Shrubs Matter: An Evaluation of the Capacity of Nine Shrub Species to Dissipate Latent Heat and to Remove CO₂ and Airborne PM

Abstract

The aim of this research was to quantify the capacity of different shrub species to remove atmospheric CO₂, to adsorb particulate matter and to dissipate latent heat through transpiration. A total of 308 established plants comprising Deutzia scabra, Elaeagnus × ebbingei, Euonymus japonicus, Forsythia × intermedia, Laurus nobilis, Ligustrum vulgare, Pittosporum tobira, Prunus laurocerasus and Viburnum tinus were selected in Lugano (Switzerland) and Bolzano (Italy). Stem diameter, crown radius, Leaf Area Index, net CO₂ assimilation per unit leaf area (A_leaf), transpiration, and stomatal conductance (g_s) were measured during spring, summer, and fall. The net CO₂ assimilation per unit of crown projection area and per plant were calculated by upscaling A_leaf using a multilayer model. Latent heat dissipation was calculated using the Penman–Monteith equation. The amount of PM trapped on leaves was measured using a gravimetric method. Differences in leaf area and leaf gas exchange among species affected their capacity to deliver specific ecosystem services. Forsythia, Pittosporum, Elaeagnus and Deutzia removed about 40% more CO₂ per unit crown projection area than Laurus, Ligustrum, and Euonymus. Latent heat dissipation by shrubs was, on average, 130 W m⁻², which is comparable to that of tree species. PM removal per unit leaf area was higher in species with sparse canopies and rough leaf surfaces.

1. Introduction

Urban vegetation can contribute to the well-being of city dwellers through the delivery of provisioning, environmental regulation and support, and cultural ecosystem services [1]. A growing body of research focuses on methods and parameters to quantify the delivery of ecosystem services (ES) by urban vegetation species [2,3,4,5]; however, most studies deal with tree species. While data about shrubs are rarer [6], the value of shrubs should not be underestimated. Unlike trees, they can be planted in very small spaces, along road edges where they do not pose any risk of interference with vehicular traffic, nor risk causing negative impact on air quality, which happens for large trees in urban street canyons [7]. Shrubs are known to effectively contribute to the removal of atmospheric particulate matter. According to the Lancet report for 2025, deaths attributable to total ambient PM_2.5 air pollution increased from 7.5 million deaths in 2010 to 8.5 million deaths in 2022 (7.6% increase); while deaths attributable to ambient PM_2.5 air pollution from human activities grew from 5.9 million in 2010 to 6.5 million in 2022 (11.3% increase) [8].

The effectiveness of shrubs to reduce atmospheric CO₂, estimated as carbon storage (C-storage) (i.e., the amount of atmospheric carbon trapped as organic carbon in the woody biomass of the tree) or C-sequestration (i.e., the change in carbon storage over 1 year) [9], is generally reported as low, because of their smaller size and biomass at maturity compared to trees [10]. Nonetheless, the contribution of vegetation to CO₂ removal may be better represented by net photosynthetic carbon gain (i.e., the amount of atmospheric CO₂ annually removed by vegetation through photosynthesis, net of respiratory losses) [4]. Shrubs can develop high Leaf Area Index (LAI) and can have high photosynthetic rate per unit leaf area, thus their role in CO₂ removal may be higher than previously thought [6]. Direct measurements of CO₂ assimilation have seldom been conducted in situ at urban sites, and in most cases, they were performed only on full-sun leaves [11]. Conversely, the parametrization of shaded leaves is crucial to upscale photosynthesis from leaf-scale to canopy-scale using sun–shade models, which were shown to be more accurate than big-leaf models [12].

Similarly, most of the research that evaluated the capacity of urban vegetation to ameliorate microclimate mostly focused on trees [13,14,15]. Due to larger canopy size and total leaf area, trees are expected to shade (i.e., reduce the amount of short-wave radiation reaching surfaces and buildings, and consequently, heat accumulation) and to dissipate latent heat through transpiration more efficiently than shrubs [16]. Nonetheless, the shrub layer can positively contribute to urban canopy cover, which is a major determinant of cooling benefits [17], and the use of shrub species that sustain a high stomatal conductance during hot periods may help the city to become more climate-proof.

Finally, the effect of shrubs on air particulate matter (PM) reduction, especially when forming ‘vegetation barriers’ or ‘green belts’ along roads, was studied in situ in a limited number of species [18,19,20]. Vegetation can passively ameliorate air quality through two main mechanisms: deposition and dispersion. Pollutant particles can be deposited on plant surfaces (i.e., bark and leaves) by rain and snow or by gravity, inertia and Brownian motion (deposition mechanism). At the same time, plants can alter air currents and turbulence simply through their physical presence, influencing the local concentration of pollutants (dispersion mechanism) [21]. The value of shrubs compared to trees in this field is particularly relevant. Shrub-specific traits, such as the canopy structure (often multi-stemmed and branched from the ground) and the limited height, allow for the creation of thick and dense green barriers very close to vehicular emission sources of air PM, efficiently reducing pedestrian exposure to air pollution in open road conditions [7]. Even in urban street canyons, shrubs are more effective than large trees. The latter of the two, even when able to trap many particles, still does not prevent the flow of air pollutants at the pedestrian level because their crowns are generally above the average human breathing height of 1.5 m [7]. For all these reasons, more attention should be paid to the evaluation of the capacity of shrub species to ameliorate air quality. Despite the relevance of these processes, relatively few studies have specifically quantified the ecosystem services provided by shrubs in urban environments, and the available literature on shrub-derived benefits remains limited.

This research used two cities (Lugano, Switzerland; Bolzano, Italy) as living laboratories to test in situ the hypothesis that the capacity to assimilate CO₂, dissipate latent heat through transpiration, and remove atmospheric PM differs among shrub species due to differences in physiological traits and leaf area. Additionally, we tested the idea that ES provisioning may vary across different seasons depending on the meteorological variables of the planting site. These findings will provide useful support for decision makers during the selection of shrub species for urban planting in order to maximize plant and human wellbeing.

2. Materials and Methods

2.1. Experimental Sites and Plant Material

The research was carried out in Lugano (Switzerland) and Bolzano (Italy). Both cities have a temperate, mild and humid climate (Cfb, according to the Koppen–Geiger classification). The average air temperature over the last 30 years is 14.0 °C in Lugano and 13.1 °C in Bolzano. January is the coldest month in both cities (average temperatures are 3.8 °C and 1.8 °C in Lugano and Bolzano, respectively), and July is the warmest month, with an average temperature of 22.6 °C in Lugano and 23.7 °C in Bolzano. Rainfall is evenly distributed throughout the year in both cities, and the average annual rainfall is substantially higher in Lugano (1658 mm per year) than in Bolzano (1111 mm per year). Over the experimental period (January 2020 to December 2020 for Lugano and January 2021 to December 2021 for Bolzano), daily weather data were obtained from the local weather stations of the two cities (MeteoSwiss for Lugano, Open data Alto Adige for Bolzano), while daily air pollutant data were obtained from local stations (OASI—Osservatorio Ambientale della Svizzera Italiana for Lugano, Agenzia provinciale per l’ambiente e la tutela del clima for Bolzano) (

Table 1. Meteorological and air quality data from the experimental sites of Lugano (a) and Bolzano (b) across a 16–17-day period (pre-sampling period of 14 days plus 2–3 sampling days).

