The rise of bilaterians

Abstract

Recently Shen et al. have argued that the Ediacaran faunas from Avalon-Charnwood [580–560 million years ago (MA)], the White Sea-Flinders Range (560–550 MA) and Namibia (550–543 MA) occupied the same morphospace even though these faunas differed in species composition, ecology, biogeography and age. The traits they used to characterise these faunas could not distinguish between important promorphological features such as radial vs. bilaterian and unitary vs. colonial animals. Their inappropriate assignment of symmetry properties led to the homogenisation of morphospace in these different faunas. Another way to examine shifts in morphospace involves sorting out radial vs. bilaterian and unitary vs. colonial genera in terms of their time of first appearance in the fossil record. While genera with both kinds of symmetry properties and unitary and colonial animals were present during the early Ediacaran, there was a large proportional increase in new bilaterian genera and a decrease in colonial genera beginning between 560–550 MA. The increase in bilaterians may have been associated with the ability to exploit food sources in and on the substrate of shallow water environments.

Keywords:

• Ediacara
• unitary and colonial animals
• radial and bilateral body symmetry

Introduction

Ediacaran animal fossils have a wide distribution and vary in age from about 620 million years ago (MA) to the beginning of the Cambrian (543 MA). There are four major assemblages: the Yangtze Gorge (620–543 MA), Avalon-Charnwood (580–560 MA), the White Sea-Flinders Range (560–550 MA), and the Nama (550–543 MA) in addition to a number of smaller assemblages ((Fedonkin et al. 2007a). In a recent study, Shen et al. (2008) have argued that three of these assemblages, the Avalon-Charnwood, the White Sea-Flinders, and the Nama occupied similar morphospaces even though they differed to a large extent in species composition, ecological setting, biogeography and age. This claim is questionable in view of the fact that there was only one putative bilaterian in the Avalon-Charnwood fauna and a significant increase of new bilaterians starting after 560 million years (Narbonne 2005). If a morphospace analysis cannot distinguish between motile bilaterian taxa and tethered colonial taxa it is not doing its job.

This study re-examines the criteria Shen et al. (2008) used to collect and categorise their data in order to expose the pitfalls inherent in the information used for their morphospace analysis. It also documents the appearance of bilaterian genera and the decline in colonial genera during the Ediacaran to provide a quantitative assessment of the changes in relative generic numbers in these subgroups but also in their morphological disparity during this period. These data indicates that there was a shift in morphospace during the Ediacaran associate starting between 550–560 MA.

Methods

Time of first appearance of Ediacaran genera and their geological setting

A number of procedures have been used to infer time of first appearance of Ediacaran genera. One or more formations at a given site frequently have an absolute date. The most trustworthy dates are based on U–Pb zircon dating associated with volcanism; however, some of these dates have a large standard error. A few dates are based on Pb–Pb isotope ratios; these measurements are inherently less accurate and generally have a larger standard error than U–Pb dates. If a fossil is in a formation that is bracketed by two relatively close dates, one can be fairly confident of its date. A fossil can be in a formation that is above or below a dated formation; in this case, the date is less certain. In cases where there is not a good date for a site but where elements of the fauna are similar to a dated site, the dated site has sometimes been used to assign a date to the site where there is a paucity of information. In some cases, the fossil site may lie immediately under a formation with characteristic Cambrian fossils. Because the base of the Cambrian is dated, these Ediacaran fossils can be dated with a fair degree of certainty. Dating of fossils in Table 1 is done in most cases within 10 million year intervals. Figure 1 gives the geographical sites and formations, where the fossils listed in Table 1 first appeared and indicates the inferred age of these formations. If a date is fairly certain, it is given superscript ‘a’; if it is less certain, it is given superscript ‘b’ and if it is an educated guess, it is given superscript ‘c’ in Table 1.

Table 1. Time of first appearance, and type of Ediacaran genus.

