Interactive comment on “ Rapid revegetation by sowing seed mixtures of shrub and herbaceous species ”

Abstract. Fast revegetation by means of sowing seed mixtures of shrub and herbaceous species is a measure to prevent bare soils from wind and water erosion. A field experiment was used to test the effect of species selection and the ratio of shrub to herbaceous species on vegetation formation and shrub growth. Results showed that herbaceous species hastened cover formation and maintained a high coverage for a longer period. However, the growth of shrubs was hindered. In the North China Plain or where the soil and climate are similar, the ratio of shrub to herbaceous seeds is proposed to be 6 : 4–7 : 3 (weight ratio). Among the herbaceous species tested, Festuca arundinacea Schreb. grows relatively slow, so it should be mixed with other fast-growing species in the practice of rapid revegetation, and a seeding density lower than 6 g m−2 is proposed when applied; Orychophragmus violaceus O. E. Schulz. wilts when the seeds are ripe, leading to a significant decrease of coverage, so other species with different phenology should be involved when it is applied; Viola philippica Car. is a good ground cover plant which grows fast and maintains a stable coverage from July to October, and a seeding density of 1.5 g m−2 is proposed for rapid revegetation. Herbaceous species have different traits. Three different types of herbs were found in our experiment: slow-growing stable species (F. arundinacea), fast-growing unstable species (O. violaceus) and fast-growing stable species (V. philippica). Shrubs, slow-growing stable species and fast-growing unstable species should not be used alone because they cannot cover the ground fast or they cannot maintain a long period of good coverage. A small seeding rate of fast-growing stable species should be used to ensure a fair coverage against erosion. Because natural environmental conditions are heterogeneous and stochastic, more species should be added to enhance the stability of plant community.


Introduction
Development and construction projects often cause damage to native vegetation.In abandoned quarries or surface mines, recolonization of plants is very difficult (Ballesteros et al., 2012) because of the destruction of natural soil structure and seed bank, as well as the limitation of nutrition and water (Jim, 2001;Haritash et al., 2007).Even though technical restoration can accelerate succession, it takes decades to achieve a complex self-sustainable ecosystem (Zhang et al., 2013).During the succession, wind or water erosion may occur when the vegetation coverage is still low, further decreasing soil nutrient (Zuazo and Pleguezuelo, 2008) and thus hindering the process of revegetation (Wang et al., 2005).Geologic hazards may also happen if no protective measures are applied (Robbins et al., 2013).Heavy metals in mineral waste may be transported by wind force and cause soil pollution (Brotons et al., 2010).Besides, during construction of roads and buildings, temporary dumps without covering may be eroded, resulting in soil loss.
Soil covering is a useful measure to protect soil from wind and water erosion (Mu, 2010), where vegetation plays an important role (Sterk, 2003).The risk and intensity of wind and water erosion decrease with increased vegetation cover (Cai, 2001;Maurer et al., 2009;Kefi et al., 2011;Houyou et al., 2014).Plants increase soil surface roughness, decrease wind speed and as a result the erosivity and erodibility decrease (Borrelli et al., 2014).Plants increase concentration time during rainfall events and increase infiltration, so less runoff is produced.
Different types of vegetation respond differently to wind and water erosion.Trees with large canopy are more effective in reducing wind speed, whereas shrubs are more effective in trapping transporting materials (Leenders et al., 2007).Compared to herbaceous species, shrubs have more developed root systems to improve soil structure and conserve water in deep layers, resulting in a better effect on soil and water conservation (Huang et al., 2006;Wei et al., 2009), and its effect is less affected by rain intensity compared to herbs (Zhang et al., 2014).Trees develop slowly (Ji et al., 2011) and have limited effect on soil protection during the early stage of development (Zhang and Shao, 2003), while herbs germinate and grow fast, rapidly covering the ground to prevent splash erosion and decrease runoff (Franklin et al., 2012).
Seed mixtures of shrubs and grasses take the advantage of both taxons, but the competition for light, water and nutrition may affect vegetation cover formation (Milton and Dean, 1995) and thus the effect of soil protection.As shown by some research, the competition from grasses might cause severe growth decline of woody plants, especially during their early stage of development (Gordon et al., 1989;Denslow et al., 2006).Because the interaction between woody species and herbaceous species is complicated, it was proposed that traits such as niche breadth and competitiveness for different resources of different species should be thoroughly studied, and the selection of species should be based on environmental condition including soil, water and light (Heneghan et al., 2008;Abe et al., 2015;Oliveira et al., 2014).By means of species selection and controlling seeding density, a positive effect can be attained for shrub establishment (Franklin et al., 2012).
In this research, a measure of fast revegetation by means of sowing seed mixtures of shrub and herbaceous species was tested using field experiment.We focused on (1) which seed mixtures of shrub and grass (which species and what proportion) could provide a fair or good coverage for a long period; (2) how different proportions of species affect the speed of cover formation and the stability of coverage (specifically, we tested the effect of the ratio of shrub to herbaceous seeds, R s/h ); and (3) how different herbaceous species affect the growth of shrub.Based on our research, advice will be proposed on species selection and determination of their proportion in seed mixtures during the practice of revegetation in (1) plains where wind erosion occurs, (2) gentle slopes where water erosion occurs and plant growth is not significantly affected by the slope and (3) seriously degraded sites such as abandoned mines where measures such as topsoil covering have been applied to improve soil quality.

