Phylogenetics of the rust fungi (Pucciniales) of South Africa, with notes on their life histories and possible origins

ABSTRACT

South Africa has an indigenous rust (Pucciniales) funga of approximately 460 species. This funga was sampled with species from as many genera as possible. The nuclear ribosomal large subunit (28S) region was amplified from samples representing 110 indigenous species, as well as the small subunit (18S) region and the cytochrome c oxidase subunit 3 (CO3) in some cases, and these were used in phylogenetic analyses. One new species is described, 12 new combinations made, six names reinstated, and two life history connections made. The life histories of this funga were summarized; it is dominated by species with contracted life histories. The majority of species are autoecious, with a small proportion being heteroecious. Of the autoecious species, many will likely be homothallic with no spermagonia. A shortened life history with homothallism allows for a single basidiospore infection to initiate a local population buildup under the prevailing unpredictable climatic conditions. Suggestions are made as to the possible origin of this funga based on the development of the modern South African flora. It is postulated that the rusts of South Africa are of relatively recent origin, consisting of three groups. Firstly, there is an African tropical element with members of the Mikronegerineae (Hemileia), the Sphaerophragmiaceae (Puccorchidium, Sphaerophragmium), and certain Uredinineae (Stomatisora). Their immediate ancestors likely occurred in the tropical forests of Africa during the Paleogene. Secondly, there is a pantropical element including the Raveneliaceae (e.g., Diorchidium, Maravalia, Ravenelia sensu lato, Uropyxis). This likely diversified during the Neogene, when the mimosoids became the dominant trees of the developing savannas. Thirdly, the Pucciniaceae invaded Africa as this continent pushed northward closing the Tethys Sea. They diversified with the development of the savannas as these become the dominant habitat in most of Africa, and are by far the largest component of the South African rust funga.

KEYWORDS:

• Evolution
• life histories
• rust fungi
• South Africa
• taxonomy
• uredinales
• 13 new taxa

INTRODUCTION

The plant diversity of South Africa is one of the richest in the world. The diversity of landscapes and climatic conditions, and the changing interactions of these through evolutionary time, is a major factor in shaping the plant diversity. Although the land surface area, at 1.22 million km², represents 0.8% of the world total, South Africa is home to 19 581 native plant species (927 bryophytes, 260 ferns, 25 lycophytes, 45 gymnosperms, and 18 324 angiosperms) (Germishuizen et al. 2006) or ca. 5.9% of the world’s total (Cowling and HiltonTaylor 1997). Three of the six major phytochoria of Africa recognized by White (1983) occur in South Africa—the Zambezian, Cape, and Karoo-Namib, as well as the southern reaches of the Afromontane archipelago-like center of endemism—and two major regional transitional zones, the Kalahari-Highveld and Tongaland-Pondoland. More recently, southern Africa was identified as one of seven major biogeographic regions in Sub-Saharan Africa based on both vascular plant and land vertebrate species distributions. Four of the five subregions in the southern African region, as well as a small area of the Zambezian region, occur in South Africa (Linder et al. 2012). Within the country, nine biomes are currently recognized (Mucina and Rutherford 2006). It has three major regions of plant endemism, the Cape Floristic Region, the Succulent Karoo, and the Maputoland-Pondoland region, and seven regional centers of plant endemism (van Wyk and Smith 2001). This plant diversity is equivalent to that of tropical regions, despite much of the country having a warm temperate, semiarid climate with mean annual rainfall below 400 mm (Cowling and Hilton Taylor 1997). The warm temperate flora in South Africa is the richest in the world (Germishuizen et al. 2006).

Rust fungi (Basidiomycota, Pucciniales) are obligate plant pathogens that infect a wide range of vascular plants. Because rust fungi are specific to their hosts, it is not unreasonable to hypothesize that a rich plant flora may also indicate a rich Pucciniales funga. To date, approximately 460 indigenous species of rust fungi are known to occur in South Africa. This number is unexpectedly low considering the rich plant diversity in the country (Berndt 2008) yet represents 5.7% of the approximately 8000 described rust fungi (Aime et al. 2017). This funga is typical of African rust fungi in general but lacks some genera that occur elsewhere in tropical Africa. The species endemism rate of this rust funga is approximately 44% (Berndt 2008). There are no genera of rust fungi endemic to South Africa, and only four small genera endemic to the whole of Sub-Saharan Africa, namely, Cumminsina (1 species), Joerstadia (4 species) (Gjærum and Cummins 1982), Stomatisora (2 species; Wood et al. 2014), and Sphenorchidium (2 species; Beenken and Wood 2015). Ypsilospora (3 species) was only known from West Africa (Ono and Hennen 1979) until the description of Y. tucumanensis from South America (Hernandez and Hennen 2003).

Very few South African species have been included in phylogenetic studies of rust fungi. A few were included in the early attempts at a phylogenetic reconstruction for Pucciniales (Wingfield et al. 2004) and Pucciniaceae (Maier et al. 2007). In addition, a few studies on specific taxa have been undertaken: a species complex of Endophyllum (Wood and Crous 2005), Macuropyxis fulva (Martin et al. 2017), Phakopsora myrtacearum (Maier et al. 2016), Puccinia on Lycieae (Ireland et al. 2019; Otálora and Berndt 2018), Solanaceae (Boshoff et al. 2022), and indigenous rye grass (Pretorius et al. 2015), Puccorchidium (Beenken and Wood 2015), Stomatisora (Wood et al. 2014), and Ravenelia (Ebinghaus et al. 2018, 2020).

In the present study, we undertook the re-collection of ca. 110 indigenous species of Pucciniales from South Africa, representative of the rust funga. Vouchers were sequenced at three loci, the nuclear ribosomal large and small subunits of the rDNA repeat and the cytochrome c oxidase subunit 3 of the mitochondria DNA, and these were used for phylogenetic reconstruction. We describe one new species, make 12 new combinations, reinstate six names, and connect two aecial stages with their sporothallus stages based on our results and discuss the origins and evolution of the South African rust funga.

MATERIALS AND METHODS

Specimens were collected on an ad hoc basis over a period of 20 years from as many regions of South Africa as possible, although many areas were not visited and may yet yield interesting examples. Herbarium specimens were prepared from the collections and maintained in paper envelopes until they could be identified. Spores were scraped from pustules and placed in a drop of 50% lactic acid aqueous solution on glass microscope slides then gently heated over an alcohol Bunsen burner, prior to microscopic examination. Free-hand cross sections of pustules were made after soaking small pieces of herbarium material in hot water for a minimum of 30 minutes. Spores and pustules were examined with either a Zeiss axioscope (Oberkochen, Germany) or a Nikon E600 microscope (Tokyo, Japan). Species were identified by comparing the observed spores and fungal structures with published descriptions.

Specimen duplicates are vouchered at the South African National Collection of Fungi (PREM), the U.S. National Fungus Collections (BPI), and/or the Arthur Fungarium at Purdue University (PUR) (SUPPLEMENTARY TABLE 1).

DNA was extracted from fresh or herbarium material with the DNeasy Plant Mini Kit (Qiagen, Germantown, Maryland), the UltraClean Plant DNA Isolation Kit (MoBio Laboratories, Solana Beach, California), or the E.Z.N.A. Plant DNA DS Kit (Omega Bio-tek, Norcross, Georgia). The first 900 bp of the nuclear large subunit (28S) region of the ribosomal DNA repeat was amplified with Rust2INV (Aime 2006)/LR6 (Vilgalys and Hester 1990) and, for weak products, nested with Rust28SF (Aime et al. 2018)/LR5 (Vilgalys and Hester 1990) following the protocols of Aime et al. (2018). The small subunit (18S) region of the ribosomal DNA repeat was amplified with NS1 (White et al. 1990)/Rust18S-R (Aime 2006) and, for weak products, nested with RustNS2-F (Aime et al. 2018)/NS6 (White et al. 1990) following the protocols of Aime et al. (2018). Cytochrome c oxidase subunit 3 (CO3) of the mitochondrial DNA was amplified with CO3_F1/CO3_R1 (Vialle et al. 2009) following the protocols of Vialle et al. (2009). For select members of Coleosporium, the internal transcribed spacer region of the ribosomal DNA (ITS) repeat was amplified with primers ITS1F/ITS2R (Toome and Aime 2015) following the protocols of Toome and Aime (2015). Polymerase chain reaction (PCR) products were sequenced with the amplification primers at Beckman Coulter Sequencing (Danvers, Massachusetts). Raw sequences were edited in Sequencher 4.5–5.4 (Gene Codes, Ann Arbor, Michigan) and verified by BLASTn against the National Center for Biotechnology Information (NCBI) database (Altschul et al. 1990). Edited sequences were deposited in GenBank (SUPPLEMENTARY TABLE 1).

For phylogenetic analyses, data sets were supplemented with publically available sequences representing South African Pucciniales or supplemented with key taxa necessary for resolving taxonomic conflicts (SUPPLEMENTARY TABLE 1). Individual data sets were constructed and analyzed (as described below) for each locus, and specimens or taxa with conflicting topologies in individual trees were re-sequenced for confirmation. For final analyses, the data sets were divided into three separate data sets for (i) early diverging suborders, (ii) Raveneliineae, and (iii) Urediniineae, following Aime and McTaggart (2021). Individual data sets for 28S, 18S, and CO3 were then realigned in Geneious Prime (Biomatters, Auckland, New Zealand) using the MUSCLE algorithm (Edgar 2004) and concatenated; the ITS data set was analyzed separately. Maximum likelihood (ML) bootstrap analyses were conducted on the final data sets using RAxML 8 (Stamatakis 2006) in the CIPRES Science Gateway (Miller et al. 2010), following the methods of Koch et al. (2018). One thousand bootstrap (BS) replicates were produced; 70% BS represents well-supported lineages.

The spore types present for all indigenous South African and Zimbabwean species were summarized (unpublished data). Rust fungas from other countries, with details of spore types present, were likewise summarized for comparison. Publications from countries and continents other than Africa, with a single reference available with life history details, were summarized. For Sub-Saharan African countries, an unpublished database was populated by reference to a wide range of literature, including numerous articles by G. B. Cummins (many African countries), D. O. Eboh (Nigeria), H. B. Gjærum (Uganda), and G. Viennot-Bourgin (Ivory Coast).

