Journal of Environmental Management 159 (2015) 1e10
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Research article
Habitat complexity influences fine scale hydrological processes and the incidence of stormwater runoff in managed urban ecosystems Alessandro Ossola a, *, Amy Kristin Hahs b, Stephen John Livesley a a
Faculty of Science, 500 Yarra Boulevard, The University of Melbourne, Richmond, 3121, VIC, Australia Australian Research Centre for Urban Ecology, Royal Botanic Gardens Melbourne, c/o School of BioSciences, The University of Melbourne, 3010, VIC, Australia
b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 January 2015 Received in revised form 24 April 2015 Accepted 4 May 2015 Available online xxx
Urban ecosystems have traditionally been considered to be pervious features of our cities. Their hydrological properties have largely been investigated at the landscape scale and in comparison with other urban land use types. However, hydrological properties can vary at smaller scales depending upon changes in soil, surface litter and vegetation components. Management practices can directly and indirectly affect each of these components and the overall habitat complexity, ultimately affecting hydrological processes. This study aims to investigate the influence that habitat components and habitat complexity have upon key hydrological processes and the implications for urban habitat management. Using a network of urban parks and remnant nature reserves in Melbourne, Australia, replicate plots representing three types of habitat complexity were established: low-complexity parks, high-complexity parks, and high-complexity remnants. Saturated soil hydraulic conductivity in low-complexity parks was an order of magnitude lower than that measured in the more complex habitat types, due to fewer soil macropores. Conversely, soil water holding capacity in low-complexity parks was significantly higher compared to the two more complex habitat types. Low-complexity parks would generate runoff during modest precipitation events, whereas high-complexity parks and remnants would be able to absorb the vast majority of rainfall events without generating runoff. Litter layers on the soil surface would absorb most of precipitation events in high-complexity parks and high-complexity remnants. To minimize the incidence of stormwater runoff from urban ecosystems, land managers could incrementally increase the complexity of habitat patches, by increasing canopy density and volume, preserving surface litter and maintaining soil macropore structure. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Habitat complexity Management Urban hydrology Soil properties Litter Vegetation
1. Introduction Urban ecosystems provide numerous environmental, socioeconomical and ecological benefits to cities and towns (Bolund and Hunhammar, 1999; Chiesura, 2004; McPherson et al., 1997). As extreme precipitation events are likely to increase with climate change (Yilmaz et al., 2014), the effective management of hydrological processes and related ecosystem services such as stormwater drainage, runoff mitigation, soil water storage and purification is becoming increasingly important to create more sustainable and resilient cities and towns (Bolund and Hunhammar, 1999; Cunningham et al., 2010; Pauleit and Duhme, 2000; Nouri et al., 2013).
* Corresponding author. E-mail addresses: [email protected] (A. Ossola), hahsa@unimelb. edu.au (A.K. Hahs), [email protected] (S.J. Livesley). http://dx.doi.org/10.1016/j.jenvman.2015.05.002 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
The traditional approach in evaluating hydrological processes and benefits within urban areas has focused on mapping and modelling the hydrology of urban land cover types at large landscape scales, such as the city or catchment scale (Pauleit and € mana and Gill, 2014; Duhme, 2000; Perry and Nawaz, 2008; Sjo Tratalos et al., 2007; Whitford et al., 2001). These models have then been applied to estimate changes in hydrological processes under climate change scenarios (Gill et al., 2007) or calculate economical benefits associated with runoff reduction (Zhang et al., 2012). However, these models can be problematic as soil hydrological properties are likely to vary with soil physical properties which are highly variable at a very fine scale (Pickett and Cadenasso, 2009). Currently, very few studies have investigated the variability of soil hydrological properties in urban ecosystems through empirical measurements (e.g. Gregory et al., 2006; Yang and Zhang, 2011).
