Introduction

Pinus wallichiana is one of the three naturally distributed pines in Pakistan and besides its relict distribution at Baluchistan, it mainly appears in Himalaya, Karakoram and Hindukush mountains from eastern Afghanistan across northern Pakistan and India to Yunnan in southwest China (Yadav et al. 1997; Ahmed et al. 2006). This species is a major component of the middle to high elevation and grows at elevations from 1800 to 3400 m (rarely as low as 1200 m), varying between 5 and 18 m in height (Farjon 2013). It covers a substantial area and is one of the characteristic elements of both dry and moist temperate forests in Pakistan (Khan et al. 2014). P. wallichiana usually grows on deep moist soils (Yadav et al. 1997) but also grows well on fertile, well drained sandy-clay to sandy-clay-loamy soils (Sheikh 1993).

One site supporting these natural forests is District Swat of the Hindukush range mountains (HRM) which has been the focus of various organizations and tourists due to the high timber quality, associated wildlife, and abundant non-timber forest products (NTFPs) (Baig et al. 2008). It forms both pure and mixed stands depending on large-scale ecological gradients (Wahab 2011). These stands have not been classified phytosociologically and their correlations with environmental parameters are not yet explored. Elucidation of such relationships is fundamental to understanding plant species composition and structure in a particular habitat, landscape, and region (e.g. Barbour et al. 1987; Mucina 1997; Burke 2001). Generally, plant communities are dynamic entities undergoing continuous changes in response to environmental factors and intrinsic dynamics within the community (e.g. Ma et al. 2012; Ren et al. 2013).

Communities and habitats in transition zones are potentially affected by climate change and thus are often the first communities to shift, especially in areas providing marginal conditions for plant growth. This phenomenon prevails in the Swat mountains which is an ecotone between the moist and dry temperate areas of Pakistan (Champion et al. 1965). These mountains form part of the Great Hindukush escarpment, an extension of the Hindu Raj and Himalaya lies in the Sino-Japanese phyto-geographical region that extended to Afghanistan and the Karakorum Range of Gilgit-Baltistan (Ahmed et al. 2006). Climatic, topographic and soil factors are believed to be the overriding factors determining vegetation patterns (Siddiqui et al. 2010; Ahmed et al. 2011). The variety of habitats, edaphic conditions and its transitional position between moist and dry parts make the Swat Hindukush range mountains an ideal study area to test whether these variables are the driving factors of vegetation structure? Steep climatic gradients, and high-elevation habitats can form an important buffer for climate-induced changes (Chapin et al. 1998; Zhang and Zhang 2007). These forests are being degraded by deforestation, overgrazing and clearing of land for terrace cultivation (Hussain et al. 1997; Sher et al. 2010; Ahmed et al. 2010; Shaheen et al. 2011), thus documenting existing plant communities and understanding their relationships with the environment is critical. Although an early inventory of plant species was compiled for the study area (Stewart 1967; Beg and Khan1984; Ali 2011; Rahman 2012; Ilyas et al. 2012), vegetation parameters were never quantified except by Siddiqui et al. (2016), Ilyas et al. (2015), and Rahman et al. (2017). These authors focused on understory vegetation, leaving woody vegetation to be either ignored or studied sporadically. Some studies of other issues such as ethnobotany and species niche modeling were also undertaken (Ahmed et al. 2010; Sher et al. 2010; Qasim et al. 2012; Akhtar et al. 2013, 2014, 2015). However, no comprehensive work has been done on the classification of woody vegetation and its correlations with environmental variables although the area has both dry and moist forest thickets with a variety of vegetation types. Thus, this study was aimed to document the existing plant communities of P. wallichiana populations and to investigate major ecological and environmental gradients responsible for its distribution in Swat Hindukush mountains. This study will help to understand to predict the development of P. wallichiana plant communities and their potential determinants because of its significant ecosystem value as a reference across the northern areas of Pakistan and adjoining countries.

Materials and methods

Study area

Our study sites were located in two contrasting climatic zones, i.e., moist and dry temperate forests, in the Swat Hindukush mountains of northern Pakistan (Fig. 1). These sampling sites were frequently distributed in the subalpine zone just above the Pinus roxburghii and admixed with dry and moist temperate evergreen forests in the Sino-Japanese phytogeographical region of Pakistan (Champion et al. 1965). The total area of the district is 5337 km2 (lat. 34°34′–35°55′N, long. 72°08′–72°50′E (Shinwari et al. 2003). The main geological formations are Hornblenditic and Schistose rocks (Zeb 1970) characterized by felsic zones. Low elevation portions of the study area were rich in alluvial soils (Hussain and Illahi 1991) and supported cereal crops, vegetable and fruit orchards (Ilyas et al. 2015).

