Phase 2:

Radiation of fast wading hominids (river apes) through adaptation to riverside gallery forest environment. Danakil sea flooding isolates a small population of bipedal apes on island habitats.

Key:

Swamp Apes - African Ape / Homo ancestor - Orrorin tugenensis  

River Apes - Ardipithecus/Australopithecus ancestor. Speciation of Australopithecus afarensis 

Orang-utan ancestor 

Pros:

• Offers a plausible habitat for the Australopithecine radiation.

• Fills the fossil gap for Gorilla and  Pan ancestors. (Absolutely no fossils have been attributed to gorillas or chimpanzees so far)

• Danakil flooding and subsequent isolation offers explanation for swimming/diving ability and also how 'baboon marker' was avoided in Homo ancestors.

• Offers plausible functional explanation (fast wading) of Australopithecine femoral-pelvic complex.

• Explains Australopithecine dentition and fossil finds.

Cons:

• Rather complex model, assuming several sub-species (although not speciated sufficiently to prevent subsequent hybridizations).
• Not clear on phylogeny of Homo/Pan. Was Australopithecus ancestral to Homo or Pan or both? This model is not clear on that. It rather sits on the fence on that subject, suggesting that A. afarensis may have been ancestral to Pan but that some of the resulting radiation from that group later bred back into the Homo group, which had been isolated on Danakil.

Phase II Story-line
Most of the coastal mangrove forest climbing-wading apes which were ancestral to all African apes/Homo have become extinct due to inter-glacial flooding and glacial desiccation cycles but at least one group survives in the coast around modern day Danakil.
The ancestors of the lineage leading to Ardipithecus ramidus and then, on the same line, to all the Australopithecines  becomes isolated in pockets of tropical rainforest dominated by river systems, rather like the Bonobo (Pan Paniscus) has today, only more so. The forests are smaller and more concentrated around rivers than the modern Congo and so, through the occasional need to wade across the water, natural selection gradually selects out those members whose traits do not allow them to wade quickly and efficiently. Already fairly good bipeds in water, these 'river apes' are now on the way to becoming specialist fast waders.
The lineage leading to Australopithecus and through them to Gorilla and Pan and possibly part of Homo become adapted to a gallery forest environment prone to flooding.
This model does not require that they spent very long in the water but that they crossed rivers frequently enough for natural selection to favour those that could do so better. A successful species quickly filled the marginal freshwater-side niches of rivers and lakes in the rift valley.
A radiation begins leading to a migration inland of fast wading  'river apes', which become more adapted to the water-side food chain.

An increase in essential fatty acids in their diet begins a change in their body chemistry, including increased fat storage and improved enamel dentition as described in Horrobin (2002) (pp 283-287.)

There are at least two proposed waves of migration away from the coastal/mangrove habitat.
The first from around 7 mya, included the such paleospecies as Sahelanthropus tchadensis (into around the 'Mega Lake Chad') and Orrorin tugenensis (into the Rift valley itself) and is postulated to represent ancestors to Gorilla perhaps through those paleospecies currently classified as robust australopithecines.
The second, included the gracile australopithecines, which are suggested in this phase as being ancestral to Pan. The model proposes that as Africa (in later phases) became drier these 'river ape' chimpanzee ancestors were increasingly out-competed from the marginal habitats by Homo newcomers, forcing them to adapt to drier conditions.

Why did the ape cross the river?
The model predicts that there would have been many reasons for these 'river apes' getting into the water:
1. Increased foraging range;
2. Access to fresh-water shellfish and other foods rich in micro-algae (and omega 3 fatty acid food chain);
3. Access to water-side fruit and other food sources;
4. To move from one clump of trees to another during floods;
5. To see further up or down stream;
6. To have sex with members of groups of river apes on the other side;
7. To escape terrestrial predators, like leopards and
8. Last, by by no means least, simply to keep cool.
 

Even if an individual only crossed the river once a week, it would still expose a young river ape to hundreds of wading events before reaching adulthood. With the danger of crocodile and other predation, as well as flash floods, the model predicts that this would be more than sufficient selective pressure for better wading to have resulted, enough to change their skeletal morphology to the one we see in the fossil record.

Selection for swimming is also likely to have occurred during this phase but it is assumed that wading came first, simply because there is no good evidence that Australopithecines swam.

Fossil Evidence
Most, if not all, the fossil evidence of the earliest bipeds is found in waterside habitats and as Horrobin (2002) describes (p 284) the evidence of the food they probably ate. "One important piece of evidence which suggests that our Australopithecine ancestors ate EFA-rich aqueous foods in the change in their tool enamel. The enamel coating of the teeth becomes much thicker than in most primates."
Much of the fossil evidence from Australopithecus afarensis suggests that the animal exhibited a very different kind of bipedalism to ours. The "bent-hip, bent-knee" gait proposed by some has been criticised because it would not be energetically efficient. But those authors all assume that they were bipedal on land. In water, aided by their natural buoyancy, bent-hip bent-knee would be exactly the kind of bipedalism predicted in a wading environment.
The strong abduction/adduction of the femur and very platypelloid pelvis of A. afarensis is compelling evidence of an adaptation to wading (perhaps with sideways gait) in order to reduce drag forces in water. See my paper on wading bipedalism for more details.

Danakil Flooding
An important part of the model is based on the fact that at around 4 mya a rise in sea level caused the Danakil depression to become flooded with sea water. Like the Baltic Sea today, this would have been a niche where the water was not too salty and was, in fact drinkable due to extensive leaching from the Rift Valley rivers.
This phase assumes that the radiation of 'river apes' included a population of hominids that became highly adapted to this estuarine environment and another that became isolated out in a much more marine habitat on the Danakil island(s).
This marine group of hominids would, it is proposed, started to become much more salt tolerant as they began to take advantage of food though coastal foraging. First through wading, then swimming and diving.
Also, isolated on Danakil, they avoided predation as the population of big cat predators their quickly died out through lack of food. This meant that they had less reason to sleep in trees and quickly became less arboreal.
The Danakil hominids also avoided contact with a virus which affected all the African apes and baboons. The model assumes that the reason the so-called baboon marker is not found in Homo is because a significant part of our ancestry did not live on the African mainland during this time. ( See Todaro et al 1974, 1976 & especially 1980 for more on this)      
The bulk of the river ape population, however, were not affected by this flooding and continued to thrive in their gallery forest habitat. They were very successful and the radiation continued, seeing them expand out of Afar, south along the Rift Valley lakes and rivers.
 

Predictions:
If this phase of the AHAH model is correct a number of observations/discoveries should be made in the future.

Specifically:

• The (now) Danakil alps should reveal fossil finds of hominid ancestors that appear 'advanced' in evolutionary terms compared to the contemporaries on the mainland.

• Human genome analysis should reveal that the 'baboon marker', present in all African apes, could have been replaced by a hybridisation event with another group of hominids which were not exposed to such a marker.

• Fossils placed ancestrally to Pan and Gorilla should be found in Africa dated at around 2-1mya showing an intermediate stage between wading-bipedalism and their present day knuckle-walking.

Specifically:

• A cast-iron model for how human ancestors' chromosome numbers changed from 48 to 46 will be published which does not involve hybridisation. This would mean that the hybrid part of the model would have to be replaced.

• If an ancestor to Homo was identified outside of Africa, perhaps on the Indian ocean coast, dated earlier than the earliest African putative hominid this would require a re-working of the whole AHAH model..

The phase ends as the radiation (Out of Afar I) matures.

The next phase describes the different resulting sub-species of 'river ape' in detail and how they diversified into a wide range of habitats.