2
A. Ossola et al. / Journal of Environmental Management 159 (2015) 1e10
Together with soils, other habitat components, such as litter and vegetation layers, are likely to exert significant effects on the hydrology of urban ecosystems (Nouri et al., 2013). The absolute amount of these habitat components (e.g. volume, surface area) also define the overall complexity of habitat patches (McCoy and Bell, 1990; Byrne, 2007), and may influence the presence of organisms, such as plant and invertebrates, able to affect hydrological processes at a fine scale (Bartens et al., 2008; Colloff et al., 2010). The hydrological impact of vegetation has been previously assessed in a few urban ecosystem studies. For example, throughfall and stemflow have been measured in some urban forests and under isolated trees and related to the complexity of vegetation layers (e.g. GuevaraEscobar et al., 2007; Inkil€ ainen et al., 2013; Livesley et al., 2014). Nevertheless, to date there have been no studies specifically investigating the hydrology of litter layers in urban ecosystems. Management practices (or lack of) can alter directly or indirectly each of these system components, determining changes in the overall complexity of urban habitats (Byrne, 2007; Gaston et al., 2013), and therefore directly impacting on the local hydrological processes (Fig.1). Building on this bi-directional interaction, this study aims to holistically investigate the effects of management-driven differences in habitat complexity upon hydrological properties for each habitat component (soil, litter and vegetation). We addressed the following research questions: How do soil physical and hydrological properties vary in urban habitats characterized by different levels of habitat complexity? What is the hydrological role of each habitat component (soil, litter, vegetation)? How can our findings inform ecologically-sensitive urban habitat management to optimize urban water conservation and retention? 2. Materials and methods 2.1. Study area The study was conducted in the south-eastern Melbourne metropolitan area, Australia. The geology of the study area was restricted to quaternary and tertiary sandstones and therefore sandy soils, such as Podosols and Tenosols (sand > 89%). The study area was also confined to a single ecological vegetation class, herb-rich, heathy
Management
Habitat complexity Vegetation
Litter
Interception Throughfall
Water storage Litter input
Evapotranspiration
Evaporation Interception
Soil
woodland (The State of Victoria, Department of Environment and Primary Industries, 2013), so that differences in soil properties amongst habitat types were brought about through management practices since the suburbs were established. The altitude is between 15 and 50 m a.s.l., the annual mean maximum and minimum temperatures are 19.7 and 10.1 C, while the average annual precipitation (1950e2014) is 711 mm (Bureau of Meteorology, 2014). We selected 3 habitat types based upon their structural complexity and management history (Fig. 2), and established 10 research plots (20 30 m) within each type. Plots were selected in flat areas to minimize soil erosion and transport due to superficial runoff. Low-complexity parks (LCP) were characterized by an overstory of various Eucalyptus species and a simplified herbaceous understory. Plots were consistently mown since the establishment of the park and the herbaceous ground layer remained 250 mm) and microaggregates (m, 250 53 mm), as the distinction between large and small macroaggregates is not informative for sandy soils (J. Six, personal communication, September 30, 2013). Soil particle density (DP) was measured using a pycnometer (Carter and Gregorich, 2007) and soil organic matter content (SOM) using loss on ignition (550 C, 3 h). Soil water holding capacity at 24 h (S_WHC) was calculated in triplicates for soil sampled (0e5 cm) from each plot (Wilke, 2005).
Surface litter mass was collected in April 2013 from three squares (50 50 cm) per plot and the material was oven dried (105 C) for 48 h before weighing. Litter was sieved (1 cm) to determine fine and coarse fractions. Leaf litter water holding capacity (L_WHC) was measured following Naeth et al. (1991). Briefly, the weighed oven dry litter was soaked in a bucket overnight, allowed to drain for 24 h undercover to prevent evaporation and then reweighed to determine L_WHC. Rainfall interception was measured over six months using a Davis™ tipping bucket (Dataflow Systems, Christchurch, New Zealand) anchored to the soil surface in 7 randomly selected plots, due to limitations in the number of instruments available. Baseline precipitation data were downloaded from the Bureau of Meteorology (2014) website for the stations n. 86,210 and 86,020. The sky view factor (0 ¼ no visible sky, 1 ¼ completely visible sky) was used as a proxy for the structural complexity of vegetation and it was measured once taking a hemispherical picture over each of the tipping buckets during summer.
