RAxML-HPC2 on XSEDE ver. 8.2.12 on the Cipres web server [40], with 1000 rapid bootstrap analyses followed by a search for the best-scoring tree in a single run.
Bayesian inference (BI) was conducted using MrBayes ver. 3.2.7a [41]. Four Markov chain Monte Carlo chains were conducted, each beginning with a random tree and sampling one tree every 1000 generations of 10,000,000 generations. Convergence among chains was checked using Tracer ver. 1.7.1 [42], and the first 25% were discarded as burn-in to ensure that stationarity in log-likelihood had been reached. The remaining trees were used to calculate a 50% majority-rule consensus topology and posterior probabilities (PP).

Divergence Time Estimations
For divergence time estimation of the gomphoid fungi, we used BEAST ver. 1.10.4 [43]. For the BEAST analysis, the uncorrelated lognormal clock model was used with nucleotide substitution model GTR+G+I and birth-death speciation (uniform prior from 0 to 10 for growth rate and 0 to 1 for relative death rate; the initial value was 1 for growth and 0.5 for relative death) was specified for the dataset. Four secondary calibration ages [4] were employed in the Bayesian analysis using the normal distribution (Table S3). Two runs of four Markov chain Monte Carlo chains with 100 million generations were run with sampling every 10,000 generations. The effective sample size (ESS) of all parameters in the combined runs was >200. Log Combiner ver. 1.10.4 and TreeAnnotator ver. 1.10.4 were used to obtain a consensus tree [43]; the first 20% were considered burn-in, and the remaining 8000 trees that were used to generate maximum clade credibility tree were combined.

Diversification Rate Analyses
Diversification rates were estimated with BAMM ver. 2.5.0 [44] which sampled from a given time-calibrated tree using a reversible jump MCMC. The BEAST consensus specieslevel tree without outgroups was applied, and 20 million generations were run, sampling every 1000 generations. Prior parameters were optimized using the "setBAMMpriors" function in BAMMtools ver. 2.5.0 [45]. Post-run analysis and visualization used the BAMMtools ver. 2.5.0 in R ver. 4.1.3 [46]. Accounting for incomplete taxon sampling, the sampling fraction was specified as 0.9, representing a sampling coverage of about 90% of gomphoid species. In addition, we visually assessed the timing and tempo of diversification by constructing lineages through time (LTT) plots using the function "ltt" in the phytools ver. 1.2 [28]. One consensus tree annotated from the BEAST analysis was used to calculate the LTT plots.

Ancestral Character Reconstruction and Rate of Evolution
We used our phylogenetic topology to reconstruct ancestral characters and assess trends across the tree in log10(fruit body size), log10(PSQ), log10(spore size) and log10(Qm). We used BayesTraits ver. 4.0 [47] and loaded the time-calibrated tree and the character file to be traced in the "variable rates model" and the options "independent contrast" and "MCMC" with 100 million iterations and 10 million of burn-in. Upon completion of the analysis, we opened the file txt.log with Tracer ver. 1.7.1 [41] to visually verify that the analysis had reached convergence and stationarity. Another output file with the extension.txt.VarRates was processed in the program PPPostProcess [48], which allowed us to obtain a new text file used for downstream analyses.
We used the parameter 'median scalar' to scale the branches of the tree, so the variable rates of individual branches could be considered in the ancestral states reconstructions [48]. Then, we employed the function "anc.Bayes" in the package phytools ver. 1.2 [28], which uses the Bayesian MCMC approach to reconstruct the ancestral states of each variable on our phylogenetic trees with 20 million generations of running and 2 million of burn-in. Subsequently, we used the "plotBranchbyTrait" function in phytools ver. 1.2 to trace the rate of evolution through time. Additionally, we plotted the rate of evolution and of each trait through time using the function "ggboxplot" in the package ggplot2 ver. 3.4.0 [49].

