From:  Climate-driven changes in pollen dynamics: increased loads and earlier, longer exposure

 Temporal trends in pollination of the Betulaceae family.

Reference (publication date)Region (number of sites)Study periodMethodologyTrend identification techniquePIn (grains per cubic meter of air per decade and significance)PSS (days per decade and significance)PSD (days per decade and significance)
Betulaceae
[16] (2016)N Italy (1)1990–2004Hirst-typeLinear regression+ (~1,400)No trend
[17] (2023)Central Europe1951–2015Phenological observationsLinear regression
[18] (2014)Europe (13)1990–2009Hirst-typeLinear regression+ (* in 2 out of 13 sites, ns in the other 11)– (** in 3 of 8 sites, ns in the other 5)+ or – (* in 2 of 8 sites, ns in the other 5)
Betula
[16] (2016)N Italy (1)1990–2004Hirst-typeLinear regression+Slightly + (1.3)
[17] (2023)Central Europe1951–2015Phenological observationsLinear regression– (1.9)
[19] (2022)United Kingdom (6)1995–2020Hirst-typeLinear regression+ (*)– (ns) at 5 of the 6 sites
[20] (2010)NW Italy (1)1981–2007Hirst-typeLinear regression+ (*)– (10.0)No trend
[21] (2020)S Germany (1)1988–2018Hirst-typeLinear regression & least squares fitSlightly +– (~2)– (0.7)
[22] (2019)France (7)1987–2018Hirst-typeLinear & polynomial regression+ (209 to 1,263)– until 2002, + thereafter– almost everywhere (2.4 to 5.7)
[23] (2021)Central Germany (4)1998–2017Hirst-typeCochran-Armitage test for trendSlightly + (ns)
[24] (2005)Austria (2)1980–2001Hirst-typeLinear regression+ (~125 in valley area, ~350 at higher altitude)– (4.0 in valley area, 7.3 at higher altitude)
[25] (2018)Brussels (1)1982–2015Hirst-typeLocal regression method (LOESS)+
[26] (2003)Switzerland (1)1979–1999Hirst-typeLinear regressionVery slightly + (ns)– (9.5; *)No trend
[27] (2025)Texas, USA (1)2009–2023Hirst-typeLeast square regression+ (0.7; **)
[28] (2021)Benelux (5)1981–2020Hirst-typeLinear regression+ in 4 sites of 5, but * in only 2– in all sites, but * in only 1– but variable and ns
[29] (1993)London1961–1990Hirst-typeLinear regression+ (especially after 1975)No trend (very strong inter-annual variation)
[30] (2014)Catalonia (8)1994–2011Hirst-typeLinear regression + non-parametric testVery slightly + (ns)
[31] (1998)Basel (1)1969–1996Hirst-typeLinear regression+ (*)No trend
[32] (2000)Basel (1)1969–1998Hirst-typeLinear regression+ (~1,700; *)
[33] (2008)Basel (1)1969–2006Hirst-typeLinear regression+ (978; *)– (~3.9, especially since the end of the 1970s; *)
[34] (2008)Switzerland (3)1969–2007Hirst-typeLinear regression+ (from 1969 to 1990), then – in Basel– (1.7 to 3.6)
[35] (2021)Basel (1)1969–2020Hirst-typeLinear regression+ (~190)– (~1.9)
[36] (2016)Spain (12)1994–2013Hirst-typeLinear regression+ (ns, but ** in Ourense)
[37] (2021)Basel (1)1969–2018Hirst-typeLinear regression+ (ns)– (~2; *)– (~0.74; ns)
[38] (2021)Switzerland (14)1990–2020Hirst-typeLinear regression + LOESS+ (*)No trend
[39] (2024)NW Spain (3)1995–2023Hirst-typeLinear regression + non-parametric test+ (*)Slightly –
[40] (2018)Brussels (1)1982–2015Hirst-typeLinear regression + Mann-Kendall test+ (ns)– (ns)