2.5. Data analysis Soil total porosity (St) was calculated using the Equation (1):
(1)
where BD is the bulk density and DP is the particle density (Carter and Gregorich, 2007). The field-saturated hydraulic conductivity (Kfs) was calculated using the model proposed by Nimmo et al. (2009):
(2)
where LG is ring-installation scaling length, tf is the final time, D0 is the initial ponded depth and the macroscopic capillary length l of the soil is 0.08. The model corrects mathematically for nonconstant falling head and subsurface lateral spreading and it does not require measurement of the water status of soils (Nimmo et al., 2009). Unsaturated hydraulic conductivity (K) was modelled according to the MiniDisk manufacturer specifications based on Zhang (1997). In particular:
K ¼ C1 =A
2.4. Measurement of litter and vegetation properties
St ¼ 1 BD=DP
. Kfs ¼ LG tf lnð1 þ ðD0 =ðLG þ lÞÞ
(3)
where C1 is the slope of the curve of the cumulative infiltration versus the square root of time, and A is a parameter which relates suction rate, disk radius and van Genuchten coefficients. Precipitation scenarios (n ¼ 28) were derived from the Intensity-Frequency-Duration rainfall data for the study area (Bureau of Meteorology, 2014) and precipitation intensity was calculated for 7 average recurrence intervals (1, 2, 5, 10, 20, 50 and 100 years) for each of 4 precipitation durations (5, 10, 20 and 30 min). Runoff (Q) was calculated as the difference between precipitation for each scenario and the saturated hydraulic conductivity of each habitat type (Yang and Zhang, 2011). Runoff (QCN) was also estimated through a second method, applying the SCS Runoff Curve Number Model (U.S.D.A, 1986):
Q CN ¼ 0 for P Ia
(4)
and
. Q CN ¼ ðP Ia Þ2 ðP Ia þ SÞ for P > Ia
(5)
where P is precipitation, S is the maximum soil moisture retention at the start of runoff (S ¼ 1000/CN10), Ia is the initial abstraction (Ia ¼ 0.2 S) and CN is the curve number, which is function of the type of habitat, its condition and soil hydrological group. Highcomplexity remnants were considered in the “woods” category, characterized by a CN ¼ 30, hydrological soil group A and in a “good” condition (no grazing or burning and with litter accumulating). Low- and high-complexity parks were considered as part of the “park” category in the “good” condition (grass > 75%), with CN ¼ 39 and within the hydrological soil group A.
4
A. Ossola et al. / Journal of Environmental Management 159 (2015) 1e10
Rainfall interception from the above ground vegetation was calculated as the difference between the total precipitation recorded at each tipping bucket and the total precipitation recorded at the closest Bureau of Meteorology station during the same period. Normalization of soil and litter variables was not possible. Therefore, a non-parametric KruskalleWallis test was used to compare untransformed variables among the three urban habitat types. A ManneWhitney test with Bonferroni correction was used as post-hoc test. Pearson's productemoment correlation test was used to assess correlation among variables. Data were analysed using the statistical software R (R Core Team, 2012) and its supplementary library “coin” (Hothorn et al., 2008). 3. Results 3.1. Soil properties Soils in the three habitat types had similar mean bulk density (BD) but values ranged between 0.4 and 1.6 g/cm3 (Table 1). Nevertheless, soil penetration depth (PEN) was significantly reduced in low-complexity parks, as compared to high-complexity parks and remnants (Table 1), indicating a higher soil strength of soils in low-complexity systems. Total porosity (St) was similar in low-complexity parks and high-complexity parks, and 15% greater than in high-complexity remnants. Conversely, soil microaggregates (m) were 10% more abundant in low-complexity parks and high-complexity remnants as compared to high-complexity parks (Table 1). Macroaggregate (M) abundance was lowest in low-complexity parks and greatest in high-complexity remnants (Table 1). Soil organic matter content (SOM) was significantly different in the three habitat complexity types (Table 1), even though its range of variation was relatively small (0.81e0.93%). The age since establishment did not significantly affect any of the soil properties investigated except for soil penetration depth (PEN), which increased significantly with age since establishment in high-complexity parks, but not low-complexity parks (Fig. 3).