City	Sampling Period	Season	Precipitation, Cumulative Values (mm)	Wind Velocity (m s−1)	PM2.5 Concentration (µg m−3)	PM10 Concentration (µg m−3)
(a) Lugano	16 to 18 June 2020	spring	294.0	1.47	4.21	8.25
	27–28 July 2020	summer	25.5	1.54	7.69	13.68
	16–17 September 2020	fall	16.8	1.23	11.47	18.67
	12 to 14 January 2021	winter	50.3	1.28	14.55	19.94
(b) Bolzano	26–27 May 2021	spring	38.2	1.58	4.26	7.62
	20–21 July 2021	summer	64.6	1.56	7.52	12.51
	30 September to 1 October 2021	fall	47.3	1.07	8.83	13.91
	26–27 January 2021	winter	32.7	1.19	15.51	23.94

).

During the experimental years (2020 and 2021), the annual average air PM₁₀ and PM_2.5 concentrations obtained from the local stations were 18.5 and 12.3 µg m⁻³, respectively, in Lugano, and 17.2 and 11 µg m⁻³ in Bolzano. In both cases, the values were above the WHO guideline recommendations (annual averages of PM₁₀ < 15 µg m⁻³ and PM_2.5 < 5 µg m⁻³; [22]), and PM_2.5 also exceeded the new thresholds recently proposed for Europe (annual averages of PM₁₀ < 20 µg m⁻³ and PM_2.5 < 10 µg m⁻³) [22]. Weather and average daily air PM₁₀ and PM_2.5 concentrations are reported for the different measurement days in Table 1 and

Table 2. Average temperature and cumulative precipitation in the experimental site of Lugano (a) and Bolzano (b) compared to the average values of the last thirty years: spring (01/05 to 21/06), summer (22/06 to 31/08), fall (01/09 to 15/10) and winter (01/01 to 29/02).

City	Year	Season	Precipitation, Cumulative Values (mm)	Precipitation, Avg Last Thirty Years (mm)	Temperature (°C)	Temperature, Avg Last Thirty Years (°C)
(a) Lugano	2020	spring	510	164	19.7	19.6
		summer	50	153	23.1	22.1
		fall	130	185	18.9	17.5
		winter	175	127	5.2	4.4
(b) Bolzano	2021	spring	98	70	17.3	18.0
		summer	133	87	24.4	24.7
		fall	47	70	21.1	18.5
		winter	180	52	2.3	1.5

. During the sampling days, air PM₁₀ concentration never exceeded the WHO recommendations (PM₁₀ < 45 µg m⁻³; 24-h average) while air PM_2.5 concentrations slightly exceeded the WHO recommendations (PM_2.5 < 15 µg m⁻³; 24-h average) in Bolzano during winter.

Six shrub species per city were selected based on their relevance in the existing plant inventories and their even distribution in the cities. The list of species, their abbreviations, the number of plants on which growth traits were assessed, and their minimum and maximum stem diameters and crown projection area are reported in

Table 3. Species selected in the experimental cities of Lugano (a) and Bolzano (b), their abbreviations, and the number of plants on which biometrics parameters were measured. D_30min (cm), D_30max (cm) and CPA_min (m²), CPA_max (m²) indicate the observed minimum and maximum stem diameters and crown projection areas, respectively. H_avg (m) indicates the average plant height recorded. Standard deviation is reported for H_avg. E, D, SD in brackets in the species column indicate their leaf persistence: evergreen, deciduous and semi-deciduous, respectively. n is the number of analyzed shrubs per species.

City	Species	Abbreviation	n.	D30min (cm)	D30max (cm)	CPAmin (m2)	CPAmax (m2)	Havg (m)
(a) Lugano	Elaeagnus × ebbingei (E)	Ee	23	4.47	18.60	0.86	20.39	2.01 ± 0.7
	Forsythia × intermedia (D)	Fi	24	3.85	15.34	1.43	19.15	3.13 ± 1
	Laurus nobilis (E)	Ln	22	4.02	42.70	0.23	16.38	2.97 ± 2.4
	Ligustrum vulgare (SD)	Lv	23	3.17	12.00	0.35	8.79	1.91 ± 0.4
	Pittosporum tobira (E)	Pt	24	3.39	36.47	1.09	35.45	2.20 ± 1.5
	Prunus laurocerasus (E)	Pl	24	7.49	27.80	0.43	9.84	2.44 ± 0.7
(b) Bolzano	Deutzia scabra (D)	Ds	30	2.48	17.70	0.28	15.16	2.20 ± 0.9
	Euonymus japonicus (E)	Ej	18	2.74	22.98	0.50	32.12	2.10 ± 1.2
	Forsythia × intermedia (D)	Fi	30	2.14	11.78	0.49	10.70	2.23 ± 0.9
	Pittosporum tobira (E)	Pt	30	6.70	25.43	0.78	35.44	2.44 ± 0.9
	Prunus laurocerasus (E)	Pl	30	2.37	31.86	0.64	77.70	2.22 ± 1.2
	Viburnum tinus (E)	Vt	30	2.33	14.48	0.33	6.67	2.02 ± 0.6

. Three of the six species [the deciduous Forsythia × intermedia (Fi), the evergreen Pittosporum tobira (Pt) and Prunus laurocerasus (Pl)] were measured in both cities. The evergreen Elaeagnus × ebbingei (Ee) and Laurus nobilis (Ln) and the semi-deciduous Ligustrum vulgare (Lv) were only measured in Lugano. The deciduous Deutzia scabra (Ds) and the evergreen Euonymus japonicus (Ej) and Viburnum tinus (Vt) were only sampled in Bolzano. To representatively sample the shrub population, the location of the selected shrub species growing in each city was retrieved using the existing city inventory implemented into GREENSPACES software (R3GIS, Bolzano, Italy). Nine (Lugano) and eight (Bolzano) blocks containing at least 2 individuals of every selected species were established in each city. The blocks were evenly distributed and included public parks, school gardens, and roadside vegetation (Figure A1 in Appendix A). Shrubs growing solitary or in shrubby patches were preferred over those growing in formal hedges.

2.2. Growth and Leaf Area Index

Biometric measurements were performed in January 2021 on all individuals of the selected species growing in the experimental areas (140 and 158 plants in Lugano and Bolzano, respectively). The number of stems and their diameter, crown radius and plant height were recorded. Stem diameters (D₃₀) were measured at 30 cm from the ground [23] with a caliper, by making two measurements perpendicular to each other and averaging the measured values. For shrubs with less than six stems, all the diameters were measured, then the radius (r = d/2) and the corresponding basal stem area (A = π r²) were calculated. The areas of the individual stems were summed (A_tot), then the overall stem radius (r_tot) and diameter (D_tot) were calculated as: r_tot = √A_tot/π; D_tot = 2 × r_tot). For shrubs with more than six stems, six representative stems were chosen and their diameters measured. The average basal area of the 6 stems was calculated, then multiplied by the number of stems to get A_tot. r_tot and D_tot were calculated as described previously. Crown projection area (CPA) was calculated from two orthogonal crown diameter measurements, performed using a mm accuracy diameter tape in the North–South and East–West direction as: CPA = π × (d_NS/2 × d_EW/2). Plant height was measured using a cm accuracy diameter tape.

