Aug 20th 2020

IThe outsiders inside

HUMANS ARE lucky to live a hundred years. Oak trees may live a thousand; mayflies, in their adult form, a single day. But they are all alive in the same way. They are made up of cells which embody flows of energy and stores of information. Their metabolisms make use of that energy, be it from sunlight or food, to build new molecules and break down old ones, using mechanisms described in the genes they inherited and may, or may not, pass on.

It is this endlessly repeated, never quite perfect reproduction which explains why oak trees, humans, and every other plant, fungus or single-celled organism you have ever seen or felt the presence of are all alive in the same way. It is the most fundamental of all family resemblances. Go far enough up any creatures family tree and you will find an ancestor that sits in your family tree, too. Travel further and you will find what scientists call the last universal common ancestor, LUCA. It was not the first living thing. But it was the one which set the template for the life that exists today.

And then there are viruses. In viruses the link between metabolism and genes that binds together all life to which you are related, from bacteria to blue whales, is broken. Viral genes have no cells, no bodies, no metabolism of their own. The tiny particles, virions, in which those genes come packagedthe dot-studded disks of coronaviruses, the sinister, sinuous windings of Ebola, the bacteriophages with their science-fiction landing-legs that prey on microbesare entirely inanimate. An individual animal, or plant, embodies and maintains the restless metabolism that made it. A virion is just an arrangement of matter.

The virus is not the virion. The virus is a process, not a thing. It is truly alive only in the cells of others, a virtual organism running on borrowed hardware to produce more copies of its genome. Some bide their time, letting the cell they share the life of live on. Others immediately set about producing enough virions to split their hosts from stem to stern.

The virus has no plan or desire. The simplest purposes of the simplest lifeto maintain the difference between what is inside the cell and what is outside, to move towards one chemical or away from anotherare entirely beyond it. It copies itself in whatever way it does simply because it has copied itself that way before, in other cells, in other hosts.

That is why, asked whether viruses are alive, Eckard Wimmer, a chemist and biologist who works at the State University of New York, Stony Brook, offers a yes-and-no. Viruses, he says, alternate between nonliving and living phases. He should know. In 2002 he became the first person in the world to take an array of nonliving chemicals and build a virion from scratcha virion which was then able to get itself reproduced by infecting cells.

The fact that viruses have only a tenuous claim to being alive, though, hardly reduces their impact on things which are indubitably so. No other biological entities are as ubiquitous, and few as consequential. The number of copies of their genes to be found on Earth is beyond astronomical. There are hundreds of billions of stars in the Milky Way galaxy and a couple of trillion galaxies in the observable universe. The virions in the surface waters of any smallish sea handily outnumber all the stars in all the skies that science could ever speak of.

Back on Earth, viruses kill more living things than any other type of predator. They shape the balance of species in ecosystems ranging from those of the open ocean to that of the human bowel. They spur evolution, driving natural selection and allowing the swapping of genes.

They may have been responsible for some of the most important events in the history of life, from the appearance of complex multicellular organisms to the emergence of DNA as a preferred genetic material. The legacy they have left in the human genome helps produce placentas and may shape the development of the brain. For scientists seeking to understand lifes origin, they offer a route into the past separate from the one mapped by humans, oak trees and their kin. For scientists wanting to reprogram cells and mend metabolisms they offer inspirationand powerful tools.

IIA lifestyle for genes

THE IDEA of a last universal common ancestor provides a plausible and helpful, if incomplete, answer to where humans, oak trees and their ilk come from. There is no such answer for viruses. Being a virus is not something which provides you with a place in a vast, coherent family tree. It is more like a lifestylea way of being which different genes have discovered independently at different times. Some viral lineages seem to have begun quite recently. Others have roots that comfortably predate LUCA itself.

Disparate origins are matched by disparate architectures for information storage and retrieval. In eukaryotescreatures, like humans, mushrooms and kelp, with complex cellsas in their simpler relatives, the bacteria and archaea, the genes that describe proteins are written in double-stranded DNA. When a particular protein is to be made, the DNA sequence of the relevant gene acts as a template for the creation of a complementary molecule made from another nucleic acid, RNA. This messenger RNA (mRNA) is what the cellular machinery tasked with translating genetic information into proteins uses in order to do so.

Because they, too, need to have proteins made to their specifications, viruses also need to produce mRNAs. But they are not restricted to using double-stranded DNA as a template. Viruses store their genes in a number of different ways, all of which require a different mechanism to produce mRNAs. In the early 1970s David Baltimore, one of the great figures of molecular biology, used these different approaches to divide the realm of viruses into seven separate classes (see diagram).

In four of these seven classes the viruses store their genes not in DNA but in RNA. Those of Baltimore group three use double strands of RNA. In Baltimore groups four and five the RNA is single-stranded; in group four the genome can be used directly as an mRNA; in group five it is the template from which mRNA must be made. In group sixthe retroviruses, which include HIVthe viral RNA is copied into DNA, which then provides a template for mRNAs.

Because uninfected cells only ever make RNA on the basis of a DNA template, RNA-based viruses need distinctive molecular mechanisms those cells lack. Those mechanisms provide medicine with targets for antiviral attacks. Many drugs against HIV take aim at the system that makes DNA copies of RNA templates. Remdesivir (Veklury), a drug which stymies the mechanism that the simpler RNA viruses use to recreate their RNA genomes, was originally developed to treat hepatitis C (group four) and subsequently tried against the Ebola virus (group five). It is now being used against SARS-CoV-2 (group four), the covid-19 virus.

Studies of the gene for that RNA-copying mechanism, RdRp, reveal just how confusing virus genealogy can be. Some viruses in groups three, four and five seem, on the basis of their RdRp-gene sequence, more closely related to members of one of the other groups than they are to all the other members of their own group. This may mean that quite closely related viruses can differ in the way they store their genomes; it may mean that the viruses concerned have swapped their RdRp genes. When two viruses infect the same cell at the same time such swaps are more or less compulsory. They are, among other things, one of the mechanisms by which viruses native to one species become able to infect another.

How do genes take on the viral lifestyle in the first place? There are two plausible mechanisms. Previously free-living creatures could give up metabolising and become parasitic, using other creatures cells as their reproductive stage. Alternatively genes allowed a certain amount of independence within one creature could have evolved the means to get into other creatures.

Living creatures contain various apparently independent bits of nucleic acid with an interest in reproducing themselves. The smallest, found exclusively in plants, are tiny rings of RNA called viroids, just a few hundred genetic letters long. Viroids replicate by hijacking a host enzyme that normally makes mRNAs. Once attached to a viroid ring, the enzyme whizzes round and round it, unable to stop, turning out a new copy of the viroid with each lap.

Viroids describe no proteins and do no good. Plasmidssomewhat larger loops of nucleic acid found in bacteriado contain genes, and the proteins they describe can be useful to their hosts. Plasmids are sometimes, therefore, regarded as detached parts of a bacterias genome. But that detachment provides a degree of autonomy. Plasmids can migrate between bacterial cells, not always of the same species. When they do so they can take genetic traits such as antibiotic resistance from their old host to their new one.

Recently, some plasmids have been implicated in what looks like a progression to true virus-hood. A genetic analysis by Mart Krupovic of the Pasteur Institute suggests that the Circular Rep-Encoding Single-Strand-DNA (CRESS-DNA) viruses, which infect bacteria, evolved from plasmids. He thinks that a DNA copy of the genes that another virus uses to create its virions, copied into a plasmid by chance, provided it with a way out of the cell. The analysis strongly suggests that CRESS-DNA viruses, previously seen as a pretty closely related group, have arisen from plasmids this way on three different occasions.

Such jailbreaks have probably been going on since very early on in the history of life. As soon as they began to metabolise, the first proto-organisms would have constituted a niche in which other parasitic creatures could have lived. And biology abhors a vacuum. No niche goes unfilled if it is fillable.

It is widely believed that much of the evolutionary period between the origin of life and the advent of LUCA was spent in an RNA worldone in which that versatile substance both stored information, as DNA now does, and catalysed chemical reactions, as proteins now do. Set alongside the fact that some viruses use RNA as a storage medium today, this strongly suggests that the first to adopt the viral lifestyle did so too. Patrick Forterre, an evolutionary biologist at the Pasteur Institute with a particular interest in viruses (and the man who first popularised the term LUCA) thinks that the RNA world was not just rife with viruses. He also thinks they may have brought about its end.

The difference between DNA and RNA is not large: just a small change to one of the letters used to store genetic information and a minor modification to the backbone to which these letters are stuck. And DNA is a more stable molecule in which to store lots of information. But that is in part because DNA is inert. An RNA-world organism which rewrote its genes into DNA would cripple its metabolism, because to do so would be to lose the catalytic properties its RNA provided.

An RNA-world virus, having no metabolism of its own to undermine, would have had no such constraints if shifting to DNA offered an advantage. Dr Forterre suggests that this advantage may have lain in DNAs imperviousness to attack. Host organisms today have all sorts of mechanisms for cutting up viral nucleic acids they dont like the look ofmechanisms which biotechnologists have been borrowing since the 1970s, most recently in the form of tools based on a bacterial defence called CRISPR. There is no reason to imagine that the RNA-world predecessors of todays cells did not have similar shears at their disposal. And a virus that made the leap to DNA would have been impervious to their blades.

Genes and the mechanisms they describe pass between viruses and hosts, as between viruses and viruses, all the time. Once some viruses had evolved ways of writing and copying DNA, their hosts would have been able to purloin them in order to make back-up copies of their RNA molecules. And so what began as a way of protecting viral genomes would have become the way life stores all its genesexcept for those of some recalcitrant, contrary viruses.

IIIThe scythes of the seas

IT IS A general principle in biology that, although in terms of individual numbers herbivores outnumber carnivores, in terms of the number of species carnivores outnumber herbivores. Viruses, however, outnumber everything else in every way possible.

This makes sense. Though viruses can induce host behaviours that help them spreadsuch as coughingan inert virion boasts no behaviour of its own that helps it stalk its prey. It infects only that which it comes into contact with. This is a clear invitation to flood the zone. In 1999 Roger Hendrix, a virologist, suggested that a good rule of thumb might be ten virions for every living individual creature (the overwhelming majority of which are single-celled bacteria and archaea). Estimates of the number of such creatures on the planet come out in the region of 1029-1030. If the whole Earth were broken up into pebbles, and each of those pebbles smashed into tens of thousands of specks of grit, you would still have fewer pieces of grit than the world has virions. Measurements, as opposed to estimates, produce numbers almost as arresting. A litre of seawater may contain more than 100bn virions; a kilogram of dried soil perhaps a trillion.

Metagenomics, a part of biology that looks at all the nucleic acid in a given sample to get a sense of the range of life forms within it, reveals that these tiny throngs are highly diverse. A metagenomic analysis of two surveys of ocean life, the Tara Oceans and Malaspina missions, by Ahmed Zayed of Ohio State University, found evidence of 200,000 different species of virus. These diverse species play an enormous role in the ecology of the oceans.

A litre of seawater may contain 100bn virions; a kilogram of dried soil perhaps a trillion

On land, most of the photosynthesis which provides the biomass and energy needed for life takes place in plants. In the oceans, it is overwhelmingly the business of various sorts of bacteria and algae collectively known as phytoplankton. These creatures reproduce at a terrific rate, and viruses kill them at a terrific rate, too. According to work by Curtis Suttle of the University of British Columbia, bacterial phytoplankton typically last less than a week before being killed by viruses.

This increases the overall productivity of the oceans by helping bacteria recycle organic matter (it is easier for one cell to use the contents of another if a virus helpfully lets them free). It also goes some way towards explaining what the great mid-20th-century ecologist G. Evelyn Hutchinson called the paradox of the plankton. Given the limited nature of the resources that single-celled plankton need, you would expect a few species particularly well adapted to their use to dominate the ecosystem. Instead, the plankton display great variety. This may well be because whenever a particular form of plankton becomes dominant, its viruses expand with it, gnawing away at its comparative success.

It is also possible that this endless dance of death between viruses and microbes sets the stage for one of evolutions great leaps forward. Many forms of single-celled plankton have molecular mechanisms that allow them to kill themselves. They are presumably used when one cells sacrifice allows its sister cellswhich are genetically identicalto survive. One circumstance in which such sacrifice seems to make sense is when a cell is attacked by a virus. If the infected cell can kill itself quickly (a process called apoptosis) it can limit the number of virions the virus is able to make. This lessens the chances that other related cells nearby will die. Some bacteria have been shown to use this strategy; many other microbes are suspected of it.

There is another situation where self-sacrifice is becoming conduct for a cell: when it is part of a multicellular organism. As such organisms grow, cells that were once useful to them become redundant; they have to be got rid of. Eugene Koonin of Americas National Institutes of Health and his colleagues have explored the idea that virus-thwarting self-sacrifice and complexity-permitting self-sacrifice may be related, with the latter descended from the former. Dr Koonins model also suggests that the closer the cells are clustered together, the more likely this act of self-sacrifice is to have beneficial consequences.

