Soil CO2, CH4 and N2O fluxes in open lawns, treed lawns and urban woodlands in Angers, France

Urban green spaces (UGSs) are mostly represented by lawns and wooded areas. These UGSs can store carbon in soil and above-ground biomass, potentially modulated by management intensity and vegetation cover. Trees in lawns can create a local microclimate modifying soil biogeochemical cycles affecting in turn greenhouse gas (GHG) emissions. The objective of this study was to assess the effects of trees on microclimate (temperature and moisture) and soil properties influencing GHG fluxes in contrasted UGS types. We monthly monitored (from March to November 2021) and compared soil CO2, CH4 and N2O fluxes simultaneously with surface temperature and moisture in treed lawns, open lawns and urban woodlands. Lawns included 4 different management intensities including mowing, irrigation and fertilization practices. Temperature was the best predictor of soil respiration in all UGS types studied and was the highest in open lawns. We showed that moisture reflected by the water filled pore space (WFPS) significantly added on variation explanation. The shading of trees decreased soil respiration by 34% in treed lawns while soil properties were similar, indicating a straightforward effect of lowering temperature. On the contrary, in woodland soils the lower rates of soil respiration were attributed to both soil properties and temperature decreasing. Urban woodlands were a sink for CH4 throughout the year (− 0.19 mg m¯² h¯¹). Methane consumption in lawns was small and even a CH4 source in irrigated parks when WFPS overpassed 75%. N2O fluxes were small (0.014 mg m¯² h¯¹) probably reflecting the transition already made from mineral to controlled-release fertilization limiting N availability.


