Narrative ruins history?

Coming from a scientific disposition, I have a scientist's rapacious desire for The Truth.


That's the beauty of science.  We get closer to Truth all the time - and the mis-steps and side alleys are far fewer and less significant than the ascientific (as are many climate change deniers) would have us believe.  Mostly, refinements are built upon refinements, and previous truths are hardly ever gainsaid - at least not significantly.  Quantum and Einsteinian physics don't negate the reality and applicability of Newtonian physics on an everyday, human scale.

 History is unlike science in so many ways, but the one that springs to mind right now is narrative.  In that sense, history is more like shoddy journalism - even good journalism - in that it tries to tell a story.  And the failing is that the whole of the truth is sacrificed: the nuance, the periphery, and the way in which life is not quite like fiction or myth; it doesn't have unity of purpose or theme, or specific point.




True history is messier, and purposes cross, narratives interact without clarity or precision.  Out of all that, historians and journalists are alike with novelists, trying to create a single strand (or multiple strands) where the full story is so much more complex, riddled with irrationality and strewn with different actors' clutter and concealment.  And of course, it's only one person - or peoples' - truth.  And even then, much of the time the truth will simply never be available.  So, from honourable motives or not, the historian as storytellers will attempt to persuade rather than prove.




I read some narrative in science.  Stephen Jay Gould and Richard Dawkins are good tellers of short stories.  But their tales are much better corroborated and agreed upon.  And if a single essay tells only part of the much wider discipline of evolution and genetics, the rest of it is there for the taking - not the disputation, disagreement, and ultimate irresolution.

Still, for the scientifically-disposed, at least history is better than fiction: there's more truth in it.  And if we long for a cracking good story, then at least we know there's more to it than we're being told.

Rees, Dawkins and Gould: finely picking the cleft between science and religion

I first encountered Martin Rees, president of the Royal Society, in the pages of New Scientist last year: a brief interview on occasion of the 350th anniversary of the Society. He talked science, but looked nothing less than an Establishment bastion (and he’s also Astronomer Royal, Master of Trinity College, Cambridge, and in the House of Lords).


Just recently, I heard him in a brief piece on the ABC Radio Science Show. He was talking on the divide between science and religion – and he rather surprised me with some non-establishment words. His conversation drilled into some finer points of this divide. Inter alia:

“I suspect my beliefs or lack of beliefs are rather similar to Richard Dawkins's…”

Then:

“I agree with Richard Dawkins that fundamentalism… is a real danger, and I think we therefore need all the allies we can muster against it, and I would see the Church of England and the Archbishop of Canterbury, for instance, as on my side against fundamentalism. Therefore it seems to me counterproductive to rubbish people like that. I'd like to see them on my side, and as a Brit who grew up in that culture, I'm rather supportive of the Church of England.
And I think there's another reason where I think his attitude is also damaging. Suppose you were teaching a group of kids in a London school and a lot were Muslims and you told them that they couldn't have their god and have Darwin. They're going to stick with their god and be lost to science, and that, again, I think is counterproductive. So I believe wherever possible one should have peaceful coexistence with mainstream religions.”

The Templeton Prize was apparently a notable point of departure between Rees and Dawkins. The prize “honors a living person who has made an exceptional contribution to affirming life’s spiritual dimension, whether through insight, discovery, or practical works”.

Rees said Dawkins labelled him a quisling because Rees was “less hostile to the Templeton Foundation than he was.” But then: “I don't go all the way with the Templeton Foundation because they believe in constructive dialogue. I think there's limited scope for constructive dialogue, I think there can be peaceful coexistence, but I don't believe theologians can help with my physics”.

Per se, that’s not too far from the resolution attempted by another scientist, Stephen Jay Gould. He coined the term Non-Overlapping Magisteria (or NOMA): “Science tries to record and explain the factual characteristics of the natural world, whereas religion struggles with spiritual and ethical questions about the meaning and proper conduct of our lives. The facts of nature simply cannot explain correct moral behaviour or spiritual meaning”.*

Of course, that satisfied few who weren’t already satisfied.  And although the two are similar, I rather like Rees' turn of phrase.