Table 1 Soil and litter hydrological and physical parameters for high-complexity parks (HCP), high-complexity remnants (HCR) and low-complexity parks (LCP). A KruskaleWallis test was used to assess the effect of the habitat type on hydrological variables. Letters indicate significant differences in the mean following a ManneWhitney test with Bonferroni correction. Parameter Soil Kfs (cm/h)
K (cm/h)
S_WHC (%)
BD (g/cm3)
PEN (cm)
St (%)
M (%)
m (%)
SOM (%)
Litter Biomass (g/m2)
L_WHC (%)
L_WHC (l/m2)
3.2. Soil hydrology Soils in low-complexity parks had field-saturated hydraulic conductivity (Kfs) ten times lower than the two more complex habitat types (Table 1). Conversely, unsaturated hydraulic conductivity (K) was significantly lower in remnants as compared to parks, independently from their habitat complexity (Fig. 4). Soil water holding capacity (S_WHC) in low-complexity parks was 18% and 32% higher than in high-complexity parks and high-complexity remnants, respectively (Table 1). Kfs was not correlated with BD (r ¼ 0.06, n ¼ 90, p ¼ 0.55) or St (r ¼ 1.50, n ¼ 90, p ¼ 0.14), but had a significant positive correlation with PEN (r ¼ 0.69, n ¼ 80, p < 0.001) and M (r ¼ 0.57, n ¼ 30, p > 0.001) and significant negative correlation with m (r ¼ 0.38, n ¼ 30, p < 0.01). Runoff (Q), calculated by subtracting saturated hydraulic conductivity from precipitation (Yang and Zhang, 2011), was generated in low-complexity parks in the 75% of the precipitation scenarios (Appendix 1) and particularly those characterized by longer average recurrence interval (10e100 years) and shorter duration (5e20 min). Following the same method, no superficial runoff was generated in high-complexity parks and remnants since their saturated hydraulic conductivities exceeded every possible precipitation scenario for the study area (Fig. 5). The U.S.D.A (1986) curve number model predicted runoff (QCN) in only 32% of scenarios for low-complexity parks and QCN was 36% lower on average compared to Q (Appendix 1). Moreover, the curve number model predicted the generation of runoff in 18% and 32% of scenarios for high-complexity remnants and highcomplexity parks, respectively. Nevertheless, the difference
UGS
N
Mean
SEM
LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR
40 40 40 60 60 60 30 30 30 30 30 30 200 200 200 30 30 30 30 30 30 30 30 30 30 30 30
4.33a 43.95b 49.01b 2.60c 2.78c 2.07d 68.62e 56.58f 46.45g 0.96 0.94 1.00 7.28h 29.00i 42.00j 57.35k 49.94l 54.98k 61.50m 72.34n 79.10o 27.03p 16.67q 25.08p 0.81r 0.86s 0.93t
0.48 4.71 4.41 0.30 0.45 0.71 5.12 5.51 4.84 0.05 0.05 0.05 0.25 1.01 0.98 1.67 1.72 1.81 1.98 1.86 2.89 1.79 2.73 1.74 0.07 0.08 0.01
LCP HCP HCR LCP HCP HCR LCP HCP HCR
30 30 30 30 30 30 30 30 30
518.53u 1282.8v 885.73w 118.84 104.45 116.04 0.59x 1.33y 1.00z
88.16 108.87 54.03 7.75 4.64 4.78 0.09 0.11 0.31
X2
df
p
73.84
2
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Research article
Habitat complexity influences fine scale hydrological processes and the incidence of stormwater runoff in managed urban ecosystems Alessandro Ossola a, *, Amy Kristin Hahs b, Stephen John Livesley a a
Faculty of Science, 500 Yarra Boulevard, The University of Melbourne, Richmond, 3121, VIC, Australia Australian Research Centre for Urban Ecology, Royal Botanic Gardens Melbourne, c/o School of BioSciences, The University of Melbourne, 3010, VIC, Australia
b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 January 2015 Received in revised form 24 April 2015 Accepted 4 May 2015 Available online xxx
