Junk DNA is inherited. Suppose we find a pattern in the junk DNA of two different species, and don't find that pattern in other species. Evolution can explain the situation by saying that the two species recently had a common ancestor, and both species inherited this pattern from their ancestor. In short, evolution suggests the family tree:
====================== species A without the pattern | common | ancestor ======| ====== species B with the pattern species | common | of A, B and C ====ancestor===X===| species | of B and C ====== species C with the pattern
"X" marks where the pattern arose.
To the best of my knowledge, Creationism does not predict or explain such situations. Since the pattern is in the junk, one cannot argue that the pattern confers any short-term benefit or meets any need.
Of course, coincidences do happen. In the example below, I discuss how scientists checked to see if the pattern they'd found was meaningful or not.
You may have heard that the AIDS virus copies itself into its host's DNA. It can do this because there is a "reverse transcription" mechanism which leaves a so-called retroposon at some randomly chosen place in your DNA. There is a category of retroposon called a SINE (Short INterspersed Element).
A recent study found two different SINEs were present in the Hippopotamus, Cow, Sperm Whale, and Humpback Whale. Neither SINE was in the Red Kangaroo, Human, Mouse, Cat, Asiatic Elephant, Domestic Horse, Pig, or Bactrian Camel. This suggests the family tree
========================= kangaroos, mice, etc | common | ancestor ====| ===== cetaceans (whales,dolphins) | common | ======ancestor======| that had the |==== ruminants (deer, cows, sheep) two reverse | transcriptions | ===== hippopotamus
There are a number of ways to check if this is really the case. The first question is whether the patterns are just some random fluctuation of junk DNA. Randomness is ruled out for at least four different reasons.
Next, we have to ask if a retrovirus could have just infected several species. This also can be ruled out.
Next, we ask if the conclusion is believable. Could a land animal have evolved into a whale? The answer is that the land ancestry of whales is a century-old idea, and well-proven without this new line of evidence. The ancestry was originally suggested based on the fact that whales are mammals, with a placenta and live birth and mother's milk. The idea is confirmed (for example) by mitochondrial DNA similarities, by protein similarities, and by fossils of small whales with legs. Also, whales with vestigial hind legs are sometimes born.
Finally, we ask if this conclusion leads anywhere. To be science, it has to make predictions, that can be tested to see if they are right.
The answer is yes. From the standard biological taxonomy, we can predict that only artiodactyls ("even toed" mammals) have the patterns. So, we can test animals that aren't artiodactyls, like cats and iguanas. We can predict that since cows have the patterns, all ruminants should. So, we can test sheep and goats. We predict that all cetaceans have the patterns, so we can test dolphins and killer whales. Furthermore, we don't predict just that dolphins have the patterns. We also predict that dolphins have the patterns at the exact same genetic locations.
In fact, many of these predictions have already been tested, and so far they have always been correct.The Minke Whale, Baird's Beaked Whale, Dall's Porpoise, Short-Finned Pilot Whale, and Bottlenose Dolphin all had both patterns, and in exactly the predicted places. Sheep, the Reticulated Giraffe, the Axis Deer, and the Lesser Malayan Chevrotain also are as predicted.
To sum up: this evidence unambiguously says that whales and cows have a common ancestor.
The diagram below shows the conclusions of the paper. This web page has only mentioned two SINEs, which are marked as B and C on the diagram. The article talks about nine SINEs, which allow us to conclude (for instance) that the hippo is closely related to whales. Which is not a surprise, since hippos can nurse infants underwater, just like whales.
This followup paper did a wider search, and found 20 independent insertion events. Click on the diagram below to see its conclusions.
Also, it is improbable that one virus could attack so many species and yet be the same: it should have had to mutate into several variants, as AIDS has. It is improbable that two viruses would both succeed in spreading widely. It is improbable that both would attack the identical set of targets.
But more importantly, if a retrovirus had infected multiple species, it should have left itself in a different place in each species. After all, leaving a pattern in junk DNA is an accident: a virus "dies" as a result. In fact, in cows, one of the patterns has been copied to five different places. Clearly, there is no single place in the junk DNA where the pattern "should" be, or "tries" to be. So, finding a pattern in the exact same place (at the same genetic locus) in two species implies that the two species inherited the pattern. Finding one pattern at the same place in five different whales, a dolphin and a porpoise settles the matter. That locus must have been affected once, in their common ancestor.