						Morphology
						Bilat.		Radial
Age	Location	Formation	Genus	Reference	Figure	Un	Co	Un	Co	Size	Hab.	#
600–615^c	S. China	Doushantuo l. phosphorite	? Planolites trace fossil	Tang et al. (2006)	4a,c	c	–			W × L 0.5 × 8 mm	M^a	1
			Sponge	Li et al. (1998)	1			–	a	D (150–750 μm)	S^b	35
580–590^b	S. China	Doushantuo m. phosphorite	Vernanimalcula	Chen et al. (2004a)	1	a	–			L 124–178 μm	M^b	10
			? Cruziana trace fossil	Tang et al. (2006)	4b	c	–			W × L 4 × 64 mm	M^a	1
			Embryos with polar lobes	Chen et al. (2006)	2	b				egg D 0.2–1.2 mm		248
			Hydrozoan	Chen et al. (2002)	11			–	a	L × W 1 × 0.2 mm	S^b	1
			Ramitubus increscens	Chen et al. (2002)	6			–	b	L × W 2 × 0.1– 0.2 mm	S^a	1
				Liu et al. (2008)								100
			Sinocyclocyclicus	Liu et al. (2008)	pl. 3	?	–	?	–	L × W 1 × 0.2 mm	S^b	200
			Quadratitubus	Liu et al. (2008)	pl. 5	?	–	?	–	D 0.2 mm	S^b	100
			Crassitubus	Liu et al. (2008)	pl. 6	?	–	?	–	D 0.18 mm	S^b	100
570–580^a	Newfoundland, CA	Drook	Thectardis	Clapham et al. (2004)	2			a	–	L 26–165 mm	S^a	205
										W 24–96 mm
			Trepassia	Narbonne et al. (2009)	10, 1–7	–	?	–	?	L × W 2 × 0.1 mm	S^a	5**
		Briscal	Fractofusus andersoni	Gehling and Narbonne (2007)	12	–	?	–	?	L × W 1.0 × 0.5 mm	S^b	22
			Beothukis	Narbonne et al.	6, 1–7	–	?	–	?	L × W 2.5 × 0.6 mm	S^a	11
			Charnia masoni	Narbonne et al. (2009)	11, 1–2	–	?	–	?	L × W 2 × 1.3 mm	S^0	1
560–570	White Sea Russia	Lamtsa	Epibaion	Ivantsov and Malakhovskaya (2002)	2, 3	a	–			L × W 44–24 cm	M^a	ca. 100
			Inaria	Gehling (1988) and Grazhdankin (2000)	6			b	–	D 21–106 mm	S^a	57
	Newfoundland, CA	Mistaken Pt.	Charniodiscus spinosus	LaFlamme et al. (2004)	4 1–4	–	?	–	?	L × W 1.4 × 2.8 mm	S^a	29
			Bradgaita	Narbonne (2004) and Hofmann et al. (2008)	3E 19	–	?	–	?	L × W 4.2 × 0.7 mm	S^a	5
			Primocandelabrum	Hofmann et al. (2008)	12	–	?	–	?	L × W 5.4 × 1.5 mm	S^a	5
			Hadryniscala	Hofmann et al. (2008)	22–1	–	?	–	?	L 1.5—2.5 mm	S^a	3
										W 1–2.5 mm
			Pectinifrons	Bamforth et al. 2008	5, 6	–	?	–	?	L × W 3.8 × 1.3 mm	S^a	102
		Trepassey	Rangeomorph plumose	Narbonne (2004)	2 c,d	–	?	–	?	L × W 1.4 × 0.3 mm	S^a	1
			Avalofractus	Narbonne et al. (2009)	3,1–6	–	?	–	?	L × W 1.5 × × 0.5 mm	S^a	9
550–560^a	White Sea Russia	Verkhovka	Kimberella	Fedonkin et al. (2007c)	1	a	–			L 2–3–150 mm	M^a	800
			Parvancornia minchami	Glaessner (1980) and Ivantsov et al. (2004)	Pl. 1	a	–			L 8.4–27 mm	M^a	60
			Dickinsonia costata	Gehling et al. (2005)	7	a	–			L 0.42–13.6 cm	M^a	8
			Andiva	Fedonkin (2002)	1–3	a	–			L 59–100 mm	M^a	4
			Yorgia	Ivanstov (1999) and Dzik and Ivantsov (1999)	1	a	–			L 9.8–215 mm	M^a	14
										W 10.3–200 mm
			Cyanorus	Ivantsov (2004)	Pl. 1, 1–6	a	–			L 3.3–11 mm	M^b	12
										W 1.9–6 mm
			Vendia rachiata	Ivantsov (2004)	Pl. 1, 10–13	a	–			L 4.5–12.5 mm	M^b	6
										W 3.8–8.5 mm
			Karakhtia	Ivantsov et al. (2004)	4	a	–			L 8–108 mm	M^b	5
			Onega	Ivantsov (2007) and Fedonkin (1990b)	Pl.1, 7–8	a	–			L 5.4–5.9 mm	M^b	8
										W 3.8–4.3 mm
			Archaeaspius	Ivanstov (2001)	Pl. 1 6–7	a	–			L 4–19.5 mm	M^b	8
										W 4.2–23.5 mm
	England	Charnwood Bradgate	Pseudovendia	Boynton and Ford (1979)	1	b	–			L × W ″25 × 20 mm	M^b	1
	S. Australia	Rawnsley Quartzite	Spriggina ovata	Glaessner and Wade (1966)	Pl. 98, 4	a	–			L × W 22 × 8.5 mm	M^b	4
	S. China	Doushantuo Miaohe	Bilaterian	Xiao et al. (2002)	11	a	–			L × W 25 × 1 mm	M^a	1
	White Sea Russia	Verkhovka	Archaeonassa trace fossil	Jensen (2003)	3A	b	–			W 4.5 mm	M^a	10
	S. Australia	Rawnsley Quartzite	Helminthoidichnites, trace fossil	Glaessner (1969) and Jensen et al. (2006)	5B	b	–			W 3–4 mm	M^a	5
			Spirorhaphe trace fossil	Runnegar, 1992	3.9	b	–			W 2–3 mm	M^a	1
			Aulichnites trace fossil	Jenkins (1995)	Pl. 2 L	b	–			W 4 mm	M^a	1
	White Sea Russia	Verkhovka	Solza	Ivanstov et al. (2004)	Pl. 1,1–4	a	–			L 7.2–10.4 mm	M^b	6
										W 5.3–8 mm
			Temnoxa	Ivantsov et al. (2004)	Pl. 1, 1–2	a	–			L 6.4 − 8.6 mm W 3.7–4.6 mm	M^b	2
			Lossiina	Ivantsov (2007)	Pl. 2, 1–3	a	–			L 3.2– 8.4 mm	M^b	3
										W 1.8–3.7 mm
			Armillifera	Fedonkin (1990b)	Pl. 7, 3	a	–			D 8 mm	M^b	1
	S. Australia	Rawnsley Quartzite	Praecambridium	Glaessner and Wade (1966)	Pl.102, 4	a	–			L × W ??5 × 4 mm	M^b	7
	White Sea Russia	Verkhovka	Veprina	Fedonkin (1990b)	Pl. 4, 6			b	–	D 6 cm	P^b	1
	S. Australia	Rawnsley Quartzite	Eoporpita	Wade (1972)	Pl. 40, 1–6			a	–	D 2–8 cm	P^b	20
			Rugoconites enigmaticus	Wade (1972)	Pl. 43, 3–4			a	–	D 16–64 mm	P^b	26
			Palaeophragmodictya	Gehling and Rigby (1996)	6			a	–	D 1.5 10 cm	S	48