The effective Plant Area Index (PAI_e, m² m⁻²) was measured using an Accupar ceptometer (Accupar LP-80, Decagon Devices, Washington, DC, USA) during bright days (PAR > 1300 μmol m⁻² s⁻¹ and progressive change in PAR < 75 μmol m⁻² min⁻¹) in July 2020 and July 2021. Measurements were carried out between 11:30 and 14:30 (CET) [24], on the same shrubs subjected to the eco-physiological measurements. Four readings of above-canopy irradiance (measured by positioning the ceptometer horizontally above the canopy) and below-canopy irradiance were collected per plant [25]. To account for Woody Area Index (WAI, m² m⁻²), a WAI/PAI ratio (α) of 0.25 was assumed based on available literature on shrubs [26,27,28]. To account for the non-normal leaf distribution within the canopy, clumping (Ω) was calculated as described by [29], assuming a uniform vertical profile of PAI_e. Leaf Area Index (LAI, m² m⁻²), i.e., half of total leaf surface area per unit ground surface area, was calculated as:

LAI = (1 − α) × PAI_e/Ω

(1)

2.3. Eco-Physiological Measurements

2.3.1. Net CO₂ Assimilation

Net CO₂ assimilation per unit leaf area (A_leaf, μmol m⁻² s⁻¹) was measured using an infra-red gas analyzer (Ciras-2, PP-System, Amesbury, MA, USA). Measurements were conducted during spring (Lugano: 16–18 June 2020; Bolzano: 26–27 May 2021), summer (Lugano: 27–28 July 2020; Bolzano: 20–21 July 2021) and fall (Lugano: 16–17 September 2020; Bolzano: 30 September–1 October 2021) between 9:00 and 17:00 (CET). Four fully expanded leaves per plant were measured (for a total of 648 and 576 leaves over the entire experiment in Lugano and Bolzano, respectively). Leaves were chosen after ideally dividing the canopy into two layers along its vertical profile: an apical (a) and a basal (b) layer, each representing 50% of the live canopy height. Within each layer, one external leaf fully exposed to solar irradiance (denoted as “ae” and “be” for apical and basal layers, respectively) and one internal leaf to the canopy subjected to self-shading (denoted as “ai” and “bi” for apical and basal layers, respectively) were analyzed. The IRGA was set to supply the leaves with a 400 ppm external CO₂ concentration, 70% of external RH, while the irradiance provided was different depending on the leaf position. External leaves (ae; be) were exposed to saturating irradiance (1300 μmol m⁻² s⁻¹), while internal leaves (ai; bi) were exposed to species-specific irradiance values, which were set as follows:

PAR_shade = PAR_sun × τ

(2)

where PAR_sun is the above-canopy PAR, while τ is the transmittance in the apical (τ_ai) and basal (τ_bi) layers of the canopy (

Table A1. The table reports the PAR values measured on apical and basal external (“ae” and “be”) and internal (“ai” and “bi”) leaves in Lugano (a) and Bolzano (b). The crown transmittance (τ), calculated as PARin/PARout, is separately reported for apical (τ_a) and basal (τ_b) leaves.

(a)
Lugano		Elaeagnus  × ebbingei	Forsythia × intermedia	Laurus nobilis	Ligustrum vulgare	Pittosporum tobira	Prunus laurocerasus
Spring	ae	1297.25	1294.44	1300.13	1298.25	1296.67	1293.33
	ai	406.63	561.44	385.50	492.75	378.11	356.44
	τa	0.31	0.43	0.30	0.38	0.29	0.28
	be	1298.13	1300.22	1297.75	1295.00	1298.78	1299.67
	bi	74.75	90.44	142.13	178.88	87.00	50.22
	τb	0.06	0.07	0.11	0.14	0.07	0.04
Summer	ae	1295.63	1295.33	1299.00	1298.63	1296.89	1298.00
	ai	441.63	469.78	397.00	464.25	369.89	350.67
	τa	0.34	0.36	0.31	0.36	0.29	0.27
	be	1294.63	1297.89	1300.00	1299.38	1296.89	1299.78
	bi	85.00	90.56	153.25	133.50	63.11	50.11
	τb	0.07	0.07	0.12	0.10	0.05	0.04
Fall	ae	1296.00	1295.22	1299.38	1292.00	1295.44	1294.89
	ai	469.13	544.56	451.50	537.00	427.89	416.33
	τa	0.36	0.42	0.35	0.42	0.33	0.32
	be	1298.63	1300.33	1297.38	1298.38	1298.89	1296.67
	bi	85.38	101.33	165.00	151.75	72.00	54.33
	τb	0.07	0.08	0.13	0.12	0.06	0.04
(b)
Bolzano		Deutzia scabra	Euonymus   japonicus	Forsythia  × intermedia	Pittosporum  tobira	Prunus  laurocerasus	Viburnum  tinus
Spring	ae	1298.43	1298.00	1296.14	1294.57	1294.29	1297.00
	ai	454.29	274.43	401.14	245.14	219.14	187.00
	τa	0.35	0.21	0.31	0.19	0.17	0.14
	be	1293.00	1295.14	1300.57	1299.14	1292.29	1300.14
	bi	175.00	90.43	156.86	86.86	56.14	83.29
	τb	0.14	0.07	0.12	0.07	0.04	0.06
Summer	ae	1295.63	1298.00	1297.63	1298.38	1296.00	1297.50
	ai	395.50	304.50	322.50	277.25	239.88	268.50
	τa	0.31	0.23	0.25	0.21	0.19	0.21
	be	1301.63	1302.38	1298.63	1297.50	1297.63	1294.25
	bi	121.75	154.75	100.88	82.25	58.00	86.50
	τb	0.09	0.12	0.08	0.06	0.04	0.07
Fall	ae	1296.50	1298.50	1295.25	1294.25	1296.43	1299.88
	ai	295.00	203.25	254.38	171.63	159.71	186.88
	τa	0.23	0.16	0.20	0.13	0.12	0.14
	be	1296.88	1296.75	1297.75	1296.75	1297.71	1293.88
	bi	99.25	84.13	94.00	53.63	52.00	60.38
	τb	0.08	0.06	0.07	0.04	0.04	0.05

in Appendix A). τ_ai and τ_bi were obtained as the ratio between the PAR measured inside the canopy (in the apical and basal layer, respectively) and above the canopy [4]. PAR_shade was in the range of 88–862 μmol m⁻² s⁻¹ (ai leaves) and 50–495 μmol m⁻² s⁻¹ (bi leaves), depending on species and solar zenith angle (θ).

According to the multilayer approach, the crown was divided into two layers, each made of sun and shaded leaves [29]. The LAI of each layer was partitioned between sun and shade using the equations:

LAI_sun = 2cosθ × (1 − e^{−0.5×Ω×LAI/cosθ})

(3)

and LAI_shade = LAI − LAI_sun

(4)

The net CO₂ assimilation per unit crown projection area (A_cpa, g m⁻² ground h⁻¹) was calculated as (More information in Appendix A):

A_cpa = 0.5 × (A_ae × LAI_sun + A_ai × LAI_shade) + 0.5 × (A_be × LAI_sun + A_bi × LAI_shade)

(5)

The total net CO₂ assimilation per plant (A_plant, g plant⁻¹ h⁻¹) was then calculated by multiplying A_cpa and CPA.

