﻿Parallel developments in floral adaptations to obligate moth pollination mutualism in tribe Phyllantheae (Phyllanthaceae)

﻿Abstract Several groups within tribe Phyllantheae (Phyllanthaceae) formed, independently, an (obligate) pollination mutualism with Epicephala moths, which originally had been parasitic. In this pollination system, female moths actively collect pollen from staminate flowers and deposit it on the stigma of pistillate flowers, after which they place at least one egg in or against the ovary. The high pollination rate makes the system beneficial for the plants, whereas the larvae are provided with food (part of the developing seeds) and some protection against predation. Qualitative comparisons are made between non-moth-pollinated lineages, used as outgroups and various, independently moth-pollinated Phyllantheae clades, used as ingroups, thereby looking for parallel developments. The flowers of both sexes of various groups display similar, convergent morphological adaptations to the pollination system, likely to secure the obligate relationship and to improve efficiency. Sepals in both sexes, free or partly to highly connate, are commonly upright and form a narrow tube. The staminate flowers often have united, vertical stamens with the anthers along the androphore or on top of the androphore. Pistillate flowers generally reduce the stigmatic surface, either by making the stigmas shorter or by uniting them into a cone with a small opening at the top for pollen deposition. Less obvious is the reduction of the stigmatic papillae; these are often present in non-moth-pollinated taxa, but absent in the moth-pollinated species. The most diverging, parallel adaptations to moth pollination are currently found in the Palaeotropics, whereas in the Neotropics, some groups continue to also be pollinated by other insect groups and are morphologically less changed.


Introduction
Obligate pollination mutualisms between insects and plants have developed in several plant families and they provide interesting study systems for reciprocal morphological evolution and adaptation between the partners. Kato and Kawakita (2017, chapter 13) provide a nice overview of the various plant groups and their pollinators. Three examples where the relationship between both parties is obligate and where a proliferation of co-speciation occurred, are famous: figs and fig-wasps, Yucca L. and Yucca moths and leafflower and leafflower moths (Kato and Kawakita 2017). The latter group forms the main topic of this paper, particularly tribe Phyllantheae Dumort. . The tribe mainly consists of the large (paraphyletic) Phyllanthus L. (Wurdack et al. 2004;Samuel et al. 2005;Kathriarachchi et al. 2005Kathriarachchi et al. , 2006, a clade that also includes Breynia J.R.Forst. & G.Forst., Glochidion J.R.Forst. & G.Forst. and Synostemon F.Muell. Kato et al. (2003) were the first to report the obligate pollination mutualism between the Phyllanthaceae plants (genus Glochidion) and the Epicephala moths. Most active research on this relationship was done by Kawakita and Kato and their results are summarised in their book (Kato and Kawakita 2017). It is now obvious that the relationship between Phyllantheae and moths originated at least six times independently (Kawakita and Kato 2009), which is confirmed by the phylogeny of Bouman et al. (2021), see the (partly) red-boxed groups in Fig. 1A, likely also several times in the genus Phyllanthus (Fig. 1B, see also Discussion). The morphology of staminate and pistillate flowers changed and adapted accordingly (for example, see Zhang et al. (2012); Yang and Li (2018)). The female moths of mutualistic species actively collect pollen from staminate flowers, whereby the sensilla or setae (hairs) on the proboscis sweep up the pollen (these hairs are absent in male moths and in (cheating) moths with no mutualistic relationship; Kato and Kawakita (2017); Yang and Li (2018); Wang et al. (2020)). The female moths then place the pollen on the stigmas of the pistillate flowers and finally place an egg either between the calyx and the ovary, or in the ovary by drilling through the stigma or through the ovary wall (Kato et al. 2003). This system provides the plants with a (usually) specific pollinator and, in return, the moths gain shelter and food for the developing larvae, which feed on some of the developing seeds. Seed consumption is variable and indicative of variations in the relationship between plant and pollinator; in some cases, only some of the seeds are consumed, while in others all can be eaten (Zhang et al. 2012).