Full text of the Rees conversation available here.


*Gould, SJ (2003): The Hedgehog, the Fox and the Magister’s Pox (p87). Jonathan Cape, London.

Evolution: Darwin's modification by descent

There's something of a flood of media relating to Darwin, as it is the bicentennary of his birth and 150th anniversary of his book On The Origin Of Species. Even something in it for me, as I heard on the radio part of a 1987 event featuring successive addresses by Richard Dawkins and Stephen Jay Gould. They both downplayed any beatup notion that there was any fundamental dispute between them, while emphasising their different perspecitives on their field. Still it was fascinating to listen to them both.

Although I am largely not one for hero-worship, Darwin must be given strong credit for introducing ideas that were revolutionary and prescient. On the latter, he broadly predicted the existence of a mechanism for evolutionary inheritance with modification, without knowing what shape it would take, and decades before it was widely verfied and accepted.

As part of the radio programme, I heard someone give a Darwinian characterisation of genetic variation as "copious, small and undirected". One of Darwin's revolutionary points was that evolution is paradoxically "neither random nor intentional". Not intentional, in that there is no overarching will at play that can make decisions about the process. And although the fodder for variation is random mutation, the effects are not, as species survival is directed by the totality of its environment.

Someone in that broadcast commented drily that although Darwin was inspired by Adam Smith's economic theory of numerous single players in a market vying for their own survival, Smith's theory "didn't work". Now that may be a bone of contention for some, but the purity of Smith's vision relied on equal players in a market where no monopoly develops - and we all know that in the absence of a directed force (ie a government), the natural tendency of capitalistic markets is towards monopoly. The analogy was imperfect; nevertheless it is the mark of a significant mind to be able to associatively draw from other fields to forge new understandings of one's own.

DNA coding - message in a bottle?

"DNA's performance as an archival medium is spectacular. In its capacity to preserve a message, it far outdoes tablets of stone. Cows and pea plants... have an almost identical gene called the histone h4 gene. The DNA text is 306 characters long... cows and peas differ from each other in only two characters out of these 306... fossil evidence suggests [their common ancestor] was somewhere between 1,000 and 2,000 million years ago... Letters carved on gravestones become unreadable in mere hundreds of years”

- Richard Dawkins, The Blind Watchmaker (p123).


The implications, at first glance, are quite spectacular too. For example, why not preserve meaningful information in such a fashion? We send out signals seeking contact with extra-solar civilisation; why not inscribe similar messages in DNA to likewise reach across the distance of time? The human genome, for one, is several billion base pairs long, plenty of space for encoding information that can be read as clear messages.



There are in fact several reasons why this is somewhat impractical.

First, we are already halfway through the effective life-supporting span of the solar system. If, for example, we were to take to the extreme this current artificially-induced extinction event (global warming and destruction of biodiversity), we may leave few species behind; humans would not ipso facto be the most robust of them. If we were to propel destruction back to the bacterial level, there could well evolve again life forms sufficiently complex to analyse and read such messages – but the timing would be quite fine. The gap between “Oh, someone's encoded a message for us in DNA” and the sun expanding to render the planet uninhabitable, could be so small that contingency might not allow for that rediscovery. A simple event on the scale of the KT event's meteorite can play havoc with such timing.

Second comes the inevitable problem with seeking to encode for two different – potentially conflicting meanings. (This is why database designers tend to create primary keys that are independent of specific data fields.) On the one hand, it would be tricky to code a section of DNA to be meaningful both genetically and as a message. And there is no guarantee that such genes would not be subject to evolutionary changes that obliterate the message.

On the other hand, large sections of genome are seen as “junk DNA”, that is, likely to be filling no purpose directly relevant to an organisms makeup. (which is not to say junk DNA is fully useless – for any organism to carry any excess baggage, there is a cost. We just don't know for sure the purpose and origin of junk DNA. It seems to consist of duplicates and misprints of DNA present elsewhere, rather like a computer's waste bin that hasn't been cleared.)