[41] (2019)Switzerland (6)1985–2014Hirst-typeBayesian statistics+ but – in recent years in 4 of 6 sites
[42] (2011)Germany (10)1988–2009Hirst-typeLinear regression + Mann-Kendall test+ in NE and SSlightly + in S
[43] (2024)United Kingdom (2)1970–2021Hirst-typeGeneralized additive model for time series+No trend
[44] (2016)Stockholm (1)1973–2013Hirst-typeLinear regression+ (~1,000; ns)– (~3.7; ***)No trend
[45] (2011)S Poland (1)1991–2008Hirst-typeLinear regression+ (*)No trend
[46] (2002)Denmark (2)1977–2000Hirst-typeLinear regression+ (~1,150 to ~2,000)– (6.2, ** in Copenhagen; 5.2, ns in Viborg)+ (2.3, ns in Copenhagen; 6.1, ** in Viborg)
[47] (2021)Bavaria (8)1989–2018Hirst-typeLinear regression+ (ns, but * in Munich)– (*)
[48] (2021)Bavaria (28)1975–2100Hirst-type & modellingLinear regression+ initially, then – in lowlands (< 800 m) and + in highlands
[49] (1995)Europe (5)1976–1993Hirst-typeLinear regression+ (*** in Basel, London, and Leiden; ns in Vienna and Stockholm)
[50] (2003)W Europe (5)1968–2001 (partly)Hirst-typeLinear regression+ in the 5 sites (* in Leiden, ** in Derby)
[51] (2024)NE Italy (20)2006–2022Hirst-typeTheil-Sen estimatorVery slightly +No trend+ (4.3)
[52] (2021)France (6)1987–2021Hirst-typeVariable+ (~440)– until 2002, + thereafter
[53] (2009)Finland (1)1974–2004Hirst-typeLinear regression & smoothing+ (*)– (4; *), especially during later yearsNo trend
[54] (2014)USA (6)1994–2011Growing degree hour modelLinear regression+ in 3 sites, – in 2 sites; ns everywhere– in 4 out of 6 sites, slightly + in the other 2; but in all cases nsGenerally – (up to ~10), but * only in Newark
[55] (2015)USA (50)1994–2010UnspecifiedLinear regression+ in 5 of the 9 regions (54.6 in the W); – in the S (47.2)– in 59% of the 50 studied sites and in 7 of the 9 climate regions (1.5 to 8.8, * in all cases)– (up to 8.1 in 4 of the 9 regions; * in S, SE, NE and NW)
[56] (2012)Europe (97)1977–2009 (partially)Hirst-typeLinear regression+ (***)
[57] (2020)N & Central Italy (9)2000–2016Hirst-typeNon-parametric Mann-Kendall testNo trendNo trend
[58] (2020)Basel (1)1969–2018Hirst-typeLinear regressionNo trend– (~2)
[59] (2021)Georgia, USA (1)1992–2018RotorodLinear regressionNo trend– (4.8)+ (3.5)
[60] (2025)Lithuania (3)2005–2023Hirst-typeLinear regression + Mann-Kendall testNo trend– (0.5 to 6; ns)Slightly + (ns)
[61] (2021)Austria (1)1993–2017Hirst-typeLinear regressionNo trend
[62] (2025)N Italy (1)1997–2023Hirst-typeLinear regression– (ns)+ (13.5; ***)– (9.3; **)
[63] (2024)Switzerland1990–2023Hirst-typeUnspecified– from 1990 to 2000, + from 2000 to 2015, then – againNo trend
[64] (2016)USA (9 regions)2000–2050ModellingUnspecified– in most climate regions in 2047–2050 compared to 2001–2004– in the NW, NE, & W– in 8 of the 9 regions
[65] (2013)Europe + USA (5)2000–2040ModellingBayesian analysis+ (~2,450 to 3,320)– (~4.75 to 5)
2000–2060ModellingBayesian analysis+ (~2,570 to 3,920)– (~3.3 to 3.5)
2000s–2090sModellingUnspecified– (1.3)
[66] (2001)
[67] (2002)
Europe1969–1998Phenological observationsUnspecified– (2.7; *)
[68] (2001)Switzerland (1)1980–1997Hirst-typeLinear regression– (10.6; *)
