NEWS & VIEWS

KONRAD WOTHE/MINDEN PICTURES/FLPA

Figure 1 | The strawberry poison frog Oophaga pumilio.  Yang et al.1 reveal that a learning process called imprinting occurs in this species, in which the
behaviour of offspring is influenced by the colour of their mother. Imprinting occurs during the period of parental care of the developing offspring. An example
of maternal care is the transport of tadpoles on their mother’s back, as shown in this image of O. pumilio from Costa Rica.

E VOLUTIO N

Frog mothers’ powerful influence

It has now been found that mothers of a species of frog affect the behaviour of their offspring — influencing female mating
preferences and aggression between males. Such behaviours might lead to the formation of new species. See Letter p.99

MACHTELD VERZIJDEN bright colours warn predators that the frogs that affect mating preference and the genes
are toxic. The colours have another important responsible for the preferred trait. A process

T
he ability of a species to maintain many
different types of individual in a popu-
lation, known as polymorphisms, is
intriguing because rarer forms must have some
selective advantage to be preserved. Establish-
ing diversity in a species through the presence
of polymorphisms can provide a step towards
speciation (the formation of new species).
On page 99, Yang et al.1 reveal a surprising
mechanism that affects the maintenance of
colour polymorphisms in the strawberry
poison frog (Oophaga pumilio). The finding
has implications for our understanding of
evolution.
Oophaga pumilio (Fig. 1) lives in Central
America, and frogs of this species have many
widely differing types of skin colour. These
role, in that females have strong preferences
for the colour of males when choosing their
mate2. Usually, for any given species, females’
mating preference for a type of male drives
a rise in the prevalence of the most popular
female-preferred polymorphism (such as a
male colour). Thus, colour selection by female
mating preference is probably not sufficient to
explain how multiple colour polymorphisms
are maintained in O. pumilio.
Polymorphisms that are perpetuated over
many generations could be a route to the
formation of a separate species if males and
females choose to mate only with their own
type. However, one issue that affects the
chances of such reproductive isolation arising
is the pattern of co-inheritance of the genes
that shuffles genes, termed recombination,
occurs when egg and sperm cells are formed,
and this results in each egg and sperm cell
having a unique combination of the parental
versions of genes. Recombination could dis-
rupt the co-inheritance of the versions of genes
affecting mating preferences and those that
affect the preferred trait3, thereby influencing
the success with which the genes giving rise to
the preference and to the preferred trait might
be co-inherited.
One mechanism that could help to boost
the likelihood of co-inheriting the preference
and the trait together is if individuals learn to
prefer to mate with individuals that are similar
to their parents, because the individuals’ off-
spring would inherit the mating preferences

3 8 | NAT U R E | VO L 5 7 4 | 3 O C T O B E R 2 0 1 9
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 NEWS & VIEWS RESEARCH

and preferred traits of their parents. Such
mating behaviour can arise through a process
called sexual imprinting, in which young off-
spring learn to recognize their parents during a
period of parental care and later use this learnt
information to select a mate similar to their
parents.
Sexual imprinting is common in bird
species, and also in some species of mammal
and fish. It provides a way of preventing
recombination from breaking the association
between mating preference and the preferred
trait genes4. Imprinting can speed the estab-
lishment of reproductive isolation, which
might lead to speciation. To identify factors
governing the establishment of a variety of
colour polymorphisms in O. pumilio, Yang and
colleagues investigated whether imprinting
occurs in this species, a behaviour that hasn’t
previously been reported in amphibians.
Females of O. pumilio lay eggs on the
ground, on a leaf covered by other foliage,
where they are fertilized by the male. During
the following week, the male ensures that the
eggs stay wet, and after the eggs have hatched,
the female takes over the parental care. She
carries each tadpole on her back (Fig. 1) to a
water-filled bromeliad plant, and then returns
to feed the tadpole with her unfertilized eggs
until it is sexually mature.
The authors studied three colour types of
O. pumilio, and carried out laboratory experi-
ments involving three set-ups: tadpoles were
raised by their biological parents, which were
both the same colour; they were raised by their
parents, which were of different colours; or
they were raised by foster frogs that were not
the same colour as the tadpoles’ parents. For
all three scenarios, when the female tadpoles
became adults, female offspring preferred to
mate with males of the same colour as the
mother that had reared them.
Yang and colleagues demonstrated that
male offspring had an imprinted behaviour,
too. These frogs biased their territorial aggres-
sion towards males of the same colour as the
mother that had reared them. Using simula-
tions, the authors showed that, over many
generations, the effect of imprinting on male
and female behaviour has opposite effects on
the fitness, and thus the prevalence, of a frog of
its mother’s colour in the population. If a male
is the same colour as a female’s mother, the
maintain multiple forms of a particular trait in
a population5.
A factor not considered by Yang and
colleagues in their work is the role of natural
selection owing to the frogs’ bright colours.
These frogs might be targeted by a range of
predators. Predators often learn to recognize
and associate particular colour patterns with
toxicity through personal experience with a
toxic prey. Thus, variation in such colours
could limit the ability of a given colour to act
as a warning because predators would need
to learn to recognize each different warning
colour6. Predation is therefore likely to boost
the selection of the most common colour.
Nevertheless, several forms of these frogs
are equally toxic and conspicuous to their
predators7. There is a hypothesis8 that when
populations are sufficiently toxic and con-
spicuous, predators will be able to generalize
across such bright colours and recognize
them as being toxic. Sexual selection would
then be free to drive the evolution of other
bright colours.
It is interesting that both natural selection
and sexual selection are affected by learning in
the various interacting individuals — the frogs
and the predators. Future experiments might
investigate to what extent predators have a role
in affecting the prevalence of the different frog
colours.
The mechanisms that Yang and colleagues
reveal to be acting in O. pumilio populations
show how intricately natural and sexual
selection affect processes that might drive
C O N D EN SE D- M AT T E R P H YS I CS
Exotic state seen at
high temperatures
The phenomenon of Bose–Einstein condensation is typically limited to extremely
low temperatures. The effect has now been spotted at much higher temperatures
for particles called excitons in atomically thin semiconductors. See Letter p.76
A N D R E Y C H AV E S & D AV I D N E I L S O N
speciation, and indicate that neither process
can necessarily be considered separately9.
In this frog species, imprinting inextrica-
bly links both female mate preferences and
inter­actions between males, ensuring that
the prevalence of these imprinted behaviours
tracks extremely closely to the frequency of the
particular parental colour form in the popu-
lation. Previous work has shown that sexual
imprinting favours leading a population on a
path towards reproductive isolation4. The evi-
dence obtained by Yang and colleagues now
shows how imprinting can also affect intra-
sexual aggression and might help to maintain
polymorphisms, thereby giving an extra boost
for conditions that favour speciation. ■
Machteld Verzijden is in the Department of
Bioscience, Aarhus University, 8000 Aarhus C,
Denmark.
e-mail: machteldverzijden@bios.au.dk
1. Yang, Y., Servedio, M. R. & Richards-Zawacki, C. L.
Nature 574, 99–102 (2019).
2. Reynolds, R. G. & Fitzpatrick, B. M. Evolution 61,
2253–2259 (2007).
3. Felsenstein, J. Evolution 35, 124–138 (1981).
4. Verzijden, M. N. et al. Trends Ecol. Evol. 27, 511–519
(2012).
5. Tinghitella, R. M. et al. Behav. Ecol. 29, 783–797
(2018).
6. Mappes, J., Marples, N. & Endler, J. A. Trends Ecol.
Evol. 20, 598–603 (2005).
7. Maan, M. E. & Cummings, M. E. Am. Nat. 179,
E1–E14 (2012).
8. Cummings, M. E. & Crothers, L. R. Evol. Ecol. 27,
693–710 (2013).
9. Maan, M. E. & Seehausen, O. Ecol. Lett. 14, 591–602
(2011).
are bound states of a negatively charged
electron and a positively charged hole