Fig. 1
figure 1

Locations of forest stands sampled in the Hindukush mountains ranges of district Swat, Pakistan

Full size image

The climate is under the strong influence of elevation gradient and generally falls in the temperate zone and a humid subtropical climate under the Köppen climate classification (Cfa) with four distinct seasons. Cold winters with heavy snowfall accumulation in the form of glaciers are the characteristic feature of the area at high elevations, where summers are mild and short (Sher et al. 2010). Annual precipitation varies from 693 to 897 mm and temperature ranges from 13.4 to 21.7 °C (data source PMD; http://www.pmd.gov.pk). November and April are the driest and wettest months, respectively, with 15 mm and 93 mm of precipitation at high elevations, respectively, and generally varying 112 mm between the driest and wettest month in the entire region. In January, the average temperature is 1.5 °C and July is the hottest month with an average temperature of 24.1 °C (Ali et al. 2014). Due to variable topography and climate, much of the area is covered by shrubland (Dodonea viscosa, Indigofera gerardiana and Parrotiopsis jacquemontiana) or broadleaved evergreen (Quercus baloot, Q. Incana, Olea ferruginea) and broadleaved deciduous (Acacia modesta, Ficus palmata, Alnus nitida) forests. Introduced Eucalyptus lanceolata and Populous nigra dominate where they occur along roads and rivers (Ilyas et al. 2013). Other vegetation includes, coniferous, meadows, wetlands and ruderal and gullies vegetation communities developed by ice melting.

Sampling and data collection

We sampled vegetation dominated by P. wallichiana in its natural zone of occurrence in both moist and dry temperate forests during the peak growing vegetation period in 2014–2016. Thirty forest stands were sampled between 1200 and 3000 m a.s.l., at sites with varied topographic and soil characters, and starting around the transition zones of subtropical pine forests (P. roxburghii) in the region. At each stand, we sampled 10 plots measuring 10 × 10 m for a total of 300 plots. We also sampled a 200 m straight transect for overstory and 5 × 5 m for understory (Uprety et al. 2014). We attempted to identify each tree species with the help of a senior taxonomist in the field and we recorded frequency, density, diameter, and height. Frequency and density for understory species were also recorded and were used as species abundance data to describe community types. Within each forest stand we recorded elevation, slope gradient, and aspect by use of GPS, clinometer and magnetic compass. Three topsoil samples of 1 kg were taken at three different positions to a depth of 15 cm (or less, if the surface soil horizon was shallow) using a soil bucket agar. The three samples were stored in polyethylene bags and bulked into one composite sample. The three soil sampling spots were taken from the edge of the study plots so as not to disturb plot vegetation as recommended by Woch et al. (2016).

pH values were measured in situ directly by making a suspension of distilled water and soil (1:5 ratio) in the field (pH meter PH119) following Khan et al. (2014). All the soil samples were brought to the Agriculture Research Station (ARS) Mingora North and were sieved (2 mm mesh) and dried at 105 °C prior to physiochemical analysis. Soil texture properties, viz. sand, silt, and clay, were quantified by a combination of sieving and sedimentation. Soil organic matter (SOM) content was determined by the Potassium dichromate (K2Cr2O7) method while a modified semi-micro Kjeldahl procedure was adopted for estimation of total nitrogen (Lu 2000). Micro diffusion and Olsen methods were used to determine the available nitrogen and phosphorus (AP) respectively, whereas, available potassium (AK) was estimated following the method of Bray and Kurtz (1945).

