Tarsier

(vagabondish.com)

I thought it would be a good idea to give a real-life example of matched-pairs nested comparisons, and this paper by Gittleman and Purvis (1998), published by the Royal Society is a great example. 

They use this technique to overcome the pitfalls in examining body size with species richness in mammals.  They collected data on body mass for 175 primate and 240 carnivore species.  They used the technique outlined in my previous blog (along with other statistical methods to test the body mass-species richness hypothesis) and found amongst high levels of heterogeneity between clades, no significant relationship exists in the primates, and a significant (but weak negative) relationship in the carnivores (especially canids), as in figure 1a  (top) and b (bottom).   

 

Figure 1 a)       Large-bodied carnivore clades tend to be significantly less species rich that small-bodied clades (black circles represent mustelids, procyonids, Ailurus, and pinnipeds), white circles are others.  b)  No significant pattern described in primates.  Least square regression line through origin. 

These are the overall trends, but certain clades e.g. the fox-like and cat-like carnivores show a stronger relationship than closer relatives.  The authors explain that the associations highlighted by this technique are like to reflect differential extinction because both clades have suffered size-selected extinctions (e.g. extant primates on Madagascar are smaller than their extinct relatives).  This study shows that body size might be important in diversification within clades (i.e. foxes), but not among them (i.e. primates).  This technique is now used to assess extinction risks, and will be discussed later.  

Reference:   Gittleman, J., Purvis, A. (1998).  Body size and species-richness in carnivores and primates.  Proceedings from the Royal Society London, 265, 113-119

Looking at diversity: phylogenetic approach

Phylogenies are used to assess the relatedness of species, genera or family to their respective ancestors and relatives.  They are particularly useful because they allow us to separate some of the different aspects of biodiversity.  For example:  there are around 1.36 million extant animals known and there are 36 animal phyla (taxonomic classification of groups in the kingdom Animalia).  If the mean number of species is around 38000 species per phylum, what do you think the MEDIAN number of species per phylum is? 

1. 3
2. 33
3. 330
4. 3300
5. 33000

Well given that over one million species are within the phylum Arthropoda, and two phyla contain just one species, the median is 330.  Species richness is not evenly distributed, meaning some taxa dominate over others.  Why might this be?  Well taxa at the same rank may be older and so have had more time to diversify.  Or maybe it’s just chance?  Phylogenies can help to explain the correlates of biodiversity, and thus expose traits which might predispose species to diversification (in this blog) and extinction (in the next blog).  For example, it’s well known that big animals are less diverse than small animals (see figure 1 for Mammalia).  We cannot simply test our hypothesis that rodents are more diverse than elephants because they are small.  The problems in testing are due to confounding variables and non-independence. 

Figure 1  Phylogeny of the mammals

One lineage might have evolved into grazers, and to be a good grazer being big is key.  This lineage might produce 5 families of big grazers (figure 2: green dots).  Another lineage from the same ancestor may have evolved strong female mate choice, which in turn speeds diversification (figure 2:  red squares).  In this lineage there’s high speciation rate and so has greater species richness. 

Figure 2  Schematic of two lineages diverging with different traits described in the text

At face value, the large grazer families are less species rich than the smaller families, but not because of body size!  Phylogenies remove this non-independence problem by comparing sister clades because they are automatically the same age.  Nested comparisons (red in figure 3a) can be made between  matched pairs. 

Figure 3a Two pairs (sister clades) with number of species S in each lineage (yellow) and total species in the sister clade (red).  The trait X is body size (yellow, red is mean body size).  3b  Correlation of species richness and trait. 

Here are two sister clades, within each matched pair, does having more species (S) correspond to having a larger or small mean trait (X)?  If there is no correlation between the two, then the null hypothesis (X is larger 50% of the time) cannot be rejected.  A regression (figure 3b) can be used to look at the:

                                                

relative rate difference (for ΔS)=

                      

These kinds of analyses are particularly useful when testing hypotheses about co-radiation (i.e. have two completely different lineages such as flowering plants and beetles driven each other’s diversification), types of selection for traits (i.e. dichromatism in birds shows strong sexual selection), and other evolutionary phenomena. 