RESULTS

DNA sequence data were generated for 166 new collections of South African Pucciniales, representing at least one species from the majority of genera with indigenous species known to be present in South Africa (SUPPLEMENTARY TABLE 1). Pucciniales from South Africa belong to 10 families fide Aime and McTaggart (2021): Coleosporiaceae, Crossopsoraceae, Milesinaceae, Phakopsoraceae, Phragmidiaceae, Pucciniaceae, Raveneliaceae, Skierkaceae, Sphaerophragmiaceae, and Zaghouaniaceae.

Spore stages present in life histories were obtained for 12 countries (TABLE 1) and compared with those of the South African species.

Table 1. Life histories of rust fungi from South Africa and elsewhere.

Country^f	(±O, I) (II, III)^a^,b	(±O, I) (III)	(±O, I, II, III)	(±O, I, III)	(±O, II, III)	(±O, III)	(±O, I^III)	(±O, I)	(±O, II)	H:A^c	Total species
South Africa (indigenous)	24 100%^d	0	49 65%	25 32%	179 5%	58 9%	13 25%	66 45%	48 0%	24:324 7%^e	461 24%
Southern Africa (alien)	20 100%	0	16 100%	3 33%	29 16%	8 0%	0	0	1 0%	20:56 36%	77 58%
Zimbabwe	15 100%	0	13 85%	6 0%	48 4%	14 36%	1 0%	24 50%	11 0%	15:82 18%	132 37%
Uganda	17 100%	0	25 79%	12 17%	155 1%	37 3%	9 22%	44 44%	38 3%	17:238 7%	337 19%
Nigeria	14 100%	0	3 66%	6 33%	59 7%	7 0%	5 40%	7 28%	11 0%	14:80 19%	112 24%
Ghana, Guinea, Ivory Coast, Sierra Leone	12 100%	0	7 83%	10 22%	104 8%	26 8%	8 38%	44 49%	46 4%	12:155 7%	257 21%
Australia	4 100%	0	16 31%	12 8%	83 16%	43 47%	0	16 44%	14 0%	4:154 3%	188 27%
New Zealand	6 100%	0	24 75%	13 46%	40 10%	17 29%	0	20 50%	20 5%	6:94 6%	140 35%
Brazil	17 100%	0	57 90%	17 47%	332 16%	95 30%	3 66%	68 62%	62 3%	17:512 3%	659 34%
Norway	71 99%	6 100%	72 96%	12 75%	40 30%	58 5%	1 100%	3 100%	3 0%	77:186 41%	266 64%
Netherlands	57 100%	6 100%	41 97%	11 54%	25 52%	23 17%	1 100%	0	0	63:101 62%	164 77%
Great Britain	70 99%	5 100%	58 97%	12 75%	43 37%	42 10%	2 100%	0	2 0%	75:157 48%	234 69%
Portugal	47 100%	6 100%	48 96%	14 86%	49 22%	30 7%	1 100%	3 33%	1 0%	53:142 37%	199 64%
U.S.A.	134 99%	29 100%	97 90%	50 78%	135 18%	129 13%	3 66%	36 94%	9 11%	163:424 38%	622 59%
U.S.A. Sonoran Desert	14 100%	0	27 100%	2 100%	35 49%	13 8%	0	0	0	14:74 19%	91 78%
U.S.A. Hawaii	2 100%	0	1 100%	0	8 25%	6 33%	2 100%	0	6 0%	2:17 14%	25 36%

Note. Total number of species within each category of life histories is given, and the percentage of these in which spermogonia are known to occur is given below. The total numbers of species are also provided for which the percentage with spermogonia is given below.

^aSpore types included in life histories, strictly according to morphology: O = spermogonia (which may be present or absent); I = catenulate, verrucose asexual spores; II = pedicelate, echinulate asexual spores; III = teliospores; I^III = endocyclic.

^bAll spore types within one set of parentheses = autoecious, those in two sets = heteroecious.

^cH = heteroecious, A = autoecious (species known only from anamorphs excluded).

^dPercent of total in which spermogonia are known to be present in the life histories.

^ePercent of heteroecious species compared with autoecious species (in bold); anamorphic species not counted.

^fReferences: African countries (unpublished data), Australia (Doungsa-ard et al. 2018; McAlpine 1906; Shivas et al. 2014), New Zealand (Crane and Peterson 2007; Cunningham 1931; McKenzie 1998, 2008; Padamsee and McKenzie 2017), Brazil (Hennen et al. 2005), Norway (Gjærum 1974), Netherlands (Termorshuizen and Swertz 2011), Great Britain (Wilson and Henderson 1966), Portugal (Talhinhas et al. 2019), U.S.A. (Arthur 1962), Sonoran Desert (Cummins 1979), Hawaii (Berndt 2011; Gardner 1994).

TAXONOMY

The recently published taxonomic scheme of Aime and McTaggart (2021) is followed below. The Rogerpetersoniaceae, Araucariomycetaceae, Pucciniastraceae, Ochrosporaceae, Tranzscheliaceae, Pileolariaceae, and Gymnosporangiaceae are not represented by any indigenous species in South Africa, nor indeed Sub-Saharan Africa as a whole.

Mikronegeriineae

Zaghouaniaceae.—

South Africa currently has a total of seven indigenous species of Hemileia. In our phylogenetic analyses (FIG. 1), Uredo ectadiopsidis resolved amongst several of these Hemileia species. This is one of two suprastomatal Uredo species on Periplocoideae (Apocynaceae) hosts in South Africa, with a similar morphology. Although the urediniospores of these two Uredo species are not reniform but rather globose to ellipsoid, they are otherwise similar to the typical morphology of Hemileia, as well as that of Uredo cryptostegiae, which also resolves in this clade and whose host is also in the Periplocoideae. Two other African species, both also known only from globose urediniospores, are already considered to belong in Hemileia, namely, H. scitula on a Periplocoideae host and H. secamones on an Apocynaceae host (Ritschel 2005). Uredo ectadiopsidis is therefore transferred, confirming that the genus concept of Hemileia is not restricted to species with reniform urediniospores. The other Uredo species, U. cryptolepidis, resolved with U. cryptostegiae in a sister clade to Hemileia, its generic placement awaits further study. Suprastomatal uredinia were observed for U. ectadiopsidis, contrasting with U. cryptostegiae, which is recorded as erumpent and rupturing the leaf epidermis (Ritschel 2005).

Figure 1. Early diverging Pucciniales in South Africa. ML topography generated from 28S, 18S, and CO3 sequencing data. The tree is rooted with Rogerpetersonia torreyea. Names in bold indicate sequences from South African material. Support for nodes is provided as ML ratios/fast bootstraps.

Hemileia ectadiopsidis (Cooke) A.R. Wood & Aime, comb. nov.

MycoBank MB843385

Basionym: Uredo ectadiopsidis Cooke, Grevillea 10(56):128. 1882.

Skierkineae

The single locally occurring representative of Skierkia, S. robusta, resolved in this clade together with the type species (FIG. 1), as previously reported (Aime and McTaggart 2021).

Melampsorineae

This suborder is predominantly a temperate Northern Hemisphere group and is poorly represented in South Africa (Berndt 2008). Of the 10 indigenous species, all but one produce only uredinia in the region. The exception is Melampsora hypericorum var. australis, known only from aecia and telia and is the only known indigenous member of the Melampsoraceae. Several indigenous species previously described in Melampsora and Schroeteriaster have all been shown to be phakopsoroid (Wood 2006). Various of the 10 indigenous species have a disjunct distribution between the temperate Northern Hemisphere and South Africa, including Milesina blechni and Thekopsora agrimoniae. Each of these has an expanded host range in South Africa, infecting plant genera that are not hosts in their native range, namely, Rumohra (Dryopteridaceae) (Berndt 2008b) and Cliffortia and Leucosidea (Berndt and Wood 2012), respectively. These may demonstrate a means of speciation amongst rust fungi when penetrating into new geographic ranges, i.e., jumping to new hosts that are similar to their original host and which do not have specific resistance mechanisms in place (new associations). The endemic Coleosporium hedyotidis, Milesina nervisequa, and Milesina silvae-knysnae may have originated in this manner.

Milesinaceae.—

South Africa has five rust fungi on fern hosts, of which two are Northern Hemisphere species with disjunct populations in South Africa (Milesina blechni, Uredinopsis pteridis), one is tentatively identified as a South American species (M. magellanica), and two are known only from South Africa (M. nervisequa, M. silvae-knysnae) (Berndt 2008b). Of these, only U. pteridis was included in the analysis, and it was found to be distinct from the only other sequence of this species available, from Australia (McTaggart et al. 2014). This raises the possibility that at least one of these is undescribed, but sequences from the native range of this rust will need to be obtained to determine the status of these collections. The only other records of a rust on a fern host from Sub-Saharan Africa is that of U. pteridis from the Democratic Republic of the Congo (Hennings 1907; Sydow and Sydow 1909), Kenya (Nattrass 1961), and Sierra Leone (Deighton 1936).

Coleosporiaceae.—

Only four species are native to South Africa, Thekopsora agrimoniae, Coleosporium clematidis, C. hedyotidis, and a taxon currently known as C. ipomoeae, of which the first and last two were included in the analysis (FIG. 1). This last taxon was distinct to authentic C. ipomoeae (FIGS. 2 and 3) and is therefore described as a new species below. Coleosporium hedyotidis is only known from South Africa on indigenous members of the Rubiaceae; all other species in this genus on members of the Rubiaceae occur in Asia (Farr and Rossman 2021). Coleosporium clematidis is also known from Ethiopia (Gjærum 1995), Kenya, Tanzania, and Uganda (Gjærum 1985) as well as through much of Asia (Farr and Rossman 2021). Thekopsora agrimoniae is not known from anywhere else in Sub-Saharan Africa.

Figure 2. Coleosporium in South Africa. ML topography generated from ITS sequencing data. The tree is rooted with C. tussilaginis. Names in bold indicate sequences from South African material. Support for nodes is provided as ML ratios/fast bootstraps.

