This is the accepted version of the journal article: Fernández-Martínez, Marcos; Peñuelas, Josep; Chevallier, Frédéric; [et al.]. «Diagnosing destabilisation risk in global land carbon sinks». Nature, Vol. 615, issue 7954 (March 2023), p. 848-853. DOI 10.1038/s41586-023-05725-1 This version is available at https://ddd.uab.cat/record/285125 under the terms of the license 1 1. Supplementary figures and tables 2 Supplementary Figure 1: Temporal evolution of annual NBP derived from CAMS and 3 CarboScope atmospheric inversion models and the Trendy ensemble of DGVMs. The 4 coefficients at the top of the panel indicate robust Theil-Sen’s slopes (± standard error of 5 the slope) with their corresponding p-values, estimated using the non-parametric Wilcoxon 6 Mann-Whitney rank sum test. Values represent spatially-weighted global means. ***, 7 P<0.001. 8 9 Supplementary Figure 2: Raincloud plots showing the global distributions of NBP, 10 NBPPV, NBPAR1 and their trends. Inversions show the results from the combination of 11 CAMS and CarboScope atmospheric inversion models. Boxes indicate the interquartile 12 ranges of the distributions, solid lines indicate the medians and the whiskers extend to 1.5 13 times the interquartile range. 14 15 16 Supplementary Figure 3: Temporal trends in the annual amplitude of atmospheric 17 CO2 concentrations, its interannual variability (PV), and its monthly CO2 temporal 18 autocorrelation. Coloured lines represent the interannual variability of the annual 19 amplitude in atmospheric CO2 concentrations (a) and the 11-year moving average of its 20 variability (b) and monthly CO2 temporal autocorrelation (at lag 1, c) for nine monitoring 21 stations distributed across the Pacific Ocean that covered the study period (1981–2018). 22 Black lines indicate the mean amplitude (a), its temporal variability (b), and its temporal 23 autocorrelation (c) across sites for the study period. Thick lines represent the smoothed 24 trend for each site. Coefficients at the right indicate robust Theil-Sen’s slopes (± standard 25 error of the slope) and their corresponding p-values, estimated using the non-parametric 26 Wilcoxon Mann-Whitney rank sum test. Mean trends were calculated over the study period 27 only (1981–2018). CHR, Christmas Island (2.0°N, 157.3°W); KER, the Kermadec Islands 28 (29.2°S, 177.9°W); KUM, Cape Kumukahi (19.5°N, 154.8°W); LJO, La Jolla Pier (32.9°N, 29 117.3°W); MLO, Mauna Loa Observatory (19.5°N, 155.6°W); NZD, Baring Head (41.4°S, 30 174.9°E); PTB, Point Barrow (71.3°N, 156.6°W); SAM, American Samoa (14.2°S, 31 170.6°W); SPO, South Pole (90.0°S). 32 33 Supplementary Figure 4: Global contribution of regions with concomitantly 34 increasing NBPPV-AR1 and concomitantly decreasing NBPPV-AR1 over time as shown by 35 the combination of CAMS and CarboScope. Aggregated annual sum of NBP from 36 regions with concomitantly increasing (red) and decreasing (blue) NBPPV and NBPAR1. 37 Smooth trends were estimated with a local regression (shaded indicate the standard error 38 of the trend), and the coefficients at the top of the panel indicate robust Theil-Sen’s slopes 39 (± standard error of the slope) and their associated significance (***, P<0.001), estimated 40 using the non-parametric Wilcoxon Mann-Whitney rank sum test. 41 42 43 Supplementary Figure 5: Regions with concomitantly increasing NBPPV-AR1 and 44 concomitantly decreasing NBPPV-AR1 and their trends in mean annual NBP from CAMS 45 and CarboScope atmospheric inversions. Panels a (CAMS) and c (CarboScope) show 46 the spatial distribution of the aggregated NBPPV and NBPAR1 trend, showing increases in 47 red (positive ΔNBPPV-AR1) and decreases in blue (negative ΔNBPPV-AR1). Spatial correlation 48 between both atmospheric inversions was rho=0.10, P<0.001 (Spearman’s correlation). 49 Panels b (CAMS) and d (CarboScope) show the temporal evolution of mean annual NBP 50 from regions increasing (red) and decreasing (blue) ΔNBPPV-AR1. Both atmospheric 51 inversions show stronger increases in NBP over regions with decreasing ΔNBPPV-AR1. Only 52 regions where trends in both indices (PV and AR1) were coincident on increasing or 53 decreasing were used. Smooth trends were estimated with a local regression (shaded 54 indicate the standard error of the trend), and the coefficients at the top of the panel indicate 55 robust Theil-Sen’s slopes (± standard error of the slope) and their associated significance 56 (***, P<0.001), estimated using the non-parametric Wilcoxon Mann-Whitney rank sum test. 57 58 59 Supplementary Figure 6: Temporal trends in mean annual temperature (MAT), mean 60 annual precipitation (MAP), and their temporal interannual variability (PV) and 61 temporal autocorrelation (AR1). Blue lines represent smooth trends (grey, 95% 62 confidence intervals). Coefficients at the top of the panels indicate robust Theil-Sen’s slopes 63 (± standard error of the slope) and their corresponding p-values estimated using the non- 64 parametric Wilcoxon Mann-Whitney rank sum test. Trends in temporal variability and 65 autocorrelation were calculated using an 11-year moving window from 1981 to 2018. Values 66 represent spatially-weighted global means. ***, P<0.001. 