Introduction

Agriculture has a great potential to contribute to climate change mitigation by improved cropping practices and increased number of trees on farm lands. Agroforestry is the third largest C sink, after primary forests followed by long term fallows in Africa (Mbow et al. 2014b). The systems have 3–4 times higher biomass carbon stocks than traditional treeless agricultural systems, with C stocks ranging from 29 to 228 Mg ha−1 (Matocha et al. 2012; Mbow et al. 2014b). On a global scale, the total land area under different types of agroforestry systems is estimated to be about 1.6 billion hectares, with a sequestration potential of 1.1–2.2 Pg C in the aboveground biomass over the next 50 years (Lorenz and Lal 2014). Agricultural land with at least 10% tree cover represents 43% of the global agricultural area, sequestering 45.3 Pg (1 Pg = 1015 g) C (Zomer et al. 2016). Besides, if the 630 million hectares of unproductive cropland and grassland could be converted to agroforestry, a further 586,000 Mg C yr−1 could be added by 2040 (Kumar and Nair 2012). Agroforestry practices also play an important role in conserving biodiversity (Bishaw et al. 2013). Other roles of agroforestry practices include increasing soil fertility and thus agricultural productivity, and controlling erosion. Moreover, the system has a significant role in minimizing vulnerability, improving resilience of farming systems and the adaptive capacity of smallholders in the face of climate variability and change (Dhyani 2014; Kahiluoto et al. 2014; Lasco et al. 2014; Mbow et al. 2014a).

Homegardens are identified by their high diversity of species with many life forms varying from herbs to trees. These create multistory canopy structure, which makes homegarden multistrata system. In the southern part of Ethiopia, a tree-ensete based homegarden agroforestry, which is dominated by native perennial trees, shrubs, “ensete” (Ensete ventricosum Welw. Cheesman), and an annual crop, is a common practice. Homegarden agroforestry is known to enhance carbon accumulation, particularly soil organic carbon (SOC) stock through continuous supply of organic matter into the ground (Tiwari et al. 2017). It can also contribute to biodiversity conservation through providing supplementary habitat for remnant native species and buffering the pressure on natural habitats (McNeely and Schroth 2006; Bardhan et al. 2012).

The quantity of C sequestered in an agroforestry system depends largely on the nature of the system, the structure and function of components, which are determined by environmental factors (Albrecht and Kandji2003). The density of trees and standing stock in homegarden agroforestry practice vary with the tree growing location, species composition, environmental variables and types of agroforestry practices (Kumar 2007). For instance, elevation is one of the environmental variables, which influences SOC, vegetation cover, and biomass and their management practices (Podwojewski et al. 2011; Smith et al. 2012). The extent to which such factors influence different parameters such as SOC needs an investigation for better planning and management. Such studies are however, limited, particularly on tree-ensete dominated homegarden agroforestry systems in Ethiopia. The highest SOC in homegarden agroforestry practice is also linked with their relatively higher species diversity (Bajigo et al. 2015). However, the relationship between carbon density and biodiversity is generally less known. Therefore, this study was conducted to: (1) quantify carbon stock, in tree-ensete dominated homegarden agroforestry practice along an elevation gradient; (2) determine the woody species diversity and (3) determine the relationship between woody species diversity and soil and biomass carbon stock.

Materials and methods

The study area

The study was conducted in Hawassa Zuria District in Sidama Zone, Southern Nations, Nationalities and Peoples’ Regional State (SNNPRs), Ethiopia (Fig. 1). Geographically, it is situated between 6° 56′ 16″ and 7° 9′ 10″ N latitude and 38° 21′ 41″ and 38° 24′ 15″ E longitude and the altitude extends from 1630 to 2000 m above sea levels. The mean minimum and maximum monthly temperature of the study area is 13.8 °C and 27.8 °C, respectively (NMA 2017). The mean annual total rainfall is 935 mm with main wet season from April to September usually preceded by a less pronounced wet period in December to February (Fig. 2). Soil type of the study sites can be broadly categorized as Nitosols (SZPED 2004).