[69] (1997)United Kingdom (3)1954–1995Hirst-typeLinear regression– (~5)
[70] (2002)Europe (6)1970–2000Hirst-typeLinear regression– in Belgium (6.3, *), in Switzerland (~10 after 1986, *), in UK (8.4 after 1982); no trend in Turku; + in Kevo (~15); + in Vienna until 1984, – thereafter (~11)
[71] (2007)Europe (10)1970–2004Hirst-typeLinear regressionSignificantly – in Turku, Zürich, London, Vienna, and Brussels; slightly – in France; no trend in Poland; + in Kevo
[72] (2010)Lithuania (3)1970–1999Hirst-typeLinear regression– (~3 to 5; *)
[73] (2002)W of the NL (1)1969–2000Hirst-typeLinear regression– (3.1; *)
2000s–2090sModellingUnspecified– (1.3)
[74] (2025)Iberian Peninsula (12)1994–2023Hirst-typeLinear regression+ (3.3)
[75] (2013)Central and S United Kingdom (3)1970–2010Hirst-typeLinear regression + distance weighted smoothing technique– in the 1980s and early 1990s, then +
[76] (2014)Ireland1990s–2030sProcess-based phenological modelLinear regression– (on average, 0.75, more in the NE than in the SW)
Corylus
[16] (2016)N Italy (1)1990–2004Hirst-typeLinear regressionNo trend– (~14 to 15)
[17] (2023)Central Europe1951–2015Phenological observationsLinear regression– (3.1)
[23] (2021)Central Germany (4)1998–2017Hirst-typeCochran-Armitage test for trendSightly + (ns)
[24] (2005)Austria, in valley area (1)1980–2001Hirst-typeLinear regression+ (~850)– (8.1)
[25] (2018)Brussels (1)1982–2015Hirst-typeLocal regression method (LOESS)+
[26] (2003)Switzerland (1)1979–1999Hirst-typeLinear regression+ (ns)No trendNo trend
[28] (2021)Benelux (5)1981–2020Hirst-typeLinear regression+ (* in 2 sites of 5)– (* in 4 sites of 5)No trend
[30] (2014)Catalonia (8)1994–2011Hirst-typeLinear regression + non-parametric testNo trend
[31] (1998)Basel (1)1969–1996Hirst-typeLinear regression+ (**)
[32] (2000)Basel (1)1969–1998Hirst-typeLinear regression+ (~450; *)
[35] (2021)Basel (1)1969–2020Hirst-typeLinear regression+ (~240)
[37] (2021)Basel (1)1969–2018Hirst-typeLinear regression+ (248; ***)– (4.5; *)+ (3.5; *)
[38] (2021)Switzerland (14)1990–2020Hirst-typeLinear regression + LOESS+ (985 to ~1,093; *)– (*)– (*)
[40] (2018)Brussels (1)1982–2015Hirst-typeLinear regression + Mann-Kendall test+ (*)– (ns)
[41] (2019)Switzerland (6)1985–2014Hirst-typeBayesian statistics+
[42] (2011)Germany (10)1988–2009Hirst-typeLinear regression + Mann-Kendall test+Slightly + in S
[44] (2016)Stockholm (1)1973–2013Hirst-typeLinear regression+ (~24; **)– (~5.1; ns)– (2.1; ns)
[45] (2011)S Poland (1)1991–2008Hirst-typeLinear regression+ (***)No trend
[51] (2024)NE Italy (20)2001–2022Hirst-typeTheil-Sen estimator+ (1,089)– (7.3)No trend
[56] (2012)Europe (97)1977–2009 (partially)Hirst-typeLinear regression+ (***)
[57] (2020)N & Central Italy (9)2000–2016Hirst-typeNon-parametric Mann-Kendall testNo trendNo trend
[58] (2020)Basel (1)1969–2018Hirst-typeLinear regression+ (~300)– (~5)
[61] (2021)Austria (1)1993–2017Hirst-typeLinear regressionNo trend
[62] (2025)N Italy (1)1997–2023Hirst-typeLinear regression+ (331; *)– (4.7; *)+ 0.6 (*)
[63] (2024)Switzerland1990–2023Hirst-typeUnspecified+ (~412)– (~5)
[81] (2007)Thessaloniki (1)1987–2005Hirst-typeLinear regression+ (~120)No trendNo trend
