Summer mass migrations were randomly directed
(Rayleigh tests; P > 0.05 in all cases) (Fig. 2A and
table S5), this indicated an absence of wind
selectivity. By contrast, fall mass migrations were
consistently directed toward the south (Rayleigh
tests; daytime: medium, 174°; large, 159°; nighttime:
medium, 181°; large, 180°) (Fig. 2A), and this indicates active selection of winds with a significantly
more northerly component than the prevailing
winds (Watson-Wheeler tests; P < 0.0001 in all
cases) (table S5). These relationships indicate
preferred movement directions during the spring
and fall, as well as selection of days and nights
with favorably directed tailwinds. Seasonally beneficial migration directions have been previously
reported in a few species of large insects, notably
pest noctuid moths (3, 14, 18), but our findings
demonstrate the ubiquity of such movements
among a diverse array of insect migrants. Because small insects fall below the VLR’s detection threshold (14) and, thus, their tracks cannot
be directly measured, we used aphid migration
intensity as representative of all small insect
migration (fig. S3) by analyzing wind directions
associated with mass migrations of aphids (15).
We found that aphid migration directions closely
match prevailing wind directions, i.e., toward the
northeast in all seasons (fig. S4), and, therefore,
conclude that these small insects do not have mechanisms for selecting seasonally beneficial winds.
The greatest amount of variation in the aerial
density of diurnal migrants was explained by
surface meteorological conditions associated with
fine weather (table S6): Migration intensity was
greatest on warm days (fig. S5A) with moder-
ate to high surface heat flux (fig. S5B) and low
surface wind speeds (fig. S5C). However, the strong
relationship with fine weather does not explain
the directed movements of the larger insects
in spring and fall. Models indicated that during
these seasons (but not summer), surface wind
direction was also correlated with migration
intensity, with high densities associated with
southerly winds in the spring and northerly
winds in the fall (table S6). Surface and high-
altitude daytime wind directions were strongly
correlated in all seasons (tests for T-linear as-
sociation; P < 0.001 in all cases) (fig. S6) but not
at night (18). Thus, surface wind direction pro-
vides a reliable cue regarding the suitability of
winds aloft for diurnal migrants at take-off but
not for nocturnal migrants, which must use other
methods for assessing high-altitude wind direction
(19). The ubiquity of tailwind selectivity in such a
diverse group indicates that compass mechanisms
must be universal in larger insect migrants.
If high-flying insects have a compass sense,
one would predict that, in addition to selecting a
favorable tailwind, they would also orientate in
the seasonally beneficial direction and, thus, actively contribute to their wind-assisted displacement. Such “common orientation” was indeed
highly prevalent in the larger insects (table S1),
and headings were close to tracks (fig. S7): northward in the spring and southward in the fall
(table S7). The close correspondence between
headings and tracks signifies that larger insects
added much of their self-powered airspeed to the
wind vector and thus achieved rapid displacement speeds [10 to 16 m/s (36 to 58 km hr−1)]
(Fig. 2B and table S7). A flight duration of 4 hours
could therefore result in transport over >200 km,
and during spring and fall, this transfer of biomass
and nutrients occurs in predictable directions.
What are the implications of this high-altitude
insect movement? Insect bodies are typically
composed of 10% nitrogen and 1% phosphorus
by dry weight (20), and as such, they represent a
rich source of nutrients that can be limiting for
plant productivity (11). The 3200 tons of biomass
moving annually above our study region contains
~100,000 kg of N and 10,000 kg of P, represent-
ing 0.2% of the surface deposits of N and 0.6 to
4.7% of P from the atmosphere, and making up
5.78 × 1012 Joules of energy (15). To put these sea-
sonal movements in context, the annual airborne
insect biomass >150 m above the southern United
Kingdom is 4.5 times the 2.2 billion (700 tons) of
bogong moths (Agrotis infusa) that migrate to the
Australian Alps every summer (7, 9), 7.7 times the
30 million songbird migrants (415 tons) that
depart the United Kingdom for Africa each fall
(table S8), and >40 times the 150 million monarch
butterflies (75 tons) that migrate between eastern
North America and Mexico (14).
If the spring and fall movements documented
here perfectly counterbalance each other, there
would be no net annual exchange of energy and
nutrients, and the principal consequences would
be the exchange of genes, pathogens, and parasites. Over the 10-year study, we found that net
northward spring movements of larger insects
were almost exactly cancelled out by net southward fall movements; however, on an annual
basis, the net flux could be up to 200 tons greater
in either direction (Fig. 3). Such insect movements
represent an underappreciated mechanism for
redistributing nutrients and energy, and if the
densities observed over southern United Kingdom are extrapolated to the airspace above all
continental landmasses, high-altitude diurnal
insect migration represents the most important
annual animal movement in terrestrial ecosystems,
comparable to the most significant oceanic migrations (21). Given the worrying declines in many
migrants (8), developing global surveillance techniques (6) for long-term observation and prediction of the impacts of mass aerial migrations at
such macrosystem scales (22) should be a priority
REFERENCES AND NOTES
1. R. A. Holland, M. Wikelski, D. S. Wilcove, Science 313, 794–796
2. S. Hahn, S. Bauer, S. Liechti, Oikos 118, 624–626 (2009).
3. J. W. Chapman et al., Science 327, 682–685 (2010).
4. J. W. Chapman et al., Proc. Natl. Acad. Sci. U.S.A. 109,
5. J. W. Chapman, D. R. Reynolds, K. Wilson, Ecol. Lett. 18,
6. R. Kays, M. C. Crofoot, W. Jetz, M. Wikelski, Science 348,
7. E. Warrant et al., Front. Behav. Neurosci. 10, 77 (2016).
8. D. S. Wilcove, M. Wikelski, PLOS Biol. 6, e188 (2008).
9. K. Green, Austral. Ecol. 36, 25–34 (2011).
10. S. Bauer, B. J. Hoye, Science 344, 1242552 (2014).
11. J. S. Landry, L. Parrott, Ecosphere 7, e01265 (2016).
12. H. Mouritsen, D. Heyers, O. Güntürkün, Annu. Rev. Physiol. 78,
13. T. Alerstam et al., Proc. Biol. Sci. 278, 3074–3080
14. V. A. Drake, D. R. Reynolds, Radar Entomology: Observing
Insect Flight and Migration (CABI, Wallingford, UK, 2012).
15. Supplementary materials and methods are available on
16. J. W. Chapman, D. R. Reynolds, A. D. Smith, E. T. Smith,
I. P. Woiwod, Bull. Entomol. Res. 94, 123–136 (2004).
17. C. Stefanescu et al., Ecography 36, 474–486 (2013).
18. J. W. Chapman et al., Curr. Biol. 18, 514–518 (2008).
19. J. W. Chapman et al., Curr. Biol. 25, R751–R752
20. J. J. Elser et al., Nature 408, 578–580 (2000).
21. O. Varpe, O. Fiksen, A. Slotte, Oecologia 146, 443–451 (2005).
1586 23 DECEMBER 2016 • VOL 354 ISSUE 6319 sciencemag.org SCIENCE
Fig. 3. Annual patterns of net directional migration. The net flow of biomass of larger insects above
the study region, in spring, fall, and the whole year. Negative values indicate a net southward movement;
positive values indicate a net northward movement. In the box plots, central bars represent median
values, boxes represent the IQR, whiskers extend to observations within ±1.5 times the IQR, and dots