Epicephala belongs to the family Gracillariidae, a family of micro-lepidoptera with ca. 100 genera and 2000 species (see Kato and Kawakita (2017): chapter 5, for a good overview and references). Most of them are leaf miners, which sometimes create galls. A few are plant-borers that attack seeds or other parts of the plant. Epicephala is part  Bouman et al. (2022)) A the new classification runs from right to left B shows the detail of the classification and phylogeny of Phyllanthus L. Red boxes in A indicate the six groups with moth pollination; in green are the clades used as outgroups with which they are morphologically compared. Boxes with green and red represent clades with first splitting off the non-moth-pollinated species and in the top part of the clade the moth-pollinated ones. In blue boxes (B), two South American groups in Phyllanthus are highlighted, which, based on morphology, might also have moth pollination. of the generally parasitic leaf-borer group, which has been found to occur also in tribe Phyllantheae (Luo et al. 2011). The phylogeny of Epicephala (Kato and Kawakita 2017: fig. 5.3) shows (disputably) Conopomorpha flueggella Li, 2011, as outgroup to the Epicephala clade. The first group to split off is the New Caledonian group (Kato and Kawakita 2017: fig. 5.3: clade 7). The species in this group are still seed parasites (like C. fluegella) (Hu et al. 2011) and lack the sensilla and ovipositor. The species in the remaining six clades have sensilla and an ovipositor and developed the mutualistic pollination syndrome. The exception is clade 2 (Kato and Kawakita 2017: fig. 5.3), where the moths returned to parasitism in the herbaceous Phyllanthus species. Conopomorpha and clade 7 (Kato and Kawakita 2017: fig. 5.3) are still seed predators, whereby seeds possibly survive due to a high mortality rate amongst the larvae. Conopomorpha flueggella feeds on the seeds of Flueggea Willd. Flueggea is sister to Phyllanthus sensu lato (Wurdack et al. 2004;Bouman et al. 2021) and did not develop pollination mutualism. Flueggea has very open flowers, whereas the moth-pollinated species generally show more closed flowers. Similarly, seed predation has only been found in developing fruits in species of Phyllanthus subgenus Macraea (Wight) Jean F.Brunel and not in flowers that were yet to be pollinated, which suggests a solely parasitic relationship (Kato and Kawakita 2017). With Breynia and (former) Sauropus Blume (now subsumed in Breynia), it was obvious that both genera were distinguished because of morphological differences in the flowers, seemingly caused by moth pollination (van Welzen 2003). A similar difference seems to be present between Glochidion and Phyllanthus subgen. Phyllanthodendron (Hemsl.) G.L.Webster.
The aims of this paper are: 1. to show the qualitative morphological changes in the various flowers of the different plant groups associated with the pollination mutualism by Epicephala moths in comparison with their non-mutualistic relatives and 2. to see if there are parallel developments in the morphological adaptations of the (confirmed) six different groups of plants that are now pollinated by the moths, with also a consideration of possibilities in the New World. Kawakita and Kato (2009: fig. 3) listed six independent origins of obligate moth pollination and these groups will be treated here (sequence as in Kawakita and Kato (2009)). For each group, simple qualitative descriptions or comparisons are provided of the general type of staminate and pistillate flowers. The species selected are those mentioned either by Kawakita and Kato (2009) or by Bouman et al. (2021) in the new phylogeny of the Phyllantheae. Qualitative descriptions of staminate and/or pistillate flowers were made if specimens were available in the herbarium material of Naturalis Biodiversity Center (L, U and WAG in Index Herbariorum, http://sweetgum.nybg.org/science/ ih/), but for many species, specimens were lacking or they were too poor to be used or they only showed one sex. The flowers in the moth-pollinated taxa (red-boxed in Fig. 1) are compared with taxa in their sister group (green-boxed in Fig. 1) to observe possible morphological changes in flowers, likely induced by the moth pollination. The groups treated here are all Old World taxa. In a more recent paper, Kawakita et al. (2019) also demonstrated moth pollination in the New World (Fig. 1B), so these taxa, blue-boxed in Fig. 1, are discussed as well.

Materials and methods
The genus Phyllanthus in its present circumscription is paraphyletic because Breynia (including Sauropus), Glochidion and Synostemon are part of it. Two views exist, either to include the four genera into one large genus (e.g. Hoffmann et al. (2006)) or to subdivide Phyllanthus in smaller, recognisable genera (e.g. van Welzen et al. (2014)). The latter option has now become feasible with a phylogeny based on a much larger sampling (Bouman et al. 2021). In the rest of the text, we will present the current names and the new classification of Bouman et al. (2022) with the newly-proposed names. Fig. 1 shows the new classification and simplified phylogeny of tribe Phyllantheae (Bouman et al. 2022). Table 1 shows the names in Phyllanthus as used by Kawakita and Kato (2009) and the new names according to Bouman et al. (2022).
The terms style and stigma need some explanation as these are sometimes used differently. The style is the united part (of the stigmas) on top of the ovary, it can be present ( Fig. 2b) or absent (Fig. 3c). The stigmas are the split part. There are usually three stigmas, which, at the end, usually split into two short lobes. The stigmatic tissue, the part receiving the pollen, is often the papillate part on the adaxial side of the stigmas (Figs 3c, 12b). However, in various species, these are absent (e.g. Fig. 14b) and, therefore, the three arms are considered as stigmas. This is concordant with the use of the terms in the Flora Malesiana revisions of the Euphorbiaceae s.l. (including Phyllanthaceae): www.nationaalherbarium.nl/euphorbs/.