Urban ecosystems have traditionally been considered to be pervious features of our cities. Their hydrological properties have largely been investigated at the landscape scale and in comparison with other urban land use types. However, hydrological properties can vary at smaller scales depending upon changes in soil, surface litter and vegetation components. Management practices can directly and indirectly affect each of these components and the overall habitat complexity, ultimately affecting hydrological processes. This study aims to investigate the influence that habitat components and habitat complexity have upon key hydrological processes and the implications for urban habitat management. Using a network of urban parks and remnant nature reserves in Melbourne, Australia, replicate plots representing three types of habitat complexity were established: low-complexity parks, high-complexity parks, and high-complexity remnants. Saturated soil hydraulic conductivity in low-complexity parks was an order of magnitude lower than that measured in the more complex habitat types, due to fewer soil macropores. Conversely, soil water holding capacity in low-complexity parks was significantly higher compared to the two more complex habitat types. Low-complexity parks would generate runoff during modest precipitation events, whereas high-complexity parks and remnants would be able to absorb the vast majority of rainfall events without generating runoff. Litter layers on the soil surface would absorb most of precipitation events in high-complexity parks and high-complexity remnants. To minimize the incidence of stormwater runoff from urban ecosystems, land managers could incrementally increase the complexity of habitat patches, by increasing canopy density and volume, preserving surface litter and maintaining soil macropore structure. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Habitat complexity Management Urban hydrology Soil properties Litter Vegetation
1. Introduction Urban ecosystems provide numerous environmental, socioeconomical and ecological benefits to cities and towns (Bolund and Hunhammar, 1999; Chiesura, 2004; McPherson et al., 1997). As extreme precipitation events are likely to increase with climate change (Yilmaz et al., 2014), the effective management of hydrological processes and related ecosystem services such as stormwater drainage, runoff mitigation, soil water storage and purification is becoming increasingly important to create more sustainable and resilient cities and towns (Bolund and Hunhammar, 1999; Cunningham et al., 2010; Pauleit and Duhme, 2000; Nouri et al., 2013).
* Corresponding author. E-mail addresses: [email protected] (A. Ossola), hahsa@unimelb. edu.au (A.K. Hahs), [email protected] (S.J. Livesley). http://dx.doi.org/10.1016/j.jenvman.2015.05.002 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
The traditional approach in evaluating hydrological processes and benefits within urban areas has focused on mapping and modelling the hydrology of urban land cover types at large landscape scales, such as the city or catchment scale (Pauleit and € mana and Gill, 2014; Duhme, 2000; Perry and Nawaz, 2008; Sjo Tratalos et al., 2007; Whitford et al., 2001). These models have then been applied to estimate changes in hydrological processes under climate change scenarios (Gill et al., 2007) or calculate economical benefits associated with runoff reduction (Zhang et al., 2012). However, these models can be problematic as soil hydrological properties are likely to vary with soil physical properties which are highly variable at a very fine scale (Pickett and Cadenasso, 2009). Currently, very few studies have investigated the variability of soil hydrological properties in urban ecosystems through empirical measurements (e.g. Gregory et al., 2006; Yang and Zhang, 2011).