Figure 3. Coleosporium hewittiae. (a) Cross section through a uredinium. (b, c) Hymenium from which urediniopores are produced. d, e. Urediniospores in median and surface views. f. One urediniospore in median and surface views to show ornamentation. Bars: a = 100 µm; b–f = 10 µm.

Coleosporium hewittiae A.R. Wood & Aime, sp. nov. FIG. 3a–f

MycoBank MB843384

sub Coleosporium ipomoeae (Schwein.) Burrill, fide Doidge (1927) Bothalia 2(1a):167. 1927.

Typification: SOUTH AFRICA. KWAZULU-NATAL: Silverglen Nature Reserve, Chatsworth, Durban, on Hewittia malabarica (L.) Suresh (Convolvulaceae), 27 May 2008, A.R. Wood 715 (holotype PREM 60119, II; isotype PUR N16340). GenBank: 28S = MF769640; ITS = OQ183718 (clone 1), OQ183719 (clone 2), OQ183720 (clone 3); 18S = OQ215129; CO3 = OR789145.

Diagnosis: Similar to C. ipomoeae but differing in host and larger urediniospore size and wall thickness, and lack of other spore stages.

Etymology: Named for the host genus Hewittia.

Spermogonia, aecia, and telia not observed. Uredinia hypophyllous, scattered or grouped, up to 1 mm diam, orange becoming pale yellow on old specimens, subepidermal, erumpent, aparaphysate, no peridial elements, spores not pedicellate, apparently catenulate. Urediniospores globose to ellipsoid, 20–30(–35) × 17–23 µm (mean 24.8 ± SE 0.35 × 20.6 ± SE 0.15 µm, n = 60); finely to coarsely verrucose, the wall 2 µm thick including verruculae, epispore approx. 05–0.75 µm thick, the verruculae approx. 1–1.5 µm high; no germ pores visible.

Other specimens examined: SOUTH AFRICA. KWAZULU-NATAL: Winkelspruit, Durban, on Hewittia malabarica (L.) Suresh (Convolvulaceae), 2 Jul 1911, I.B. Pole-Evans (PREM 1598); Winkelspruit, Durban, on Hewittia malabarica (L.) Suresh (Convolvulaceae), 27 May 1915, E.M. Doidge (PREM 9100).

This taxon has long been known from South Africa but was originally assumed to belong to the American species C. ipomoeae, based on the morphology of the pustules and spores. It differs from C. ipomoeae in having slightly larger urediniospores (18–27 × 13–21 µm, wall 1–1.5 µm in C. ipomoeae). As C. ipomoeae has also been recorded in Uganda (Wakefield and Hansford 1949) and Tanzania (Riley 1960), it is possible that these records also represent C. hewittiae. The spores are verrucose, typical of uredinia in Coleosporium.

Raveneliineae

Phakopsoraceae.—

The Chaconiaceae, Phakopsoraceae, and Uropyxidaceae sensu Cummins and Hiratsuka (2003) are polyphyletic (Aime 2006; Aime and McTaggart 2021). Two genera have recently been described to accommodate species originally placed in Phakopsora and Crossopsora but that are phylogenetically distinct, namely, Neophysopella (Ji et al. 2019) and Crossopsorella (Souza et al. 2018), respectively. These studies suggest that there are additional distinct lineages for species originally placed within the Phakopsoraceae, as well as the Chaconiaceae, and which likely represent new genera. This situation has since been partially resolved, with parts of each of these families and their constituent genera split between the Raveneliineae and the Uredinineae (Aime and McTaggart 2021). However, many more species need to be included in phylogenetic analyses before generic limits can be set and new genera described for distinct clades.

Phakopsoraceae in South Africa include several species of Phakopsora, Uredopeltis, and Masseeëlla capparis (FIG. 4). Phakopsora stratosa is congeneric with Bubakia argentinesis; therefore, the name B. stratosa is reinstated. Phakopsora ziziphi-vulgaris and P. microspora are congeneric with Uredopeltis species and are therefore transferred to this genus. In our analyses, Phakopsora combretorum is not a member of Phakopsora sensu stricto, however, no taxonomic novelties are proposed at this time until the correct placement of P. combretorum can be ascertained with additional analyses.

Figure 4. Raveneliineae in South Africa. ML topography generated from 28S, 18S, and CO3 sequencing data. The tree is rooted with Phragmidiaceae (Uredinineae). Names in bold indicate sequences from South African material. Support for nodes is provided as ML ratios/fast bootstraps.

Bubakia stratosa (Cooke) Dietel, in Engler & Prantl, Nat Pflanzenfam Edn 2 (Leipzig) 6:48. 1928.

Basionym: Melampsora stratosa Cooke, Grevillea 10(no. 56):128. 1882.

Synonyms: Schroeteriaster stratosus (Cooke) P. Syd. & Syd., Monogr Uredin (Lipsiae) 3(2):402. 1914; Phakopsora stratosa (Cooke) Arthur, Bull Torrey Bot Club 44:508. 1917.

Bubakia doidgeae (Syd.) A.R. Wood & Aime, comb. nov. FIG. 5a, b

Figure 5. (a, b) Bubakia doidgea. Cross sections through telia. (c, d) Uredopeltis microspora. (c) Cross section through a telium. (d) Teliospores. (e, f) Uredopeltis corbiculoides. Cross sections through telia. Bars: a–c, e, f = 100 µm; d = 10 µm.

MycoBank MB843386

Basionym: Schroeteriaster doidgeae Syd., Annls mycol 24(3/4):265. 1926.

Synonym: Uredo doidgeae (Syd.) A.R. Wood, S Afr J Bot 72(4):541. 2006.

Telia hypophyllous, forming hyaline or light yellow crusts, subepidermal but developing above the leaf surface, epidermal covering persistent, completely hidden by the plant’s large overlapping disc-shaped stellate hairs covering the abaxial leaf surface, associated with old highly melanized uredinia, eliptic in transverse section, 125–263 × 88–155 µm. Teliospores 1-celled, sessile, irregularly arranged and densely aggregated, irregularly shaped, broadly cuboid to elliptic, ends rounded, 22–38 × 10–17 µm (mean 30.9 ± SE 0.77 × 17.8 ± SE 0.85 µm, n = 25), wall hyaline to lightly yellow pigmented with age, smooth, uniformly 1 µm thick.

Specimen examined: SOUTH AFRICA. MPUMALANGA: Nelspruit, Lowveld National Botanical Garden, on Croton gratissimus Burch. (Euphorbiaceae), 29 Apr 2022, A.R. Wood 1246 (PREM 63408).

This is the first description of the telial stage (FIG. 5a, b), which is cryptic, as the small telial crusts remain covered by the plant’s shield-like leaf hairs and hidden from view.

Uredopeltis ziziphi-vulgaris (Henn.) A.R. Wood & Aime, comb. nov.

MycoBank MB843424

Basionym: Uredo ziziphi-vulgaris Henn., Hedwigia 41(Beibl.):21. 1902.

Synonym: Phakopsora ziziphi-vulgaris (Henn.) Dietel, Annls mycol 8(3):469. 1910.

Uredopeltis corbiculoides (Cummins) A.R. Wood & Aime, comb. nov. FIG. 5e, f

MycoBank MB847186

Basionym: Uredo corbiculoides Cummins, Bull Torrey Bot Club 72:219. 1945.

Telia hypophyllous, not on leaf spots, scattered, subepidermal in origin but developing above the surface of the leaf. In surface view appearing as Ravenelia-like round, single, black bodies. Basal structure of telia similar to uredinia of this species, but not developing from uredinia. Arise from an elongate cellular stipe-like structure, with peripheral paraphyses as described for the uredinia (Cummins 1945). The telia consists of numerous adherent teliospores, round in surface view and oval with stipe in transverse view, emerging far beyond paraphyses which are inconspicuous, 196–508 µm wide, 235–390 µm high with stipe, without stipe 145–290 µm high. Teliospores unicellular, irregularly arranged, rectangular to cuboid, 15–33 × 7–13 µm (mean 23.4 ± SE 0.99 × 9.4 ± SE 0.37 µm, n = 25), distal spores darkly pigmented appearing black, inner spores and stipe cells hyaline, spore walls smooth, 1 µm thick, distal periclinal wall of distal cells 5–8 µm.

Specimen examined: SOUTH AFRICA. LIMPOPO: Kruger National Park, Punda Maria Camp, on Grewia damine Gaertn. (= Grewia bicolor Juss., Malvaceae), 25 Apr 2021, A.R. Wood 1158 (PREM 63391).

This is the first description of the telia (FIG. 5e, f) for this species, which are very similar to those of Uredopeltis atrides (Wood 2007), differing in being distinctly stipitate and larger. Based on morphology and hosts, there is no reason to doubt that this species belongs to Uredopeltis.

Uredopeltis microspora (Cummins) A.R. Wood & Aime, comb. nov. FIG. 5c, d

MycoBank MB847187

Basionym: Phakopsora microspora Cummins, Bull Torrey Bot Club 87(1):37. 1960.

sub Uredopeltis cf. chevalieri fide Mennicken et al., Mycol Prog 4:72. 2005.

Specimens examined: SOUTH AFRICA. LIMPOPO: Abel Erasmus Pass, N of Ohrigstad, on Grewia flavescens Juss. (Malvaceae), 26 Jul 2008, A.R. Wood 728 (PREM 62324; PUR N16539; 28S OQ215090); MPUMALANGA: Berg-en-dal, Kruger National Park, on Grewia flavescens, 26 Jan 2011, A.R. Wood 827 (PREM 62323).

Cummins (1960) distinguished Phakopsora microspora from Uredopeltis chevalierii by having smaller urediniospores and teliospores (TABLE 2), the difference being pronounced only for urediniospores. Cummins (1960) also described P. microspora as subepidermal, whereas U. chevalierii was described as subepidermal becoming erumpent (Cummins 1945).

Table 2. Spore dimensions of Uredopeltis microspora and Uredopeltis chevalieri according to the literature, and for two specimens collected in South Africa.

