Effect of tree canopy on species composition, diversity and biomass of herbaceous vegetation and soil characteristics in semi-arid forests of the

Background: Herbaceous vegetation is of great ecological importance and responds quickly to environmental changes. Present study was aimed to understand the variation in herbaceous species composition, diversity, biomass and soil physicochemical properties in canopy gaps and under the canopy of semi-arid forests in the Aravalli hills. Methods: Four permanent plots consisting of canopy gaps and under canopy were established. To study herbs phytosociological attributes and biomass, 10 quadrats of 0.5 x 0.5 m size were laid in each plot. For soil physico-chemical properties, 5 soil samples were collected randomly at 0-10 cm depth from each plot. Relationship between biodiversity indices, herbs biomass and soil physicochemical properties was determined using Redundancy Analysis (RDA) and correlation analyses. Results: The sites differed in terms of herbaceous species composition, diversity, biomass and soil physico-chemical properties. Canopy gaps were dominated by grasses while forbs were higher under the canopy. Across the sites, poaceae was the dominant family. All species showed contagious distribution pattern. Oplismenus burmannii reported highest IVI (138.0; 230.5) under the canopy while Achyranthes aspera (56.15) and Eragrostis ciliaris (53.1) had highest IVI in canopy gaps. Diversity indices were higher in canopy gaps. Herbaceous community biomass in canopy gaps ranged from 700-900 gm -2 while under the canopy it ranged from 30-70 gm -2 . Cenchrus ciliaris (597.97 gm -2 ) and Chrysopogon sp. (391.2 gm -2 ) constituted major proportion of biomass in canopy gaps. The soil of under canopy regions had lower pH and bulk density, but higher soil moisture (%), electrical conductivity, soil organic carbon (SOC), soil total nitrogen (TN), SOC stock and soil TN stock than the canopy gaps. Linear positive relationship between species diversity and biomass; negative relationship between soil chemical properties and herbaceous biomass was found in study area. Even though canopy had positive effects on soil physico-chemical properties, overall negative effect on herbaceous diversity and biomass was seen. Conclusions: Reduced light for photosynthesis and tree root competition for soil resources may be the major factors negatively affecting herbaceous diversity and biomass.

micaceous rocks (Sinha, 2014). Its soil is mainly dry and sandy, lacking humus and supporting sparse vegetation (Sinha, 2014). The Indian State of Forest Report, 2015, published by FSI records a total of 299.77 km 2 of forest and tree cover in Delhi, which accounts to 20.22% of its geographical area.
Sampling design, data collection, and laboratory analysis After reconnaissance and eld observations, two representative sites on the campus were chosen for the study: Site 1: Parth Sarthi Rocks (PSR); Site 2: Poorvanchal (PRV). At each site, soil and herbaceous communities of both canopy gaps (CG) and under canopy (UC) areas were studied. The study area was divided into four permanent plots, viz., PSR CG, PSR UC, PRV CG, and PRV UC and phytosociological study was conducted during September 2019 using 10 random quadrats of 0.5 m x 0.5 m size. Soil sampling was done randomly with a metallic tube 50 cm high and 5 cm in diameter. After removing the litter layer carefully by hand, 5 samples from each of the four plots were collected at 0-10 cm depth. The samples were collected in zip lock bags, labeled and brought to the laboratory. Before the physicochemical analyses, all soil samples were air-dried and then sieved through a 2 mm sieve to remove the stones, litter, and plant roots.
For soil bulk density, soil samples were collected separately. Five soil samples were collected from each of the 4 permanent plots using cylindrical soil corer of diameter 5 cm and height 50 cm from the depth of 0-10 cm. The samples were then brought to the laboratory, sieved (2 mm) and oven-dried.
For biomass estimation of the herbaceous vegetation, aboveground and belowground parts of the plants were separated, cleaned and wrapped together in paper and labeled. The samples were then kept in oven at 65 C till constant weight. Dry weight was then taken.
