Evolution isn't supposed to run backwards – yet an increasing number of examples show that it does and that it can sometimes represent the future of a species

进化不应该向后退的——然而，有越来越多的例子显示：它确实会后退，而且有时候这还代表着一个物种的未来

The description of any animal as an 'evolutionary throwback' is controversial. For the better part of a century, most biologists have been reluctant to use those words, mindful of a principle of evolution that says 'evolution cannot run backwards'. But as more and more examples come to light and modern genetics enters the scene, that principle is having to be rewritten. Not only are evolutionary throwbacks possible, they sometimes play an important role in the forward march of evolution.

将任何动物描述为“进化倒退” 的说法都是有争议的。大半个世纪以来，大多数生物学家都不太情愿使用这些词语，因为他们头脑中有这样一条进化原则，认为“进化不能倒退”。但是随着越来越多的例子出现以及现代基因学的诞生，那条原则正不得不被重写。进化倒退不但是可能的，而且有时候还在进化的向前发展中扮演着重要角色。

用来描述进化倒退的技术性术语是“返祖现象”，它来自拉丁语词汇atavus,意思是“祖先”。这个词的暗示含义很不好，这在很大程度上要拜一位19世纪的意大利医生Cesare Lombroso所赐，他认为犯罪在出生的时候就并未发育完整，可以通过某些生理特征来辨认他们，而这些特征就是向原始的、低于人类级别状态的退化。

The technical term for an evolutionary throwback is an 'atavism', from the Latin atavus, meaning forefather. The word has ugly connotation thanks largely to Cesare Lombroso, a 19th-century Italian medic who argued that criminals were born not made and could be identified by certain physical features that were throwbacks to primitive, sub-human state.

就在Lombroso忙于评估罪犯的同时，一位名为Louis Dollo的比利时古生物学家正在研究化石记录并得出了相反的结论。他在1890年提出，进化是不可逆的：“一个有机体不可能返回到其祖先们已经实现过的进化各阶段中的任何一个先前的状态，甚至部分地返回也不可能。”20世纪早期生物学家也得出了类似的结论，不过他们是以可能性高低的方式来进一步说明这一规律的，指出没有什么理由为什么进化一定不能倒退——只不过这种情况不太可能发生罢了。就这样，进化不可逆的理论便延续了下来，逐渐被称为“Dollo法则”。

如果Dollo法则正确无误的话，那么返祖现象即使会出现，也应当只是罕见的例外而已。然而几乎就是在这一理论开始扎根的同时，反例就一直层出不穷。比如，1919年在加拿大温哥华岛外围捕捞到一头座头鲸，它长着一对足有一米多长的像腿般的附肢，且其中的整套腿骨系统都是发育完全的。探索家Roy Chapman Andrews当时认为，那头鲸鱼一定是退化成了其陆地存活祖先的某个状态。“我想不到任何其他解释，”他在1921年写道。

While Lombroso was measuring criminals, a Belgian palaeontologist called Louis Dollo was studying fossil records and coming to the opposite conclusion. In 1890 he proposed that evolution was irreversible: that 'an organism is unable to return, even partially, to a previous stage already realized in the ranks of its ancestors'. Early 20th-century biologists came to a similar conclusion, though they qualified it in terms of probability, stating that there is no reason why evolution cannot run backwards – it is just very unlikely. And so the idea of irreversibility in evolution stuck and came to be known as 'Dollo's law'.

从那时起，还发现了如此众多的其他案例，认为进化是不可逆的理论再也没意义了。而这又提出了一个谜题：那些几百万年前便已消失的特点是如何又突然重新出现的呢？在1994年，美国印第安纳大学的Rudolf Raff及其同事们决定利用遗传学为进化返祖现象所可能发生的时限确定一个具体的数字。他们是这样分析的：虽然一些进化改变涉及基因的丢失，因此是不可逆的，然而另外一些改变也许是由于基因作用被关掉了。他们认为，如果这些沉默的基因通过某种方式被重新开启，那些消失已久的特征就有可能重现。

Raff的团队进而计算了这种情况发生的可能性。他们如此认为：沉默基因会积累随机性的突变，最终使得它们自身变得毫无用处。因此，如果一份基因不再被使用，它能在这个物种中存活多久？此团队分析，沉默基因很有可能在每个群体中的至少几个个体身上存在至多达六百万年之久，有一些甚至可能延续一千万年那么长。换言之，退化是可能发生的，但是仅仅只能退化到某个距今相对较近的先前进化阶段。

If Dollo's law is right, atavisms should occur only very rarely, if at all. Yet almost since the idea took root, exceptions have been cropping up. In 1919, for example, a humpback whale with a pair of leg-like appendages over a metre long, complete with a full set of limb bones, was caught off Vancouver Island in Canada. Explorer Roy Chapman Andrews argued at the time that the whale must be a throwback to a land-living ancestor. 'I can see no other explanation,' he wrote in 1921.