Variations in Reproductive Traits through Time
The ancestral state of fruit body size, PSQ, spore size and Qm with age were obtained from "anc.Bayes" in the package phytools ver. 1.2 [28] to plot the characters changed through time by using the function "ggboxplot" in the package ggplot2 ver. 3.4.0 [49]. To plot the data through time, we binned species into 10-million-year temporal bins, which include the average of all the values for that specific temporal interval. A loop automatically classified each average age in the appropriate 10-million-year bin [50]. A column was included with the average of all bins and calculated the mean, standard deviation and total number of specimens per bin to obtain the interval of confidence per bin [50]. Additionally, the function "ggboxplot" from the package ggplot2 ver. 3.4.0 was used to visualize differences in fruit body and spore values among Mesozoic, Paleogene, Neogene and Quaternary taxon samples.

Phylogenetic Regressions
Phylogenetic generalized least squares (PGLS) regressions were used to determine the correlations among fruit body size, spore size, PSQ and Qm in the Mesozoic, Cenozoic and present, with associated models of Brownian motion, the Lambda model, Early Burst and Ornstein-Uhlenbeck: (a) log10(fruit body size) vs. log10(PSQ); (b) log10(fruit body size) vs. log10(spore size); (c) log10(fruit body size) vs. log10(Qm); (d) log10(spore size) vs. log10(Qm). The function "gls" and "anova" in the nlme package ver. 3.1 [51] was used to select the best model for each regression based on the lowest AIC value. Eventually, model lambda was determined for all four regressions. The package ggplot2 ver. 3.4.0 [49] was used to plot the regressions with the "ggplot" function.

Phylogeny of Gomphoid Fungi
A total of 242 sequences from 39 species were included, of which 62 of ITS, 66 of nrLSU and 56 of tef1-α were newly generated in this study. Comparisons of three topologies from the ML analyses of the individual nuclear markers identified no well-supported conflicts. Thus, the three nuclear datasets were concatenated. The topologies of the ML and BI trees based on the concatenated datasets showed no conflicts and generally increased support values; thus, only the tree inferred from ML analysis was displayed ( Figure 1).
Based on our molecular analyses (ML and BI), Gomphus and Turbinellus are each distributed in the different monophyletic clade with strong support (MLBS: 92%, BIPP: 0.99 for Gomphus; MLBS: 99%, BIPP: 0.99 for Turbinellus). The clade that combined both Gomphus and Turbinellus was resolved as a monophyletic group independent of the Ramaria sensu lato with strong support (MLBS: 97%, BIPP: 0.99). Furthermore, Gloeocantharellus, once included in gomphoid fungi in the broader sense [1], was resolved as a group paraphyletic to the gomphoid fungi in the narrower sense, including Gomphus and Turbinellus.
Our three nuclear datasets combined resolved 26 collections of Gomphus into five clades and one single-accession clade including seven species (with G. brunneus BR034190-46 clustered in the lineage of G. clavatus); 66 collections of Turbinellus into 11 clades and three single-accession clades including 14 species, of which one species of Gomphus and nine species of Turbinellus from China are putatively new to science; and two species of Gomphus originally described from southwestern China which should be transferred to Turbinellus. Finally, 15 species (including three known, two new combinations and 10 novel species described here) from China were recognized and elucidated mainly based on the phylogenetic analysis. Further morphological analyses of the related species were consistent with supporting the classification of these 10 new species and two new combinations.

Divergence Time Estimation
The results of the BEAST analysis based on three combined nuclear genes (Figure 2A,B) showed the origin of gomphoid fungi in the narrower sense during the Lower Cretaceous about 142.85 Ma, which is consistent with Sánchez-García et al. [4]. The inferred node heights of the genus-level clades of the gomphoid fungi s. str. range from 78 Ma to 87 Ma, which places the origin of genus-level clades in the Upper Cretaceous: the Gomphus in the age of 87.2 Ma and Turbinellus in the age of 78.47 Ma. These numbers are older than some previous estimates [5,52] but later than or close to the divergence time of host plants in the families Pinaceae, Fagaceae and Myrtaceae [53][54][55][56].