Data analysis

Frequency, density and basal area for different tree species were calculated and sorted according to Mueller-Dombois and Ellenberg (1974). Relative values were calculated by dividing individual values for frequency, density and basal area by the sum of frequencies, densities and basal areas of all species in a plot, then multiplying by 100. Importance value index (IVI) for each species was calculated as the sum of the relative frequency, density and basal area values (Cottam and Curtis 1956). We used importance values in the analysis as it discloses the ecological importance of a species and gives more information than any other single ecological attribute (Brown and Curtis 1952). Data matrices of tree species IVI and the corresponding environmental variables were prepared in Microsoft Excel and were used in the quantitative multivariate analyses using PC-ORD software version 5.10 (McCune and Mefford 2005). In the assignment of slope and aspect, 4 and 8 grades were set and assigned with figures respectively as suggested by Khan et al. (2013) for better interpretation. We used Ward’s Agglomerative cluster analysis (WACA) for objective classification of forest vegetation (McCune and Grace 2002). We used the quantitative Sorensen (Bray–Curtis) distance measure, an effective distance measure for ecological community analysis (McCune and Grace 2002) and flexible beta (β = − 0.25) as a linkage method (Lance and Williams 1967) which is often compatible with Sorensen distance and is space-conserving (Legendre and Legendre 1998). Community type was declared based on the first two species with highest average importance values in the groups following Chai and Wang (2016). The phytosociological parameters of vegetation and associated environmental variables in groups identified by cluster analysis were subjected to one-way analysis of variance (ANOVA) to compare means. The significant differences (p < 0.05) among different parameters in vegetation groups (community types) and habitat properties were further examined using a Tukey’s HSD post hoc test for multiple comparisons using SPSS 18.0. The variability of vegetation parameters and topographic variables between groups were tested with Box–Whisker diagram following Kavgaci et al. (2013). Non-metric Multinational Scaling (NMS) ordination was applied to identify important ecological gradients determining plant species distribution and to verify the classification of the forest vegetation as the two results are often complementary (Atkinson and Treitz 2012; Khan et al. 2014). We preferred the use of NMS-ordination as it is a superior ordination method because (1) it is not based on the assumption of multivariate normality, and (2) it is robust to problems caused by large numbers of zero truncation (Minchin 1987). We employed the low stress-value in the analysis as lower values of stress indicate a more reliable ordination (Benítez-Malvido et al. 2014). Intra- and inter-set correlations among ordination scores and environmental factors were performed with Pearson’s r (Peterson and McCune 2001).

Results

Classification of vegetation

We recorded 105 plant species of 90 genera and 50 families from both overstory and understory in the 300 sampling plots. Dicotyledonous angiosperms were the most abundant category with 90 species (Table 1). Of these species, Gymnospermae included 5 species of 2 families, and Pteridophyta included 3 species of 3 families. The plant species in this region mainly represented 50 families, including Asteraceae and Rosaceae which were rich species families with twelve species each. Lamiaceae was the second most rich family with 9 species, whereas, the remaining families had 4 or less species each.

Table 1 Overall floristic composition and major taxonomic divisions of the recorded plant species
Full size table

Our 30 forest stands were comprised of 12 tree species that we classified into three ecologically distinct vegetation types using Ward’s agglomerative cluster analysis (Fig. 2). Dominant and co-occurring species of these populations types with their importance values (IV) are listed in Table 2, and the corresponding topographic and soil variables are listed in Table 3.

Fig. 2
figure 2

Classification of 30 sample stands dominated by Pinus wallichiana into three groups (vegetation types) using Ward’s Agglomerative clustering procedure. Groups were extracted at 70% information and arranged from top (Group-I) to bottom (Group-III) on the Dendrogram

Full size image
Table 2 Mean (± SE) importance values of different tree species in three groups identified by Ward’s agglomerative cluster analysis
Full size table
Table 3 Average values (Mean ± stand error) of the topographic and edaphic parameters in the three vegetation groups isolated by cluster analysis
Full size table

The first division of the clustering dendrogram generated a group of 11 forest stands dominated by P. wallichiana (IV = 76 ± 2.7%) and Quercus dilatata (IV = 6.4 ± 2.5%). This group was both floristically and environmentally heterogeneous and consisted of 12 trees and 66 understory species. P. smithiana, A. pindrow, and Q. incana were co-occurring species with importance values ranging from 4.3 to 3.1%. T. fauna, C. deodara, A. indica, J. regia, P. cornuta, P. pashia and P. duthiei contributed < 5% of importance value in this group (Table 2). The understory flora included 66 species, of which Fragaria nubicola was abundant, while Dryopteris stewartii was a frequently occurring species. Rabdosia rugosa, Viola canescens, Viburnum grandiflorum, Indigofera heterantha, Rubia cordifolia, Arisaema flavum, Bistorta amplexicaulis and Berberis lyceum were in the frequency range from 20 to 40%. The remaining 56 species were rare with average frequency ranging from 0.4 to 20%.

The second group generated a homogenous cluster and we considered this to be a monospecific community of P. wallichiana (IV = 95%). Each co-dominant species contributed < 2% of the importance value (Table 2). This group constituted a single cohort stand structure and generally formed mature stands in the region. In comparison to groups 1 and 3, this group was floristically rich in terms of understory species (80 species). Fragaria nubicola (70%) and Dryopteris stewertii (52%) were frequent species in the understory. The small number of species was occasional and occurred at a frequency ranging from 22% to 38%, whereas 69 species recorded in this group occurred rarely.