Study area
The research was conducted in the Ecological Restoration Research Base of Environmental Protection Research Institute of Light Industry (EPRILI), located in Changping County, Beijing (40 • 9 56.73 N, 116 • 9 1.04 E, 57 m a.s.l.).Beijing has a continental monsoon climate with a rainy season from June to September.The mean annual precipitation is 620 mm (historical data).Monthly precipitation and average temperature during the experimental period were measured using Davis Vantage Pro2 Weather Station, and the data are shown in Fig. 1.
The study area was a flatland.The local soil used for the experiment was sandy loam.The pH value was 7.44.The chemical properties of the soil are shown in Table 1 (Liang, 2013).

Experimental design
Four native species were studied, including a shrub species, Amorpha fruticosa L., and three herbaceous species, Festuca arundinacea Schreb., Viola philippica Car. and Orychophragmus violaceus O. E. Schulz.These species are commonly seen in the North China Plain, and former research has shown their tolerance against water or nutrient deficiency.The characters of target species are shown in Table 2 and the designs of seed mixtures are shown in Table 3.
Every design of the seed mixtures was tested in a 4.5 m long, 1.3 m wide plot, so there were altogether 41 plots.Seed mixtures with a seeding density of 15 g m −2 were manually sowed without fertilizer in May 2013.Non-woven fabrics (a planar, permeable, polymeric textile material) were used as soil cover to protect the seeds from erosion and enhance humidity.Irrigation was applied until mid-June, after which precipitation became the only water source for plants.The design of T3 was the same with T1.T4 A. fruticosa : V. philippica The design of T4 was the same with T1.T0 A. fruticosa Note: the thousand-grain weight of A. fruticosa, F. arundinacea, O. violaceus and V. philippica are 6.163, 2.814, 2.175 and 0.981 g, respectively.Each plot is denoted as Txty.The capital letter T indicates species, and the following x ranges from 0 to 4, indicating different combinations of species.The small letter t indicates the proportion of shrub seeds, and the following y ranges from 0 to 9, indicating the percentage of A. fruticosa in the seed mixtures by mass, which equals y/10.Data of T2t0 are deleted because of deficient setting of the experimental plot.

Data collection and analysis
From July to October 2013, three 1 × 1 m 2 sample plots were randomly taken in each plot three times a month to measure the coverage of A. fruticosa and the total coverage of all species.Invaded native species were recorded.The same measurement was also made from April to August 2014.In this study, a coverage of 60 % was assumed to be fair, and a coverage of 80 % was assumed to be good, because erosion risk was low in slopes under 30 • with a fractional vegetation cover of 60-80 % based on an erosion model (Vrieling et al., 2006).The duration of total coverage higher than 60 and 80 % were calculated by the following equation.