67 68 Supplementary Figure 7: Global maps showing trends in mean annual temperature (MAT), mean annual precipitation (MAP), and their 69 temporal interannual variability (PV) and temporal autocorrelation (AR1). Panels a, b and c show trends in mean, PV and AR1 of mean 70 annual temperature; panels d, e and f show trends in mean, PV and AR1 for mean annual precipitation. See Methods for further details. 71 72 Supplementary Figure 8: Maps showing the spatial patterns of the aggregated PV- 73 AR1 trends in temperature and precipitation. Regions with contrasting trends in PV 74 and AR1 appear in grey. See Methods for details regarding the calculation of the 75 aggregated PV and AR1 trends in annual temperature and precipitation. 76 77 Supplementary Figure 9: Response curves for net biome production (NBP), its 78 variability (NBPPV), and its temporal autocorrelation (NBPAR1) showing patterns in 79 atmospheric inversions and TRENDY. Panels a, b, and c show the relationships of 80 NBP with biodiversity, mean atmospheric nitrogen (N) deposition, and the percentage of 81 forested area. Contrasting patterns between NBP derived from atmospheric inversions 82 and from the Trendy ensemble occur mainly for biodiversity and N deposition. Panels d, 83 e, and f show the relationships of NBPPV with biodiversity, N deposition, and the temporal 84 variability of mean annual temperature (MATPV), respectively. The interactive effects of 85 biodiversity and N deposition (see Figure 3) are also shown in panels a, b, d, and e. 86 Panels g, h, and i show the relationships of NBPAR1 with biodiversity, N deposition, and 87 the temporal autocorrelation in monthly temperatures (MATAR1), respectively. Shaded 88 areas represent the 95% credible interval of the slope. Plant biodiversity is shown in units 89 of species per geographic degrees. Effect sizes are shown in Figure 3 and in 90 Supplementary Materials – Model summaries (1-3). 91 92 93 Supplementary Figure 10: Spatial correlations of NBP, NBPPV and NBPAR1 and their 94 trends with their predictors. Points and confidence bands indicate, respectively, the 95 50% percentile and the 95% credible interval of the regression model parameter 96 estimates. Solid dots indicate that the 95% of the distributions of the parameter estimates 97 do not cross zero. See Methods for details on statistical models. 98 99 Supplementary Figure 11: Response curves for trends in net biome production 100 (ΔNBP), its variability (ΔNBPPV), and its monthly temporal autocorrelation (ΔNBPAR1). 101 Panels a, b, and c show the relationships of ΔNBP with N deposition, trends in mean annual 102 temperature (ΔMAT), and trends in mean annual temperature (MAT), respectively. For 103 ΔNBPPV and ΔNBPAR1, the third column (panels f and i) indicates trends in MAT temporal 104 variability and autocorrelation, respectively. Shaded areas represent the 95% credible 105 interval of the slope. Effect sizes are shown in Figure 3 and in Supplementary Materials 106 – Model summaries (4-6). 107 108 109 110 Supplementary Figure 12: Spatial controls of aggregated trends in NBPPV-AR1. Points 111 and confidence bands indicate, respectively, the 50% percentile and the 95% credible 112 interval of the regression model parameter estimates. Solid dots indicate that the 95% of 113 the distributions of the parameter estimates do not cross zero. See Methods for details on 114 statistical models. 115 116 117 Supplementary Figure 13: Sensitivity analyses showing minimal differences when 118 selecting different window lengths (7, 11 and 15 years) to estimate trends in NBPPV 119 and NBPAR1. Blue, black and red lines show, respectively, results for 7, 11 and 15-year 120 window lengths. Coefficients show the linear Theil-Sen’s slopes of the results (± standard 121 error of the slope) and their corresponding p-values estimated using the non-parametric 122 Wilcoxon Mann-Whitney rank sum test. Values represent spatially-weighted global means. 