The third group was dominated by P. wallichiana (IV = 52.8%) and C. deodara (IV = 24.5%) whereas, P. smithiana, Q. dilatata and A. pindrow locally join the population with importance values ranging from 6.3% to 8.7%. This community was represented by mature stands of a multi-cohort structure. The community was relatively poor with 28 understory species, of which F. nubicola (69%) was abundant and Rumex dentatus (41%) was frequent. V. canescens (29%) and I. heterantha (24%) were occasional species. Remaining 24 species were rare with average frequency ranging from 0.4 to 20%.

Environmental characteristics of the vegetation

The environmental variables topographic, edaphic and soil variables varied widely (Table 3). Forest stands mainly were located in the middle of the forest vegetation belt (2144–2361 m a.s.l.) with moderate (26°) to steep (32°) slope inclination and south to east aspect. Clay and sand particles, and pH progressively increased with increasing elevation. However, silt, organic matter and inorganic ions of soil exhibited a reverse trend. Vegetation group I occurred at low elevation on steep slopes with fertile soils containing high concentrations of N, P, K, and organic matter. The soil type was loamy and acidic.

Aspect varied greatly among stands of group I, i.e., north, north-west, west, north-east, and south-east. Group II stands were mostly located at mid elevation (average 2257 m), with steep slopes (32°) and south, east, and west aspects. Among soil nutrients, organic matter, N, P, and K+ contents were medium, while the concentration of lime was high. The soil type was loamy-sand with medium pH. Stands of group III were located at high elevation (2361 m) with comparatively gentle slopes (26°) on north-west aspects. In comparison to groups I and II, organic matter and the concentration of essential nutrients (N, P, and K+) were lower with slightly acidic soils.

Univariate analysis of variance (ANOVA)

Environmental variables corresponding to the three identified vegetation groups were analyzed using one-way ANOVA (Table 4). Among the topographic variables, slope gradient differed significantly (Tukey HSD test, F = 10.89; p < 0.05) by group but elevation and slope aspect did not. Edaphic variables organic matter (%), soil nitrogen (%), lime (%) and phosphorus (mg kg−1) did not vary by group but soil pH did.

Table 4 Intra-set correlation among the different topographic and soil attributes
Full size table

Correlation among the environmental variables

The intra-set correlation among different environmental factors exhibited significant relationships on various probability levels (Table 4). Among the 12 variables, elevation showed a negative relationship (r = − 0.419; p < 0.05) with slope whereas; aspect, lime and potassium contents were positively correlated. pH showed a positive relationship with aspect and clay content of the soil. Strong negative relationship of silt was also observed with sand, pH, and lime (%) respectively.

The sand particles showed a strong negative relationship with clay (r = − 0.553; p < 0.05) and silt particles (r = − 0.901; p < 0.05). This relationship was also observed in phosphorus contents with sand contents and soil pH. Lime, nitrogen and potassium content showed a positive correlation with soil organic matter, while nitrogen content of the soil was positively correlated with potassium.

Ordination and gradient analysis

To detect the significance of individual environmental factors, Non-metric multidimensional scaling produced a 2-dimensional solution (Fig. 3) with a stress value of 11.70 (p < 0.01) based on 250 iterations with randomized data and a final instability of 0.01019. The ordination explained 79% of the variance in the original dissimilarity matrix, with axes 1 and 2 accounting for 51% and 28% of the total. In the NMS-ordination, three groups identified by cluster analysis were superimposed into the graph configuration quite neatly and could be distinguished along both axes 1 and 2. P. wallichiana–Q. dilatata (Group-I) dominated forests appears at the upper middle part of the axes and P. wallichiana (Group-II) pure population forest, located at the extreme top of the 2-D NMS-ordination along axis 1 and 2. P. wallichiana–C. deodara (Group-III) dominated forests plot on the lower middle part of the graph configuration.

Fig. 3
figure 3

Non-metric multidirectional scaling ordination (NMS) showing the pattern of the 30 stands dominated by P. wallichiana along axes 1 and 2. Groups i.e., Pinus–Quercus; Pinus (pure community) and Pinus–Cedrus community obtained from cluster analyses were superimposed in the ordination space

Full size image

In the NMS ordination diagram of the forest stand dominated by P. wallichiana, a clear gradient of the floristic variations in the communities were evident along axes 1 and 2 (Fig. 3). The ordination axes 1 and 2 is positively correlated with the elevation (r = 0.3778; p < 0.05), however, slope (r = 0.5548; p < 0.01) and pH (r = 0.49750; p < 0.01) were correlated strongly with ordination axis 2 only (Table 5). Aspect exhibited a significant relation with both the axes whereas, significant relationships of soil clay (r = 0.2782; p < 0.05) and silt (r = 0.32251; p < 0.05) contents with axes 2 were also observed (Table 5).