Species	Urediniospores	Teliospores	Reference
Uredopeltis microspora	16–19 × 14–17 μm, wall 1 μm	8–20 × 6–13 μm, wall 1–1.5 μm	Cummins 1960
On Grewia flavescens	17–26(–33) × 11–18 μm, wall 1(–2) μm	12–23 × 10–15 μm, wall 1–2 μm	PREM62323/4
“Uredopeltis cf. chevalieri”^a	16–29 × 12–17 μm, wall 0.5–1.5 μm	10–29 × 6–16 μm, wall 0.5–2 μm	Mennicken et al. 2005
“Phakopsora grewiae”^b	20–28 × 15–21 μm, wall 1 μm	11–23 × 8–14 μm, wall 1.5–2 μm	Cummins 1945
“Dasturella grewiae”^b	20–26 × 15–20 μm, wall 1 μm	11.5–21 × 8–12 μm	Thirumalachar 1946
Uredopeltis chevalieri	22–33(–37) × 17–22 μm, wall 1 μm	15–23 × 10–15 μm, wall 2(–2.5) μm	Walker and Shivas 2004

^aHere considered as Uredopeltis microspora.

^bA synonym of Uredopeltis chevalieri.

Phakopsora microspora was first recorded as uredinia only from South Africa in McTaggart et al. (2017) (see https://collections.daf.qld.gov.au/web/key/africarust/Media/Html/index.html); however, Mennicken et al. (2005) had previously assigned a specimen collected in Namibia on Grewia flavescens to U. cf. chevalierii, whilst noting that the dimensions of the spores include smaller ones than described for that species. In fact, the urediniospore and teliospore dimensions span that described for both P. microspora and U. chevalierii. Two more specimens collected in South Africa, also on G. flavescens, similarly have dimensions spanning the ranges of both these rust fungi (TABLE 2). Because the urediniospores of these specimens on G. flavescens from southern Africa consistently have a broader range of spore sizes that includes smaller ones than those of U. chevalerii (Walker and Shivas 2004; Wood 2007), and the telia are initially subepidermal becoming erumpent (FIG. 5c, d) but not as prominently erumpent as those of U. chevalerii, these specimens are referred to P. microspora. Mennicken et al. (2005) provides a full description and illustration (as Uredopeltis cf. chevalieri). It was considered better to expand the species concept of P. microspora rather than describe another species, due to the small number of samples. So far, U. chevalerii has not yet been recorded on G. flavescens, but this plant is also a host to U. atrides (Wood 2007).

Phakopsora microspora clearly belongs to the complex of Uredopeltis species on Grewia species in southern Africa (Wood 2007) and is therefore transferred to this genus.

Raveneliaceae.—

Aime (2006) circumscribed this family as having the majority of species on hosts in the Fabaceae, particularly the mimosoid clade of the Caesalpinioideae. The family concept has since been modified (Aime and McTaggart 2021), but generic limits are still problematic. Ebinghaus et al. (2018, 2020) have recently revised the taxonomy of Ravenelia species on Senegalia and Vachelia from South Africa, bringing the total number of species present in the country to 30 (15% of the described species in the genus). Two other species, both parasitic on members of the Papilionoideae, also belong to this family, namely, Diorchidium woodii and Maravalia lonchocarpi (FIG. 4). Diorchidium woodii is the type of this genus. Beenken and Wood (2015) resolved this species within the Pucciniaceae, presumably as a result of not having as extensive a sampling of species for their analysis. Studies have shown that Maravalia as currently circumscribed is not monophyletic (e.g., Aime and McTaggart 2021); in our analyses, M. lonchocarpi is clearly not congeneric with M. mimusops and M. limoniformis, the latter of which was considered a proxy for the type species due to host and morphological similarities (Aime and McTaggart 2021). Therefore, the name Angusia lonchocarpi is reinstated for this taxon.

Angusia lonchocarpi (Doidge) G.F. Laundon, Trans Br Mycol Soc 47(3):327. 1964.

Basionym: Uredo lonchocarpi Doidge, Bothalia 2(1a):195, 213. 1927.

Synonym: Maravalia lonchocarpi (Doidge) Y. Ono, Mycologia 76(5):899. 1984.

In the analyses of Ebinghaus et al. (2020), Ravenelia resolved into two major clades, one of which has now been separated as Cephalotelium, with C. macowanianum as the type species (Aime and McTaggart 2021). The remaining species of Ravenelia resolve in various clades within the Raveneliaceae (Aime and McTaggart 2021; Ebinghaus et al. 2020). Various names have been proposed for sections (Dietel 1906; Long 1903; Sydow 1921) and are available for use. The type species of Cystingophora, R. hieronymi, resolved in the Cephalotelium clade (clade VII of Ebinghaus et al. 2020), making this a synonym of the latter genus, which having been mentioned prior in the same publication takes precedence. Longia is available for R. natalensis; however, this apparently rare species has so far not been included in phylogenetic analyses. The following combinations are made or reinstated.

Cephalotelium acaciae-arabicae (Mundk. & Thirum.) A.R. Wood & Aime, comb. nov.

MycoBank MB843425

Basionym: Ravenelia acaciae-arabicae Mundk. & Thirum., Mycol Pap 16:17. 1946.

Cephalotelium hieronymi (Speg.) A.R. Wood & Aime, comb. nov.

MycoBank MB843426

Basionym: Ravenelia hieronymi Speg., Anal Soc Cient Argent 12(1):66. 1881.

Synonyms: Cystingophora hieronymi (Speg.) Arthur, N Amer Fl (New York) 7(2):131. 1907; Pleoravenelia deformans Maubl., Bull Soc mycol Fr 22:73. 1906; Ravenelia deformans (Maubl.) Dietel, Beih Bot Zbl Abt 2 20:404. 1906; Cystingophora deformans (Maubl.) Syd., Annls mycol 19(3–4):165. 1921.

African material still requires to be subjected to phylogenetic analysis and may prove to be distinct to this otherwise American species.

Cystotelium inornatum (Kalchbr.) Syd., Annls mycol 19(3–4):165. 1921.

Basionym: Aecidium inornatum Kalchbr., Grevillea 11(no. 57):25. 1882.

Synonym: Ravenelia inornata (Kalchbr.) Dietel, Hedwigia 33:52, 61. 1894.

In Ebinghaus et al. (2020) and the analysis presented here, this species resolved as the sister to Cephalotelium.

Neoravenelia holwayi (Dietel) Long, Bot Gaz 35:131. 1903.

Basionym: Ravenelia holwayi Dietel, Hedwigia 33:61. 1894.

Neoravenelia dichrostachydis (Doidge) A.R. Wood & Aime, comb. nov.

MycoBank MB843427

Basionym: Ravenelia dichrostachydis Doidge, Bothalia 2(1a):148. 1927.

This species resolved in a clade with N. holwayi, clade V of Ebinghaus et al. (2020).

The remaining species in Ravenelia resolved in three different clades (FIG. 4) (clades I, II–IV, and VI of Ebinghaus et al. 2020); however, until such time as molecular data become available for the type species of Ravenelia (R. glandulosa) and Hapaloravenelia (H. indica), no further combinations are made. No type species was listed for Pleoravenelia; however, this is likely a synonym of Ravenelia (Sydow 1921). Of interest, one of these clades occurs on Senegalia (clade I of Ebinghaus et al. 2020), one on various genera in the Caesalpinioideae (sensu LPWG 2017) (clades II–IV of Ebinghaus et al. 2020), and the last on various genera in the Papilionoideae (clade VI of Ebinghaus et al. 2020). As the type of Ravenelia, R. glandulosa, occurs on Tephrosia, this last clade may be the one that is Ravenelia sensu stricto, as also indicated by Ebinghaus et al. (2023). It is probable that at least one if not more additional genera will need to be described to accommodate all ex-Ravenelia species.

Ravenelia ornata resolved as three different subclades, one on each of two species of Abrus from South Africa and another from the Philippines, further investigation is required.

Uredinineae

Phragmidiaceae.—

This family has both tropical and temperate Northern Hemisphere genera, all associated with the Rosaceae. Only three species are indigenous to South Africa, namely, Hamaspora longissima, Kuehneola uredinis, and Trachyspora intrusa. The first two were included in the analysis (FIG. 6). All three are widespread species, the first in the old world tropics, the second throughout much of temperate Northern and Southern Hemispheres, and the last with a disjunct distribution in temperate Europe and South Africa (Farr and Rossman 2021). The Sub-Saharan African endemic genus Joerstadia (Gjærum and Cummins 1982) is not known from southern Africa.

Figure 6. Uredinineae in South Africa. ML topography generated from 28S, 18S, and CO3 sequencing data. The tree is rooted with Sphaerophragmiaceae and Crossopsoraceae fide Aime and McTaggart (2021). Names in bold indicate sequences from South African material. Support for nodes is provided as ML ratios/fast bootstraps.

Figure 6. Continued.

Crossopsoraceae.—

Crossopsora gilgiana and Uredo sekhukhunenensis formed a clade with several phakopsoroid species all parasitic on members of the Asteraceae, namely, P. vernoniae, Uredo tarchonanthi, and U. brachylaenae (FIG. 6). Stomatisora psychotriicola resolved in a sister clade to Kweilinga divina, Angiopsora apoda, and A. paspalicola, together with several species of Aecidium. All but the Angiopsora species have hosts in the Rubiaceae (the aecial host in K. divina). There are other morphologically very similar Aecidium species on the Rubiaceae in southern Africa that also likely belong here. This group includes A. pachystigmae, Aecidium sp. (on Afrocanthium), A. transvaaliae, and A. vangueriae, and possibly other species. However, another morphologically similar species, A. krausii, resolved within Puccinia. Several of these species produce multiple generations of aecia during the year, suggesting that they either are asexual species or have a parasexual life history (the exception being A. transvaaliae).

Crossopsora brachylaenae (Doidge) A.R. Wood & Aime, comb. nov.

MycoBank MB847189

Basionym: Uredo brachylaenae Doidge, Bothalia 2(1a):190, 213. 1927.

Crossopsora sekhukhunenensis (Berndt & A.R. Wood) A.R. Wood & Aime, comb. nov.

MycoBank MB843428

Basionym: Uredo sekhukhunenensis Berndt & A.R. Wood [as ‘sekhukhunensis’], Mycol Progr 11(2):492. 2012.

