Structural Monophyly Analysis Allows Estimation of Self-Sustainability at the Supraspecific Level over 88 Million Years in Mosses

Abstract

A new method of macroevolutionary analysis—high-resolution phylogenetics, integrating both morphological and molecular traits—has revealed well-supported evidence of complexity-based processes generating and controlling biodiversity. A novel technique of using evolutionary rates following a strict morphological clock, at least approximately, may allow detailed information on speciation and extinction events across geologic time. Branching series of minimally monophyletic genera are used to characterize in detail the branching lineage of the widely distributed moss family Streptotrichaceae. A strict morphological clock is calibrated by timing of genera new to recently exposed islands, molecular scaling against fossil taxa, and fossil evidence of the origin of the modern bryoflora. The numbers of genera generated in each 22-million-year interval are similar, while only one genus is inferred as extinct. The general outline of the phylogeny is tadpole-shaped because cumulative extinction is less than cumulative speciation, thus sustaining the family over vast time spans. Extant species per genus increase significantly over time, not through the proliferation of secondary descendants (i.e., more than four species per ancestral node), but through the selective preservation of lineages via extinction. Ancient traits are preserved throughout the lineage. It is hypothesized that descendant species are protected from coeval competition through bursts of speciation. This study supports a complexity-based explanation of the interaction of major evolutionary processes resulting in sustainability.

1. Introduction

High-resolution phylogenetics is a new field using multi-disciplinary methods possibly unfamiliar to working taxonomists with an interest in evolution. It has been previously named macroevolutionary systematics, but with the added ability to identify the same ancestral taxa using both morphological and molecular data [1], the origin in numerical taxonomy is suitably recognized. Several published papers and books [2,3,4,5,6,7] have explained and provided examples for this set of powerful techniques of analysis of relationships between taxa. Because high-resolution phylogenetic analysis uses taxonomically important morphological features (a term including all expressed traits), direct inferences of macroevolutionary processes—as trait changes—are possible regarding potentially adaptive character states. In addition, the high Bayesian posterior probability support for inferred evolutionary relationships allows the results of molecular phylogenetics to be treated as primary data serving as a molecular tree onto which morphological traits are mapped.

The basic unit of evolution is considered to be the minimally monophyletic genus, here referred to as a microgenus. It is basically one ancestral species and a few descendants. The microgenus is identified in two ways: (1) it is the most similar of the ingroup species to an outgroup taxon, and (2) it is least highly modified among ingroup species, i.e., most generalist in the genus. Concatenating microgenera results in an evolutionary diagram of sequential monophyly, a caulogram of structured monophyly. Series of minimally monophyletic genera, deprecated by cladists as paraphyly, actually reflect a genuine process in nature—descent with modification—in this case operating above the level of species. A mesogenus is a standard result of evolutional taxonomy, typically comprising one or a few species, occasionally a moderate number, and rarely rather large. A macrogenus is a genus that is already moderate to large in species number but becomes further expanded with addition of related genera, synonymized because they are embedded in (descended from) another genus, according to the cladistic classification principle of holophyly.

For context, standard genera in the taxonomic literature commonly include more than one ancestral species, and relationships are thus necessarily general. These often large classical, heterogeneous groups are referred to here as mesogenera. For example, the average number of species per genus among vascular plants is 23, and 15 in the Asteraceae [8]. Average species per genus for several selected angiosperm families ranges from 27 to 6 [9]. With animals, perhaps more intensive study has revealed fewer species per genus: in 31 British insect orders and suborders, the average species per genus ranges between 1.3 and 7.3 [10]. If the world is split into small and large islands and portions of continental areas [11], the average number of species per genus in floras ranges from 1.9 to 4.4, 3.0 to 5.4, and 2.2 to 8.9, respectively [12], indicating that speciation is affected, within limits noted, by local competition, and that much speciation is allopatric.