For such profound propinquity, move from the free-flowing oceans to the more structured world of soil, where potential self-sacrificers can nestle next to each other. Its structure makes soil harder to sift for genes than water is. But last year Mary Firestone of the University of California, Berkeley, and her colleagues used metagenomics to count 3,884 new viral species in a patch of Californian grassland. That is undoubtedly an underestimate of the total diversity; their technique could see only viruses with RNA genomes, thus missing, among other things, most bacteriophages.

Metagenomics can also be applied to biological samples, such as bat guano in which it picks up viruses from both the bats and their food. But for the most part the finding of animal viruses requires more specific sampling. Over the course of the 2010s PREDICT, an American-government project aimed at finding animal viruses, gathered over 160,000 animal and human tissue samples from 35 countries and discovered 949 novel viruses.

The people who put together PREDICT now have grander plans. They want a Global Virome Project to track down all the viruses native to the worlds 7,400 species of mammals and waterfowlthe reservoirs most likely to harbour viruses capable of making the leap into human beings. In accordance with the more-predator-species-than-prey rule they expect such an effort would find about 1.5m viruses, of which around 700,000 might be able to infect humans. A planning meeting in 2018 suggested that such an undertaking might take ten years and cost $4bn. It looked like a lot of money then. Today those arguing for a system that can provide advance warning of the next pandemic make it sound pretty cheap.

IVLeaving their mark

THE TOLL which viruses have exacted throughout history suggests that they have left their mark on the human genome: things that kill people off in large numbers are powerful agents of natural selection. In 2016 David Enard, then at Stanford University and now at the University of Arizona, made a stab at showing just how much of the genome had been thus affected.

He and his colleagues started by identifying almost 10,000 proteins that seemed to be produced in all the mammals that had had their genomes sequenced up to that point. They then made a painstaking search of the scientific literature looking for proteins that had been shown to interact with viruses in some way or other. About 1,300 of the 10,000 turned up. About one in five of these proteins was connected to the immune system, and thus could be seen as having a professional interest in viral interaction. The others appeared to be proteins which the virus made use of in its attack on the host. The two cell-surface proteins that SARS-CoV-2 uses to make contact with its target cells and inveigle its way into them would fit into this category.

The researchers then compared the human versions of the genes for their 10,000 proteins with those in other mammals, and applied a statistical technique that distinguishes changes that have no real impact from the sort of changes which natural selection finds helpful and thus tries to keep. Genes for virus-associated proteins turned out to be evolutionary hotspots: 30% of all the adaptive change was seen in the genes for the 13% of the proteins which interacted with viruses. As quickly as viruses learn to recognise and subvert such proteins, hosts must learn to modify them.

A couple of years later, working with Dmitri Petrov at Stanford, Dr Enard showed that modern humans have borrowed some of these evolutionary responses to viruses from their nearest relatives. Around 2-3% of the DNA in an average European genome has Neanderthal origins, a result of interbreeding 50,000 to 30,000 years ago. For these genes to have persisted they must be doing something usefulotherwise natural selection would have removed them. Dr Enard and Dr Petrov found that a disproportionate number described virus-interacting proteins; of the bequests humans received from their now vanished relatives, ways to stay ahead of viruses seem to have been among the most important.

Viruses do not just shape the human genome through natural selection, though. They also insert themselves into it. At least a twelfth of the DNA in the human genome is derived from viruses; by some measures the total could be as high as a quarter.

Retroviruses like HIV are called retro because they do things backwards. Where cellular organisms make their RNA from DNA templates, retroviruses do the reverse, making DNA copies of their RNA genomes. The host cell obligingly makes these copies into double-stranded DNA which can be stitched into its own genome. If this happens in a cell destined to give rise to eggs or sperm, the viral genes are passed from parent to offspring, and on down the generations. Such integrated viral sequences, known as endogenous retroviruses (ERVs), account for 8% of the human genome.

This is another example of the way the same viral trick can be discovered a number of times. Many bacteriophages are also able to stitch copies of their genome into their hosts DNA, staying dormant, or temperate, for generations. If the cell is doing well and reproducing regularly, this quiescence is a good way for the viral genes to make more copies of themselves. When a virus senses that its easy ride may be coming to an end, thoughfor example, if the cell it is in shows signs of stressit will abandon ship. What was latent becomes lytic as the viral genes produce a sufficient number of virions to tear the host apart.

Though some of their genes are associated with cancers, in humans ERVs do not burst back into action in later generations. Instead they have proved useful resources of genetic novelty. In the most celebrated example, at least ten different mammalian lineages make use of a retroviral gene for one of their most distinctively mammalian activities: building a placenta.

The placenta is a unique organ because it requires cells from the mother and the fetus to work together in order to pass oxygen and sustenance in one direction and carbon dioxide and waste in the other. One way this intimacy is achieved safely is through the creation of a tissue in which the membranes between cells are broken down to form a continuous sheet of cellular material.

The protein that allows new cells to merge themselves with this layer, syncytin-1, was originally used by retroviruses to join the external membranes of their virions to the external membranes of cells, thus gaining entry for the viral proteins and nucleic acids. Not only have different sorts of mammals co-opted this membrane-merging trickother creatures have made use of it, too. The mabuya, a long-tailed skink which unusually for a lizard nurtures its young within its body, employs a retroviral syncytin protein to produce a mammalian-looking placenta. The most recent shared ancestor of mabuyas and mammals died out 80m years before the first dinosaur saw the light of day, but both have found the same way to make use of the viral gene.

This is not the only way that animals make use of their ERVs. Evidence has begun to accumulate that genetic sequences derived from ERVs are quite frequently used to regulate the activity of genes of more conventional origin. In particular, RNA molecules transcribed from an ERV called HERV-K play a crucial role in providing the stem cells found in embryos with their pluripotencythe ability to create specialised daughter cells of various different types. Unfortunately, when expressed in adults HERV-K can also be responsible for cancers of the testes.

As well as containing lots of semi-decrepit retroviruses that can be stripped for parts, the human genome also holds a great many copies of a retrotransposon called LINE-1. This a piece of DNA with a surprisingly virus-like way of life; it is thought by some biologists to have, like ERVs, a viral origin. In its full form, LINE-1 is a 6,000-letter sequence of DNA which describes a reverse transcriptase of the sort that retroviruses use to make DNA from their RNA genomes. When LINE-1 is transcribed into an mRNA and that mRNA subsequently translated to make proteins, the reverse transcriptase thus created immediately sets to work on the mRNA used to create it, using it as the template for a new piece of DNA which is then inserted back into the genome. That new piece of DNA is in principle identical to the piece that acted as the mRNAs original template. The LINE-1 element has made a copy of itself.

In the 100m years or so that this has been going on in humans and the species from which they are descended the LINE-1 element has managed to pepper the genome with a staggering 500,000 copies of itself. All told, 17% of the human genome is taken up by these copiestwice as much as by the ERVs.

Most of the copies are severely truncated and incapable of copying themselves further. But some still have the knack, and this capability may be being put to good use. Fred Gage and his colleagues at the Salk Institute for Biological Studies, in San Diego, argue that LINE-1 elements have an important role in the development of the brain. In 2005 Dr Gage discovered that in mouse embryosspecifically, in the brains of those embryosabout 3,000 LINE-1 elements are still able to operate as retrotransposons, putting new copies of themselves into the genome of a cell and thus of all its descendants.

Brains develop through proliferation followed by pruning. First, nerve cells multiply pell-mell; then the cell-suicide process that makes complex life possible prunes them back in a way that looks a lot like natural selection. Dr Gage suspects that the movement of LINE-1 transposons provides the variety in the cell population needed for this selection process. Choosing between cells with LINE-1 in different places, he thinks, could be a key part of the process from which the eventual neural architecture emerges. What is true in mice is, as he showed in 2009, true in humans, too. He is currently developing a technique for looking at the process in detail by comparing, post mortem, the genomes of different brain cells from single individuals to see if their LINE-1 patterns vary in the ways that his theory would predict.

VPromised lands

HUMAN EVOLUTION may have used viral genes to make big-brained live-born life possible; but viral evolution has used them to kill off those big brains on a scale that is easily forgotten. Compare the toll to that of war. In the 20th century, the bloodiest in human history, somewhere between 100m and 200m people died as a result of warfare. The number killed by measles was somewhere in the same range; the number who died of influenza probably towards the top of it; and the number killed by smallpox300m-500mwell beyond it. That is why the eradication of smallpox from the wild, achieved in 1979 by a globally co-ordinated set of vaccination campaigns, stands as one of the all-time-great humanitarian triumphs.

Other eradications should eventually follow. Even in their absence, vaccination has led to a steep decline in viral deaths. But viruses against which there is no vaccine, either because they are very new, like SARS-CoV-2, or peculiarly sneaky, like HIV, can still kill millions.

Reducing those tolls is a vital aim both for research and for public-health policy. Understandably, a far lower priority is put on the benefits that viruses can bring. This is mostly because they are as yet much less dramatic. They are also much less well understood.

The viruses most prevalent in the human body are not those which infect human cells. They are those which infect the bacteria that live on the bodys surfaces, internal and external. The average human microbiome harbours perhaps 100trn of these bacteria. And where there are bacteria, there are bacteriophages shaping their population.

The microbiome is vital for good health; when it goes wrong it can mess up a lot else. Gut bacteria seem to have a role in maintaining, and possibly also causing, obesity in the well-fed and, conversely, in tipping the poorly fed into a form of malnutrition called kwashiorkor. Ill-regulated gut bacteria have also been linked, if not always conclusively, with diabetes, heart disease, cancers, depression and autism. In light of all this, the question who guards the bacterial guardians? is starting to be asked.

The viruses that prey on the bacteria are an obvious answer. Because the health of their hosts hostthe possessor of the gut they find themselves inmatters to these phages, they have an interest in keeping the microbiome balanced. Unbalanced microbiomes allow pathogens to get a foothold. This may explain a curious detail of a therapy now being used as a treatment of last resort against Clostridium difficile, a bacterium that causes life-threatening dysentery. The therapy in question uses a transfusion of faecal matter, with its attendant microbes, from a healthy individual to reboot the patients microbiome. Such transplants, it appears, are more likely to succeed if their phage population is particularly diverse.

Medicine is a very long way from being able to use phages to fine-tune the microbiome. But if a way of doing so is found, it will not in itself be a revolution. Attempts to use phages to promote human health go back to their discovery in 1917, by Flix dHrelle, a French microbiologist, though those early attempts at therapy were not looking to restore balance and harmony. On the basis that the enemy of my enemy is my friend, doctors simply treated bacterial infections with phages thought likely to kill the bacteria.

The arrival of antibiotics saw phage therapy abandoned in most places, though it persisted in the Soviet Union and its satellites. Various biotechnology companies think they may now be able to revive the traditionand make it more effective. One option is to remove the bits of the viral genome that let phages settle down to a temperate life in a bacterial genome, leaving them no option but to keep on killing. Another is to write their genes in ways that avoid the defences with which bacteria slice up foreign DNA.

The hope is that phage therapy will become a backup in difficult cases, such as infection with antibiotic-resistant bugs. There have been a couple of well-publicised one-off successes outside phage therapys post-Soviet homelands. In 2016 Tom Patterson, a researcher at the University of California, San Diego, was successfully treated for an antibiotic-resistant bacterial infection with specially selected (but un-engineered) phages. In 2018 Graham Hatfull of the University of Pittsburgh used a mixture of phages, some engineered so as to be incapable of temperance, to treat a 16-year-old British girl who had a bad bacterial infection after a lung transplant. Clinical trials are now getting under way for phage treatments aimed at urinary-tract infections caused by Escherichia coli, Staphylococcus aureus infections that can lead to sepsis and Pseudomonas aeruginosa infections that cause complications in people who have cystic fibrosis.

Viruses which attack bacteria are not the only ones genetic engineers have their eyes on. Engineered viruses are of increasing interest to vaccine-makers, to cancer researchers and to those who want to treat diseases by either adding new genes to the genome or disabling faulty ones. If you want to get a gene into a specific type of cell, a virion that recognises something about such cells may often prove a good tool.

The vaccine used to contain the Ebola outbreak in the Democratic Republic of Congo over the past two years was made by engineering Indiana vesiculovirus, which infects humans but cannot reproduce in them, so that it expresses a protein found on the surface of the Ebola virus; thus primed, the immune system responds to Ebola much more effectively. The World Health Organisations current list of 29 covid-19 vaccines in clinical trials features six versions of other viruses engineered to look a bit like SARS-CoV-2. One is based on a strain of measles that has long been used as a vaccine against that disease.

Viruses engineered to engender immunity against pathogens, to kill cancer cells or to encourage the immune system to attack them, or to deliver needed genes to faulty cells all seem likely to find their way into health care. Other engineered viruses are more worrying. One way to understand how viruses spread and kill is to try and make particularly virulent ones. In 2005, for example, Terrence Tumpey of Americas Centres for Disease Control and Prevention and his colleagues tried to understand the deadliness of the influenza virus responsible for the pandemic of 1918-20 by taking a more benign strain, adding what seemed to be distinctive about the deadlier one and trying out the result on mice. It was every bit as deadly as the original, wholly natural version had been.