123 124 Supplementary Figure 14: Example of calculation of PV and AR1 metrics (simulated 125 NBP time series). Panel a shows the raw data of NBP (blue dots) and the model adjustment 126 (red line) when controlling for the trend (smoothed term) and the seasonality of the time 127 series using a generalised additive model (GAM). Temporal variability is calculated from 128 raw data, by means of the proportional variability index (PV), as the average across the 129 twelve months of the PV of each month across years (as an example, September data 130 points are shown in red). Panel b and c show, respectively, the partial residuals of the fitted 131 trend and seasonality estimated by the GAM model. Panel d shows the residuals of the 132 model from which we finally calculated the temporal autocorrelation at lag 1 for the period. 133 134 Supplementary Figure 15: Map showing the approximate magnitude of the change in 135 NBP due to changing temporal interannual variability over the study period. See 136 Supplementary text, section 1 for further details on the calculation of the magnitude of 137 change. 138 139 Supplementary Table 1: Sensitivity analyses showing strong spatial correlations 140 amongst the results obtained using different window lengths (7, 11 and 15 years) to 141 estimate trends in NBPPV and NBPAR1. Values indicate Spearman’s correlations (rho). The 142 upper part of the diagonals shows correlations amongst AR1 results from different widow 143 lengths, the lower part of the diagonals shows results for PV. d) Show the spatial correlation 144 for the aggregated trends in PV and AR1 (see Figure 2). 7 years 11 years 15 years a) CAMS + CarboScope 7 years 1 0.81 0.68 11 years 0.90 1 0.90 15 years 0.83 0.94 1 b) CAMS 7 years 1 0.77 0.62 11 years 0.91 1 0.89 15 years 0.76 0.93 1 c) CarboScope 7 years 1 0.76 0.59 11 years 0.83 1 0.86 15 years 0.69 0.90 1 d) CAMS + CarboScope (PV + AR1 trends) 7 years 1 0.88 0.78 11 years - 1 0.94 15 years - - 1 145 146 147 2. Supplementary notes 148 How much does the temporal behaviour of regional NBP change given the reported 149 increase in PV? 150 We can illustrate the magnitude of change in NBP given an increase in variability with an 151 example, but because PV is calculated by the proportional comparison of all pair-wise 152 values in a time series, and the example is based on average values, the results may not 153 be exact. For example, the Iberian Peninsula shows a value for NBPPV of around 0.25 and 154 a trend in NBPPV of around 0.003 y-1 (Figure 1). Considering that average NBP for that 155 region during the study period is around 20 gC m-2 y-1, the increase in variability over 28 156 years (1981 to 2018 with an 11-year moving window) would be around 0.084, which 157 represents an increase of 8.4% in variability. That increase means that if NBP would have 158 remained constant, and initial NBPPV would have been 0.208 (increasing to 0.292 by the 159 end of the study period at a rate of 0.003 per year), the average range of variability would 160 have gone from 20 x 0.208 = 4.16 gC m-2 y-1 to 20 x 0.292 = 5.84 gC m-1 y-1. Given the large 161 heterogeneity of trends in NBPPV at the global scale, the reported average global increase 162 in PV as a result of the combination of CAMS and CarboScope is small (8.3·10-4, Figure 1f) 163 which represents an increase of PV of 0.023 over the study period (2.3% increase in 164 variability). So, if NBP would have remained constant at 10 gC m-2 y-1 at the global scale, 165 the reported increase in NBPPV over an average NBPPV of around 0.28, would have entailed 166 an increase in variability from 10 x 0.238 = 2.38 gC m-2 y-1 to 10 x 0.262 = 2.62 gC m-2 y-1 167 over the entire world. Supplementary Figure 15 shows the regional magnitude of change 168 in NBP with changing temporal variability following the methodology explained above. 169 For temporal autocorrelation, however, there is literally no possible way to assess how an 170 increase in autocorrelation would affect values of NBP in the previous manner. That metric 171 describes how strongly correlated are consecutive numbers. Therefore, our results 172 combining trends in NBPPV and NBPAR1 have the same drawback due to the impossibility to 173 assess the impact of AR1 on values of NBP. 174 175 176 3. Model summaries 177 Variable names 178 Biodiversity: plant species richness (biodiver), I(biodiver^2) indicates the x2 part of a 179 polynomial fit (i.e., biodiversity + biodiversity2). 180 Atmospheric N deposition: mean annual atmospheric N deposition (mean.ndep.tot). 181 Climate: mean annual temperature (tmps.mean), mean sum of annual precipitation 182 (pres.mean), temporal variability in temperature (tmps.pv), temporal variability in 183 precipitation (pres.pv), temporal autocorrelation in temperature (tmps.ar1), temporal 184 autocorrelation in precipitation (pres.ar1), aggregated index combining trends in PV and 185 AR1 (temperature: tmps.pv.ar1.trend; precipitation: pres.pv.ar1.trend). 