In molecular phylogenetics, groups are commonly made maximally monophyletic by lumping several mesogenera together under the cladistic principle of holophyly, a nomenclatural rule that reflects no empirically determined process in nature. Such groups are here termed macrogenera and are unwieldy, maximally monophyletic units for detailed analysis of evolutionary relationships. Particularly strong examples of recently minted highly heterogeneous macrogenera in the moss family of this author’s expertise, Pottiaceae, include Chionoloma s.lat. [1,13,14] and an expansive treatment of Syntrichia [15]. Given that cladistics operates by estimation of common ancestry through cluster analysis on a dichotomous dendrogram using synapomorphies, the ancestor–descendant patterns required for modeling evolution by structured monophyly are not, and a priori cannot be, evident in macrogenera. The only true evidence of descent with modification (ancestral signature) is that of exemplars of the same species occurring at different nodes, reflecting mutations accumulated by morphologically static ancestral species between the generation of multiple descendant species. This is paraphyly in cladistic studies, and it is usually absent or not recognized as evolutionarily informative.

The microgenus, the basic building block of structural monophyly, consists of optimally one ancestral species and four descendant species. This was determined in a cumulative study of 30 genera of Pottiaceae that revealed 4 as the most common number of immediate descendants. NK analysis using a random Boolean network model [5,16,17] demonstrated that the constant appearance of the most recent traits of the ancestral species in all the descendant species gave the microgenus a fractal nature [18,19] of fractal dimension 1.16 (equivalent to a Pareto distribution of 1:4) that broadened the range of resilience to competition.

The traits of any one species may be divided into three groups that are apparently effective in evolutionary processes. The novon is the set of traits (character states) newly evolved in the descendant species. These traits may be modified during generation of secondary descendants (descendants of descendants). The ancestron is the set of the remaining traits. This is subdivided into two groups: (1) The immediate ancestron is the set of traits identical to the new traits evolved in the ancestral species and is empirically determined as optimally four. The immediate ancestron is the same in each of the immediate descendants. (2) The reserve ancestron is the set of ancient traits that is the source of new traits for the novons of the descendant species. The immediate ancestron is not used as a source of new traits. Another way to look at this is that the immediate ancestron is possibly responsible for stabilizing selection, and the novon versus the reserve ancestron are active in balancing selection and sustainability of the family. Given that these phenomena are found in 36 genera studied, it is hypothesized that they are widespread among mosses and likely present throughout the global flora.

A caulogram is an evolutionary tree with named ancestors at the nodes, i.e., a stem-taxon tree. Because there are optimally four descendant species, a caulogram is not dichotomous as in a cladogram but may have one or more branches from a node. Character state changes, habitats, and geographic range distributions may be mapped omto a caulogram as has been done for the taxa involved in the present study (species of Streptotrichaceae worldwide and Neotrichostomum and related genera in the West Indies) [5,9], allowing for direct analysis of evolutionary changes. Nomenclatural authorities for the taxa named in this study are given in the aforementioned papers and are also available online at the Missouri Botanical Garden’s taxonomic data center, Tropicos (https://tropicos.org/home). As the present paper focuses on evolution, nomenclatural authorities are not included herein.

2. Materials and Methods

The basic method of macroevolutionary systematics using morphological traits and a stem-taxon tree has been detailed at length in the papers by the author listed in the Introduction. The evolution of the Streptotrichaceae [3] is the focus of this study. The time intervals for evolution of the lineage are inferred from calibration with West Indian endemic genera of Pottiaceae [3,7] of 22 mya (Figure 1) and from a molecular study of Syntrichia (Pottiaceae) [15] for separation of the worldwide family Streptotrichaceae [3] from Pottiaceae. In the present paper, and commonly in the field, “mya” means million years ago. Details of the calibration are provided in a separate paper, but in short, two genera, mostly endemic to the West Indies, are considered to have been generated around the middle of the maximum period of emergence of the West Indies as a floristic unit, about 45 mya, as estimated by Rodriguez and Strong [20] and Rodriquez-Silva and Schlupp [21]. The estimate of separate generation of these two genera is 22 mya for each, with considerable margin of error. This value is, however, bracketed by the estimate that the modern bryoflora first appeared during the mid-Cretaceous [22,23,24], about 80–100 mya, although originating much earlier in the Permian [25]. The caulogram is four ancestral genera deep, and dividing the existence of the modern bryoflora into four intervals of 22-million-year intervals aligns well across time. In addition, the Jauregui-Lazo et al.’s [15] study suggests that the Pottiaceae separated from the Streptotrichaceae (as two species of Leptodontium) about 120 mya, perhaps deep within a shared Ur group.