Introduction
Urban areas cover 3% of the Earth's land surface and may contain significant amounts of organic carbon contributing to the mitigation of increasing urban greenhouse gas (GHG) emissions (Churkina et al. 2010;Lal 2012;Canedoli et al. 2020).To counterbalance these emissions, it is important to maintain and even increase C stocks in urban areas, especially in soil and vegetation that contribute the most to C storage (Churkina et al. 2010;Churkina 2012).Urban green spaces (UGS) could be therefore an important lever to this mitigation (Lal 2012) but C fluxes need to be evaluated.
Vegetation cover is distinguished in this study as open and treed lawns (Fig. 1).Open lawns are almost permanently exposed to the sun with inputs of labile organic matter mostly coming from mowed grass and root exudates turnover (Hamido et al. 2016;Law et al. 2017;Thompson and Kao-Kniffin 2019).Treed lawns, formed at least by several trees, can create a microclimate by lowering soil temperatures and moisture, but also by locally modifying biogeochemical cycles and litter decomposition processes (Livesley et al. 2016;Nidzgorski and Hobbie 2016).Tree shading also limits the quality and quantity of light, reduces the biomass production of lawns and thus organic matter retention compared to open lawns (Bell et al. 2000;Miller and Edenfield 2004;Dudeck et al. 1992).Moreover, under deciduous trees, the litterfall in parks is regularly raked in autumn, thus litter does not decompose under trees or is concentrated under few trees for commodity.By controlling these environmental factors, treed lawns can therefore modify the C and N cycles and thus the driving processes of CO 2 , CH 4 and N 2 O emissions from roots and soil microbes.Studying the influence of treed lawns adjacent to open lawns on soil C and N outputs therefore is important to understand the potential contribution of UGSs functioning and management types to GHG emissions in ever-expanding cities.
Herbaceous and tree-dominated UGSs have been well studied in urban areas, through the study of lawns (Livesley et al. 2010;Qian and Follett 2012) and urban woodlands (Yesilonis and Pouyat 2012).Intensively managed lawns (irrigated, fertilized, frequently mowed with mulching of mowing residues) have been identified as systems with high C sequestration potential (Pouyat et al. 2009;Qian and Follett 2012;Wang et al. 2022) with important indirect CO 2 emission from fuel, water and fertilizer uses (Townsend-Small and Czimczik 2010; Braun and Bremer 2019).These lawns potentially present negative impact with an increase in CO 2 and N 2 O emissions from soil especially when mineral fertilization and irrigation is not adjusted to plant needs (Livesley et al. 2010;Hamido et al. 2016a Braun andBremer 2018a, b;Law et al. 2021).Urban woodlands, provide great services such as storing similar C amounts than lawn soils due to high C inputs from litter (Yesilonis and Pouyat 2012).Moreover, soil studies rarely take into account the huge C biomass stock in woody plants leading to missing component to compare ecosystems processes (Sun et al. 2019).Furthermore, soil from urban woodlands compared to intensively managed lawns typically are a low source of CO 2 and N 2 O (Groffman et al. 2009;Decina et al. 2016;van Delden et al. 2018) as well as a CH 4 sink (Groffman and Pouyat 2009;Costa and Groffman 2013;van Delden et al. 2018).These effects are due to lower soil temperature limiting soil respiration, high C:N of litterfall favoring N immobilization and lower bulk density as well as lower moisture content increasing the rate of diffusivity of CH 4 and its consumption into the soil.
The biogeochemical processes in UGSs have rarely been investigated when herbaceous and trees are interacting as a treed lawn.It is necessary to investigate the impact of this UGS type on soil C and N cycles under temperate climate.Indeed, the influence of trees on CO 2 and CH 4 fluxes in lawns has been studied recently by Lu et al. (2021) under boreal climate and they found that CO 2 fluxes were twice lower in deciduous treed lawns compared to open lawns and they attributed this effect to the lower decomposability of tree litter (i.e.mainly fine roots).These authors also found higher CH 4 consumption in treed lawns than in open lawns, but the potential reasons for these differences were not discussed.Literature has not yet assessed the microclimatic effects of lawn trees on the soil respiration process in an urban environment, and how much shading can drive this soil respiration lowering.We suspected that microclimatic effects of trees inducing changes in soil temperature and moisture (Nidzgorski and Hobbie 2016), could explain soil respiration variations (Oertel et al. 2016).In urban (Groffman et al. 2009;Chun et al., 2014) and in non-urban environment (Smith and Johnson 2004) studies showed that the microclimatic effect of tree in a woodland could reduce Fig. 1 The three studied urban green space types (urban woodland, treed lawn and open lawn) and their specific characteristics related to sun exposure, grass biomass production and tree litterfall soil respiration by 17 to 38% compared to a grassland.This reduction was mainly attributed to the shading of the trees decreasing soil temperature.Thus, there is a need to address the question of how the presence of trees in lawns can change the soil surface temperature and moisture influencing soil respiration and C stocks.
Studies on UGS have mostly focused on intensively managed lawn (i.e.irrigated, fertilized and mulched) systems (Frank et al. 2006;Qian and Follett 2012) thus, with strong stimulation of C and N cycling.However, municipalities worldwide are currently changing the way they manage UGSs, by selecting sustainable action plans preventing to waste natural resources (Ignatieva et al. 2020;Pantaloni et al. 2022), such as soil, water and nutrients use.Specific management practices (low or no irrigation and fertilization, controlled-release or organic fertilization, varying mowing frequencies, mulching, etc.), are therefore implemented to obtain the desired landscape and maintain it, in the long term.UGS management can be classified into increasing levels of intensity depending on several factors potentially nested (Table 1): the size of the UGS, plant cover type, UGS functions (e.g.leisure, aesthetic, environmental or ecological), location in the urban area (e.g.city center or periphery) and their landscape impact.Data of the impact of management practices on soil properties leading to GHG emissions are still lacking (Hamido et al. 2016a Braun andBremer 2018b;Thompson and Kao-Kniffin 2019;Law et al. 2021).
The objective of this study was to measure soil GHG emissions in treed lawns and open lawns under a gradient of management intensity, and to relate the effect of trees (soil temperature and moisture) on microclimate and soil properties to soil GHG emissions in a temperate urban context.We mainly hypothesized that (1) shading of tree in lawns limits soil respiration by lowering soil surface temperature (2) changes in soil biochemical properties (i.e.C:N increase) under trees can slow down soil respiration; (3) urban woodlands and treed lawns with lower bulk density and both soil resource and moisture content exhibit higher CH 4 consumption; (4) N 2 O emissions increase with management intensity of lawns because of fertilization and irrigation that increase both nutrient availability and soil water content.
According to the Köppen-Geiger classification (Kottek et al. 2006), Angers' climate is warm and temperate (type Cfb, warm temperate -fully humid -warm summer).Over the period 1981-2010, Angers had an average annual temperature of 11.5 °C.Annual rainfall averaged 693 mm and was well distributed throughout the year.