2.3.2. Stomatal Conductance and Latent Heat Dissipation

Daily latent heat dissipation by transpiration was calculated using a one-step Penman–Monteith approach. Transpiration per unit leaf area (E_leaf, mmol m⁻² s⁻¹) and stomatal conductance to water vapor (g_s, mmol m⁻² s⁻¹) were measured as previously described for A_leaf in Section 2.3.1. Nighttime measurements of g_s were also conducted between 23.00 and 01.00 (CET). The environmental variables needed to run the equation were obtained from the local weather stations of the two cities (MeteoSwiss for Lugano, Open data Alto Adige for Bolzano). The latent heat flux due to transpiration per soil square meter (LH_cpa; W m⁻²) was calculated on a daily scale, according to [30] (modified from [31]):

$$LHcpa={\displaystyle \frac{∆Q+{\rho }_a{C}_p\frac{VPD}{R_a}}{∆+\gamma \left(1+\frac{R_s}{R_a}\right)}}$$

(6)

where Q is the available energy in W m⁻²; ρ_a is the mean air density at constant pressure (1.22 kg m⁻³); C_p is the specific heat of air (1.013 kJ kg⁻¹K⁻¹); ∆ represents the slope of the saturation vapor pressure–temperature relationship; R_a is the aerodynamic resistance; and R_s is the surface resistance.

The contribution of each type of leaf (sun/shade) in each layer (apical/basal) to latent heat dissipation was calculated from independent measurements of g_s values conducted on ae, ai, be and bi leaves, both during the day and at night (LH_cpa,day and LH_cpa,night). LH_cpa was then calculated as (more information in Appendix A):

LH_cpa = LH_cpa,day + LH_cpa,night

(7)

2.3.3. Air Particulate Matter Removal

Air particulate matter removal was assessed by in situ leaf sampling and subsequent ex situ laboratory analysis. Sampling was performed in spring (Lugano: 16–18 June 2020; Bolzano: 26–27 May 2021), summer (Lugano: 27–28 July 2020; Bolzano: 20–21 July 2021), fall (Lugano: 16–17 September 2020; Bolzano: 30 September–1 October 2021) and winter (Lugano: 12–14 January 2021; Bolzano: 26–27 January 2021) on the same shrubs selected for biometrics and eco-physiological traits. A leaf sample representative of the entire crown was collected from each shrub, resulting in 54 samples in Lugano and 48 in Bolzano for each of the spring, summer, and fall seasons. During winter, only evergreen and semi-deciduous species were sampled, for a total of 45 and 32 samples in Lugano and Bolzano, respectively. Thus, a total of 207 samples in Lugano and 176 samples in Bolzano were analyzed over the entire experiment. Samples were manually collected, immediately stored at −20 °C and analyzed within one week. Each sample, made of about 300–400 cm² leaf area, was analyzed with a gravimetric technique described in detail in [32]. The number of leaves needed to cover the required sampled area depended on leaf size. To quantify PM_x accumulation per unit leaf area (µg cm⁻²) (i.e., the amount of PM_x on the leaf surface at the time of sampling), each leaf sample was subjected to a washing procedure with 250 mL of deionized water for 60 s. The washing solution was then first filtered through a metal sieve (Fritsch analysensieb, Idar-Oberstein, Germany) to eliminate particles larger than 100 µm and sequentially filtered using 3 filters with increasing retention capacities (type 1288, retention: 10 µm; type 391, retention: 2.5 µm; and PTFE membrane, retention: 0.2 µm) (Sartorius AG, Goettingen, Germany). Filtration was performed using a filtration apparatus equipped with a 47 mm glass filter funnel (Sartorius stedim, Sartorius AG, Goettingen, Germany) connected to an MV-50 vacuum pump (Comecta-Ivymen, Barcelona, Spain). Before filtration, the PTFE membranes were moistened with isopropyl alcohol to facilitate the process. Prior to filtration, each filter was dried at 60 °C for 30 min in a drying chamber (WTB Binder 7200, Tuttlingen, Germany) and then left to stabilize (at 50% air relative humidity) for 120 min before being pre-weighed on a high-precision balance (ED224S-OCE, Sartorius, Germany). After filtration, filters were dried again, the weight was stabilized, and then post-weighed. To consider potential balance drifts, the tare was reweighed every 15 min. The amount of PM was determined by the difference between the post-weight and pre-weight. Three fractions of PM_x were collected during the entire filtration procedure: (1) large (PM_10–100), (2) coarse (PM_2.5–10) and (3) fine (PM_0.2–2.5). The PM₁₀ fraction was obtained by adding the second to the third fraction, while the PM_2.5 fraction was considered equivalent to the third fraction. After filtration, the leaf area of each sample was measured using an A3 scanner (HP OfficeJet Pro 7740, HP Development Company, L.P., Palo Alto, CA, USA) and a leaf area software (Leaf Area Measurement, version 1.3, University of Sheffield A.P., Askew, UK). The PM_x accumulation was then expressed per unit leaf area (µg cm⁻²).

To obtain PM deposition per unit leaf area (µg cm⁻² day⁻¹) (i.e., the change in accumulation over time), a linear trend of deposition over the accumulation measurements was assumed. Thus, deposition was calculated as accumulation divided by the number of days without rainfall since the last effective rain (i.e., the one assumed to wash off the particles from the leaves, considered here higher than 7 mm day⁻¹). Up to a maximum of 15 days from the effective rain was observed in this experiment. Thus, the amount of deposited PM₁₀ (depPM₁₀) and PM_2.5 (depPM_2.5) was expressed per unit leaf area and time (µg cm⁻² day⁻¹). depPM_x was upscaled to PM deposition per unit crown projection area (depPM_x,cpa, g m⁻² ground day⁻¹) as follows:

depPM_x,cpa = depPM_x × LAI × 10⁻²

(8)

where 10⁻² is the conversion factor from μg cm⁻² to g m⁻².

2.4. Statistics

All statistics were performed using SPSS software (IBM SPSS Statistics for Windows, Version 29.0 Armonk, NY, USA), while figures were drawn with SigmaPlot 12.5 (Systat Software Inc., San Jose, CA, USA). Data were analyzed using linear mixed models within each city, estimated using restricted maximum likelihood (REML), with denominator degrees of freedom for tests of fixed effects obtained via the Satterthwaite approximation. In the model “species”, “leaf position” and “season” were considered fixed factors, while “block” was considered a random factor. A homogeneous ante-dependence correlation structure (AD1) was chosen based on the best fitting. Leaf position was not considered when analyzing benefits expressed per unit crown projection area. Only the species factor was considered for Leaf Area Index (LAI, m⁻² m⁻²). The assumptions of normality and homoscedasticity were verified by visual inspection of the diagnostic plots. Homogeneous subsets were identified using Tukey’s and Sidak’s post hoc tests. The growth curves DBH-R_crown and DBH-H_tot were fitted using the curve fitting tool included in SPSS. Different functions (e.g., linear, power and logarithmic functions) were tested for fitting.

3. Results

3.1. Growth Curves

Leaf Area Index (LAI), stem diameter and crown radius of the different species were monitored in Lugano and Bolzano. LAI significantly differed among species both in Lugano (p < 0.001) and Bolzano (p < 0.05) (

Table 4. For each species and experimental site, Leaf Area Index (LAI, m² m⁻²) values are shown, along with parameters (scaling constant (a) and exponenet (b)) and coefficient of determination (R²) of the regression curves (Y = a X^b) describing the relationships between stem diameter at 30 cm height (D₃₀, m) and crown radius (R_crown, m), and between D₃₀ and total plant height (H_tot, m). In the LAI column, different letters indicate significant differences among species.