The intervals are as follows: 2–22 mya: Holocene, Pleistocene, Pliocene, and Miocene (ice caps, strong cooling); 22–44 mya: Oligocene, mid-Eocene (warm but cooling, no ice caps); 44–66 mya: mid-Eocene, Paleocene (Eocene–Paleocene Thermal Maximum); 66–88 mya: last portion of the Late Cretaceous, Maastrichtian through Coniacian epochs (global temperatures nearly those of Eocene–Paleocene Thermal Maximum); and 88–100 mya: early portion of the Late Cretaceous, Turonian, and Cenomanian epochs (global temperatures warm across a broad latitudinal range).

As previously established [3,4,5], four descendants per ancestral species is assumed to be optimal. Only one complete lineage of several microgenera, that of Streptotrichaceae (Bryophyta) [3], has yet been published. The lineage is here reproduced (Figure 2) annotated with the number of estimated extinct species. There is one internal branch with a leap of 11 traits between Streptotrichum ramicola (66–88 mya) and Williamsilla araucarieti (44–66 mya). This supports interpolation of an extinct genus of optimally five species, one ancestor and four descendants, a genus here assigned to the more distant time interval of 66–88 mya to allow more time to go extinct, thus joining the single extant genus, Steptotrichum, of that time interval.

There are three “unknown” ancestral species inferred for the genera Leptodontium, Microleptodontium, and Trachyodontium. These are needed as shared ancestors to account for somewhat disparate descendants and are scored as extinct species. The interpolation of an extinct genus and three unknown ancestral species is quite like the familiar technique of phylogenetic bracketing for elucidation of soft tissues in the study of fossils [26]. Terminal genera with less than four descendant species are assumed to be undergoing extinctive reduction in numbers, not being simply young and unfinished in speciation. This is because the two or three extant descendant species are more probably evidence of a past burst of speciation, rather than an indication that we, at this one moment in geological time, are in the midst of such a burst.

The number of extinct species in each genus is simply the optimal number four minus the number of extant species that are immediate descendants of the inferred ancestral species. Species that are descended from immediate descendants, by secondary ancestry, are considered potential seeds of new genera, and are generally allopatric. The ancestor in a genus derived from the ancestor of another genus is scored as an immediate descendant of that other genus.

A species extinction is implied when a presumed ancestral species is absent. If too great a number of traits is found between two genera, then an entire extinct genus of five species is interpolated.

3. Results

A caulogram (Figure 1) of the “Weissia Probe” of Zander [5] emphasizes that two sub-lineages, the genera Chionoloma and Tainoa, follow the rule of one ancestor with optimally four descendant species as established by Zander [5] based on 30 different genera of Pottiaceae. There is only one apparently extinct species among these two small mostly endemic genera. The three species estimated as extinct in the Hymenostomum sub-lineage are doubtless an artifact, as that group is more speciose and of worldwide distribution. Self-sustainability of Chionoloma and Tainoa is strong and estimated at 22 million years.

Evaluation of past extinction rates must reflect the fact that geologic time affects apprehension of geologic space. Over 22 my, for example, the geographic region of distribution of a species may be large, scattered, and variable, and the terms sympatric and allopatric are less distinct.

The results of the analysis of extinction in Streptotrichaceae are given in

Table 1. Comparison of extinction per 22 my intervals for extant taxa in the Streptotrichaceae over the past 88 my. Not included in this table is an Ur group of this family, assigned to the 88–100 mya interval, which gave rise to two main generic lineages but about which little else is known.

Time Interval, Past to Present	66–88 mya (Yellow)	44–66 (Blue)	22–44 mya (Green)	2–22 mya (Red)
(1) Cumulative total extant genera	1	4	7	10
(2) Cumulative extant species	1	7	15	27
(3) Cumulative extant species per genus	1	1.75	2.15	2.7
(4) Total genera for that interval	2 (incl. 1 interp.)	3	3	3
(5) Total extant species for that interval	1	6	8	12
(6) Extant species per genus	0.5	2	2.67	4
(7) Species extinct in each interval	6 (incl. 5 interpol.)	9	8	8
(8) Cumulative extinct species	6	15	23	31
(9) Immediate descendant species	3 (all inerpol.)	3	5	5
(10) Immediate descendants per genus	0.5	1	1.67	1.67
(11) Secondary descendant species	0	1	2	7
(12) Secondary descendant per immediate descendant species	0	0.33	0.4	1.4

, which evaluates the annotated caulogram (Figure 2). This study is based on a world monograph [3] and includes all known species. There are 10 extant genera and 27 extant species. The family lineage extends back to the Late Cretaceous, with roots much earlier, as implied by the sub-basal connection of the somewhat anomalous genus Trachyodontium.