probability that the female will mate with the
male is boosted. However, when that colour
becomes the most common type in the popu-
lation, such males incur a survival penalty by
being subject to competitive aggression from
other males of the same colour. This aggres-
sion could explain how an alternative rare
colour could persist in a population because,
compared with males of the common colour,
males of the rare colour would instead spend
less time and energy on territorial defence, and
presumably expend this energy and time on
attracting females, increasing their chances of
mating. Such ‘rare-male advantage’ can help to
A
t sufficiently low temperatures, large
assemblies of particles that are classi-
fied as bosons condense into a single
quantum state. This remarkable phenom-
enon, known as Bose–Einstein condensation
(BEC), can allow the particles to become a
superfluid, whereby they flow without fric-
tion. Super­fluidity has been seen in gaseous
helium-4 and in ultracold atoms, but only at
extremely low temperatures (a few kelvin). In
the past few decades, there have been many
attempts to achieve high-temperature BEC
in semi­conductors using electrically neutral
composite particles called excitons, which
(electron vacancy). On page 76, Wang et al.1
report compelling experimental evidence
that charge-separated excitons in a pair of
atomically thin semi­conductors can exhibit
BEC at temperatures as high as 100 K.
When an electron is excited from the
‘valence’ energy states of a semiconductor
material to higher-energy conducting states,
it leaves behind a hole. The electrostatic
attraction between electrons and holes can
bind them into excitons. Separately, electrons
and holes are particles that are classified as
fermions, which cannot form Bose–Einstein
condensates. But because a bound state of two

3 O C T O B E R 2 0 1 9 | VO L 5 7 4 | NAT U R E | 3 9
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 RESEARCH NEWS & VIEWS