However, junk DNA looks to be more susceptible to mutation than purposive genetic material. Why? If mutation is steady and equally likely throughout the genome (say, for instance, that solar radiation causes a slow but steady rate of damage – a small percentage of miscoding – in haploid genetic material), DNA that has purpose is more subject to error-correction – via the decrease in viability of mutated, ie DNA-damaged, individuals. Thus junk DNA mutation – coding errors – at the individual level is more readily retained, and the information inherent in that junk code would change more frequently. An ideal vacant repository for information, but not as secure.

So a genetic designer could conceivably store non-core information in DNA, but couldn't reliably expect it to last through an evolutionary time scale. However, I can picture the technology being developed to enable insertion of signature or copyright information in junk DNA that would last the required human timeframe.






References

Dawkins, R (1986): The Blind Watchmaker. Penguin, London.

Evolution: sex scandal: how rotifers get away without it - more questions than answers

Richard Dawkins discusses the scandal of sex in his book The Ancestor's Tale.

Dawkins originates the concept with John Maynard Smith: that sex itself is an evolutionary scandal. At its simplest, one would expect that sexual reproduction is counterintuitive in a theoretical sense. In an evolutionary context, it is in the individual's best interests to pass on as many genes as possible, yet sexual reproduction requires a halving of the amount of genetic information passed on.

It would make sense if there was some value in sharing the load of passing on genes, such that genetic survival was enhanced if the two parents worked together.

Yet it has been found fairly consistently that the male does much less work - so the workload is not shared to the benefit of the genes.

That is one scandal. The other could be characterised as a scandal within a scandal (Dawkins: a paradox within a paradox). That is, the bdelloid rotifer.


Rotifers are tiny marine organisms (as are many odd creatures). Although many animals have asexual reproduction within a broader repertoire, bdelloid rotifers consistently reproduce without sex. That is, as far as their natural history can be traced - at least 40 million years - there has never been any evidence of two sexes (effectively classifying them as female, and eliminating the concept of a male).

Intuitively, I would expect sexual reproduction to make sense: halving the genetic information passed on would stabilise a population that experiences constant random mutation - as all species do*. This because most mutations are harmful to individual and species survival, but constant melding of genes can spread the load and weed out the unviable outliers.

So bdelloids are a scandal: how can they consistently reproduce, without succumbing to gradual genetic (and thus individual and species) degradation into unviability?

A recent study (reported here) has gone some of the way to unravelling the mystery. In the journal Science, Harvard scientists reported (abstract here) that bdelloids "steal" (absorb) genetic material from a whole host of other organisms. Not just animals: they also found genes from bacteria, fungi and plants. They have "relaxed the barriers to the incorporation of foreign genetic material".

This definitely gives some answers, but opens up many more questions. Including:
1) how do they stabilise as a species despite the absorption of foreign genes?
2) how do they "resist radiation" that affects the genetic information?
3) how does this tie in with the understanding that complex multicellular life developed through the unification of genetic material from diverse basal organisms?


I suspect that these three questions are wrapped up in each other, and I believe this is a highly significant find. I get the feeling we're on the verge of a more comprehensive understanding of how evolution (via genetic information) works.


*What causes mutation? Intuition again tells me this is directly due to radiation, probably mainly solar, which is otherwise beneficial - and still is, I contend. But this is a whole other story.



Reference
Dawkins, B (2004): An Ancestor's Tale. Phoenix, London.

Evolutionary explanations with heterochrony

Neoteny is a very interesting concept in evolution. It's the process by which an animal [which differs between juvenile and adult form] becomes sexually mature in the larval (or juvenile) form. Thus can be set off a divergent evolutionary path.

The Axlotl (sic) is probably the most well-known example. It's a member of the salamander's order, Urodela.