Results
The general flower type in the Phyllantheae shows usually six sepals (in two whorls of three), a nectar disc (annular or consisting of three or six separate glands) and then either three or more stamens in the staminate flowers (free or variously connate; Fig. 2a) or an ovary with usually three locules with two ovules per locule and on top a short style branching into the stigmas (Fig. 2b). The latter are often split at the top (Fig. 5c). For each group, the possible changes in these organs are discussed and Table 2 shows short descriptions of both sexes of flowers for the species mentioned in the text and an indication whether or not they are (likely) moth-pollinated.
Both genera were formerly separated because of the strong differences in floral morphology in both sexes. The flowers of both sexes in former Sauropus are usually flat, open and disc-like. The staminate flowers (Fig. 3a) have six sepals, basally to almost completely united; they lack nectar glands, but (usually) have scales inside, which are  probably transformed disc glands. The scales keep the flower closed when the stamens are still immature. The filaments of the three stamens are basally united into a column, but at the top they form three, ± horizontal, free arms with the anthers underneath. The ovary of the pistillate flowers (Fig. 3b) has a flat top, on which there are three, welldeveloped stigmas that are usually flat and split at the apex and resemble a crescentmoon with papillae to catch pollen (Fig. 3c). Exceptions are that, in some species, the stigmas are erect and, in some, the staminate flowers lack the scales, but instead they have three sepals that are folded inwards and the stamens are enlarged and point diagonally upwards. The potential pollinators of these 'Sauropus-flower' species have not yet been elucidated by any study.
Breynia subgen. Breynia sect. Breynia (Fig. 1A) is pollinated by moths Kato 2004b, 2009). The staminate flowers are narrow and campanulate, with united sepals forming a tube, with the disc scales on the top inside (sometimes the sepal lobes are reduced to a thickened ring and the scales resemble the calyx lobes; Fig. 4a). Within the tube, three stamens are united into a massive stalk (androphore) with the anthers orientated vertically along it (Fig. 4b). The pistillate flowers are less open than in former Sauropus, but the main difference is in the stigmas. The style is lacking and the stigmas are greatly reduced in size, sometimes completely split (Fig. 4c). In some species, there are six short vertical stalks on top of the ovary without any stigmatic papillae. Only two species in this section, B. fruticosa (L.) Müll.Arg. and B. glauca Craib (Fig. 4), still have larger stigmas, but these are not united in a style. Furukawa and  showed that a gynophore (of variable length) often develops after fruit set in B. officinalis Hemsl. on Amami-Oshima Island (Ryukyu Islands, Japan), also mentioned by Zhang et al. (2012), which likely precludes overconsumption by Epicephala vitisidaea Li, Wang & Zhang, 2012, as the larvae have to eat their way up through the gynophore (unfortunately, they used the name B. vitisidaea (Burm.f.) C.E.C.Fisch., a species not present in the Ryukyu Islands and southern China and which definitely lacks a gynophore; see van Welzen and Esser (2005)).
Seemingly, in comparison to the outgroup, the flowers in the ancestral species of sect. Breynia evolved in response to moth pollination: the staminate flowers became closed, campanulate and, inside, the anthers became vertical with united stamens (Fig. 4a, b); in the pistillate flowers, the stigmas are strongly reduced (Fig. 4c). Breynia retusa shows a reversal towards a normal style and stigma (Fig. 5c), thus likely this species underwent a host switch. Kato and Kawakita (2017: (Figs. 5a,b), but more open than other flowers of the moth-pollinated species.
Stigmatic papillae are often present in Breynia subgen. Breynia sect. Cryptogynium and Breynia subgen. Sauropus (Fig. 2c), but absent or strongly reduced in Breynia subgen. Breynia sect. Breynia (Fig. 4). The disappearance of the papillae may be another adaptation to moth pollination. Fig. 1A; Table 2) A similar situation as with Breynia occurred in these two groups.  (2017)). The staminate flowers of former Phyllanthodendron are rather open, with five or six basally connate sepals with an attenuate apex ( Fig. 6a, b), the disc glands are often large and somewhat petal-like, the generally three stamens have connate filaments and diverging upper parts and the connectives have an apical appendage. The pistillate flowers (Fig. 6c, d) have similar sepals and disc glands as the staminate flowers and on top of the ovary are well-developed stigmas (often with a short style) that are apically entire or shortly bifid.