Originally identified as Crossopsora ziziphi (Doidge 1927), the type species of this genus (Sydow and Sydow 1918), Berndt and Wood (2012) separated these two species based on slight differences in size and that C. sekhukhunenensis had apparently 5–6 germ pores compared with 3 in C. ziziphi. A phylogenetic comparison between authentic C. ziziphi and South African material will clarify as to whether the South African taxon is a distinct species or a geographic variant. Elsewhere in Africa, C. ziziphi is known from Uganda (Gjærum 1998) and Zambia (Berndt and Wood 2012) on Ziziphus abyssinica whereas C. sekhukhunenensis is only known from Ziziphus mucronata.

Crossopsora tarchonanthi (Mennicken & Oberw.) A.R. Wood & Aime, comb. nov.

MycoBank MB847188

Basionym: Uredo tarchonanthi Mennicken & Oberw., Mycotaxon 90(1):10. 2004.

Crossopsora vernoniae (Jørst.) A.R. Wood & Aime, comb. nov.

MycoBank MB847218

Basionym: Phakopsora vernoniae Jørst., Ark Bot Ser 2 3(no. 17):567. 1957 [1956].

Sphaerophragmiaceae.—

McTaggart et al. (2016) and Beenken and Wood (2015) resolved a monophyletic clade sister to the Pucciniaceae, which is also resolved in this study (FIG. 6). Beenken (2017) formally resurrected the family name Sphaerophragmiaceae for this clade. A high proportion of species occur on hosts in the Annonaceae (Beenken 2014; Beenken and Berndt 2010; Beenken and Wood 2015; Beenken et al. 2012), the most diverse of the early diverging flowering plant families (Guo et al. 2017). Thirteen of 21 Sphaerophragmium species (Beenken and Berndt 2010; Lohsomboon et al. 1994), Puccorchidium popowiae, and both Sphenorchidium species (Beenken and Wood 2015) occur in Sub-Saharan Africa. Austropuccinia psidii (Beenken 2017) and 11 Dasyspora species (Beenken et al. 2012) are native to South America, and the remaining eight Sphaerophragmium species (Lohsomboon et al. 1994) and Puccorchidium polyalthiae (Beenken and Wood 2015) occur through tropical Asia and the Americas. A total of 16 (43%), of the 37 known species in this family, occur in Sub-Saharan Africa. This is, by far, the highest proportion of species within any of the rust families that occurs in Africa. This does raise the possibility that this family arose in Africa. Aime and McTaggart (2021) suggest an origin in the late Cretaceous to Paleogene, 75–50 million years ago (mya), which is approximately around the end Cretaceous extinction event.

Pucciniaceae.—

Early phylogenetic work concluded that there were two main clades within the Pucciniaceae, one associated with telial hosts in the Cyperaceae, Fabaceae, and Asteraceae, and one with telial hosts on the Poaceae (Maier et al. 2007, 2008; van der Merwe et al. 2007), with other host plant families in both. Within both groups were species of Puccinia and Uromyces, and scattered amongst these were all other genera within the Pucciniaceae and Pucciniosoraceae of Cummins and Hiratsuka (2003). Maier et al. (2007) included a subclade of South African species within the Poaceae clade, which they named the “African clade.” Dixon et al. (2010) and Minnis et al. (2012) demonstrated a third, smaller clade, also with telial hosts in the Poaceae. Recently a fourth clade, a “Macuropyxis” clade, has been differentiated (Martin et al. 2017). All these four clades were resolved in Aime and McTaggart (2021), as well as another clade that included Desmella and Sphenospora. The four main clades were also resolved in our analyses (FIG. 6). No distinct clade equivalent to the “African clade” of Maier et al. (2007) was distinguished. The majority of the local species included in the clade equivalent to clade 2 of Dixon et al. (2010) (= clade II of Maier et al. 2007, = clade B of van der Merwe et al. 2007, 2008) are endemic in the southern and/or western parts of South Africa, and many have hosts in the Asteraceae and Fabaceae, which is unusual for this clade. The flora of these areas is dominated by two major regions of endemism, the Cape Floristic Region and the Succulent Karoo (van Wyk and Smith 2001), and is predominantly within a winter rainfall Mediterranean-like region. In contrast, other species in the Asteraceae and Fabaceae, but mainly from the summer rainfall region, resolved within the clade equivalent to clade 1 of Dixon et al. (2010) (= clade I of Maier et al. 2007, = clade A of van der Merwe et al. 2007; 2008), as would be predicted from the earlier published work.

Within the clade equivalent to clade 2 of Dixon et al. (2010), which includes U. ixiae, there are two species complexes with highly reduced life histories distributed within the Mediterranean-like winter rainfall Cape region, one on each of the Asteraceae and Fabaceae: Endophyllum dimorphothecae, E. elytropappi, E. metalasiae, E. osteospermi (Wood and Crous 2005), Uromyces euryopsidicola, and Aecidium relhaniae on Asteraceae and Uromyces bolusii, Aecidium dielsii, A. resinaecolum, and A. viborgiae, all on Crotalarieae, Papilionoideae, Fabaceae. All produce perennial galls or witches’ brooms on their hosts, and presumably this is the manner whereby they survive through the long hot and dry summers in their native range. Two specimens identified as E. elytropappi do not appear conspecific with our data, despite repeated sequencing, suggesting that there remain additional cryptic species requiring description in this complex. The Aecidium species on Crotalariae do not germinate in an Endophyllum-like manner (i.e., they did not produce basidospores or structures that can be interpreted as such) (Wood, pers. observ.), nor do they have spermogonia. It is postulated that these will all prove either to be asexual species (type X in Ono 2002) or to have a parasexual nuclear cycle.

An available sequence (KY575114, accessed from https://www.ncbi.nlm.nih.gov/genbank/) of the type specimen of Uromyces pedicellatus was distinct from U. eragrostidis (KY575110), with which it is currently considered a synonym (Doidge 1927), and this species is therefore reinstated.

Uromyces pedicellatus Pole-Evans [as ‘pedicellata’], Bull Misc Inf Kew:229. 1918.

Other members of this clade occur on a wide range of hosts, although predominantly on monocots, and exhibit the full range of life histories. An Aecidium from Oxalis (Oxalidaceae) had an identical sequence to that of Uromyces ixiae from a wide range of Iridaceae. These two rusts were on a number of occasions collected within one to several meters of each other. Thümen (1876) described A. oxalidis from a specimen on Oxalis bowiei collected in the Eastern Cape, South Africa. Subsequently, specimens have been collected from various indigenous species of Oxalis (Doidge 1950; Wood, pers. observ.), which is particularly speciose in the Fynbos biome and the species frequently grow in association with various members of the Iridaceae but where Zea mays is not cultivated. Oxalis has approximately 800 species, with two major centers of endemism, these being South-Central America (with over 500 species) and South Africa (with ±270 species) (Oberlander et al. 2002). Arthur (1904) first demonstrated that Puccinia sorghi was heteroecious, with O. stricta as an aecial host. This has been confirmed several times (Arthur 1905, 1906; Hecke 1906; Rice 1933). Arthur (1904) assumed that all aecial stages on Oxalis then known were the same species and applied the earliest available name—A. oxalidis. Later, it was also shown in South Africa that O. corniculata was another aecial host (Dunhin et al. 2004; Pole-Evans 1923b), as well as in various other countries (Rice 1933; Mahindapala 1978 and references therein). This plant is a widespread and common introduced weed in South Africa. However, it was also shown that a number of Oxalis species indigenous to South Africa were not aecial host plants of P. sorghi (Pole-Evans 1923b). Attempts to infect the type host of A. oxalidis, O. bowiei, have failed (Arthur 1906; Hecke 1906). Recently, Guerra et al. (2019) has confirmed O. conhorriza as another aecial host of P. sorghi. Therefore, only American species of Oxalis, as well as the related cosmopolitan O. corniculata, have been confirmed as aecial hosts of P. sorghi, whereas all South African species tested have proven to be not susceptible. Doidge (1950) maintained A. oxalidis as separate from P. sorghi, only listing aecial specimens collected in southern Africa on O. corniculata as the aecial stage of P. sorghi. All other collections on various indigenous species of Oxalis were assigned to A. oxalidis, not P. sorghi. Aecial collections on Oxalis pes-caprae all clustered amongst collections of U. ixiae from various Iridaceae hosts; therefore, A. oxalidis is correctly the aecial stage of this rust and not P. sorghi, confirming Doidge’s opinion. Aecidium peyritschianum, whose type host is O. corniculata, is available as the name for the aecial stage of P. sorghi (Arthur 1904).

Uromyces ixiae (Lév.) G. Winter, Flora Regensburg 67:262. 1884.

Basionym: Uredo ixiae Lév., Annls Sci Nat Bot sér 3 3:70. 1845.

Gametothallus: Aecidium oxalidis Thüm., Flora Regensburg 59:425. 1876.

Synonyms: Uromyces gladioli Henn., Hedwigia 34:326 (1895); Uromyces geissorhizae Henn., Hedwigia 39(Beibl):(153). 1900; Uromyces melasphaerulae Syd. & P. Syd., Annls mycol 2(1):28. 1904; Uromyces ecklonii Bubák, in Sydow & Sydow, Monogr Uredin (Lipsiae) 2(2):253. 1910; Uromyces delagoensis Bubák [as ‘delagoënsis’], in Sydow & Sydow, Monogr Uredin (Lipsiae) 2(2):255. 1910; Uromyces romouleae Van der Byl & Werderm., Feddes Repert Spec Nov Regni Veg 19:64. 1923; Uromyces romuleae Doidge, Bothalia 2:36, 208. 1927; Uromyces babianae Doidge, Bothalia 2:31, 206. 1927.

Morphologically, Aecidium oxalidis and the gametothallus of P. sorghi are difficult to distinguish. Aeciospores of A. oxalidis measured were 15–25 × 12–21 µm (mean 20 ± SE 0.38 × 17 ± SE 0.36 µm, n = 40), whereas the original description gave 24–28 µm (Thümen 1876). Those of P. sorghi were 18–26 × 13–19 µm (Arthur 1962) or 20–30 × 17.5–22.5 µm (Dunhin et al. 2004), on average 21 × 19.5 µm (Guerra et al. 2019). The aeciospore wall of A. oxalidis measured 1 µm or less in thickness, whereas that of P. sorghi was 1.5 µm thick (Arthur 1962).