The use of engineered pathogens as weapons of war is of dubious utility, completely illegal and repugnant to almost all

Because such gain of function research could, if ill-conceived or poorly implemented, do terrible damage, it requires careful monitoring. And although the use of engineered pathogens as weapons of war is of dubious utilitysuch weapons are hard to aim and hard to stand down, and it is not easy to know how much damage they have doneas well as being completely illegal and repugnant to almost all, such possibilities will and should remain a matter of global concern.

Information which, for billions of years, has only ever come into its own within infected cells can now be inspected on computer screens and rewritten at will. The power that brings is sobering. It marks a change in the history of both viruses and peoplea change which is perhaps as important as any of those made by modern biology. It is constraining a small part of the viral world in a way which, so far, has been to peoples benefit. It is revealing that worlds further reaches in a way which cannot but engender awe.

Editors note: Some of our covid-19 coverage is free for readers of The Economist Today, our daily newsletter. For more stories and our pandemic tracker, see our hub

This article appeared in the Essay section of the print edition under the headline "The outsiders inside"

Original post:
Viruses have big impacts on ecology and evolution as well as human health - The Economist

A few months before SARS sparked major outbreaks in several countries in early 2003, there were a series of small clusters of infection in Guangdong province, in southern China. Between November 2002 and January 2003, seven clusters were reported in Guangdong, ranging from one to nine cases in size. Based on the scope of these outbreaks, researchers later estimated that the reproduction numberR0, which is defined as the average number of new infectious caused by a typical casemay have been around 0.8 during this period. However, by the time the first major outbreak occurred in Hong Kong a couple of months later, SARS had an R0 of more than 2.

There are several reasons the R0 of an infection may increase, and we can understand them better by breaking transmission down into four main components. Consider the steps that lead an infectious person to spread the infection to others. It depends on A) the duration of time someone is infectious, B) the opportunities for transmission during this period (such as social interactions), C) the transmission probability during each opportunity (such as whether someone coughs or sneezes), and D) the probability that the person on the other end of the interaction is susceptible to that infection. I like to call these four things the DOTS for short, and if we multiply them together we get the value of the reproduction number:

For example, if a case is infectious for a week on average (i.e. duration), interacts with around five new people per day (i.e. opportunities), has a 5 percent chance of infecting each of these contacts during an interaction (i.e. transmission probability), and everyone is susceptible, wed expect 7 x 5 x 0.05 x 1 = 1.75 secondary cases on average.

As I explain in my latest book, The Rules of Contagion, for infectious diseases, each of the DOTS can influence the extent of contagion. One 2016 study estimated that viruses that transmitted successfully among humans tended to cause longer infections (that is, larger duration value) and spread directly from one person to another, rather than via a secondary source (that is, more opportunities). Transmission probability can also make a difference; avian influenza viruses, such as H5N1, have struggled to gain a foothold in people because they cant latch onto the cells in our airway as easily as human viruses can. Thats why there was concern in 2011, when teams at Erasmus Medical Center and the University of WisconsinMadison announced that they had each created new strains of H5N1. Among other changes, mutations in these viruses could make them more capable of infecting the easy-to-access upper portion of the mammalian airway. Unlike circulating avian influenza viruseswhich had only spread between humans whod come into close contact with each otherthese mutant versions could go airborne, successfully spreading between ferrets in the lab. An increase in susceptibility can also nudge the reproduction number higher, such as when vaccination coverage falls for diseases such as measles, which means outbreaks may follow.

As well as understanding how infections spread, the DOTS can provide insights into how we might control them. In the case of HIV, for example, treatment can reduce the duration of time someone is infectious, and fewer sexual partners can reduce the number of opportunities for transmission. The latter generally isnt an appealing option for most people, so health agencies have instead encouraged people to use condoms, which reduces the transmission probability during sexual encounters. In recent years, there have also been successful programs of pre-exposure prophylaxis, whereby HIV-negative individuals take anti-HIV drugs to reduce their susceptibility to infection.

For SARS-CoV-2, there isnt yet a way to reduce susceptibility other than through natural infection. Even with infection, the duration of subsequent immunity remains unclear. This means control measures have focused on decreasing the effective duration of infectiousness (by isolating people with symptoms and quarantining their contacts), the number of opportunities (by reducing social interactions), or limiting the probability of transmission with personal protective equipment. Just as theres not a single component of transmission that made the COVID-19 pandemic take off, there wont be a simple solitary measure that will keep SARS-CoV-2 under control in the long term. To prevent large future outbreaks, countries will have to connect together measures that can target several of the DOTS.

Adam Kucharskiis an associate professor in the department of infectious disease epidemiology at the London School of Hygiene and Tropical Medicine.

Read more from the original source:
Connecting the DOTS - The Scientist

Ten eggs have been harvested from the last two remaining northern white rhinos in an effort to save the near extinct subspecies. Natural reproduction is now impossible as the last male of the subspecies died in March 2018.

Northern white rhinos are considered critically endangered by the IUCN Red List. Poaching and civil war in the countries they call home have decimated populations over the last three decades. Rhinos are particularly vulnerable because they are relatively unaggressive and are hunted for their horns which are used in Traditional Chinese Medicine.

It is work that is becoming more and more important as the human race continues to ravage the natural world, said Richard Vigne, managing director of Ol Pejeta Conservancy. We very much hope that our efforts keep drawing attention to the threats posed to biodiversity across the globe.

After a delay due to COVID-19, conservationists have renewed their programme to try and save the rhinos from dying out completely.

The team of international scientists working in the Ol Pejeta Conservancy in Kenya hope that they will be able to use frozen sperm from now dead male rhinos to create viable embryos.

The news comes almost exactly a year after the teams first groundbreaking move to save the subspecies using artificial reproduction methods. This is the third ovum pickup and plans are now in place to move to the next phase of the mission.

As neither of the two remaining females is able to carry a pregnancy to full term, once fertilised, the eggs will then be implanted into a surrogate. A female southern white rhino from a herd at Ol Pejeta Conservancy will be selected and then the team will create the ideal hormonal environment for the embryos to survive.

Time works against us as the oocytes that are not harvested will be lost physiologically anyway so we must try to do as many collections as possible in absolute safety, said Cesare Galli, director of Avantea - the lab in Italy where the embryos are being created.

But collecting oocytes in Ol Pejeta is only the tip of the iceberg. A lot of work is taking place behind the scenes in European zoos to be able to establish the first pregnancy with southern white rhino embryos as this will be instrumental before we thaw and transfer any northern white rhino embryos.

Only time will tell if the conservationists can manage to bring the species back from the brink of extinction.

Link:
Scientists harvest eggs in last bid to save northern white rhinos - Euronews

Many women believe that their menstrual cycles are in sync with the Moons cycle, meaning they menstruate and ovulate at specific times of the Moons cycle. There are four phases to our menstrual cycles and four main phases of the Moon, so it would be silly not to even consider the connection.Express.co.ukspoke to the worlds first period astrologer Priscila Gonsalez, who has been appointed by Intimate healthcare brand INTIMINA, to reveal the link between the Moon and your period.

The Moon has long been seen as linked with females and fertility, Priscilla said.

If you dont believe us, take a look at the etymology of menstruation and menses

The terms come from Latin and Greek words meaning month (mensis) and Moon (mene).

Priscilla said: If your cycle is in sync with the Moon you will menstruate around the New Moon and ovulate around the Full Moon.

The link is also suggested by studies in chronobiology, the field that studies our biological rhythms.

Priscilla explained: "These rhythms are the natural variations in the body that occur at regular intervals aligned to environmental changes.

Many environmental cycles exist, including the ocean tides, day and night, the lunar cycle, and the four seasons.

All living creatures have internal biological clocks that allow them to adjust their activities to changing environmental conditions.

The lunar cycle has an impact on human reproduction, in particular fertility, menstruation, and birth rate.

READ MORE-New Moon August 2020 horoscope: How will the New Moon affect YOU

Some people will find that their periods do sync up with the Moon, and they are bleeding around the New Moon and ovulating around the Full Moon.

However this is not always the case.

Priscilla noted: Syncing your menstrual cycle with the Moon has been impacted by the introduction of the pill, electricity, computers and mobile phones that have managed to unplug us from our direct connection to the phases of the Moon and to the nature of our own bodies.

On top of these modern inventions, everyones body is unique.

Priscilla pointed out: Everyones body is different and our menstrual cycle can suffer the influence of many things such as our stress levels, the environment and endocrine activity of each individual.

Whatever is your bodys pattern, trust that it is the right for you.

The two cycles of menstruation and of lunar phases dont always align with each other.

We are all different and have a myriad of personal energies acting inside and outside of our bodies.

Priscilla believes that which lunar phase you have your period during will impact your energy, mood and productivity.

Your body and psyche decides which phase you bleed during.

Priscilla said that menstruating on the Crescent Moon means you are being called to touch connect with your inner child.

She explained: As menstruation brings renewal, it is common to access patterns developed when you were a child.

If you are aligned with this phase, it is because your body and psyche are asking that you connect with your inner child and release old patterns that are stopping you from achieving what you want.

During this time, youre about to think about your dreams and desires and decide how to get them.

She said: New ideas will sprout like seeds in the spring. It is the time to take action and sow new seeds.

Start a new project or change a habit. This is the time for playing and dreaming about projects without restrictions.

Write your dreams, make plans, the sky is the limit.

Start projects, its time for a fresh start. Organise, prioritise, and get some spring cleaning done.

DON'T MISS...New Moon 2020: Is it a New Moon tonight?[INFORMER]Black Moon spiritual meaning: What does last night's Black Moon mean?[INSIGHT]Black Moon meaning: Why is the New Moon 'black' and can you see it?[EXPLAINER]

If you ovulate during the Full Moon, it is about to get really emotional.

Priscilla said: The Full Moon influences our emotions and internal waters.

Menstruating during this phase can facilitate the healing of wounds related to mother-daughter relationship, infertility and help you to connect with your loving and caring side.

This Full Moon energy enhances psychic perception, favours healing rituals related to fertility, abundance, prosperity, nutrition and creativity. Let your heart guide you.

The good news is, youre probably having a good hair day, a good skin day, or a good everything day, really!

Priscilla said: Youre likely to look and feel your best.

This is the ideal time to commit to what is rising in you and go about making it happen for real.

Launch the program you have been creating, write the book, host the event.

Work hard, birth creative projects, stay up late!

You will probably feel harmony with nature and other mothers, specially yours if you are lucky to still have her around.

If you bleed with the waning mood, your body is calling for a slower pace and your mind wants you to turn inwards.

Priscilla said this is the best time to refocus and evaluate the previous weeks.

She suggests dancing, singing, or doing whatever you can to express your emotions.

Priscilla added: The Waning Moon brings the strength of the Wild woman.

She is not scared of the underworld and your shadows.

If you menstruate with the Waning Moon, your are being called to do some shadow work.

Ask yourself what parts of you that you are hiding or not accepting.

If you are menstruating around the New Moon, this is your body telling you to get rid of what is no longer serving you.

Priscilla said: Both the New Moon and menstruation itself bring the energy of death and rebirth.

The shedding of the uterus will fertilize the soil so that the new can blossom in the future.

When menstruating on the New Moon you will be nurturing the archetypal image of the Wise Woman, and at times you may feel more tired, more introspective and feel that you need to restmore.

If you bleed with the New Moon your psyche is calling you to explore fears of getting old, letting go of old energies, people, and jobs that do not fulfil you.

You heard her, take this time to retreat into yourself and dream.

She added: "This phase symbolises the potential of new life hidden in the dark ground beneath.

Only do what is essential, now is not the time to take on any new projects.

You should delay important decisions or stressful conversations.

Although you cant change when you ovulate and menstruate, there are ways to reconnect yourself with the Moon phases.

Priscilla believes that being aware of the Moon phases can bring a shift to your cycle making your cycle less painful and conception easier.

By journalling, you can look back and check how your mood, emotions and any other patterns differ around the lunar cycle.

For example, you may have more energy around a Full Moon but feel more tired around a New Moon.

Priscilla said: Its best to listen to your body and honour your cycle, thats the most important advice to follow.

You can focus on practices that can help regulate your cycle.

Understand and make friends with menstruation and it will reward you with deeper self knowledge, stacks of energy, and the radiant health that a balanced body and mind can bring.

Read the rest here:
Does the Moon affect your menstrual cycle? - Express

The sightings have been noted since early in the pandemic critters everywhere, including here in Connecticut. Its not quite deer roaming through Japanese subway stations or monkeys hanging out in Thai cities, but theres enough bear, deer, chipmunks, rabbits, raccoons and more to make life a little, well, wilder than usual.

But a good bit of whats going on in nature this summer is nothing close to cute. Climate change which was responsible for this past winters warmth, the recent heat wave, repetitive windstorms and other weather events, including the many early-season hurricanes this year like Isaias has brought a host of flying, crawling, hopping, buzzing, burrowing and otherwise mobile creatures, many of which can cause serious and even deadly diseases.

Layer COVID-19 on top and the folks who sound the alarms on these things are sounding them louder than usual.