UGS types: urban woodlands, open and treed lawns
In 2021, the city presented 1 500 ha of UGSs, among which 49% (735 ha) included grasslands mainly composed of lawns and 17% (225 ha) are wooded areas mainly composed of urban woodlands more or less artificial, but also including treed lawns.Lawns were mainly composed of cool season grasses (Lolium perenne, Festuca arundinacea, Festuca rubra) that could include some spontaneous species in the less intensive managed lawns.The selected 27 sites within the 15 UGSs (Table S1), included 3 urban woodlands, 12 treed lawns associated to 12 open lawns.All sites were established for at least 18 years.Within each site, we chose to focus mainly on deciduous treed lawns because they are predominant in Angers.at the 27 sites (a duration of 3 weeks was necessary to sample the 27 sites).The sampling spots were selected to be a representative area of the UGS types (slope, vegetation).In open lawns, we placed the sampling spot at least 18 m from a landscape element (e.g.tree, hedge, wall) whenever possible.As Viaud and Kunnemann (2021) we estimated that this distance minimized the effect of the trees on the soil.In treed lawns and urban woodlands, we placed the sampling spot directly under the canopy of a tree, at a distance of 3 m from the trunk to avoid the effect chemically enriched rain from stemflow on soil properties (Van Stan et al. 2018).
For the measurement of CO 2 , we used 2 automated infrared analyzers CFLUX-1 (PP Systems, Amesbury, USA) with a sampling surface of 0.032 m 2 and volume of 2.3 L. For the measurement of CH 4 and N 2 O, we used a Fourier transform infrared analyzer DX4040 (Gasmet Technologies Oy, Helsinki, FINLAND) associated with an opaque manual dynamic closed chamber system (6.6 L) with a sampling surface of 0.042 m 2 .All gas samples were taken between 09:00 and 13:00 because this time was considered to be representative of daily fluxes (Kaye et al. 2004(Kaye et al. , 2005)).At each sampling spot we placed one chamber for CO 2 measurements and one chamber for CH 4 and N 2 O measurements.We chose not to cut the grass before measuring GHGs to avoid an artificial temperature elevation or an artificial microbial soil respiration decrease (Craine et al. 1999;Bahn et al. 2006).Further experiments that supported that latter decision (Supplementary Material and Methods) have shown that the contribution of dark foliar respiration to the CO 2 fluxes was not significant and that clear-cutting the grass causes an artificial 2 °C increase of soil temperature.CO 2 measurements were made every 15 min.For the measurement of CH 4 and N 2 O, two measurements were made per spot (between 9:00 and 11:00 and between 11:00 and 13:00) to cover the daily temporal increase in GHG fluxes.In order to obtain linear increases of GHG fluxes in the measurement chamber, the time of chamber closure was 5 min for CO 2 measurements and 15 min for CH 4 and N 2 O measurements.
Soil surface temperature and volumetric humidity at 5 cm depth were monitored for each sampling point with a TRIME-PICO 32 TDR probe (IMKO Micromodultechnik GmbH, Ettlingen, GERMANY).Water-filled pore space (WFPS) was calculated using the following equation (Robertson and Groffman 2015): where SW C is volumetric soil water content (vol%), BD is bulk density (g cm − 3 ) and PD is particle density (2.65 g cm − 3 ).

UGS management
The 12 treed and open lawns presented four increasing management intensities regarding of mowing frequency and cutting height, retention or removal of grass clippings, irrigation and controlled-release fertilization (Table 1).More specifically, the fertilizer was organic polymer-coated (12-5-22, 90-d release, Kabel Cote Special K, Kabelis) and applied in April, June and October.The coated nitrogen form was 2.5% nitric, 5.5% ammoniacal and 4% ureic.In treed lawns litterfalls are raked every 15 days during the defoliation period (from October to January).Urban woodlands were not managed.

Soil properties
According to the French soil classification (Baize and Girard 2009), soils of the 15 UGS were classified into four soil types (Table S1).These soils had mainly a sandy loamy and loamy texture.We analyzed soil biophysicochemical properties (total organic C, total N, total P, microbial biomass, pH, bulk density, soil texture, CEC, EC and exchangeable elements).Soils were sampled between 0 and 10 cm depth with an auger (Ah horizon) after GHG measurements (February 2022).Soil sampling for C stocks calculation were performed (April 2021) between 0 and 30 cm depth, followed by a visual inspection to determine horizon depths.Soil samples were divided based on their horizons, air dried and sorted to remove coarse roots and rocks (> 2 mm).In addition, soils cores were collected between 0 and 6 cm using a steel cylinder (diameter: 5.5 cm, height: 6 cm), oven dried (105 °C) and weighted for bulk density calculation 272).Bulk density (g cm − 3 ) was determined from the oven dried mass of soil cores and core volume.The following equation was used to calculate the soil organic C stocks (SOC kg m − 2 ; Poeplau et al. 2017): where SOC is the soil organic C concentration (g C kg − 1 ), BD is bulk density (g cm − 3 ) and D (m) the sampling depth.
We verified that the level of urbanization, defined as difference in temperature between the rural and urban environment (Δ T urban -T rural ) (Peng et al. 2012;Heisler and Brazel 2015), was not related to soil surface temperature or soil N content (Table S2).

In situ GHG fluxes, soil temperature and moisture measurements
The soil CO 2 , CH 4 and N 2 O flux were monthly measured in 2021 (i.e.March, April, May, June, July and November) The sensitivity of CO 2 fluxes to temperature variation can be described by the Q 10 factor (Oertel et al. 2016).The following equation was used to calculate Q 10 values in the three studied UGS types: where R T +10 is the soil respiration at the initial soil temperature (R T , T = 15 • C ) plus 10 °C (Smith et al. 2003;Brisson and Launay 2008).We calculated these values using the linear equations from the regression of soil respiration with soil surface temperature.