Soil moisture (%) was measured gravimetrically as the percent weight of water lost after the fresh soil sample was oven-dried. For this, 20g (W1) of fresh soil sample was weighed in a beaker. It was then oven dried at 105 C till constant weight was attained. The samples were weighed again (W2). Soil moisture (%) was calculated using following formula: Soil bulk density by core sampling was determined (Lichter and Costello, 1994) using the following formula: In equation (2), BD is the bulk density of the < 2mm fraction (g cm -3 ), W is the oven dry mass of ne fraction < 2mm in gram, CV is core volume (cm 3 ) = 785 cm 3 , CF is mass of the coarse fragment (> 2mm) in gram, PD is density of rock fragment or particle density given as 2.65 g cm -3 . Core volume is calculated as ℼr2h where, r is the radius of the cylindrical core = 5 cm, h is depth of soil taken = 10cm.
The soil texture was measured using the sieving method (Gee and Or 2002). Soil pH was measured using control dynamics pH meter (Rhoades 1982). The soil to distilled water ratio taken was 1: 2.5. Soil electrical conductivity (EC) was measured using the Wensar Multiparameter water quality meter (Rhoades 1982).
The soil to distilled water ratio taken was 1: 2.5. Soil organic carbon (SOC) was calculated using the modi ed Walkley Black titration method (Walkley and Black 1947). Percent of soil organic matter (SOM) was obtained by multiplying percent soil organic carbon by a factor of 1.724 following the assumption that organic matter is composed of 58% carbon. Soil total nitrogen as calculated using Kjeldahl method (Kirk 1950) (digestion, distillation, and titrimetric analysis).
Soil C: N ratio was calculated using following formula: The SOC and soil TN stocks (Mg ha −1 ) were calculated as follows: In equation (4) and (5), where SOC is soil organic carbon concentration (%), TN is the soil total nitrogen concentration (%), D is the sampling depth (cm), BD is the bulk density (g cm -3 ) and G refers to the gravel content (%).

Data analysis
Quantitative parameters such as percentage of frequency, density and abundance of each herbaceous species present in quadrats were recorded and analyzed as per the methods of Curtis and McIntosh (1950). The importance value index of species was calculated by summing the three relative values, viz., relative frequency, relative density and relative dominance of the species. The following formula were used: The ratio of abundance to frequency (A/F) was worked out to interpret the distribution pattern of species (Whitford 1949;Curtis & Cottam 1956), Accordingly, the value for regular dispersion is < 0.025, random -0.025 to 0.05 and contagious ->0.05. The species richness was calculated using Menhinick Index (Menhinick 1964) and Margalef Index (Margalef 1958). The species diversity was computed by using the Shannon-Wiener Diversity Index (Shannon-Wiener 1949) and Simpson's Index of diversity (Simpson 1949). Species evenness was calculated using Shannon Equitability index (E H )/ Pielou's Evenness Index (Pielou 1966). The formula used were as follows: In the above equations (7-11), pi or (ni/N) = proportion of individuals belonging to species 'i' where ni is number of individuals of species 'i', N is the total number of individuals in the plot, S = total number of species in the plot.
The soil texture was computed using Gradistat 8.0. Descriptive statistics were performed using R studio and MS Excel (2013) analytical software and expressed as the mean value± standard error. One-way analysis of variance (ANOVA) and Tukey's post hoc test was performed to test the within and between signi cant differences of the different soil variables at p<0.05. Pearson correlation coe cient was calculated between biodiversity indices, biomass, and soil physicochemical properties of the open canopy and under canopy sites using R studio software. Redundancy analysis (RDA) (by Vegan package in R) was used to examine changes in herbaceous species biomass about open and under canopy sites.

Vegetation characteristics
A total of 19 herb species were recorded from the study area. 11 were perennial and 8 were annual herbs. Grasses (11) were more in number compared to forbs (8). Grasses were higher in canopy gaps (9) than in under canopy (4) while forbs were higher in under canopy (7) than canopy gaps (6). Out of 6 families recorded for the representative species, Poaceae (9 species) was observed to be dominant at all study sites followed by Amaranthaceae, Acanthaceae, and Malvaceae (2 species). Details of herb species attributes are given in Table 1.   Chloris barbata - Chrysopogon sp.