186 Climate change: trend in mean annual temperature (tmps.mean.trend), trend in mean sum 187 of annual precipitation (pres.mean.trend), trend in the temporal variability in temperature 188 (tmps.pv.trend), trend in the temporal variability in precipitation (pres.pv.trend), trend in the 189 temporal autocorrelation of temperature (tmps.ar1.trend), trend in the temporal 190 autocorrelation of precipitation (pres.ar1.trend). 191 Land use: mean percentage of forested land (mean.forested), mean percentage of 192 agricultural land (mean.croped), mean percentage of urbanised land (mean.urbaned). 193 Land use change: trend in mean percentage of forested land (trnd.forested), trend in mean 194 percentage of agricultural land (trnd.croped), trend in mean percentage of urbanised land 195 (trnd.urbaned). 196 Additional estimated parameters: nu2, tau2 and rho are estimated by the spatial 197 regression model. 198 Bold coefficients indicate that 95% of the posterior distribution did not overlap 0. All variables 199 were standardised before fitting the model (mean=0, sd=1). 200 201 1. Spatial variability in mean NBP 202 1.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0524 0.0288 0.0763 3851.9 1.9 biodiver 0.1708 0.097 0.2435 2612.2 0.6 mean.ndep.tot 0.2257 0.1538 0.3016 1858.8 0.4 I(biodiver^2) -0.1222 -0.1461 -0.0984 3607.7 -1.2 tmps.mean -0.0263 -0.1144 0.0643 481 -0.1 pres.mean -0.0909 -0.1524 -0.0288 3079.9 -1.2 mean.forested 0.1019 0.0682 0.1366 3776.4 -1 mean.croped 0.0716 0.0389 0.1047 5161.3 0.6 mean.urbaned -0.0243 -0.0529 0.0045 4875.2 -0.2 biodiver:mean.ndep.tot 0.1324 0.0958 0.1687 3407.5 -0.8 nu2 0.0023 0.0012 0.0046 5322 0.2 tau2 5.4106 5.0299 5.7928 63883 -1 rho 0.9642 0.9114 0.9910 35213.9 -1 DIC = -5081.876 p.d = 2038.19 LMPL = 1522.32 203 204 1.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0983 0.0686 0.1277 4470.3 0 biodiver 0.3968 0.3013 0.4915 2000.2 0.8 mean.ndep.tot 0.1167 0.0215 0.2102 1761.5 -0.6 I(biodiver^2) 0.0193 -0.0111 0.0496 4230.2 -0.6 tmps.mean 0.1600 0.0409 0.2787 464.1 0.9 pres.mean 0.0551 -0.024 0.1323 3505.8 -1.1 mean.forested 0.1238 0.0790 0.1681 4292.4 -0.3 mean.croped -0.2862 -0.3290 -0.2443 5336.2 1.1 mean.urbaned -0.0120 -0.0489 0.0252 5519.4 -0.5 biodiver:mean.ndep.tot -0.2231 -0.2678 -0.1781 4112.1 0.8 nu2 0.0045 0.0018 0.0126 2698.1 -1 tau2 8.7307 8.0748 9.3708 24383.1 1.1 rho 0.9614 0.9031 0.9905 11797.1 1.3 DIC = -3732.725 p.d = 1885.291 LMPL = 736.85 205 206 207 1.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.1175 0.0888 0.1471 3372.3 2.7 biodiver 0.2078 0.1173 0.2987 2291 1.7 mean.ndep.tot 0.3021 0.2140 0.3880 1714.9 0.8 I(biodiver^2) -0.1635 -0.1927 -0.1341 3298.5 -0.8 tmps.mean -0.1294 -0.2406 -0.0153 435 1.1 pres.mean -0.1418 -0.2179 -0.0678 2852.8 -0.5 mean.forested 0.0717 0.0287 0.1155 3144.2 -0.6 mean.croped 0.0798 0.0395 0.1204 4509 -0.6 mean.urbaned -0.0147 -0.0503 0.0210 4343.7 -0.4 biodiver:mean.ndep.tot 0.0872 0.0425 0.1308 3074.3 -2.1 nu2 0.0033 0.0014 0.0080 3523.7 -0.8 tau2 7.9095 6.9563 8.6804 28803.1 1.6 rho 0.9088 0.7984 0.9752 22688.7 0.7 DIC = -4355.835 p.d = 1964.491 LMPL = 1111.89 208 209 1.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0533 -0.0719 -0.0350 3999.6 0.1 biodiver 0.0537 -0.0048 0.1127 2469.3 -1.2 mean.ndep.tot 0.0439 -0.0138 0.1009 1752.6 -0.3 I(biodiver^2) -0.0213 -0.0402 -0.0024 3416.4 0 tmps.mean 0.1142 0.0432 0.1860 466.3 -1.9 pres.mean 0.0107 -0.0388 0.0591 3182 0.9 mean.forested 0.1048 0.0773 0.1327 3681.8 0.8 mean.croped 0.0334 0.0078 0.0597 4810.7 1.1 mean.urbaned -0.0267 -0.0498 -0.0039 4595.8 0.5 biodiver:mean.ndep.tot 0.1417 0.1132 0.1695 3146.8 -0.1 nu2 0.0015 0.0009 0.0026 8304.2 1.1 tau2 3.5164 3.3101 3.7390 70000 -1.7 rho 0.9871 0.9660 0.9970 68207.9 -2.3 DIC = -6014.875 p.d = 2091.966 LMPL = 2042.7 210 211 212 2. Spatial variability in NBPPV 213 2.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.2392 0.2166 0.2616 5138 -0.3 biodiver 0.9535 0.8929 1.0144 4127.4 -1.3 mean.ndep.tot 0.2172 0.1500 0.2828 2403.9 1.3 I(biodiver^2) -0.1459 -0.1684 -0.1233 4616.6 0.7 tmps.pv 0.5513 0.4586 0.6416 602.6 -0.7 pres.pv 0.0319 0.0010 0.0633 4429.1 -1.2 mean.forested 0.0510 0.0171 0.0848 4499 1.4 mean.croped 0.0719 0.0405 0.1033 7022.6 0.1 mean.urbaned -0.0292 -0.0564 -0.0019 6408.1 0 biodiver:mean.ndep.tot -0.1772 -0.2111 -0.1435 4075.4 -0.7 nu2 0.0029 0.0014 0.0064 4124.9 0.2 tau2 4.7717 4.3670 5.1426 35670.3 -0.2 rho 0.9485 0.8762 0.9870 30714.9 -0.6 DIC = -4635.541 p.d = 1991.098 LMPL = 1290.43 214 215 2.