				Rcrown = a × D30 b			Htot = a × D30 b
	Species	LAI (m2 m−2)		a (m)	b	R2	a (m)	b	R2
Lugano	Elaeagnus × ebbingei	5.60	a	4.182	0.604	0.375	6.839	0.573	0.579
	Forsythia × intermedia	3.73	b	8.335	0.455	0.139	8.335	0.455	0.139
	Laurus nobilis	4.20	ab	4.300	0.675	0.523	10.167	0.633	0.518
	Ligustrum vulgare	3.39	b	5.088	0.684	0.429	1.912	0.007	0.000
	Pittosporum tobira	4.63	ab	4.065	0.699	0.552	12.122	1.044	0.568
	Prunus laurocerasus	5.80	a	7.886	1.017	0.619	9.231	0.711	0.483
Bolzano	Deutzia scabra	3.38	ab	10.646	0.975	0.687	9.142	0.564	0.275
	Euonymus japonicus	3.07	b	10.270	0.955	0.938	7.519	0.561	0.802
	Forsythia × intermedia	3.54	ab	6.710	0.684	0.218	5.201	0.31	0.058
	Pittosporum tobira	3.83	ab	4.164	0.582	0.215	7.681	0.592	0.409
	Prunus laurocerasus	4.50	a	4.749	0.693	0.588	8.966	0.635	0.703
	Viburnum tinus	3.94	ab	3.749	0.582	0.623	20.684	0.912	0.747

). In Lugano, Ee, Ln, Pt and Pl showed higher LAI than Fi and Lv. In Bolzano, LAI was similar among species except for Pl, which showed a higher LAI than Ej (Table 4). The three species that were monitored in both cities (Fi, Pt and Pl) exhibited, in general, higher LAI in Lugano than in Bolzano (p < 0.05).

D₃₀ and species were significant predictors of the crown radius (R_crown) (p < 0.001). The best fitting between D₃₀ and R_crown was obtained in both cities using a power function (R_crown = a × D₃₀ ^b) with average R² of 0.439 and 0.545 in Lugano and Bolzano, respectively (Table 4). D₃₀ and species were significant predictors of plant height (H_tot) (p < 0.001). The power function (H_tot = a × D₃₀ ^b) best-fitted the observed data in all species with an average R² of 0.381 and 0.499 in Lugano and Bolzano, respectively. The specific coefficient values and R² of both regressions, categorized by species and city, are shown in Table 4.

3.2. CO₂ Assimilation

3.2.1. CO₂ Assimilation per Unit Leaf Area

Photosynthesis of the different species was measured in sun and shaded leaves in Lugano and in Bolzano, during the different seasons. In Lugano, the net CO₂ assimilation per unit leaf area (A_leaf, μmol m⁻² s⁻¹) significantly differed among species, seasons and leaf positions (p < 0.001). Fi and Ee showed higher A_leaf than Ln, Lv, and Pl. Pl had significantly lower A_leaf compared to all the other species investigated (Figure 1A). Apical external leaves (ae) had 6.94% higher A_leaf than basal external leaves (be). Shaded apical and basal leaves (ai and bi) had 60% and 83.2% lower A_leaf, respectively, compared to ae (Figure 1C). A_leaf declined in all species from spring to summer (−16.16%) and recovered in fall (+21.3% compared to summer) (Figure 1E).

In Bolzano, A_leaf was significantly affected by species and leaf position (p < 0.001) but not by season (p = 0.241). Fi and Pt showed higher A_leaf than all the other species investigated. Pl showed lower A_leaf than all the other species except for Ej (Figure 1B). A_leaf was similar in ae and be leaves. Ai and bi leaves had 51.7% and 77.5% lower A_leaf compared to ae (Figure 1D). Full data of A_leaf for each species, leaf position, season, and city can be found in

Table A2. The stomatal conductance to water vapor (g_sw; mmol H₂O m⁻² s⁻¹) and the net CO₂ assimilation per unit of leaf area (A; μmol CO₂ m⁻² s⁻¹) in different leaf positions and different seasons for each city. ae = apical external; be = basal external; ai = apical internal and bi = basal internal.

	gsw
	Spring				Summer				Fall
Lugano	ae	ai	be	bi	ae	ai	be	bi	ae	ai	be	bi
Elaeagnus × ebbingei	135.00	61.88	119.13	44.25	116.00	46.00	106.25	31.13	173.50	73.50	178.00	48.13
Forsythia × intermedia	229.56	134.44	201.78	102.78	132.00	112.11	149.22	59.11	215.11	140.11	205.56	131.89
Laurus nobilis	142.38	68.13	144.88	72.00	103.50	55.50	94.88	40.00	150.00	77.50	118.63	62.00
Ligustrum vulgare	98.63	80.38	104.63	67.63	95.38	51.13	70.13	45.13	137.50	94.13	102.38	71.50
Pittosporum tobira	152.56	57.00	151.22	44.22	100.44	58.89	100.44	38.67	133.78	51.89	139.22	42.78
Prunus laurocerasus	106.56	64.33	109.33	49.44	82.33	42.56	74.78	25.56	124.67	49.44	106.44	46.67
Bolzano	ae	ai	be	bi	ae	ai	be	bi	ae	ai	be	bi
Deutzia gracilis	63.57	65.14	66.57	47.43	49.00	43.63	59.25	42.38	81.25	81.88	89.63	71.88
Euonymus japonicus	53.86	38.00	47.86	38.14	59.75	39.63	69.88	26.13	71.63	46.00	58.75	42.00
Forsythia × intermedia	58.29	60.14	84.57	53.29	77.88	72.50	119.25	56.88	66.75	83.50	99.25	73.50
Pittosporum tobira	66.71	47.43	48.00	21.43	72.38	55.13	80.38	41.38	113.50	58.00	81.50	27.13
Prunus laurocerasus	54.57	42.43	44.86	25.71	62.63	46.38	65.25	23.75	56.71	41.86	68.29	37.14
Viburnum tinus	77.14	42.71	76.14	41.43	70.63	44.25	70.38	34.63	76.63	52.50	80.75	30.00
	A
	Spring				Summer				Fall
Lugano	ae	ai	be	bi	ae	ai	be	bi	ae	ai	be	bi
Elaeagnus × ebbingei	11.40	3.28	9.78	1.44	10.05	3.33	8.39	1.13	12.23	3.76	13.05	1.39
Forsythia × intermedia	12.42	6.41	12.39	2.57	9.86	6.02	10.99	1.32	12.41	4.97	11.14	2.46
Laurus nobilis	8.65	3.88	9.15	1.44	8.06	3.33	7.35	1.89	8.89	3.18	7.64	2.35
Ligustrum vulgare	7.74	4.51	8.28	3.09	7.41	2.95	5.51	1.78	11.01	5.71	8.19	2.33
Pittosporum tobira	10.28	3.32	10.64	1.36	8.91	3.54	8.32	1.22	10.79	3.58	9.98	1.58
Prunus laurocerasus	8.76	3.23	8.58	0.72	7.76	2.66	6.12	0.71	8.47	2.49	7.44	0.90
Bolzano	ae	ai	be	bi	ae	ai	be	bi	ae	ai	be	bi
Deutzia gracilis	6.77	4.67	6.04	2.43	4.49	2.98	5.23	1.58	4.99	2.63	5.35	1.25
Euonymus japonicus	5.29	2.44	4.86	1.14	5.08	2.21	5.60	0.89	5.89	2.18	5.13	1.13
Forsythia × intermedia	5.99	4.21	7.36	2.44	7.89	3.89	9.04	1.59	5.58	3.39	6.28	1.66
Pittosporum tobira	6.76	3.16	5.14	1.23	6.83	3.60	7.03	1.48	8.81	3.08	7.06	1.06
Prunus laurocerasus	5.27	2.53	4.19	0.74	5.26	2.74	5.10	0.86	4.71	1.77	5.46	0.86
Viburnum tinus	6.73	1.69	6.57	1.43	6.05	2.18	5.43	1.13	6.41	2.58	6.24	0.84

in Appendix A.

