Foraminiferid Architectural History; A review using the MinLOC and PI Methods

Geometrical models of unit volume are used to examine the effects of differing rates of chamber or test volume expansion, growth plan, chamber shape and apertural form upon internal-external lines of communication within foraminiferid tests. The main quantitative measure is the minimum line of communication (MinLOC) from the back of the proloculus to the nearest aperture in contact with the external milieu. The Parsimony Index (PI) is a qualitative measure, here used to illustrate some basic changes in foraminiferid architecture through time. Three general trends from longer to shorter MinLOC are indicated, particularly in shallow water tropical carbonate facies, with climaxes in the Devonian, the Carboniferous to Permian and the Cretaceous to Recent.


INTRODUCTION
The aim of this paper is to outline some broad scale evolutionary changes in foraminiferid architecture, through the use of two methods of analysis: tlhe MinLOC method (illustrated in detail by Brasier, 1982) and the Parsimony Index (given here for the first time). Together they suggest that internal-external lines of protoplasmic communication, o r related factors such as test shape, test compactness and osmoregulation, deserve more study as factors in foraminiferid ecology and evolution.
A general tendency for increased constructional strength, from feeble non-septate tubes to compact multilocular coils, was postulated long ago by Rhumbler (1.895) but the idea was contradicted by Galloway ( I 933), partly on the mistaken belief that foraminifera from the Malverns were Cambrian (Chapman, 1900;see Haynes, 1981, p. 72). Scepticism about adaptive morphology was also voiced in the works of D'Arcy Thompson (1942), Glaessner (194.5) and Smout (1954). But more recent biological and ecological research has brought forward the question of test function (e.g. Marszalek et al., 1969;Hottinger, 1978;Haynes, 198l), from which it seems that the test may serve to reduce biological, physical and chemical stresses. to enhance particular kinds of feeding and to regulate buoyancy. For example, Lipps (197.5) suggests that different test shapes are adaptations to different feeding methods. Erect tubular and branched forms fixed to hard surfaces or embedded in sediment may be suspension feeders, with their pseudopodia in the water column (e.g. Homotrema). Active lenticular forms living o n soft substrates and weed, o r elongate forms feeding passiveby near the sediment/water interface may be detrital scavengers (e.g. Elphidium, miliolids, nodosariids, buliminids). Trochoid o r flattened forms temporarily fixed, browsing on algae o r other substrates, are mostly herbivores, with pseudopodia streaming in all directions (e.g. Cibicides, Discorbis). Larger foraminifera tend to have high surface area t o volume ratio plus other features adaptive for their endosymbiotic algae (Haynes, 1981, p. 5.5); discoidal forms may be attached to plants (e.g. Marginopora) o r reclining o n the sea bed (e.g. Nummulites, orbitoids); fusiform tests may be passive rollers in more turbulent conditions (e.g. fusulinids, alveolinids). A review of test form and habitat in recent Caribbean carbonate facies supports these generalisations (Brasier,197.5).
Much of the emphasis to date has therefore fallen upon external constraints on test form, such as hydrodynamic properties, substrate type and food source. Without dismissing the significance of these, it is the intention of this paper to focus upon an internal constraint: the lines of communication within the test. That communication may be an important factor in test form and foraminiferid evolution is suggested simply by the fact that they are unicellular. This means there will be constraints upon maximum size and upon the degree t o which the protoplasm can be stretched out o r partitioned by the skeletal elements. Myers (1936) noted that a ratio exists between the volume of test protoplasm and the number (or surface area) of nuclei. Mobility of the nuclei may therefore be especially important where there is much protoplasm and this is suggested by the relatively short lines of communication in larger foraminifera (Hottinger, 1978). These lines of communication must also affect the movement of other organelles, food vacuoles, osmotic vacuoles, symbionts and chemical signals around the test. To these internal factors may be added the external ones of physical and chemical stress: long lines of communication in the form of tubular chambers may help to keep these stresses away from sheltering protoplasm (Marszalek et al., 1969). Hence there may be a correlation between lines of communication and ecology or palaeoecology of foraminifera, as has already been suggested by Chamney (1976).