Although salamanders look rather like lizards, that's just due to convergence (see previous evolution post), and they're actually amphibians. Colin Tudge refers to that shape as the "archetypal tetrapod" - albeit one whose legs sprawl out the side, as in many reptiles, as opposed to being positioned beneath the body as with most mammals.
The axlotl is effectively a salamander that never made it to land. The adult form retains a number of the features of the salamander's early water-based existence, such as gills, while having grown the limbs that would be expected in an adult salamander.

Richard Dawkins (The Ancestor's Tale) details experiments that verified this neotenous nature - separately conducted by Vilem Laufberger and Julian Huxley (brother of Aldous). They each used hormone injections (thyroxine, the absence of which is said to have set off the neoteny), and induced the axlotl to lose its gills and become a fairly normal salamander.

Neoteny is actually the slowing down of other aspects of development relative to sexual maturity. This is subtly different from the acceleration of sexual maturity relative to the rest of the body, called progenesis.

Dawkins refers to the general case as heterochrony, which is the change of pace (a slowdown or a speeding up) of one developmental process relative to other developmental processes. He suggests that this must be behind a high proportion of evolutionary changes in anatomical shape.

Dawkins then mentions the newt, a type of salamander which does this twice, so to speak. It matures to a land-based adult form, but doesn't reproduce. It only becomes sexually mature when it returns to the water, and reclaims some (but not all) of its larval features. Strangely enough, one aspect that doesn't roll back is the gills. It's easy to postulate that the land-based form is not as redundant as it seems: it may have been quite useful at some point to ride out dry seasons - or, more plausible to me, it may have been evolutionarily advantageous to be able to travel a fair distance between ponds or other such water bodies. And it may have been easier to retain water-based birth practices (a somewhat harder call at why it switched the second time - it may have just reverted).

It is also speculated that humans underwent several differeny kinds of heterochrony - for example, the human brain keeps developing for several years after birth, which doesn't happen with chimpanzees. In fact, Wikipedia's Heterochrony entry claims (based on Penin, Berge, Baylac, 2002 and Mitteroecker, 2004) that humans demonstrate 30 different neotenies compared to chimps.


Reference

Dawkins, B (2004): An Ancestor's Tale. Phoenix, London.
Mitteroecker, P et al (2004): "Comparison of cranial ontogenetic trajectories among great apes and humans" in Human Evolution (2004) Volume 46, pages 679-697.
Penin, X, Berge, C and Baylac, M (2002): "Ontogenetic study of the skull in modern humans and the common chimpanzees: neotenic hypothesis reconsidered with a tridimensional Procrustes analysis" in American Journal of Physical Anthropology (2002) Volume 118, pages 50-62.
Tudge, C (2000): The Variety of Life. Oxford University Press, Oxford.

Convergent evolution

This is always such a fascinating subject. Earth's natural history is positively littered with examples of different species that develop remarkable similarities to each other in complete evolutionary isolation.

An easy example is the body shapes of sharks (fish), cetaceans (mammals: whales, dolphins, etc), and ichthyosaurs (porpoise-like reptiles): all evolved fairly similar adaptions for swimming, although a notable difference is the tailfins of cetaceans, giving rise to an up-and-down tail movement, rather than side-to-side for the others. A ready observation would be that they show similar vertebrate adaptions to the similar demands of the environment.

There are so many other examples: similar-shaped and/or similar-functioned animals around the world. Try unrelated fish in the arctic and antarctic waters each with an equivalent anti-freeze-like chemical in their blood, with genetic origins quite unconnected.

Try re-invention of the eye several times. Albeit the squid's, for example, is better engineered than ours, since our retinal nerve endings emerge between lens and retina, making for less efficient vision that the squid's whose nerve endings sensibly emerge behind the retina. Moving ahead of myself in Richard Dawkins' The Ancestor's Tale (currently at p344), Dawkins mentions on p602 an expert in comparative zoology of eyes, Professor Michael Land, who identifies nine independent principles of optical mechanics, "each of which has evolved more than once".