It certainly at least anecdotally seems like were seeing more of these species come in, said Jason White, who took over as director of the Connecticut Agricultural Experiment Station on April 1 as COVID was exploding here. Obviously the thing were eyeing right now is all the tick-borne diseases and mosquito viruses.

The Experiment Station, a state agency with a wide portfolio of research and monitoring, is best known for tracking Lyme disease, West Nile virus and Eastern Equine Encephalitis EEE, referred to as triple-E. These are normally the summers most worrisome health issues.

Yehyun Kim :: ctmirror.org

Scott Williams, agricultural scientist at the Connecticut Agricultural Experiment Station, sedates a mouse to check for ticks on Friday, Aug. 7 in North Branford. Some of these people who are recreating outdoors now who wouldnt otherwise without the pandemic, Williams. You just need to be aware and check yourself and take precautions and check your kids because it would be terrible to be social distancing and then succumb to some other diseases.

White and other researchers are worried about the coronavirus impact on other diseases, however.

People are getting kind of disease weary, White said. This is all theyve been hearing about for months and months and months. You get somebody who has flu-like symptoms and what are you going to do this year? Youre going to get a COVID test. Well, your COVID test comes back negative so you dont think about it anymore. What if its Lyme? What if its one of those mosquito-borne illnesses? If it were any summer other than this one, those are the things youd be thinking about.

But there is way more than usual to think about out in nature this summer for both people and plants. Mosquitoes are at the top of the list.

Nine months ago, municipalities in Connecticut were frantically cancelling evening activities obviously not for COVID but for EEE. It has a human mortality rate of about 35%, and the fact that there were four human cases in the state all but one fatal - made it an off-the-charts year.

The first mosquitoes to test positive for EEE this season were found in Stonington on Aug. 5. Thats early. If they show up at all its usually not until late August. I dont think it will be as big as last year, said Phil Armstrong, director of the mosquito surveillance program at the Experiment Station. Last year was a record. He said EEE usually lingers for a couple of years after a big outbreak like last years. A second positive mosquito was identified this week in Hampton.

Yehyun Kim :: ctmirror.org

Jamie Cantoni points to nymph, an immature form of the tick, on Aug. 7 in North Branford. There hasnt been a consistent increase of ticks, but ticks that used to live in the south have been found in the north, including Connecticut, said Scott Williams, agricultural scientist at the Connecticut Agricultural Experiment Station.

Its been in Massachusetts since early July and mosquito spraying is planned for more than two-dozen communities. Last year that state had 12 human cases, six of them fatal.

On the other hand, the recent stretch of hot, dry weather does not allow mosquitoes to thrive.

Wet conditions are the best setup for mosquitoes, Armstrong said. But he cautioned: We have 50 different mosquito species and they all have different relationships to water.

Increased rainfall and temperatures will accelerate mosquito development and theyll bite more when its humid.

Mosquitoes that carry EEE tend to turn up in freshwater swamps and areas with a high water table like those found in eastern Connecticut, which is where EEE tends to surface. Long periods of sustained rainfall like those in early spring not periodic extreme rains like the state has had lately are what help them proliferate.

But theres another big factor the birds that carry EEE in their blood. Thats what the mosquitoes feed on before they bite humans or horses, another mammal in which EEE can be fatal.

EEE can kill birds, but theres some evidence theyre developing immunity. And if theyre able to stay alive, the disease can be transmitted more easily to mosquitoes. Not only that, one of the most prolific carriers is that most common of birds the robin.

Connecticut Agricultural Experiment Station

EEE cases in Connecticut last year a record year with four human cases, three of which were fatal.

The other major disease concern from mosquitoes remains West Nile Virus. But its generally from a different mosquito Culex pipiens. It prefers torrential downpours and dirty standing water. It primarily develops in catch basins and storm drains, so heavy rain will flush them out.

But spring/early summer rain followed by drought like Connecticut had this year is a set-up for West Nile, and indeed the first mosquitoes that tested positive for West Nile were found on July 8 in Newington and have been increasing in number since. A human case was identified this week.

To keep track of all these tiny vectors, the state has added 16 new mosquito trapping stations all in eastern Connecticut bringing the total number of testing sites to 108. Each site has at least two traps that are cleared every 10 days. If West Nile or EEE is found, clearing changes to twice a week.

At some sites theres a third trap to catch Asian Tiger mosquitoes, a species that has been creeping northward and, as climate change has made winters warmer, has overwintered in southern Connecticut. It can transmit the dangerous Zika and Chikungunya viruses.

Armstrong said the data the lab has collected provide a picture of how mosquitoes here are changing over time. Climate change seems to be playing a role. Weve seen a lot of other species that are new arrivals into the state that likely resulted from northward range expansion from the south, he said. He said thats different from some of the climate change-induced insect trends that are reducing diversity. If you just look at overall mosquito abundance and mosquito diversity in the state its going up.

I always get asked is it a bad tick year, said Kirby Stafford. He is the state entomologist and as chief entomologist at Experiment Station runs the Center for Vector Biology and Zoonotic Diseases. Every year is a bad tick year. Some are just more so than others.

Yehyun Kim :: ctmirror.org

Jamie Cantoni, agriculture research assistant at the Connecticut Agricultural Experiment Station, walks around the woods in North Branford to collect ticks.

This years degree of bad is yet to be determined. But at one of the states two tick testing labs the one known as a passive lab because the public brings ticks in to be examined ticks were coming in all winter at a heavy clip, according to its director Goudarz Molaei.

From November to mid-March, Molaei said, they used to receive 50-100 ticks. But thats ballooned up to hundreds some 800 in 2017.

Last year the station started active tick testing,collecting them from 40 sites across eight counties. But a ticks two-year life cycle makes prediction complicated. Ticks in any given year reflect how many were born the previous year. And the ticks born in any given year will inform what happens the following year.

Among the other pieces of the tick puzzle is the number of hosts the ticks need some for reproduction, some for transmitting the disease. And in the case of the rodents and large mammals that handle those duties, there are a lot more at the moment.

Yehyun Kim :: ctmirror.org

John Shepard, mosquito biologist at the Connecticut Agricultural Experiment Station, identifies mosquitos. Climate change that includes longer summers, mild winters and extreme weathers, is one of the factors that led to the increase of new species of mosquitos from the south, said Philip Armstrong, director of mosquito surveillance program at the Experiment Station.

The changing climate is also starting to alter the profile of ticks appearing in the state, with southern species working their ways north as the climate warms. The Lone Star tick is exhibit A, Stafford said. It is an aggressive biter that can spread several dangerous diseases including ehrlichiosis, a bacterial infection that can produce flu-like symptoms; or cause Alpha-gal syndrome a recently identified allergy to red meat.

The Lone Star tick is probably responsible for about 90% of tick bites in the southeastern United States but its been moving northward with our warmer winters, Stafford said. It began showing up on Long Island in the early 1990s and is now abundant there.

In the past decade, Lone Star tick submissions to the passive tick lab have increased from about 2-3% to more than 4% this year. In 2017 Stafford discovered a large population of Lone Stars on a peninsula in South Norwalk.

Connecticut Agricultural Experiment Station

Asian Longhorned Ticks are in nearby states, but not Connecticut yet a prospect that has officials worried.

More ominously, Stafford and a colleague conducted overwintering survival studies with adult stage Lone Star ticks in Maine. While the survival rate was low, some nevertheless weathered the winter.

Stafford also has his eye on the Asian Longhorned tick, which infests deer and livestock and can transmit some nasty diseases including a hemorrhagic one. Its not in Connecticut but has infested sheep in Hunterdon County, NJ since 2017 and is in Westchester County, N.Y. and on Staten Island. It is parthenogenic which means females dont need males to reproduce.

Overall his tick concerns are similar to Jason Whites that folks fixated on COVID may ignore symptoms of Lyme disease, which Stafford said tends to be under-reported anyway. And with people using parks and outdoor areas more as a respite from being stuck home due to the pandemic, he worries more people will suffer tick bites.

Mosquitoes and ticks may be the most worrisome of summer visitors, but they are not the only changes in the critter world that Connecticut is experiencing.

There is one bit of good news. Gypsy moths shouldnt be too bad this year. (Of course there wasnt much left for them to eat.)

Spring rains like the ones we had this year and last are good for proliferating a soil fungus that can keep the oak-munching critters in check. But the three years before that were absolute devastation.

Weve had drought before and weve had gypsy moths, but the two came together in three consecutive years and the trees just couldnt take it, said Chris Martin, director of the Division of Forestry at the Department of Energy and Environmental Protection and the state forester.

Yehyun Kim :: ctmirror.org

Alyssa Marini, seasonal resource assistant at the Connecticut Agricultural Experiment Station, identifies mosquitos. Mosquitos are more active late summer to early fall, especially in August and September, said Philip Armstrong, director of mosquito surveillance program at the Experiment Station. Armstrong recommended wearing long sleeve shirts and avoiding outdoor activities between dusk and dawn to avoid mosquito-borne diseases.

Well over hundreds of thousands of trees covering 90,000 acres have been defoliated, rendering many of them hazards to safety, not to mention utility lines in storms such as the recent tropical storm Isaias.

There wont be aerial surveillance of trees this year. The Cessna aircraft used to do conduct it are too small to allow social distancing, but last years was brutal enough for both years. And there are plenty more insects causing trouble.

Emerald Ash Borer arrived less than a decade ago. Efforts to keep it from spreading largely failed and it has marched west to east to cover the entire state. It takes three to five years to kill a tree once its established.

Were going to lose most all of our ash on account of this, Martin said. The insidiousness of this insect is that its small, it spreads itself quickly, its hard to detect and by the time you notice it, its too late.

While the ash borer was introduced through global trade, the arrival of Southern Pine Beetle is clearly climate change-driven. Its been throughout the south for decades hitting the New Jersey pine barrens as far back as the 1980s and turning up in Connecticut in 2016.

Another newcomer is Beech Leaf Disease caused by a nematode. It was found in one tree in Stamford last year, but when scientists began monitoring, they discovered that wasnt the only one by a long shot.It had come from Westchester and Dutchess Counties in New York and headed all the way through Rhode Island and into Massachusetts, said Bob Marra, a forest pathologist at the Experiment Station.

Robert Marra, Connecticut Agricultural Experiment Station

The dark leaves show infestation by the nematode that causes Beech Leaf Disease.

The nematodes inside the leaves make the foliage look dark. And thats where they breed. Tons and tons of eggs. Anything were seeing now this year, had to have come from over-wintered buds that were already infected, Marra said. The leaves emerge fully symptomatic.

Marra is also facing the potential of Oak Wilt arriving in Connecticut from where its currently holed up in Brooklyn, N.Y. Its a fungus but the vector is a native bark beetle. Oak Wilt operates similarly to Dutch Elm disease in that it blocks the trees vascular system, killing it within a season even a large mature oak.

Marra hasnt found Oak Wilt in the state, but hes not saying its not here. Probably is, he said. We just have not seen it yet.

But if youre looking for good critter news try the Asian Long Horned Beetle, a pest that is happy to munch on any number of hardwood trees. Connecticut has managed for the better part of a decade to keep it from making the move from the Worcester, Mass. area across the border to this state.

I think the lesson learned is we just dont know whats next, DEEPs Martin said.

Big animals, little animals

Gale Ridge, an associate scientist at the Experiment Station, isnt entirely sure which category the Cicada Killer wasp is in big or little. As wasps go, its big. More to the point,it is much the same size and look as the so-called murder hornet thats become a media sensation. It is causing Ridge to take call after call after call after call.

Connecticut Agricultural Experiment Station

The Connecticut Agricultural Experiment Station has received many calls from people who think theyve seen the Giant or Murder Hornet.

Thats probably one right now, she said as one of her phones started ringing, as if on cue.

The murder hornet, officially the Asian giant hornet, was found in Washington state. Thats 3,000 miles away, she exclaims. And so shes providing a handy how-to-tell-them-apart from the cicada killer diagram, since the cicada killer (which might have its name changed to the somewhat more benign cicada hunter) is around here.

Other than killing cicadas and dragging them off to a hole in the ground to feed her young, these wasps are harmless. In her 20 years at the Experiment Station, Ive never known anyone to get stung by them. Males dont even have stingers, Ridge said.

Of course you dont have to be big to do damage, which home gardeners and even seasoned farmers are learning this year more than usual, the experts say.

Asiatic garden beetles are out there in huge numbers hiding in the dirt during the day and feeding at night on just about anything. (So if youre wondering what has chewed up every leaf in your vegetable bed this is a pretty good bet, Ridge said.)

Shuresh Ghimire, vegetable specialist with UConn Extension service, said squirrels, chipmunks and deer are what hes hearing about. Im getting queries from growers and gardeners who have been farming for many years and have never encountered such a big problem with squirrels, he said.

Hes also getting many more weekly calls about European corn borer problems than he has in previous years.

The fruit fly known as spotted wing drosophila continues its march begun about a decade ago through the states berry crops in particular. Were still in a situation thats pretty dire for fruit growers, said Richard Cowles, a scientist with the Experiment Station.