Divergence Time Estimation
The results of the BEAST analysis based on three combined nuclear genes ( Figure  2A,B) showed the origin of gomphoid fungi in the narrower sense during the Lower Cretaceous about 142.85 Ma, which is consistent with Sánchez-García et al. [4]. The inferred node heights of the genus-level clades of the gomphoid fungi s. str. range from 78 Ma to 87 Ma, which places the origin of genus-level clades in the Upper Cretaceous: the Gomphus in the age of 87.2 Ma and Turbinellus in the age of 78.47 Ma. These numbers are older than some previous estimates [5,52] but later than or close to the divergence time of host plants in the families Pinaceae, Fagaceae and Myrtaceae [53][54][55][56].

Correlations among the Fruit Body Size, Spore Size and Spore Shape
PGLS analysis of reproductive traits showed that there was a significantly positive correlation between the fruit body size and the spore size of gomphoid fungi in the narrower sense in the Cenozoic (Paleogene, Neogene and Quaternary) and present but a negative correlation in the Mesozoic ( Figure 3B). The correlation between the fruit body size and Qm was not significant in the Cenozoic but positive in the Mesozoic ( Figure 3C). The spore size of gomphoid fungi s. str. was negatively correlated with Qm in the Mesozoic and Cenozoic ( Figure 3D), indicating that larger spores were more spherical in these periods. Gomphoid species with larger fruit bodies produced larger and more spherical spores in the Cenozoic but smaller and narrower spores in the Mesozoic ( Figure 3B,D).

Correlations among the Fruit Body Size, Spore Size and Spore Shape
PGLS analysis of reproductive traits showed that there was a significantly positive correlation between the fruit body size and the spore size of gomphoid fungi in the narrower sense in the Cenozoic (Paleogene, Neogene and Quaternary) and present but a negative correlation in the Mesozoic ( Figure 3B). The correlation between the fruit body size and Qm was not significant in the Cenozoic but positive in the Mesozoic ( Figure 3C). The spore size of gomphoid fungi s. str. was negatively correlated with Qm in the Mesozoic and Cenozoic ( Figure 3D), indicating that larger spores were more spherical in these periods. Gomphoid species with larger fruit bodies produced larger and more spherical spores in the Cenozoic but smaller and narrower spores in the Mesozoic ( Figure 3B,D).     Basionym-Merulius clavatus Pers., Observationes Mycologicae 1:21 (1796). Description-Basidioma up to 17 cm tall, unipileate at base and then merismatoid with 5-15 subpilei. Pileus up to 15 cm wide; irregularly fan-shaped; surface orangish brown to vinaceous brown to creamy violet; glabrous or covered with minute patches toward the crenate or undulate margin. Hymenium surface gray-violet to violet or vinaceous brown; wrinkled; irregularly reticulate to almost poroid and extending to the upper stipe. Stipe is off-white to pale violet; glabrous at apical part but tomentose to hispid toward the white base; context off-white.

Taxonomy of Gomphoid Fungi in the Narrower Sense
Basidiospores ( Notes-This species was reported from southwestern China under either the name of G. orientalis or G. clavatus [57,58]. Our molecular and morphological study of 13 collections of this group confirmed the occurrence of G. clavatus in southwestern China, which was often confused with G. orientalis (see below) due to their similar characteristics of merismatoid basidiomata, irregularly fan-shaped pileus and purplish hymenium. However, G. clavatus generally possesses an off-white context, smaller basidiospores (10-15 × 5-7.5 µm) and the common presence of pileocystidia in pileipellis, while G. orientalis has a grayish violet context, larger basidiospores (15-18 × 7-9 µm) and the rare presence of pileocystidia in pileipellis. Our phylogenetic analysis resolved G. orientalis as a lineage independent of G. clavatus (Figure 1).