Try the proliferation of parallels between eutherian (placental) mammals and metatherian (marsupial) equivalents. Thylacine and timber wolf skulls, for example are said to be nearly identical, despite being completely unrelated.

It sounds like a rash of grossly unlikely coincidences, ready-made for creationists to fan argumentative flames. Yet in some ways it makes perfect sense: evolution is about species expanding to fill environmental niches, and
a) Over many times and locations, there are sets of all but identical niches
b) What developed and survived was what was most successful in the environment, so the directive forces of nature operated in a similar way on what was often a relatively similar genetic material.

This is obviously quite a source of frustration for taxonomy. Whereas in the relatively recent past, classification was based largely on morphology (body shape and features), modern molecular analysis has demonstrated that many of the connections drawn in the past were examples of convergence.

Dawkins also describes three different moles. The African golden mole (Chrysochloridae) was grouped with the Eurasian mole (Talpidae) in the defunct order Insectivora on the basis of such close similarity as burrowing machines: forepaws modified as spades; atrophied eyes (superfluous underground) and no visible ears. Then there's the marsupial mole: Wikipedia says they are so similar to golden moles that they were once thought to be related despite the marsupial/placental divide. Of course, all three are now classified quite separately; molecular analysis has in fact torn down the whole disputed Insectivora order altogether.

Interesting to note convergent evolution depicted as a criticism of Stephen Jay Gould. One of Gould's consistent themes was the contingent nature of evolution: how randomness played such a role that any slight re-alignment would have resulted in totally different outcomes.


I don't see these concepts as being entirely in opposition. Gould's theme plays out on a much larger tapestry than the micro-evolutionary outcomes. Convergence doesn't necessarily mean, for example, that had the K-T meteor not wiped out non-avian dinosaurs, reptilian humans would have evolved. In the absence of that meteor, dinosaurs could have remained successful in their niches for many millions more years given no other environmental pressures.


I suspect that certain paths are available only for species at a certain level of complexity; yet evolution is not specifically directional towards complexity (merely towards variation in complexity)...


How convergence does and doesn't work, what it does do and what it doesn't, is a fascinating area of study.

Reference
Dawkins, R (2004): The Ancestor's Tale. Phoenix, London.

Evolution: Gondwana vs New Zealand

Gondwana* was a large land mass that tended to reside in the southern hemisphere from Cambrian times to mid-Jurassic times (c.500 million years ago to c.160mya). It consisted, in the main, of Antarctica, Africa, South America, India, Australia, and New Zealand. For part of its existence, it was united with its northern equivalent into the supercontinent Pangea.

Through this time, Gondwana experienced a number of climate changes, due in part to the movement of Earth's techtonic plates which formed then ultimately broke up the continent.
Africa moved away first, followed by South America and India (the latter has been a particularly fast mover, crashing spectacularly into Asia to form the Himalayas).



Of course, disintegration took many millions of years, so the dates are approximate. The dispersal (and later speciation) of fauna is not entirely stopped by such splits. For some time after complete separation, islands are typically scattered between the land masses. This more or less allows for some island hopping from one land mass to another, typically through "rafting" - the carriage of fauna across on masses of vegetation, often after storms. Richard Dawkins treats this well in The Ancestor's Tale, and makes the point that even if such an event sounds unlikely, given the time scales involved - millions of years - it's unlikely that it wouldn't happen. And all it takes for a population to be established is one pregnant female.


I always thought that New Zealand broke away from Australia, because they seem to be such a neat fit. However, New Zealand is usually cited as splitting from Antarctica/Australia earlier. Ultimate separation was achieved around 82mya. Dawkins says Australia was finally sufficiently free of Antarctica to obviate island hopping around 55mya, although estimates vary a fair bit; Wikipedia suggests it was still freeing itself at 40mya.
The point about this is that New Zealand was isolated for substantially longer than Australia. At the time of the dinosaur-extinction K-T meteor - 65mya - New Zealand was already isolated, and Australia probably was not.