Interpretation of the data in Table 1 details the speciation and extinction rates across geologic time in a large family of bryophytes, representing the first study to do so. The cumulative values span the full time frame, and the rates are interpretative summaries of the main evolutionary trends.

3.1. Cumulative Total Extant Genera, Past to Present

The 10 extant genera have increased by about 3 genera per interval. Only one genus is assumed to have gone extinct, although there are three ancestral species (labeled “unknown”) inferred as probably extinct, leaving descendants with new traits considered advanced.

3.2. Cumulative Extant Species, Past to Present

The 27 extant species accumulate by doubling in each of the last three intervals.

3.3. Cumulative Extant Species/Extant Genera

Extant species per genus increase with recency.

3.4. Total Genera for That Interval, Including One Interpolation

Total genera, including an implied genus, are the about same for the last three intervals. If the implied extinct genus of five species is included, then the total number of genera generated per interval is about the same for each interval. The rate of genus generation is about three genera per 22 my intervals, and the only evidence the extinction of an entire genus is one in the 66–88 mya interval, as interpolated in Figure 2.

3.5. Total Extant Species for That Interval, Not Including Unknown GENERA

Total extant species increase by two for each of the last three intervals.

3.6. Extant Species/Genera

Extant species per genus increase significantly over time through extinction, not as an increased number over the optimal four descendants per ancestor, i.e., not as secondary descendant species.

3.7. Extinct Species per Interval

This is calculated as the optimal four minus the number of actual descendant species for each genus, including unknown genera, and extinction of numbers of species is about the same during each interval.

3.8. Cumulative Extinct Species

The total estimated extinct species is 31, compared to 27 extant species. These balance in number, but there are more ancient extinct species than there are recent extinct species, and more recent extant species than ancient extant species.

3.9. Immediate Descendants for That Interval, Including Interpolations

Total immediate descendant species are about the same in each interval.

3.10. Immediate Descendants/Genera for That Interval

Immediate descendant species per genus are about the same in each interval.

3.11. Secondary Descendants for That Interval

Total secondary descendant species are most prevalent in the most recent interval. Secondary descendants are scored from species in that genus; the ancestor of a different genus is not considered a secondary descendant. Secondary descendants are here assumed to be the potential beginnings of new genera.

3.12. Secondary Descendants/Immediate Descendant Species for That Interval

Secondary descendant species are more prevalent compared to immediate descendants in the most recent interval. This implies that secondary descendants go extinct more rapidly than immediate descendant species. This may be due to loss of critical traits (partial overwriting of immediate ancestron) or simply pressures of unusual allopatric habitats, or both.

4. Discussion

Why is the number of four descendants e considered optimal, as suggested by NK evaluation with random Boolean network analysis [5,16,17]? Four seems to be a “magic number”, even in inorganic chemistry, where many compounds have a multiple of four atoms in their basic unit structure. This has been termed the rule of four, and there is apparently no convincing explanation beyond a correlation with loose packing arrangements maximizing free volume [27]. The appearance of an optimal four descendant species, and four traits per species in evolution, is apparently an emergent process in complexity theory. Perhaps this number is optimum on account of the selective value of punctuated equilibrium? This has been discussed for molecular trees [28,29], where a sequence of molecular lineages leads only to several exemplars of that species, showing molecular differentiation but no evolutionarily important phenotypic differentiation.

Punctuated equilibrium in the usual sense of a burst of fossil evolution [30,31,32] can be invoked because increasing time between speciation events reduces available habitat space logarithmically by competition with other immediate descendants. Of course, at geologic time scales, coeval descendants may be generated over the course of a million years, accumulating DNA mutations, yet still have the same immediate ancestrons. Short branches on a phylogram imply a burst of descendant origination followed by stasis. If the four descendants were generated all at once (“once” being within a short geological time span like a million years), then each would address 25% of available sympatric habitat. On the other hand, if not at once, but generated serially, and each distant in time, then if the first occupied all the habit, there would be no space for the remaining three.