该团队列举了墨西哥和加利福尼亚的钝口螈属为例证。与大多数两栖动物类似，这些生物以一种年幼的“蝌蚪”形态开始自己的一生，然后变形为成年的形态——但有一个品种例外：蝾螈，从所周知它的一生都保持了幼年形态。对此最简单的解释便是：只有蝾螈这一支系丧失了变形的能力，而其他品种则保留着这种能力。然而，从对蝾螈目动物族谱的深入细致分析中可知，很显然其他支系所从之进化而来的祖先本身当时就已失去了变形的能力。换句话说，钝口螈属的变形行为是一种返祖现象。这个蝾螈目例证正好符合Raff的一千万年时限理论。

但是，更近些年来，一些例子的出现打破了这个时间限制，从而表明沉默基因也许不能解释这种现象。在去年发表的一篇论文中，耶鲁大学生物学家Gunter Wagner公布了自己对于一种名为Bachia的南美洲蜥蜴的进化历史所做的一些研究。这类蜥蜴中的许多个体都长着微小的四肢；有一些看起来不像蜥蜴而更像蛇，还有一些后腿上已经完全没有了足趾。然而，其他蜥蜴目物种却在后腿上进化出了最高可达四个的足趾。最简单的解释就是：有趾的这些支系其足趾从未退化丢失过，但Wagner持有不同看法。根据他对Bachia蜥蜴家族的分析，那些有趾物种是从无趾的祖先哪里重新进化出了爪趾，而且更有甚者，这些爪趾的减少和增加在数千万年的时间里发生过不止一次。

Since then, so many other examples have been discovered that it no longer makes sense to say that evolution is as good as irreversible. And this poses a puzzle: how can characteristics that disappeared millions of years ago suddenly reappear? In 1994, Rudolf Raff and colleagues at Indiana University in the USA decided to use genetics to put a number on the probability of evolution going into reverse. They reasoned that while some evolutionary changes involve the loss of genes and are therefore irreversible, others may be the result of genes being switched off. If these silent genes are somehow switched back on, they argued, long-lost traits could reappear.

那么到底是怎么回事呢？可能性之一是：这些特征先是丧失，然后仅仅就是又重新出现了而已，就好像彼此完全无关的物种身上会独立地各自出现相似的结构，例如鲨鱼和虎鲸的背鳍。另一种更加引人深究的可能性则是：那些产生足趾所需要的基因信息由于某种原因在蝾螈身体中存在了几千万年乃至数亿年的时光又被重新激活了。这些返祖特征提供了一项优势并在群中散播开来，有效地扭转了进化的进程。

但是如果沉默基因会在六百万到一千万年的时间内退化，那些早已消失很久的特征是如何经历了更长的时期而被重新激活呢？答案也许在于子宫。许多物种的早期胚胎都发展出了其祖先的性状。例如，蛇的胚胎会伸出后腿的突起物。在后来的发展中，这些性状又消失了，多亏其发育机制说：“丢掉那些腿”。如果出于某种原因这个机制没有起作用，祖先的特征就可能不会消失，导致个体出现返祖现象。

Raff's team went on to calculate the likelihood of it happening. Silent genes accumulate random mutations, they reasoned, eventually rendering them useless. So how long can a gene survive in a species if it is no longer used? The team calculated that there is a good chance of silent genes surviving for up to 6 million years in at least a few individuals in a population, and that some might survive as long as 10 million years. In other words, throwbacks are possible, but only to the relatively recent evolutionary past.

As a possible example, the team pointed to the mole salamanders of Mexico and California. Like most amphibians these begin life in a juvenile 'tadpole' state, then metamorphose into the adult form – except for one species, the axolotl, which famously lives its entire life as a juvenile. The simplest explanation for this is that the axolotl lineage alone lost the ability to metamorphose, while others retained it. From a detailed analysis of the salamanders' family tree, however, it is clear that the other lineages evolved from an ancestor that itself had lost the ability to metamorphose. In other words, metamorphosis in mole salamanders is an atavism. The salamander example fits with Raff's 10-million-year time frame.

More recently, however, examples have been reported that break the time limit, suggesting that silent genes may not be the whole story. In a paper published last year, biologist Gunter Wagner of Yale University reported some work on the evolutionary history of a group of South American lizards called Bachia. Many of these have minuscule limbs; some look more like snakes than lizards and a few have completely lost the toes on their hind limbs. Other species, however, sport up to four toes on their hind legs. The simplest explanation is that the toed lineages never lost their toes, but Wagner begs to differ. According to his analysis of the Bachia family tree, the toed species re-evolved toes from toeless ancestors and, what is more, digit loss and gain has occurred on more than one occasion over tens of millions of years.

So what's going on? One possibility is that these traits are lost and then simply reappear, in much the same way that similar structures can independently arise in unrelated species, such as the dorsal fins of sharks and killer whales. Another more intriguing possibility is that the genetic information needed to make toes somehow survived for tens or perhaps hundreds of millions of years in the lizards and was reactivated. These atavistic traits provided an advantage and spread through the population, effectively reversing evolution.

But is silent genes degrade within 6 to 10 million years, how can long-lost traits be reactivated over longer timescales? The answer may lie in the womb. Early embryos of many species develop ancestral features. Snake embryos, for example, sprout hind limb buds. Later in development these features disappear thanks to developmental programs that say 'lose the leg'. If for any reason this does not happen, the ancestral feature may not disappear, leading to an atavism.