The hypothesis that lines of communication are factors in foraminiferid ecology and evolution deserves more cytological and ecological research, but the wide architectural range o f fossil foraminifera provides ideal material for modelling lines of communication. This paradigm principle has been widely applied to macrofossils (e.g. Raup & Stanley, 1978) but less so in microfossils, though Berger (1969) has modelled the problems of ontogeny and buoyancy in planispiral planktonic foraminifera, and Scott (1974) has modelled the Orbulina lineage. Brasier (1980) modelled multilocular tests from the interaction between three variables during growth: the rate of growth translation (i.e. the net movement along the growth axis to the movement away from the growth axis), the rate of chamber volume expansion and the chamber shape. T h e aim of modelling here is to show that, for a fixed volume, lines o f communication may vary with growth plan, apertural position and chamber shape.

MODELS AND THE MinLOC METHOD
The models in Figs. 1 and 2 illustrate, theoretically, a range of foraminiferid tests of identical volume but with varying growth plan, chamber shape, apertural position or rate of test volume expansion. These allow measurement of the minimum line of communication (MinLOC) from the most remote point of the proloculus to the n e a rest ape r t u re in contact with the ex t e rn a I mi 1 i e u .
Other comparative mcasurements discussed by Brasier (1982), such as the maximum line o f communication (MaxLOC) from the most remote point of a distal chamber to its nearest aperture in contact with the cxtcrnal milieu, may be relevant to the number and placement of apertures but are n o t discussed in detail here.
The unit of MinLOC measurement in Figs. 1 and 2 is a n arbitrary one, taken as the diametcr of the proloculus. In each multilocular model the proloculus is assumed t o be spherical with a single aperture, followed by six chambers o f varying shape, but with a relatively high rate of chamber volume expansion E, where E = 2.9. This is calculated from the formula E = V, -, where V represents thc volume of the nfh or n -1 chamber. The chosen rate o f expansion, E = 2.9, compares with that of Horrnosina globuliferu figured by Loeblich & Tappan (1Y64,fig. 128.4a). This was chosen as a model because the uniserial growth and sphacroidal chambers allow easy calculation of test and chamber volume, while the relatively high rate of chamber volume expansion enhances the differences in MinLOC and MaxLOC found in the modified models. Fig. 1 shows how these parameters can be measured and compared. Models were first drawn to a large scale on metric graph paper; simple geometry and arithmetic allowed the plotting of most architectural forms sketched in Fig. 2. T h e various apertural positions were then plotted and the MinLOC and MaxLOC were measured directly from the diagram with an opisometer, taking a reading for each chamber. In trochospiral models several elevations were plotted and the MinLOC calculated by simple geometry. The glomospiral growth of Fig. 2 (7) was modelled in plasticene and then dissected; whorl portions were then measured and the MinLOC was cumulatively plotted from these. The overall test volume is standard in all the models, allowing comparison between the other parameters.
Below is a review of the MinLOC in models of tests with continuous o r septate growth, this being found in the majority of living and fossil foraminifera. I t is assumed here that the aperture is the primary means of communication with the extcrior and that mural pores, where present, are only used for osmoregulation, gas exchange and the passage of dissoved organic substances (Haynes, 1981, p. 51). Numbers in brackets refer to each of the models in Fig. 2.
Non-septate, continuous growth is shown in column 1 o f Fig. 2 and clearly demonstrates how a greater rate of test (or chamber) volume expansion can shorten the MinLOC. Considerable reduction in this parameter is also achieved by meandering ( 2 ) , planispiral ( 3 ) , helical (4), zig-zag (S), tight helical (6) and, ultimately, by glomospiral (7) growth, demonstrated here in tubular chambers in which the width W = I unit. Such nonseptate tests are seldom built with walls tapering at an apical angle of more than lO", hence their MinLOC is long relative to many septate tests. Thus one function o f the septum and aperture may therefore be to allow greater rates of volumetric expansion without exposing a larger area of protoplasm.