	White Sea Russia	Verkhovka	Bonata	Fedonkin (1990b)	Pl. 8, 4			a	–	D 40 mm	S	4
			Vendoconularia	Ivanstov and Fedonkin (2002)	Pl. 1			a	–	W 54 mm max.	S^b	1
	Newfoundland, CA	Femeuse	Triforillonia	Gehling et al. (2000)	17			a	–		S^c	10
	S. Australia	Rawsley Quartzite	Tribrachidium	Glaessner and Wade (1966)	Pl. 101, 5			a	–	W 9–400 mm	S^c	40
			Conomedusites	Glaessner and Wade (1966)	Pl. 99, 3–4			a	–	D 21–22 mm	S^c	3
			Arkarua	Gehling (1987)	4–5			a	–	D 3.5–10.2 mm	S^c	57
			Skinnera	Wade (1969)	Pl. 69, 8–12			a	–	D 3.9–32 mm	S^b	27
	White Sea Russia	Verkhovka	Albumares	Fedonkin (1990b)	Pl. 9, 1, 4			a	–	D 13 mm	S^b	3
			Anfesta	Fedonkin (1990b)	Pl. 9, 2			a	–	D 18 mm	S^b	2
	S. China	Doushantuo Miaohe	Calyptrina	Xiao et al. (2002)	2.3–6	?	–	?	–	D 3–10 mm	S^b	10
			Protoconites	Xiao et al. (2002)	7.9–7.1	?	–	?	–	D 1–4 mm at aperture	S^b	11
	White Sea Russia	Verkhovka	Ventogyrus	Fedonkin and Ivanstov (2007b)	3–7	–	?	–	?	D 5–7 mm	P^b	25
										W 3–4 mm
	S. Australia	Rawsley Quartzite	Funisia	Droser and Gehling (2008)	1	–	?	–	?	L up to 30 cm	S^a	1000
										W 2–12 mm
			Gehlingia	Gehling (1988) and Fedonkin et al. (2007a)	7	–	?	–	?	L × W 2.1 × 0.7 mm	S^b	1
	White Sea Russia	Verkhovka	Ramellina	Fedonkin (1990b)	Pl 14, 1	–	?	–	?	L 4.5–6 mm	S^b	2
										W 1–1.5 mm
			Swartpuntia	Narbonne et al. (1997)	6, 9, 10	–	?	–	?	W 2–3 mm	S^a	10
	S. Australia	Rawsley Quartzite	Pambikalbae	Jenkins and Nedin (2007)	3–6	–	?	–	?		S^a	5
			Pteridinium	Jenkins (1992) and Narbonne et al. (1997)	5	–	?	–	?	L × W 12.8 × 3.6 mm	S^a	10
			Phyllozoon	Jenkins and Gehling (1978)	7	–	?	–	?	W 2–3 mm	S^a	15
543–550^a	White Sea Russia	Yorga	Paravendia	Ivantsov (2001) and Ivantsov (2004)	Pl. 1, 2, 3	a	–			L 13.5–36 mm	M^a	5
										W 7–18 mm
			Ivovicia	Ivantsov (2007)	Pl. 1, 3, 4	a	–			L 13–19 mm	M^a	2
										W 9–13 mm
			Platypholinia	Fedonkin (1990b)	Pl. 19, 5, 6	a	–			L 6 cm max	M^b	2
										W 2.8 cm max
			Vladimissa	Fedonkin (1990b)	Pl. 19, 7	a	–			L × W 4.5 × 3.2 cm	M^b	1
			Tamga	Ivanstov (2007)	Pl. 2, 5–7	a	–			L 3.8–5.4 mm	M^b	5
										W 2.7–3.8 mm
	Aus area Namibia	Dabis	Protechiurus	Glaessner (1979)	1	b	–			L × W 7.4 × 1.9 cm	M^b	1
		Urusis	Archaeichnium	Glaessner (1977)	1–2	a	–			L 35 mm max	M^a	73
										W 4.5 mm max
	N.W. CA	Blueflower	Windermeria	Narbonne (1994)	3, 2–4	b	–			L × W 1.6 × 0.8 cm	M^b	1
			Palaeophycus trace fossil	Narbonne and Aitken (1990)	Pl. 3, 6–8	b	–			W 3.7–7.1 mm	M^a	34
			Torrowangea trace fossil	Narbonne and Aitken (1990)	7, A–C	b	–			W 0.8–2.7 mm	M^a	160
			Helminthopsis irregularis trace fossil	Narbonne and Aitken (1990)	PL. 4, 5.6	b	–			W 1.4–3.1 mm	M^a	ca. 100
	Bhutan	Deschiling	Gordia trace fossil	Tangri et al. (2003)	3	b	–			W 1, 7–4.0	M^a	ca. 10
	Nama Namibia	Urusis	Treptichnus trace fossil	Jensen et al. (2000)	2	b	–			W × L 2 × 10 mm	M^a	ca. 7
		Dabis	Namacalathus	Grotzinger et al. (2000)	9			a	–	cup 2.5 cm stem 3 cm	S^a	105
			Ernietta	Pflug (1972)	Pl. 34, 5, 7, 8			b	–	W × L 62 × 29 mm	S^a	12
	White Sea Russia	Yorga	Nemiana	Leonov (2007)	10			a	–	D 18 mm max	S^b	60
			Vaveliksia	Ivanstov et al. (2004)	Pl. 2, 3–10			b	–	L 35–86 mm	S^b	5
										W 8–19 mm
	Uruguay	Yerbal	Soldadotubulus	Gaucher and Sprechmann (1999)	Pl. 9, 3–6	?	–	?	–	W × L 50 × 400 μm	S^b	10
	Great Basin US	Lower Woods	Onuphionella	Hagadorn and Waggoner (2000)	5:7	?	–	?	–	W × L 6 × 35 mm	S^b	1
	S. China	Dengying	Sinotubulites	Hua et al. (2000)	Pl. 2, 6	?	–	?	–	W 1.3 mm	S^b	2
			Goojiashania	Hua et al. (2000)	Pl. 1, 4b, 7b	?	–	?	–	W 4 mm	S^b	2
			Ningqiangella	Hua et al. (2000)	Pl. 1, 4b, 7b	?	–	?	–	W 5.5 mm	S^b	2
			Conotubus	Hua et al. (2000)	Pl 1, 7a	?	–	?	–	W 4 mm Aperture	S^b	1
			Chenella	Hua et al. (2000)	Pl. 2, 11	?	–	?	–	W 0.5 mm	S^b	1
			Paracharnia	Sun (1986)	01-Feb	–	?	–	?	W ca 1 mm	S^b	2
	Nama Namibia	Dabis	Rangea schneiderhoehni	Jenkins (1985)	2	–	?	–	?	L × W 3 × 1 mm	S^b	23
		Urusias	Nasepia	Germs (1973)	2 A–H	–	?	–	?	W 0.1–1 mm	S^c	6
	S. Brazil	Tamango	Corumbella	Babcock et al. (2005)	4, 5, 7	–	?	–	?	W 2.4 mm	S^a	11
	Uruguay	Yerbal	Waltheria	Gaucher and Sprechmann (1999)	Pl. 9	–	?	–	?	W 120 μm	S^b	57
	S. China	Dengying	Cloudina	Hua et al. (2005)	1	–	?	–	?	W 2–3 mm	S^b	>500
	Nama Namibia	Zaris	Namapoikia	Wood et al. (2002)	1			–	a	W 1.5–5 mm	S^a	>500
Notes: Un, unitary; Co, colonial; Hab, habitat; M, motile; S, sessile; P, pelagic; #, number of cases; ?, cannot be certain if animal is radian or a bilaterian W, width; L, length and D, diameter.