-   According to Pearson correlation analyses, herbaceous AGB, BGB, and TB were positively correlated with the diversity indices (Menhenick Index, Shannon Weiner Index, Margalef, Simpson Index, and Shannon Equatibility Index) (Fig. 3). This indicates higher species diversity supported greater productivity.

Soil physico-chemical properties
Soil moisture was lower in canopy gaps as compared to under canopy while the bulk density showed the reverse trends (Fig. 4). Soil clay, silt, and sand content did not vary between canopy gaps and under canopy sites. Soil pH, soil EC, SOC, TN, SOM, SOC/TN, and soil carbon and nitrogen stock varied little both among and within sites (Fig. 5). The soil of under canopy regions had lower pH than canopy gaps of the same site, thus pH varied signi cantly (p < 0.05).
The soil of under canopy regions had higher EC than canopy gaps at both sites. Also, under canopy sites had signi cantly (p < 0.05) higher SOC, TN, and SOM contents and soil carbon and nitrogen stock than canopy gaps. Soil SOC/TN ratio was higher in PSR under canopy sites. Thus soils of under canopy regions were acidic, had greater porosity and soil moisture, and were richer in organic matter and nutrients compared to canopy gaps. Pearson correlation analyses showed that SOC, TN, SOM contents, SOC stock and TN stock were positively correlated with each other (Fig. 3). However, SOC, TN, SOM contents, SOC stock and TN stock were negatively correlated with bulk density and soil pH and signi cantly correlated with soil moisture contents (Fig. 3). This indicates that greater soil organic matter retained more soil moisture, reduced pH, and increased the porosity of the soil. Redundancy analysis (RDA) con rmed the correlation between soil physicochemical properties and biomass (Fig. 6). RDA axes 1 and 2 shows 74.14 and 9.49% variation in composition respectively. chemical properties (SOC, TN, SOM, SOC stock and TN stock). Chloris barbata, Lathyrus palustris, Ageratum houstonianum, Chrysopogon sp., Arundinella pumila correlated with soil bulk density. As shown in Fig. 6, Oplismenus burmannii and Dipteracanthus prostratus showed positive relationship with soil chemical properties (SOC, TN, SOM, SOC stock and TN stock).

Discussions Species composition and diversity
The species found in the present study belonged to either of the 6 families -Poaceae, Malvaceae, Amaranthaceae, Asteraceae, Acanthaceae, and Fabaceae which are among 10 dominant families in the region (Mishra et al., 2015b). The dominant family was Poaceace (grass family) which is the dominant family of herb species in Aravalli hills (Gaury and Devi, 2017). In the present study, herbaceous species composition and dominant species varied in canopy gaps and under the tree canopy. Under canopy was dominated by Oplismenus-Dipteracanthus community while canopy gap was dominated by Cenchrus-Chrysopogon community. Several other studies also reported differences in herbaceous communities in the and canopy gaps (Sagar et al., 2012;Singh et al., 2017).
In another study, clear harvesting of trees in oak-hornbeam forests changed the understory vegetation considerably as compared to intact tree stands (Tinya et. al., 2018). Similar to the present study, canopy gaps hosted a higher number of colonizing species with greater plant height in temperate r-beech forests (Kermavnar et. al., 2019). Increasing light availability followed by warming led to establishment of taller herb community in temperate deciduous forest (Blondeel et. al., 2020). Plants that are shaded by similar sized neighbors (as in the case of canopy gaps), produce longer internodes (resulting in increased plant height) and larger leaves to maximize light capture (Poorter et. al., 2019). According to Bonser and Geber 2005, erect growth forms are favored in lowlight habitats and rosette or prostrate growth forms under high-light habitats. However present study showed reverse trend. In a study, erect species showed greater competitive ability (in terms of leaf number) than other growth forms (Fazlioglu et. al., 2016). In present study also, species having erect growth form dominated the study area. Also, higher number of C 4 plants were present in canopy gaps while higher number of C 3 plants under the canopy in our study. C 4 photosynthesis evolved as an adaptation to high light intensities, high temperatures and dryness. Therefore, C 4 plants dominate grassland and biomass production in tropical and subtropical regions (Edwards et al., 2010).