the exciton density. This intensity also has a

large enhancement that is strongly peaked at
equal electron–hole densities. Such sensitiv-
ity of the electroluminescence enhancement
to charge imbalance is consistent with exciton
condensation.
Graphene gate These tunnelling and electroluminescence
Hexagonal characteristics persist up to temperatures of
Tungsten Hole boron nitride
diselenide about 100 K. Wang et al. therefore interpret
this temperature as the transition temperature
for BEC in this system, consistent with previ-
Exciton ous predictions5. Earlier this year, a theoreti-
Tunnelling cal investigation of superfluidity specifically
Molybdenum for this molybdenum diselenide–tungsten
diselenide diselenide system7 reported properties that
Electron Light
quantitatively align with those in the current
experiment, lending further credence to the
authors’ conclusions.
In 2D systems, the transition temperature
for BEC is generally higher than that for
Figure 1 | Excitons in a semiconductor device.  Wang et al.1 study monolayers of the semiconductors superfluidity. Above the superfluid transition
tungsten diselenide and molybdenum diselenide that are separated by sheets of the two-dimensional temperature, there are disconnected regions of
electrical insulator known as hexagonal boron nitride. The two ends of this device consist of electrodes superfluidity that persist up to the BEC transi-
called gates that are made from graphene (a 2D form of carbon). The device contains composite particles tion temperature. Both of these temperatures
known as excitons, which are bound states of an electron vacancy (a hole) and an electron. A hole in one increase linearly with exciton density, at rates
monolayer can tunnel to meet an electron in the other monolayer, and merge with it, releasing energy in
the form of light. The authors demonstrate that, at temperatures up to about 100 kelvin, several features
that depend on the electron and hole effective
of this light depend only on the exciton density. This observation suggests that the excitons are in a single masses, but the BEC temperature increases
quantum state — a phenomenon known as Bose–Einstein condensation. much more quickly than does the superfluid
temperature. For this reason, Wang and col-
leagues’ measurements, which are sensitive
fermions is a boson, excitons can condense. the quantum wells with atomically thin sheets to BEC, reveal condensation at high exciton
The effective masses of electrons and of the semiconductor graphene (a two- densities (of about 1012 per square centimetre)
holes (the masses that these particles seem dimensional form of carbon), the separation at temperatures up to 100 K.
to have when responding to electrical forces) between the layers could be dramatically Superfluidity could not be probed directly
are much smaller than those of atoms. As a reduced, thereby greatly strengthening the elec- in the current experiment. But this is an
result, excitons can condense at much higher tron–hole pairing interactions3. To decrease exciting challenge for future work. Conclu-
temperatures than can ultracold atoms. In the detrimental effects of screening, research- sive evidence for superfluidity could come
addition, the energies needed to split excitons ers proposed alternative atomically thin semi- from direct measurements of electric current
into electrons and holes are greater than the conductor layers; namely, two sheets of bilayer flowing in opposite directions along the two
thermal energy of the excitons even at room graphene4 or of molyb­denum disulfide5. For layers, using independent electrical contacts
temperature, so excitons can be stable at this the bilayer graphene, strong experimental for these layers.
temperature. signatures of BEC were reported last year6, but The authors’ molybdenum diselenide–
When electrons and holes are in the same at temperatures of only about 1 K. tungsten diselenide double layer is a straight-
spatial region of a semiconductor, there is a Wang and colleagues studied charge- forward semiconductor system. As a result, it
high chance that they will recombine (merge), separated excitons in a device that comprises is suitable for future condensate-based opto-
releasing energy in the form of light through a two semiconducting monolayers (one of electronics and ultrafast devices, and paves
process called electroluminescence. Such elec- molybdenum diselenide and the other of the way for the search for exciton-mediated
tron–hole recombination usually happens so tungsten diselenide), separated by a few atomic high-temperature superconductivity. ■
quickly that BEC cannot occur, but recombina- layers of the 2D electrical insulator known as
tion can be avoided by separating the electrons hexagonal boron nitride (Fig. 1). This material Andrey Chaves is in the Department of
and holes into two adjacent semiconductor limits the tunnelling of charges between the Physics, Federal University of Ceará, Fortaleza
layers. However, despite expectations, experi- two monolayers to suppress electron–hole 60455-900, Ceará, Brazil. David Neilson is
ments over the past few decades that have recombination. The two ends of the device in the Department of Physics, University of
searched for high-temperature BEC of excitons consist of electrodes called gates that are made Antwerp, 2020 Antwerp, Belgium.
have been largely unsuccessful. from graphene. By combining voltages applied e-mails: andrey@fisica.ufc.br;
In 2002, charge-separated excitons were to the two gates and a voltage applied to the david.neilson@uantwerpen.be
produced by confining electrons and holes tungsten diselenide layer, the electron and
1. Wang, Z. et al. Nature 574, 76–80 (2019).
in two disconnected slabs (quantum wells) hole densities in each monolayer can be tuned 2. Butov, L. V., Lai, C. W., Ivanov, A. L., Gossard, A. C. &
of the semiconductor gallium arsenide, independently. Chemla, D. S. Nature 417, 47–52 (2002).
separated by a slab of another semiconduc- The authors detected a large electric cur- 3. Min, H., Bistritzer, R., Su, J.-J. & Macdonald, A. H.
tor, aluminium gallium arsenide2. However, rent associated with tunnelling of charges Phys. Rev. B 78, 121401(R) (2008).
4. Perali, A., Neilson, D. & Hamilton, A. R. Phys. Rev.
the electron–hole pairing interactions in between the layers. This current depends only Lett. 110, 146803 (2013).
such systems are weak because it is difficult on the exciton density, suggesting that the 5. Fogler, M. M., Butov, L. V. & Novoselov, K. S. Nature
to get the quantum wells close to each other excitons coordinate strongly with each other. Commun. 5, 4555 (2014).
6. Burg, G. W. et al. Phys. Rev. Lett. 120, 177702 (2018).
and because the inter­actions are screened The electron–hole recombination induces 7. Conti, S., Van der Donck, M., Perali, A.,
(weakened) by the presence of the other charges. strong electroluminescence, the intensity of Peeters, F. M. & Neilson, D. Preprint at http://arxiv.
It was later recognized that, by replacing which has a critical threshold that depends on org/abs/1909.03411 (2019).