Fig. 1
figure 1

Map of the study area (the solid area in the study district’s map represents the sample plots)

Full size image
Fig. 2
figure 2

Temperature and rainfall of nearest metrological station of Hawassa Zuria District (NMA 2017)

Full size image

Sampling techniques

A stratified sampling method was used to collect vegetation and soil samples. The study site was stratified into three elevational category; lower (1630 to 1753 m), middle (1754 to 1876 m) and upper (1877 to 2000 m) to obtain homogenous units. After stratification, two hamlets (Table 1) were selected randomly from each elevation category. Seven homegardens, dominated by trees, coffee (Coffea arabica L.) and ensete, were randomly selected per hamlet and for each of these, a plot of 20 m × 20 m was sampled to inventory of woody species. This in turns helped to determine the C stocks and diversity, and the relationship between them, Area of the homegardens ranged from 0.18 to 7.5 ha.

Table 1 Villages selected for sampling along elevation gradients of Hawassa Zuria District
Full size table

Sampling design, and woody species identification and measurement

All woody species with a diameter at breast height (1.3 m above the ground) ≥ 5 cm in the 400 m2 plot (Macdicken 1997) were measured using a caliper or a diameter tape except for coffee. The diameters of coffee and ensete were measured at 40 cm (Negash et al. 2013) and 10 cm heights above the ground (Negash et al. 2013), respectively. For smaller ensete individuals and shrubs, five plots of 5 m × 5 m, placed at the four corners and in the center, were laid within the 400 m2 plot (Molla et al. 2017). Elevation was recorded using GPS in every sampling plot and height of woody species was measured with a clinometer (Suunto PM-5/360 Pc, Finland). Woody species determination was done with the help of local communities, and identification of the scientific names of species was carried out using books of Flora of Ethiopia and Eritrea (Edwards et al. 1995; Hedberg et al. 2004).

Soil sampling design

Soil samples for soil OC determination were collected using “X” design at 0–30 cm and 30–60 cm depths in every plot. The samples were taken from five sampling points within a 1 m × 1 m area with an auger in the middle of each of the 400 m2 plot. The five soil samples were then composited at each depth within a plot (Roshetko et al. 2002; Takimoto et al. 2008). Therefore, 84 composite (3 kebele * 2 hamlet * 7 plot * 2 depth) soil samples were used for soil OC determination. For bulk density (BD) measurement, soil samples were taken by using 10 cm length and 7.5 cm internal diameter of core sampler (Roshetko et al. 2002).

Biomass and C stocks estimation

We used locally developed general multi-species generic allometric equations to estimate above ground biomass (AGB) and below ground biomass (BGB) of the plant species in the agroforestry systems. We were unable to sample destructively as the owners of the homegardens did not allow us to do so. Moreover, because of high species richness in tropical agroforestry, it is difficult to use species-specific regression models (Brown and Schroeder 1999). Therefore, nondestructive and generic biomass estimating models were used for both AGB and BGB estimation. The best equations for this study were selected based on similarity in rainfall distribution between the study area and where the model was developed, prediction errors, coefficient of determination (R2), simplicity of the models to use, and sample size. Accordingly, AGB and BGB of all trees and shrubs were calculated using the equations developed in western Kenya by Kuyah et al. (2012), respectively. The AGB and BGB for coffee and ensete were estimated using allometric equations developed in southern Ethiopia by Negash et al. (2013), respectively. Biomass C stock was estimated as 48% of the biomass stock for woody species (Kuyah et al. 2012), 49% for coffee (Negash et al. 2013) and 47% for ensete (Negash et al. 2013).

  • AGB = 0.091 * d2.472, R2 = 0.98, n = 72; (Kuyah et al. 2012)

  • BGB (including coffee) = 0.048 * d2.303; R2 = 0.96, n = 72; (Kuyah et al. 2012)

  • AGBcoffee = 0.147 * d240; R2 = 0.80, n = 31; (Negash et al. 2013)

  • ln(AGBensete) = − 6.57 + 2.316ln(d10) + 0.124ln(h), R2 = 0.91, n = 40; (Negash et al. 2013)

  • BGBensete = 7 * 10−6 * d4.08310; R2 = 0.68; n = 40; (Negash et al. 2013)

where AGB, aboveground biomass, BGB belowground biomass, kg/plant, d, d10, d40 diameter at respectively 130 cm, 10 cm, and 40 cm height, total height, m.