Understorey vegetation serves as a good indicator of the ecological condition of a system (Schmidt, 2005;Mataji et. al. 2010). Most of the species (Achyranthes aspera, Brachiara ramosa, Cynodon dactylon, Cenchrus ciliaris, Chrysopogon sp., Dactyloctenium aegyptium, Dipteracanthus prostatus, Eragrostis ciliaris, Sida sp. and Setaria verticillata ) found in the present study are species of disturbed areas such as roadsides, agricultural elds and wastelands (Mishra et al. 2015b) and are ruderals or weeds. Chloris barbata and Sida sp. are non-native species (Mishra et. al., 2015a). Presence of such species in the study area suggests it is a disturbed site, however the extent of disturbance was not evaluated. Across the sites, all species showed contagious pattern of the distribution. In natural conditions, contagious distribution is the most common type of distribution and is performed due to small but signi cant variations in environmental conditions (Odum, 1971). In our study, several species reproduced through vegetative propagation through stolon, rhizomes and adventitious roots. Many species had tufted or clumped growth forms. The dominance of contagious distribution may be due to the fact that majority of grass species reproduce vegetatively in addition to their sexual reproduction (Shameem et al., 2017).
Species diversity was higher in canopy gaps than at under the canopy area. Singh et al., 2017 also reported that the distribution and number of unique species were higher in grassland vegetation as compared to under forest in dry tropical forests of India. Studies in a temperate deciduous forest by Vockenhuber et al., also showed a negative response of herbaceous species richness to increasing canopy cover. The herb density and species richness were higher in the open canopy of sal forest of central Himalayas, India due to maximum space, less canopy cover, low density of trees and proper sunlight. Herbaceous density was lower in sal mixed canopy forest due to the closed canopy, higher tree density and presence of more number of broad-leaved associated tree species (Tectonis grandis, Mallotus philippensis, and Terminalia sp.) interrupting light to reach the ground vegetation (Adhikari et al., 2018). Some studies suggested that the decrease in species richness and diversity under the canopy is due to an increase in soil nitrogen input which leads to the loss of many nitrogene cient species and brings about the dominance of few high nitrogen-requiring species (Gilliam 2006). Species richness and simpson diversity index increased with increasing gap area and increasing light availability in beech stands (Vajari et al., 2012). In a natural disturbance-based silvicultural system, harvest gaps conserved understory plant diversity by promoting moderate disturbance. However, it also led to introduction of invasive plant species (Bolton and Amato 2019). Rainforest trees invasion in open forests led to decline in understory plant diversity and cover in open forests in Australia (Baker et al., 2020). Light availability was positively correlated with understory plant species richness and with understory cover in temperate forest also (Dormann et. al., 2020). Biomass The total herbaceous biomass in canopy gaps ranged from 700-900 gm − 2 while in under canopy area it ranged from 30-70 gm − 2 . Varshney (1972) recorded maximum shoot biomass 771 gm − 2 and a minimum of 9 gm − 2 for the grasslands of the region. In the semi-arid grasslands of Khirasara, the maximum shoot biomass recorded was 201 gm − 2 (Singh and Krishnamurthy, 1981). The semi-arid region of India has Cenchrus-Dicanthium dominated grasslands (Dabadghao & Shankarnarayan, 1973;Chandran, 2015). In the present study also, Cenchrus ciliaris constituted major proportion of herbaceous community biomass in canopy gaps of the semi-arid forests.
Singh et al. 2017 also reported higher herbaceous biomass in open grasslands compared to under the canopy of the trees. In another study, moist dense forest canopies showed negative effect on herbaceous biomass (Shirima et. al., 2015). Similar trend was seen in Mediterranean dehesas (López-Carrasco et. al., 2015). Canopy plants absorb a large proportion of the available irradiance, water and nutrients, so there is always likely to be competition for resources between canopy and understorey plants (Valladares et al., 2016). However, other studies (Das et al., 2008;Sagar et al., 2012) reported positive effects of tree canopy on herbaceous diversity and production in dry tropical forests of Northern India.