4 0 | NAT U R E | VO L 5 7 4 | 3 O C T O B E R 2 0 1 9
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 NEWS & VIEWS RESEARCH

FIS H ERIES with international development organizations

or institutions such as the United Nations —

Micronutrient richness would be doing everything possible to encour-

age the domestic consumption of fish caught
in the EEZs of these countries. However, most

of global fish catches

economic-development policies, including
those of these countries themselves, are geared
towards promoting fish exports to match the
insatiable demand for fish in the markets of
Analysis of the nutrient composition of fish caught around the globe reveals high-income Western countries and East Asia3.
locations where the retention of fish for consumption by local populations could The waters surrounding developed
help to tackle human disease caused by nutrient deficiencies. See Letter p.95 countries became overfished before over­
fishing began to occur in other countries.
For example4, the combined fisheries’ catch
D A N I E L PA U LY from foods such as maize (corn) or rice. in the North Atlantic peaked in 1975, and the
Hicks and colleagues’ data demonstrate world’s catch peaked in 1996. The catch limits

E
ating fish is good for us. Fish are a source
of micronutrients that help to prevent
nutrient-deficiency diseases, which are
a leading cause of infant deaths worldwide.
Determining whether the consumption of
locally caught fish could reduce the incidence
of nutrient-deficiency diseases in countries
particularly affected by this problem requires
having access to the relevant data. On page 95,
Hicks et al.1 report their assessment of the
nutritional content of 367 species of fish. For
43 countries, the authors mapped the relation-
ship between the fish-derived nutrients avail-
able from fisheries’ catches and the prevalence
of nutrient-deficiency diseases in communities
living within 100 kilometres of the coast.
When assessing the nutritional composition
of fish stocks, Hicks and colleagues focused on
six crucial micronutrients: calcium, iron, zinc,
selenium, omega-3 and vitamin A. They also
considered protein content. Using some previ-
ously available data, the authors generated a
model that could correctly predict the levels
of these nutrients in different species of fish.
By mining databases containing information
SYLVAIN CHERKAOUI/REUTERS
that fisheries’ catches in some developing
countries should be enough to meet the key
micronutrient needs of their populations. For
example, more than 75% of the population in
Namibia is at risk of calcium deficiency, even
though enough fish is caught there to remedy
this situation. In some cases, ensuring that
even a fraction of a country’s total fish catch
is retained for local consumption could have
a substantial impact on public health. This is
particularly true for children under five years
old, during a crucial stage of their develop-
ment when micronutrient deficiencies have
a severe effect. For 22 of the countries that
Hicks and colleagues studied, 20% or less of the
fish caught could provide enough key micro­
nutrients to meet the needs of all children
under five years old.
Not only do nutrient shortages harm public
health, but this problem has a knock-on effect
of lowering gross domestic product. It might
be supposed, then, that the governments of
developing countries in the tropics — along
placed on overfished regions has led such
regions on a quest to obtain their fish from
other sources. These days, much of the haul in
many parts of the developing world is either
caught by local fishermen and exported, or
taken by foreign fleets — which, by paying a
nominal fee to access the EEZs of developing
countries, catch fish for their own markets.
Such actions contribute to the scarcity of
nutrients in many developing countries.
This problem is perhaps greatest for coun-
tries on the northwestern coast of Africa.
There, fishing by fleets from the European
Union, Russia and East Asia — and high
fish exports to the EU — have led to local
fish scarcity and price increases that have
made fish increasingly inaccessible to local
consumers5. In Senegal in western Africa,
one of the countries studied by Hicks and col-
leagues, a small micronutrient-rich, herring-
like fish called sardinella has been a staple for
centuries. A 2016 documentary film called
An Ocean Mystery: The Missing Catch (see

about fisheries’ catches taken between 2010

and 2014, the authors gathered information
about the amount and type of fish caught
in each country’s exclusive economic zone
(EEZ) — the area of its coastal waters over
which it has sovereign fishing rights. Hicks
and colleagues used their model to estimate
the nutrients available from these fish catches
and thus determine the spatial pattern of this
nutrient availability in global fish catches. For
example, they noted that tropical fish have
higher concentrations of calcium, iron and
zinc than have fish from other regions.
In developing countries around the tropics,
fish are not usually just another healthy
complement to an already rich assortment
of foodstuffs. Rather, for millions of people
living in these regions, fish add the missing
micro­nutrients and proteins to what would
otherwise be an un­balanced diet. In many
developing countries, fish are the food source2
that provides the majority of the inhabitants Figure 1 | Sardinella fish being processed in Mauritania to generate fishmeal. In many developing
with most of the micronutrients studied by tropical countries, a substantial proportion of local fish catches are either exported for human consumption
the authors. The protein from fish boosts the or processed locally to generate fishmeal that is then exported and used, for example, to feed farmed fish.
nutritional content of typical diets in such Hicks and colleagues’ analysis1 suggests that the retention of fish for local consumption could help tackle
countries, where calories are obtained mainly human disease associated with nutrient deficiencies in countries where such conditions are prevalent.

