Why cities fly first, and by how little

Cities are warmer than the countryside around them. Ants are cold-blooded and their flights key off warmth. So flights should happen earlier in town, and they do.

That's the easy part, and it's been measured. The interesting question is how much earlier, and the honest answer is less than almost anyone guesses.

The urban heat island is one of the older results in applied climatology and one of the more variable. Angilletta and colleagues, opening a 2007 paper on São Paulo's leafcutter ants, put the upper end of the published range at as much as 12 °C hotter than the surrounding country, which is nearly 22 °F. That's an extreme, measured under the conditions that maximise it. A couple of degrees is a much more ordinary figure for a mid-sized city on a mid-sized day.

What sets the size of it took a while to pin down. Manoli and colleagues worked through summertime urban and rural surface temperatures worldwide for a 2019 Nature paper and found the intensity rises non-linearly with rainfall, because what a city mostly does is stop the ground from evaporating water. Vegetation sweats and cools itself. Tarmac doesn't. In a wet climate you've replaced a landscape that was doing a lot of evaporative cooling; in a dry one there wasn't much to replace. Population size matters too, but it's modulated by that.

The edge is also softer than a map suggests. Bassett and colleagues put 29 weather stations across Birmingham at roughly 3 km spacing, just under 2 miles, and ran them for 20 months, and found the city's heat gets carried downwind, contributing up to 1.2 °C, about 2.2 °F, to areas outside it. So the line between "urban" and "rural" that any study has to draw is somewhat fictional, and the fiction is in the direction of shrinking whatever difference you find.

The Royal Society of Biology's Flying Ant Survey ran across 2012, 2013 and 2014 and produced 13,394 analysable public records, which Hart and colleagues wrote up in Ecography in 2018. Participants could optionally say whether they were in a rural or urban area, and describe the nest's setting.

Urban observations came in earlier. The survey numbered its days from 1 June, and the mean urban observation landed on day 55.5, against day 58.5 for rural. That's 25 to 26 July against 28 to 29 July, a gap of about three days, and it's statistically solid on a sample of that size.

Nest setting did the same thing and slightly harder. Colonies in what the authors call heat-retaining structures, meaning pavements, stones and brick walls, or heat-producing ones, meaning compost heaps, were seen a little over three days ahead of everything else. Same direction, same magnitude, and it makes obvious sense: a nest under a slab is in a warmer place than a nest under a lawn, which is the same point the neighbour post makes at the scale of one street.

Now the part that gets left out of the summary. Habitat and microhabitat together explained about 1% of the variance in observation date. The authors say so plainly and don't dress it up.

It's worth sitting with that for a second, because the popular version of this claim is much stronger than the evidence for it.

The same dataset found no significant spatial clustering of observations across the UK at a national level, and, when the authors zoomed in on Greater London specifically, none there either. A handful of individual days did cluster, but across the survey as a whole flying ants could turn up anywhere in the country on more or less any day in the window. Even the broad geographic gradient, which does exist, is modest: comparing six separated areas, Exeter in the far south ran about nine days ahead of the more northerly stations across a whole country.

So a city buys you three days out of a season that runs from June to September, in a country where the north-south spread is nine. It's real. It's just nowhere near enough to build a "London flies on this date, Cumbria on that one" calendar out of, which is more or less what people try to do with it every July.

Which makes the other main result on this look strange at first.

Chick and colleagues, in the Journal of Thermal Biology in 2019, took acorn ants from urban and rural sites in three eastern US cities along a latitudinal gradient, reared them in the lab under five temperature treatments, and looked at the timing of reproduction. At the northernmost and southernmost of the three, 6° of latitude apart, both urbanisation and warmer rearing significantly advanced it, and translated back to the field the urban and rural populations differed by roughly a month.

A month against three days is a big gap to explain, and I don't think it's a contradiction so much as two different clocks being read.

Chick's experiment measures when a colony produces reproductives. Hart's data measures when a member of the public saw a swarm outside their house. A colony can have winged ants finished and sitting in the nest for weeks, waiting. Boomsma and Leusink noticed the same thing back in 1981, that flights late in the season happen in worse weather because the colonies are running out of time and lowering their standards. Production and departure aren't the same event, and the gap between them absorbs a lot.

The two studies also differ in nearly every other way that matters: different continents, different species, different nesting habits (an acorn on a woodland floor bakes in a way a deep soil nest doesn't), a controlled common garden versus 13,000 members of the public. Both can be right.

Hart and colleagues offer the explanation themselves, and it fits both results: colonies in warmer places develop faster and so have winged ants ready to fly at an earlier date.

That's a claim about readiness, not about triggering. The trigger is still the afternoon, and the afternoon arrives across a whole region at once. A city colony reaches the starting line first and then waits, along with everyone else, for a warm still day. If the first suitable day after the urban colony is ready is also suitable for the rural one twenty miles out, they fly together and the head start vanishes from the record entirely. You'd only see the urban advantage on the years and in the places where a suitable day lands inside that window. Averaged across three summers and a whole country, that washes down to about three days.

Something I did on the back of an envelope while writing this, offered as arithmetic rather than as a result.