Results
CO 2 , CH 4 and N 2 O fluxes according to the type of UGS or management intensity CO 2 fluxes were significantly affected by the UGS type depending on the time (Repeated measure ANOVA, F = 2.22, P < 0.05, Table 2).Soils in open lawns had systematically higher fluxes than soil from treed lawns and urban woodland soils, the difference increasing with time from March to July (Fig. 2a).CO 2 fluxes in treed lawns did not differ from urban woodlands except in March.On average, the CO 2 fluxes in open lawns (1 091 ± 61 mg CO 2 m¯² h¯¹) were 34% higher than CO 2 fluxes from treed lawns (715 ± 46 mg CO 2 m¯² h¯¹) and 52% higher than CO 2 fluxes from urban woodlands (521 ± 59 mg CO 2 m¯² h¯¹) (Rm-ANOVA main effect, F = 18.8, P < 0.001, Table 2).
The CH 4 fluxes were significantly higher in urban woodlands than in open lawns and in treed lawns (Fig. 2b) during the 3 consecutive months recorded (June, July and November).On average, the CH 4 fluxes in urban woodlands (− 0.19 ± 0.04 mg m¯² h¯¹) were higher than treed and open lawns (− 0.019 ± 0.02 mg m¯² h¯¹ and − 0.033 ± 0.01 mg m¯² h¯¹, respectively).
No significant differences of N 2 O fluxes were found between the 3 UGS types (Fig. 2c).On average, N 2 O fluxes in this study reached 0.014 ± 0.002 mg m¯² h¯¹.Nevertheless, the variability of N 2 O fluxes was high with rates as high as 0.087 mg m¯² h¯¹, especially in June.
Management intensity had no significant effect on CO 2 and CH 4 fluxes (Fig. S1).In July, N 2 O fluxes in management intensity 2 (0.006 ± 0.001 mg m¯² h¯¹) were significantly lower to those of management intensity 3 and 4 (0.04 ± 0.003 mg m¯² h¯¹ and 0.024 ± 0.003 mg m¯² h¯¹ respectively) (Fig. 3).In June, several high fluxes were found especially in management intensity 1 reaching 0.087 mg m¯² h¯¹.

Data processing and statistical analysis
GHG fluxes were calculated from linear increase (CO 2 and N 2 O) or decrease (CH 4 ) in gas concentration per unit of time (Butterbach-Bahl et al. 2016), corrected for chamber volume, sampled surface area, air temperature and atmospheric pressure (Barton et al. 2007).The atmospheric pressure could not be assessed in the manual chamber so we did not integrate it in the CH 4 and N 2 O flux calculation.GHG fluxes, soil surface temperatures and WFPS collected between 09:00 and 13:00 were averaged to obtain the mean daily values.These daily values were averaged for each UGS type or management intensity and were used as the average values of the month.
One-way-repeated measures analysis of variance (rm-ANOVA) was used to determine the effects of UGS type and management intensity across time on GHG fluxes, soil surface temperature and WFPS.When a significant interaction was found, we separately analyzed the effects of UGS type (urban woodland, treed lawn, open lawn) or management intensity (intensity 1, 2, 3, 4) for each sampling time by one-way ANOVA followed by the Tukey post-hoc tests (at P < 0.05) to analyze in detail the variations between each UGS type or management intensity and for each sampling time.One-way ANOVA followed by Tukey post-hoc tests (at P < 0.05) was also used to determine the effects of UGS type or management intensity on soil properties.Microbial biomass was 'n + 3' log normal transformed prior to statistical analyses to meet the assumption of normality and homogeneity of variances.For CH 4 fluxes, N 2 O fluxes, organic C, total N and total P contents data normality could not be obtained.Thus, a Kruskal-Wallis test followed by Mann-Whitney post-hoc tests were carried out to test the effects of UGS type and management intensity on the non-normal datasets.To test the effect of grass clipping retention (only within management intensity 2) on GHG fluxes, total N and microbial biomass contents we used a homoscedastic student t-test if normality of datasets could be obtained.For non-normal datasets (N 2 O fluxes of June, organic C content) we used a Wilcoxon rank-sum test.
Simple linear and polynomial regression models were used to analyze the effects of temperature or WFPS on GHG fluxes.To model the soil respiration with the specific effect of temperature and with the combined effects of temperature and WFPS, we compared 9 models referenced by Weissert et al. (2016) (Table S4) based on statistical criteria such as R 2 and RMSE.
We performed Spearman correlation to test the links between GHG fluxes and soil properties as well as between GHG fluxes, soil surface temperature and urbanization level.All statistical analyses were conducted on R software, version 4.1.3.