Various sources suggest New Zealand eroded since isolation, by up to 80% - some claim it disappeared below the waves altogether for a time, although this seems unlikely. The question is around the sustainability of populations in isolation. This is mainly relevant for large terrestrial tetrapods. Conventional wisdom is that the number of unique flightless birds in New Zealand reflects the absence of terrestrial predatory species until human introduction in the past thousand years.

There remains a few species in New Zealand's islands that are "anomalous" to the global narrative, including:

• the living Tuatara - only lizard-like - actually a distant lizard relative called a sphenodont.
  
• fossil dinosaurs and crocodilia from well before NZ's isolation
  
• and the newly-discovered sb mammal or "waddling mouse" from NZ's very recent evolutionary past, c.16 million years ago (main discussions here and here), interpreted as non-therian, ie egg-laying.

These need to be drawn into the narrative one way or another.


Related discussions:

• The evolution of moas, kiwis and other ratites;
• More detail on New Zealand's evolution as a land mass;
• Extinctions in New Zealand at early human arrival;
  
• The emergence and survival of early mammal forms, here and here.
  

*Gondwana means "land of the Gonds"; the term Gondwanaland is both redundant and obsolete.


Reference

Dawkins, R (2004): The Ancestor's Tale. Phoenix, London.

Evolution: Mammals 2: Monotremes

Odd fact of the day: echidnas derived from platypus-like creatures.


Monotremes are the only group of extant mammals that are non-therian, i.e. non-placental. (The name refers to their peculiar single opening for intestinal/urinary/reproductory functions.)



There are five identified species: four are echidna, and one is a platypus.

I happen to be reading Richard Dawkins' Ancestor's Tale. Dawkins mentions platypus fossils (Obdurodon) older than the point at which echidna and platypus diverged, along with molecular analysis that strongly implies the echidna is derived from a platypus! - effectively, over time its ancestors gradually abandoned water, changed the form of their feeding appendage (duckbill to tubular) and acquired protective quills.

Monotremes are non-therian mammals, which would mean they lay eggs. I am casting aside the oft-used term Prototherian ('first animals') as not being truly cladistic, since it excludes descendant Eutherians ('placentals') and Metatherians (marsupials).

So to other non-therians: early mammals and their non-placental descendants.
Research is hard. The situation is not as simple as Tudge's statement (dated 2000) that there are eight types of non-therian mammals - that takes a judgment call, and a lot has happened since then.

Non-therians are typically taken to include at the least: monotremes, multituberculates, Morganucodonts, Triconondonts, Doconondonts... and a few others.
How they're grouped is a matter of current debate. It would take several posts, so I intend to lay this aside for the moment.

Most of the names above are based on characteristics of their teeth - I imagine for several reasons:
- teeth are often all that remains
- they're a useful differentiating characteristic
- they give some indication of diet, and thus niche, adaption, selectivity.

Another debate also seem to revolve around the time frames that they each occupy. Of particular interest are those mammal-like creatures (mammaliformes) that existed before the K-T boundary - the 65mya dinosaur extinction event that enabled mammals to proliferate. True mammals were all shrew-sized before then, barely eking an existence in the dominant shadow of the dinosaurs.

All this is relevant because it traces the evolution of mammals, how they developed their features over time (and how fast), and how successful they were in their environmental niches against competition from non-mammals and from later developments.

More to come on non-therians. I might take a break, however, to discuss the formation of New Zealand.

The SB mammal: New Zealand's non-therian mammal

The traditional view of New Zealand fauna is that its wide range of unique bird life flourished due to the absence of mammals to prey on or compete against them. That is, until the arrival of humans (who also brought rats and dogs) around a thousand years ago.