3.2.2. CO₂ Assimilation per Unit Canopy Cover

In Lugano, the net hourly CO₂ assimilation per unit crown projection area at saturating irradiance (A_cpa, g m⁻² ground h⁻¹) was significantly affected by species (p < 0.001). Season (p = 0.089) and the interaction between species and seasons (p = 0.948) were not significant. Ee, Fi, and Pt had higher A_cpa than Lv (Figure 2A). A_cpa was about 14.3% higher in spring than in summer (Figure 2C).

In Bolzano, A_cpa was affected by species (p < 0.001), while season (p = 0.122) and the season × species interaction were not significant (p = 0.506). Fi and Pt displayed higher A_cpa compared to Ej (Figure 2B). A_cpa was about 13.8% higher in spring and summer than in fall (Figure 2D).

CO₂ assimilation per plant during an hour of saturating sunlight irradiance, calculated as the product of A_cpa and CPA, varied between 6.18 and 41.42 g plant⁻¹ h⁻¹ in Lugano and between 4.29 and 18.15 g plant⁻¹ h⁻¹ in Bolzano (

Table 5. Average crown projection area (CPA_avg, m²) and net CO₂ assimilation per plant (A_plant, g plant⁻¹ h⁻¹) of the species investigated in Lugano and Bolzano.

		CPAavg	Aplant
Lugano	Species	m2	g plant−1 h−1
	Elaeagnus × ebbingei	5.35	27.29
	Forsythia × intermedia	9.25	41.42
	Laurus nobilis	4.65	16.21
	Ligustrum vulgare	1.99	6.18
	Pittosporum tobira	6.19	26.15
	Prunus laurocerasus	4.87	19.10
Bolzano	Species	m2	g plant−1 h−1
	Deutzia scabra	3.35	6.84
	Euonymus japonicus	8.06	12.96
	Forsythia × intermedia	3.10	7.84
	Pittosporum tobira	7.11	18.15
	Prunus laurocerasus	6.22	12.81
	Viburnum tinus	2.36	4.89

).

3.3. Latent Heat Dissipation

3.3.1. Transpiration per Unit of Leaf Area

Transpiration and stomatal conductance of sun and shaded leaves of the different species were measured in different seasons in Lugano and in Bolzano. In Lugano, transpiration per unit leaf area (E_leaf, mmol H₂O m⁻² s⁻¹) was significantly affected by species, leaf position and season (p < 0.001). Fi displayed higher E_leaf than the other species evaluated (Figure 3A). Pl was the species with lower E_leaf and transpired 42.8% less water than Fi. All species experienced a temporary slight drop in E_leaf during summer (−14.9% compared to spring), then E_leaf fully recovered during fall. The interaction between species and leaf position was not statistically significant (p = 0.094).

In Bolzano, E_leaf was significantly affected by species, leaf position and season (p < 0.001). Fi showed 33% higher E_leaf than Pl. E_leaf of Vt and Pt did not differ from Fi. E_leaf increased by 42.8% from spring to summer, then dropped by 40.2% during fall (Figure 3D). The interaction between the leaf position and the seasons was significant (p = 0.01).

3.3.2. Latent Heat Dissipated per Unit Canopy Cover

In Lugano, the latent heat dissipated per unit crown projection area (LH_cpa, W m⁻²) was significantly affected by the species (p < 0.005) and season (p < 0.001). The interaction between species and seasons was not significant (p = 0.929). Ee showed 27.7% higher LH_cpa than Ln (Figure 4A). LH_cpa was statistically similar in spring and summer, then it decreased by 32.2% in fall (Figure 4C).

In Bolzano, only season significantly affected LH_cpa (p < 0.001), while species (p = 0.088) and the interaction between species and seasons (p = 0.946) did not (Figure 4B). LH_cpa increased by 26.7% from spring to summer, and then it dropped by 48.7% in fall (Figure 4D).

3.4. Air Quality Amelioration

3.4.1. PM Accumulation and Deposition per Unit of Leaf Area

Particulate matter accumulated on leaves of the different species growing in Lugano and Bolzano was quantified. In Lugano, species affected the accumulation of PM_2.5–10 (p = 0.008). Lv accumulated 33% more PM_2.5–10 than Fi and Ln. No significant difference was observed among species for PM_10–100 (p = 0.160) and PM_0.2–2.5 (p = 0.173) accumulation, with annual averages of 8.798 µg/cm⁻² and 0.869 µg cm⁻², respectively (Figure 5A).

In Bolzano, species affected the accumulation of PM_10–100 (p < 0.001) and PM_2.5–10 (p = 0.001), while the accumulation of PM_0.2–2.5 was unaffected (p = 0.326). The season of sampling significantly affected the accumulation of PM_10–100 (p = 0.042) and PM_2.5–10 (p < 0.001), but did not affect the accumulation of PM_0.2–2.5 (p = 0.096). Evergreen species (Ej, Pt, Pl, and Vt) were 48% more efficient, on an annual basis, in PM_10–100 accumulation than deciduous species (Ds and Fi). Moreover, Ej was found to be 35% more effective in accumulating PM_2.5–10 than Ds and Fi (Figure 5B). No interaction was observed between species and season for the accumulation of any PM fraction in either of the two cities studied, so species were not analyzed separately by season.

Species (p = 0.006) and season (p = 0.030) affected PM₁₀ accumulation in Lugano, whereas species did not affect depPM₁₀ (p = 0.115). Lower PM₁₀ accumulation was observed in spring compared to winter. In contrast, depPM₁₀ (p < 0.001) was higher in spring compared to summer, when, in turn, higher depPM₁₀ than in fall and winter was observed (Figure 6A,B). PM_2.5 accumulation (p = 0.749) was unaffected by season, while depPM_2.5 was significantly higher in spring (p < 0.001) (Figure 6C,D).

In Bolzano, the accumulations of PM₁₀, depPM₁₀, and depPM_2.5 were significantly affected by season (p < 0.001 for all traits). Species affected neither depPM₁₀ (p = 0.131) nor depPM_2.5 (p = 0.318). PM₁₀ accumulation was lower in summer compared to all other seasons (Figure 6E). depPM₁₀ was higher in spring than winter, when, in turn, higher depPM₁₀ was observed compared to summer and fall (Figure 6F). The accumulation of PM_2.5 was unaffected by season, whereas DepPM_2.5 was higher in spring compared to other seasons (Figure 6G,H).

If total PM accumulation is compared between cities using data from the three species that were both in Lugano and Bolzano, Pt and Pl accumulated 14% more PM in Bolzano than in Lugano. Conversely, Fi accumulated 19% more PM in Lugano than in Bolzano.

Seasonal and annual accumulation and deposition values expressed per unit of leaf area for the different PM fractions by the six shrub species sampled in Lugano and Bolzano are provided in Appendix A.

3.4.2. PM Deposition per Unit Canopy Cover

In Lugano, PM₁₀ deposition per unit crown projection area (depPM_10,cpa) was significantly affected by species (p < 0.001) and season (p < 0.001). Ee had significantly higher depPM_10,cpa compared to Ln. depPM_10,cpa was significantly higher in spring than in summer, when, in turn, higher depPM_10,cpa was observed compared to fall and winter. depPM_2.5,cpa was affected by season (p < 0.001), but was not affected by species (p = 0.059).