Duration of fair or good coverage =
the number of days when total coverage ≥ 60 or 80% the number of days for coverage measurement × 120 days (1) Coefficient of variation (CV) of total coverage during the experimental period was calculated to describe the stability of total vegetation coverage.Each CV value of different R s/h was taken as a sample, and the Friedman test for nonparametric paired samples was used to test the significance of variation between the CV values of different combinations of shrub and herbaceous species.
At the end of October 2013, 15 individuals of A. fruticosa in each plot were randomly taken to measure height and ground diameter.ANOVA was used to test the effect of herbaceous species and R s/h on the growth of A. fruticosa where T0 was used as control.Normality of samples was tested before the significance test, and when the effect was significant (P < 0.05) Least Significant Difference Test (LSD) was used to test comparisons among different seed mixture designs.Statistic analysis was performed using SPSS program.

The effect of species on total coverage
As shown in Fig. 2, from July to October 2013, T4 had the highest total coverage, regardless of the ratio of shrub to herbaceous seeds.The performance of other seed mixtures differed with time.In July, when R s/h s were 1 : 9, 3 : 7, 4 : 6 and 5 : 5, T2 had the second highest total coverage, and when R s/h s were 2 : 8, 6 : 4, 8 : 2 and 9 : 1, T3 had the second highest total coverage.T1 had the lowest total coverage in July.From August to October, when R s/h s were 1 : 9-3 : 7, T2 had a higher total coverage than T1, and when R s/h s were 6 : 4-9 : 1, T1 had a higher total coverage value than T2.T3 had a relatively low total coverage from August to October, which was also shown in T2t8 and T2t9.

The effect of R s/h on total coverage
As shown in Fig. 3, R s/h had different effects on the dynamics of total coverage in different species combination.
1. T0: A. fruticosa took a longer time to form a fair coverage and maintained a fair or good coverage for a much shorter period compared to herbaceous species.Total   coverage of T0 was lower than 60 % in July and October but higher than 80 % in August and September.
2. T1: in July, 6 out of 10 plots including t2 and t5-t9 had a total coverage higher than 60 %, among which t7 had a total coverage higher than 80 %.In August, except for t0 and t1, all plots had a total coverage higher than 80 %.From September onward, all plots had a total coverage higher than 80 %.
3. T2: in July, 7 out of 9 plots including t1 and t3-t8 had a total coverage higher than 60 %, among which t3 and t5 had a total coverage higher than 80 %.In August, except for t9, all plots had a total coverage higher than 80 %.From September onward, t1-t7 had a total coverage higher than 80 %.The total coverage of t8 and t9 was good in September, but both fell to 77 % in October.4. T3: in July, 9 out of 10 plots including t0 and t2-t9 had a total coverage higher than 60 %, among which t6 had a total coverage higher than 80 %.In August, all plots had a total coverage higher than 60 %, among which t0 and t6-t9 had a total coverage higher than 80 %.The total coverage of most plots was maintained until October except for t2, which enhanced total coverage from September, t1 and t3-t5, which enhanced total coverage in October, and t9, which decreased total coverage to a value lower than 60 % in October.
5. T4: since July, all plots achieved a total coverage higher than 80 %.

Duration and stability of total coverage
From July to October (counted as 120 days), duration of fair coverage was 76, 107, 112, 112 and 120 days (mean values of different R s/h s, the same below) from T0 to T4, respectively.Duration of good coverage was 65, 84, 95, 82 and 109 days from T0 to T4, respectively.In this respect, T4 had the best performance, followed by T2, T3, T1 and T0.T1 and T3 had relatively poor performance compared to T4 and T2, but T1t5, T1t7, T3t7 and T3t8 maintained a good coverage of more than 100 days.Even though T2 had the second best performance in general, T2t2 and T2t9 maintained a shorter period of fair or good coverage compared to T1 or T3 of the same R s/h .Remarkably, when R s/h was 6 : 4 and 7 : 3, all combinations of shrub and herbaceous seeds maintained a fair coverage for 120 days, i.e., the whole experimental period.As a result, this ratio of shrub to herbaceous seeds is proposed for seed mixtures applied in rapid revegetation.T0 not only had the shortest duration of fair or good coverage but also had the highest coefficient of variation (46 %), indicating that it had the least stability among all plots.The coefficient of variation from T1 to T4 were 19, 15, 19 and 9 %, respectively.The coefficient of variation of T4 was significantly lower than those of T1, T2 and T3 (P < 0.05).In T1, plots with a R s/h of 5 : 5-9 : 1 had a relatively low coefficient of variation, ranging from 10 to 16 %.In T2, plots with a R s/h of 1 : 9 and 3 : 7-8 : 2 had a relatively low coefficient of variation, ranging from 10 to 15 %.In T3, plots with a R s/h of 2 : 8, 3 : 7, 6 : 4-8 : 2 had a relatively low coefficient of variation, ranging from 11 to 15 %.