2
A. Ossola et al. / Journal of Environmental Management 159 (2015) 1e10
Together with soils, other habitat components, such as litter and vegetation layers, are likely to exert significant effects on the hydrology of urban ecosystems (Nouri et al., 2013). The absolute amount of these habitat components (e.g. volume, surface area) also define the overall complexity of habitat patches (McCoy and Bell, 1990; Byrne, 2007), and may influence the presence of organisms, such as plant and invertebrates, able to affect hydrological processes at a fine scale (Bartens et al., 2008; Colloff et al., 2010). The hydrological impact of vegetation has been previously assessed in a few urban ecosystem studies. For example, throughfall and stemflow have been measured in some urban forests and under isolated trees and related to the complexity of vegetation layers (e.g. GuevaraEscobar et al., 2007; Inkil€ ainen et al., 2013; Livesley et al., 2014). Nevertheless, to date there have been no studies specifically investigating the hydrology of litter layers in urban ecosystems. Management practices (or lack of) can alter directly or indirectly each of these system components, determining changes in the overall complexity of urban habitats (Byrne, 2007; Gaston et al., 2013), and therefore directly impacting on the local hydrological processes (Fig.1). Building on this bi-directional interaction, this study aims to holistically investigate the effects of management-driven differences in habitat complexity upon hydrological properties for each habitat component (soil, litter and vegetation). We addressed the following research questions: How do soil physical and hydrological properties vary in urban habitats characterized by different levels of habitat complexity? What is the hydrological role of each habitat component (soil, litter, vegetation)? How can our findings inform ecologically-sensitive urban habitat management to optimize urban water conservation and retention? 2. Materials and methods 2.1. Study area The study was conducted in the south-eastern Melbourne metropolitan area, Australia. The geology of the study area was restricted to quaternary and tertiary sandstones and therefore sandy soils, such as Podosols and Tenosols (sand > 89%). The study area was also confined to a single ecological vegetation class, herb-rich, heathy
Management
Habitat complexity Vegetation
Litter
Interception Throughfall
Water storage Litter input
Evapotranspiration
Evaporation Interception
Soil
woodland (The State of Victoria, Department of Environment and Primary Industries, 2013), so that differences in soil properties amongst habitat types were brought about through management practices since the suburbs were established. The altitude is between 15 and 50 m a.s.l., the annual mean maximum and minimum temperatures are 19.7 and 10.1 C, while the average annual precipitation (1950e2014) is 711 mm (Bureau of Meteorology, 2014). We selected 3 habitat types based upon their structural complexity and management history (Fig. 2), and established 10 research plots (20 30 m) within each type. Plots were selected in flat areas to minimize soil erosion and transport due to superficial runoff. Low-complexity parks (LCP) were characterized by an overstory of various Eucalyptus species and a simplified herbaceous understory. Plots were consistently mown since the establishment of the park and the herbaceous ground layer remained 250 mm) and microaggregates (m, 250 53 mm), as the distinction between large and small macroaggregates is not informative for sandy soils (J. Six, personal communication, September 30, 2013). Soil particle density (DP) was measured using a pycnometer (Carter and Gregorich, 2007) and soil organic matter content (SOM) using loss on ignition (550 C, 3 h). Soil water holding capacity at 24 h (S_WHC) was calculated in triplicates for soil sampled (0e5 cm) from each plot (Wilke, 2005).
Surface litter mass was collected in April 2013 from three squares (50 50 cm) per plot and the material was oven dried (105 C) for 48 h before weighing. Litter was sieved (1 cm) to determine fine and coarse fractions. Leaf litter water holding capacity (L_WHC) was measured following Naeth et al. (1991). Briefly, the weighed oven dry litter was soaked in a bucket overnight, allowed to drain for 24 h undercover to prevent evaporation and then reweighed to determine L_WHC. Rainfall interception was measured over six months using a Davis™ tipping bucket (Dataflow Systems, Christchurch, New Zealand) anchored to the soil surface in 7 randomly selected plots, due to limitations in the number of instruments available. Baseline precipitation data were downloaded from the Bureau of Meteorology (2014) website for the stations n. 86,210 and 86,020. The sky view factor (0 ¼ no visible sky, 1 ¼ completely visible sky) was used as a proxy for the structural complexity of vegetation and it was measured once taking a hemispherical picture over each of the tipping buckets during summer.