In Bolzano, species had no effect on depPM_10,cpa (p = 0.154) or depPM_2.5,cpa (p = 0.457). Conversely, the season significantly affected both depPM_10,cpa and depPM_2.5,cpa (p < 0.001).

Seasonal and annual deposition values expressed per unit canopy cover for both fractions by the six shrub species in Lugano and Bolzano are provided in Appendix A.

4. Discussion

4.1. CO₂ Removal from the Atmosphere

When estimated as CO₂ storage and sequestration, the capacity of shrubs to remove CO₂ is generally believed to be low, due to their small biomass at maturity [33,34]. In this research, CO₂ removal was estimated using a different approach that considers the photosynthetic capacity of the different species, as previously done with trees [4]. We conducted leaf gas exchange and Leaf Area Index measurements to calculate CO₂ assimilation per unit crown projection area (A_cpa). A_cpa is the net CO₂ uptake from the atmosphere (i.e., the photosynthetic CO₂ gain minus leaf autotrophic respiration) per unit canopy cover. Because shrubs are often planted and managed as masses rather than as individual plants, A_cpa may particularly be valuable for shrub species, since it provides an area-based metric for CO₂ uptake [34]. A_cpa of the shrub species investigated here ranged between 1.6 and 5.1 g CO₂ m⁻² ground h⁻¹, depending on species. One m² ground planted with Fi, Pt, Ee, Ds removed about 40% more carbon compared to the same ground area planted with Ln, Lv, Ej.

The main determinants of A_cpa are Leaf Area Index (LAI) and net photosynthetic rate per unit leaf area [35]. Shrubs often display high canopy density, which can be further increased when they are regularly pruned [30,36], as it occurs in most municipalities [37]. High canopy density can result in similar LAI in shrubs compared to trees, although canopy height is substantially lower [36]. Consistently, shrub species investigated here displayed, on average, 10% higher LAI compared to tree species growing in similar climates [4]. Dense canopies with LAI > 4.5 were found in Ee, Pl, and Pt, especially when grown in Lugano, where high rainfall during sprouting may have favored shoot growth [38,39]. Increasing LAI enhances photosynthetic area per unit ground area, but the amount of canopy subjected to self-shading and shade intensity increases as well [35]. Plants can acclimate the morpho-anatomy and physiology of their leaves to maintain a positive carbon balance in shade, although the capacity to acclimate depends on species and shade density [40,41]. As an example, the contribution of shaded leaves to overall canopy photosynthesis was 7% lower in Lugano than in Bolzano, because the denser canopies in the former city reduced light penetration to the inner canopy.

The contribution of shaded leaves to canopy photosynthesis did not significantly differ among species, although photosynthetic rate per unit leaf area did. Fi, a fast-growing deciduous species with thick palisade tissue and high chlorophyll levels [42], had higher photosynthesis per unit leaf area than most of the other species in both cities. Such a difference may be due to lower diffusive limitations to photosynthesis and to higher nitrogen allocation to the photosynthetic apparatus occurring in deciduous than in evergreen species [43]. The 37% difference in A_leaf observed in Fi between Lugano and Bolzano suggests opportunistic water use by Fi, which maximizes photosynthesis under high moisture availability and strictly modulates stomatal opening to avoid dehydration in drier conditions [44]. Pt, an evergreen heat-tolerant Mediterranean species with a more conservative water-use [44], was as effective as Fi for CO₂ assimilation when planted in the drier conditions of Bolzano. At the other end of the spectrum, Pl, a species known for its high stomatal sensitivity to moisture availability and high hydraulic limitations under water stress [45], displayed the lowest photosynthetic rate among the species investigated in both cities. The A_leaf values observed in this research were consistent with previous literature [45,46,47], although we found that A_leaf was, in general, lower in Bolzano than in Lugano. Such a result was unexpected, because the cities fall within the same climatic zone, but it may be due to higher soil pH in Bolzano than in Lugano and to the differing total annual rainfall and rainfall distribution between the cities. Although total rainfall was 40% lower in Bolzano than in Lugano, the latter city experienced a relatively dry summer, which did not occur in Bolzano, where summer rainfall exceeded the 30-year average by 36%.

Modeling the change in whole-plant net CO₂ assimilation as a function of plant age or stem diameter, as previously done with tree species [4], may be of little practical significance and subject to high uncertainty in shrubs for several reasons: (1) stem diameter of shrubs are seldom measured and recorded in municipal inventories [48]; (2) criteria for stem diameter measurement are often inconsistent among municipalities and tree workers, because shrubs are usually multi-stemmed and branched below 1.3 m [23]; (3) stem diameter may be a poor predictor of crown projection area and plant height in shrubs, as displayed by the low R² of the allometric relationship obtained here, because crown radius and height were greatly affected by planting distances, planting layout, and pruning regime [49]. On average, shrubs measured in this research assimilated 6.2 to 41.4 g CO₂ h⁻¹ plant⁻¹ in Lugano and 4.9 to 18.2 g CO₂ h⁻¹ plant⁻¹ in Bolzano. This is 97.8% lower than trees under a similar climate [4]. However, such a large difference between trees and shrubs is mostly due to differences in canopy size. Indeed, when estimated per square meter of canopy cover, CO₂ assimilation of shrubs accounted for 23.6% of that observed in tree species in a similar climate zone [4]. This indicates that CO₂ uptake by the urban forest can be increased by designing multi-layered green areas which integrate a shrub layer made of species with high A_cpa, as also suggested by [34].

4.2. Latent Heat Dissipation

Only a few studies evaluated the capacity of shrubs to improve the microclimate. Most of these studies assessed microclimate amelioration by measuring temperature using infrared or remote sensing [50,51]. In this study, latent heat dissipation (LH_cpa, W m−²) was evaluated using a Penman–Monteith approach based on direct measurements of leaf gas exchange. LH_cpa by shrubs ranged between 87 and 136 W m⁻² over the 24 h, depending on the species. The latent heat dissipation by Ln was approximately one-third that of the same surface area planted with Ee (Figure 4A,B). Main determinants of LH_cpa are stomatal and aerodynamic resistances to transpiration, weather parameters and LAI.

Transpiration per unit of leaf area (E_leaf) differed significantly among species in both cities. Fi had the highest transpiration rate of the investigated species and transpired 52.3% more water than Pl, the species with the lowest transpiration rate. In Lugano, Fi outperformed other species in terms of E_leaf. The 54% difference in E_leaf observed for Fi between Lugano, characterized by higher total annual rainfall, and Bolzano confirms its opportunistic water use, which maximizes stomatal opening under higher water availability, but strongly regulates water loss during heat and drought stress. In contrast, Pt, a species characterized by a more conservative water use, showed only 14% variation in E_leaf between the two cities despite their different microclimatic conditions during the measurements [44]. Regardless of the city, E_leaf measured in this experiment were lower than those found on well-watered plants of the same species [45,46,47,52] and were similar to those found on Fi, Ln, Pt and Vt in studies under drought stress in pot-grown plants [46,53].