3 O C T O B E R 2 0 1 9 | VO L 5 7 4 | NAT U R E | 4 1
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 RESEARCH NEWS & VIEWS

go.nature.com/2kyjv51) shows sardinella
being smoked, dried and hand-processed by
Senegalese women and then trucked to the
interior of the country, where these fish are
the only affordable main source of micro­
nutrients and animal protein. The leader of
these workers emphasized in an interview in
the documentary that it would be a catastro-
phe if the sardinella supply was interrupted,
because they would have no fish to process.
Since then, this feared catastrophe has
begun to happen. Despite much local
consterna­tion, more than 40 industrial fish-
processing plants have been built, mainly by
Chinese enterprises, along the coast of Senegal
(see go.nature.com/2kva8bu) and neighbour-
ing countries (see go.nature.com/2jtmcjq).
These plants process sardinella (Fig. 1) and
similar small fish into an animal-feed product
called fishmeal. Many of the local fisheries,
which had traditionally supplied the regional
markets with sardinella for human consump-
tion, now instead supply the fishmeal plants.
These factories export their product mainly
to China, which is the world’s largest fishmeal
importer, and it is commonly used there to
feed farmed fish.
Thoughtful consumers often insist that
they eat fish certified as sustainably caught.
This nebulous term often implies a hope that
such fish suffered as little as possible, and that
their stocks are somehow being managed to
ensure the continuation of an abundant supply.
If such fish come from fish farms, as is the case
for most salmon on offer, this, too, is consid-
ered a good thing, because it is widely thought
that fish farming relieves pressure on capture
fisheries. However, using sardinella to make
fishmeal for farmed fish does not reduce the
pressure on wild fish. Moreover, it deprives
people in the developing world on low incomes
of previously affordable, nutritious local fish
— to aid the production of costly farmed fish
that is mainly consumed in high-income
countries2.
When considering what fish we should eat,
given that fish is good for us, it is time to take
a broader perspective about how “us” is being
defined. Hicks and colleagues’ work points
a way forward. The information they have
provided could be used to put a spotlight on
fish availability when thinking of ways to pre-
vent human disease caused by micronutrient
deficiencies. ■
Daniel Pauly is at Sea Around Us, Institute
for the Oceans and Fisheries, University
of British Columbia, Vancouver, British
Columbia V6T 1Z, Canada.
e-mail: d.pauly@oceans.ubc.ca
1. Hicks, C. C. et al. Nature 574, 95–98 (2019).
2. Golden, C. D. et al. Nature 534, 317–320 (2016).
3. Swartz, W., Sumaila, U. R., Watson, R. A. & Pauly, D.
Mar. Policy 34, 1366–1373 (2010).
4. Pauly, D. & Zeller, D. Nature Commun. 7, 10244
(2016).
5. Thiao, D., Leport, J., Ndiaye, B. & Mbaye, A.
Aquat. Living Resour. 31, 25 (2018).
This article was published online on 25 September 2019.

SYNT H ESIS product quickly), and generates new chemical

bonds between two molecules. Ideally, the

Chemical libraries reactants should be used in a one-to-one ratio,

rather than with an excess of one or more com-
ponents (which is a common requirement

from a double click

for many reactions). Click reactions must be
high-yielding, applicable to a broad range of
compounds, and yet exceptionally selective,
meaning that the chemical groups that undergo
Operationally simple chemical reactions, termed click reactions, are widely used the reaction must react only with each other,
in many scientific fields. A streamlined synthesis of compounds called azides and not with any other groups in the reactants.
looks set to expand the role of click chemistry still further. See Letter p.86 The product should also be easy to isolate or
use without extensive purification. Although
many synthetic reactions meet some of these
JOSEPH J. TOPCZEWSKI & EN-CHIH LIU that optimization can involve the laborious, criteria, surprisingly few meet all of them.
resource-intensive synthesis of hundreds, or In 2002, two research groups independently

G
enerating molecules and materials
that have desirable functional prop-
erties is arguably the central goal of
synthetic chemistry. For example, drugs are
developed to have a set of physical and phar-
macological properties that can treat a specific
disease safely. On page 86, Meng et al.1 report
a reagent that greatly simplifies the synthesis
of compounds known as azides, and thereby
opens up a remarkably straightforward route
to making libraries of compounds that might
have useful biological functions.
Altering the structures of molecules to tune
their properties is much more complicated
than modifying objects in the everyday world.
In carpentry, for instance, the same starting
materials (timber, nails and screws) and tools
(saws, hammers and screwdrivers) can be used
to construct objects that have diverse shapes
and functions, such as chairs, doors and crates.
By contrast, building structural analogues of
molecules often requires very different starting
materials (reagents) and tools (reactions). The
need to develop a range of synthetic routes to
such analogues can be a bottleneck when opti-
mizing functional molecular properties2, given
even thousands, of structural analogues. reported4,5 that copper(i) salts are effective
A way of streamlining the optimization of catalysts for reactions known as alkyne–azide
desired functional properties was formalized cycloadditions (the copper-catalysed reaction
in 2001, in a concept known as click chemi­ is abbreviated as CuAAC). These reactions link
stry3. A reaction is defined as click chemistry an azide group (N3) with the carbon–carbon
if it is operationally simple, is ‘spring-loaded’ triple bond in an alkyne compound to form
(thermo­dynamically driven to produce a single a triazole ring (Fig. 1). Because the CuAAC
–
– N
N
+
N N N
N+
O R NH2 a R N b R N
N + R′′
S
F A few minutes, Copper(I) salt,
O R′ room temperature R′ R′
R′′
Fluorosulfuryl Primary Azide Triazole
azide amine
Figure 1 | A two-step click-chemistry sequence.  a, Meng et al.1 report that a reagent called fluorosulfuryl
azide rapidly converts almost any primary amine into an azide at room temperature — a type of reaction
known as diazotransfer. The reactions are fast and high yielding, and the reagent does not react with
chemical groups other than amines; they therefore fulfil the criteria to be categorized as ‘click’ reactions.
b, The authors show that the resulting azide solution can be used without purification in a copper(i)-
catalysed click reaction with alkynes (compounds that contain carbon–carbon triple bonds) to produce
products called triazoles, which are potentially useful in drug discovery. R, Rʹ and Rʹʹ represent any
chemical group or molecular fragment.