Soil organic carbon (SOC) determination

For soil C analysis, soil samples were air-dried and passed through a 2 mm sieve. The, SOC was determined according to the Walkley–Black method (Walkley and Black 1934) using the standard procedures followed by the Wondo Genet College of Forestry and Natural Resources Soil laboratory. Soil BD was determined using core sample method (Blake 1965). The soil samples were oven dried at 105 °C for 24 h. The mass of oven dried soil divided by the total volume of core was used to get the BD. SOC was estimated using Pearson et al. (2005). The C stock values for the two layers (0–30 cm and 30–60 cm) were converted into Mg ha−1 and summed to give the C stock for the 0−60 cm layer (Negash and Starr 2015). The total C stock of the tree-ensete dominated homegarden agroforestry practices were the summation of AGBC, BGBC and SOC stocks. CO2 equivalent of C is calculated as the ratio of molecular weight of CO2 (44) to molecular weight of carbon.

Data analysis

Woody species diversity was analyzed by using Shannon diversity index (), evenness index (), species richness (), and Simpson diversity index. Species richness is the total number of species in the community (Krebs 1999). Shannon–Wiener diversity index, species richness and evenness were computed to describe species diversity (Shannon and Weiner 1949). Similarity indices measure the degree to which the species compositions of different system are alike. The Sorensen and Jaccard coefficient of similarity (Ss) was computed using Kent and Coker (1992). Importance Value Index (IVI) was computed by summing up relative density, relative frequency and relative basal area (Saikia et al. 2012). The Important value index (IVI) is a composite index based on the relative measures of species frequency, abundance, and dominance (Kent and Coker 1992). It indicates the significance of species in the system.

Statistical analysis

A one-way analysis of variance (ANOVA) was used to test for significant sources of variation in AGB and BGB carbon, soil organic C stock, diversity index, and IVI along the elevation gradient. In order to meet the assumptions of normal distribution and homogeneity of variances, data on biomass C and total C stock were ln-transformed before statistical analysis was undertaken. The significance difference between the means was tested using the Tukey. Pearson’s correlation test was also run in order to determine the relation between woody species diversity and C stock. All statistical computations were made using Statistical Package for Social Science (SPSS) software version 20.

Results

Woody species diversity

A total of 49 woody species that belong to 31 families were identified and recorded in the tree-ensete dominated agroforestry practices. There were 34 tree species (69%) and 15 shrub species (31%) recorded in the study site. The families Fabaceae and Rutaceae were represented by 5 (10%) species each, while the families Euphorbiaceae, Rosaceae, Myrtaceae and Moraceae were represented by 3 (6%) species each. The families Boraginaceae and Musaceae were represented by 2 (4%) species each and the remaining 23 families were represented by only 1 (2%) species each. Woody species richness in the middle elevation was significantly (p < 0.05) different from both upper and lower elevations. In terms of woody species distribution across elevation, upper, middle and lower elevation category had 36 (36%), 28 (27%) and 37 (37%) species, respectively.

Diversity indices along elevation gradient

The diversity indices, Shannon, Simpson and evenness significantly (p < 0.05) varied between the three elevation gradients. The lower elevation was more diverse and significantly higher (p < 0.05) in Simpson and Evenness index than the two other studied elevation gradients (Table 2). The pooled Shannon diversity index of the tree-ensete dominated agroforestry was 1.87 with the evenness of 0.81. The tree-ensete dominated homegarden agroforestry practice had high similarity index between the elevation gradients. The middle and lower elevation categories were particularly more similar (74%) in species composition (Table 3).