Glochidion versus former Phyllanthus subgenus Phyllanthodendron (Group 2 in
In subgenus Glochidion, which is moth-pollinated, flowers of both sexes lack a disc. The staminate flowers (Fig. 7a) are campanulate and the androecium resembles that of Breynia (Fig. 5b) as the filaments are vertical and tightly together with the anthers along them and the connectives have apical appendages that form a pyramidal cone. When wilting, the stamens start to detach from each other and bend outwards and then resemble the stamens of the staminate flowers in Glochidion subgenera Phyllanthodendron and Pseudoactephila. In the pistillate flowers (Fig. 7b, c), the stigmas are upright and united into a cone terminating in the often slightly split apices of the stigmas that form a cavity in the middle where the female moth deposits the pollen (Fig. 7b, c).
As with the first group (Breynia, including former Sauropus), the differences in flower morphology led to this group being separated into two genera. Based on the phylogeny (Bouman et al. 2021) and the resulting new classification (Bouman et al. 202), both former genera are now united under Glochidion (Phyllanthodendron was paraphyletic and Glochidion is the older name). The staminate flowers in subgenus Glochidion (Fig. 1A) are narrower than in the other two subgenera with more strongly united stamens and the connective appendages touching, while in the pistillate flowers, the stigmatic surface is reduced by uniting the stigmas into an erect pyramidal structure. In both sexes, the disc/disc glands are gone, as seemingly the moths do not need a nectar stimulus. As in Breynia, there is one Glochidion species, G. sericeum (Blume) Zoll. & Moritzi, showing a reversal, not in the staminate flowers ( Fig. 8a, b) but in the pistillate flowers, where a well-developed style (covered by the sepals) with free, spreading stigmas is present (Fig. 8c, d); Kato and Kawakita (2017: fig. 10.2) did not record any Epicephala moth on this species.   Table 2) Cicca L. is separated from Phyllanthus (Bouman et al. 2022) and restricted to a few subgenera, most of which are only found in Madagascar. Of the local outgroup, Cicca subgenus Betsileani (Jean F.Brunel) R.W.Bouman, only C. betsileani (Leandri) R.W.Bouman could be studied, a non-moth-pollinated species. Its flowers are open with six thin sepals; the staminate flowers have six disc glands and three free stamens; the pistillate flowers have a 3-locular ovary bearing terminal, well-developed, spreading stigmas that are split in the upper 2/3.
Cicca subgenus Menarda (Comm. ex A.Juss.) R.W.Bouman has staminate and pistillate flowers with 5(6) sepals; the staminate flowers have separate disc glands and three or five stamens; the pistillate flowers show a circular disc or separate disc glands and a 3-locular ovary (a revision and drawings can be found in Ralimanana and Cable (2020)). In Kato and Kawakita (2017), subgenus Menarda is represented by Cicca humbertii (Leandri) R.W.Bouman and C. marojejiensis (Leandri) R.W.Bouman (latter not included in this study) as being moth-pollinated. Various species of subgenus Menarda were investigated. Morphologically, Cicca humbertii (Leandri) R.W.Bouman, C. perrieri (Leandri) R.W.Bouman (Fig. 9) and C. sambiranensis (Leandri) R.W.Bouman have staminate flowers with (rather) stiff, upright sepals, making the flowers narrow and three united filaments with anthers vertically along the androphore in C. humbertii, Unlike in the previous two groups, the switch to moth pollination resulted in less distinct differences in the flowers; therefore, the moth-and non-moth-pollinated species had not been divided into separate genera. In comparison to the former two groups, C. betsileani, C. coodei and C. cryptophila conform in their morphology to non-mothpollinated with open flowers with either well-developed stigmas or free, spreading stamens. Cicca cryptophila is in the basal clade of subgenus Menarda (Bouman et al. 2022: suppl. fig . 1). The other three species, C. humbertii, C. perrieri and C. sambiranensis, Table 2. Possible adaptations or lack of adaptations to moth pollination in staminate and pistillate flowers. Group refers to the groups as discussed in the text and in Fig. 1, I means Ingroup, O = local outgroup (related clade/taxon with which the ingroup is compared). The classification follows Bouman et al. (2022). Adapted: -= primitively not adapted to moth pollination; + = adapted to moth pollination, obs.+ = moth adaptation observed (Kawakita et al. 2019), but morphologically not adapted (morph.-); R = reversal to non-moth pollination. are in higher clades. Their pistillate flowers are similar to those of the moth-pollinated subgenus Glochidion in the cone-like united stigmas. Their staminate flowers have narrow, stiff sepals, but differ in the degree of connation of the stamens. As few species could be observed, it is not possible to show where in the Cicca group moth pollination started, but likely between the basal group (with C. cryptophila) and the upper clades (which partly form a trichotomy).