Herbaceous peak biomass is intimately related to annual production and can be used as a substitute for the measurement of productivity (Sagar et al., 2012).
In the present study positive correlation was found between species diversity and biomass (under canopy areas had lower species diversity and biomass while reverse was seen in canopy gaps). A linear relationship between productivity (peak biomass) and diversity was also found in several other studies (Abrams 1995;Bai et al., 2007;Sagar et al., 2012;Singh et al., 2017). Our study supported the diversity-stability hypothesis of Elton (1958) i.e. higher species diversity supports greater productivity in the ecosystem. However, reverse trend was seen in a study of mountain forests along the elevation gradient. There, increase in the diversity, richness, and species evenness of herbaceous plants resulted in a decrease in the plant dry biomass (Shahriari et. al., 2019).

Soil Physico-chemical Properties
In this study, the soil moisture content was higher in under canopy region and ranged between 8-20% while in canopy gaps it ranged from 3-10%. Savanna trees and shrubs may reduce the sub-canopy solar radiation by 45 to 60%, which may lower the soil temperature and evapotranspiration, which may subsequently increase the soil moisture content (Vetaas 1992). Studies have shown that soils developing under the tree canopy have a greater water-holding capacity and a macroporosity favorable to in ltration and redistribution of soil water (Joffre and Rambal 1993). Direct solar radiation, due to canopy opening, increases evaporative demand and reduction in soil moisture (Joshi et al., 2001;Pausas and Austin, 2001;Sagar et al., 2012;Shirima et al., 2015;Singh et al., 2017). Lower water holding capacity and soil moisture could be due to synergistic events of direct solar radiation, canopy gaps and soil structure mediated by the local anthropogenic disturbances on the sites (Singh et al., 2017). In this study, the soil bulk density was slightly higher in canopy gaps. This is due to soils beneath the canopy are more fertile and less compact than those in the canopy gaps (Gómez-Aparicio et al., 2005).
As in the present study, Joshi et al., 2001;Das and Chaturvedi 2008;Sagar et al., 2012;Singh et al., 2017;Desta et al., 2018;Amolikondori et. al., 2020 also reported higher soil moisture, soil organic carbon, soil nitrogen, and lower soil pH in under canopy regions compared to canopy gaps. This may be due to greater litter inputs, tree roots turnover and dust accumulation on the canopy which gets transported to the soil by through-fall and/or stem-ow. In another study of dry lands in Ethopia, effect of Acacia saligna canopy on soil physic-chemical properties was seen. Soil bulk density increased as the distance from the tree trunk increased. However, soil total nitrogen, soil organic carbon, moisture content, silt and clay percentages and soil electrical conductivity were greater under the canopy compared to outside the canopy (Gebretsion et. al., 2019). In the present study, there was a negative correlation of herbaceous biomass with soil physicochemical properties. Singh et al., 2017, found an increase in herbaceous biomass with a decrease in soil water and nutrients in grasslands. This may be due to the higher e ciency of C 4 grasses to photosynthesize in low moisture and nutrient conditions. However, in other studies (Das et al., 2008;Sagar et al., 2012) positive relation of soil moisture and nutrient availability with herbaceous diversity and productivity was seen.

Conclusion
Herbaceous species composition, diversity, biomass, and soil physicochemical properties showed signi cant variations in canopy gaps and under tree canopy sites of semi-arid forests. Canopy gaps and under canopy sites were dominated by different species. Species diversity and biomass were signi cantly higher in canopy gaps. C 4 plants (grasses) had a major contribution to species diversity and biomass of canopy gaps. A linear relationship occurred between species diversity and biomass. Canopy cover had positive effects on soil moisture and nutrients. However, soil chemical properties (SOC, TN, SOM, SOC stock, and TN stock) were negatively related to herbaceous species biomass. The increase in herbaceous diversity and biomass with reduced water and nutrient conditions in canopy gaps can be utilized for the management of forest under the present changing climate scenario where C sequestration can be enhanced with fewer inputs (Singh et al. 2017