Representation of temporal and geographical limits characterising the ecological niche of a hypothetical odonate species. Points show 250 individuals according to their Julian day of emergence, latitudinal position, and temperatures to which each individual is exposed. (A) Points represent historical observations, triangle symbols show observations following a shift towards higher latitudes, and plus signs show observations following a shift towards earlier emergence dates, necessary to maintain temperatures to which species were exposed historically, after temperatures warmed across the species’ range. Warm and cool colors show hot and cold temperatures, respectively. (B) Grey points represent historical observations, and the colored points show observations after warming without allowing shifts through time or space. In this scenario, we assumed that the hottest temperature tolerated by the species corresponds to the hottest temperature experienced in its historical niche. In this case, individuals are exposed to hotter temperatures overall while declining in areas outside species’ thermal tolerances after warming.

Richness of 76 odonate species sampled in North America and Europe between 1980 and 2018 per 100 × 100 km quadrat. Dark red indicates high species richness, while light pink indicates low species richness.

Relationship between range shifts and emergence phenology shifts among North American and European odonate species (N = 66). The shaded area shows mean latitudinal range shifts of terrestrial taxa (Lenoir et al., 2020), calculated as the yearly mean dispersal rate of 1.11 +/- 0.96 km per year over 38 years.

Fixed effects estimates and associated statistics from the generalized linear model and generalized mixed effects model (accounting for phylogeny; for credible intervals, see Table S3) of the relationship between range shifts and emergence phenology change. The continent term shows effects of the North American continent compared to the European continent as the reference level. N gives the number of species involved in the model, and an asterisk indicates statistical significance of the variable in question (p-value < 0.05). The pseudo R2 type is Nagelkerke (Nagelkerke, 1991).

Fixed effects estimates and associated statistics from the generalized linear model and generalized mixed effects model (accounting for phylogeny; for credible intervals, see Table S3) of drivers of odonate range shifts. N indicates the number of modelled species, an asterisk indicates statistical significance of the variable in question, and a dash symbol shows that the variable was excluded from the final model. The pseudo R2 type is Nagelkerke (Nagelkerke, 1991). For the categorical variables breeding habitat type and range geography, we used lotic habitat type and Northern range as reference levels, respectively. Phenology shifts were unrelated to any of the predictor variables that we tested.

76 species sampled across North America and Europe between 1980 and 2018 followed our criteria for quality observation records for inclusion in our analysis of geographical shifts. Species Northern Range Limits (NRL) are shown in this table, as well as range limit shifts. All range limit values are shown in kilometers from the equator. We used the 10 most northern points of sampling in each time period to identify species’ NRL, as detailed in the Methods section of the main text.

66 species sampled across North America and Europe between 1980 and 2018 followed our criteria for quality observation records for inclusion in our analysis of emergence phenology shifts. Mean phenological shifts (PS) is measured in the number of Julian days comparing both time periods, as estimated using the Weibull distribution (See Methods). We also report the number of 200 × 200 quadrats used to calculate phenology estimates per species.

Ecological and geographical traits of 76 North American and European odonate species used in this work. Field guides (Cannings, 2002; Jones et al., 2008; Paulson, 2012) and existing trait databases (Powney et al., 2014; Waller et al., 2019) were used to build this dataset. Habitat type represents species’ breeding habitat, and can have a value of lentic, lotic, or both types. Distribution shows the general geographic position of each species’ range, which can be widespread (W), southern (S), northern (N), southern and widespread (SW), or northern and widespread (NW). Oviposition type corresponds to egg laying inside plants (endophytic) as opposed to directly in water or on plants (exophytic). Body size is measured as body length in mm. In the case that body length was given as a maximum and minimum value, we used the average of both values.

Credible intervals of all MCMCglmm models testing predictions regarding the range and phenology shifts across 66 odonate species in North America and Europe. These models are detailed in Model information and statements of the Supplementary Information.

P-values and coefficients of 1000 GLM iterations testing whether range shifts as calculated from random datasets predict the range shifts measured in the study. Each point shows the results of a single GLM model, with measured range shifts as the dependant variable and randomized range shifts as the independent variable.

Observed range shifts in km from the equator, against randomized predicted values according to 4 random datasets. Points represent species and each pane contains a different set of random data in calculations of randomized range shifts. There is no consistent relationship among 1000 iterations.

P-values and coefficients of 1000 GLM iterations testing whether phenology shifts as calculated from random datasets predict the phenology shifts measured in the study. Each point shows the results of a single GLM model, with measured phenology shifts as the dependant variable and randomized phenology shifts as the independent variable.

Observed phenology shifts in Julian day, against randomized predicted values according to 4 random datasets. Points represent species and each pane contains a different set of random data in calculations of randomized phenology shifts. There is no consistent relationship among 1000 iterations.

Distribution of values computed for range shifts of 76 European and North American species between a recent time period (2008 - 2018) and a historical time period (1980 - 2002). See Methods section of the main text for full details on data assembly, and steps undertaken to produce these preliminary results.

Distribution of values computed for phenology shifts of 66 European and North American species between a recent time period (2008 - 2018) and a historical time period (1980 - 2002). See Methods section of the main text for full details on data assembly, and steps undertaken to produce these preliminary results.

Panels A and B show the trace and density estimates of a phylogenetic mixed effects model exploring the relationship between range and phenology shifts in North American and European odonates (N = 66). 150,000 iterations were run to produce these results. These plots verify model convergence and absence of autocorrelation within the explanatory variables.

Panels A and B show the trace and density estimates a phylogenetic mixed effects model testing whether ecological traits, and geographic and climatic attributes predict range shifts in North American and European odonates (N = 76). 150,000 iterations were run to produce these results. Results of the best model, according to DIC, are shown here. These plots verify model convergence and absence of autocorrelation within the explanatory variables.