4 2 | NAT U R E | VO L 5 7 4 | 3 O C T O B E R 2 0 1 9
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 NEWS & VIEWS RESEARCH

reaction fulfils all of the click criteria, it has
become the poster child for click chemistry.
It was the first click reaction to be widely
adopted, and is now used in applications span-
ning many disciplines, from materials science
to chemical biology6,7.
Several other click reactions have emerged
over the past few years. Of particular note is
one known as sulfur(vi)–fluoride exchange
(SuFEx), which links an oxygen or nitrogen
atom to an SO2F group. SuFEx is generally
recognized as a second category of click
reaction8,9 (unlike other click reactions, it is not
a cyclo­addition process), and has been used in
a diverse range of chemical transformations9,10.
Despite the power of CuAAC reactions, their
applications would be even broader if structur-
ally complex, azide-containing compounds
were more widely available. Conventionally,
organic azides are synthesized by replacing
a molecular fragment called a leaving group
with an azide group; the leaving group can be
a variety of chemical groups or just a single
atom. However, the azide anions used in these
substitution reactions are highly nucleophilic
(electron-rich) and therefore very reactive.
Substitutions with azide anions are thus often
incompatible with having other chemical
groups in the molecule. Furthermore, the leav-
ing group often needs to be made in advance
from an alcohol group (OH), which can be dif-
ficult or impossible to achieve selectively on
molecules that contain many chemical groups.
Alternatively, azides can be prepared from
primary amines (compounds that contain
NH2 groups) by a ‘diazotransfer’ reaction. Until
now, the state-of-the-art reagent used to carry
out diazotransfer had been trifluorometh-
anesulfonyl azide11 (CF3SO2N3). However,
the reactions often require an excess of this
re­agent, are slow, and do not always proceed to
door to a highly efficient and general two-step
method for converting primary amines — a
common chemical group in organic mol­
ecules — into triazoles. Notably, this method
does not require the amines to be modified in
advance to prevent unwanted side reactions at
other chemical groups; nor does it require the
intermediate azides to be purified.
Triazoles are functional mimics of the
amide bond12, which is found in many phar-
maceutical agents and in all proteins. Triazoles
can also function as surrogates for sugars in
poly­s accharides13. Meng and co-workers’
chemistry could therefore be used to synthesize
well-characterized libraries of complex small
molecules and biomacromolecules from read-
ily available precursors. More broadly, the work
brings us a step closer to the vision laid out by
the pioneers of click chemistry3,5: the develop-
ment of a few operationally simple reactions
that use common precursors to rapidly gener-
ate diverse libraries of (bio)molecules that have
desirable functional properties. ■
N EUR O SC I E NCE
A daily rhythm in
colour preference
Behavioural and genetic experiments have revealed that fruit flies prefer green
light over other colours in the morning and evening, and always avoid blue.
These colour preferences rely on different mechanisms. See Letter p.108
CHARLOTTE HELFRICH-FÖRSTER
Joseph J. Topczewski and En-Chih Liu are
in the Department of Chemistry, University of
Minnesota, Minneapolis, Minnesota 55455,
USA.
e-mail: jtopczew@umn.edu
1. Meng, G. et al. Nature 574, 86–89 (2019).
2. Campos, K. R. et al. Science 363, eaat0805 (2019).
3. Kolb, H. C., Finn, M. G. & Sharpless, K. B.
Angew. Chem. Int. Edn 40, 2004–2021 (2001).
4. Rostovtsev, V. V., Green, L. G., Fokin, V. V. &
Sharpless, K. B. Angew. Chem. Int. Edn 41,
2596–2599 (2002).
5. Tornøe, C. W., Christensen, C. & Meldal, M.
J. Org. Chem. 67, 3057–3064 (2002).
6. Meldal, M. & Tornøe, C. W. Chem. Rev. 108,
2952–3015 (2008).
7. Hein, J. E. & Fokin, V. V. Chem. Soc. Rev. 39,
1302–1315 (2010).
8. Dong, J., Krasnova, L., Finn, M. G. & Sharpless, K. B.
Angew. Chem. Int. Edn 53, 9430–9448 (2014).
9. Barrow, A. S. et al. Chem. Soc. Rev. 48, 4731–4758
(2019).
10. Schimler, S. D. et al. J. Am. Chem. Soc. 139,
1452–1455 (2017).
11. Cavender, C. J. & Shiner, V. J. J. Org. Chem. 37,
3567–3569 (1972).
12. Pedersen, D. S. & Abell, A. Eur. J. Org. Chem. 2011,
2399–2411 (2011).
13. Tiwari, V. K. et al. Chem. Rev. 116, 3086–3240 (2016).
blue, green or red filters. The authors placed
food at one end of each tube, and varied the