Specimens examined: SOUTH AFRICA. WESTERN CAPE: Traveler’s Rest, Agterpakhuis, NE of Clanwilliam, on Oxalis pes-caprae L. (Oxalidaceae), 13 Jul 2001, A.R. Wood 329 (PREM 63301); SE of Bonnievale, on Oxalis sp. (Oxalidaceae), 5 Jun 2008, A.R. Wood 833 (PREM 63302); Stettynskloof Dam, Liemietberg nature reserve, SW of Worcester, on Oxalis sp. (Oxalidaceae), 25 Aug 2011, A.R. Wood 855 (PREM 63299); Bracken Nature Reserve, Brackenfell, on Oxalis pes-caprae L. (Oxalidaceae), 14 Jul 2013, A.R. Wood 929 (PREM 63296; PURN23556; 28S OQ215109); 12 Jul 2014, A.R. Wood 949 (PREM 63297; PURN23557; 28S OQ215108).

Puccinia sorghi Schwein., Trans Am Phil Soc New Series 4(2):295. 1832.

Gametothallus: Aecidium peyritschianum Magnus, Ber Naturw-med Ver Innsbruck 21:34. 1893.

Uromyces transversalis was a sister species to U. ixiae. The heteroecious nature of U. ixiae reported here may be why this species has not become invasive in other parts of the world where its sporothallus hosts have been introduced via the horticultural trade, whereas U. transversalis has.

Doidge (1948) differentiated South African collections of Puccinia tetragoniae from this Australian taxon as P. tetragoniae var. austroafricana. Berndt (2009) raised this variety to species level as P. austroafricana. In our analysis, a specimen from South Africa was very close to an Australian specimen, and it is concluded that specimens from these two regions are separated at the variety rather than the species level; therefore, Doidge’s variety name is reinstated. Long-distance aerial transport of rust spores between southern Africa and Australia has been demonstrated (Visser et al. 2019).

Puccinia tetragoniae var. austroafricana Doidge [as ‘austro-africana’], Bothalia 4:904. 1948.

Synonym: Puccinia austroafricana (Doidge) Berndt [as ‘austro-africana’], Mycol Progr 8(2):100. 2009.

In our analyses, Endophyllum macowanii, on Rhamnus prinoides, is sister to P. coronata, confirming these as correlated species, with the first derived by the endocyclic pathway from the aecial stage of the crown rust, whose gametothallus host is Rhamnus species.

Aecidium evansii (on Lantana rugosa, Verbenaceae) is sister to P. versicolor. Thirumalachar and Narasimhan (1950) demonstrated by inoculation that A. plectroniae, on members the Rubiaceae, was an aecial stage of this grass rust. Cummins (1953) suggested that A. evansii (described from Lippia, Verbenaceae) may also be the aecial stage of this species, based on morphological similarities between these two Aecidium species. Patil and Thirumalachar (1964) confirmed Lantana (Verbenaceae) as an alternate host by inoculation but did not name the aecial stage as A. evansii. The molecular data here, together with the previous inoculation experiments, connect these sporothallus and gametothallus stages. This is therefore another example of a heteroecious species with its gametothallus on more than one plant family. Puccinia lippiicola, a microcyclic species on Lippia, also clustered with the above, demonstrating these as correlated species. These species belonged to the Macruropyxis clade and likely should be transferred.

Puccinia versicolor Dietel & Holw., in Holway, Bot Gaz 24(1):28. 1897.

Gametothallus: Aecidium plectroniae Cooke, Grevillea 10(no. 56):124. 1882; Aecidium evansii Henn., Bot Jb 41:272. 1908.

Aecidium crini has been claimed to be an aecial stage of Uromyces clignyi (Patil and Thirumalachar 1969); however, in our analysis, it grouped closest to U. pedicellatus. Likewise, A. habunguense has been considered to be the aecial stage of P. agrophila (Patil and Thirumalachar 1972); however, aecial specimens on Solanum that were morphologically consistent with this aecial stage grouped with P. penicillariae and P. digitariae, as has been recently published by Boshoff et al. (2022). This suggests that both these gametothalli from southern Africa have been connected to the incorrect sporothallus names following inoculation studies in India. There is no reason to doubt the validity of the cultural work done, but greater circumspection should be applied when associating one taxon native to one continent with another on a different continent.

DISCUSSION

Life histories.—

Only three species, Puccinia digitariae (Boshoff et al. 2022), Puccinia tristachyae (Pole-Evans 1923), and Uromyces aloes (Putterill 1918), of all the South African species have had their life history proven locally by inoculation experiments, and another nine species have been demonstrated to be endocyclic. For the majority of species, however, complete life histories are not demonstrated. Heteroecism is rare amongst the rust fungi recorded in southern Africa. The following indigenous heteroecious species are known to have their gametothalli recorded in the subcontinent: P. aristidae var. chaetariae, P. digitariae, P. duthiae, P. magnusiana, P. phragmitis, P. tristachyae, and P. versicolor. All have their sporothalli on host plants belonging to the Poaceae. In addition, we have shown that U. ixiae, on Iridaceae hosts, is also heteroecious. It is also possible that A. antherici, A. crini, and A. hartwegiae are the aecial stage of species within a P. pedicellatus complex. The following indigenous species, known to be heteroecious elsewhere, have not had their gametothalli recorded from the subcontinent: Puccinia cacao, P. chloridis, P. cynodontis, P. dietelii, P. penicillariae, P. rufipes, P. vilfae, Uromyces setariae-italicae (all on Poaceae), P. caricina, P. cyperi, U. lineolatus (Cyperaceae), U. commelinae (Commelinaceae), Milesina blechni, and Uredinopsis pteridis (ferns). Altogether a total of 22 Pucciniales species, representing only 4% of the rust funga of South Africa, are confirmed as or can be presumed to be heteroecious. Of these species, only P. tristachyae and U. ixiae are endemic to southern Africa. The remainder are all widespread species, with five surviving in their uredinial stage only in South Africa.

Even if the number of heteroecious species in South Africa is increased somewhat, there are far fewer species than would be expected if the proportion of heteroecism in the Northern Hemisphere (approximately 30%; TABLE 1) were to be used as a guide. Brazil has only approximately 17 heteroecious species, all of which are widespread in distribution (Hennen et al. 2005). New Zealand has two endemic heteroecious species that complete their life histories there (Mikronegeria fuchsiae, P. otagensis; Crane and Peterson 2007; Padamsee and McKenzie 2017), one widespread species that does complete its life histories there (Puccinia caricina), and another three that don’t (Puccinia polygoni-amphibii, P. scirpi, P. zoysiae; McKenzie 1998). Australia has four widespread species, of which three complete their life histories there (Puccinia caricina, P. magnusiana, P. scirpi; McAlpine 1906) and one that produces only urediniospores in Australia (Uredinopsis pteridis; McTaggart et al. 2014). Schmiedeknecht (1986) and Buriticá (2003) noted that heteroecious species are almost absent from the Neotropics. The above examples (TABLE 1) indicate that these low levels of heteroecism apply to the tropics and the Southern Hemisphere in general. Heteroecious species still do occur, mostly species on some grasses and sedges that occur in areas with long dry seasons and therefore for the majority of the year these survive as teliospores on dead plant material (Buriticá 2003). For heteroecious species to survive, both host plants need to occur in relative close proximity and to have a similar growth seasonality (Anikster and Wahl 1979). The gametothallus host needs to provide an advantage by producing aeciospores early in the growth season, most are perennial fast sprouting bulbs (Anikster and Wahl 1979), perennial plants that produce flushes of leaves rapidly with onset of favorable conditions, or woody plants that allow for perennial aecial infections (Saville 1953). However, the requirement for both the hosts to frequently occur in relative close proximity occurs less frequently in the tropics and Southern Hemisphere with the highly heterogenous and diverse plant floras of these areas than in the temperate Northern Hemisphere.

A large proportion—more than 50% in several European countries and 48% in the U.S.A.—of rust fungi in the Northern Hemisphere are macrocyclic, with all spore stages present (including both heteroecious and autoecious species) (TABLE 1). In contrast, this life history is much less frequent in the Southern Hemisphere, where contracted life histories (with only uredinia or aecia in addition to telia) dominate. The most common life history amongst South African species includes uredinia and telia only, as also occurs in Neotropical species (Buriticá 2003; Piepenbring et al. 2011; Schmiedeknecht 1986) and for which the majority of species have no known spermogonia present (TABLE 1). There are also many asexual species (in Aecidium and Uredo, or similar segregated anamorph genera) present. Jackson (1931) proposed that the major evolutionary tendencies in the rust fungi were toward a reduction in spore types in their life histories, and that this was frequently associated with the change from heterothallism to homothallism. Buller (1950) concluded that the presence of well-developed spermogonia was associated with heterothallism, whereas the lack of spermogonia was correlated with homothallism. This is a general principle rather than a rule, as exceptions occur. It is likely that additional stages will be identified with time for a number of the species currently known with assumed contracted life histories in South Africa. A number of connections have already been suggested (e.g., in Doidge 1927) but require testing. However, only 28 (45%) of the South African Aecidium species are known to have spermogonia (implying that they may be the aecial stage of species where the connections have yet to be proven). By comparison, there are 171 sporothallus species with only uredinia and telia known in their life histories. This implies that many are likely to prove to be homothallic, with spermogonia and aecia absent.