Consider the case of four descendants from a shared ancestral species. If generation occurred at the time the first descendant occupied 1/2 the habitat space, then the second —assuming only two descendants—would occupy one-quarter of the total space (as the first receives 1/2, and the two then split the remaining 1/2). Following this logic, each successive descendant receives half of the remaining space. The third would receive 1/8 (with the first receiving 1/2, the second 1/4, and the third 1/8), and the fourth only 1/16 This represents a worst-case scenario for serial origination. If all four descendants were generated in closer succession—but not all at once—then the fourth descendant would receive somewhere between 25% and 6.25% of the available habitat. This decline in available space follows a logarithmic pattern, decreasing by powers of two.

In the case of five descendants, the range is between 20% and 3.25% (1/32). But this assumes that the ancestor is absent, which it seldom is. In fact, it lingers longest in the tail of a lineage. Thus, optimal sharing of sympatric space for descendant species is when each species has about the same amount of habitat. The fewer the descendants, the more habitat each species has, while a short time between speciation events avoids logarithmic reduction in available habitat for new species. But the fewer the descendants, the greater the chance that the genus will go extinct.

The rate of extinction, if high, trims species, particularly secondary descendants, from genera over time. Optimum lineage-level selection for survival resulting in long-term sustainability is when a lineage balances the rate of descendant survival with the rate of extinction over geological time such that the lineage retains ancient genetic traits in remnants of ancient genera. It comes down to a high degree of vagility and a low rate of extirpation, operating under natural selection at species, genus, and ecosystem levels. Protecting the availability of ancient traits enhances the evolvability of the whole lineage that supports survival during serial repetitions of similar large-scale or global environmental perturbations.

In other words, natural selection occurs at the lineage level affecting ecosystems through time. Examples of ancient trait combinations contributing to lineage survival are the generation of robust genera from antique genera—such as Crassileptodontium from Trachyodontium (its unknown ancestor), Williamsiella from Streptotrichum ramicola, and Microleptodontium from Leptodontiella apiculata.

Lovejoy and Spiridonov [33] evaluated the balance of relative effects on macroevolution biodiversity of the environment (with climate as proxy) and of intrinsic life processes (extinction and speciation) over geologic time. They found that 34 my was a critical crossover time with “Red Queen” life processes operating at shorter intervals and “Court Jester” environmental processes at longer times. These crossover times occurred across 1 to 400 my. Their stochastic model emphasizes a probabilistic expression of changes in biodiversity at geological time levels. Rampino and Caldeira [34] reviewed several reports of environmental oscillations with cycles of around 30 my, including sea-level changes, episodes of volcanism, climate and biological extinctions, occurring over hundreds of millions of years. They suggest a shared modulation in the earth’s mantle as a potential cause. Substituting 30 my for the presently estimated 22 my interval for genus formation would push the origination of the Streptotrichaceae back from 88 mya to 120 mya, still within the envelope of calibration.

François [35] introduced an N-epoch stochastic model of species generation from an ancestor that is like a Yule birth–death model but with unequal diversification rates backward in time such that each ancestral species leaves more descendant species in recent time periods than in ancient time periods. This produces a hollow curve that fits actual data better, with reptiles as examples. Most taxa apparently follow a species-per-genus hollow curve [36]. According to François, Reptilia consists of about 10,885 species in 1,196 genera, averaging 9.1 species per genus. The hollow curve of frequencies of numbers of species per genus bends most sharply at five species per genus, both with standard modeling and more strongly using the N-epoch model. Thus, although the average number of species per genus is about nine, most genera have one to five species and the fewer genera of greater numbers of species level off as an asymptote. The import of the present study is that the complexity-based evolutionary rule of four descendants per ancestor may have a major probabilistic component [37] that works well with punctuated equilibrium to stabilize the lineage through geologic time. Nottale et al. [38] have shown that evolutionary events mapped to evolutionary trees also follow a power law resulting in a hollow curve, and that logarithmic evolutionary behavior is self-similar across scales of organization.