Relationships between temperature, WFPS and soil properties with GHG fluxes
CO 2 fluxes were significantly correlated to soil temperature with 54 to 58% of the variation explained in the 3 UGS types, urban woodlands with the lowest and open lawns with the highest slopes (Fig. 6a, b, c).The WFPS did not correlate with CO 2 fluxes but a site effect was found showing either positive or negative correlations in the three UGS types (data not shown).CH 4 fluxes showed no clear link with soil surface temperature in urban woodlands and treed lawns but a slight negative correlation was found in open lawns (R 2 = 0.14, P < 0.05).CH 4 fluxes were positively correlated to the WFPS (Fig. 7a, b, c) with the highest variation explained in urban woodlands (R 2 = 0.51, P < 0.001), intermediate in treed lawns (R 2 = 0.45, P < 0.

01) and weak in open lawns
The water filled pore space (WFPS) significantly changed with time (Rm-ANOVA main effect, F = 8.23, P < 0.001, Table 2) but no influence of UGS type was found.The WFPS was significantly affected by the management intensity depending on the time (Rm-ANOVA, F = 2.35,

Soil properties and GHG fluxes depending on UGS types and management intensities
Soil properties were significantly different depending on UGS types and management intensities: total P, pH and bulk density were significantly lower in urban woodlands than in treed and open lawns (Table 3).Total organic C, total N, total P were significantly lower and bulk density was significantly higher in management intensity 2 than management 4 (Table 3).

GHG emission overview in urban green spaces (UGS)
In the current study, soil CO 2 fluxes recorded in open lawns were in the upper range (Fig. 2a) of the those recorded in various contexts (from 118 to 1 649 mg CO 2 m¯² h¯¹; e.g.2021) under boreal climate (from 35 to 80 mg CO 2 m¯² h¯¹) and were largely lower than those observed in the current study under a warm temperate climate.In the current study, roots may have significantly contributed to soil respiration (Hanson et al. 2000a, b;Ryan et Law 2005), especially during growing season where its contribution could reach 50% of soil respiration in a fertilized temperate grassland (Byrne and Kiely 2006) or 76% in a tropical low management intensity lawn (Ng et al. 2015).In addition, the contribution of dark foliar respiration accounted for 10% of (R 2 = 0.21, P < 0.05), all revealing different UGS type behaviors but all regressions increasing.
N 2 O fluxes and soil surface temperature or WFPS showed no general correlation for treed and open lawns and a significant and positive correlation with temperature was found in urban woodlands r²=0.14, P < 0.05, polynomial regression).Individual correlations by sites showed very contrasted behaviors with positive or negative correlations from non-significant to very significant correlations (r 2 = 0.01 to 0.96) in both treed and open lawns.During November, data under 10 °C in open lawns showed a significant and negative correlation between temperature and N 2 O fluxes (R 2 = 0.40, P < 0.05).During June and July, data over 10 °C showed after removing 4 extreme values of N 2 O (above 0.040 mg N 2 O m 2 h − 1 , see discussion), a general increase in N 2 O fluxes positively correlated with WFPS (R 2 = 0.35, P < 0.01).
By separately analyzing irrigated and non-irrigated lawns (Table S4), the best modelling of CO 2 fluxes in irrigated lawns were also obtained with the power-logistic equation (R 2 = 0.65, RMSE = 200 mg CO 2 m − 2 h − 1 for treed lawns and R 2 = 0.78, RMSE = 249 mg CO 2 m − 2 h − 1 for open lawns).CO 2 fluxes of non-irrigated lawns were better modelled with the power-logistic equation (R 2 = 0.55,