It has long been recognized that microform rust fungi, with telia only or spermogonia and telia, have a wide variety of nuclear cycles, including sexual and apomitic variations, with homothallism being common (Jackson 1935; Peterson RH 1974; Ono 2002). However, the nuclear cycle in species with long and contracted life histories was considered to be uniform with little or no variation (Jackson 1935; Ono 2002). Hiratsuka (1973) defined the aecial stage as the spore-producing structure that developed after dikaryotization and usually associated with spermogonia. In general, finding spermogonia by means of inoculation studies, or repeated field observations, has been taken as the criterion for proving the life history of a species. For species known only from uredinia and telia, these have been assumed to be either heteroecious species for which the spermogonial/aecial host has yet to be discovered or autoecious species whose spermogonia have yet to be discovered (Jackson 1931). In the temperate Northern Hemisphere, this generalization has largely held, with the majority of species with long and contracted life histories proven to be heterothallic and producing spermogonia. The majority of species shown to be homothallic are microforms (TABLE 1). Yet there are examples of species with long or contracted life histories not known to have spermogonia and proven to be homothallic by means of cultural or cytological investigation or by detailed field observation, including (i) heteroecious species: Pucciniastrum goeppertianum (Faul 1939), Puccinia coronata f. sp. elaeagni (Fraser and Ledingham 1933) and f. sp. bromi (Anikster et al. 2003), Uromyces christensenii, and U. viennot-bourginii (Anikster et al. 1980); (ii) autoecious with long life histories: Melampsora hirculi, Phragmidium arcticum, Puccinia karelica (Parmelee and Corlett 1996), and U. danthoniae (Cunningham 1931; McAlpine 1906); and (iii) autoecious with contracted life histories: Nyssopsora cedrelae (Kakishima et al. 1984), P. allii p.p. (Anikster et al. 2004), P. variabilis (= P. insperata) (Jackson 1931; Wilson and Henderson 1966), P. oxyriae (Gäuman and Müller 1957), U. behenis, U. scrophulariae (Dietel 1895), U. suksdorfii (Jackson 1931), and Trachyspora intrusa (Wilson and Henderson 1966). It is only recently that the first tropical rust with a homothallic contracted life history has been demonstrated, Austropuccinia psidii (McTaggart et al. 2020, 2018). It is postulated that the proportion of homothallic species is much greater in the tropics and Southern Hemisphere than in the temperate Northern Hemisphere; therefore, it is to be expected that spermogonia will not usually occur in their life histories. It can also be expected that modifications of the nuclear cycle as known for microcyclic species (Ono 2002) will also be discovered occasionally in species with contracted life histories.

It was early recognized that there was a tendency for life histories to be shorter in warmer climates, and colder alpine regions (Bisby 1920) and the arctic (Arthur 1928), compared with the well-known temperate Northern Hemisphere rust funga. Both shortened life histories and homothallism are common adaptions in arctic (Savile 1953, 1972) and desert conditions (Anikster and Wahl 1979). Another adaptation noted for arctic rusts is for perennial infections, allowing for a rapid response to a favorable but short growing season (Saville 1972). Climatic conditions through the tropics and Southern Hemisphere are highly variable. Much of the tropics, such as the tropical grasslands, savannas, dry forests, and arid regions outside of the rainforests, have short growing seasons due to pronounced rainfall seasonality (Berndt 2012). Prolonged droughts are also common in many regions. A shortened life history, together with homothallism, allows for an initial single basidiospore infection to initiate a local buildup of these pathogens under unpredictable climatic conditions. This confers a survival benefit under these conditions, especially when the host plant is ephemeral such as annuals or perennials that annually die back. On plants that maintain green leaves throughout the year but where rainfall is seasonal, species with long life histories are common (Buriticá 2003). In the humid tropics, where suitable infection conditions occur all year, rust species tend to perpetuate themselves as urediniospores; telia are seldom formed (Buriticá 2003; Piepenbring et al. 2011).

It is concluded that in the tropics and temperate zones of the Southern Hemisphere, including South Africa, every possible variation of the life histories of rust fungi will occur, including those that are rare or unknown in the temperate Northern Hemisphere. This includes heteroecious (rare), autoecious long and contracted life histories (with many homothallic), short and endocyclic life histories (mostly homothallic), and asexual species. Many of these variations have been found to co-occur in individual localities in South Africa (Wood, pers. observ.).

Origin and evolution of South African rust fungi.—

The Pucciniomycotina likely originated during the Carboniferous, approximately 290–359 mya (Berbee and Taylor 1993; Lutzoni et al. 2018; Taylor and Berbee 2006). Estimates of when the rust fungi themselves, the Pucciniales, originated are approximately 230–215 mya (Aime et al. 2018), 203 mya (Lutzoni et al. 2018), 115–113 mya (McTaggart et al. 2016), and approximately 175 mya (Aime and McTaggart 2021), depending on calibration points utilized. These dates coincide with the possible origin of the angiosperms in the Triassic (248–206 mya) or alternatively during their major diversification during the Cretaceous (142–65 mya) (Li et al. 2019; Magallón et al. 2015, 2013). Thus, the rust fungi originated more or less at the same time that angiosperms were originating, although it is uncertain whether the original rusts were on the early angiosperms or the gymnosperms, the latter being much more diversified and ecologically dominant from the Triassic to the early Cretaceous in comparison with today. Gymnosperms were highly diverse during the late Triassic, with many orders and families present that were lost in the extinction events that ended the Tiassic and Cretaceous (Anderson et al. 2007). The modern gymnosperm species are largely of relatively recent origin, diverging after the angiosperm radiation of the Cretaceous, with most extant genera and species originating during the Paleogene (65–24 mya) and the Neogene (24–2 mya) (Crisp and Cook 2011; Lu et al. 2014). Although various and opposing viewpoints of the earliest diverging rust fungi have been proposed (e.g., Evans 1993; Hart 1988; Hennen and Buritica 1980; Leppik 1953; Saville 1976), phylogenetic data support only Rogerpetersonia (= Caeoma) torreyae, a species known only from the gametothallus, as the earliest known extant rust species (Aime 2006). R. torreyae is the only rust known on a Taxaceae host, and the only known member of its family (Aime 2006; Aime et al. 2018; Aime and McTaggart 2021; Aime et al. 2017). Peterson RS (1974), who suggested that the species (as C. torreyae) was the precursor to other rusts genera then considered to be primitive (Milesina and Gymnosporangium), assumed R. torryea to be heterecious and postulated that the unknown sporothallus would be similar to Mikronegeria or the “Stomatosporae.” However, it is also possible that this rust is endocyclic, or asexual, derived from a heteroecious species now extinct along with the original telial host. Nuclear studies and the elucidation of the full life history of this rust fungus are of particular interest.

Various members of the next early diverging clade, the Mikronegeriineae, are heteroecious and others are autoecious. In the Zaghouaniaceae, heteroecious species have their gametothalli on the gymnosperm families Cupressaceae (Blastospora betulae (Peterson RS 1974), Mikronegeria alba (Peterson and Oehrens 1978)), Araucariaceae (Mikronegeria fagi (Cummins and Hiratsuka 2003)), and Podocarpaceae (Mikronegeria fuchsiae (Crane and Peterson 2007)) or on angiosperms (Blastospora smilacis (Ono et al. 1986), B. itoana (Ono et al. 1987)). Others are autoecious (Cystopsora oleae (Thirumalachar 1945), Cystopsora antidesmatis (Ramakrishnan and Sundaram 1952), Zaghouania phillyreae (Patouillard 1901), Elateraecium salaciicola and E. divinum (Thirumalachar et al. 1973), and Achrotelium species (Cummins and Hiratsuka 2003); all on angiosperms). Hemileia (and Uredo cryptostegiae) appears to be a recent derived group originating approximately 20.1–22.2 mya (McTaggart et al. 2016), occurring predominantly on the families Rubiaceae and Apocynaceae, which themselves originated approximately 67.7 and 52.1 mya, respectively (Magallón et al. 2015).

In the next two lineages, both small, one (Araucariomycetineae) consists of aecial species on gymnosperms (Araucariomyces balansae and A. fragiforme on Agathis (Araucariaceae)) (Peterson 1968), and the other (Skierkineae) with autoecious species on angiosperms (Aime and McTaggart 2021). The earliest suggested rust fungus fossil is that of aecia on leaves of Agathis from Patagonia, Argentina, from the early Eocene, ca. 52.2 Ma (Donovan et al. 2020), based on the similarity of the fossil to extant Araucariomyces, showing a long association of Pucciniales and gymnosperms.

The evolution of the flora of South Africa has been reviewed many times (e.g., Linder 2014; MacRae 1999; McCarthy and Rubridge 2005; Scott et al. 1997); a brief synopsis based on these references of the relevant recent past is provided here. Africa has since the breakup of Gondwana been separated from all other continents and has therefore had a distinct evolutionary path to that of South America, Antartica, and Australia from the Middle Cretaceous to the late Paleogene. However, as Africa drifted northward and approached Laurasia, interchanges have occurred intermittently with the Laurasian flora and fauna via land bridges that occurred at times of low sea level (Gheerbrant and Rage 2006). By the mid- to late Paleogene, this interchange was occurring between Africa and Europe via the Mediterranean Tethys Sill and between Africa and Asia via southwestern Asia (Gheerbrant and Rage 2006). The current vegetation patterns, and by implication the host associated pathogenic rust funga, originated during the Neogene. Prior to this, the whole of Africa was fairly uniformly covered by forest; the Guineo-Congolan forests are relics of this forest (Dagallier et al. 2020). Globally, there was a reduction in temperature from the Paleogene through the Neogene, culminating in the last ice age, this continuous cooling being associated with reducing rainfall in Africa. The development of the Sub-Saharan African savannas and grasslands developed during this period of gradual aridification, coming into their current form between 8 and 3 mya (Edwards et al. 2010; Osborne 2008). In South Africa, major geological uplifting occurred approximately 20 and 5 mya, most prominently in the east. This, together with the development of the warm Agulhas current along the east coast and the cold Benguella current along the west coast, produced a significant gradient of rainfall, highest in the east and least in the west. During the Pleistocene ice age (2–0.01 mya), desert conditions occurred over most of the interior of South Africa and the whole Kalahari basin (northwestern South Africa to the edge of the Congo basin). On the coastal side of the mountains, sea levels fell 130 m, resulting in an expanded Agulhas coastal plain with grassland or savanna vegetation (Cowling et al. 2020). The end of the ice age, and the warmer wetter interglacial Holocene since, has resulted in the retreat of the desert, and the expansion of grasslands and savannas in the northern and eastern parts of South Africa, but also the inundation of the Agulhas coastal plain. Relict Afromontane forest patches still occur in the southern temperate area or at high altitudes, while tropical forests occur along the northeastern coastal areas that have affinities with the East African and Congo forests. The Cape Floristic Region occurs in the temperate south; it has been ecologically distinct to the rest of Africa since the late Miocene (Linder 2003).

