The oldest compelling evidence for sponges is still early Cambrian in age – reply to Love and Summons (2015)

‘Sponge biomarkers’ can either be treated as fossils or not. Let me start by agreeing with Love and Summons (2015) that biomarkers can be treated as fossils. As such, they must be subjected to the same scrutiny and analysis as normal body fossils. They are subject to natural selection and can therefore change over time. We must consider in detail their diagnostic and unique characters, and how diagenesis and taphonomy may have affected the characteristics of these fossils. Biomarkers are unique in comparison to all other fossils in that they only have one character. This is the structure of a particular organic compound, which implies in its entirety everything about the substructure of the compound. Body fossils have a suite of characters available for analysis, with characters that can have many different forms. For instance, there are very many different ways to be a sponge spicule, and these suites of substructures tell you much more about the nature of the organism. In comparison, if even one bond changes in a biomarker, then it is a different biomarker, we must call it something different, and it may only be found in completely different taxa (compare 24-ipc to 24-npc, see Antcliffe 2013). Further, when you find a disarticulated sponge spicule if it is well preserved, it will be made up of characters that individually are apomorphic for sponges and will confidently allow placement of the fossil among sponges. When you find an articulated whole sponge body fossil, individual spicules may have unique arrangements that form even more apomorphies allowing identification as a sponge. In the case of the 24-ipc marker, Love and Summons (2015) asserted that it does not matter that the biomarker character is known to be convergent between a group of modern sponges and a group of modern algae, simply because it is only produced accidentally as a by-product in algae. But the function of the character does not matter, only the presence or absence of the character and whether it is unique and diagnostic. How was this character acquired in these two very unrelated groups? Either, it evolved separately on each of the lineages leading to these groups or it is a deeply shared character that has been lost in many other groups. Either way, rewinding the clock 600 million years makes it obvious that there are many more potential producers of 24-ipc than just these modern groups. The unique nature of biomarkers as fossils with only one character prevents us from investigating character acquisition in respect to other characters of the organisms to which they belong. Love and Summons (2015) stated that it is ‘of course, self-evident’ that other Neoproterozoic organisms that might have produced 24-ipc are unavailable for analysis. On this we can agree. Fossils tell us about extinct forms, we cannot simply infer everything about biology and evolution from the modern world. When we cannot place fossil characters in the context of character acquisition, they are meaningless. When such weak evidence stands in sharp contrast to what is known from the body fossil record of sponges (Antcliffe et al. 2014; Muscente et al. 2015; Xiao et al. 2005) then it is reasonable to look at the biomarker evidence with some degree of scepticism. Arguments about the relative proportions of 24-ipc and related compounds require careful consideration of taphonomy and diagenesis, as for all very ancient fossils. 24-ipc found in a modern sample could be derived from either algal or poriferan input. How then can we use this compound to make uniformitarian inferences over the last c. 650 million years? This is particularly worrisome given that all of the concerns raised by Antcliffe (2013) concerning diagenesis and taphonomy were simply dismissed by Love and Summons (2015), without evidence or argument. I will say no more here save that these problems (i.e. differential uptake in kerogen of the biomarker compounds; the alleged exclusive link between hopane maturity proxies and that of 24-ipc; production of 24-ipc from the alteration of compounds known to be produced by nonporiferan organisms) appear not to have been remedied and that these problems are a meaningful direction for future research. One concern raised by both Antcliffe (2013) and Brocks and Butterfield (2009) is the apparent absence of Phanerozoic sponge biomarkers despite the abundance of sponge body fossils during this time. Love and Summons (2015) mentioned that there is 24-ipc from Permian sponge beds (with no data provided) in an attempt to solve this bizarre iniquity. This creates somewhat of a double-edged sword. Love et al. (2009) rejected 24-ipc contamination from the lubricating oils used during coring because these lubrication oils were drawn from Phanerozoic reservoirs, which are supposedly devoid of this particular marker. So, if there are sponge biomarkers from the Permian and the rest of the Phanerozoic (indeed 24-ipc has already been confirmed in the Cambrian and Ordovician by Love et al. 2009), then contamination from the Phanerozoic lubricating oils could be occurring in the Cryogenian samples. Love and Summons (2015) cannot have it both ways. Either there is no Phanerozoic 24-ipc, and this compound confusingly only occurs at times with no sponge fossils and never alongside them. Or there are Phanerozoic sponge biomarkers, and it is not possible to reject contamination from Phanerozoic-derived drill lubrication during core extraction. Future research should focus on the biomarker suites found associated with strata with well-documented sponge fossils and carefully control for the lubrication fluids used in the coring. That way we could build a picture of how biomarker compounds have changed through time and more importantly how they are related to character acquisition in metazoan phylogenies. Turning now to the correlation between a number of molecular clocks (Sperling et al. 2010; Erwin et al. 2011) and the biomarker study (Love et al. 2009). Contrary to Love and Summons (2015), the choice is not between a c. 700 Ma maximum calibration point in a molecular clock and no maximum at all. This is a false dichotomy, as a younger maximum could be selected (e.g. 550 Ma, see Lee et al. 2013). Molecular clock analyses of basal animal divergences either run right up to the maximum constraint date or, if unconstrained, until rate parity is achieved across the base of the tree. In the case of the latter, basal animal divergences are estimated at c. 1.5 Ga (e.g. Wray et al. 1996), which problematically probably predates the origin of eukaryotes (Antcliffe and McLoughlin 2008). In contrast, the differences in evolutionary rates between clocks constrained at 713 and 550 Ma are relatively negligible (Lee et al. 2013). Regardless of the maximum date employed, accelerated rates of evolution for basal animal phylogeny are unavoidable. There is nothing convincing about correlation between the statistical inferences of molecular clocks and the 24-ipc record; one certainly does not provide validation of the other. The oldest reliable sponge fossils are just above the base of the Cambrian in Iran (Antcliffe et al. 2014). These Iranian spicules are definitively sponge, but could be basal on the hexactinellid stem lineage. They date from c. 535 Ma but with a geochronological minimum of the Zhujiaqing Carbon Isotope Excursion (ZHUCE), dated to c. 529 Ma. They are roughly 5 million years older than the next sponge fossils from the lower Cambrian of Siberia, which are also stem hexactinellids. There is still no compelling evidence for crown group demosponges before the lower Cambrian, 535 million years ago (Antcliffe et al. 2014; Muscente et al. (2015).