Helms's 2022 survey of a century of US museum records found ant mating dates falling about 0.9 days later per 100 m of elevation gained, which is a shade over 300 feet. Air cools with height at very roughly 0.65 °C per 100 m, about 1.2 °F. Divide one by the other and you get something like 1.4 days of phenological shift per degree Celsius, which for a city running 2 °C, or 3.6 °F, above its countryside predicts a shift of just under three days.

Hart measured three. I would not lean on that agreement at all, since it's two unrelated datasets, two continents and a lapse rate I plucked from a textbook. But when a crude sum and a real measurement land in the same place it's usually worth noticing, and here it suggests the three days isn't a fluke of the British survey.

Slightly off the main point, but too good to leave out.

Because urban heat islands are decades old and ant generations are short, cities have become an accidental evolution experiment. Diamond and colleagues ran acorn ants from urban and rural populations in three cities through a common garden and found evolved losses of cold tolerance, and a narrowing of thermal tolerance overall, in two of the three. Angilletta's São Paulo leafcutters took 20% longer to lose mobility at 42 °C, which is 107.6 °F, if they came from inside the city.

None of that changes a forecast. It does mean the urban and rural populations of the same species are drifting apart, and Chick's paper raises the possibility that if their reproductive timing separates far enough they'd stop meeting each other in the air.

Heaviside and colleagues make a point in passing, in a 2016 paper on heat mortality in the West Midlands, that anyone using weather data in a city should probably have tattooed somewhere: temperature monitoring sites are traditionally placed outside city centres. Airports, mostly, and open ground.

So the station that nominally represents a city is often standing in exactly the conditions the city isn't in. Any urban-versus-rural comparison built on standard station data is comparing a real rural reading against a slightly-less-rural one, and the effect gets flattened again. This is a cousin of the resolution problem in why the forecast and your garden disagree, where the issue is the size of the box rather than where it's sitting.

Dealate does not model the urban heat island. There's an impervious-surface proxy in the design notes as a feature worth adding, and the notes also say, in as many words, that the phenological effect size for ants is not yet well quantified. Both of those are still true, and a three-day shift explaining 1% of variance is not a thing to bolt onto a forecast in a hurry.

What the app does instead is score the weather where you actually are, hour by hour, for the species active near you, and show the reasoning behind the score. If your street runs warm, that's already partly in the local weather it's reading. The rest of it, the couple of degrees your particular postcode holds onto after dark, is honestly beyond what a public weather model can see.

A reason to look up. Not a guarantee.

Also worth reading: ants are mating a day earlier every decade on the climate-change version of the same shift, and ants don't read the barometer on which weather variables actually move a flight.

Hart, A.G., Hesselberg, T., Nesbit, R. & Goodenough, A.E. (2018). The spatial distribution and environmental triggers of ant mating flights: using citizen-science data to reveal national patterns. Ecography 41(6): 877 to 888. Source of the urban and rural mean observation days, the heat-retaining structure result, the 1% variance figure, the absence of spatial clustering nationally and within Greater London, and the Exeter comparison.

Chick, L.D., Strickler, S.A., Perez, A., Martin, R.A. & Diamond, S.E. (2019). Urban heat islands advance the timing of reproduction in a social insect. Journal of Thermal Biology 80: 119 to 125. Source of the approximately one month advance in urban acorn ant reproductive phenology.

Diamond, S.E., Chick, L.D., Perez, A., Strickler, S.A. & Martin, R.A. (2018). Evolution of thermal tolerance and its fitness consequences: parallel and non-parallel responses to urban heat islands across three cities. Proceedings of the Royal Society B 285: 20180036. Source of the evolved cold-tolerance losses in two of three cities.

Angilletta, M.J., Wilson, R.S., Niehaus, A.C., Sears, M.W., Navas, C.A. & Ribeiro, P.L. (2007). Urban Physiology: City Ants Possess High Heat Tolerance. PLoS ONE 2(2): e258. Source of the upper-end figure for urban heat island intensity and the São Paulo leafcutter knock-down result.

Manoli, G., Fatichi, S., Schläpfer, M., Yu, K., Crowther, T.W., Meili, N., Burlando, P., Katul, G.G. & Bou-Zeid, E. (2019). Magnitude of urban heat islands largely explained by climate and population. Nature 573: 55 to 60. Source of the evapotranspiration and precipitation controls on heat island intensity.

Bassett, R., Cai, X., Chapman, L., Heaviside, C., Thornes, J.E., Muller, C.L., Young, D.T. & Warren, E.L. (2016). Observations of urban heat island advection from a high-density monitoring network. Quarterly Journal of the Royal Meteorological Society 142(699): 2434 to 2441. Source of the Birmingham network and the downwind advection figure.

Heaviside, C., Vardoulakis, S. & Cai, X-M. (2016). Attribution of mortality to the urban heat island during heatwaves in the West Midlands, UK. Environmental Health 15 (Suppl 1): S27. Cited for the point that temperature monitoring sites are traditionally sited outside city centres.

Helms, J.A. (2022). Climate, geography, and the mating phenology of ants. Insectes Sociaux 70(1): 119 to 125. Source of the 0.9 days per 100 m, or 330 feet, of elevation figure used in the rough check above.

Boomsma, J.J. & Leusink, A. (1981). Weather conditions during nuptial flights of four European ant species. Oecologia 50(2): 236 to 241. Cited for late-season flights occurring in progressively worse conditions.