The effect of herbaceous species on the coverage of A. fruticosa
The average coverage of T0 during the experiment period was 74.6 %.Assuming the coverage was proportional to the amount of seeds we sowed, some seed mixtures had a positive effect on the coverage of A. fruticosa, including T1t1-T1t7, T3t8-T3t9 and T4t7-T4t9, while other seed mixtures had a negative effect on the coverage of A. fruticosa, as shown in Fig. 4. When A. fruticosa was sowed alone, fair coverage was achieved on 30 July.When herbaceous species were sowed with A. fruticosa with a R s/h range of 1 : 9 to 3 : 7, the coverage of A. fruticosa was lower than 60 % during the whole experimental period in any combination of seed mixtures.When the R s/h ranged from 4 : 6 to 7 : 3, the coverage of A. fruticosa in T1 reached 60 % first, on 30 August, 20 July, 30 July and 10 August.When the R s/h ranged from 8 : 2 to 9 : 1, the coverage of A. fruticosa in T4 reached 60 % first, on 20 July.In plots of T1t9, T2t7-T2t9, T3t6-T3t9 and T4t6-T4t7, A. fruticosa also achieved a coverage of 60 % but at a later period of the rainy season.

The effect of R s/h and herbaceous species on the growth of A. fruticosa
There was a negative effect of herbaceous species on the growth of A. fruticosa, as shown in Table 4.In T1, height growth of A. fruticosa was significantly lowered when R s/h s were 1 : 9-3 : 7 and 6 : 4-8 : 2, while ground diameter was significantly lowered when R s/h s were 1 : 9, 2 : 8 and 7 : 3 compared to T0.In T2, height and diameter growth of A. fruticosa were significantly decreased in all R s/h s compared to T0.In T3, height growth of A. fruticosa was significantly lower than T0 in all R s/h s, while ground diameter was significantly lower than T0 when R s/h s were 2 : 8-6 : 4 and 9 : 1.In T4, height growth of A. fruticosa was significantly lowered when R s/h s were 3 : 7-5 : 5 and 7 : 3, while ground diameter was significantly lowered when R s/h was 4 : 6, compared to T0.
When different combinations of species with the same R s/h were compared (T1-T4), the values of height and ground diameter were the highest in T3 when the R s/h s were 1 : 9 and 2 : 8.When the R s/h s were 3 : 7-5 : 5, T1 had highest value of height and generally the highest value of ground diameter.When the R s/h s were 6 : 4-9 : 1, T4 had the highest values of height and ground diameter.

Dynamics of the established plant communities in the subsequent year
T1 was damaged in 2014 so the data are not reported.As to other plots, in April 2014, T2t1-T2t5, T3t0-T3t8 and T4t0-T4t5 had a total coverage higher than 80 %, and T3t9 had a total coverage higher than 60 %.In May, except for T2t6 and  T3t2-T3t5, all plots had a total coverage higher than 80 %.
In June, the total coverage of T2t1, T2t5 and T3t0-T3t5 decreased and was lower than 60 % because of the wilting of O. Violaceus since April.Since May, T0 and all plots of T4 achieved a total coverage higher than 80 %.Since July, all plots of T2 achieved a total coverage higher than 80 %.In August, T3t0-T3t1 had a total coverage ranging from 70 to 73 %, while the other plots of T3 achieved a total coverage higher than 80 %.

