Wind drives drought responses of green roof vegetation in two substrates

The multifunctionality and delivery of ecosystem services from green roofs is improved by biological diversity of the roof vegetation. However, the frequency and intensity of drought episodes on extensive green roofs may limit the use of non-succulent species and the potential functional and phylogenetic diversity of the vegetation. Wind accelerates water use by plants and desiccation of the green roof substrate, and may be a key factor in selection of non-succulent plant species for green roofs. In this study, we tested wind interactions with green roof substrate composition and the effects on plant and substrate water balance, overall plant performance, and wilting and survival of three non-succulent species (Plantago maritima L., Hieracium pilosella L., and Festuca rubra L.) under realistic prolonged water deficit conditions. We found that, regardless of species or substrate tested, wind accelerated drought response. Drought-stressed plants exposed to wind wilted and died earlier, mostly due to more rapid desiccation of the growth substrate (critical substrate moisture content was 6-8%). The moderate wind levels applied did not affect plant performance when not combined with drought. Species with contrasting growth forms showed similar responses to treatments, but there were some species-specific responses. This highlights the importance of including wind to increase realism when evaluating drought exposure in non-succulent green roof vegetation.


32
Green roofs are engineered ecosystems representing an effective strategy to address some of the most challenging 33 environmental issues in the urban areas [1,2]. They provide several urban ecological goods and services, including 34 thermal insulation to buildings, extension of roof lifespan, mitigation of the urban heat island, aesthetics, 35 promotion of biodiversity (space, habitats and food) and stormwater management [1,2]. This multifunctionality 36 in delivery of ecosystem services by green roofs is improved by biological diversity of the roof vegetation [3].

37
Green roofs are however harsh environments, where the negative effects of drought, temperature extremes,  The most critical factor for long-term performance of non-succulent species is the intensity and frequency of 47 drought episodes, even in cold humid regions [13]. The water balance of the substrate is determined by the input 48 through precipitation, retention (storage) and loss through evapotranspiration. Retention is determined by the 49 thickness of the substrate and the water holding capacity as affected by porosity and organic matter content, while 50 evapotranspiration is affected by environmental factors such as irradiance, temperature and wind.

51
The impact of wind on plants is extremely variable and largely depends on its speed, duration, and frequency 52 combined with plant characteristics [14,15]. The effect of wind on green roof vegetation can include acceleration 53 of substrate desiccation through evaporation, induction of mechanical and morphological responses, changes in 54 leaf microclimate, and disturbance of leaf gas and heat exchange [15,16,17,18,19]. Despite the importance of 55 wind for water loss and its effect on vegetation performance, there is a lack of knowledge about how wind affects 56 green roof vegetation either directly or through effects on the water balance.

57
We sought to address this knowledge gap by examining how wind interacts with the water holding capacity of the 58 substrate to affect water balance and performance of non-succulent vegetation under prolonged drought. We also 3 59 examined how selected leaf chlorophyll a fluorescence parameters are related to desiccation and whether these 60 parameters can be used to predict plant mortality on wind exposed green roofs.

62
Experimental approach

63
To test the interactions between wind and substrate composition and their effects on drought response in green 64 roof non-succulent species, we conducted a pot experiment under greenhouse conditions to monitor plant wilting 65 symptoms and performance. Three species, representing different growth forms, were established in two different 66 green roof substrates and exposed to simulated wind, while watering was withheld to mimic prolonged drought.

67
The experimental design was a combination of two levels of wind (moderate wind or calm conditions), two levels 68 of watering (withheld or control), and two different growth media (water-holding capacity 37% or 46%) for each 69 of the three species, giving a total of 12 treatment combinations and 240 pots. Pot was the experimental unit, with 70 4 and 16 replicates per species and wind combination for control and drought treatments, respectively.

72
The three species selected for the experiment (Plantago maritima L., Hieracium pilosella L., and Festuca rubra 73 L.) are non-succulent candidates for green roofs representing different growth forms and phylogenetic lines with 74 relatively high tolerance to abiotic stresses typical for green roofs conditions. Individual plants were established 75 from large seedlings in pots (10 cm x 10 cm x 11 cm) containing two contrasting green roof substrates designed 76 for this experiment, substrate A (more mineral, 36.7% water-holding capacity) or substrate B (more organic, 77 46.1% water-holding capacity). Substrate characteristics are given in Table 1. All plants were allowed to establish 78 to greenhouse conditions for 6 weeks before the start of the experiment, during which time they were watered to 79 approximately field capacity 2 times per week to achieve realistic drought acclimation before the experiment. The

150
Temporal changes in observed responses, measured parameters and differences between treatments were 151 compared using 95% confidence intervals.

153
Overall, the three tested species showed similar responses to wind. Moderate wind increased the rate of water loss

188
Although responses to drought were species-specific, responses to wind were surprisingly similar between species, 189 as no strong species by wind interactions were recorded (Table 2)

229
We found that wind had a strong negative effect on drought response in non-succulent candidate species for green 230 roofs, and that this effect was not modified by the water-holding capacity of the two substrates used. We expected 231 responses to wind to be slower for plants growing in the substrate holding more water, but this was not the case.

232
Under water deficit conditions, wind increased the rate of water loss from the system, leading to higher midday

243
In general, wind was the factor that influenced drought-stressed plants the most, followed by differences between 244 plant species, while substrate composition had no effect. Wind exposure accelerated wilting symptoms under 245 drought conditions by 3-8 days and mortality by 1-10 days. The reason for the faster mortality of wind-exposed 246 plants was increased water loss from both substrates. The critical substrate moisture level for all species and wind 247 treatments was 6-8% (calculated by weight). Low water content and availability in the substrates most probably 248 led to stomatal closure and decreased transpiration cooling, as indicated by the higher midday temperatures 249 recorded in wind-exposed leaves, what was recorded throughout the drought period even if measurements were 250 taken during periods when wind was turn off. The increased leaf temperature under abiotic stresses due to 251 decreased gas exchange, particularly transpiration, is well known phenomenon [22]. In contrast, other studies have 252 reported that wind (higher than 1.5 m s -1 ) usually reduces leaf temperature [23,24]  conditions must include wind effects, as otherwise the negative effect of water deficit/or drought periods may be 301 underrated. There still is a potential to design substrates with high water holding capacity using water absorbing 302 gels and similar, but our results indicated that for substrates with representative water holding capacities, this 303 effect may be marginal.

304
Another objective of this study was to test whether selected leaf chlorophyll a fluorescence parameters can be 305 used to predict plant mortality on windy exposed green roofs. We found that decreases in the value of Fv/Fm and 306 PI abs accompanied morphological changes recorded with the help of a wilting scale. Plant response to the 307 combined action of drought and wind was very rapid, and increased wilting score and declining values of 308 chlorophyll a fluorescence parameters coincided with serious, probably irreversible, changes in green roof plants.

310
We found that wind accelerated the drought response of non-succulent candidate species for green roofs and this 311 effect was not modified by higher substrate water-holding capacity (46% compared with 37%). Drought-stressed 312 plants exposed to wind wilted and died faster, mostly due to more rapid desiccation of the substrate, while wind

322
We are grateful to Monika Małecka-Przybysz for laboratory support and for technical editing of the entire 323 manuscript.