completion, with 60–70% as the typical yield.
Meng et al. have addressed these limitations
by developing a more efficient diazotransfer
reagent, fluorosulfuryl azide (FSO2N3). They
report that it reacts with almost any primary
amine in a one-to-one ratio, achieving a nearly
100% yield of the corresponding azide. The
authors demonstrated the reagent’s substrate
scope and practicality by using it to make a
library of 1,224 azides in 96-well plates. It is
notable that 49% of these azides had not been
synthesized before, according to the authors’
literature search.
The number of azides synthesized is
impressive (see Supplementary Information
Section 6 of the paper1 for a full list), but the
most striking aspect of this study is the sub-
strate scope: the reaction works for different
amine subclasses, on complex molecules, and
in the presence of various chemical groups.
More­over, Meng and colleagues’ diazotrans-
fer reaction meets the speed, breadth and
efficiency criteria for click chemistry.
In addition, the authors demonstrated
that the prepared azide solutions can be used
directly in CuAAC reactions. This opens the
C
olour vision helps animals to find
nutritious food, to avoid poisonous
animals and, in some cases, in social
interactions1. Colour can affect people’s mood,
and their colour preferences might reflect
current emotional and physiological states2.
Colour preferences also seem to vary through
the seasons3. Lazopulo et al.4 show on page 108
that the fruit fly Drosophila avoids blue light,
and prefers green light to red light at different
times of the daily 24-hour cycle. The authors
also pinpoint separate mechanisms for these
behavioural responses.
Light influences various behaviours in
insects, and fruit flies serve as a model
in which to study the mechanisms underlying
this effect. Flies avoid or are attracted to light
depending on its intensity and colour and the
duration and time of day of the exposure5,6.
However, it is unclear whether fruit flies have
intrinsic colour preferences, as do primates7
and, if so, how these preferences are mediated.
Lazopulo et al. analysed video recordings of
the position and movements of individual flies
living in glass tubes, each of which contained
three equally sized zones that were covered by
order of the coloured zones along the tubes to
avoid misinterpreting flies’ preference for the
zone that contained food as reflecting a colour
preference. To simulate the day–night cycle,
the flies were kept in light–dark conditions
(12 hours of light and 12 hours of darkness),
and thus the colours were visible only
during the light phase. Consistent with this,
the flies showed no preference for any par-
ticular coloured part of the tube during the
dark phase.
During the light phase, however, the flies
exhibited a complex, systematic pattern of
colour preference (Fig. 1). They consistently
avoided the blue-light zone; furthermore, they
spent more time in the green zone than in the
red zone in the early morning and late after-
noon, when the flies showed bursts of activity8.
Such timed preferences are intuitively advanta-
geous, because some of this activity is devoted
to searching for food, and flies often find food
in or under green trees and bushes.
Flies lack photoreceptors (light-sensitive
cells) that are specifically sensitive to red light,
although their green-light photoreceptors
show some sensitivity to red light9; thus, they

3 O C T O B E R 2 0 1 9 | VO L 5 7 4 | NAT U R E | 4 3
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 RESEARCH NEWS & VIEWS