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.2090 0.1893 0.2290 4641.8 1.5 biodiver 0.6324 0.5784 0.6866 3462.1 1.3 mean.ndep.tot 0.1970 0.1392 0.2564 2365.3 0.3 I(biodiver^2) -0.1331 -0.1535 -0.1130 4481.8 -1 tmps.pv 0.2602 0.1825 0.3374 561 0.3 pres.pv 0.2171 0.1889 0.2447 3906.8 0.6 mean.forested 0.1522 0.1222 0.1816 3999.7 -0.5 mean.croped 0.1508 0.1232 0.1783 6345.2 -0.9 mean.urbaned -0.0132 -0.037 0.0114 5775 0.5 biodiver:mean.ndep.tot -0.1442 -0.1742 -0.1150 3963 -0.6 nu2 0.0021 0.0011 0.0040 6053 -0.3 tau2 3.7970 3.5037 4.0750 55672.7 -0.4 rho 0.9566 0.8943 0.9890 34951 -1.2 DIC = -5299.126 p.d = 2047.861 LMPL = 1673.68 216 217 218 2.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.2299 0.2056 0.2543 4502 -2 biodiver 0.7554 0.6900 0.8216 3865.6 0.5 mean.ndep.tot 0.3605 0.2896 0.4319 2217.5 -1.5 I(biodiver^2) -0.1281 -0.1521 -0.1035 4410.8 0.8 tmps.pv 0.4070 0.3089 0.5015 537.1 1.9 pres.pv 0.0012 -0.0328 0.0353 4087.3 0.8 mean.forested 0.0845 0.0482 0.1201 4224.9 -0.6 mean.croped 0.0747 0.0408 0.1083 6304 1.4 mean.urbaned -0.0168 -0.0466 0.0131 6037.6 1.9 biodiver:mean.ndep.tot -0.1937 -0.2303 -0.1567 3672.8 1.5 nu2 0.0031 0.0014 0.0070 4263.4 -0.6 tau2 5.4439 4.8453 5.9396 25754.6 -2.8 rho 0.9201 0.8194 0.9786 21242.2 -3 DIC = -4477.512 p.d = 1987.668 LMPL = 1196.64 219 220 2.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.1763 0.1520 0.2014 4026.2 0.4 biodiver 0.9621 0.8956 1.0308 3266.8 1.4 mean.ndep.tot -0.0828 -0.1527 -0.0089 2033.8 -1.8 I(biodiver^2) -0.1272 -0.1520 -0.1027 3983.2 -1.8 tmps.pv 0.6152 0.5184 0.7115 493.8 -1.8 pres.pv 0.0712 0.0374 0.1056 3738.4 1.7 mean.forested -0.0179 -0.0546 0.0190 3634.1 0.1 mean.croped 0.0425 0.0083 0.0763 5690.4 1.3 mean.urbaned -0.0391 -0.0694 -0.0092 5095.2 1.7 biodiver:mean.ndep.tot -0.0934 -0.1298 -0.0569 3218.4 1.8 nu2 0.0028 0.0013 0.0059 4535 0.3 tau2 5.7641 5.3175 6.1845 57264.5 0 rho 0.9560 0.8938 0.9891 47159.3 1.3 DIC = -4718.69 p.d = 2011.388 LMPL = 1295.89 221 222 223 3. Spatial variability in NBPAR1 224 3.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0499 0.0240 0.0753 4334.5 -1.1 biodiver 0.1758 0.1061 0.2445 3131 1 mean.ndep.tot -0.0805 -0.1545 -0.0051 2551.5 -2.5 I(biodiver^2) -0.0378 -0.0638 -0.0115 4484.6 -1.5 tmps.ar1 0.0830 0.0288 0.1367 1875.5 1.1 pres.ar1 -0.0044 -0.0317 0.0231 6768.8 -1.2 mean.forested -0.0215 -0.0596 0.0155 3670 1.5 mean.croped 0.0637 0.0280 0.1000 6385.1 0 mean.urbaned 0.0121 -0.0193 0.0432 6006.9 -0.3 biodiver:mean.ndep.tot -0.0229 -0.0624 0.0160 3931.2 3 nu2 0.0037 0.0016 0.0092 3385.7 0.3 tau2 6.4355 5.9658 6.8948 41076.7 1.1 rho 0.9632 0.9089 0.9910 64138.6 -0.2 DIC = -4123.148 p.d = 1939.636 LMPL = 974.04 225 226 3.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0193 -0.0395 0.0008 5100.6 0.4 biodiver -0.1311 -0.1853 -0.0769 3620.2 1 mean.ndep.tot 0.0810 0.0234 0.1377 2858.3 -0.7 I(biodiver^2) 0.0259 0.0060 0.0459 5600.9 -0.4 tmps.ar1 0.1433 0.1016 0.1853 2131.7 -0.1 pres.ar1 0.0899 0.0690 0.1109 8145.7 0.7 mean.forested -0.1492 -0.1791 -0.1198 4217.7 1.3 mean.croped 0.0144 -0.0130 0.0423 7772 -1.6 mean.urbaned 0.0354 0.0112 0.0595 6928 0.3 biodiver:mean.ndep.tot -0.0129 -0.0423 0.0172 4539.6 0 nu2 0.0026 0.0013 0.0052 5069.3 -0.6 tau2 3.8896 3.6475 4.1442 56501.3 0.6 rho 0.9816 0.9521 0.9956 65592 -1.4 DIC = -4875.772 p.d = 2014.984 LMPL = 1443.5 227 228 229 3.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0844 0.0462 0.1236 2880.9 -0.4 biodiver 0.2919 0.1871 0.3994 2315 -1.1 mean.ndep.tot -0.1473 -0.2588 -0.0373 1872.3 0.5 I(biodiver^2) -0.0559 -0.0954 -0.0168 3090.6 0.3 tmps.ar1 0.1320 0.0536 0.2137 1367.9 -0.3 pres.ar1 0.0065 -0.0352 0.0477 4507.2 0.3 mean.forested -0.0466 -0.1037 0.0120 2365.4 1.7 mean.croped 0.0990 0.0438 0.1536 4197.3 0.3 mean.urbaned 0.0192 -0.0279 0.0683 4161.1 0.6 biodiver:mean.ndep.tot -0.0539 -0.1119 0.0028 3170.8 0.1 nu2 0.0056 0.0020 0.0181 2147.1 0.5 tau2 11.703 9.0062 14.309 15815.7 0.9 rho 0.7356 0.5563 0.9017 15312.7 0.9 DIC = -3325.409 p.d = 1813.241 LMPL = 433.5 230 231 3.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0055 -0.0123 0.0012 6870.7 1.9 biodiver -0.0213 -0.0395 -0.0031 5413.6 1.4 mean.ndep.tot 0.0290 0.0096 0.0487 4105.6 1.2 I(biodiver^2) -0.0021 -0.0089 0.0046 7637.1 -1.4 tmps.ar1 -0.0076 -0.0217 0.0061 2633.2 -1.1 pres.ar1 -0.0123 -0.0193 -0.0054 11649.3 0 mean.forested 0.0134 0.0035 0.0232 5859.7 -1.9 mean.croped 0.0017 -0.0076 0.0110 11310.5 -0.8 mean.urbaned -0.0006 -0.0088 0.0075 10564.5 0.2 biodiver:mean.ndep.tot 0.0145 0.0043 0.0245 6630.4 -0.7 nu2 0.0004 0.0003 0.0006 27719.2 -1 tau2 0.4438 0.4182 0.4713 68976.4 1.2 rho 0.9989 0.9971 0.9998 69210 1.1 DIC = -8832.31 p.d = 2122.28 LMPL = 3598.25 232 233 4. Spatial variability in trends of mean NBP 234 4.