Gomphus matijun J.W. Liu & F.Q. Yu, Mycoscience 63: 293-297 (2022). (Figure 4). Description-Basidioma up to 14 cm tall; unipileate to merismatoid; clavate or urceiform when young and hippocrepiform with age. Pileus up to 9 cm in diam.; gradually inflated upwards and depressed at center when mature; grayish purple; pileal margin undulate; rarely lobed. Hymenium surface grayish blue to bluish purple when young but fading to vinaceous gray or pale lilac when mature; deeply wrinkled; irregularly reticulate and extending to the stipe base. Stipe about 2-7 cm long; 1-3.5 cm in diam.; subclavate; upper part vinaceous brown or grayish purple, becoming whitish toward the base; context white to gray or slightly grayish blue.   Notes-Gomphus orientalis is distinguished by the combination character lowish brown and lightly funnel-shaped or irregularly fan-shaped pubescent the violet hymenium, stipe and context. Gomphus orientalis is similar and closel G. matijun. However, the former possesses a violet context, 4-spored basidia basidiospores (15-18 × 7-9 μm), while the latter has a white to gray or sligh blue context, generally 2-spored basidia and smaller basidiospores (9-11 × 6 addition, G. orientalis and G. matijun occur in different habitats: Gomphus orien tributed in forests of Abies, Picea, Pinus and Betula at 2450-3400 m altitude, whil Notes-Gomphus orientalis is distinguished by the combination characters of the yellowish brown and lightly funnel-shaped or irregularly fan-shaped pubescent pileus, and the violet hymenium, stipe and context. Gomphus orientalis is similar and closely related to G. matijun. However, the former possesses a violet context, 4-spored basidia and larger basidiospores (15-18 × 7-9 µm), while the latter has a white to gray or slightly grayish blue context, generally 2-spored basidia and smaller basidiospores (9-11 × 6-7 µm). In addition, G. orientalis and G. matijun occur in different habitats: Gomphus orientalis is distributed in forests of Abies, Picea, Pinus and Betula at 2450-3400 m altitude, while G. matijun is found in fagaceous forests at 1100-1200 m altitude.
Zang et al. [59] described basidiospores of G. orientalis as 10.3-15.5 (-16.8) × 4.3-7.5 µm, significantly smaller than those of the isotype measured by us (15-17 × 7-9 µm), which were consistent with other two collections and were adopted here. Gomphus orientalis has a relative wide range of habitats. The isotype was collected in forests of Abies and Picea at 3000-3400 m altitude and the other two collections were collected in forests of Pinus and Betula at 2450-2800 m altitude. Nevertheless, these collections were treated here as the same species because molecular evidence indicated that they represented the same species  Diagnosis-Turbinellus flavidus is very similar to T. longistipes in having sparsel pileal surface, yellow hymenium, yellow cylindrical stipe and large basidiospores ever, T. flavidus possesses a smaller basidioma (less than 5 cm tall) with a yellow p short stipe and a yellow context, while T. longistipes has a larger basidioma (up to tall) with an orange pileus, a long stipe and a white context. Description-Basidioma 3.5-5 cm tall; unipileate; tapering downward. Pileus cm in diam.; funnel-shaped; depressed at center; surface light yellow (3A4) to o yellow (4A4-6); sparsely covered with minute appressed scales; scales more or les ally arranged; pileal margin uplifted and undulate. Hymenium surface creamy (3 Diagnosis-Gomphus violaceus is similar and closely related to G. clavatus. However, the former possesses a merismatoid basidioma with 2-5 subpilei, a bluish violet pileus, a violet context and the rare presence of pileocystidia in pileipellis; the latter has a merismatoid basidioma with 5-15 subpilei, a brown or grayish purple pileus, an off-white context and the common presence of pileocystidia in pileipellis. G. violaceus has an allopatric distribution with G. clavatus, in which G. violaceus occurs in forests with Pinus in 2000-2150 m altitude, while G. clavatus is found in forests with Abies and Picea at 2700-3600 m altitude. Description-Basidioma 6-13 cm tall; unipileate at base and then merismatoid with 2-5 subpilei; and two subpilei common. Pileus 2-8 cm in diam.; irregularly fan-or funnelshaped; depressed to one side; surface bluish violet (18A5) to violet (16C4); thick in the center; thin toward the ascending; pileal margin uplifted and undulate. Hymenium surface bluish violet (18A4-B4); deeply wrinkled; irregularly reticulate to almost poroid and extending to the stipe base. Stipe about 3-5 cm long; 1-3 cm in diam.; tapering downward; dilating obconical into the pileus; solid; bluish violet (18A5) to violet (16C4) or dark violet (14D4); context bright violet (18B4); unchanging on exposure; basal mycelium white (1A1).