Figure 1 Time line showing the ages of the sites where Ediacaran fossils were collected and where the age of the deposit has been established by measuring the products of radioactive decay. An ‘ × ’ marks the U–Pb zircon dates and a ‘0’ marks the Pb–Pb dates. The superscipt identifies the authors who established the dates: (1) (Condon et al. (2005), (2) Zhang et al. (2005), (3) Barford et al. (2002), (4) Chen et al. (2004), (5) Bemus (1988), (6) Bowring et al. (2003a), (7) Bowring et al. (2003b), (8) Compston et al. (2002), (9) Martin et al. (2000), (10) Grazhdankin (2004), (11) Preiss (2000) and (12) Grotzinger (1995). A solid line indicates that the date for a given formation or a member in the formation is relatively secure; while a dashed line indicates that the dating is not secure.

The Ediacaran fauna in the Yangtze Gorge is found in two formations. The upper Dengying Formation extends from the base of the Cambrian to about 551 MA (Condon et al. 2005; Zhang et al. 2005). The Doushantuo Formation extends from the base of the Dengying formation to about 635 MA. The Marinoan glaciation, which was a global event, was coming to an end at the base of the Doushantuo Formation (Condon et al. 2005; Zhang et al. 2005). All of the fossiliferous sites in both formations appear to have been shallow marine basins. Fossils found primarily in the lower part of the Dengying Formation are referred to as the Gaojiashan biota; a number of these fossils have thin mineralised shells. There are three fossil containing phosphorite strata within the Doushantuo Formation. At these sites, the soft parts of organisms have been phosphatised. Preservation can be superb allowing one to see tissues and cellular detail; this is the Ediacaran Lagerstätten. Anywhere between 1 and 3 phosphorite containing strata are found in all of the extended sections of the Doushantuo Formation that have been examined. The upper Phosophorite unit is frequently referred to as the Miaohe biota and lies just below the Dengying Formation. Its upper boundary is well defined, however, its lower boundary has not been determined; it may extend to 560 MA. At Weng'an the Doushantuo Formation can be about 40 m thick (Dornbos et al. 2006). In some sections, the region just above the Middle Phosophorite unit has been dated at 576 MA and the region just below this unit has been dated at 599 MA (Barford et al. 2002; Chen et al. 2004); the dating of this part of the formation is poorly constrained. The upper part of the Basal Phosphorite unit terminates some time before the Middle Phosphorite unit and begins after 621 MA. The dating of this unit is even less well constrained than the middle unit. Another set of fossiliferous sections associated with Sandouping is 188 m thick (Tang et al. 2006); none of its fossiliferous units have been dated, however all three phosphoirite containing strata are present.

The Avalon fauna occurs in strata on the east coast of Newfoundland and an almost identical fauna is found in Charnwood Forest in the English midlands. The Ediacaran is truncated at the basal end of the Drook Formation by the Gaskiers glaciation (ca. 580 MA), which appears to have been a regional event. Four dates constrain the ages of these sites, including the Drook (Bowring et al. 2003a, 2003b), Mistaken Point (Bemus 1988) and Bradgate Formations (Compston et al. 2002). The top ends of the Bradgate and Femeusse Formations are not constrained. It has been assumed that there is a partial chronological overlap between one or more formations in the Avalon fauna and the Charnwood fauna. Ecologically, both of these faunas existed in relatively deep water below the photic zone on the continental shelf (Narbonne 2005). All of the fossils at these sites are casts.

The Russian White Sea and Australian Flinders Range faunas are also quite similar; both faunas lived in shallow near-shore environments; all of the fossils at these sites are casts. There are two dates that constrain the age of the Verkhovka Formation; at the same time they define the top of the Lamtsa and the base of the Zimnegory Formations (Martin et al. 2000; Grazhdankin 2004) at the White Sea sites. The date of 556 MA defines the base of the Rawnsley Quartzite in the Flinders Range (Preiss 2000); however, this date has a large standard deviation associated with it. The top of the Rawnsley Quartzite and the Yorga Formations are not constrained. The fact that there is a significant overlap of the White Sea and Flinders Range faunas suggests that they were largely contemporaneous.

The Nama fauna of Namibia also lived in a shallow water near shore environment; the fossils at this site are either casts or composed of shell. It is constrained by three dates (Grotzinger et al. 1995). The upper end of the Urusis Formation, which abuts the lower Cambrian is dated at 543 and 545 MA and the base of the Nudaus and boundary of the Zaris Formation is dated at 548.8 MA. The bottom of the Dabis, which is the oldest formation, is not constrained by a date. At both the Yangtze Gorge and Nama sites there is a shelly index fossil, Cloudina that is only found in the strata just below the Cambrian boundary (Geyer and Uchman 1995). At Nama this genus extends from the top of the Urusis to the Dabis Formation (Vickers-Rich 2007).

There are additional sites where dates are not available, but the formation lies just below the Cambrian and Cloudina is present. These include the Blueflower Formation in the Canadian northwest, the Lower Woods Formation in the great basin of the US, the Tamango Formation of Brazil, the Yerbal Formation of Uruguay and the Deschiling Formation of Bhutan.

Frequently, a given Ediacaran genus is found at more than one stratigraphic level at a given site or at multiple sites. An atlas of Precambrian metazoans in (Fedonkin et al. 2007a) was used in Table 1 to establish the earliest geographic site and formation of a given genus. The earliest record for a given genus does not always provide the best description of the genus, as a consequence, the reference for the genus is frequently based on material that dates from a later horizon.

Morphological characterisation of Ediacaran genera

Fossils were characterised as either unitary individuals or colonies composed of multiple individuals (zooids). Unitary individuals or zooids that make up a colony can be either radially or bilaterally symmetrical. A radially symmetrical individual has only one axis of symmetry which is frequently referred to as its anterior–posterior axis. Bilaterally, symmetrical individuals have an anterior–posterior axis and a plane that bisects the dorsal and ventral region that extend along the anterior–posterior axis; this plane divides the organism into left and right halves along its anterior–posterior axis. Extant radially symmetrical animals are composed of two cell layers; an outer epithelium and a gastrodermis. Extant bilaterians are composed of an outer epidermis, a gastrodermis and an intermediate tissues layer between them, the mesoderm that forms the muscles and the excretory system. The distinction between radial and bilateral symmetry is robust. During embryogenesis in extant bilaterians, the anterior–posterior axis is always specified prior to the specification of the dorsal–ventral axis (Slack 1991).

Three classes of fossils make up the Ediacaran fauna: body fossils, trace fossils and embryos. Body fossils provide the most information on the morphology of a given organism. Trace fossils of trails made by animals moving through a substrate or feeding have frequently been used to assign that fossil to the Bilateria because it takes muscle, a mesodermal derivative, operating on a hydrostatic skeleton to make a track. In a number of cases, the body fossil that was responsible for making a given trace is present with its trace. However, there are circumstances involving living animals where traces have been generated by an organism that was not a bilaterian (Jensen et al. 2005; Bengston and Rasmussen 2009). Two additional problems are associated with trace fossils: (1) short morphologically simple traces may be artifacts and (2) because trace fossils have a relatively simple morphology it is easy to confuse one ichnogenus for another. Jensen et al. (2006) have compiled and assessed the published records on Ediacaran trace fossils. Almost all of the trace fossil citations in Table 1 are based on their records. There are plausible claims for cleavage stages, gastrula stages and larval fossils (Li et al.1998; Chen et al. 2000; Chen et al. 2006). However, in most cases these stages do not contain enough information to indicate whether the developing animal would be radially or bilaterally symmetrical. Because a fossil of a ’gastrula’ is essentially a snapshot of an individual at a given stage of development, that particular fossil may only have two germ layers while at slightly later stage of the development that species may have three germ layers. This would change the diagnosis of the embryo from a radial to a bilaterian form. Usually, asexual reproduction does not begin until after embryogenesis is over; therefore it cannot be diagnosed on the basis of an embryonic stage.