Weather patterns drive infestations, especially if its humid at ripening, a circumstance he called pure hell for growers. This summer he could see problems starting to increase duringstrawberry season. But now with the sustained hot and humid conditions, he said: Its going to be a wild ride for blueberry growers.

But what really has him worried is a new arrival the Spotted Lanternfly after one was sighted in Southbury fall. It will eat pretty much anything, but has a special taste for vineyards, which growers in Pennsylvania learned the hard way. It sucked the living daylights out of the vineyards and killed them outright, Cowles said. They dont have a significant number of enemies to keep population in check.

If they arrive in Connecticut, he worried, there would be mind-boggling impacts.

So far there have been no additional sightings, but hes anticipating there will be. Were just waiting for other shoe to drop.

Connecticut Agricultural Expeeriment Station

This year, West Nile Virus began to show up in mosquitoes in early July. A human case was reported this week.

Beyond all the hand-wringing over small but troublesome critters, its the big eye-popping animals that have gotten a lot of the publics attention. Reports of bears, bobcats and coyotes in particular, along with smaller mammals have been nonstop.

Whats actually different and whats perceived to be different those are two different things, said Jenny Dickson, director of DEEPs wildlife division. She attributes a good bit of it to people being at home more due to the pandemic. Theyre noticing more things than they do normally.

There are definitely more bears.

Some of the wildlife activity has to do with people staying indoors, especially this spring when hibernating animals reappeared, giving them human-free spaces to roam in. Some of it has to do with active restoration efforts, such as with osprey. But a lot of it has to with climate change.

It was a warm winter, making it easier for species to survive and their food more plentiful. Ranges and seasons for animals have expanded generally due to the warming climate.

I think there are populations that are just expanding, and its not necessarily tied entirely to mild climates or food availability, although both of those things help, Dickson said.

But whatever the animal for whatever reason, Dickson is good with it.

What has stood out for us are the number of calls weve gotten from folks who are sort of newly discovering the wildlife around them, she said. For them its a combination of new and scary and fascinating all at once.

See the article here:
Where Connecticut, COVID, climate change and critters intersect - The CT Mirror

INTRODUCTION

Classical approaches to urban theoryin economic geography (13) and, more recently, in complex systems (4)often treat cities as spatial equilibria, where a balance of benefits and costs is achieved out of a set of social and economic exchanges, including wages, land rents, and transportation costs (1, 35). While these modeling approaches have proven powerful for generating quantitative predictions in agreement with many observed properties of cities (35), they leave unresolved two fundamental problems: the problem of statistics and the problem of growth.

Both growth and statistics denote a broad set of phenomena that must be unpacked so that we can fully appreciate what is at stake. By statistics, we mean that in dealing with real cities, we must appreciate the wide variation between individuals and places. This variation has positive manifestations in that cities are extremely diverse in terms of the types of the people and lifestyles they support, including a broad set of coexisting cultures, professions, languages, races, and ethnicities (69). This interdependent functional diversity is what J. Jacobs famously called organized complexity and is at the heart of the kind of problem a city is (10). Negative expressions of these same heterogeneities are also familiar, such as ethnic, racial, and economic segregation (11, 12), inequality, and variable access to justice and opportunity. Moreover, it has been observed that these differences between places and people within each city are persistent over time (9, 12) and do not have the fleeting character of noisy fluctuations in statistical physics. Instead, they can pile up over time and lead to patterns of cumulative advantage and disadvantage (9, 13), which are at the root of most challenges of human development. Thus, the problem of statistics in cities deals not only with the existence of structural differences on how the same quantity is distributed across different people and places but also with the temporal persistence and amplification of these effects.

By growth, we mean that (modern) cities are characterized by fast, typically exponential change across many variables. On the one hand, modern cities tend to experience annual population growth rates between about a fraction of 1 and 3 to 4%, as we shall see below. Exceptions exist at either end at least over some periods of time, as different places experience contextually specific factors. However, the principal change in modern cities is the fast pace of their economic growth and technological transformations. Across the world today, we observe rates of urban economic growth that are typically larger than those in their corresponding populations, reaching in some cases 10% a year, with 2 to 4% being typical (14). These growth rates mean that the size of a citys economy doubles every few decades, making it possible to transition from poverty to wealth in one or two generations, as has happened in many places over the last century. With such fast growth rates at play, how is it tenable to model cities as spatial equilibria? Even more importantly, how do different resource growth rates, experienced by different households, neighborhoods, or cities, shape the heterogeneity (inequality) of outcomes for different people? Why are cities not torn apart by differential growth more often?

It turns out that these two problems, of statistics and of growth, are intimately connected and must be tackled together. This is a twist on classical statistical mechanics linking the strength of fluctuations to dissipation around equilibrium (15), which, in the context of exponential growth processes in populations, takes a character that we may call fluctuation-amplification, typical of evolutionary processes (16). The literature of complex systems applied to urban growth, especially from the perspective of geography, has demonstrated the importance of this type of stochastic nonlinear dynamics with strong feedbacks for a number of decades (17, 18).

To continue to make progress on these issues, an analytical approach is necessary that identifies and articulates the essential joint mechanics of scaling, growth, and statistics in cities. Here, we show how this synthesis can be constructed and illustrate its theoretical and empirical implications using stochastic simulations and a particularly long time series of wages and populations of U.S. metropolitan statistical areas (MSAs) covering nearly five decades. The central insight is the realization that the budget constraint used to define functional cities (14) is not static or homogeneous across agents but must be managed adaptively to promote growth and avoid instability. This is because what is being controlled are stochastic net resource flows over time, as incomes minus costs, not static quantities such as forces. It is precisely this fluctuating difference that accumulates to drive resource growth for each agent and for cities as populations of agents, with implications for both aggregate growth and inequality.

The manuscript is organized following the scheme of Fig. 1. We first show how scaling analysis isolates the parameters that control urban growth, providing a number of quantitative targets for explanation and prediction. We then introduce standard theory for stochastic geometric growth and outline its properties. This allows us to establish the connection between well-known models of cities as spatial equilibria and processes of stochastic growth for individual agents. This connection disaggregates a city-wide budget condition to the level of agents and requires that they act strategically in their own self-interest to control fluctuations in net incomes, which is naturally handled via adaptive temporal averaging of expenditures. This is the central assumption of this manuscript, which leads to the expectation that net income volatilities are kept finite and small and that temporal averages of incomes and expenditures become statistically dependent. A number of results follow from standard limit theorems: The statistics of resources, incomes, and costs at the agent and group level become asymptotically lognormal, even as these quantities grow exponentially. The self-similarity of growth processes across group sizes also emerges, defining running couplings characterizing the mean growth rate for populations and its associated volatility. Explicitly computing these quantities allows us to identify the circumstances when urban scaling is conserved by the system dynamics. Conversely, we show when these conditions are violated, creating corrections to mean-field scaling exponents when volatilities are scale dependent but small and the breakdown of scaling if they become large. These procedures are illustrated using data on wages for U.S. metropolitan areas. We finish with a discussion of the significance of the results toward a general statistical dynamics of cities and its relation to analogous dynamics of resource flows in other complex systems.

Basic assumptions are shown as blue boxes, and derived results are shown as red boxes; arrows indicate outcomes, while dashed lines represent alternative scenarios. The budget condition, y c, is the common basic assumption for urban agents, generalizing energy conservation in simpler systems. Recognizing its dynamical, stochastic nature leads to the central assumption of the manuscript that agents must actively control its associated volatility; the simplest way to do this is through the time averaging of expenditures (consumption smoothing). Then, the resource growth rate volatility, 2, becomes finite and small, both at the individual and group levels. This leads to emergent stochastic geometric dynamics of resources both at the individual and population levels with exponential growth and lognormal statistics observable at long times (right). Averaging over populations derives the growth rate statistics for cities, which determines when dynamics become self-similar across scales and preserve urban scaling (left): First, if under group averaging variations of the growth rate () are correlated to those in agent resources (r) inequality will change within the population. Second, if effective growth rates are independent of population size, the dynamics becomes self-similar and urban scaling is preserved over time. Alternatively, if the growth rate volatility(2) is population size dependent, corrections to mean-field exponents result: They are calculable via B 0 and are controlled by the volatilitys magnitude. For large 2(N), the statistics become dominated by fluctuations, and urban scaling breaks down. The existence of strong group volatilities contradicts the assumption of effective control at lower levels. This regime would be unstable, signaling the loss of control over resource flows for most of the population and entailing wide-spread crises and eventual collapse. See text for detailed notation.

Scaling analysis provides a simple and straightforward way to characterize urban quantities (19) and extract average agglomeration effects at play across cities of different sizes in an urban system. This section also shows that scaling analysis provides an efficient parameterization of growth processes, different from growth accounting in economics (20).

The starting point is to write an extensive positive quantity, Yi(t), here the total wages paid in city i over a time period t, asYi(Ni,t)=Y0(t)Ni(t)ei(t)(1)where Y0(t) is time dependent but independent of city population size Ni(t), which also changes over time due to standard demographic processes. The parameter is the scaling exponent, and the quantity i(t) is the time-dependent deviation from the average scaling prediction for city i. This expression is exact because any deviation from the average scaling relation, Yi(Ni, t) = Y0(t)Ni(t), is absorbed in the residuals i(t) (see section S1).

Averaging over cities can now be used to isolate a few quantities of interest. To do this, let us first take the logarithm of Eq. 1lnYi(Ni,t)=lnY0(t)+lnNi+i(t)(2)followed by the average over cities ln Y(t) = ln Y0(t) + ln Ni. This ensemble average is defined explicitly by lnY(t)=1Nci=1NclnYi(Ni,t), lnN(t)=1Nci=1NclnNi(t), where Nc is the total number of cities in the urban system such as the United States. We will refer to the quantities ln Y(t) and ln N(t) as centers (21), which are collective coordinates tracking the temporal motion of the entire system of cities (yellow symbols in Fig. 2, A and B) analogous to center-of-mass coordinates in many-body physics.

(A) Total wages for U.S. metropolitan areas 19692016. Each circle is a city in a given year from blue (1969) to brown (2015). Yellow squares show the urban systems centers ( ln N, ln Y), which account for collective economic and population growth (movement, upward and to the right). Urban scaling relations for each year (black lines) are derived through the consideration of a short-term spatial equilibrium (inset), which changes on a very slow time scale. (B) Centered data, obtained from (A) by removing the centers motion (inset). This allows the decomposition of temporal change into two separate processes: collective growth (centers motion) and deviations from scaling, i(t), characteristic of each city i. We see that scaling with a common exponent (global fit = 1.114, 95% confidence interval = [1.111, 1.117], R2 = 0.935) is preserved over time, and net growth is a property of the urban system and not of individual cities. (C) The statistics of deviations, obtained from the residuals of the centered scaling fit of (B). While the distribution is well localized and symmetrical, it is not very well fit by a normal distribution (blue line). Instead, the red dashed line, which follows from theory developed in the paper, produces a much better account of the data. (D) The deviations i(t) of a few selected cities: Silicon Valley (San JoseSanta Clara MSA) and Boulder, CO show two of the more exceptional trajectories in wage gains for their city sizes, whereas Las Vegas, NV and Havasu, AZ illustrate wage losses. New York City, Los Angeles, and the exceptionally poor McAllen, TX show no relative change in their positions over nearly 50 years.

By definition, the ensemble average of the deviations is zero, (t) = 0, so that the second (variance) and higher moments of the become the leading quantities of interest. From these expressions, we can write the deviations i(t) asi(t)=lnYi(Ni,t)Y0(t)Ni(t)=[lnYi(Ni,t)lnY(t)][lnNi(t)lnN(t)](3)

The first expression is the most common interpretation of the i as (multiplicative) residuals from the scaling relation, whereas the second makes their status as deviations from the collective coordinates (centers) explicit. For these reasons, the i(t) provides a city size-independent measure of city performance and are also known as scale-adjusted metropolitan indicators (SAMIs) (22). Characterizing these three quantities, the two centers and the statistics of the , gives a complete description of growth and deviations in a system of cities and separates collective effects from idiosyncratic events in each city.

Figure 2 illustrates the meaning of these quantities. Figure 2A shows the total wages, Yi(t), for U.S. MSAs between 1969 and 2016 (47 years), year by year (colors light blue to brown). The growth trajectory of some specific cities, such as New York City, Los Angeles, Chicago, or Silicon Valley (San JoseSanta Clara MSA), is easily visualized in this way. The solid lines show the scaling relation for each year (see caption for details). We see how scaling is a good fit to the data each year, reproducing a slowly shifting spatial equilibrium in each instance (inset) (4). We also see how the position of the center (yellow squares) moves from year to year, reflecting the overall growth in population (shifts to the right) and especially in wages (movement upward). These results include both real and nominal growth of wages due to inflation.

