Results and Discussion

Mean CH₄ uptake rates were 1.3 mg CH₄ m⁻² d⁻¹ in the Baltimore rural forest soil (Fig. 1A) and 1.0 mg CH₄ m⁻² d⁻¹ in the Hubbard Brook reference forest soil (Fig. 1B) over the full study period.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Annual CH₄ uptake. (A) CH₄ uptake (±SE, n = 8–16) in urban and rural forests in Baltimore, Maryland, from 1999 to 2016 (annual means calculated for a “water year” from October to September). (B) CH₄ uptake (±SE, n = 12) in calcium (Ca) fertilized and reference forests at Hubbard Brook, New Hampshire, from 2002 to 2015. Differences between forest types are significant in Baltimore (all P < 0.05 except in 2000 and 2013) but not significant (all P > 0.05) at Hubbard Brook for each year. The Insets represent average CH₄ uptake during the study period in urban (Left) versus rural (Right, n = 18, P < 0.001) or Ca-fertilized (Left) versus reference (Right, n = 14, P = 0.87) forests. The boxes show median and 5th and 95th percentiles. All trends with time are statistically significant (P < 0.01) (SI Appendix, Table S4).

In Baltimore, CH₄ uptake significantly decreased by 62% (from 1.2 to 0.46 mg CH₄ m⁻² d⁻¹) and 53% (from 2.0 to 0.96 mg CH₄ m⁻² d⁻¹) in the urban and rural forest soils, respectively, over the 18-y study period (Fig. 1A and SI Appendix, Tables S3 and S4). These declines were significant over all individual sites (SI Appendix, Figs. S3 and S4 and Table S5). Methane uptake varied seasonally with high values in summer and autumn (SI Appendix, Fig. S5) due to higher temperatures and lower soil moisture (SI Appendix, Fig. S6).

At Hubbard Brook, CH₄ uptake significantly decreased by 74% (from 1.7 to 0.44 mg CH₄ m⁻² d⁻¹) and 89% (from 2.0 to 0.22 mg CH₄ m⁻² d⁻¹) in the Ca-fertilized and reference forest soils, respectively, over the 14-y study period (Fig. 1B and SI Appendix, Table S4). If data from 2002 were removed, the decreases were 55% (from 0.98 to 0.44 mg CH₄ m⁻² d⁻¹) and 80% (from 1.1 to 0.22 mg CH₄ m⁻² d⁻¹) in the Ca-fertilized and reference forest soils, respectively, over the 13-y study period.

Global mean CH₄ uptake in forest soils (derived from our literature analysis) over the period from 1988 to 2015 was 1.03 mg CH₄ m⁻² d⁻¹ (Fig. 2A), similar to the global mean uptake level (1.15 mg CH₄ m⁻² d⁻¹) reported in a recent metaanalysis (4). Our analysis also shows a 77% decrease in CH₄ uptake over the past three decades, from 1.4 mg CH₄ m⁻² d⁻¹ in 1988 to 0.31 mg CH₄ m⁻² d⁻¹ in 2015. These declines were only observed in forests located in the 0–30 °N and 30–60 °N latitude bands (Fig. 2B); they were not present in the Arctic or in the Southern Hemisphere (Fig. 2C and SI Appendix, Table S4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Methane uptake in forest soils retrieved from published studies. (A) Global annual CH₄ uptake in forest soils from 1988 to 2015. The Inset represents average CH₄ uptake for the full period from 1988 to 2015. The boxes show median and 5th and 95th percentiles. Error bars denote SEs. (B) Annual CH₄ uptake in forest soils from 0 to 30 °N and 30–60 °N latitude. (C) Annual CH₄ uptake in forest soils from 60 to 90 °N, 0–30 °S and 30–60 °S latitude. All trends with time are statistically significant in A and B (P < 0.05) (SI Appendix, Table S4) but are not significant in C. The data in A were actually measured annual CH₄ uptake and those in B and C include both actually measured and estimated annual CH₄ uptake (SI Appendix, Supplementary Methods). The data collected in Baltimore and Hubbard Brook were excluded from this literature analysis.