Basidiospores Notes-This species almost has the smallest basidioma in the genus and is closely related to T. tomentosipes and T. solidus in our phylogenetic analysis (Figure 1). However, T. flavidus differs from the latter two due its smaller yellow basidioma with cream-colored context and large basidiospores. Moreover, T. flavidus has only been collected in forests with Castanopsis, Pinus kwangtungensis and Schima superba in central China, while the latter two species occur in forests with Fagaceae and Pinus yunnanensis in southwestern China (see below Diagnosis-Turbinellus fulvus is very similar to T. solidus and T. tomentosipes in that it has a sparsely scaly pileal surface, sparse hymenium, and solid stipe. However, T. fulvus possesses a smaller basidioma (less than 9 cm tall), a khaki pileus, a short and pubescent stipe and smaller basidiospores (11.5-14 × 6-7 µm), while T. tomentosipes has a larger basidioma up to 15 cm tall, an orange-yellow pileus and a tomentose cylindrical stipe. T. solidus has a glabrous stipe and larger basidiospores (13.5-16 × 6.5-9 µm).
diam.) and some inflated hyphae (7-12 μm in diam.); gloeoplerous hyphae abund low in KOH. Pileus and stipe context composed of simple branched hyphae; a mately parallel; hyaline to yellow in KOH. Hymenial trama composed of subpa interwoven, undifferentiated and narrow hyphae. Clamp connections absent in a of basidioma.
Description-Basidioma 6-11 cm tall; unipileate; tapering downward. Pileus 1-7 cm in diam.; subcylindrical when young and narrow funnel-shaped with age; depressed at center; surface light brown (5B4); radial bulge of the surface densely covered with large scales; scales with radial stripes and rolling up in waves; pileal margin extending over hymenium and undulate. Hymenium surface off-white (1A1); slightly wrinkled; irregularly reticulate and extending to the upper stipe; sparse; thin to 0.2-0.4 cm. Stipe about 1.8-4.1 cm long; 0.8-1.5 cm in diam.; tapering downward; dilating obconical into the pileus; solid; white (1A1); glabrous above the base; context pliable; thin to 0.1-0.  Notes-Turbinellus longistipes is easily distinguished by its creamy to light orang large basidioma, long stipe (5-10 cm) with slightly swollen base and large basidiosp This species may also be confused with T. verrucosus in the wild. However, T. long differs from T. verrucosus by its minute appressed scales sparsely covering the pilea face and long stipe with slightly swollen base (see below). Diagnosis-Turbinellus parvisporus is similar to T. fujisanensis in having large d scales on pileal surface and small basidiospores, but the former has a smaller basid less than 9 cm tall, with the reticulation of hymenium becoming stronger upward the common presence of inflated hyphae (7-24 μm in diam.) in pileipellis and stipitip The latter has a larger basidioma up to 15 cm tall, an uniformly forked hymenium an rare presence of inflated hyphae in all parts of the basidioma.
of the surface densely covered with large deltoid scales especially in the center; s rolling up in waves; pileal margin uplifted and undulate. Hymenium surface white to creamy (3A2); strongly wrinkled; irregularly reticulate and extending to the u stipe. Stipe about 1.5-3.5 cm long; 0.    Diagnosis-Turbinellus parvisporus is similar to T. fujisanensis in having large deltoid scales on pileal surface and small basidiospores, but the former has a smaller basidioma less than 9 cm tall, with the reticulation of hymenium becoming stronger upwards and the common presence of inflated hyphae (7-24 µm in diam.) in pileipellis and stipitipellis. The latter has a larger basidioma up to 15 cm tall, an uniformly forked hymenium and the rare presence of inflated hyphae in all parts of the basidioma.