Figure 2B shows the result of removing collective growth by moving all data clouds so that their centers coincide at the origin (0,0) (21). We see that removing the centers motion (inset) results in the reproduction of the same scaling pattern at each time, with additional small and slow moving deviations, i(t), changing only slightly from year to year. Figure 2C shows the histogram of these deviations (gray) about the overall best fit scaling relation (Fig. 2B), pooled across all years. We observe that the statistical distribution is well localized and symmetrical about the origin, but that is not very well fit by a normal distribution (blue line). A better fit is provided by another model (dashed read line), which will be derived below. Last, in Fig. 2D, we see the change in i(t) over time for some of the most extreme trajectories. Specifically, some cities became substantially richer in relative terms over this period (Silicon Valley and Boulder), some experienced loss in economic status (Las Vegas, NV and Havasu, AZ), and a few others, such as New York, Los Angeles and McAllen, TX (one of the worst performers, by this measure), have not changed much. These trajectories also show how slow the change in the s is. Particular events that affect different cities at specific times are easily identifiable, such as the dot-com economic boom and bust around 2000, specifically for Silicon Valley and Boulder.

We now seek to connect the macroscopic statistics of entire cities to a microscopic general model of single agents behavior. We note that the budget condition, which is taken as the starting point of an important set of urban theories (14), is a version of the fundamental law of energy conservation. Hence, it must apply to every single agent and to cities as collections of agents.

To see this, let us start by introducing a variable, r(t), denoting the accumulation of the net quantity of y(t) over time t. For example, if y stands for an agents income, then r becomes its monetary wealth, but we should think of r more generally as resources that can be grown over time and used in turn (reinvested) to generate more y and so on. An important noneconomic example, which applies to premodern human societies and other biological populations, is when r is stored energy and y is an energy income per unit time. We write the dynamics of r, given y, asdr(t)dt=y(t)c(t)=r(t)r(t)(4)where r is the stochastic growth rate of resources. The first equality in Eq. 4 is pure accounting, stating that resources grow by the difference between income and costs, c, (i.e., net income) over some time interval, dt. Costs in cities are local and include real estate rents, transportation, and consumption, as well as others, such as health care and losses resulting from crime or poor urban services. In this sense, costs and benefits are also affected by migration decisions about where to live and work. The centerpiece of this equation is the difference between income and costs y(t) c(t), which must be balanced by all agents in their specific environments. For urban agents, this difference is the budget condition for the spatial equilibrium that defines a city according to the Alonso model of economic geography and urban scaling theory (1, 3, 4). In the original version, this difference is typically set to zero, although the meaning of incomes and costs is rather flexible and can include savings (23). In urban scaling theory, this difference is nonzero in general (4) and becomes the target of maximization through the self-consistency of infrastructural and social network properties. This implies that a positive difference between incomes and transportation costs is necessary for cities to exist (Fig. 2A, inset) and to generate exponential resource growth. It also implies that the scaling of resources, incomes, and costs has the same population size dependence characterized by a single common exponent for all these quantities, > 1.

The second equality in Eq. 4 is a definition of the growth rate r implying that r(t)y(t)r(t)c(t)r(t). Equation 4 is not an arbitrary modeling choice: It is the standard starting point for modeling population growth and human behavior where time, effort, or resources are invested strategically. Among many examples, it is the standard model for city population growth (24), the standard model of financial mathematics and asset pricing (25, 26), and the one good wealth accumulation model, which is the basic tool in economics to analyze dynamical issues of wealth inequality (23). Stochastic proportional growth and resulting lognormal (and power law) distributions are associated with many forms of human behavior, including the statistics for the time to complete a task, epidemic dynamics, demography and even the statistics of marriage age [see (27) for a review].

Let us see how this model works in practice. Because the equation is nonlinear (the stochastic term is multiplicative), we have to be careful and use the rules of stochastic (It) calculus to integrate its solution in time, leading tolnr(t)r(0)=(rr22)t+(t)(5)where the randr2 are the mean and variance of r, respectively, in the usual sense of those obtained over the probability density of r. Now, let us define the average effective growth rate r=rr22. This quantity is fundamental in geometric random growth models and will recur in the discussion below. Keeping track of physical dimensions tells us that randr are temporal rates and have dimensions of (time)1. Thus, the standard deviation (SD) r (known as the volatility) has dimensions of (time)1/2. The stochastic noise (t) is the sum over the integration time, t, or more explicitly, (t)=l=1tr(tl), with r(tl)=r(tl)r. This is a random variable with zero mean. Because it is the sum of stochastic variables, we expect (t) to express universal behavior as a consequence of the central limit theorem (25). In the simplest case, where r is statistically independent across time with finite variance, we obtain that = rW(t) is a Wiener process, which is a normal variable with zero mean and variance 2(t)=r2t. This will later define the property of ergodicity for stochastic growth, which means that for long-time averages of growth rates, the mean dominates the variance. This is not to be confused with the more general property with the same name in statistical physics, which means that all allowable spaces of a dynamical system will be sampled subject to constraints. The point of the present paper is to show that path dependence for particular cities can coexist with simple emergent statistics for the ensemble of cities.

A number of key general results follow from this solution and associated limiting theorems. First, (i) the central limit of implies that lnr(t)r(0) approaches, in the same limit of long times, a Gaussian variable with time-dependent mean rt and variance r2t. This implies, in turn, that (ii) r(t) is asymptotically distributed as a lognormal variable, a result that will become important later. In turn, (iii) the temporal mean growth rate 1tlnr(t)r(0)=r+(t)tr, for long times, as a result of the behavior of t1/2r. Last, (iv) the characteristic time, t*=r2(rr22)2, marks the interval necessary for net exponential growth to become apparent over the (shorter-term) effect of fluctuations, which average out for longer times.

These properties are illustrated in Fig. 3 (A to C), obtained from numerical simulations of Eq. 4, with r taken as Gaussian white noise. The asymptotic behavior of all quantities depends on whether the effective growth rate is positive r > 0 or equivalently r>r2/2 (Fig. 3, A and C). When this condition holds, there is net growth (Fig. 3A, blue and orange trajectories). Growth of r becomes apparent on a time scale longer than t* (Fig. 3A), which can be very short when the volatility is small. In this regime, the distribution narrows on the scale of the mean as t becomes larger, and predictable exponential growth emerges. However, when r0 is not sufficient to guarantee long-term growth; instead, a finite threshold r>r2/2 must be overcome. Approaching this threshold from a regime with growth, an agent will experience wild fluctuations as t* goes to infinity (Fig. 3C, inset) and will struggle to tell whether growth persists and estimate its time scale to plan. As a consequence, low volatility and positive average rates are necessary for sustained growth (Fig. 3C). Given these results for individual agents, it becomes critical to establish the conditions for these dynamics to apply also for populations such as for entire cities (Fig. 3D) and to determine how corresponding growth rates change across scales.

(A) Example of growth trajectories for a simple process of geometric Brownian motion (Eq. 4). The blue trajectory shows typical growth with small fluctuations and positive effective growth rate, the orange line shows a similar situation with larger fluctuations, and the green line shows a trajectory with critical r = 0. The purple and red lines illustrate negative effective growth rate trajectories. The critical growth time, t*, is shown for growing trajectories. (B) An ensemble of trajectories with stochastic growth rates similar to those of U.S. MSAs, starting with the same initial conditions. The yellow line shows the temporal trajectory of the ensemble average, and the black lines show the 95% confidence interval. Note that both the mean and the SD are time dependent (see text). The inset shows the resource distribution at a later time, which becomes asymptotically lognormal (red line). (C) The general properties of stochastic growth imply that a positive growth rate is necessary to overcome temporal decay due to rate fluctuations. If volatility increases, growth will ultimately stop, and decay will ensue. The critical point r=rr22=0 is characterized by large fluctuations with a diverging t* so that agents will not be able to tell whether they are experiencing growth and may be unable to exert effective control (see text). (D) Under general conditions, multiplicative random growth can be self-similar across group sizes, providing a simple theory that applies at all scales, from individual agents to populations and cities (see section S3) (29). However, the key parameters of the theory run across scales and are in general sensitive to both group size, temporal averaging, and inequality. These dependences define urban scaling as a dynamical statistical theory beyond the mean-field approximation.

From the general properties of stochastic growth processes, we can conclude that any agent seeking growth must aim at a positive mean growth rate and small volatility. The conundrum is that the volatility and the mean growth rate are, to a large extent, properties of the environment, outside the agents control. What is under the agents control, however, are his/her own actions, which we show next can adapt to extrinsic circumstances via processes local in time so as to produce low volatility and stable growth.

It is important to realize that, besides levels of population aggregation, there is also a hierarchy of time scales involved in the process of balancing costs and benefits and observing growth (Fig. 3D). Over the very short term, there will be moments when the agent is resource flow negative, e.g., while shopping. However, judicious choices over time should result in more even positive net flow over the longer term, integrating together periods when incomes are larger than costs (at work) and vice versa (at home, socializing, etc.). This process of balancing costs and benefits over time is necessary in dissipative complex systems because there are always resources lost in any activity or exchange. Balancing costs and benefits over time creates strong correlations between y, c, and r and results on ratios, y/r and c/r, that can become independent of the level of wealth, as we show next.

Consider the basic accounting (Eq. 4) for a single agent, y(t) c(t) = r(t)r(t). As we have seen, dividing by r(t) > 0 gives us the definition of the growth rate r(t). Defining the two resulting ratios as b(t) y(t)/r(t) and a(t) c(t)/r(t) and averaging over time leads to1t0tdt[b(t)a(t)]=ba+1t0tdt(r(t)r)ba=r(6)

This means, in general, that we can also define r(t)=r+r(t), where r(t) is the error (or fluctuations) away from the growth rates temporal mean such that 1t0tdtr(t)0, as we have seen for (t) in the previous section.

What kind of process sets the statistical properties of these fluctuations? On a short-term basis, fluctuations will be large if a, b vary strongly and independently of each other. Then, the amplitude of r will be large over some period of time and, if negative, may deplete stored resources (r 0), placing the agent at risk of death or bankruptcy. Thus, it is in the vital self-interest of the agent to act so as to minimize, or at least control, fluctuations.

How is this to be achieved? The point is that the variations in expenditures, a(t), should not just be seen as passive costs but rather as strategic dynamical investments under the agents control. Conversely, the returns on this investment, b(t), are stochastic and will always fluctuate because of environmental factors (Fig. 4A). Thus, a(t) should be chosen to generate a target growth rate and reduce fluctuations, in other words, to achieve stable and predictable growth (Fig. 4, B and C).

(A) Example trajectories for the income-to-resources and costs-to-resources ratios, b(t) (red) and a(t) (blue), respectively. Note that when income is larger than costs, there can be growth, but fluctuations need to be controlled. (B) Control scheme to deliver average growth rate and tame fluctuations r(t). Costs a(t) become a control variable that, in part, adapts to environmental fluctuations to generate r(t) with small, known variance. (C) The dynamics of the resulting error r(t) is now centered around zero and (D) displays a Gaussian distribution (red line) with variance given by the ratio of the environmental variance to control parameters (see text). In this way, adaptive agents behavior can lead to predictable growth in stochastic environments with a chosen variance.

To demonstrate how this can be achieved, we write the returns as b(t)=b+v(t) and the investment as a(t)=a+u(t). Here, v(t) are (stochastic) variations in returns, whereas u(t) will play a role of a control variable adjusted by the agent. This leads toba+v(t)u(t)=r+r(t)r(t)=v(t)u(t)(7)

We must now specify how control is implemented to tame the errors. Most general practical controllers are in the Proportional-Integral-Derivative (PID) class (28), which specifies u(t) as a function of the error, r(t), asu(t)=kPr(t)+kI0tr(t)dt+kDdrdt(8)where kP, kI, and kD are constants (in time) to be chosen by the agent. These three terms allow for different kinds of strategy to reduce fluctuations: kP is the magnitude of an instantaneous response against the fluctuation, kI refers to averaging of the error over time (known as smoothing, because averaged errors are smaller and converge to zero), and kD describes a corrective reaction in the direction of the temporal change in the error. Of these, only smoothing by time integration will prove essential. Note also that u(t) is a simple quantity that can be updated locally in time via the current observed error, r(t), and its addition and subtraction to the integral and difference, which requires remembering only two numbers. The stochastic dynamics of the errors is best captured via the derivative of Eq. 8dudt=kPdrdt+kIr(t)+kDd2rdt2kDd2rdt2+(kP+1)drdt+kIr=dvdt(9)

This equation for the error describes a simple driven oscillator: It is familiar from stochastic calculus when we take dvdt to be white noise with variance 2. The solution is provided in section S2, showing that r converges to a normal distribution with zero mean and variance r2=22kP+1kI (see Fig. 4, C and D). Making kI larger has the double effect of accelerating the temporal convergence to a time-independent distribution and narrowing the error variance. The other parameters, kP and kD, can be set to zero, leading to very simple control based on the temporal averaging of the fluctuations. The effect of the environmental variance 2 is simply to increase the error variance proportionally. Thus, the control process effectively filters out environmental shocks and makes the net-income variance smaller as a function of parameters chosen by the agent. This is a very simple general mechanism that allows agents to cope with environmental uncertainty and generate stable growth by adjusting their expenditures over time. Much more sophisticated strategies are possible that can maximize growth rates if more of the structure of returns, b, are known (28).