[82] (2024)Rome (1)1997–2016Hirst-typeLinear regressionNo trend
[83] (2002)Central Italy (1)1982–2001Hirst-typeLinear regression+ (1.0; ns)No trend
Alnus
[16] (2016)N Italy (1)1990–2004Hirst-typeLinear regressionSlightly –
[17] (2023)Central Europe1951–2015Phenological observationsLinear regression– (3.6)
[23] (2021)Central Germany (4)1998–2017Hirst-typeCochran-Armitage test for trendNo trend
[24] (2005)Austria (2)1980–2001Hirst-typeLinear regression+ (~1,100 in valley area, ~3,000 at higher altitude)– (4.7 in valley area, 6.0 at higher altitude)
[25] (2018)Brussels (1)1982–2015Hirst-typeLocal regression method (LOESS)+
[26] (2003)Switzerland (1)1979–1999Hirst-typeLinear regression+ (*)No trendNo trend
[28] (2021)Benelux (5)1981–2020Hirst-typeLinear regression+ (*) in 4 sites of 5– in 3 sites of 5, but * in only 2+ in 4 sites of 5, but * in only 1
[32] (2000)Basel (1)1969–1998Hirst-typeLinear regression+ (~1,100; *)
[37] (2021)Basel (1)1969–2018Hirst-typeLinear regression+ (~480; **)– (ns)– (ns)
[38] (2021)Switzerland (14)1990–2020Hirst-typeLinear regression + LOESS
[40] (2018)Brussels (1)1982–2015Hirst-typeLinear regression + Mann-Kendall test+ (*)– (ns)
[42] (2011)Germany (10)1988–2009Hirst-typeLinear regression + Mann-Kendall test+ in NE and S– in NW (1.4)+ in NW
[44] (2016)Stockholm (1)1973–2013Hirst-typeLinear regression+ (~30.5; ns)– (~2.4; ns)– (~1.1; ns)
[45] (2011)S Poland (1)1991–2008Hirst-typeLinear regressionNo trendNo trend
[47] (2021)Bavaria (8)1989–2018Hirst-typeLinear regression+
[51] (2024)NE Italy (20)2006–2022Hirst-typeTheil-Sen estimator+ (326)– (4.4)No trend
[56] (2012)Europe (97)1977–2009 (partially)Hirst-typeLinear regression+ (***)
[57] (2020)N & Central Italy (9)2000–2016Hirst-typeLinear regression + non-parametric Mann-Kendall test+ (*)– (*)
[60] (2025)Lithuania (3)2005–2023Hirst-typeLinear regression + Mann-Kendall testNo trend– (6.5 to 17; *)No trend in Siauliai and Klaipeda, + (> 5.0) in Vilnius
[61] (2021)Austria (1)1993–2017Hirst-typeLinear regressionNo trend
[81] (2007)Thessaloniki (1)1987–2005Hirst-typeLinear regression+No trendNo trend
[83] (2002)Central Italy (1)1982–2001Hirst-typeLinear regression+ (2.0; ns)No trend
[84] (2023)NW Spain (1)1993–2018Hirst-typeLinear regression+ (1,120; **)+ (ns)+ (ns)
[85] (2021)Slovakia (1)2002–2019Hirst-typeLinear regression + Mann-Kendall test+
[86] (2020)Andalusia (3)1994–2017Hirst-typeLinear regression+ (* in Granada, ns elsewhere)+ in Córdoba (*) and Málaga (**), ns in Granada

The grey backgrounds correspond to studies that have made projections about the future. +: trend towards increased pollen concentration (PIn), earlier onset of pollination (PSS), or longer pollen season (PSD); –: trend towards decreased pollen concentration (PIn), later onset of pollination (PSS), or shorter pollen season (PSD); *: p < 0.05; **: p < 0.01; ***: p < 0.001. PIn: pollen integral; PSS: pollen season start date; PSD: pollen season duration; N: north; ns: non-significant; NW: northwest; S: south; LOESS: locally estimated scatterplot smoothing; NE: northeast; W: west; SE: southeast; SW: southwest; NL: Netherlands.