Basidiospores [ Notes-Our phylogenetic analysis showed that T. squamosus shares, with high support, a most recent common ancestor with T. fujisanensis, T. imbricatus, T. kauffmanii and T. parvisporus (Figure 1). The large deltoid scales on pileal surface are a synapomorphy for these five species and appear to have evolved independently in gomphoid fungi. Description-Basidioma 14-25 (-30) cm tall; unipileate; tapering downward. Pileus up to 17 cm in diam.; funnel-shaped; surface orange-ochre to brownish orange; pellicle-like. Hymenium surface clay colored; strongly wrinkled; irregularly reticulate and extending longitudinally to the upper stipe; dense. Stipe approximately cylindrical; rounded at base; hollow; upper part off-white and becoming reddish toward the base; glabrous above the base; context white.    Figures 1, 4 and 5); thus, a new combination was proposed. In order to fix the concept of the species, an epitype for the species was selected in this paper.
As mentioned by Petersen [6], T. szechwanensis is characterized by its larger basidioma exceeding 30 cm in height, broader pileus up to 17 cm in diam., pellicle-like surface of the pileus and coarsely decorated larger basidiospores. Turbinellus floccosus differs from this species by its significantly smaller basidiospores (10-16 × 5-8 µm in Arora [61]; 11.5-14.5 × 7-8 µm in Bessette et al. [62]). This species is closely related to T. verrucosus in the phylogenetic analysis (Figure 1), but the latter has a smaller basidioma less than 7 cm tall, a narrower pileus (2-4.5  Diagnosis-Turbinellus tomentosipes is similar to T. solidus, but the former possesses a larger basidioma up to 15 cm tall, a densely tomentose stipe and profusely inflated hyphae in stipitipellis. The latter has a smaller basidioma less than 10 cm tall, a glabrous stipe and little inflated hyphae in stipitipellis.
Notes-Turbinellus tomentosipes is similar to T. solidus, but the latter po smaller basidioma, a glabrous stipe and filamentous hyphae in stipitipellis. G ically, T. tomentosipes occurs in northern parts of Yunnan, while T. solidus fru southern parts. Diagnosis-Turbinellus verrucosus is similar to T. solidus in having small b orange pileus and small basidiospores. However, the former has a pileus with verrucous protuberances, a hollow stipe and little inflated hyphae in stipitipelli ter has a pileus with sparsely appressed scales, a solid stipe and plenty of inflate in stipitipellis.
Basidiospores [   Diagnosis-Turbinellus verrucosus is similar to T. solidus in having small basidioma, orange pileus and small basidiospores. However, the former has a pileus with densely verrucous protuberances, a hollow stipe and little inflated hyphae in stipitipellis. The latter has a pileus with sparsely appressed scales, a solid stipe and plenty of inflated hyphae in stipitipellis.

Phylogeny and Diversity of Gomphoid Fungi
Gomphus harbors conspicuous members of Gomphaceae and are often confused with other genera in the family. Recent analyses [3,[63][64][65] indicated that Gomphus, in its broader sense, constitutes a paraphyletic group of fungi with Ramaria sensu lato. To maintain the monophyly of Gomphus, some species were transferred to Gloeocantharellus, and Turbinellus was resurrected [1,3]. Gomphus and Turbinellus, as defined here, are monophyletic. This classification is consistent with some previous studies [1 ,3]. Our phylogenetic analysis supports a sister relationship between Gomphus and Turbinellus, contrasting with those studies in which Turbinellus, together with two or more genera of Gomphaceae, were found to be sisters to either Gomphus alone or Gomphus together with several species of Ramaria [2,[63][64][65]. The previous definition of gomphoid fungi [1], including Gloeocantharellus, turned out to be unnatural. Our analyses supported the monophyly of the currently defined gomphoid fungi in the narrower sense, only including Gomphus and Turbinellus.