That this is a process occurring at large scales agrees with the study of Plotnick and Sepkoski [39], which found that the Self-Organized Criticality of chaos theory, which derives large changes from small beginnings, may not be operative. The result of natural selection does not then necessarily depend on cascades of fitness but on feedback at the ecosystem level. Instead, patterns of self-similarity appear to change at periods of about 25 my as a periodic imposition on a random background. Sepkoski’s suggestion is curiously similar to the present model of 22 my intervals for genus generation. Complexity theory, in this case, operates while chaos theory does not or does so in shorter time periods. For the estimated 100 my of the existence of the Streptotrichaceae, biodiversity is apparently governed by Red Queen effects although there may have been two or three major environmental cross-overs imposing changes from environmental perturbations.

5. Conclusions

The main findings are summarized (Figure 3) as a tadpole-shaped “space-time ship Earth” or at least a large lifeboat of such. It has been suggested that there are no biological laws relevant to evolutionary systematics as reliable as there are in physics, except possibly that of Dollo parsimony [40] (p. 101). The present paper presents evidence that the Streptotrichaceae lineage travels through time and space by very gradually losing ancient genera and species and building ordered sets of new genera and species following complexity-based rules, these powered by natural selection [5,41,42,43]. In this case, these minimal rules seem to be (1) an optimum of four immediate descendants, (2) the projection of all the most recently evolved traits of a genus’s immediate ancestor into its immediate descendants (developing a shared novon), (3) descendant species in a genus tending to appear all at the same time (across a geological time scale), and (4) ancient genera both persisting and generating new genera. The resulting lineage maintains an optimal ability to regenerate across time, is, by its very existence, resilient to a series of recurring environmental perturbations, and contributing to differential survival of ecosystems.

One might speculate that if complexity-based evolutionary processes are common to all lineages, then those lineages of ancient taxa existing prior to the modern bryoflora throughout the Mesozoic may perhaps concatenate in the same tadpole-like phylogenetic conformation. Evidence of the existence of a tadpole lineage penultimate to Pottiaceae and Streptotrichaceae is the appearance of the apparently ancient genera Luisierella and Timmiella at the very base of the molecular [44] tree somewhat below the insertion of the Pottiaceae although both are clearly pottiaceous in character but either much reduced (the former) or much elaborated morphologically (the latter). Timmiella is also at the base of the morphological cladogram [45], but Luiserella is not, probably because of the extreme reduction in informative morphological traits. Other pre-middle Cretaceous intermediate and possibly more informative taxa in this lineage ancestral to the Pottiaceae, Streptotrichaceae, and perhaps other families, may be assumed extinct.

Survival of a robust lineage will theoretically contribute to the entropic balance (Brooks and Wiley) [46] of the inhabited ecosystem—in the present case of the wide-ranging Streptotrichaceae, the páramos and largely montane forests, soils, and lavas of the tropical and sub-tropical ecosphere [3]. Raup [47] suggested that new species simply fill ecological voids left by extinct species. Many species in the family share the same habitats, and although the genera are morphologically disparate, the idea cannot be dismissed. He further supported the view that extinctions are point events and not a time-continuous process, but the tadpole shape of the Streptotrichaceae (Figure 3) suggests otherwise, excepting the catastrophe of 65-million years ago. Wilson [48] (p. 6) cited evidence that in many groups species diversity apparently remains much the same throughout geological time. The tapering shape of the lineage in the present study suggests a gradual die-off of this moist-tropical and montane-temperate group of taxa perhaps largely because of gradual global drying during the Miocene.

The above detailed implementation of a strict morphological evolutionary clock presented here, to my knowledge, has not been conducted before, yet it integrates coherently into a causal whole, embodying a form of uniformitarianism. There are distinctive similarities between intervals that imply overarching processes. The evolutionary rule of four (e.g., an optimal four descendants and four traits in the novon) applies to each interval. The 27 extant species accumulate by doubling in each interval, and total new genera are about three per interval. New extant species increase by two during the last three intervals. The number of extinct and extant species nearly balance across the ages. The number of immediate descendants is about the same for each interval. Even with the absence of data on extinct species, it is clear that the groups discussed here are massively self-sustaining. The underlying processes may be the same for all lineages of bryophytes and would support the use of a strict morphological tree. It would be helpful, of course, if in the future one could find geographic regions isolated for intervals longer than those of the West Indies and provide additional direct calibrations for the evolutionary timeline of morpho-genera.