We see how averaging expenditures over time (known as consumption smoothing in economics) gives a general mechanism whereby agents can make their average resource growth rate take on a target value, up to stochastic fluctuations with variances given by the balance between the unpredictability of the environment and the quality of their control. Effective control generates strong statistical correlations between income and costs over time, which constitute the basis for (a spatial) equilibrium. In this light, variations between agents may persist as the result of differences in their specific experienced environments and/or the quality of their control. Exposing these issues requires the consideration of averages over populations of agents as in Fig. 3D, to which we now turn.

To compute the growth dynamics for a city, we now define the averages over a population of size G, rG=1Gj=1Grj, where rj are individual js resources and so on for growth rates, incomes, and costs (see section S6 for a summary of notation). To derive the corresponding dynamics, we take these averages over Eq. 4drGdt=yGcG=(r)G(10)where we dropped the r subscripts on the rate, for simplicity, so that in this section, rG G. The average of the product is(r)G=1Gj=1Gjrj=GrG+covarG(,r)=[G+covarG(,r/rG)]rG(11)

The quantity GG+covarG(,r/rG) is the effective stochastic growth rate for the group average resources, rG. This quantity equals the simple arithmetic group average, G, plus a correction due to the fact that growth rate variations may not be statistically independent from variations in resources across individuals. The covariance term is familiar from evolutionary theory in the context of the Price equation (16) and signals selection. For example, if richer individuals experience higher growth rates across the group, then the average growth rate will be higher and vice versa. This flags the important issue that pursuing the highest possible group-level growth rates in a heterogeneous population will increase inequality. Conversely, pursuing growth such that poorer individuals enjoy higher rates leads to more equitable outcomes in distribution but subtracts from the average G because the covariance is negative.

To complete the derivation, we now characterize the mean and stochastic components of G. We express the individual growth rate, as in the previous section, j=j+j, which leads to G=1Gj=1Gj=1Gj=1G(j+j)=G+G, where G is the group mean of individual temporal means and G is a stochastic noise term resulting from the group average of the errors for each individual. The properties of G are inherited from those of each agent and their statistical correlations. The mean remains zero, while the variance is given by G2=1G2j,kGjkjk, where j and k are the volatilities for agents j and k and jk is the correlation matrix between them. (The correlation matrix is symmetric, with 1 jk 1 and with ones along the diagonal, corresponding to each agents squared volatilities).

In the simplest case, when errors are statistically independent across agents, jk = 0 for k j, and if all SD are the same, k2=r2, we have that G=1Gj=1GjG2=1Gr2. Then, the magnitude of fluctuations is reduced by group size and vanishes in the infinite G limit. Thus, if errors are independent across individuals, both long times and large-population pooling leads to a convergence to the behavior set by the temporal means. This, curiously, implies that the group average grows faster than the agents temporal average in general and provides a strong quantitative argument for pooling resources either via government action or risk management instruments, such as insurance (26).

The case of nonindependent variables is interesting because the treatment of the last section suggests that it would follow from different agents either experiencing correlated fluctuations and/or generating coordinated institutional control responses, which is likely in many circumstances. When all variables are fully correlated jk = 1 and G2=r2, the volatility associated with rG becomes independent of group size. In urban settings, we may expect some correlation between agents as they experience a common spatial and socioeconomic environment of the city. For U.S. MSAs, G2 is approximately constant in G (see fig. S3).

The covariance term between individual growth rates and resources adds additional correctionscovarG(,rrG)=[1Gj=1G(jG1)(rjrG1)]G+[1Gj=1G(jG1)(rjrG1)]G=covarG(G,rrG)G+covarG(G,rrG)G(12)

With these results in hand, we can now write the time evolution of average group resources asdrGdt=GrG,withG=G+G,G=[1+covarG(G,rrG)]G,G=[1+covarG(G,rrG)]G(13)

We see that the statistical behavior of rG is set by the dependencies of these quantities (see section S3 for discussion). When G and G2 are independent of rG (but may depend on G and t) and G obeys the conditions of the central limit theorem, the population average resources rG will follow a multiplicative random growth process (Fig. 3D). This process, similar to Eq. 4, will then integrate to givelnrG(t)rG(0)=(GG22)t+GW(t)(14)showing that if W(t) converges to a normal variable as the result of the central limit theorem, then the statistics of rG(t) become lognormal at long times (see section S3 for an example and further discussion of necessary conditions, exceptions, and related results) (29). It is important to stress that growth rates and volatilities now run (i.e., change) with group sizes, G, and time, t, depending on the correlations captured by the several covariance terms (see Fig. 3D).

We are now ready to express the quantities in scaling relations as functions of stochastic growth rates. This will provide us with a statistical theory that derives urban scaling beyond mean-field calculations (4). To keep the notation simple, the index i denotes cities. We take each city to be a group with G = Ni and write the simplified notation i=Ni, i=Ni, and so on. We will also write the averages of these quantities over the ensemble of cities as = i (see section S6 for a summary of notation).

Running scaling exponents and the emergence of scale invariance. Let us see when a power law scaling relation is a conserved quantity of the stochastic growth dynamics. We start with the integral trajectory for total resources, Ri(t), lnRi(t)Ri(0)=it+iW(t). This equation is ergodic in the sense of stochastic population dynamics because long-time averages coincide with ensemble averages (30)(1tlnRi(t)Ri(0)i)2=i2W2(t)t2i2t0(15)

This property of long-time means specifies necessary conditions for scaling to hold over time. To see this, define BiBi(lnNi)didlnNii(t)=BidlnNi(0)+BilnNi+O[(lnNi)2](16)where (0) is independent of time and scale. Bi varies slowly with lnNi so that Bi is also independent of scale but could depend on time. Bi is analogous to a beta function expressing the change (running) of a coupling with scale in statistical physics (15). Replacing it into Eq. 15 obtainsRi(t)Ri(0)eitY0(t)Ni(t)+Bit(17)which shows that if Bi is nonzero, then the scaling exponent (t)+Bit becomes time dependent in general and is not conserved by the dynamics of growth. Scaling relations will then vary over time, becoming steeper (larger exponent), if Bi>0, or shallower, if Bi<0. It is also possible that the integral Eq. 16 yields a more complicated function of lnNi and time. Under time averaging and control, it is natural for Bi1/t as we have seen, resulting in a time-independent change of scaling exponent.

To see this, consider that the volatility i22 in the effective growth rate is, in general, both time and population scale dependent, while the mean i is independent of both. This means that, in most circumstances, Bi(lnNi)=12di2dlnNi, which should be small because of the agents control over fluctuations. Consider the example i2(Ni)=r2tNi, Bi=2i2(Ni), which leads to the exact result, r22NlnN. This shows that the scaling exponent , while time independent, increases with city size, N. In this case, only at sufficiently large N>>(r2/2)1/ will the value of coincide with that predicted by mean-field scaling theory (4). This is not an issue if r2 is small. Otherwise, for smaller cities, may become measurably smaller than for larger ones. Because the magnitude of variations away from scaling is urban system and quantity dependent, this may help account for some variations of observed scaling exponents in different nations and for different urban properties (31, 32). It also implies correlations between the behavior of the prefactor, Y0(t), the variance, and the scaling exponent , as noted recently in (33).

These results show that strict scaling invariance is predicated on Bi=didlnNi0, which is analogous to a renormalization group fixed point in statistical mechanics (34) applied to the population growth rate. Away from this fixed point, we have now shown how to compute corrections to scaling exponents, which are the result of the scale-dependent statistics of growth rates. Last, note that the scale independence of growth rates for cities is a standard assumption known as Gibrats law (or law of proportional growth) (24). This assumption is necessary to derive Zipfs law for the statistics of city sizes. Figure S3 illustrates this general analysis with the growth rates and variances for wages in U.S. metropolitan areas since 1969, showing that the effective growth rates are city size independent to an excellent approximation, justifying the observed persistence of scaling with a time-invariant exponent.

Equations of motion for prefactors and scaling residuals. We now translate stochastic growth into equations of motion for both scaling prefactors and residuals; details of the derivations are given in section S4. For the prefactors, we obtaindlnR0dt=dlnRdtdlnNdt(18)which is a function of only the centers dynamics. Because the centers are averages over all cities, no higher order statistics plays out in these quantities. This dynamics of the scaling prefactor is important because it measures the urban system (nation) wide per capita baseline growth, a form of endogenous intensive economic growth.

For the residuals, we obtaindirdt=(i)ddt(lnNilnN)+(i)=i(NiN)+(i)(19)where Ni=ddtlnNi and N=ddtlnN. This equation has a number of interesting properties: The most important is that it essentially describes a random walk driven by the terms, which set the variance, (r)2. The two other terms enforce the convergence to the population averages in terms of growth rates of resources and population and guarantee that r = 0 is preserved by the urban systems growth dynamics.

The emergent statistics of urban indicators. The statistics of resources follows from integrating Eq. 19, leading to the general expectation that the statistics of the r become normal at long times (see section S4). This means, in turn, that cumulative urban indicators (stocks) are expected by the same argument to be lognormal, as we saw more directly above. Flow quantities, such as income or costs, are often more accessible empirically (Fig. 2). Their statistics follow from the analysis of the previous sections, where we wrote Yi=biRi=(bi+vi)RilnYi=lnRi+ln(bi+vi). Substituting the scaling relations for Ri, Yi, this implies that i(t)=ir(t)+lnR0lnY0+lnbi. Taking averages over cities obtains the constraint lnY0 ln R0 = ln b, which allows us to writei(t)=ir(t)+lnbilnb(20)

This shows that the statistics of income are set by two different processes, the first resulting from the statistics of associated resources and the second due to stochastic returns. The first piece is characterized by the accumulation of variations over time, which entails time averaging and is expected to become approximately normal. The second term is instantaneous and consequently not subject to limit theorems. Hence, it can have more arbitrary statistics.

To see this, we return to the analysis of stochastic returns bi under agents adaptive control to obtain the explicit time evolution equationdi(t)dir(t)+(lnbilnb)dt+[ibidWi(t)bdW(t)](21)where the force dvi/dt was taken here to be white noise dWi (the differential of the Wigner process Wt) with variance i2, as above. If dvi/dt has nonrandom components, the expression is similar but more complicated. Here, dW is the average stochastic force over cities, and we assumed that fluctuations are uncorrelated to population variations in and b. This also implies that the quantity i=ir(lnbilnb) inherits the property of ergodicity from ir. Figure S4 shows the income growth rates for U.S. metropolitan areas over time, including its noise-driven equation of motion and the property for wages where fluctuations away from the mean trajectory of growth fall over time (roughly as 1/t, inset) to become negligible for long times.

Note that in the limit of strong control, at the individual level and/or as an emergent average within cities when the i/bi<<1, the stochastic terms will be small, and the statistics of income will approximate that of resources as a normal distribution for the i. In addition, this derivation leads to a set of quantitative expectations that can be checked against the data: Figure 5A shows that the quantity t22 2t behaves approximately like the displacement of a one-dimensional (1D) random walk. This is well described by the straight line in time with slope given by the variance, although empirically we also observe shorter periods of acceleration or deceleration relative to the main trend. Figure 5B shows an analogous picture depicting each SAMI, i, trajectory, starting all cities at i = 0 in 1969. This demonstrates the spread of the SAMIs over time according to the behavior of a 1D random walk (red line, the same as straight line in Fig. 5A). Figure 5C shows the volatility and mean growth rates for all cities over the 47 years and corresponding estimates from measurements of dispersion over time (Fig. 5, A and B) and over the ensemble of cities: The observed statistical agreement of these two strategies for measuring the square volatility demonstrate the ergodicity of the statistics of i(t) once drift has been removed. Last, Fig. 5 (A and D) shows that the income residuals variance is actually time dependent, spreading very slowly over time as predicted by the derived lognormal part of the distribution. The overall distribution is better described, however, by the sum of two Gaussians, one broad and one narrow, corresponding to the two terms in Eq. 21 (red dashed line in Fig. 2C). It is only because the annual volatilities are so small that this temporal pooling and a deduction of a pure lognormal behavior appeared reasonable for flow variables in earlier work (22, 35).

(A) On the average over cities, the displacement from their initial deviations in 1969 grows linearly (red line) (gradient = 0.00108, 95% confidence interval = [0.00102, 0.00115]; intercept = 2.13279, 95% confidence interval = [2.25885, 2.00672], R2 = 0.93), as expected from pure random diffusion of the growth rates. Note that this is a mean temporal behavior and that there are periods when deviations grow faster or slower. Periods of economic recession are shown in gray. (B) The trajectory of deviations for all cities (different colors) but having set all deviations in 1969 to zero so that all trajectories depart from a common origin. The red line indicates the diffusive behavior, same as in (A), clearly showing that deviations tend to increase in magnitude over time. (C) The prediction of the wage growth volatility for U.S. MSAs by three methods: the fit of (A) and (B) and the averages over time and sets of cities, demonstrating the ergodic character of the statistical dynamics. Shaded areas show the overlapping 95% intervals in these estimates. (D) The distribution of deviations, year by year, using the same color scheme as in Fig. 2 (A and B). We see that, unlike our first approach in Fig. 2C, the width of the distributions is increasing slowly over time (brown most recent) and that the data for wages (a flow) should be fit by a distribution that is well described as the sum of two Gaussians: a universal broad distribution due to resource compounding and a contingent short-term narrow distribution (Eq. 20), which depends on most recent environmental shocks.

