Think of the tropics, and you'll no doubt picture lush, dense greenery with huge leaves. Venture towards the poles or arid regions, however, and the leaves on vegetation become small and narrow.

So it's easy to assume plants evolved different sized leaves to cope with differences in temperature and moisture. Now scientists have a clearer picture, and it appears there's a little more to the story.

Leaf sizes all over the globe differ in size more than 100,000 times, from less than a square millimetre to over a square metre (10 square feet) in area. Surprisingly, there's little research to confirm exactly why this variation exists.

New research by an international team of scientists has determined night time temperatures and risk of damage from ice are the key factors that determines why the Phillipino banana tree (Musa textilis) has large, wide leaves, while the camel thorn (Acacia erioloba) has tiny, relatively narrow leaves.

"The conventional explanation was that water availability and overheating were the two major limits to leaf size," says lead researcher Ian Wright from Macquarie University.

All leaves lose water through transpiration, which helps water move up the stem and also cools down the leaf much in the same way as sweating keeps us from overheating.

Given plants can't simply move into the shade at will, this kind of temperature regulation is important in the heat of the tropics.

But bigger leaves are also usually thicker leaves, which makes it harder for the plant to shed heat as efficiently.

The solution would be to pump up more water. Problem solved, but now there's a price on cooling that creates stiff competition over a resource that isn't always plentiful.

Clearly where it was wetter, this might not be a big problem. But sunlight isn't exactly in short supply in the tropics, either.

According to the models based on "classic" energy budget theory, leaves in cooler climates should be scooping up sunlight without the same concern for overheating as big leaves in warm, sunny areas.

But the data from previous studies had confirmed as the average temperature went up in an environment, so did the average leaf size.

To address the question of which factors were actually responsible for putting pressure on the leaf's evolution, the team went big, collecting measurements from over 7,600 plants taken from hundreds of non-agricultural locations worldwide.

Tens of thousands of leaf measurements later, the researchers had a sizeable database they could analyse for significant relationships.

"By sampling across all continents, climate zones, and plant types we were able to show that simple 'rules' seemingly operate across the world's plant species, rules that were not apparent from previous, more limited analyses," says Write.

As expected, leaves were once again shown to be larger around equatorial regions and smaller as you headed further from the tropics.

The average amount of precipitation was also taken into account, as was the general amount of moisture in the soil.

There was no surprise in finding combinations of climate factors such as site temperature, moisture, and sunlight were big matches for leaf size.

The details, however, revealed something new.

"The most surprising result was that over much of the world the maximum size of leaves is set not by the risk of overheating, but rather by the risk of damaging frost at night," says researcher Colin Prentice from Imperial College London.

In other words, the main thing that prevents leaves from stretching out to absorb as much sunlight as possible isn't overheating or even losing too much water, but rather staying warm at night when the temperature drops.

Large leaves have wide areas that grip onto a thin slice of air called a boundary layer.

As heat radiation pours out of the leaves at night, the plants can't replace it as easily with heat from the environment thanks to that insulating layer, which acts like a warm jacket in reverse.

The discovery can help ecologists update their models on how plants cope with different conditions, something that could be super important as many of the world's diverse climates rapidly change in coming years. 

This research was published in Science.