A given fossil was only used in Table 1, if it could be dated with some degree of confidence, had characters that could be used to distinguish unitary and colonial forms and a statement could be made about its symmetry properties.

For each genus a reference to a figure showing a typical fossil and description of the genus is given. If there was more than one species in the genus, the species used was named. The reference was used to obtain an approximate measure of the size of an average member of a genus. When no measurements were given in the reference, the length and width or diameter of the fossil from the figure was measured where possible. This is only an approximate measure of the size of a typical member of a genus since different species within a genus can differ in size. Within a species there are also size differences within a population; in a number of cases these size differences represent different growth stages. In colonial organisms this measurement was made on the size of the largest zooids in the colony. The habitat of each genus: sessile, motile or pelagic is also given and the approximate number of specimens of that genus that have been examined. A number of genera are only known from one or a small number of specimens; one has less confidence in a generic description based on a small specimen number. In the categories Morphology and Habitat in Table 1, ‘a’ indicates that the judgment is supported by the fossil evidence, ‘b’ indicates a judgment that is only partially supported by the evidence and ‘c’ indicates a judgment that is an educated guess.

Results

Critique of the work of Shen et al. (2008)

The data set for the Shen et al. (2008) paper was published as supporting online material. Fifty morphological characters were used to quantify the symmetry properties and morphology of species from three Ediacaran faunas. The characters used (# in parenthesis) delimited overall shape (6), symmetry (7), structure of the central region (23), structure of the intermediate region (7), structure of the peripheral region (3) and other structures (4). The two most important character sets are overall shape and symmetry because they are used to characterise every fossil.

There is a formal language zoologists use to characterise the symmetry properties of animals and there is simple descriptive language. Shen et al. (2008) have opted for simple descriptive language. For example, the symmetry term ‘radial’ means going out from a centre, while ‘bilateral’ indicates extending out on each side of a median line. For zoologists an extant multicellular animal is either radially symmetrical or it is a bilaterian (Beklemishev 1969). These categories also accommodate the Ediacaran fauna (Fedonkin 1990a). A radially symmetric animal has an anterior–posterior axis; each of these axial landmarks differs in morphology and the animal is symmetrical along the axis separating the two end points. Bilaterally, symmetrical animals have an anterior–posterior axis in addition to dorsal and ventral sides that differ and extend along this axis. The plane that defines the dorsal–ventral sides also defines the left and right sides of the animal. When descriptive language is used, the term bilateral refers to an organism that has left-right symmetry but does not take into account dorsal–ventral differences. One problem involved in using the informal language of the authors to describe fossils is that the description is not commensurate with a statement about whether or not a fossil is radially symmetrical or a bilaterian. The morphological characters necessary to make a distinction between a radially symmetrical animal and a bilaterian are not included in the authors data matrix and they do not discuss these morphological categories.

Another problem in Shen et al. (2008) is based on their treatment of unitary individuals and colonies composed of zooids that were generated by asexual reproduction. All of the individuals in a given colony can be either radially or bilaterally symmetrical, however, in almost all Ediacaran colonies, while one can locate where the zooids were, in most cases it is not possible to specify their symmetry properties because of their small size. The symmetry properties of a colonial species are defined by the symmetry properties of its zooids. While the shape of the colony as a whole can be described as having a given kind of symmetry, these symmetry properties may not be the same as the symmetry properties of the zooids that make up the colony. Shen et al. (2008) do not make a distinction between colonies and unitary individuals even though colonies make up a significant part of the Ediacaran biota (Fedonkin 1990b). The symmetry properties of the whole colony are treated as if they were equivalent to the symmetry properties of a unitary individual. Since many colonial species have a bilateral colony morphology, they are lumped with disparate unitary bilaterians that also exhibit bilateral symmetry. This effectively makes their morphospaces similar.

In addition to these major methodological errors, there are categories of information that the authors have purposely excluded that would help to establish differences between radial and bilaterian taxa. Most bilaterians are capable of locomotion, while locomotion on substrates in radially symmetrical animals either does not occur or is restricted. There are a number of cases where a trail is coupled with a fossil animal. This kind of information can be used to define the front end of the animal in cases where both ends appear to be the same. There are also a number of structural differences that emerge when both the dorsal and ventral sides of a given fossil are examined and internal structures can become visible in compressed fossils that are potentially useful for establishing dorsal–ventral and anterior–posterior polarity differences. Different growth stages of a fossil genus can be used in order to make inferences about the generation of structural features from a growth zone.

These excluded categories of information can change the morphospace diagnosis of a fossil. One example involves the bilaterian Dickinsonia (Figure 2A). According to Shen et al. (2008), this genus is defined by the following character states: overall shape – oval, symmetry – bilateral and bipolar, and structures in the central region – central groove plus primary branches (segments) curved toward both ends. As Dickinsonia moved it could leave resting traces on its substrate. The set of resting traces that follow a body fossil are oriented in the same way as the body fossil allowing one to infer that the body fossil moved with only one bilateral end in front (Gehling et al. 2005). As Dickinsonia grew it added superficial segmental structures to its external epithelium. Segments are only added to one end of the animal (Runnegar 1992). This end is opposite the front end of the animal during locomotion. Studies of compressed specimens have revealed structures interpreted as an anterior ventral mouth-pharynx that connects to an alimentary canal (the central groove of Shen et al. 2008) and paired caecae that extend anterior laterally from the alimentary canal (Jenkins 1992; Dzik and Ivantsov 2002). Some of these fossils show characteristic wrinkles on part of their surface, indicating muscle contractions, muscle is indicative of bilaterian status (Runnegar 1982). This additional information shows that the animal had an anterior–posterior axis and that there are dorsal–ventral differences along this axis; in other words it was a bilaterian (Figure 2B). It is important to note that Shen et al. (2008) would have not been able to infer that this genus was a bilaterian because they do not use a character set for defining dorsal–ventral differences.