In septate tests o f identical volume, great variation in the MinLOC may occur through changes in chamber shape and in apertural position (Figs. I , 2) as well as through changes in the rate of chamber volume expansion. In uniserial tests, for example, flattening of the chambers towards the proloculus in the models (21), (25), (34) and (35) results in a great reduction in the MinLOC. Flattcning and arching of the chambers to a flabelliform test (24) may assist adherence or endosymbiosis but the addition o f many peripheral apertures (31) also greatly reduces the MinLOC.
Growth plan also modifies the MinLOC, as shown by comparing a simple uniserial test having sphaeroidal chambers and a terminal aperturc (2 1) with comparable planispiral (22), trochospiral(23) and biserial(26) forms (see Fig. 1). Modification of chamber shape and aperture can reduce the MinLOC, as for example in biserial tests with the change from terminal (26) to basal (27) to a long basal slit (29) aperture, and in planispiral or trocho-   Fig. 2 . Geometrical models of foraminifera with one, two o r seven chambers (including proloculus) and identical chamber volume and total volume. The chamber shape, growth plan and aperture have been varied to modulate the quantitative MinLOC parameter (vertical axis, 1-1000) and the qualitative Parsimony Index (horizontal axis, 1-10).  (43). The growth plans adopted by most larger foraminifera are relatively economical in their MinLOC values, as for example in discoidal annular (38), flaring annular (39), sphaeroidal annular (40), orbitoline conical (41) and fusiform planispiral (42) growth ; multiple apertures are necessary in most of these to avoid the MaxLOC parameter exceeding the MinLOC.
A number of multilocular test types have relatively high MinLOC values; e.g. periloculine (14), biloculine and spiroloculine (15, 16), triloculine (17) and quinqueloculine (1 8) streptospiral growth of miliolids. L,enticulinu, which is planispiral with a single, excentric aperture (cf. 20), also has a relatively high MinLOC value for its growth plan. Multiform tests that uncoil may also lengthen the MinLOC, though this is not necessarily the case since there may be a modification in chamber shape or apertures (Brasier, 1982, fig. 14)

PARSIMONY INDEX (PI) AND PI SPECTRUM
Studies of the ecological and evolutionary potential of this MinLOC spectrum should, ideally, begin with measurements of absolute test volume and MinLOC in living and fossil foraminiferid populations. But this is an immense labour and it is necessary to have a reconnaissance method. The Parsimony Index (PI) has therefore been devised to permit a rapid ranking of foraminiferid tests in rough proportion to their potenfiaf for shortened (parsimonious) MinLOC values, without reference to any absolute measurements (see Fig. 2, horizontal axis, columns 1-10). The PI of a foraminiferid test is the sum of separate point scores for growth plan, chamber shape and aperture; only one score can be achieved in each category. Multiform tests may have mean scores for each category. Foraminiferid populations can thus be separated into a PI spectrum ranging from 0 (unilocular spheres) to 10 (multilocular complex and others) as shown in Figs.

PI MAXIMUM AND ARCHITECTURAL EV 0 L UT I 0 N
A general trend towards shortened MinLOC through the Palaeozoic has been inferred from the MinLOC models by Brasier (1982) but the PI method now allows a more graphic plot of foraminifera with the highest PI from each of eight suborders through time (Fig. 3). The classification adopted here follows Hohenegger & Piller (1975, 1977 in distinguishing hyaline Lagenina,    (Amsden et al., 1980); type Couvinian of Belgium (Bultynk, 1970); Frasnian of North America (Toomey, 1972); Famennian to basal Carboniferous of West Germany (Eickhoff, 1979) Jenkins & Murray (1981). The vertical scale is 0-20 species, except in the smaller samples where it is 0-10 species.