The effect of species selection and R s/h on total coverage
Vegetation cover is one of the main factor controlling the effect of soil protection from wind and water erosion (Ferreira and Panagopoulos, 2014).An early recovery of vegetation cover can prevent water erosion during the rainy season, while the stubble and litters can prevent wind erosion during the following dry season.Two months after sowing, total coverage of T0, T1t0, T1t1, T1t3, T1t4, T2t2, T2t9 and T3t1 were lower than 60 %, so they are not proposed for rapid revegetation.
Based on the speed and the stability of coverage, sowing seed mixtures performed better than sowing shrubs alone, which was consistent with Gilardelli et al. (2013).Among the combinations of shrub and herbaceous species, T4 showed its excellency in fast ground cover formation and high coverage maintenance around the whole experimental period, most attributed to V. philippica.According to our results, sowing V. philippica with a seeding rate of 1.5 g m −2 is efficient in rapid revegetation in northern China or regions where the climate and soils are similar.A higher seeding rate may be a waste of seeds and more seriously, the dense ground cover may hinder the recolonization of other native species.In plots where O. violaceus instead of V. philippica was sowed with A. fruticosa, a coexistence with local annual or perennial herbs such as Bidens pilosa L., Acalypha australis L., Amaranthus retroflexus L., Euphorbia humifusa Willd., Abutilon theophrasti Medic., Artemisia annua L., Convolvulus arvensis L. and Polygonum lapathifolium L. was observed but not in T4.
Other than T4, T2 had a fast cover formation when R s/h was low, and T3 had a fast cover formation when R s/h was high.Sowing F. arundinacea alone was not appropriate for rapid revegetation because it covered the ground slowly.But considering the whole experimental period, T1 had a relatively high total coverage when R s/h was high.As a result, F. arundinacea should be mixed with other fast-growing species and a seeding rate of 1.5-6.0g m −2 is proposed in order to achieve high value of total coverage.T3, i.e., O. violaceus, covered soil rapidly but had the lowest total coverage considering the whole experimental period.Because O. violaceus wilted when the seeds were ripe, a significant decline of O. violaceus was observed, though total coverage was hardly affected thanks to the development of A. fruticosa.
Before the experiment, we supposed that the stability of total coverage was correlated with the tolerance to environmental stress.For example, because of the stochastic nature of precipitation, wilting, defoliation or die off during water deficiency may weaken the protective effect of vegetation when a rain storm finally occurs (Zuazo and Pleguezuelo, 2008).Compared to herbs, woody species were supposed to maintain a more stable coverage because they could use the resource in the deep soil layers or at least they have longer life (Wang et al., 2005).Contrary to expectation, results showed that T0 had the highest coefficient of variation among all plots.If A. fruticosa could use the water stored in the deep layers, its coverage would not fluctuate in spite of the temporal water deficiency (the longest interval between rainfall events was 17 days during our experiment), and thus the coefficient of variation would be small.The high value of coefficient of variation in T0 indicated that the ability to conserve water and the resistance against environmental stress was not fully developed in A. fruticosa.
Some plots, including T2t2, T2t9 and T3t1, had a low total coverage, and no pattern was observed between these plots and the adjacent plots.It was supposed that random factors such as the variation of seeds and microsite conditions accounted for the poor performance of these plots.However, natural ecosystems are much more diverse than our study plots.Microsites are spatially heterogeneous, weather events are stochastic by nature and the inter-or intraspecific relationship may vary in different stages of individual development and community succession (Zanini et al., 2006).To deal with the spatial and temporal heterogeneity, more species should be used in artificial revegetation because of their adaptation to different niches and thus the reconstruction of the whole plant community is more likely to succeed even if some species fail (Sheley and Half, 2006).