Figure 2 A) Drawing of Dickinsonia showing the bipolar features that were used by (Shen et al. (2008). (B) Drawing of Dickinsonia showing that it has an anterior (a)–posterior (p)-axis, and dorsal and ventral sides based on inferences about the polarity of locomotion, the pattern of growth and the anatomy of internal and ventral organs.

In a number of cases, separated parts of an Ediacaran body fossil have been assigned to different taxa. This was a frequent occurrence for sessile colonial forms where the holdfast that tethers the colony to the substrate was preserved while the frond-like structures with the zooids was lost. After this condition was recognised (Gehling et al. 2000) a number of disc shaped fossils were identified as holdfasts (see atlas of Precambrian fossils in Fedonkin et al. 2007a). Shen et al. (2008) have included these holdfasts in their morphospace analysis. There are 32 cases that were used in Shen's analysis of the Avalon fauna, 31% of these cases are holdfasts; there are 195 cases in their White Sea-Flinders Range fauna, 36% of these are holdfasts; there are 32 cases in the Nama fauna, 32% of them are holdfasts. Since holdfast morphospace is rather uniform it dampens the impact of the morphospace values of intact fossils.

Time of first appearance of Ediacaran genera

One can get a picture of the changes in the Ediacaran fauna through time and changes in morphospace by noting when members of this fauna first appear in fossil record and the promorphological nature of these genera (Table 1).

600–615 MA: There are two fossils from the lowest fossiliferous unit of the Doushantuo formation. One fossil is a crown group poriferan. Thin sections of these fossils show monaxial spicules with associated sclerocytes. There is an epidermis with porocytes and an internal spongocoel. There is suggestive evidence that these animals were engaged in asexual reproduction. These fossils were found in association with embryos and a possible parenchymella type sponge larva. Recent work on the distribution of a sponge biomarker in Ediacaran and Cryogenian strata has provided evidence for the appearance of sponges just prior to the Marinoan glaciation at ca. 635 MA (Love et al. 2009).

The other fossil resembles the ichnogenus Planolites. This type of trace is found on sediment surfaces. Its round cross-section suggests peristaltic locomotion which is associated with a bilaterian body plan. Given the fact that there is only one representative of this ichnogenera it is not a compelling case for an early Ediacaran bilaterian.

580–590 MA: This time interval contains a crown group cnidarian body fossil that resembles part of a modern thecate hydrozoan polyp colony (Chen et al. 2002). Two zooids are connected by a stolon. Underneath the perisarc there appears to be an epithelial layer that surrounds a gastric region. There are also embryos with two germ layers associated with the body fossils that resemble some hydrozoan gastrulae, however, these gastrulae do not have enough diagnostic features to give them a phylum identity. Chen et al. have also described the fossil Ramitubus (which Chen et al. refer to as Sinocylocyalicus). This tubular fossil, which can reproduce asexually, is composed of a series of compartments enclosed in an extracellular matrix. Only the terminal compartment fed. In one case, two tentacles emerge from the top compartment of a sectioned fossil. The skeleton of this fossil has the same organisation of a tabulate coral, however, the evidence that it represents a crown group Anthozoan cnidarian is not as compelling as it is in the case of the hydrozoan because the body of the fossil was not sectioned even though it can be seen in outline. Rambitubus has also been found in younger strata in the Doushantuo Formation that comprise the Miaohe biota (Xiao et al. 2002). Another genus, Funsia (Droser and Gehling 2008) from the Ediacara of Australia appears to be an enlarged version of Rambitubus.

There are cases involving a body fossil, a trace fossil, and embryos during this time interval that indicate the presence of bilaterians. The most convincing case involves the body fossil Vernanimalcula (Chen et al. 2004a). A population of these small fossils has been examined in section. They have three germ layers, an anterior mouth and a through gut. They have clear-cut anterior–posterior a dorsal–ventral axes and anterior dorsal structures that may have had a sensory function. Bengston and Budd (2004) have argued that this fossil genus is an artifact on the basis of both taphonomic and diagenetic alterations. Chen et al. (2004b) replied that they have a number of cases that show the complex morphology that they report and as a consequence the possibility that one is dealing with an artifact is slim.

The trace fossil ?Cruziana has parallel edges and faint transverse ridges at the bottom of a shallow longitudinal furrow. Only one case of this ichnogenus has been found. It resembles Cruziana, a trace fossil, typical of the early Cambrian, that is thought to have been generated by an arthropod (Jensen 2003).

One class of cleavage stage fossil embryos has been found with polar lobes at the 2 and 4 cell stages of early embryogenesis (Chen et al. 2006). The polar lobe is a transitory structure that forms from the vegetal pole of the egg during cleavage and shunts cytoplasm to one of the blastomeres at the four-cell stage. The descendants of this cell play a major role in setting up the dorsal–ventral axis of the embryo in some genera of extant lophotrochozoan animals (Freeman and Lundelius 1992). These embryos appear to be derived from different species because the egg sizes of embryos producing polar lobes vary.

In addition to Ramitubus, there are three other tubular genera, Sinocyclocylicus, Quadratitutbus and Crassitubus that first make their appearance during this period. Sinocyclocylicus and Quadratitutbus have internal cross walls; however, in these three genera it is not clear if the animals that lived in the tube had radial or bilateral symmetry. All three genera were probably sessile.

570–580 MA: Four colonial and one unitary genera appeared during this period, all of these genera were sessile. In colonial forms, there is a central vane; primary branches come off the central vane, secondary branches come off the primary branches, third order branches can come off the secondary branches and zooids (frondlets) form on third order branches. (See Brazier and Antcliffe (2009) for the nomenclature used to describe branching and Narbonne et al. (2009) for photographs of zooids of these frond-like animals.) Because of the small size of the zooids one cannot say with certainty whether or not they are radially or bilaterally symmetrical. Trepassia was formerly called Charnia wardi. These frond-like colonial animals are referred to as rangeomorphs. Thectardis has a cone-like morphology with an attachment site at the apex of the cone. Other genera with this morphology appear later in the Ediacaran fossil record.

560–570 MA: During this period one bilaterian and one radially symmetrical unitary genus appeared in the fossil record. Epibaion was a large flattened bilaterian that presumably made its living moving over and grazing on surfaces. Feeding trails left behind by these animals, directly adjacent to a body fossil help to define their anterior–posterior axis. (This genus resembles Dickinsonia which is found in the 550–560 MA interval.) It appears that during life the base of Inaria was partly covered by the substrate and there was an apical extension of the animal with a retractable feeding apparatus that probably had a tentacle-like morphology. The other seven genera were sessile colonial organisms with a rangeomorph body plan with zooids distributed along their terminal branches; it is not clear whether their zooids were radially or bilaterally symmetrical.