This has been a brief introduction into phylogenies, and they can be used to predict correlates of extinction in a very similar way that they predict correlates of diversification.   Using the fossil record, we look at traits which are predisposed to extinction and extrapolate those traits into contemporary diversity. 

Reference:  All personal communication Prof A Purvis.

The species problem

Biological biases in the fossil record manifest themselves in different ways.  Firstly, we should consider species concepts.  Traditionally, a species was defined by its morphology.  That is, a species is a collection of individual organisms with similar morphology (or other traits) which differ from other such groups.  Most of us can quite confidently tell that the animal in figure 1a is not in the same species as 1b.  Other than being striped, furry and legged, they are completely different morphological entities. 

Figure 1            a) tiger                                                                  b) bumblebee

The biological species concept was borne from modern day biology, when species were studied in their environments and is defined by their breeding habits.  A species being groups of [potentially and/or actually] interbreeding populations which are reproductively isolated from other such groups.  In practice, this concept is not used in the fossil record because we lack information on breeding habits.  The modern-day phylogenetic approach is used in extant species using DNA sequencing.  The phylogenetic species is an irreducible cluster of individuals which can be diagnosed as completely different from other such groups (this can be traits or genes).  Recently this approach has uncovered the cryptic species concept, a group of organisms which are morphologically very similar, but genetically very different to related species.  

Figure 2         Male (columns 1 and 3) and females (columns 2 and 4) of four Perichares species.  Left is ventral and right is dorsal view (from http://www.pnas.org/content/105/17/6350/F3.expansion.html)

Defining extant species seems to be quite difficult.  Defining extinct species from fossils is even more of a challenge because we are left with ancient remains, no or little information about breeding, and usually no DNA sample.  This leaves the morphological concept.  Unfortunately, the morphology of some organisms can reduce their fossilisation potential.  Species without identifiable hard parts are may only be fossilised in specific conditions (e.g. jellyfish are best fossilised by the hardening of their surrounding substrate, which then leaves a natural mold).  This creates a bias in the paleorecord towards those organisms which have a more ‘fossilisable’ anatomy, or indeed those fossils which are easy to identify and find (e.g. bivalve molluscs). 

Due to the complexities in identifying distinct extinct species, past diversity is measured using higher taxonomic classifications (such as genus or family).  This automatically fills in the gaps because a genus is still alive until the last individual species goes extinct.  This also has its drawbacks because some genera are not uniformly species rich, meaning a genus of 2 species can be more susceptible to extinction than a genus with 200 species (figure 3).  Therefore genera with only few species (who may also be rare) are likely to be under-represented by the fossil record and potentially, more likely to go extinct without either being fossilised, or being discovered as a fossil.  Furthermore, higher taxonomic units are not comparable between groups because there is no consistent genus/family concept (i.e. it is not temporally standardised). 



 Figure 3   Uneven distribution of species among: a, eutherian orders, with rodents being the dominant group; b, rodent families, with murids being dominant; and c, murid genera (from Purvis and Hector, 2000)

Given the issues highlighted, it is unsurprising that it is difficult to pin down the vital statistics of past global biodiversity.  Paleobiologists have been aware of some of them for some time, and thus potential solutions have been suggested.  But given the uncertainty of our past, how can we compare the extant with the extinct when we have two completely datasets?  I will highlight some of the issues in comparing past with present, and some of the ways the aforementioned biases have been addressed. 

References

Barnosky, A., et al. (2011), Has the Earth’s sixth mass extinction already arrived?  Nature, 471, 51-57.

Purvis, A., and Hector, A., (2000), Getting the measure of biodiversity.  Nature, 405, 212-219

Purvis, A., Jones, K., Mace, G. (2000), Extinction.  Bioessays, 22, 1123-1133