Genus/Species Infrageneric taxon Group
Due to the high similarity with the pistillate flowers of Glochidion, several of the species in Cicca were formerly described or transferred to Glochidion (see Leandri (1937); Hoffmann and McPherson (2003)), but generally differ in the presence of disc glands, which are absent in Glochidion (Table 2). Fig. 1A, Table 2) With this group, Cicca betsileani can serve as a non-moth-pollinated outgroup; see the previous group for a short description.

Dendrophyllanthus section Dendrophyllanthus (Group 4 in
Dendrophyllanthus S.Moore sect. Dendrophyllanthus is a New Caledonian group that comprises moth-and non-moth-pollinated species (Table 2) based on the comparison with the outgroup. However, the staminate flowers were often not seen or were young and in bud. They do not present a clear picture of morphological adaptations. The stamens are generally erect and free to completely united ( Table 2). The pistillate flowers show either large, well-developed stigmas, generally free and sometimes recurved, or the stigmas are erect, free or united, sometimes small, in a tight upright circle with the stigmatic tissue on the inside (Table 2). Dendrophyllanthus clamboides (F.Muell.) R.W.Bouman (Fig. 10) shows united stamens and stigmas and is likely moth-pollinated like D. glochidioides (Elmer) R.W.Bouman with partly free stamens (Fig. 11). The interpretation that united stamens and stigmas are indicative of moth pollination is not necessarily always correct. Dendrophyllanthus buxoides (Guillaumin) R.W.Bouman and D. caudatus (Müll.Arg.) R.W.Bouman are recorded by Kawakita and Kato (2004a) to be moth-pollinated, but their sturdy stigmas point at the opposite conclusion, though the stigmas are more or less upright (slightly spreading) and, in case of D. caudatus, bent inwards towards each other (only seen in young fruit). Dendrophyllanthus wilkesianus (Müll.Arg.) R.W.Bouman (Fig. 12) shows more or less united stamens (but small anthers with horizontal slits), but free, recurved, papillate stigmas and likely is not moth-pollinated.   Unfortunately, the published phylogenies for this group either only contain mothpollinated species (Kawakita and Kato (2009), as Phyllanthus subgen. Gomphidium sect. Adenoglochidion) or show a basal polytomy (Bouman et al. 2021). This means that it is uncertain where moth pollination started in the phylogeny and if there are species reversing to non-moth pollination. Additionally, due to lack of material, not all species mentioned by Kato (2004a, 2009) could be analysed. Fig. 1A; Table 2) Dendrophyllanthus S.Moore sect. Leptonema (Baill.) R.W.Bouman is an Australian-New Caledonian group that shows the same developments as the previous group. Again, C. betsileani, which has free stigmas, can serve as the outgroup. The staminate flowers do not show any obvious morphological adaptations to moth-pollination and the pistillate flowers show either well-developed spreading stigmas or cone-like, united or free, upright stigmas (Table 2). Thus, some of the species in this group may already have adapted to moth pollination, some not. Kawakita and Kato (2004a, as Phyllanthus subgen. Gomphidium sect. Gomphidium) listed a number of moth-pollinated species and three non-moth-pollinated species. The overlap with the species investigated here is very small with only two moth-pollinated species, of which D. aeneus (Baill.) R.W.Bouman has pistillate flowers that are adapted to moth pollination, but D. vulcani (Guillaumin.) R.W.Bouman seemingly has not (yet) adapted morphologically as it has well-developed, spreading stigmas. The phylogeny of Kawakita and Kato (2009) included only moth-pollinated species, whereas the phylogeny of Bouman et al. (2021) showed that, perhaps in part of the clade, with D. ligustrifolius (S.Moore) R.W.Bouman at the base, a switch to moth pollination occurred, because D. ligustrifolius likely is moth-pollinated (Fig. 13) and D. hypospodius (F.Muell.) R.W.Bouman (Fig. 14) and D. serpentinus (S.Moore) R.W.Bouman likely are not moth-pollinated. Unfortunately, the upper part of the clade comprises a few polytomies and weakly-supported branches, thus this may change in the future when more species and/or markers are added. In this group, D. bupleuroides (Baill.) R.W.Bouman, D. kanalensis (Baill.) R.W.Bouman and D. loranthoides (Baill.) R.W.Bouman are probably not moth-pollinated; D. vulcani is disputable (it seems to have moth pollination Kato 2004a, 2009) Table 2) In Kirganelia, section Pseudomenarda (Müll.Arg.) R.W.Bouman is the non-mothpollinated outgroup to section Kirganelia, where moth pollination is recorded. Section Pseudomenarda is a small taxon with two species, of which K. somalensis (Hutch.) R.W.Bouman (Fig. 16) was seen. It shows staminate flowers of 2.5-3 mm in diameter with five sepals, five disc lobes, two outer free and three inner united stamens. The pistillate flowers are very open, ca. 3.5 mm in diameter, with five relatively large sepals, a broad, ring-like disc and a 5-locular (can be 3-locular; see Hutchinson (1912)) ovary with a terminal, ca. 0.3 mm long style and then five horizontally spreading stigmas ca. 1 mm long, split in the upper third.