2.5. Data analysis Soil total porosity (St) was calculated using the Equation (1):
(1)
where BD is the bulk density and DP is the particle density (Carter and Gregorich, 2007). The field-saturated hydraulic conductivity (Kfs) was calculated using the model proposed by Nimmo et al. (2009):
(2)
where LG is ring-installation scaling length, tf is the final time, D0 is the initial ponded depth and the macroscopic capillary length l of the soil is 0.08. The model corrects mathematically for nonconstant falling head and subsurface lateral spreading and it does not require measurement of the water status of soils (Nimmo et al., 2009). Unsaturated hydraulic conductivity (K) was modelled according to the MiniDisk manufacturer specifications based on Zhang (1997). In particular:
K ¼ C1 =A
2.4. Measurement of litter and vegetation properties
St ¼ 1 BD=DP
. Kfs ¼ LG tf lnð1 þ ðD0 =ðLG þ lÞÞ
(3)
where C1 is the slope of the curve of the cumulative infiltration versus the square root of time, and A is a parameter which relates suction rate, disk radius and van Genuchten coefficients. Precipitation scenarios (n ¼ 28) were derived from the Intensity-Frequency-Duration rainfall data for the study area (Bureau of Meteorology, 2014) and precipitation intensity was calculated for 7 average recurrence intervals (1, 2, 5, 10, 20, 50 and 100 years) for each of 4 precipitation durations (5, 10, 20 and 30 min). Runoff (Q) was calculated as the difference between precipitation for each scenario and the saturated hydraulic conductivity of each habitat type (Yang and Zhang, 2011). Runoff (QCN) was also estimated through a second method, applying the SCS Runoff Curve Number Model (U.S.D.A, 1986):
Q CN ¼ 0 for P Ia
(4)
and
. Q CN ¼ ðP Ia Þ2 ðP Ia þ SÞ for P > Ia
(5)
where P is precipitation, S is the maximum soil moisture retention at the start of runoff (S ¼ 1000/CN10), Ia is the initial abstraction (Ia ¼ 0.2 S) and CN is the curve number, which is function of the type of habitat, its condition and soil hydrological group. Highcomplexity remnants were considered in the “woods” category, characterized by a CN ¼ 30, hydrological soil group A and in a “good” condition (no grazing or burning and with litter accumulating). Low- and high-complexity parks were considered as part of the “park” category in the “good” condition (grass > 75%), with CN ¼ 39 and within the hydrological soil group A.
4
A. Ossola et al. / Journal of Environmental Management 159 (2015) 1e10
Rainfall interception from the above ground vegetation was calculated as the difference between the total precipitation recorded at each tipping bucket and the total precipitation recorded at the closest Bureau of Meteorology station during the same period. Normalization of soil and litter variables was not possible. Therefore, a non-parametric KruskalleWallis test was used to compare untransformed variables among the three urban habitat types. A ManneWhitney test with Bonferroni correction was used as post-hoc test. Pearson's productemoment correlation test was used to assess correlation among variables. Data were analysed using the statistical software R (R Core Team, 2012) and its supplementary library “coin” (Hothorn et al., 2008). 3. Results 3.1. Soil properties Soils in the three habitat types had similar mean bulk density (BD) but values ranged between 0.4 and 1.6 g/cm3 (Table 1). Nevertheless, soil penetration depth (PEN) was significantly reduced in low-complexity parks, as compared to high-complexity parks and remnants (Table 1), indicating a higher soil strength of soils in low-complexity systems. Total porosity (St) was similar in low-complexity parks and high-complexity parks, and 15% greater than in high-complexity remnants. Conversely, soil microaggregates (m) were 10% more abundant in low-complexity parks and high-complexity remnants as compared to high-complexity parks (Table 1). Macroaggregate (M) abundance was lowest in low-complexity parks and greatest in high-complexity remnants (Table 1). Soil organic matter content (SOM) was significantly different in the three habitat complexity types (Table 1), even though its range of variation was relatively small (0.81e0.93%). The age since establishment did not significantly affect any of the soil properties investigated except for soil penetration depth (PEN), which increased significantly with age since establishment in high-complexity parks, but not low-complexity parks (Fig. 3).