Table 5 Correlation coefficients (Pearson’s r, n = 30) between topographic, soil physical and chemical attributes with vegetation scores for 2-D NMS ordination
Full size table

The soil variables organic matter, nitrogen, phosphorus, and potassium did not show any significant relationships. However, among the edaphic factors, pH and lime (%) affected vegetation in the study area (Table 5).

Discussion

The quantitative clustering and ordination approach in the present study clearly describe the distribution pattern of the tree vegetation composition. Although the vegetation types were dominated by P. wallichiana, they showed substantial structural and floristic heterogeneity and complex relationship with the local environment. The three community types were dominated by P. wallichiana and were distributed at elevations from 2144 to 2361 m a.s.l., a middle elevation in the study area. The species rich Pinus–Quercus community grew best at low elevations, whereas, the pure population was located at medium elevation (2257 m) on steep slopes (32°). The second species rich Pinus–Cedrus community occurred at high elevation (2361 m) and on moderate (26°) slopes. Many studies have reported similar results in classifying the coniferous forests in Himalayan, Hindukush and Karakoram ranges of northern Pakistan. Akbar et al. (2017) described the coniferous forests of Gilgit-Baltistan and concluded that P. wallichiana dominated pure and mixed forest stands at low and mid elevation zones on medium to steep slopes (26°–33°). Ahmed et al. (2010) studied P. wallichiana from different climatic zones of Pakistan at elevations from 1950 to 2700 m and slopes of 23°–45° gradient. They concluded that Betula utilis and P. wallichiana prefer moist areas of dry temperate zones. Ahmed et al. (2006) recorded P. wallichiana in pure stands at elevations of 2770 m and 3100 m on south exposures in Naltar and Takht-e-Suleiman area of Baluchistan Province.

The results of cluster analysis have generally suggested a vertical zonal distribution of P. wallichiana dominated vegetation types in the Swat Hindukush range mountains. However, the vertical zonal spectrum is incomplete due to the occurrence of sampling forests only in middle elevation range. Forest area in this region is comparatively small and mainly composed of coniferous mixed forests (Champion et al. 1965; Ahmed et al. 2006).

The ordination results were complementary to classification and ecologically sound as the floristic variations in communities dominated by P. wallichiana showed a clear gradient of environmental variables along the axes 1 and 2 (Fig. 2). Among the environmental variables elevation (r = 0.3778; p < 0.05) and aspect showed significant positive correlation with both of the axis, whereas, slope showed statistical strong correlation (r = 0.5548; p < 0.01) with axis 2 (Table 5). These results indicated that topographical variables have strong influence on vegetation, because the change in physiography leads to change in temperature and humidity gradient (Zhang et al. 2006). Several authors have discussed the prominent role of elevation, aspect and slope in vegetation development and distribution in mountains (e.g. Fisher and Fuel 2004; Kabrick et al. 2004; Jonathan and Givnish 2007; Mahdavi et al. 2013; Khan et al. 2014), and therefore, it is significant in vegetation restoration in such microhabitats.

The results of NMS-ordination also suggest that soil textural and chemical properties i.e. clay (r = 0.278), silt (r = 0.323) and pH (r = 0.49750) have significantly positive correlations with ecological groups of the species (Table 5). Our findings are in line with several studies (e.g. Badano et al. 2005; Small and McCarthy 2005; Mataji et al. 2009), that soil texture also greatly influences the formation of plant communities. Additionally, pH (r = 0.4975) and lime (r = 0.398) are important in defining the ecological groups and species richness in the present study as it plays an important role in regulating photosynthesis, and other physiological processes (Xu et al. 2006; Zhang et al. 2012; Sekulova et al. 2013). The availability of essential nutrients like potassium, phosphorus, organic matter and nitrogen etc., remained inactive in the floristic variation of the P. wallichiana dominated forest of the study area. Although the role of these factors in the distribution of plant species is described by several researchers in their studies (e.g., Enright et al. 2005; Zare Chahoki et al. 2007) as these factors make easy the transformation of water and nutrients and thus increase soil fertility (e.g. Biggelow and Canham 2002; Amorin and Batalha 2007).

Conclusions

Three different tree communities dominated by P. wallichiana were identified and elevation-coupled soil textural and physiochemical properties were strongly correlated with vegetation distribution and must be considered if the goal is the restoration and conservation of these forests. This study provided a better understanding of the mechanisms driving vegetation distribution in the similar or other vegetation types at adjoining high mountain ranges in Pakistan, India, and Afghanistan. We recommend a more comprehensive study at both local and regional scales to expose the vegetation–environment relationships of these forests by considering several other biotic and abiotic factors for better understanding.