Dendrophyllanthus section Leptonema (Group 5 in
Section Kirganelia, Kirganelia reticulata (Poir.) Baill. was reported to be mothpollinated (Kawakita et al. 2015;Fig. 15). The staminate flowers are very small and inconspicuous (ca. 2 mm in diameter) with five or six sepals in two whorls, five or six small disc glands and five or six stamens in two very tight whorls, outer two or three free, inner three with filaments basally connate; these staminate flowers are usually  found unopened in dried specimens. The pistillate flowers have the same number of sepals and disc lobes as the staminate flowers, the ovary is small with four or five short stigmas, slightly split apically, bent towards each other and forming a circular cone that is open on the inside. Kawakita et al. (2015) reported that six moth species visit the flowers of K. reticulata, of which three have become parasitic again, causing the ovules/ fruits to gall and one species also induces the fruit to become inflated. A possible reason for the reversal to parasitism is protection against predation by specialised braconid wasps (Kawakita et al. 2015). Predation by the wasps is likely as K. reticulata has relatively small fruits (ca. 4.5 mm in diameter), thus long ovipositors are not needed.
Adaptation in flower morphology of K. reticulata for moth pollination is seemingly only visible in the pistillate flowers; the staminate flowers with tightly, vertically grouped stamens are, at most, smaller than those in other species and then less accessible for various groups of insects. As in Breynia section Breynia, the stigmas in the pistillate flowers are shorter and, as in Glochidion, the stigmas are united, upright and short, forming an erect ring on top of the ovary and forming a depression inside. In contrast, in section Pseudomenarda, the stigmas are still well-developed and spreading. Additionally, some species within the genus, belonging to the now-subsumed group Phyllanthus section Hemicicca (Baill.) Müll.Arg., possess different flower features; especially K. flexuosa (Siebold & Zucc.) R.W.Bouman, which has staminate flowers with reddish, spreading sepals and two free stamens (see Kato and Kawakita (2017)). Their pollinators are unknown.

Palaeotropics
As a result of obligate leafflower moth pollination, adaptations in the morphology of staminate and pistillate flowers were expected because they can improve pollination efficiency and the fit with the moths, while also optimising pollen uptake, pollen deposition and oviposition for the pollinators. However, because the mutualistic relationship has evolved several times in tribe Phyllantheae, there could be differences in morphological adaptations or similar morphological patterns in the different groups could have arisen through convergent evolution. Recent studies have shown that the mutualism between leafflowers and leafflower moths is more complex than first described (Kato et al. 2003) as there are differences in the level of species specificity Kato 2004a, 2006;Zhang et al. 2012;Li et al. 2015;Yang and Li 2018), cheaters of the mutualism (Kawakita et al. 2015;Wang et al. 2020) and defences against this antagonistic relationship Furukawa and Kawakita 2017). Furthermore, possible reversals in flowers are indicated here.
Flowers associated with the obligate moth pollination mutualism show, as a visual stimulus to attract the nocturnally active Epicephala moths, a contrast between the light (green to yellowish) sepals and the dark night sky, but this is also observed in nonmoth-pollinated flowers. Olfactory cues were detected in a number of studies, differences that likely promoted speciation as they differed either in strength or composition between the sexes (Svensson et al. 2010;Okamoto et al. 2013;Okamoto 2017;Huang et al. 2015;Zhang et al. 2016). Okamoto et al. (2007; see also Okamoto (2017)) showed that Glochidion subgenus Glochidion produces floral odours, often typical of each species, but different between flower sexes, that attract the female moths. Additionally, Breynia vitis-idaea (Burm.f.) C.E.C.Fisch. produces two floral odours. Attractants are likely crucial, especially during the nocturnal pollination and differences in smell between the sexes can help guide the moth first to the staminate and later the pistillate flowers. Therefore, nectar glands like the disc or disc glands may still be as important as the tissue to emit the odours; the moths themselves do not seek nectar and the plants likely do not attract other insects. The glands/discs are present in most Phyllantheae, except in Glochidion subgenus Glochidion and Breynia; in these taxa, perhaps the receptacle produces the odour. The two other subgenera of Glochidion (Pseudoactephila and Phyllanthodendron) have disc glands in the flowers, but these have become almost petal-like and likely are not producing nectar any longer.