Most gomphoid species are known from the northern hemisphere, especially North America, where Corner [57] and Petersen [6] considered to be the diversity center of this group. However, according to the latest studies [1, 66,67], only four species of this group, namely G. clavatus, G. ludovicianus, T. floccosus and T. kauffmanii, have been recognized in North America. Our study first included a large number of Asian samples and clarified the phylogenetic relationships of most known species of gomphoid fungi. Based on our study and previous ones [1, 66,67], 23 gomphoid species are currently recognized from Africa, Asia, Europe and North America. Our study indicated that gomphoid species in Asia are abundant and different from those in the other continents. Except for the widely distributed species G. clavatus, G. crassipes and T. floccosus, which have been reported in two or three continents, the remaining species are endemic either to Africa, including G. brunneus, or to North America, including G. ludovicianus and T. kauffmanii or to East Asia (or China), including G. matijun, G. orientalis, T. fujisanensis, T. szechwanensis, T. yunnanensis and the new species described here.
Turbinellus floccosus, easily identified due to the funnel-shape and orange color system of its basidioma, was supposed to be widely distributed in the northern hemisphere [1, 57,68].
However, our study indicated that its occurrence in China or in East Asia needs to be verified. Gomphoid fungi in southwestern China, including the Yunnan-Guizhou Plateau, the Hengduan Mountains region and the Qinghai-Tibet Plateau, show a very high diversity with at least 14 species, accounting for 61% of the total, thus indicating another hotspot of gomphoid fungi in the northern hemisphere. However, geographic configuration of the whole group, as well as the comprehensive phylogenetic relationship, remain unresolved. Increasing taxon sampling in different regions and enlarging the molecular information are essential for further studies.

Trade-Offs among Traits Related to Reproduction in Evolution
Our results showed a close relationship between spore size and spore shape in gomphoid fungi in the narrower sense. When the spores expand in size, they become more spherical, and these changes occurred almost simultaneously ( Figure 3D,K,L), suggesting that the spore size and spore shape are evolutionarily synchronized. That could be due to internal physiological or anatomical constraints, making it difficult to produce both large and narrow spores [13]. That argument is also supported by the fact that spore expansion is the result of simultaneous expansion of both length and width based on our analysis of character evolution, rather than simply length or width extension (Figure S1A-C). As for the tendency for spore evolution to be larger and more spherical beginning in the Cenozoic era, it could be a survivability trade-off driven by global cooling and global aridity [69]. Large spores which are spherical with smaller surface to the volume ratio, compared to small and narrow spores, will contain more water and more nutrients to fight the cold and avoid large amounts of water evaporation [13,26], which can be essential for successful germination [70]. The nutritional mode also accounted for a significant part of the fluctuation in spore size and shape [12,13], with ectomycorrhizal species producing significantly large and spherical spores, in line with the fact that ectomycorrhizal spores need to survive until conditions for root colonization are suitable.