Spirillinina and Involutinina from hyaline Rotaliina ;
it also follows Poyarkov (1979) in recognition of the distinct early history of the calcareous Semitextularia, Nanicella and relatives (here called Semitextulariina). This classification is believed to be more phylogenetic than that of Loeblich & Tappan (1964) or Haynes (1981) and interprets wall composition and ultrastructure as having evolved gradually through various similar grades in divergent stocks. It also places stronger emphasis on test architecture, particularly the change from non-septate tubes or coils to septate testsa "Rubicon" that seems to have been crossed separately in each of the suborders shown in Figure 3, except for the Rotaliina. These may owe their septation to the Fusulinina, via Tetrataxidae, Duostominacea and Robertinacea (e.g. Loeblich & Tappan, 1974) or, less favoured, to the spirillinacean Patellina (e.g. Haynes, 1981). In general, it can be seen from Fig. 2 that each presumed lineage has advanced from tubular forms with low PI to coiled forms with chamberlets or open umbilici and a high PI. But the rate of change and times of climax have not always coincided in these lineages.
The Semitextulariina rapidly reached a high PI in mid Devonian tropical carbonates but were eclipsed in the end Frasnian extinctions (e. g. Toomey & Mamet, 1978;Poyarkov, 1979). The Fusulinina developed rapidly in the late Devonian to early Carboniferous, of which the lineages leading to Bradyina (with open sutural apertures) and to the Fusulinacea (with a spindle shape and mural pores) appear to be classic examples of MinLOC shortening trends, connected perhaps with their increasing size. Hottinger (1978, fig. 11) has illustrated how the numerous short chambers and row of frontal foramina stretching from pole to pole may together give short lines of communication between tne first and last shell compartments of fusiform tests. Tl;e iterative evolutionary trends to extend the biomass along the growth axis may also be seen as MinLOC shortening trends, since the length of the spiral canal is thereby shortened for a unit volume of protoplasm. These tropical cabonate foraminifera with high PI suffered extinction through the late Permian. Involutinina with low PI were common through the late Palaeozoic and Triassic and may be considered to have reached a climax in the Norian with the involute, septate, chamberlet-bearing Triasina (e.g. Piller, 1978). This reefal form did not survive the end of the Rhaetian, though simpler forms remained into the Jurassic and Cretaceous (ibid. ) Rotaliine ancestors such as Tetrataxis appear to have survived the late Permian crisis and, through progressive changes in perforation, wall ultrastructure and composition, may have led ultimately to the canaliculate Rotaliacea and other larger foraminifera in the late Cretaceous and Cainozoic (e.g. Tappan, 1976). The canal system of the Rotaliacea may be a device for maintaining mobility of the test when the protoplasm has been withdrawn from the later chambers.
They are able to do this because the canals connect the protoplasm of the inner chambers with the ambient environment (Hottinger, 1978). The spiral canals in the marginal cord of nummulites and the radiating stolons of orbiotoids were also features that allowed for shorter MinLOC in larger discoidal foraminifera.
A late Cretaceous to Recent climax is also found in the imperforate Miliolina and less spectacularly in the perforate Spirillinina (Patellina). Extinctions of complex larger foraminifera in the late Cretaceous embraced the Textulariina, Rotaliina and Miliolina but these were not a great set back because Palaeocene tropical carbonates may contain complex larger forminifera with high PI, such as Discocyclina and Glomalveolina.
Unlike the above groups, the fossil record suggests that the perforate Lagenina did not develop from coiled ancestors but from rectilinear ones (e.g. Hohenegger & Piller, 1975;Tappan, 1976) which, combined with their conservative retention of a single, often excentric aperture, may have discouraged the development of many new forms with high PI. Pachyphloia and Colaniella of the late Permian may belong to this lineage, however, and some polymorphinids and glan$ulinids of Triassic to Recent times have comparatively high PI values.

PI SPECTRUM THROUGH TIME
The summary presented above is a great simplification, however, because it overlooks fluctuations that have taken place in the PI spectrum (and by implication, in the MinLOC) through time, as shown for example in Fig.4. This has been compiled from the major foraminiferidbearing stages of the British Isles or adjacent regions and it must be emphasised that it does not represent population studies or complete faunal lists but simply the number of cited taxa in each P I class, from the sources mentioned. But it does illustrate the pattern of architectural change in the British area, with three rather unequal cycles from low to high P I dominance: Wenlock to Frasnian ; Famennian to Upper Namurian ; Zechstein to Quaternary. And the high PI climaxes in the Couvinian to Frasnian, Carboniferous, and Albian to Recent noted in Fig.3 are even more clearly illustrated. Some of the main foraminiferid changes are noted below, with PI subscores for growth plan, chamber shape and aperture.