The effect of herbaceous species on the growth of shrub
The coverage of A. fruticosa in T1t5, T4t8 and T4t9 reached 60 % 10 days earlier than T0, even though fewer seeds were sowed in these plots, indicating a positive effect of herbaceous species on the coverage of A. fruticosa.Compared to T0, the average coverage of A. fruticosa during the study were higher in T1t1-T1t7, T3t8-T3t9 and T4t7-T4t9, but average height and ground diameter were lower in these plots, indicating that the individuals were smaller, but the number of individuals was higher when herbaceous species were sowed together.The result was consistent with the research by Mason et al. (2013), which showed that ground cover was favorable for shrub germination but disadvantageous to growth.Moreover, when a field study was made in May 2014, it was observed that the stem number of each individual of A. fruticosa ranged from three to five in T0, but more than six in T1t9 and T4t9, which may partly account for the inconsistency between high coverage and low growth in these plots.Competition for resources, such as water, may explain the decline of growth of A. fruticosa.Soil water content is determined by the input, such as precipitation and irrigation, together with the output, such as infiltration and evapotranspiration.Plants can increase infiltration rate (Ji et al., 2008) and water holding capacity but also consume a large amount of water during transpiration.As a result, soil water content may be increased or decreased by coexisting species (Bréda et al., 1995, D'Odorico et al., 2007).Competition may also exist for nutrition or light, but the relationship differs among different species (Denslow et al., 2006;Mendoza-Hernández et al., 2014).Research indicated a very comprehensive relationship between different coexisting species; not only negative but also positive relationships were shown in different studies (Harmer et al., 2011, Ballesteros et al., 2012;Zhang et al., 2013;Oliveira et al., 2014).
Other than interspecific competition, intraspecific competition exists.Competition for light between individuals of A. fruticosa was more intense in T0 than other plots, especially when R s/h was low.In T0, short and weak individuals may be weeded out and only the tall and strong ones which have access to light survive, leading to a higher mean value of growth.Compared to height, ground diameter was less correlated to the competition for light, so it was also less corrected to R s/h .However, this hypothesis needs to be tested.

Conclusions
Firstly, shrub cover was formed slower than ground cover and was maintained for a shorter period at least in the early stage of development.When herbaceous species were sowed with shrubs, total coverage increased and was maintained for a longer period, but the growth of shrubs was hindered.Secondly, in the practice of rapid revegetation in the North China Plain or wherever the soil and climate are similar, the ratio of shrub to herbaceous seeds is proposed to be 6 : 4-7 : 3 by mass.Thirdly, herbaceous species have different traits.In our experiment, three different types of herbs were found, i.e., slow-growing stable species (F.arundinacea), fast-growing unstable species (O.violaceus) and fast-growing stable species (V.philippica).Slow-growing stable species and fast-growing unstable species should not be used alone because they cannot cover the ground fast or they cannot maintain a long period of coverage.A small seeding rate of fast-growing stable species should be used to ensure a fair coverage against erosion, and other species with different traits should be added to enhance the stability of plant community.Fourthly, in the practice of rapid revegetation in the North China Plain or wherever the soil and climate are similar, seeding density of F. arundinacea is proposed to be lower than 6 g m −2 and the seeding density of V. philippica is proposed to be 1.5 g m −2 .

Figure 1 .
Figure 1.Monthly precipitation (bars) and average temperature (dots) during the experimental period.

Figure 2 .
Figure 2. Total coverage of different combinations of shrub and herbaceous seeds (±SE) from July to October 2013; seeTable 3 caption for code identification.

Figure 3 .
Figure 3. Dynamics of total coverage (±SE) from July to October 2013; seeTable 3 caption for code identification.

Figure 4 .
Figure 4. Coverage of A. fruticosa sowed with different herbaceous seeds.Note: each spot with an error bar is the mean value of coverage of A. fruticosa during the experimental period.The dotted line indicates a predicted coverage of A. fruticosa under different seeding densities based on the assumption that the coverage is proportional to the amount of seeds sowed.

Table 1 .
Some chemical properties of the local soil used in the field experiment.

Table 2 .
Characteristics of target species.

Table 3 .
Designs of seed mixtures.O. violaceus : V. philippica The ratio of shrub to herbs was the same with T1, and the masses of O. violaceus and V. philippica were the same.

Table 4 .
Average height and ground diameter of A. fruticosa.: the superscript * indicates a significant difference compared to T0 (P < 0.05).The subscript of the same letter or the absence of subscript indicates that the mean values of height or ground diameter in the same row were not significantly different.No A. fruticosa survived in T4t1 and T4t2, and only five and two individuals of A. fruticosa survived in T2t1 and T3t1, respectively. Note