Shading reduced temperature and soil respiration in treed lawns and urban woodlands
In treed lawns and urban woodlands, the canopy of woody species (deciduous trees) limited soil warming through shading of the soil surface and consequently reduced temperature and soil respiration (Wan and Luo 2003;Smith and Johnson 2004).This effect has been found lasting but least intensive during the non-growing season (Smith and Johnson 2004) with evergreen trees (Lu et al. 2021).The correlations found in the current study, between soil respiration and soil surface temperature (Fig. 6a, b, c) with and without shading, confirmed that soil temperature is a dominant driver explaining a larger part of the CO 2 flux variations in urban lawns and urban woodlands (Chen et al. 2013;Shchepeleva et al. 2019).
More specifically, the presence of trees in lawns reduced soil respiration by 34% (Fig. 2a) paralleling with a decrease in temperature of 2 °C for the same period.The close Q 10 values (Fig. 6a, b) of treed (1.88) and open lawns (1.90) indicated similar temperature dependence and confirm our hypothesis that tree shading in lawns reduces soil respiration.Furthermore, moisture was found to be particularly improving soil respiration variation in lawns (Table S4) and even better, by distinguishing irrigated and non-irrigated lawns, revealing a significant contribution of WFPS the total CO 2 fluxes in this study, possibly leading to a small overestimation of CO 2 fluxes in lawns, but this was not significant (Supplementary Material and Methods).
CH 4 consumption (i.e.negative fluxes) in open and treed lawns (Fig. 2b) was in the upper range of fluxes in literature (from 0.000 to -0.027 mg CH 4 m¯² h¯¹; i.e.Kaye et al. 2004;Groffman and Pouyat 2009;van Delden et al. 2018;Shchepeleva et al. 2019).Urban woodlands were an important CH 4 sink as its consumption was 3 times higher than the average consumption rate attributed to temperate forest systems (-0.06 mg m¯² h¯¹ according to a review of Dalal and Allen 2008).Compared to other urban woodlands, CH 4 consumption was also largely above the literature (from − 0.010 mg CH 4 to -0.158 mg CH 4 m¯² h¯¹; i.e.Goldman et al. 1995, Groffman and Pouyat 2009, Zhang et al. 2014, Lu et al. 2021).
With an average N 2 O flux of 0.014 mg m¯² h¯¹ in the current study, these rates were small and well below the values reported for open lawns in other studies (from 0.031 to 0.276 mg N 2 O m¯² h¯¹; e.g.Kaye et al. 2004;Groffman et al. 2009;Livesley et al. 2010;Gillette et al. 2016;Braun andBremer 2018a, 2019;Law et al. 2021) and for urban woodlands (0.044 mg N 2 O m¯² h¯¹) (Groffman et al. 2009).One reason of the small N 2 O fluxes may have been the low measurement frequency that is discussed latter (see management intensity section below).C m − 2 ) showed high variability between sites, which may be explained by an insufficient number of sampled sites and strong heterogeneity in UGS history management.The soil C stock of treed lawns 0-30 cm (10.6 ± 1.6 kg C m − 2 ) also showed higher variability than those of open lawns (9.2 ± 0.8 kg C m − 2 ).The tree species and age may have contributed to the high variability of C stocks in treed lawns (Scharenbroch 2012;Lu et al. 2021).The similar C stocks observed between the three UGS types are in line with Canedoli et al. (2020) who found similar C stocks between lawns and urban woodlands (6 kg C m − 2 for 0-40 cm).To explain these results, the authors suggested that factors such as the higher initial C content in lawn soils and management history could have a predominant influence on C stocks distribution prior to vegetation types (Bae and Ryu 2015;Canedoli et al. 2020).Raciti et al. (2011) and Weissert et al. ( 2016) also found lower C stocks than expected in urban woodlands (5 and 3 kg C m − 2 respectively) compared to lawns (7 and 5 kg C m − 2 respectively).Indeed, these authors also suggested that the high C and N inputs in intensively managed lawns (Qian et Follett 2012;Wang et al. 2022), and the ability of herbaceous systems to store C deep in the soil (Campbell et al. 2014) could explain this difference.