The rust funga of South Africa is dominated by the Pucciniaceae, with 275 species. There are 35 species in the Raveneliaceae, 19 in the Phakopsoraceae, 10 in the Crossopsoraceae, and 9 in the Hemileia complex, as well as 96 anamorph species. There are 22 genera each with only 1, 2, or 3 species. These represent 30 of the 49 teleomorph genera recorded from Sub-Saharan Africa (TABLE 3). The affinities of this funga are either pantropical or paleotropical, or with the northern temperate rust funga. The development of the modern plant flora, together with the rust funga present, suggests three ecologically and phyllogenetically distinct groups of rust fungi. Firstly, there is the African tropical element with members of the Mikronegerineae (Hemileia), the Sphaerophragmiaceae (Puccorchidium, Sphaerophragmium), and other genera of the Uredinineae (Stomatisora). Their immediate ancestors likely occurred in the tropical forests that covered most of Africa during the Paleogene, although these lineages may well have originated elsewhere. Hemileia has 31 species recorded from Africa, 7 in Asia, 4 in both Africa and Asia, and only one in South America (Ritschel 2005), suggesting that it may have originated in the African tropical forests. The Sphaerophragmiaceae may also have originated in Africa, suggested by the, at present, greater generic diversity present on the continent. The Sphaerophragmiaceae appears to be associated with the Annonaceae, in both South America and Africa (Beenken and Wood 2015; Beenken et al. 2012). This is a large pantropical family of trees and lianas and is the most diverse within the Magnoliales. It is particularly associated with tropical rainforests (Courvreur et al. 2011). It has been postulated that it originated in Africa and from there radiated through the pantropics via the Boreotropical route to the Americas in the Paleocene, and a number of subsequent dispersal events to Asia in the Miocene (Couvreur et al. 2011; Thomas et al. 2015). Secondly, there is a pantropical element including the Raveneliaceae (Diorchidium, Maravalia, Ravenelia sensu lato, Uropyxis) that is predominantely associated with the Caesalpinioideae (which includes the mimosoids). The Raveneliaceae is particularly diverse in the Neotropics, and more diverse in the Asian tropics than Africa (Berndt 2012). It likely originated outside of Africa but spread there and diversified during the Neogene, when mimosoids became the dominant trees of the developing African savannas. South Africa has the highest known diversity of Ravenelia sensu lato in any Sub-Saharan African country (Wood, unpubl. data). Various other genera have a similar distribution, e.g., Bubakia, Chaconia, Crossopsora, Goplana, Olivea, Phakopsora, Skierka, Scopella, Sorataea, and Uredopeltis, each with only a few to many species in Sub-Saharan Africa, whereas some genera occur in both Asia and Sub-Saharan Africa only, e.g., Didymopsorella and Elateraecium. Thirdly, the Pucciniaceae likely originated in the Northern Hemisphere but invaded Africa as this continent pushed northward closing the Tethys Sea and colliding with Europe (Gheerbrant and Rage 2006). They diversified with the development of the grasslands and savannas as these become the dominant habitat in most of Africa, particularly on the grasses and associated herbaceous Asteraceae and Fabaceae. The greatest diversity at both genus and species levels of rust fungi is found in the warmer, wetter, northeastern regions of South Africa. The successive biotic interchanges between Africa and Europe and Asia (Gheerbrant and Rage 2006) have resulted in a number of groups of rust fungi entering into and then diversifying within Africa at different times, resulting in the low level of endemicity at the genus level but high levels of endemism at the species level.

Table 3. Biodiversity of rust fungi in Africa, showing number of species for all genera recorded in representative countries.

Taxon	Rust genus	South Africa	Zimbabwe	Uganda	Nigeria	West Africa^a
Mikronegerineae	Achrotelium					1
	Elateraecium	2				1
	Hemileia + “Uredo”	9	3	18	7	18
Skierkineae	Skierka	1	1	2	1	1
Melampsorineae	Coleosporium	3		2		1
	Melampsora	1	1	1
	Milesina	4
	Thekopsora	1
	Uredinopsis	1				1
Raveneliineae	Angusia	1
	Bubakia	2
	Cerotelium			3	3	5
	Goplana					2
	Maravalia	2
	Masseeela	1	1			1
	Phakopsora sensu lato	7	2	11	3	5
	Phragmidiella			1	1	1
	Uredopeltis	6		3	1	3
	Chaconia					1
	Diorchidium	1	1		1	2
	Didymopsorella	1
	Hapalophragmium				2	5
	Kernkampella				1
	Newinia		1		1
	Olivea			1	1	1
	Phragmopyxis					1
	Ravenelia sensu lato	32	6	8	5	5
	Sorataea				2	2
	Spumula		1	1
	Ypsilispora				1	2
	Uropyxis	1	1
Uredinineae	Kuehneola	1		1		3
	Hamaspora	1		2		1
	Trachyspora	1	1	1
	Endophylloides					1
	Puccorchidium	1
	Sphaerophragmium	4		2	2	9
	Sphenorchidium					1
	Angiopsora		3	3	1
	Crossopsora	5	3	4	1	2
	Kweilinga				1	1
	Stomatisora	1
	Dietelia	1		1	1	1
	Endophyllum	9	1	7	2	3
	Macuropyxis	1
	Miyagia			1
	Puccinia	185	51	131	48	70
	Pucciniosira	2		1	2	3
	Uromyces	77	26	53	10	24
Form genera	Aecidium	64	24	44	7	42
	Uraecium			1		2
	Uredo	32	8	34	7	34
Totals		461	132	337	112	257

^aIncludes Ivory Coast, Ghana, Guinea, Siera Leone, and Senegal.

The diverse flora of the Cape Floristic Region (largely congruent with the Fynbos biome, with ~9000 plant species, 69% endemism) and Succulent Karoo (~6300 plant species, 40% endemism), as well as the temperate Afromontane forests (~1400 plant species, low endemism) in South Africa (Mucina and Rutherford 2006), likely is an example of a flora that has been recently invaded by the Pucciniales. These biomes have a depauperate rust funga, with only 81, 37, and 17 rust species recorded, respectively (SUPPLEMENTARY TABLE 2). This rust funga consists of 7 species in the Milesinaceae (representing the greatest diversity of this group in Africa), 98 species in the Pucciniaceae, and only 9 non-Pucciniaceae species. Yet 55 (49%) of these species are endemic to one (or occasionally two) of these biomes. The majority of species are in the clade including U. ixiae of the Pucciniaceae (FIG. 6). This pattern of restricted diversity at the family and genus levels is very likely the result of a recent spread, and subsequent diversification, into these biomes. In addition to the Pucciniaceae, the relatively few members of the Melampsorineae in South Africa are also likely recent additions due to extreme long-distance dispersal events and largely confined to the most temperate areas within South Africa.

Conclusions

A phylogeny was generated from a total of 166 specimens, representing 110 species of South African rust fungi. Included were representatives of all families and most genera present in the country. Genera not included were Didymopsorella, Dietelia, Elateraecium, Melampsora, Milesina, Trachyspora, and Uropyxis. The diversity is representative of Sub-Saharan Africa as a whole. One new species is described, 12 new combinations made, six names reinstated, and two aecial stages were connected with their sporothallus stages based on our results. The life histories of the South African rusts were summarized and compared with those of other countries from around the world. In comparison with countries in the temperate Northern Hemisphere, a high proportion of the South African funga are only known from uredinia and telia (170 species, 37% of the total diversity). It is postulated that many of these will prove to be homothallic and will not produce spermogonia nor an aecial stage. A shortened life history, together with homothallism, allows for a single basidiospore infection to initiate a local buildup of these pathogens under unpredictable climatic conditions, as is prevalent for much of South Africa. It is further postulated that some of the variations in nuclear cycle known for microcyclic rust fungi will occasionally be found to also occur in some of these species.

From the current diversity of rust fungi, together with the patterns of plant (i.e., their hosts) evolution and development of the current broad plant communities (phytochoria, biomes) in Sub-Saharan Africa, it is concluded that the rust funga of South Africa is predominantly of relatively recent evolutionary origin, expanding from tropical Africa (Hemileia and Sphaerophragmiaceae), and for many groups originally from Europe or Southeast Asia, speciating within the relatively recent past. In particular, it is likely that rust fungi have only penetrated the Fynbos and Succulent Karoo biomes within the late Pleistocene, or even the Holocene, diversifying with frequent host jumps (especially the clade including U. ixiae of the Pucciniaceae) (Aime et al. 2018; McTaggart et al. 2016) as well as radiation within host groups by pseudo-cospeciation (radiation of pathogens postdating the original radiation of the hosts, fide Choi and Thines 2015). The Raveneliaceae and Sphaerophragmiaceae likely have had a long association with the tropical African vegetation that extends southward into South Africa. The Raveneliaceae is likely a good candidate group to test for cospeciation, host jumps, or pseudo-cospeciation as paths of diversification amongst Sub-Saharan African rusts, as it is both diverse and largely restricted to the Fabaceae. Likewise, the representatives of the Melampsorineae in the Southern Hemisphere would be a good candidate group to test for speciation following extreme long-distance dispersal coupled with both host jumps to plant species related to their parental hosts and shortening of the life history.

In comparison with the rust funga of the Northern Hemisphere, little work has been done on the rust funga of the southern continents, other than basic morphological descriptions. There are notable recent papers that are addressing this lack of studies, but almost none from Sub-Saharan Africa. This gap should be addressed; it can be expected that the funga of South Africa, and Africa as a whole, will provide useful comparative data on rust fungus origins, evolution, mycogeography, speciation mechanisms, and life history variations.

ACKNOWLEDGMENTS

The authors are grateful to Lucy Liu, Elena Karlsen-Ayala, Raman Kaur, and Gwen Samuels for technical laboratory assistance and to the staff of the Arthur Fungarium and the U.S. National Fungus Collections for specimens.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

SUPPLEMENTARY MATERIAL

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00275514.2024.2334189

Additional information

Funding

Funding in part was provided by the U.S. National Science Foundation [DEB-2127290 and DEB-1458290], the U.S. Department of Agriculture (USDA) [AP20PPQS and T00C077], and a USDA Hatch grant [no. 1010662].