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0078 -0.0270 0.0415 2779.4 -0.3 biodiver 0.0113 -0.1036 0.1224 1779 -1.2 I(biodiver^2) -0.0078 -0.0414 0.0269 2881.2 0.3 mean.ndep.tot 0.2035 0.1098 0.2932 1357 -0.4 tmps.mean -0.2907 -0.4304 -0.1531 362.5 -1.7 pres.mean 0.0196 -0.0735 0.1170 2281.3 0.8 tmps.mean.trend -0.0968 -0.1626 -0.0297 1888 -0.4 pres.mean.trend -0.0114 -0.0519 0.0287 3230.2 0.2 mean.forested 0.0843 0.0265 0.1433 2819.1 -0.7 mean.croped 0.0222 -0.0287 0.0721 3739.5 0.2 mean.urbaned 0.1426 0.0591 0.2264 2979 0.4 trnd.forested -0.0641 -0.1149 -0.014 2827.9 0.9 trnd.croped 0.0447 -0.0019 0.0920 2794.5 -0.5 trnd.urbaned -0.1041 -0.1815 -0.027 3019.7 -0.1 nu2 0.0040 0.0016 0.0103 3141.1 0.6 tau2 11.795 10.192 13.082 31132.6 1.3 rho 0.8896 0.7656 0.9693 26303.5 2 DIC = -3990.262 p.d = 1933.688 LMPL = 879.6 235 236 4.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0013 -0.0372 0.0334 2925.6 0.9 biodiver 0.1143 -0.0047 0.2304 1755.4 0.5 I(biodiver^2) 0.0013 -0.0334 0.0371 2902.8 -0.9 mean.ndep.tot 0.1469 0.0544 0.2419 1655.4 -0.6 tmps.mean -0.0575 -0.1879 0.0799 467.6 -0.3 pres.mean -0.1119 -0.2058 -0.0167 2419.3 -0.1 tmps.mean.trend -0.2134 -0.2814 -0.1473 2234.3 -1.1 pres.mean.trend 0.3061 0.2661 0.3457 3874.3 0.4 mean.forested 0.0774 0.0171 0.1374 2705.3 0.9 mean.croped -0.0884 -0.1384 -0.0382 4399.3 0 mean.urbaned 0.0254 -0.0615 0.1115 3613.8 0 trnd.forested -0.0473 -0.0982 0.0037 3313.3 1.5 trnd.croped -0.1335 -0.1808 -0.0853 3056.3 0.8 trnd.urbaned -0.0863 -0.1657 -0.0068 3811.3 0.3 nu2 0.0048 0.0018 0.0135 2543.9 -0.1 tau2 11.009 8.7874 12.939 11120.3 -0.5 rho 0.7949 0.625 0.9308 10326.3 -0.9 DIC = -3616.401 p.d = 1882.808 LMPL = 658.06 237 4.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0036 -0.0379 0.0325 2739.9 -0.3 biodiver -0.0011 -0.1164 0.1166 1884.9 -0.2 I(biodiver^2) 0.0036 -0.0324 0.0378 2797.1 0.3 mean.ndep.tot 0.1898 0.0917 0.2824 1337.4 1.8 tmps.mean -0.2747 -0.406 -0.1335 451.3 0.4 pres.mean 0.0164 -0.0846 0.1139 1986.3 -0.5 tmps.mean.trend -0.0777 -0.1464 -0.0070 1750 1.6 pres.mean.trend -0.0126 -0.0539 0.0286 3101.6 0.6 mean.forested 0.0821 0.0220 0.1438 2512.5 0.7 mean.croped 0.0249 -0.0267 0.0764 3526.1 -1.1 mean.urbaned 0.1607 0.0718 0.2465 2844.5 0.2 trnd.forested -0.0661 -0.1192 -0.0148 3171.9 -0.5 trnd.croped 0.0337 -0.0160 0.0838 2635.5 0.2 trnd.urbaned -0.1195 -0.1985 -0.0364 2837.6 -0.8 nu2 0.0041 0.0017 0.0108 2868.9 1 tau2 12.272 10.223 13.956 23157.1 -0.5 rho 0.8469 0.6993 0.9538 25272.5 -0.5 DIC = -3908.619 p.d = 1927.172 LMPL = 867.15 4.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0421 0.0285 0.0557 5015.3 0.8 biodiver 0.0475 0.0031 0.0913 3377.6 0.1 I(biodiver^2) -0.0421 -0.0557 -0.0286 4800.1 -0.8 mean.ndep.tot 0.0687 0.0322 0.1052 2111.1 -1.7 tmps.mean -0.1164 -0.1704 -0.0632 633.1 0.8 pres.mean 0.0230 -0.0148 0.0609 3605 0.2 tmps.mean.trend -0.0815 -0.1084 -0.0552 3042.2 0 pres.mean.trend 0.0006 -0.0151 0.0164 5546 1.2 mean.forested 0.0287 0.0054 0.0517 4644.6 -1 mean.croped -0.0040 -0.0236 0.0157 6429.8 1.6 mean.urbaned -0.037 -0.0701 -0.0034 5105.2 -1 trnd.forested -0.0044 -0.0244 0.0157 5541.3 0.4 trnd.croped 0.0474 0.0289 0.0661 4651.5 0.3 trnd.urbaned 0.0331 0.0021 0.0642 5167.9 0.1 nu2 0.0011 0.0007 0.0017 11516.9 0.8 tau2 2.0656 1.9462 2.1941 67101.3 -1.2 rho 0.9938 0.9833 0.9985 70000 -1.6 DIC = -6706.438 p.d = 2108.892 LMPL = 2458.83 238 239 5. Spatial variability in trends of NBPPV 240 5.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0235 -0.0101 0.0573 3223 -0.3 biodiver 0.0017 -0.0946 0.0997 2342.4 0.1 I(biodiver^2) -0.0235 -0.0571 0.0100 3206.6 0.2 mean.ndep.tot -0.1634 -0.2516 -0.0769 1712.3 0.3 tmps.mean.trend -0.1002 -0.1638 -0.0341 1665.9 0.6 pres.mean.trend -0.0290 -0.0672 0.0085 3916.1 -0.6 tmps.pv.trend 0.1644 0.1036 0.2236 1785.7 1.1 pres.pv.trend -0.0974 -0.1388 -0.0566 3773.7 -0.1 mean.forested -0.0264 -0.0824 0.0316 2328.8 1.2 mean.croped 0.1013 0.0529 0.1507 4144.7 -0.4 mean.urbaned -0.0398 -0.1222 0.0426 3403 1.9 trnd.forested -0.0602 -0.1094 -0.0093 3595.3 1 trnd.croped 0.0534 0.0067 0.0992 2949 0.8 trnd.urbaned 0.0637 -0.0132 0.1395 3461.7 -1.2 nu2 0.0044 0.0017 0.0116 2939.1 0.8 tau2 11.102 9.4125 12.455 35986.7 0.9 rho 0.8686 0.7316 0.9614 34169.8 1 DIC = -3804.119 p.d = 1910.947 LMPL = 784.98 241 242 5.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0231 -0.0067 0.0521 3693.2 0.7 biodiver -0.0084 -0.0926 0.0753 2781.5 1.7 I(biodiver^2) -0.0231 -0.0519 0.0064 3663.5 -0.8 mean.ndep.tot -0.0121 -0.0861 0.0631 2432.8 -1.2 tmps.mean.trend 0.2373 0.1813 0.2946 2074.6 -0.6 pres.mean.trend -0.1402 -0.174 -0.1064 4973.1 -0.1 tmps.pv.trend 0.1183 0.0641 0.1711 2355.2 0.4 pres.pv.trend 0.3919 0.3558 0.4278 4755.3 0.3 mean.forested 0.0185 -0.0317 0.0687 3400 -0.3 mean.croped 0.0388 -0.0042 0.0823 5495 0.6 mean.urbaned -0.0656 -0.139 0.0063 4268.7 -1.4 trnd.forested -0.0432 -0.0872 0.0018 4495.5 1.1 trnd.croped 0.0657 0.0252 0.1059 3858.1 0.2 trnd.urbaned 0.0627 -0.0046 0.1301 4531.2 1.1 nu2 0.0045 0.0018 0.0127 2625 -0.7 tau2 7.9472 6.351 9.3386 22991.3 0 rho 0.7953 0.6263 0.9308 22049.9 0.3 DIC = -3732.416 p.d = 1880.682 LMPL = 742.6 5.