The majority of the moth-pollinated flowers are small with no bright colours, while the narrow shape of many is geared more towards a mechanical fit to the pollinator. The adaptations in staminate flowers that seem most common are a tight, cylindrical ring of sepals, with united, or at least upright, stamens with the anthers vertically along the androphore (opening towards the sepals) or with the anthers at the top of the androphore. This is especially demonstrated in Breynia section Breynia and Glochidion subgenus Glochidion (groups 1 and 2, respectively). With herbarium material, it is difficult to access the tightness of the sepals; therefore, this is often not noted, except when the sepals are thick and upright like several species in Cicca subgenus Menarda (group 3), Pyllanthus graveolens (Phyllanthus subgenus Conami; group 7) and several species in Phyllanthus subgenus Microglochidion (Group 8). If the staminate flowers produce a tight calyx cylinder, then as soon as the moth probes with its proboscis, covered with sensilla, the sensilla become dusted with pollen. This seemingly is a very effective mechanism.
Female moths have to transfer the pollen to the pistillate flowers, which is generally facilitated by the plant as most Phyllantheae species are monoecious and may have fascicles with both sexes (often many-flowered in Glochidion), whereby the sexes can be receptive at different moments, or the sexes are separated in space on one branch with the staminate flowers proximally and the pistillate flowers distally (often in Breynia and Phyllanthus s.l.). A few species are reported as dioecious, but this is not easy to judge from herbarium specimens; it can also be the result of dichogamy (an extended separation in time as the staminate flowers usually develop later than the pistillate flowers). For the placements of the different sexes along branches in the various species in Thailand, see for Breynia: van Welzen and Esser in van Welzen (2007), for Glochidion: van Welzen (2007) and for Phyllanthus s.l.: Chantaranothai (2007).
In the pistillate flowers, the upright, closely set sepals and the reduced stigmatic surface may serve two functions, to facilitate the pollen transfer and to preclude visits by insects other than leafflower moths. Two reversals (Breynia retusa and Glochidion sericeum) show that well accessible, completely developed and spreading stigmas no longer attract moths. A reduction in the style length and stigmas spreading was shown as an indicator of Epicephala pollination by Kawakita and Kato (2009: fig. 2). The reduction of the stigmatic tissue is achieved in two ways, either by shortening the stigmas or by folding the stigmas into a vertical (pyramidal) cone with a small, apical opening where the female moth can deposit the pollen. A reduction of the stigmas is seen in Breynia section Breynia. Kirganelia reticulata seems to follow both strategies: the stigmas are short and folded together. All other groups show cone-like united stigmas as in Glochidion. Several mothpollinated South American species of Phyllanthus may still have well-developed, spreading stigmas (Kawakita et al. 2019;see below). Additionally, the stigmatic papillae seem to disappear in the moth-pollinated species; compare The taxa included here have been shown by other authors to be pollinated and/ or parasitised by Epicehala. However, several taxa display similar morphological adaptations to those treated here and future studies might uncover new instances of plants in tribe Phyllantheae in a mutualistic relationship with Epicephala moths. The main pollination system in Nymphanthus Lour. has not yet been uncovered. Kawakita and Kato (2009) (Bouman et al. 2018). Other species of Emblica usually have well-developed spreading stigmas, but whether they are associated with Epicephala is not well known.
Adaptation to moth pollination seems to be comparatively more pronounced in the Palaeotropics (and Pacific islands) in what was formerly known as Breynia (B. subgen. Breynia section Breynia) and Glochidion (G. subgenus Glochidion) (Kawakita et al. 2019). It is often speculated that the obligate moth pollination contributed to a proliferation in species (Kato et al. 2003;Kawakita and Kato 2004a), especially in Glochidion (> 300 spp.) and New Caledonian Dendrophyllanthus (~ 160 spp.), which are speciose groups that have radiated relatively recently (see Kawakita and Kato (2009)). Breynia is somewhat less species rich, thus other factors than moth pollination likely contributed to the high speciation rate seen in these groups. The olfactory attractants may play a role (see above), but also the greater variation in flower shapes in Glochidion and Dendrophyllanthus.
In conclusion, these Palaeotropical groups show some general trends in morphological adaptation in flowers resulting from obligate moth pollination: -Sepals are usually stiff and upright and in a tight group around either the stamens or the ovary.
-Disc and disc glands are reduced.
-Stamens are united with the anthers upright along the androphore.
-Stigmatic surface is reduced, either by reducing the length of the stigmas to short stigmas or by uniting them into a cone with an apical round opening.
However, these trends are not always completely present, for example, mutualistic relationships were found in species with well-developed stigmas (see some discussions above and Table 2).