In fact, there are at least three species of native mammals - but they are excluded in the above narrative - bats, seals and sealions.

taking a long step back, New Zealand was once part of the great southern land, Gondwana, finally separating around 85 million years ago (see Wikipedia and University of Waikato) - the former indicates New Zealand separated specifically from the Antarctic land mass before Australia did; however, it's not clear whether Australia was at the time counted as Antarctica. It seems to me most likely that Australia was its last port of call, since the coastlines match so neatly.


In 2006, a mouse-sized mammalian fossil 16 to 19 million years old was discovered in New Zealand's St Bathans fossil bed, in the south of the South Island. This has been written up in PNAS (Proceedings of the National Academy of Sciences [of the USA], available here.  Interestingly, in the paper it is said to be "nontherian". That is, it split from the mammal lineage before marsupials: therians are defined as non-egglaying mammals: thus this one (the 'SB mammal') would be defined as an egglayer.


Colin Tudge's 2000 book The Variety of Life details the metatherians (marsupials) and eutherians (so-called placentals), then mentions eight groups of nontherians, of which only the monotremes (Australia's platypus and echidna) survive; of the others, only multituberculates and morganucodonts are mentioned. Wikipedia has an entry on Prototherians (aka non-therians), but Tudge doesn't recognise this as a clade, only as a small-p colloquial term for a non-therian mammal. Wikipedia adds to the group Triconodonta and Docodonta; as you can see, the classifications are mainly based on tooth characteristics. Why? Tudge: "Most of the extinct lineages are known only from a few teeth and jaws" - prescient, as this is also the case for the SB mammal, with the addition of a bit of hip. (Braincase is also mentioned as a differentiating characteristic.) I've seen no word on the placement of this creature within non-therians (Wikipedia's entry says the SB mammal is not a monotreme) - probably not an easy task, and the discovery's only about a year old.
I remain curious, too, about the other three of Tudge's non-therian clades; however, Tudge constantly reminds us that at this level of taxonomy, judgment calls are inevitably involved.

Eutherians ('placentals') and marsupials split 100 to 125 million years ago; this specimen dates from 16 to 19 million years (see the entry in New Scientist), suggesting this creature survived for at least 100 million years. Notably, this is the first land mammal fossil found in New Zealand.

The ubiquitous Mike Archer, current UNSW Science Dean, was involved in this study. He suggests it goes against the conventional wisdom that New Zealand temporarily disappeared below the water from 25 to 30 million years ago [due to some combination of rising oceans and/or erosion] after separation from Gondwana, and re-emerged due to its location on the boundary of two techtonic plates pushing against each other. Archer reckons it was too small to succeed at rafting [so it must have remained indigenous since before the split; others disagree.
So, did New Zealand stay above water, or did the SB mammal raft?
I'm currently reading Richard Dawkins' The Ancestor's Tale (2004, Phoenix, London), and he is a great believer in the ability of rafting to explain movements. In a nutshell, rafting is the drift of land-based animals on floating vegetation masses, typically due to the effects of storms. Dawkins (p150) details a large-scale eyewitness event (iguanas in the Carribean) and makes the case that, although this may call for highly improbable events, over a long enough period of time these can come to pass.

However, New Zealand's submergence (as well as erosive forces - see the U Waikato link) does explain the lack of landbased fossils.
What then, of the Tuatara? When I was growing up, the Tuatara was represented on the five cent piece, and typified as a rare lizard - preserved due to isolation on several island just off New Zealand's mainlands. They are actually uniquely-surviving lizard-like Sphenodontians, creatures that flourished 200 million years ago. Further, they show unique cold-climate adaptions over the original warm-climate sphenodontians. That's not unexpected in the time that they've been isolated, but it suggests they were in New Zealand when it was in a warmer climate. It seems to me less likely than that they rafted as recently as 25 million years ago, subsequently disappearing in their home location.
Neither explanation is fully satisfactory. On the one hand, New Zealand entirely lost its land-based fauna. On the other hand, two uniquely survived - at least until recently, and certainly after New Zealand's purported submergence.

It remains a puzzle for the moment.

Further discussion of this unusual find includes a number of reference links.