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0082 -0.0267 0.0432 3302.4 -0.4 biodiver 0.0217 -0.0806 0.1239 2235 -0.4 I(biodiver^2) -0.0082 -0.0431 0.0265 3307.1 0.4 mean.ndep.tot -0.1599 -0.2522 -0.0670 1636.4 0 tmps.mean.trend 0.0274 -0.0409 0.0934 1700.9 -1.6 pres.mean.trend -0.0333 -0.0735 0.0075 3607.9 -1.3 tmps.pv.trend 0.1295 0.0676 0.1926 1802.9 0.4 pres.pv.trend -0.0945 -0.1369 -0.0510 3558.6 -0.3 mean.forested -0.0531 -0.1133 0.0068 2333 -0.7 mean.croped 0.0938 0.0427 0.1465 4124.6 -1.3 mean.urbaned -0.0665 -0.1528 0.0200 3365.4 -0.2 trnd.forested -0.0276 -0.0782 0.0250 3561.9 -0.3 trnd.croped 0.0595 0.0103 0.1073 2657 1.8 trnd.urbaned 0.0773 -0.0030 0.1566 3558.6 1.4 nu2 0.0048 0.0018 0.0134 2416.3 0.4 tau2 11.855 9.7117 13.637 29616.6 -1.6 rho 0.8282 0.6720 0.9454 28108.7 -1.6 DIC = -3623.725 p.d = 1881.522 LMPL = 633.15 5.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0478 0.0237 0.0719 3247 -1.5 biodiver -0.0430 -0.1136 0.0266 2451.3 -0.6 I(biodiver^2) -0.0478 -0.0718 -0.0237 3211.3 1.5 mean.ndep.tot -0.0617 -0.1267 0.0003 1475.3 -0.9 tmps.mean.trend -0.3634 -0.4103 -0.3166 1773.4 -0.1 pres.mean.trend 0.0023 -0.0255 0.0303 3592.2 0.3 tmps.pv.trend 0.1401 0.0959 0.1843 1690.7 0.9 pres.pv.trend -0.0376 -0.0670 -0.0075 3636.2 1.1 mean.forested 0.0646 0.0238 0.1048 2660.3 0.4 mean.croped 0.0464 0.0104 0.0816 4379.3 1.8 mean.urbaned 0.0579 -0.0011 0.1167 3584 1.3 trnd.forested -0.1000 -0.1349 -0.0649 3831 0.1 trnd.croped 0.0016 -0.0314 0.0341 2941.2 0.2 trnd.urbaned -0.0174 -0.0710 0.0372 3681.2 -0.8 nu2 0.0023 0.0012 0.0045 5539.6 -1.5 tau2 6.3211 5.8927 6.7523 68535.6 -1 rho 0.9677 0.9195 0.9922 64800 1.2 DIC = -5124.036 p.d = 2049.197 LMPL = 1556.57 243 6. Spatial variability in trends of NBPAR1 244 6.1. Atmospheric inversions (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0285 -0.0605 0.0035 3453.9 0.9 biodiver -0.0431 -0.1400 0.0512 2275.8 1.7 I(biodiver^2) 0.0284 -0.0034 0.0604 3417.2 -0.9 mean.ndep.tot -0.0912 -0.1741 -0.0092 1914.3 1.4 tmps.mean.trend 0.1082 0.0475 0.1711 1980.3 -0.6 pres.mean.trend -0.0194 -0.0568 0.0176 4247.3 0.4 tmps.ar1.trend 0.0353 -0.0167 0.0867 2257.8 -0.3 pres.ar1.trend 0.0343 0.0017 0.0673 5461.3 -0.8 mean.forested -0.1445 -0.1990 -0.0893 2593.4 -0.9 mean.croped 0.0518 0.0043 0.0984 4920.1 -2.6 mean.urbaned 0.0181 -0.0617 0.0961 4030.4 0.7 trnd.forested -0.0022 -0.0494 0.0453 4065 -1.2 trnd.croped 0.0214 -0.0230 0.0654 3397.7 0.7 trnd.urbaned 0.0064 -0.0648 0.0801 4020.7 -0.2 nu2 0.0047 0.0018 0.0134 2580.7 -1 tau2 10.858 9.6371 11.864 48380.5 -0.2 rho 0.9173 0.8162 0.9778 58817.2 0.2 DIC = -3666.425, p.d = 1870.205 LMPL = 701.98 245 6.2. TRENDY Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0591 0.0240 0.0946 4016.3 0.9 biodiver 0.2276 0.1260 0.3303 2810.5 -0.8 I(biodiver^2) -0.0591 -0.0945 -0.0242 4062.1 -0.9 mean.ndep.tot -0.0969 -0.1864 -0.0084 2278.3 0.2 tmps.mean.trend 0.1473 0.0803 0.2140 2320.1 1 pres.mean.trend -0.1171 -0.1586 -0.0764 4601.9 2.1 tmps.ar1.trend 0.2419 0.1854 0.2993 2742.8 -0.1 pres.ar1.trend 0.2639 0.2272 0.2997 6121 -0.5 mean.forested 0.0146 -0.0462 0.0757 2701.5 -0.1 mean.croped 0.1118 0.0592 0.1638 5469.4 0.7 mean.urbaned 0.0042 -0.083 0.0884 4521.7 -1.8 trnd.forested -0.0220 -0.0755 0.0314 4351.7 -0.9 trnd.croped -0.0376 -0.0863 0.0109 3779.3 -0.5 trnd.urbaned 0.0064 -0.0726 0.0875 4440.2 1.3 nu2 0.0067 0.0022 0.0237 1869.9 0.3 tau2 11.513 9.1933 13.559 20720.6 0.5 rho 0.7919 0.6237 0.9291 20608.4 0.5 DIC = -3000.355 p.d = 1733.364 LMPL = 260.9 246 247 6.3. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0309 -0.0661 0.0046 3205.2 -0.7 biodiver -0.0347 -0.1392 0.0675 2207 -1.2 I(biodiver^2) 0.0309 -0.0047 0.0660 3163.8 0.7 mean.ndep.tot -0.0933 -0.1858 -0.0042 1737.4 -1.2 tmps.mean.trend 0.1042 0.035 0.1696 1936.4 0.8 pres.mean.trend -0.0023 -0.0442 0.0385 3544.3 0.4 tmps.ar1.trend 0.0436 -0.0119 0.0998 2114.6 -0.8 pres.ar1.trend 0.0290 -0.0072 0.0649 5097.1 -0.9 mean.forested -0.1508 -0.2095 -0.0893 2499.3 1 mean.croped 0.0508 -0.0025 0.1024 4236.4 1 mean.urbaned 0.0305 -0.056 0.1169 3782.5 0.8 trnd.forested -0.0131 -0.0636 0.0395 3886.6 1 trnd.croped 0.0137 -0.0357 0.0620 3080.7 1 trnd.urbaned -0.0033 -0.0827 0.0768 3789.1 -0.7 nu2 0.0054 0.0019 0.0167 2316.3 -1.9 tau2 12.085 10.093 13.725 34690.7 0.9 rho 0.8486 0.7034 0.9537 33756.8 0.5 DIC = -3392.847 p.d = 1830.146 LMPL = 506.34 248 249 6.4. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.0032 -0.0185 0.0121 3955.7 -0.2 biodiver -0.0319 -0.0767 0.0132 2748.7 0.1 I(biodiver^2) 0.0032 -0.0121 0.0184 3926.3 0.2 mean.ndep.tot -0.0171 -0.0571 0.0219 1966.6 0 tmps.mean.trend 0.0453 0.0152 0.0753 2047.8 -0.8 pres.mean.trend -0.0388 -0.0567 -0.0210 4475.8 0.9 tmps.ar1.trend -0.0075 -0.0313 0.0171 2459.4 -0.1 pres.ar1.trend 0.0192 0.0035 0.0348 5465.9 -0.9 mean.forested -0.0209 -0.0465 0.0051 2861.9 -0.2 mean.croped 0.0168 -0.0057 0.0388 5265.1 1.3 mean.urbaned -0.0191 -0.0559 0.0186 4299.2 -1.2 trnd.forested 0.0213 -0.0012 0.043 4649.5 0.9 trnd.croped 0.0188 -0.0022 0.0398 3289.4 0.2 trnd.urbaned 0.0197 -0.0157 0.0541 4087.8 0.6 nu2 0.0011 0.0007 0.0018 11387.3 1.9 tau2 2.6533 2.5009 2.8180 70022.4 1.1 rho 0.9923 0.9796 0.9982 70000 0.3 DIC = -6626.257 p.d = 2113.523 LMPL = 2409.82 250 251 7. Spatial variability in trends of the aggregated NBPPV-AR1 index 252 7.1. Atmospheric inversions combined (CAMS + CarboScope) Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.004 -0.038 0.030 3447.5 1.1 biodiver -0.001 -0.112 0.109 1998.1 2 I(biodiver^2) 0.004 -0.029 0.038 3419.7 -1 mean.ndep.tot -0.190 -0.279 -0.103 1933.9 -0.7 tmps.mean 0.217 0.077 0.352 468.9 1.5 pres.mean -0.026 -0.117 0.066 2335.6 -0.9 tmps.mean.trend 0.027 -0.038 0.093 2307.1 0.4 pres.mean.trend -0.034 -0.072 0.005 3891.4 0.9 tmps.pv.ar1.trend 0.078 0.024 0.133 2321 -1.1 pres.pv.ar1.trend 0.034 -0.001 0.069 5130.4 -0.1 mean.forested -0.129 -0.187 -0.072 3129.6 -0.9 mean.croped 0.082 0.034 0.131 4593.3 -0.5 mean.urbaned -0.009 -0.090 0.073 3779.7 -2.1 trnd.forested -0.046 -0.095 0.003 3992.4 -1.6 trnd.croped 0.041 -0.004 0.086 3641.6 1.8 trnd.urbaned 0.039 -0.036 0.113 4034.9 1.9 nu2 0.005 0.002 0.012 2762.2 -0.3 tau2 11.296 9.841 12.461 41447.4 0.7 rho 0.898 0.781 0.971 31696.2 0.8 DIC = -3712.121 p.d = 1901.937 LMPL = 741.53 253 254 7.2. CAMS Median 2.50% 97.50% n.effective Geweke.diag (Intercept) -0.016 -0.051 0.020 3246 -0.9 biodiver 0.006 -0.112 0.120 2273.9 -3.3 I(biodiver^2) 0.016 -0.020 0.051 3214.7 0.9 mean.ndep.tot -0.185 -0.280 -0.091 1804 2.4 tmps.mean 0.141 0.003 0.282 516.3 -1.8 pres.mean -0.017 -0.114 0.080 2531.4 0.6 tmps.mean.trend 0.091 0.020 0.159 2239.4 0.7 pres.mean.trend -0.025 -0.066 0.018 3826 1.3 tmps.pv.ar1.trend 0.075 0.019 0.131 2093.2 1.2 pres.pv.ar1.trend 0.034 -0.003 0.072 4918 1.4 mean.forested -0.149 -0.211 -0.089 3057.8 0.5 mean.croped 0.082 0.030 0.133 4725.6 0.9 mean.urbaned -0.020 -0.107 0.068 3849.2 -0.4 trnd.forested -0.033 -0.085 0.020 3871 0.1 trnd.croped 0.040 -0.009 0.090 3192.8 -1.2 trnd.urbaned 0.044 -0.036 0.125 3828.9 0.2 nu2 0.005 0.002 0.015 2466.8 0.1 tau2 11.846 9.649 13.720 27259.2 -1.4 rho 0.818 0.658 0.941 26704.7 -1.3 DIC = -3547.387 p.d = 1850.76 LMPL = 602.75 255 7.3. CarboScope Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.0408 0.0214 0.0598 4349.9 1 biodiver -0.0146 -0.0792 0.0484 2202.3 2.9 I(biodiver^2) -0.0408 -0.0597 -0.0216 4313.9 -1 mean.ndep.tot -0.0913 -0.1435 -0.0385 1921.5 -1 tmps.mean 0.283 0.2037 0.3699 439.8 2.1 pres.mean -0.0146 -0.066 0.0374 3297.8 -1.6 tmps.mean.trend -0.2173 -0.2551 -0.1788 2453.9 -2.8 pres.mean.trend -0.0301 -0.0522 -0.0081 4963.5 -0.2 tmps.pv.ar1.trend 0.0435 0.0127 0.0743 2548.5 -1.1 pres.pv.ar1.trend 0.0007 -0.0191 0.0207 5762.7 -0.4 mean.forested 0.0273 -0.0053 0.0596 3835.7 -1 mean.croped 0.0284 0.0011 0.0562 5623.6 -0.2 mean.urbaned 0.0299 -0.0165 0.0759 4528.6 -2.7 trnd.forested -0.0676 -0.0962 -0.0393 4533.3 -1.3 trnd.croped 0.0117 -0.015 0.0384 3886.3 2.4 trnd.urbaned -0.0013 -0.0441 0.0418 4647.8 3 nu2 0.0017 0.0010 0.0031 7614.8 -0.9 tau2 4.0587 3.8213 4.3143 72063 2.3 rho 0.9870 0.9661 0.9970 41442.2 2.8 DIC = -5692.17 p.d = 2079.633 LMPL = 1867.3 256 257 258 8. Additional model to check the potential confounding effect of climate on 259 the relationship between mean NBP and biodiversity (both atmospheric 260 inversions combined) 261 Median 2.50% 97.50% n.effective Geweke.diag (Intercept) 0.215 0.18 0.2503 1595.8 0.7 biodiver 0.2641 0.1905 0.3399 2294.5 -1.3 mean.ndep.tot 0.2307 0.1626 0.3003 1570.2 2 I(biodiver^2) -0.0827 -0.1215 -0.044 3956.1 2 tmps.mean 0.0558 -0.0394 0.1514 1091 -2 pres.mean 0.0936 0.0118 0.175 2150.6 1.5 tmps.pv 0.7465 0.6494 0.84 913.1 1.5 pres.pv -0.0334 -0.0673 0.0006 3368.3 0.5 mean.forested 0.0531 0.0185 0.087 2729.8 0.8 mean.croped 0.05 0.0185 0.0808 4414.8 0.1 mean.urbaned -0.0147 -0.041 0.0119 4086.5 0 biodiver:mean.ndep.tot 0.1085 0.0736 0.1444 2105.5 -0.9 biodiver:tmps.mean 0.0028 -0.0903 0.0978 2635.1 1.3 biodiver:pres.mean 0.0823 0.0387 0.1261 3413.3 -0.3 tmps.mean:pres.mean -0.1749 -0.2623 -0.0875 2662.4 -1.1 biodiver:tmps.pv 0.3363 0.2493 0.426 3002.7 1.9 biodiver:pres.pv 0.0289 -0.0023 0.0601 4288.4 0.7 tmps.pv:pres.pv -0.0424 -0.0788 -0.0057 3295 1.1 nu2 0.0025 0.0013 0.0052 3174.2 0.7 tau2 4.5314 4.1535 4.8771 34656.6 0.4 rho 0.9487 0.8773 0.9873 30361.2 1.3 DIC = -4910.2 p.d = 2022.36 LMPL = 1464.75 262 263