550–560 MA: During this period 22 bilaterians, 15 unitary genera that do not have the overt characteristics of bilaterians and eight colonial genera appeared in the fossil record.

Two of the bilaterians appear to be members of stem lineages that may have evolved to form extant phyla. Parvanocornia has a dorsal shield that covers the body; there is a ventral median region that paired ?limbs emerge from; it closely resembles and has a growth trajectory similar to the Cambrian arthropod Skania (Lin et al. 2006). Another bilaterian, Kimberella, is a putative stem group mollusc (Fedonkin et al. 2007c). This organism has a stiff non-mineralised dorsal shell with a characteristic ornamentation. A ventral foot is present; locomotory traces are associated with the foot in some specimens. There is an anterior proboscis that could be extended beyond its shell; there can be feeding traces associated with some animals suggesting the presence of hard claws or a rasp-like feeding structure associated with the proboscis. The animal has an internal axial structure interpreted as an alimentary canal. In the groove between the shell and the foot there are bilaterally arranged ridges on each side of the animal that could have functioned in respiration.

The affinities of the other bilaterians to modern animal phyla are more problematic. There have been a number of attempts to assign Ediacaran genera to groups on the basis of their characters; however, there have been no formal cladistic studies of these animals. A group of bilaterians that first appear in the fossil record at this time had an anterior head region and a set of somites which made up the posterior part of their body and are referred to as Vendomorpha. The genera included in this group include, Cyanorus, Vendia, Karakhtia, Onega, Pseudovendia, Spriggina, Lossinia, Armillifera,Yorgia Archaeaspius and Praecambridium. Some of the members of this group including Cyanorus and Vendia had a tubular digestive system that ran along the animal's anterior–posterior axis with bilateral caeca that extended from the digestive tube to the somites that make up the posterior part of the body. Yorgia left locomotory traces. Temnoxa which had a head region and an unsegmented body region probably also belongs to this group.

Another group of bilaterians are referred to as the Dipleurozoa; it includes Dickinsonia, Andivia and Solza from the interval 550–560 MA and Epibaion from the interval 560–570 MA. In these species, cephalisation is minimal. The organisation of Dickinsonia has previously been discussed in the critique of the work of Shen et al. Andivia and Solza had non-mineralised dorsal shells and a ventral foot. There are five worm-like genera; four of these genera are trace fossils. Another genus labelled ‘bilaterian’ has a through gut with a bulb at one end and appears to have been in the process of making a locomotory trace.

The unitary genus Palaeophragmodictya is a poriferan with a reticulate mesh of spicules and a centrally placed osculum. There is a bag-like unitary fossil Bonata that appeared to be attached to and partly covered by the substrate. Inaria from 560–570 MA has a similar morphology; it was also embedded in the substrate. There is a set of disc shaped genera that may have been attached to the substrate that had higher order radial symmetry; five genera, Albumares, Anfesta, Skinnera, Triforillonia and Tribrachidium exhibited triradial symmetry, one low cone-shaped genus, Conomedusites exhibits tetraradial symmetry and one genus Arkarua exhibits pentameral symmetry. Vendoconularia has a conical form and resembles Thectardis from the interval 570–580 MA. Three disc-like genera with tentacles may have been pelagic: Veprina, Eoporpita and Rugoconites. Both Calyptrina and Protoconites lived in tubes so that it is not possible to say if they were radially symmetrical or bilaterians.

Ramellina and Gehlingia are colonies with multiple zooids along their arms that extend from its central axis. The rangeomorph colonies that first appeared between 560 and 580 MA are also organised in this way. Pteridinium, Pambikalbae, Phyllozoon and Swartpuntia have a central axis with vanes made up of double rows of closely packed tubular elements extending out of it. The ends of these tubular elements furthest from the insertion of the vanes on the axis are open and presumably contained the zooids. While the sessile tubular colonial genus Funisia is much larger than Ramitubulus it had the same body plan and reproduced asexually in the same way. Ventogyrus is a polar form composed of three radial modules organised around an axis. Each module has a large basal chamber and is divided into two lateral compartments that are divided into smaller chambers of zooids. Because of the orientation of members of this genus and their organisation they are thought to have been pelagic. The zooids of these colonial animals could either be radially or bilaterally symmetrical.

543–550 MA: Thirteen bilaterians, four unitary genera with radial symmetry and one colonial genus with radial symmetry made their appearance in the fossil record during this period. In addition, seven unitary tubular genera, two colonial tubular genera and four colonial genera appeared; in these cases, it is not clear if the organisms were radially symmetrical or bilaterians One new trait that appeared during this period was biomineralised shells.

Among the bilaterians the genera Paravendia and possibly Tamga were vendomorphs. Windermeria and Ivovicia were members of the Dipleurozoa. Platypholinia and Vladimissa were flatworm-like and might be considered as dipleurzoans. Archaeichnium was a tubular fossil with a wall composed of grains of sediment; it left locomotory trails. Protechiuris was an elongated body fossil with a narrow spatulate anterior end, a flattened lobe on one side and a convex lobe on the other side. There are five trace fossils; three are from the Blue Flower Formation in Northwest Canada. Two of these trace fossils, Palaeophycus and Torrowangea, are also present at Nama where these trace fossils are associated with the late Precambrian index fossil Cloudina.

All four of the unitary radially symmetrical genera, Nemiana, Ernietta, Vaveliksia and Namacalathus were sessile. Nemiana lived in close proximity to its conspecifics. Ernietta was partly buried in its substrate. Namacalathus are relatively small animals, the body of which is connected to a stalk attached to the substrate; parts of the animal were mineralised. Namapoikia is a cnidarian that built large calcareous reefs as such it is a radially symmetrical genus that reproduced asexually. Valveliksia has a holdfast and an elongated radially symmetrical bag-like body.

The unitary tubular genera are: Soldadotubulus, Onuphionella, Sinotubulities, Gaojiashania, Ningqiangella, Conotubus and Chenella. Sinotubulities lived in a mineralised tube that was open at both ends (Chen et al. 2008). The two colonial tubular genera, Waltheria and Cloudina also had mineralised shell walls. These tubular genera could have been either bilaterally or radially symmetrical animals.

Three of the colonial frondose fossils, Paracharnia, Rangea and Nasepia were rangeomorphs. Corumbella had vanes composed of tubes attached to a central axis with zooids at the end of each tube; its organisation is similar to Pteridinium and Swartpuntia.

Discussion and conclusions

Shen et al. (2008) have reported that the full range of Ediacaran morphospace was established in the Avalon fauna and that these morphospace parameters remained essentially unchanged in the younger White Sea-Flinders range and Nama faunas. They have also suggested that there was a special set of circumstances analogous to those that might have caused the Cambrian explosion that generated the Avalon fauna.