Lawns are weak CH 4 sinks compared to urban woodlands
In our study, positive correlation between WFPS and CH 4 fluxes in the 3 UGS types (Fig. 7a, b, c) suggest that CH 4 consumption was favored by the low WFPS values (i.e.strong aerobic conditions).The decrease in WFPS is related to an increase in gas diffusion (CH 4 used as C energy source and O 2 as electron acceptor) and is necessary to methanotroph bacteria oxidizing CH 4 into CO 2 (Costa and Groffman 2013;Serrano-Silva et al. 2014;van Delden et al. 2018).Indeed, at the drier conditions of the study (i.e.20% of WFPS) woodland soils consumed up to 6 times more CH 4 than lawns (Fig. 7a, b, c).Other factors such as nature of the herbaceous vegetation whose litter is rich in N (Li et al. 2013;Hamido et al. 2016b;Nataningtyas et al. 2017), fertilization and irrigation (van Delden et al. 2018;Law et al. 2021) could have limited CH 4 consumption in lawns.
The similar CH 4 consumption in treed and open lawns showed that tree presence affecting microclimate (i.e.temperature and moisture effects) but not soil properties had little influence on the CH 4 consumption process.One explanation could come from the regularly raked litterfall under deciduous trees would slow down soil enrichment and thus would limit the methanotrophic activity (Le Mer and Roger 2001).Lu et al. (2021) showed that coniferous treed lawns, but not deciduous treed lawns, had higher soil organic into modelling.It is estimated that the maximum of CO 2 production in soil by heterotrophic respiration is reached when macro pores are filled with air and micro pores are filled with water, i.e.WFPS of the soil is 50-60% (Linn and Doran 1984;Luo and Zhou 2006).Xu et al. (2004) observed that ecosystem respiration is slowed down at a WFPS below 30% whereas above a WFPS of 80%, soil O 2 becomes limiting for soil biological activity.Other studies (e.g.Smith and Johnson 2004;Shchepeleva et al. 2019) concluded that soil moisture was not a determining factor to explain variations of soil respiration in UGS and did not show significant differences.We demonstrated in the current study that separating irrigated and non-irrigated lawns led to better predict soil respiration between April and July.Whether we cannot exclude a potential role of soil properties on soil respiration, the lack of differences between open and treed lawns (Table 3) and no correlations found (data not shown) could confirm the almost exclusive driver force of soil respiration by temperature and moisture in urban lawns (Oertel et al. 2016).
Trees in urban woodlands intercept the sunlight and resulted into a temperature decrease of 3.2 °C compared to open lawns paralleling with a 52% decrease of soil respiration (Fig. 2a).The lower Q 10 in urban woodlands (1.76) than treed (1.88) and open lawns (1.90) (Fig. 6a, b, c) this time clearly showed a temperature sensitivity influenced by factors, such as soil properties such as organic matter content and quality (Davidson et al. 2006;Conant et al. 2008).This was confirmed in the current study, especially with the lower bulk density, pH, CaO and phosphorus content (Table 3 and Table S3).The differences of soil respiration between urban woodlands and lawns, probably resulted from combined effect of shading and soil biochemical differences.Indeed, the literature has also shown that the recalcitrant woodland litter can slow down the C cycle and thus could limit the observed CO 2 fluxes (Livesley et al. 2016;Lu et al. 2021) while intensively management practices of lawns can stimulate soil CO 2 fluxes in lawns compared to urban woodlands (Decina et al. 2016;Groffman et al. 2009).The lower soil respiration in urban woodlands could also be attributed to the dense surface litter layer inducing an additional shading effect (Chun et al. 2014;Groffman et al. 2009;Sayer 2005).

The presence of trees did not change soil C stocks across UGS types
We expected to find significantly higher C stocks in urban woodlands than in lawns due to the high retention of recalcitrant aerial litter to the soil and the low net losses of C as CO 2 (Yesilonis and Pouyat 2012;Livesley et al. 2016).However, C stocks (0-30 cm) did not show significant differences.The soil C stock of urban woodlands (9.5 ± 2.6 kg high N 2 O fluxes confirming that irrigation was the primary factor promoting N 2 O emissions (Livesley et al. 2010).Moreover, high P content was related to high N 2 O emissions by stimulating mineralization and nitrification (Mori et al. 2010) and would need to be followed by anoxic conditions for denitrification potentially favored by heterotrophic respiration (Mori et al. 2013).The lower soil resources in management intensity 3 along with a weak WFPS while still presenting strong N 2 O fluxes (Fig. 3), could not be explained in this study.The preferential emission of urine by domestic animals at the sampling point could also have been a significant contributor to occasionally high N 2 O fluxes by influencing the available N in soil (Allen et al. 2020).Literature also showed that mulching of grass could also explain higher N 2 O fluxes in lawns, because grass clippings are made of N-rich organic matter that could be easily mineralized and integrated into denitrification process (Li et al. 2013;Nataningtyas et al. 2017).In our study, the retention of grass clippings in lawns (management intensity 2) had no effect on N 2 O fluxes (data not shown) as well as on total C, N and microbial biomass in soil.These results are in line with Bremer (2006) and Law et al. (2021) who found that retention and removal of grass clippings had negligible effects on N 2 O fluxes.
Woodlands are not N 2 O sources as they present aerobic conditions, low pH and N availability necessary for N 2 O and N 2 produced during denitrification (Van Den Heuvel et al. 2011).Moreover, the strong C content (7.8%) in urban woodland soils could play as a C-based sink for inorganic N, limiting N mineral availability and thus denitrification process (Qian and Follett 2012).

Conclusions
The shading of trees strongly decreased soil respiration in treed lawns while soil properties were similar to the open lawns indicating a straightforward effect of temperature.In urban woodland soils, both temperature and soil properties resulted in the lowest rates of soil respiration, making these systems the most conservative for C emissions, whereas above-ground biomass is an additional C sequestration pool.Woodland soils were strong sink for CH 4 throughout the year.On the contrary lawns, both open and treed were weak CH 4 sinks or even temporal sources of CH 4 in irrigated parks where WFPS overpassed 75%.The rather small N 2 O emissions in this study probably reflected the low GHG measurement frequency.The transition already made from mineral to controlled release fertilization limiting N availability (microbial immobilization) probably have also contributed to the low N 2 O fluxes whereas soil water logged was almost never met even in UGS with irrigation.matter than open lawns and exhibited higher CH 4 consumption and could confirm this hypothesis.
We also found that some irrigated treed lawns presenting a WFPS exceeding 75% were a source of CH 4 (Fig. 7b), whereas open lawns especially in dry conditions (20% WFPS), could reach similar levels CH 4 consumption of woodlands (Fig. 7a, c).