The Carboniferous PI spectra are distinctly bimodal in Fig. 3, with numerous Archaediscidae (001) and Endothyru sp. (332). A shift towards higher PI after the Chadian is related to the appearance of Eostafella (333), Tetrataxis (333), Bradyina (334), Janichewskina (334) and Valvulinella (433). Dwindling faunas in the Namurian may reflect the more brackish arid regressive conditions, and Westphalian foraminifera are only simple ammodiscaceans (e.g. Calver, 1969). But the carbonate platforms of North America and Tethyan regions continued to support diverse assemblages in which the Courceyan to Brigantian pattern of Fig. 3 was sustained and developed (e.g. Toomey & Winland, 1973) culminating in the fusulinacean assemblages of Permian times.
While these PI spectra are suggestive of evolutionary change, they are an incomplete record since the data chosen may ignore some low PI taxa of little biostratigraphic interest ; independant study of selected assemblages and discussions with foraminiferid workers spccialising in different parts of the column confirm, however, that the pattern is quite representative. The PI spectra illustrate, at least, the value of the PI method for preliminary investigations and the potential of the MinLOC method for more accurate evolutionary research.

DISCUSSION
Although the emphasis has been placed on trends from longer to shorter MinLOC, there are certainly some counter-trends and some anomalies.
Firstly there is the question of survival in many ancient genera with relatively long MinLOC, such as Bathysiphon, Ammodiscus and Nodosaria. It might be thought that their longevity argues against any major functional significance in the MinLOC, though there are several reasons to refute this. These survivals may be seen as the counterpart to the more rapid evolution and extinction of larger foraminifera, planktonics and others with shortened MinLOC; i.e. evolutionary rate, survival and lines of communication may be related in some way. They may also owe their survival to a shift from shallow to deeper or marginal marine habitats, where longer lines of communication are not disadvantageous. It is noteworthy that many of the forms evolving a short MinLOC have lived in shallow marine conditions, but any relationship between habitat and MinLOC, particularly of habitat shift in archaic forms, requires much more research.
A second problem concerns the interpretation of supposed MinLOC lengthening trends, such as the "uncoiling" of Lenticulina (Barnard, 1960). Various interpretations have been put on these uncoiling trends, ranging from adaptation to deposit feeding (e.g. Haynes, 1981), evolution in stable marine conditions (e. g. Chamney, 1976) to non functional explanations (e.g. Hofker, 1954). Related to this is the problem of dimorphism, in which the gamont and schizont generations of a species may have quite different architecture, and hence have MinLOC values which are different. Both uncoiling and dimorphism require more study, but the MinLOC method may give these trends new significance since they should throw further light on the adaptive and non-adaptive features of test architecture.
There is also the problem of the Saccamminidae (Cambrian to Recent) and other non-septate, non tubular rhizopods. These do not show a clear MinLOC shortening trend because they had relatively short internal-external lines of communication from the start, though the Saccamminidae declined in importance after the Silurian. Their displacement by foraminifera with septate growth is not hard to explain, since these gained the advantages of size increase, tests reinforced by spiral walls and septa, and novel architectural forms with a wider adaptive range. Even so, the Saccamminidae evolved from simple forms with n o aperture, to forms with multiple apertures, stellate arms and labyrinthic walls, through the course of the early Palaeozoic (Brasier, 1982).
In summary, the MinLOC method provides another way of looking at foraminifera, from the inside. The MinLOC and P I methods can be used to study patterns of architecture through time (evolution) or in space (ecology) but these are likely to be the outcome of both internal factors (such as MinLOC and MaxLOC) and external factors (such as osmotic stress, substrate type, hydrodynamic properties and food source). The effects of these external factors could also be studied by modelling. But interpretation of the architectural patterns is a complex matter that will also involve further biological and ecological work. Toomey, D.F. 1972. Distribution  1053-1 074.