We showed how quantitative urban theory can be taken beyond a stationary approach based on an average budget constraint, characteristic of spatial equilibrium. In its place, we proposed the primacy of stochastic growth processes and agents strategic behavior as the dynamical statistical theory from which more particular results follow (Fig. 1). This provides a common foundation for nonequilibrium modeling of cities across scales (17, 18) and shows how these processes are associated with urban scaling and agglomeration effects. From this point of view, we see how the budget condition of spatial equilibrium models becomes the emergent property of a much more fundamental process, whereby agents subject to stochastic resource flows (incomes and costs) must develop adaptive strategies to reduce potentially fatal volatility. This point of departure is both necessary for dissipative complex systems and is very general so that it offers a number of connections with the statistical dynamics of other natural and engineered systems (28, 36, 37).

The key advantage of this bottom-up stochastic approach is that it naturally unifies processes of resource flow management (equilibrium), growth, and statistics. Hence, the framework emphasizes the critical role played by growth rate variations in a number of important urban phenomena. Specifically, we showed how the properties of the growth rate volatility are implicated in the (non-)preservation of urban scale invariance and set the boundary between growth and decay regimes, including the time scale for exponential growth to become manifest as fluctuations average out. In particular, we demonstrated how the general property of ergodicity in population dynamics and formal demography (30) is intimately connected, together with a renormalization fixed point condition on the growth rates, to the emergence of mean-field scaling relations (4).

There are a number of important consequences for urban theory that these results clarify and unify. First, they show how spatial equilibrium is, after all, consistent with observed exponential growth in cities both economic and demographic, which has been an assumption in previous models. Second, they show how to derive macroscopic statistical behavior for cities and urban systems from microscopic strategic choices at the agents level and provide expressions for how to aggregate growth rates over time and populations. Third, this process exposes issues of inequality of wealth and income and how they are compounded over time (23). Specifically, the quantities discussed here show how policies aimed at maximizing aggregate economic growth may naturally deemphasize the relative growth of poorest sections of the population. Last, and in many ways the central motivation of the paper, the results derived here demonstrate that the statistics of most urban indicators are not universal in a simple sense. Rather, they are emergent as the consequence of limit theorems under stochastic (exponential) growth. In particular, the statistics of urban indicators that account for incomes and costs are the result of a mixture of a more universal component, inherited from their association to accumulating quantities and corresponding limit theorems, and a nonuniversal part, arising from the short term hustle (accidents and the quality of the agents control) in variable stochastic environments. Thus, statistical tests to evaluate the Gaussianity of urban (log-)quantities (22, 35, 38), to be meaningful, must be performed with care and explicitly acknowledge the distinct distributions of different urban indicators.

The models for the budget constraint, the growth of resources, and associated control strategies introduced here are standard starting points in demography, geography, economics, and finance (2325). They can clearly be made more complex and, where necessary, also more realistic. The concept of resources and incomes is not 1D. Issues of energy, monetary wealth, knowledge, and social capital all contribute to resource growth in human populations. The extent to which these quantities, which can all be accumulated, interact with each other is critical for a general understanding of human development. Models for the dynamics of volatilities and more sophisticated control of fluctuations and maximization of growth rates may also become important. Some inspiration should be derived from population biology and mathematical finance (16, 25), where such models are more developed. Last, data on detailed expenditures, wealth, and other financial and social characteristics are becoming increasingly available for households at finer temporal resolutions (14) and will be critical to test and improve the ideas introduced here and to identify systematic heterogeneities in agents behavior, e.g., associated with conditions of poverty and uncertainty.

The approach developed here can be applied to other contexts beyond the contemporary United States but requires appropriate contextualization. In all societies, household adaptive management of resources is likely to remain important. However, in more collectivist societies or in those with stronger top-down governance, the aggregate management of benefits and costs will replace, to a larger extent, bottom-up agency. As shown, managing aggregate costs at the societal level can achieve considerable benefits because this strategy minimizes some risk. Its success hinges, however, on the effective investment of resources that generate society-wide benefits and their redistribution and related inequality. In circumstances of low growth, such as in most preindustrial societies, adaptive control of resources and the associated dynamics of volatility provide us with an important window into their (in)stability. This allows us to connect more proximate explanations of collapse, e.g., related to environmental stresses or violence, to the broader collective and political dynamics of societies, expressed as the capacity to manage shocks or disintegrate instead.

Empirically, the U.S. urban system, at least in terms of changes in total wages in MSAs, turns out to be very well-behaved: Its growth volatilities are almost always very small, fluctuations converge to limiting statistics quickly, and scaling relations are conserved over time. However, our theoretical results show that these properties pertain only to quantities and systems of cities with small, population sizeindependent growth rate volatilities. In the United States, over the past nearly 50 years, despite a number of notable events, observed average square volatilities associated with wages and population growth are about one order of magnitude smaller than average growth rates, making their effects almost negligible. It will be interesting in the future to investigate other urban systems and quantities characterized by larger volatility, such as crime or innovation (22, 33, 35), for which the present framework makes a number of testable predictions.

The flip side of the observed constancy and stability of growth rates in American cities is that extant wage disparities become very slow to reverse. The typical square displacement in over nearly five decades (Fig. 5A; 2t) is just 0.054. Assuming a similar rate of change in the past means that the observed variance in deviations from scaling at the beginning of our dataset (in 1969, about 0.043) would have been the product of the previous 40 years, taking us back to the time of the roaring 1920s and the subsequent great depression. Thus, the answer to the question at the beginning of this paper about predicting the magnitude of deviations from scaling in any given year is now recast not so much in terms of parameters of stationary statistics (33). Rather, this variance is the result of accounting for the accumulation of much smaller accidents and variations that make up the stochastic history of cities, which compound short-term noisy growth under partial control of heterogeneous agents over entire urban areas and long-time periods of decades (22). This is the quantitative sense in which history matters for cities, and their development becomes path dependent (18, 39, 40).

Continue reading here:
Urban growth and the emergent statistics of cities - Science Advances

What does your voice say about you?

Your accent can nod to where you come from; the pace of your speech can reveal your emotional state; your voiceprint can be used to identify you.

Linguists, companies and governments are now parsing our voices for these details, using them as biometric tools to uncover more and more information about us.

While a lot of this information is used to make our lives easier, it has also been used to controversial and worrying effect.

And the next frontier of voice technology means we may not be able to trust what we hear even if we appear to have said it ourselves.

Much like a fingerprint, we all have a unique voiceprint.

Its distinctiveness is a product of our differently-shaped bodies, our throats, our larynxes, mouths and tongues.

These days it's common to confirm your identity with your voice to log into your bank, access your ATO file, or unlock your phone.

But the use of voiceprints has a longer history than you might think.

"By the early 20th century, vocal portraits were added to archives of criminal records in police departments across Europe and the United States," says Xiaochang Li, assistant professor of communication at Stanford University.

"These recordings were used for a number of forensic purposes, from identification to physiognomy, which is tied to eugenics."

The portraits were used not so much to identify a single person, but to try and identify characteristics of criminality across a population.

Researchers believed the voice could reveal age and class but also uncertainty, obfuscation or lies.

But use of these portraits failed to take hold, in part due to the technical constraints of reproducing the recordings.

Fast forward to 1944 and the mighty R&D company Bell Labs credited with developing radio astronomy, the transistor, the laser and the photovoltaic cell was researching the sound spectrograph.

It's a visual representation of sound that maps frequencies and their intensities.

They were attempting to map qualities of the voice pitch, rate of speech, missing harmonics to identity.

They couldn't conclusively prove which features signalled markers of identity, but the sound spectrograph would become instrumental in the use of the voiceprint for identification because it's much faster for a computer program to look at an image of sound than to listen in real time.

And now voice ID systems such as Nuance, Verint and Pindrop are replacing customer service representatives in banks, call centres and government agencies around the world.

But what if you are looking to identify qualities in the voice that don't present in a sound snapshot? Qualities that shift and change over your lifetime like your accent?

In the early 2000s, as part of the asylum seeker application process to Australia, applicants who arrived without papers were asked to take a language test.

Not a test to confirm their ability to speak English, but a linguistics test to confirm their country of origin.

Tim McNamara, a professor of applied linguistics at the University of Melbourne, says it was first introduced in Sweden in 1993.

"The Swedish government decided to use the way people spoke as part of the process the likelihood that the person is from where they claim to be from," he says.

If done properly, a Language Analysis for the Determination of Origin (LADO) test would employ a trained linguist to elicit speech samples from the applicant that would place them on one side of a border or another.

Your phone and other personal computing devices know an awful lot about you.

However, in Australia the test was often conducted by another asylum seeker who could speak the language, or a translator who spoke a similar dialect.

"I think language verification results are appalling," says Sanmati Verma, accredited immigration law specialist at Clothier Anderson Immigration Lawyers.

"In the context of displaced refugee communities folks have not been to school, they are living in ethnocised ghettos, they haven't been interacting with members of local population there is no way to determine through language where they are truly from."

Professor McNamara, with a group of international linguists, wrote a set of guidelines that say if LADO is to be used, there are ways to sharpen the tool: use qualified linguists, better understand how language moves across borders, and allow for uncertainty.

"The real problem with LADO is that it is done badly, and it's done cheaply. And the courts want black and white, but there has to be room for doubt," he says.

LADO is still being used by several governments in Europe and the UK, but Australia has exchanged the test in favour of constructing a person's narrative, a process that has also attracted criticisms.

Nonetheless, the decisions within the application process are still being made by humans. What happens when we hand over the reins to automated systems?

Public services such medical systems and government departments are turning to automation to find efficiencies and to meet increasing demand.

Dial the ATO and your voiceprint can be matched in less than 30 seconds by a computer algorithm, forgoing minutes of security questions.

Beyond the simple voiceprint, companies at the frontier of voice tech suggest they can sort for all kinds of different traits, such as our emotional states.

Earshot is about people, places, stories and ideas, in all their diversity.

The next time you dial a call centre you could be triaged by an AI voice for how stressed you sound over the phone.

One company, Clearspeed, even promises to be able to hear risk in the voice, through an automated questionnaire.

CEO Alex Martin says its algorithms are looking for the "acute threat response, in concert with what is known as cognitive effort".

"These are the things that collide and impact speech phonation, resonating impacts of speech and that's what we're detecting and measuring."

For instance, a job applicant answers a set of tailored yes/no questions over the phone, and they are flagged low to high risk. Low risk, they go onto an interview; high risk, their application is reviewed.

"Everyone should be able to say no and have absolutely no problem in their response. Their no looks normal in our model and we score it as low to average risk. But if there is neurophysiological reaction, depending on the amount of reaction it could be scored as potential risk or high risk," Mr Martin explains.

Clearspeed is concerned with clearance problems, assessing hundreds of applications quickly, automatically, and says its tool should be used in constellation with standard practises.

But as more systems are automated, the role of humans and human judgement is diminishing.

Is artificial intelligence sexist?

"Judgement is what happens when we face a decision that doesn't have a pre-coded answer," says Mark Andrejevic, professor of communication at Monash University.

"Most of the meaningful decisions we make in our lives are not syllogistic, [meaning] they cannot be reduced to a set of logical principles building on a few givens."

He cautions against automating too many decisions in our lives.

"Automation is really good for those syllogistic decisions. But we want to avoid handing over the non-syllogistic decisions because it reduces the capacities we need to live a life in common."

And then there's next frontier in voice technology, where we do away with our bodies altogether.

In the past couple of weeks, a new voice synthesis product was released.

Overdub proposes to create ultra-realistic text-to-speech of your own voice.

Using 10 minutes of recorded audio of your voice, it creates a clone of your voice, which you can make say anything you want.

All you have to do is type the desired sentence to the web app, and out it comes your voice, as if you'd said it using your throat, larynx, mouth and tongue.

The implications of text-to-speech synthesis are manifold, exciting and terrifying.

Photo editors made us doubt images, video editors resulted in deep fakes soon we will not be able to trust what we hear. Even if we appear to have said it ourselves.

I created a voice clone to test whether "she" would pass as the narrator of my Earshot documentary.

And the process revealed the elegance and the clumsiness of the tech. To train the AI, I had to repeat prescribed sentences and inflect them with joy or anger or sadness, as if the AI needed emotional training too.

When my voice had been cloned and I heard the first utterances of "her", I was amazed at the reproduction but listening more closely the odd vowels and strange phrasings came into focus, and the uncanniness I first heard started to come undone.

But, in terms of this technology, we are only at the beginning.

Our voices are shifting from representations of us to technological tools. Tools that we can learn how to use to redefine ourselves, for us and for who is listening.

Get more stories that go beyond the news cycle with our weekly newsletter.

Link:
Your voice says a lot about you and AI is listening - ABC News