Management intensity induced variation in N 2 O fluxes among lawns
We observed no effect of management intensity except in July, showing a cut off separating intensity 1 and 2 from intensity 3 and 4 with higher N 2 O fluxes (Fig. 3).Despite fertilization events (April, June and October) in intensity 4, no burst of increase was detected.The monthly GHG measurement frequency did not account for daily variability of N 2 O fluxes which can be important in UGSs with high management intensity.Indeed, the N 2 O flux peaks occurring directly after fertilization or irrigation events (Kaye et al. 2004;Livesley et al. 2010) have probably not been captured leading to an underestimation of N 2 O fluxes.Therefore, at least a weekly sampling frequency should be carried out for more robust monitoring of N 2 O fluxes (Barton et al. 2015;Braun and Bremer 2018a, b).Furthermore, a specific site management history could have some importance as we found individual and atypical behaviors strongly related to N 2 O fluxes modulated by moisture first (above 75% WFPS) and then by temperature with different behaviors whether temperature was below 10 °C (i.e.negative correlation) and or above 20 °C (i.e.positive correlation).
Due to the unbalanced number of UGS in each management intensity (Table 1), we suggest that interpreting the difference between intensity 2 and 4 (n = 11 and n = 8, respectively) was the most meaningful in this study.N mineral fertilization associated with irrigation has been identified as practices promoting N 2 O fluxes in urban lawns by denitrification (Braun and Bremer 2018a, b;Kaye et al. 2004;Livesley et al. 2010) indicating the role of N availability.We did not follow N mineral in the current study, but lawns intensively managed (i.e.intensity 4: fertilized and irrigated) presented greater N 2 O fluxes paralleling with soil C, N and P content than intensity 2. However, N 2 O in this study were rather low compared to other studies.Soil organic C has been reported to create a C-based sink for inorganic N and thus, limiting N mineral availability and then denitrification process (Qian and Follett 2012).The controlled release fertilizer used in urban parks could have reduced the immediate N availability and thus have alleviated N 2 O fluxes (Braun and Bremer 2018a, b).Nevertheless, the combination of high N content and WFPS above 75% in June and July in some irrigated parks resulted into Further research is necessary to understand the soil contribution to GHG fluxes.We need to better take into consideration the C sequestration of trees as well as above and below ground litter and its turn-over and this should be done (i) regarding the N cycle especially kinetic N availability along with GHG emissions; (ii) by improving of the comprehensive contribution of living roots and exudate production to soil respiration; and (iii) by considering the role of soil microbial community and functions (catabolic and enzymatic) and should be measured under exotic plant species frequently met in UGSs.

Fig. 3 Fig. 2
Fig. 3 Monthly N 2 O fluxes (mg m¯² h¯¹) for lawns distributed in 4 management intensities (Table1).In each box-plot the central bar of the graph is the median.The cross is the mean value and upper and lower edges are the quartiles.The ends of the whiskers are calculated using 1.5 times the interquartile range.Letters indicate significant differences between treatments (Mann-Whitney post-hoc tests); n.s., not significant; *P < 0.05

Fig. 6 Fig. 5 Fig. 4
Fig. 6 Linear regression of CO 2 fluxes (mg m¯² h¯¹) with soil surface temperature (°C) in urban woodlands (a), treed lawns (b) and open lawns (c).R² and Q 10 refer to the coefficient of determination and sensitivity of CO 2 fluxes to temperature variation.***P < 0.001 Fig. 5 Monthly water-filled pore space (WFPS; %) for lawns distributed in 4 management intensities (modalities details are in Table1).In each box-plot the central bar of the graph is the median.The cross is the mean value and upper and lower edges are the quartiles.The ends Livesley et al. 2010;Ng et al. 2015;Hamido et al. 2016a;Shchepeleva et al. 2019;Law et al. 2021) and higher than those recorded in urban woodlands byGroffman et al. (2009) andChen et al. (2013) (366 and 199 mg CO 2 m¯² h¯¹, respectively).CO 2 fluxes in treed lawns were only studied byLu et al. (

Table 2
Results of one-way-repeated measures ANOVA testing urban green space (UGS) type or management intensity (Management) and time on CO 2 fluxes, soil surface temperature and water-filled pore space (WFPS).F and P values are provided for each property for the significance of

Table 3
Soil properties shown according to either urban green space (UGS) types or management intensities.Values are means and standard errors in parentheses.Bold and different letters indicate significant differences between modalities