The results of this experiment indicate how differences in LAI and canopy density among species can affect cooling capacity. Transpiring leaf area per ground square meter increases indeed with LAI, with positive effects on cooling. Thus, species with high LAI, such as Ee, were extremely effective in cooling, although their E_leaf was relatively low. Conversely, low LAI prevented Fi from providing higher LH_cpa than most of the other species, despite its higher E_leaf. On the other hand, light transmittance to the inner canopy decreases with increasing LAI, thus increasing the relative amount of leaves subjected to self-shading while decreasing the irradiance reaching shaded leaves (Appendix A). As a consequence, shaded leaves of the high-LAI species Ee transpired 60% less water than their full-sun counterparts, whereas transpiration of shaded leaves of the sparse-canopy low-LAI Lv was only 34% lower than full-sun leaves.

LH_cpa increased with temperature and vapor pressure deficit (VPD), being higher in summer and lower in fall. This underlines the high potential for cooling of shrubs, which were capable of higher latent heat dissipation when it is more beneficial to the population. Apart from temperature, the drop of LH_cpa in fall may be associated with a higher frequency of self-shaded leaves within the canopy, due to wider solar zenith angle.

Schwaab et al. [54] showed how green spaces with trees cool down 2–4 times more than treeless green spaces. Latent heat fluxes around 100–200 W m⁻² are reported in the literature for urban trees, measured both with direct [30,55] and indirect [56] methods. We show here that the latent heat flux by the shrub species investigated was in the same order of magnitude as that of trees. Thus, although shrubs are much less effective at casting shade than trees, because of their smaller size [30], their capacity to dissipate latent heat through transpiration should not be underestimated.

4.3. Particulate Matter Removal

Although bark contributes to PM capture, leaves are the main plant organs responsible for atmospheric PM accumulation [57]. Leaf micromorphological traits, such as trichomes, stomata, and epicuticular waxes, are often linked to higher PM retention [58,59,60]. Baraldi et al. [61] used Scanning Electron Microscopy to examine Ligustrum spp., revealing peltate trichomes, waxes, and cuticular characteristics (ridges, grooves, prominent stomata, and irregular wax deposits), which may explain the higher PM_2.5–10 and PM₁₀ accumulation observed in Lv in Lugano. In the same study [61], Ln showed a stomatal density similar to Ligustrum spp. (200–300 stomata per mm²) but lacked trichomes. While higher stomatal density is generally linked to greater PM adsorption on leaves [62], the absence of additional contributing factors in Ln may explain its low PM accumulation in Lugano. This trend was also confirmed by other studies, where Ln was among the least efficient species in capturing PM [63]. Ee, which showed intermediate PM_2.5–10 and PM₁₀ accumulation in Lugano, was identified by Mori et al. [63] as the most efficient species in PM capture due to high leaf area, biomass, and favorable leaf traits like trichomes [6]. Fi was among the least efficient species for PM_2.5–10 and PM₁₀ accumulation in Lugano and also performed poorly in Bolzano, including for PM_10–10. Dzierzanowski et al. [64] found that Fi leaves had 2 to 8 times more waxes than other species analyzed; however, this did not significantly enhance PM retention. The potential of epicuticular waxes to trap PM depends on their chemical composition and structure, which vary depending on species [65]. Additionally, Fi lacks ultrastructures, such as thin films and tubules, that promote PM accumulation [65], potentially explaining its low efficiency observed in this study.

Macroscopic leaf properties (e.g., shape and size) and canopy structure can impact leaf PM accumulation capacity [60,66], interacting with climatic and environmental factors [67]. Leaf density within the canopy also significantly affects PM adsorption. In Lugano, Lv had the highest PM accumulation but the lowest LAI (33% below the average of other species). Such a result may be explained by the micro-turbulence generated in porous canopies, which can enhance PM_2.5–10 and PM₁₀ capture. Canopy density influences wind turbulence, a key factor for PM deposition, especially fine particles [68]. Excessive leaf density reduces canopy porosity, limiting airflow and consequently its filtering effect [69,70]. Similarly, in Bolzano, Ej, which had 20% lower LAI than the other species analyzed, showed high accumulation of PM_10–100, PM_2.5–10, and PM₁₀. As with Lv in Lugano, higher canopy porosity in Ej likely enhanced PM capture.

Evergreen species capture PM throughout the year, including winter when air PM levels are higher, unlike deciduous species that shed their leaves annually. However, since evergreens retain their leaves for years, they may accumulate less new particulate matter each season [58,71]. In Bolzano, Vt was highly efficient in capturing PM_10–100 per unit of leaf area, consistently with previous research [63]. This efficiency is likely due to its abundant wax layer [61] and a dense trichome indumentum, which enhances PM retention.

Comparing the cumulative adsorption of the three species grown in both Lugano and Bolzano with the atmospheric PM concentrations during the sampling and pre-sampling periods in both cities, some trends emerged. The use of evergreen species such as Pt may be particularly effective in sites where air PM concentration peaks during winter, as was the case in Bolzano. In contrast, deciduous species can be used effectively when the peak in air PM concentration occurs during the growing season, such as in Lugano.

Meteorological and environmental factors can influence the complex process of accumulation, wash-off and re-suspension of PM [72,73]. In both Lugano and Bolzano, PM₁₀ accumulation increased over the year, coherently with previous studies [74,75]. This is probably linked to higher winter air PM concentrations, primarily due to heating-related combustion. High winter accumulation is also attributed to wet leaf surfaces and fog, which favor wet deposition [76,77]. While high pollution levels can promote leaf PM accumulation [78,79], heavy rain and strong winds can wash off or resuspend accumulated PM [80]. In Bolzano, summer rain corresponded to the lowest leaf accumulation, while in Lugano, spring rainfall likely washed away PM, explaining lower accumulation in that season. The high PM accumulation on leaves observed in Bolzano can be explained by the dominance of evergreen species in the sample, which capture PM year-round, and by much less rainfall experienced by that city during spring, compared to Lugano [81]. Deciduous species, which shed leaves, tend to capture less PM and, among them, those that retain their leaves for longer periods are more efficient.

The opposite seasonal trend of accumulation and deposition (both per unit of leaf area and per unit canopy cover) supports the idea that particulate matter is primarily adsorbed onto specific leaf microstructures (e.g., cuticular characteristics, trichomes, glands, etc.) [59]. As these microstructures become saturated, the deposition rate decreases [82], resulting in fewer particles that can accumulate daily on leaf surfaces that are nearing saturation.

4.4. Limitations

In this research, net photosynthetic rate, stomatal conductance and PM accumulation measured per unit leaf area were upscaled to calculate the benefits provided by some shrub species. We independently measured photosynthesis and stomatal conductance in sun and shaded leaves to upscale the data using a sun–shade model, which is more accurate than the traditional big-leaf model [12]. Nonetheless, we are aware that the estimation of CO₂ assimilation and latent heat dissipation per unit crown projection area relies on strong assumptions in the upscaling procedure. Similarly, the methods to calculate the PM deposition relied on the assumption that a rainfall event with intensity >7 mm could completely wash off PM from leaves.

5. Conclusions

This research shows that, although their benefits per plant are inextricably lower than in tree species because of smaller size, shrubs can provide significant ecosystem services in terms of CO₂ assimilation, latent heat dissipation, as well as accumulation and deposition of atmospheric particulate matter (PM). Per m² of crown projection area, shrubs dissipated up to 130 W m⁻² latent heat per day and assimilated up to 5 g CO₂ m⁻² ground h⁻¹ of full sun, which can be comparable to tree species. Moreover, shrubs were efficient filters for particulate matter due to their growth form, indicating possible benefits of planting them near roads, one of the most significant emission sources. Nonetheless, the provision of such benefits varies between species, emphasizing the need for a science-based design of urban vegetation that takes into account the capacity of a given species to provide the desired benefit.