Species producing large spores might need large fruit bodies to produce a high number of spores [13,71,72]. However, the fruit body size of gomphoid fungi s. str. did not vary with spore size in that correlation, as expected during the Mesozoic, when gomphoid species had relative larger fruit bodies with smaller spores ( Figure 3B,J,K). This trend has surprisingly reversed in the Cenozoic, when gomphoid fungi s. str. have relative smaller fruit bodies with larger spores than their Mesozoic ancestors (Figure 3B,F,G). An explanation for gomphoid species with smaller fruit bodies after the Mesozoic is the need to adapt to aridification, especially during the late Eocene and beyond, when it was generally more arid than before due to global cooling [73][74][75]. Water supply is crucial for fungal growth [76], as mentioned by Halbwachs et al. [20]; many macrofungi experience considerable dwarfing in dry years. However, the decrease in fruit body size was not followed by stasis or acceleration; rather, fruit body size enlarged through the Quaternary ( Figure 3F,J). Fruit body size increase should not be the result of spore size expansion; there were no lineage-consistent changes in fruit bodies and spores of phylogenetic character mapping ( Figure S1E,F). However, this variation is in accordance with water availability, as shown by Halbwachs et al. [20]. As the Himalaya orogeny entered its late-stage, the East Asian monsoon was strengthened and the precipitation in the collision zone of the Indo-Asian plate increased, allowing the fruit body to enlarge. A larger fruit body may be a greater buffer against cold stress, which was also beneficial for survival during periods of sharply global cooling.

Larger Spores Were More Important Than Larger Fruit-Bodies
Based on analyses of characters combined with their evolution, we found that Cenozoic and present gomphoid fungi s. str. had larger spores relative to their fruit body size than those of Mesozoic taxa ( Figure 3B,G), and the spores showed a trend of rapid enlargement in the Cenozoic excluding the late Eocene ( Figure 3G,K). When expressed as a phylogenetic sporulation quotient (PSQ), spore sizes of Cenozoic taxa developed more rapidly than those of Mesozoic taxa, and high rates continued through the Quaternary (Figure 2A). All data available suggested that spores have expanded dramatically after the end-Cretaceous extinction, as a critical innovation enhancing spore survival and germination through time and space, and thereby insuring their persistence and ecological success. Conversely, the fruit body size of Paleogene species contracted rapidly and continued into the Neogene ( Figure 3F,J), although briefly enlarged during the middle Eocene, when there was an episode of Eocene climatic optimum with humid conditions [69,77,78]. They were essentially smaller than their sizes in the Mesozoic era ( Figure 3F,J), which could have meant larger fruit body numbers [12]. However, our study found no direct evidence that these innovations after the end-Cretaceous extinction contributed to the diversification of gomphoid fungi s. str. Rates of speciation and net diversification of this group increased steadily through time ( Figure 2C), which were also consistent with LTT plots ( Figure 2D). There was no rate shift within the gomphoid fungi s. str., suggesting no radiative evolution event occurred. That is, these innovations did not lead to rapid speciation after the end-Cretaceous extinction.
As the Mesozoic transitioned to the Cenozoic, the mode of spore and fruit body evolution shifted. Relative spore size greatly increased, in both average and variance, but the fruit body size continued to decrease, signaling a new regime in which spore size enlargements were more paramount than those in fruit body size. Greater fruit body numbers were also more advantageous than those in large fruit bodies. The growth of spores was not only in size, but also in shape, i.e., being more spherical, thus greatly increasing capacity of water retention. The rate of change in PSQ was stable across the Mesozoic but increased dramatically near the beginning of the Cenozoic or immediately after the end-Cretaceous extinction ( Figure 2A); this preceded the increase in the rate of change in fruit body size ( Figure 2B). This suggests that selection acted differently on spores and fruit body size, and, at the very least, relatively large fruit bodies were not necessary for gomphoid fungi s. str. after the end-Cretaceous extinction. Smaller fruit bodies, alternatively meaning a greater number of fruit bodies, were considered to have an advantage when colonizing patchy niches as extinction survivors, with option for a finer-grained response to reproductive investment over larger fruiting species and reducing the risk of demising before sporulation [8,12]. The majority of branches exhibited faster rates of fruit body size increase in the late Miocene. This reached its maximum in the Quaternary ( Figure 2B), during which fruit body size continued to enlarge ( Figure 3F,J), thus indicating that the larger fruit body was more likely driven by the sharply global cooling and the strengthening of the East Asian monsoon.