Neotropics
Based on the morphological trends found in the Palaeotropics, the situation in the Americas is evaluated here. Moth pollination in the Neotropics was recently reported by Kawakita et al. (2019). The moth Epicephala chancapiedra Kawakita & Kato, 2019, was found on three herbaceous species: Moeroris amara (Schumach. & Thonn.) R.W.Bouman, M. stipulata Raf. and Phyllanthus orbiculatus Rich. (Fig. 17). These three herbaceous species are pollinated by ants and as female E. chancapiedra does not have sensilla on the proboscis, the moths are (secondary) parasites, only laying eggs in the pistillate flowers without active pollination (Kawakita et al. 2019). The flowers of both sexes in these three species are small, hanging, quite open and the stigmas are generally short, but well-developed and spreading. Only the short stigmas might be an adaptation to moths, but it is more likely an adaptation to the ants or no pollination is needed as all pistillate flowers develop fruits (also in a hothouse without pollinators, but with limited seed production, pers. observ.).
In the genus Phyllanthus L. subgenus Ciccastrum (Müll.Arg.) R.W.Bouman (Fig. 1B) can act as a local outgroup for comparisons. This group consists of two species, P. purpusii Brandegee (Fig. 19) and P. riedelianus Müll.Arg. Both species have free stigmas that recurve over the ovary (P. purpusii) or that are bent towards each other in fruit (P. riedelianus). Only the staminate flowers of P. purpusii, each consisting of a narrow and tight cylinder of stiff upright sepals and three united stamens with the anthers apically along the androphore (like Breynia sect. Breynia), may point at moth pollination. The staminate flowers of P. riedelianus show no distinct adaptations to moth pollination (more like Breynia subg. Sauropus). However, P. riedelianus (not in the phylogeny of Bouman et al. (2021)) is doubtfully placed in this group and is strongly divergent from P. purpusii (Webster, 2002); see Torres et al. (2020). Kawakita et al. (2019) also looked for moth pollination in four woody species in the subgenera Conami (Aubl.) G.L.Webster and Xylophylla (L.) Pers. of Phyllanthus: P. acuminatus Vahl, P. graveolens Kunth, P. huallagensis Standl. ex Croizat and P. salviifolius Kunth (Fig. 18). In all cases, the stigmas do not show reduced surfaces and the stamens are often not upright (except more or less in P. graveolens) and are, thus, usually not readily accessible to moths. The pollination relationship does not appear to be obligate as other pollinators are present (gall midges and thrips; Kawakita et al. (2019)), which may mean that the relationship with the moths is rather recent and perhaps still developing and morphological adaptations are still lacking.   Arg.) Jean F.Brunel as a group with likely mutualistic pollination as there is some floral resemblance with Glochidion. Typical in the group is that the leaves often have a subapical extrafloral nectary abaxially, generally subapical above the mid-rib. In this group, as defined by Bouman et al. (2022), there are generally six sepals in two whorls; staminate flowers with six disc glands and three free, upright stamens; and pistillate flowers generally with a circular disc and a 3-locular ovary with the stigmas variously arranged. The situation is not completely clear as a large number of taxa could not be sampled. Not all taxa show visible morphological adaptations to moth pollination. Phyllanthus lediformis Jablonski, P. majus Steyerm., P. myrsinites Kunth (Fig. 20) and P. neblinae Jablonski have pistillate flowers with well-developed stigmas and thin sepals that generally are spreading and also have relatively large anthers and upright stamens. Phyllanthus duidae Gleason, P. pycnophyllus Müll.Arg. and P. vacciniifolius (Müll.Arg.) Müll.Arg. (Fig. 21) show an apical cone of united stigmas with three small lobes apically around a narrow opening, the typical situation as in Glochidion subgenus Glochidion; their staminate flowers have upright stamens and high stiff, upright sepals. Very likely, the species with united stigmas are moth-pollinated as the staminate and pistillate flowers show adaptations like those in other groups: limited stigma surface and narrow staminate flowers. The phylogeny of Bouman et al. (2021) only shows one representative for this group (P. vacciniifolius); thus, it is not clear if a single adaptation to moth pollination occurred in an ancestral species or multiple adaptations. Phyllanthus L. subgenus Xylophylla (L.) Pers. section Epistylium (Sw.) Griseb. was another taxon mentioned by Kawakita et al. (2019) as a Central American (West Indies) group that might have mutualistic relationships with Epicephala moths. Herbarium material was largely lacking; we saw only a single sheet of P. axillaris with young pistillate buds (Proctor 18349, Jamaica, U), which seem to have a closed stigma cone as in Glochidion subgen. Glochidion, which may be indicative of mutualistic pollination.
Thus, it appears that, in the Neotropics, the relationship with Epicephala moths probably is evolutionarily younger than in the Palaeotropics, because many plant species do not show strong adaptations to moth pollination and, as shown by Kawakita et al. (2019), several groups are likely not obligately moth-pollinated as other insect groups also pollinate the flowers. However, the pollination systems in the Neotropics are very understudied in comparison to those in the Palaeotropics.