The morphospace analysis that Shen et al. (2008) did was flawed because the character set used was selected without considering the possible promorphologies of the animals whose fossils were being examined. They made no attempt to distinguish between unitary and colonial animals. The symmetry characters they used could not distinguish between animals that are radially symmetrical and bilaterians. The authors make the statement that they ‘make no inference on phylogenetic homology and functional biology of the coded characters’; this approach led to data sets that were ambiguous because a given character state could have different meanings in different contexts.

Three practices the authors used tended to homogenise the morphospace envelopes for the fauna's they analysed. (1) A given part of a fossil, such as a holdfast, if it was separated from the rest of the fossil, was treated as a species even though it had a much simpler morphology than the animal it came from. (2) A given genus was used to construct the morphospace of a younger fauna, even if it had already appeared in an older fauna. This carry-over tends to make the faunas similar and masks morphospace change. (3) Even though the Shen et al. paper is primarily concerned with the Avalon-Charnwood, White-Sea-Flinders and Namibia faunas, data was also collected on the faunas from the Makenzie Mountains, Great Basin, British Columbia, Wernecke Mountains, Olenek uplift, Finmark and Podolia. The only fauna excluded from the study was from the Yangtse Gorge. The reason given for excluding this fauna was its compressed state. Much of the Yangtse Gorge fauna is three-dimensional and is better preserved than the classic Ediacaran assemblages; it includes the oldest and most informative Ediacaran fossils.

When the appearance of new Ediacaran genera is plotted through time in ca. 10 million year bins, between 570–615 MA only a relatively small number of genera appeared compared to later in the Ediacaran Period. This could indicate inadequate sampling; it could also reflect a low intrinsic level of genus origin. Of far greater importance is the appearance of representatives of two radially symmetrical crown group phlya: Porifera and Cnidaria, prior to the appearance of Avalon fauna fossils. There is evidence for new genera in these phyla at later points in the Ediacaran period. Molecular phylogenetic analyses (Dunn et al. 2008) indicate that radially symmetrical phyla are the sister group of the bilaterians. There are several lines of evidence, body fossils, trace fossils and embryonic stages, that bilaterians were present during this early period; however, a significant number of bilaterian genera did not appear in the fossil record until 550–560 MA and crown group bilaterians probably did not appear until the Cambrian.

The increase in the proportion of new bilaterian genera and the decrease in colonial genera during the Ediacaran period are shown in Figure 3. The bar on the left side of each set of graphs covers a time period that is 4.5 times as long as each bar on the right. The graphs on the left (A) shows the relative increase in new unitary genera and the decrease in colonial and tubular genera with time. Since colonial and tubular genera can be either radially or bilaterally symmetrical the graph at the right (B) shows the percentage increase in new bilaterian unitary genera and the relative decrease in radially symmetrical unitary genera and colonial genera that are known to be composed of radially symmetrical zooids. During the periods 615–560 to 550–543 MA, the percentage of new bilaterian genera went from 50 to 72% and the percentage of colonial genera went from 65% to less than 20%. The change from a radial to a bilateral body plan and from a colonial to a unitary body plan are each major changes in morphospace. One does not need 50 characters to define Ediacaran morphospace; a few characters that will distinguish between the promorphological features of body plans will do.

Figure 3 A) The percentage of new genera in each time interval that are either unitary or colonial. Both categories may include some bilaterians. (B) The percentage of new bilaterian genera at each time interval included in radially symmetrical unitary and colonial genera with radially symmetrical zooids. The top of each graph gives the number of genera.

The new bilaterian genera that originated between 560 and 543 MA exhibited a variety of different morphologies that could be characterised in part as vendomorphs, Dipleurozoa with forms covered with a dorsal carapace and worms. Given the genesis of possible stem group Arthropoda and Mollusca during this period, possible groups would have to be around that the precursors of these extant phyla would evolve from; this was certainly the case. At the same time there was a marked increase in the size of bilaterians even though small forms still persisted (Table 1).

Why was it that the proportion of new bilaterian genera did not begin to increase significantly until 560–550 MA? There were two sources of food for Ediacaran animals: bacteria, small unicells and multicellular organisms in the pelagic environment and (2) benthic organisms on and in the substrate (Narbonne 2005). The first animals to form extant phyla were the Porifera and Cnidaria. These animals were filter feeders (Porifera) and possible passive trappers of small pelagic animals (Cnidaria). The stem groups that gave rise to these extant groups had probably been using these feeding mechanisms for millions of years and would continue to use them. The early bilaterians were small worm-like (trace fossils) and flatworm-like (Vernanimalcula) animals that had recently evolved from ancestors that were radially symmetrical. The part of the ecosystem that they would feed in as grazers would be a benthos dominated by bacteria coated substrates and algae. It probably took these initial bilaterians a period of time to adapt to this new niche within their ecosystem. This may explain the delay in the proportional generic increase in bilaterians. Near the end of the Ediacaran period another modification took place in this ecosystem involving the introduction of predation by bilaterians on other metazoan animals, documented by the appearance of bore holes in Cloudina shells (Bengston and Zhao 1992).

Shen et al. (2008) argue that the morphospace of Ediacarans did not change through time even though the species that make up the fauna do. This suggests that there were no major changes in their utilisation of the ecosystem. The picture that emerges in the data presented here demonstrates that there was a shift in the proportionate generation of new bilaterian genera during the late Ediacaran that may have been tied to their ability to utilise the benthic environment.

Postscript

Xiao and Laflamme (2009) have recently produced a review that includes a section on the evolution of the Ediacaran fauna. This review focuses on the Avalon, White Sea and Nama assemblages. A possible phylogenetic placement of radial and bilaterally symmetrical fossils is given in their Figure 2. Avalon Rangeomorphs are the oldest Ediacaran clade shown and are indicated as the stock that the sponges and cnidarians evolved from. Sponges and cnidarians are not given crown group status until the Cambrian. Early bilaterians are shows evolving from stem group cnidarians during the late Ediacaran. This speculative phylogeny does not fit the facts. The fossils in the early Doushantuo Formation of the Yangtse Gorge predate the Avalon assemblage and include crown group sponges and cnidarians in addition to early bilaterians. It is probable that many of the bilaterian fossils found during the late Ediacaran (560–543 MA) evolved from these early bilaterians. If one is interested in the origin and early evolution of fossil bilaterians a good place to look for the appropriate fossils is the early Doushantuo Formation.

Acknowledgements

I want to thank Dennis Trombatore, the Geoscience librarian for his help, Marianna Grenadier for preparing the figures, and Judith Lundelius for reading the manuscript.