Methane uptake has been found to be strongly affected by several factors (8, 24, 25), several of which have varied in surprising ways over the course of our long-term studies (18, 26) (SI Appendix, Fig. S7). The declines in CH₄ uptake that we observed at two LTER sites coincided with increases in atmospheric CH₄ concentration (SI Appendix, Fig. S8A), marked declines in atmospheric N deposition (13) (SI Appendix, Fig. S8B), and increases in air temperature (SI Appendix, Fig. S7A), all of which should have stimulated soil CH₄ uptake (27, 28). Previous research at the Baltimore sites found that strong differences in CH₄ uptake between urban and rural forest soils (15) were driven by higher rates of N cycling, especially nitrification, in the urban forest soils (29). The results here confirm the large differences in N cycling and availability between urban and rural forest soils (SI Appendix, Table S3), but there was no significant long-term temporal change in either the concentration or flux of inorganic N in soil leachate over the 18-y time period (both P > 0.05). At Hubbard Brook, soil N cycling decreased during the study period (30). We therefore cannot attribute the declines in CH₄ uptake in these temperate forest soils to changes in either N deposition or cycling.

There is a long-term trend of increased precipitation in the northeastern United States (26) (SI Appendix, Fig. S7B). In Baltimore there was also a marked increase in leachate volume (the water collected in zero tension lysimeters) (SI Appendix, Fig. S9A) and significant temporal variation in soil moisture (SI Appendix, Table S3) over the 18-y time period. A multivariate analysis of 28 variables found that leachate volume and soil moisture were the dominant factors influencing CH₄ uptake in Baltimore forest soils (SI Appendix, Fig. S10). At Hubbard Brook we have observed significant increases in soil moisture over the past 14 y (SI Appendix, Fig. S9B). The trend in soil moisture at Hubbard Brook was only significant in the Ca-fertilized watershed, but trends in this variable are difficult to see in infrequently sampled plots (one to five times per year). More comprehensive analyses, based on continuous data, show general increases in soil moisture at Hubbard Brook (26). These results suggest that increases in soil water flux underlie the long-term observed decline in CH₄ uptake at both sites (SI Appendix, Fig. S11), either by inhibiting the flow of substrate to consumers or by increasing production in anaerobic microsites (7). The fact that CH₄ uptake decreased in both urban and rural forest soils in Baltimore, and in both Ca-fertilized and reference forest soils at Hubbard Brook with very different temperature (SI Appendix, Fig. S7A) and N availability regimes (22, 29), supports the idea that CH₄ uptake in forest soils is more sensitive to soil moisture than to temperature or N cycling (3).

At a much larger scale, the decline in CH₄ uptake in forest soils in the latitudes of 0–60 °N (Fig. 2B and SI Appendix, Fig. S12) was coincident with increases in precipitation in these areas (SI Appendix, Fig. S13). This result is in accordance with another global analysis (6). In addition, our analysis around the world also found that CH₄ uptake was consistently lower at high soil moisture levels (SI Appendix, Fig. S14), further supporting the idea that changes in precipitation and hydrological flux are the key drivers of regional-scale changes in CH₄ uptake in forest soils.

Declines in CH₄ uptake were not observed in arctic areas or in the Southern Hemisphere (Fig. 2C). One possible reason for this is that there have been few in situ measurements in these areas (SI Appendix, Fig. S12), hindering our ability to assess the long-term trends in CH₄ uptake in these forest soils. Moreover, some studies have found that soil CH₄ uptake in subtropical and temperate forests is lower in the Northern Hemisphere than in the Southern Hemisphere (3, 6). Thus, the effects of changes in hydrological flux (i.e., increase in precipitation) on CH₄ uptake in Southern Hemisphere forest soils may be important and worthy of further study.

Global budgets show that upland soils are a major sink for atmospheric CH₄ (3⇓–5), and some analyses suggest that this sink has increased in recent decades (6). Here we find that soil CH₄ uptake in two temperate forests in the northeastern United States has declined to rates of 0.18–0.22 mg CH₄ m⁻² d⁻¹, and our regional-scale analysis shows a marked decline in annual mean CH₄ uptake over the past three decades to 0.31 mg CH₄ m⁻² d⁻¹ in 2015. These uptake rates are much lower than current estimates for temperate forests (3, 4, 6) (0.65–1.6 mg CH₄ m⁻² d⁻¹). Our results therefore suggest that the current soil CH₄ sink may be overestimated in areas where precipitation and soil hydrological flux are increasing, particularly for the tropical/subtropical and temperate forests located between 0 and 60 °N latitude. Moreover, our results suggest that soil CH₄ uptake is much more sensitive to subtle shifts in soil hydrology than to marked changes in N cycling, increases in atmospheric CH₄ concentrations, and increases in temperature. These results need to be considered in assessments of the global CH₄ budget and of how this budget is likely to change in the future along with climate and atmospheric deposition.