probably experience the red light as dim green
light or just dim light. Lazopulo et al. observed
that the flies tended to rest under red or green
light during their midday ‘siesta’ (Fig. 1). How-
ever, when they could choose between bright
green light and dim white light, the flies gen-
erally chose to have their siesta in dim white
light. These results suggest that flies prefer to
rest under dim light.
Nonetheless, when the flies could choose
between bright green and dim blue light for
their siesta, the flies always preferred green.
This implies that flies decide between green
and blue on the basis of colour (wavelength)
of the light, rather than intensity. Indeed, their
strong aversion to blue light during the siesta
makes sense because blue light (and ultraviolet
light, which has a slightly shorter wavelength
and higher energy) can damage DNA and can
therefore be harmful10.
Light-activated pigments called rhodopsins
are expressed by photoreceptor cells in the eyes
of animals. The authors monitored the colour
preferences of fly strains that were geneti-
cally engineered to have mutations in genes
expressed by photoreceptor cells, including
those encoding rhodopsins. They found that
rhodopsin 1, which is the most abundant rho-
dopsin expressed in the compound eyes of
fruit flies, is necessary for green-light prefer-
ence, whereas rhodopsin 6 contributes to the
preference for dim light during siesta time.
Both rhodopsins are more sensitive to green
light than to other colours, and photoreceptor
cells that express these rhodopsins signal to a
group of neuronal cells in the brain — known
as the central clock11 — that regulate many bio-
logical processes over the 24-hour (circadian)
period. Thus, the central clock might time the
changes in green-light preference.
‘Clock’ genes expressed by neurons in
the central clock interact with one another
in a complex feedback loop that serves as a
timekeeping mechanism. The authors tested
clock-gene mutants12 whose genetic clock
did not function, or ran faster or slower than
normal. Flies that lacked a functioning clock
did not show morning and evening peaks in
green-light preference and instead preferred
green light throughout the entire light period.
By contrast, mutants whose internal clocks ran
faster or slower than normal had an abnor-
mally timed preference to green light. Notably,
however, the clock-gene mutants still avoided
blue light, indicating that this behaviour does
not require a functional circadian clock.
Unexpectedly, the authors established that
the photoreceptor cells responsible for blue-
light avoidance are not located in the head;
neither the compound eyes, nor the blue-
light-sensitive protein cryptochrome, which
is present in the circadian clock neurons, were
needed for this response. Instead, rhodopsin 7,
expressed by pain-sensing neurons of the
body wall, was required for blue-light avoid-
ance. These neurons mediate escape responses
of the fly to high temperatures, potentially
Dark Light Dark
a
Preference
Wild-type
flies
Avoidance
b choose between different colours. There-
Clock-gene
mutants
c of the flies’ eyes5,6,16,17. Thus, the mechanisms
↓ Rhodopsin 1
d the avoidance of blue light work together in
↓ Rhodopsin 7
Figure 1 | Flies show differences in colour
preference during the day. Lazopulo et al.4
assessed the colour preferences of fruit flies
(Drosophila) at various times of day as they
moved freely between three differently coloured
zones — blue, red and green — of glass tubes. a,
Wild-type flies showed no preference for any of
the three zones during the 12-hour dark period.
But, during the light phase, they avoided the blue
zone and preferred the green zone to the red zone
during their morning and evening activity bouts.
b, Flies that were genetically engineered to lack a
24-hour biological rhythm (clock-gene mutants)
preferred green light to red light across the whole
light phase. c, By contrast, flies that expressed less
of a light-sensitive pigment called rhodopsin 1 in
their eyes than normal flies did not prefer green
light to red light (which flies perceive as dim light).
d, Unexpectedly, flies that expressed less-than-
normal levels of another rhodopsin pigment,
rhodopsin 7, in pain-sensitive neuronal cells in the
body wall did not avoid blue light.
harmful chemicals and mechanical stimuli.
Rhodopsin 7 is present at low levels in several
fly tissues, and its sensitivity to ultraviolet and
blue light was characterized only a few years
ago13,14. Its function has been debated, but
Lazopulo et al. seem to have found a crucial
function of rhodopsin 7.
The study shows for the first time that pain-
sensing neurons in the body wall of adult flies
can detect light and mediate consistent avoid-
ance of short-wavelength light. Until now, this
had been reported only in larvae15. Previous
studies measured adult flies’ colour prefer-
ences by placing them between two tubes lit
with different colours (generally UV, green or
blue)5,16,17. Depending on the available choice,
the flies moved towards UV or blue light. By
contrast, the flies in Lazopulo and colleagues’
study were not attracted to blue or UV light at
any time of the day.
This difference in result is probably due to
the fact that in the Lazopulo et al. experimental
set-up, the flies could freely choose a preferred
place, completely undisturbed. In previous
studies, flies were aroused either by being
placed together in a new environment5,16,17 or
by acute exposure to light6 before they could
fore, these previous experiments measured
phototactic responses — that is, movements
towards or away from a light source — rather
than intrinsic preferences for certain colours.
Phototactic responses depend on the function
that control phototactic behaviour and the
avoidance of blue light are distinct.
It remains to be seen how the two pathways
that control the preference for green light and
fruit flies. Lazopulo and colleagues’ findings
also raise the question of how other animals,
including mammals, respond to different
colours at different times of day, and whether
similar pathways exist in other animals.
Rhodopsin 7 is present in most arthropods18
(joint-legged invertebrates) and might have
a similar role in these animals. However, it is
absent in vertebrates, and humans don’t seem
to have a sensor in their skin that enables them
to avoid blue light. If they did, they probably
wouldn’t spend hours in the sun getting a tan. ■
Charlotte Helfrich-Förster is in the
Department of Neurobiology and Genetics,
Theodor Boveri Institute, Biocenter, University
of Würzburg, Am Hubland, 97074 Würzburg,
Germany.
e-mail: charlotte.foerster@biozentrum.uni-
wuerzburg.de
1. Osorio, D. & Vorobyev, M. Vision Res. 48,
2042–2051 (2008).
2. Whitfield, T. W. & Wiltshire, T. J. Genet. Soc. Gen.
Psychol. Monogr. 116, 385–411 (1990).
3. Schloss, K. B., Nelson, R., Parker, L., Heck, I. A. &
Palmer, S. E. Cogn. Sci. 41, 1589–1612 (2017).
4. Lazopulo, S., Lazopulo, A., Baker, J. D. & Syed, S.
Nature 574, 108–111 (2019).
5. Yamaguchi, S., Desplan, C. & Heisenberg, M. Proc.
Natl. Acad. Sci. USA 107, 5634–5639 (2010).
6. Baik, L. S., Recinos, Y., Chevez, J. A., Au, D. D. &
Holmes, T. C. J. Biol. Rhythms 34, 391–400 (2019).
7. Wells, D. L., McDonald, C. L. & Ringland, J. E.
J. Comp. Psychol. 122, 213–219 (2008).
8. Hamblen-Coyle, M. J., Wheeler, D. A., Rutila, J. E.,
Rosbash, M. & Hall, J. C. J. Insect Behav. 5, 417–446
(1992).
9. Salcedo, E. et al. J. Neurosci. 19, 10716–10726
(1999).
10. Kielbassa, C., Roza, L. & Epe, B. Carcinogenesis 18,
811–816 (1997).
11. Alejevski, F. et al. Nature Commun. 10, 252 (2019).
12. Konopka, R. J. & Benzer, S. Proc. Natl Acad. Sci. USA
68, 2112–2116 (1971).
13. Ni, J. D., Baik, L. S., Holmes, T. C. & Montell, C.
Nature 545, 340–344 (2017).
14. Senthilan, P. R., Grebler, R., Reinhard, N., Rieger, D.
& Helfrich-Förster, C. Biology 8, 6 (2019).
15. Xiang, Y. et al. Nature 468, 921–926 (2010).
16. Heisenberg, M. & Buchner, E. J. Comp. Physiol. 117,
127–162 (1977).
17. Gao, S. et al. Neuron 60, 328–342 (2008).
18. Senthilan, P. R. & Helfrich-Förster, C. PeerJ 4, e2427
(2016).
This article was published online on 18 September 2019.

4 4 | NAT U R E | VO L 5 7 4 | 3 O C T O B E R 2 0 1 9
©
2
0
1
9
S
p
r
i
n
g
e
r
N
a
t
u
r
e
L
i
m
i
t
e
d
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.