Table 1 Soil and litter hydrological and physical parameters for high-complexity parks (HCP), high-complexity remnants (HCR) and low-complexity parks (LCP). A KruskaleWallis test was used to assess the effect of the habitat type on hydrological variables. Letters indicate significant differences in the mean following a ManneWhitney test with Bonferroni correction. Parameter Soil Kfs (cm/h)
K (cm/h)
S_WHC (%)
BD (g/cm3)
PEN (cm)
St (%)
M (%)
m (%)
SOM (%)
Litter Biomass (g/m2)
L_WHC (%)
L_WHC (l/m2)
3.2. Soil hydrology Soils in low-complexity parks had field-saturated hydraulic conductivity (Kfs) ten times lower than the two more complex habitat types (Table 1). Conversely, unsaturated hydraulic conductivity (K) was significantly lower in remnants as compared to parks, independently from their habitat complexity (Fig. 4). Soil water holding capacity (S_WHC) in low-complexity parks was 18% and 32% higher than in high-complexity parks and high-complexity remnants, respectively (Table 1). Kfs was not correlated with BD (r ¼ 0.06, n ¼ 90, p ¼ 0.55) or St (r ¼ 1.50, n ¼ 90, p ¼ 0.14), but had a significant positive correlation with PEN (r ¼ 0.69, n ¼ 80, p < 0.001) and M (r ¼ 0.57, n ¼ 30, p > 0.001) and significant negative correlation with m (r ¼ 0.38, n ¼ 30, p < 0.01). Runoff (Q), calculated by subtracting saturated hydraulic conductivity from precipitation (Yang and Zhang, 2011), was generated in low-complexity parks in the 75% of the precipitation scenarios (Appendix 1) and particularly those characterized by longer average recurrence interval (10e100 years) and shorter duration (5e20 min). Following the same method, no superficial runoff was generated in high-complexity parks and remnants since their saturated hydraulic conductivities exceeded every possible precipitation scenario for the study area (Fig. 5). The U.S.D.A (1986) curve number model predicted runoff (QCN) in only 32% of scenarios for low-complexity parks and QCN was 36% lower on average compared to Q (Appendix 1). Moreover, the curve number model predicted the generation of runoff in 18% and 32% of scenarios for high-complexity remnants and highcomplexity parks, respectively. Nevertheless, the difference
UGS
N
Mean
SEM
LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR LCP HCP HCR
40 40 40 60 60 60 30 30 30 30 30 30 200 200 200 30 30 30 30 30 30 30 30 30 30 30 30
4.33a 43.95b 49.01b 2.60c 2.78c 2.07d 68.62e 56.58f 46.45g 0.96 0.94 1.00 7.28h 29.00i 42.00j 57.35k 49.94l 54.98k 61.50m 72.34n 79.10o 27.03p 16.67q 25.08p 0.81r 0.86s 0.93t
0.48 4.71 4.41 0.30 0.45 0.71 5.12 5.51 4.84 0.05 0.05 0.05 0.25 1.01 0.98 1.67 1.72 1.81 1.98 1.86 2.89 1.79 2.73 1.74 0.07 0.08 0.01
LCP HCP HCR LCP HCP HCR LCP HCP HCR
30 30 30 30 30 30 30 30 30
518.53u 1282.8v 885.73w 118.84 104.45 116.04 0.59x 1.33y 1.00z
88.16 108.87 54.03 7.75 4.64 4.78 0.09 0.11 0.31
X2
df
p
73.84
2