Table 2 Woody species diversity indices (mean ± SE) per plot of the tree-ensete dominated homegarden agroforestry practice in Hawassa Zuria District
Full size table
Table 3 Sorensen’s similarity Index (%) among elevation gradients in Hawassa Zuria District
Full size table

Important value index (IVI)

The species that were identified in the tree-ensete dominated homegarden agroforestry practices showed a variation in IVI in each elevation category (Table 4). Within each elevation category Ensete ventricosum plant species had highest IVI than the other species.

Table 4 Jaccard’s similarity Index (%) among elevation gradients in Hawassa Zuria District
Full size table

Aboveground and belowground biomass C stock along elevation gradient

The C stock in AGB and BGB of woody species varied along the elevation gradients. Both above and belowground biomass C stocks significantly (p < 0.05) increased with increasing elevation gradient (Table 6). The total biomass C stocks (above and belowground) were estimated to 50, 30 and 20 Mg ha−1 for upper, middle and lower, respectively. The maximum AGB and BGB C stocks were recorded in the upper elevation category followed by middle category and lower elevation category. Upper elevation gradient had 42% and 26% higher mean AGB C stock than lower and middle elevation categories, respectively (Table 5). Similarly, the mean BGB C stock in the upper elevation category was 38% and 22% higher than the lower and the middle elevation categories, respectively (Table 5). In each elevation gradient, 98% of the total biomass carbon was attributed to trees (Table 6). In the lower elevation category, ensete and coffee contributed only 1.2% and 0.7% to the total biomass C, respectively. Similarly, the contribution of ensete and coffee to the total biomass C were 0.7% each in both the upper and middle elevation categories.

Table 5 The five top species with the highest IVI in each elevation category
Full size table
Table 6 The above and belowground biomass carbon stock (Mg ha−1) along elevation gradients (mean ± SE)
Full size table

Soil organic carbon stock along elevation gradient

The mean SOC stock was significantly different (P < 0.05) along the elevation gradients. The upper elevation had 20% and 4% more SOC than lower and middle elevation categories, respectively (Table 7).

Table 7 Contribution of trees, ensete and coffee to total biomass carbon (Mg ha−1) in each elevation gradient
Full size table

Total carbon stock along elevation gradient

The total C stock estimates varied among the elevation categories. The total C stock significantly (p < 0.05) increased with increasing elevation gradient (Table 8). The upper elevation gradient has 26% higher C stock than the lower elevation and 10% higher than the middle elevation. The middle elevation had 16% more total C stock than the lower elevation gradient (Table 8).

Table 8 The SOC stock (Mg ha−1, 0–40 cm soil layer) along elevation gradients (mean ± SE) in Hawassa Zuria District
Full size table

Correlation between carbon stock and woody species diversity

The woody species diversity and C stock had a negative correlation with each other (Table 9). However, the correlation was not significant (p > 0.05).

Table 9 Total C stocks (mean ± SE) (Mg ha−1) along the elevation gradient in Hawassa Zuria District
Full size table

Discussion

Woody species diversity

The number of species found in this study area is comparable with that of the ensete-coffee based agroforestry practices (52 species) reported from Dilla Zuria District of southern Ethiopia (Tefera et al. 2016) and 55 woody species of traditional agroforestry practices in Dellomenna District, southern Ethiopia (Molla and Kewessa 2015). It was however; lower than the number of woody species (64) found in homegarden agroforestry practices in south-central highlands of Ethiopia (Tolera et al. 2008) (Table 10).

Table 10 Pearson correlation of woody species diversity index and C stock (Mg ha−1)
Full size table

The species richness was higher than the 24 woody species found in the Wenago District, Ethiopia (Seta and Demissew 2014), and 32 woody species found in Gununo of Wolayitta Zone, southern Ethiopia (Bajigo and Tadesse 2015). According to Wade et al. (2010), variation in woody species richness between different places could be due to site characteristics, management approach, socioeconomic issues, and farmers’ preferences for tree species. For instance, the upper story of the homegardens was dominated with different fruit and timber trees. These are related to farmers’ tree species preference. Cordia africana, Hagenia abyssinica, Croton macrostachyus, Persea americana and Mangifera indica mainly dominated the upper story of the tree-ensete dominated homegarden. Persea americana and Cordia africana were also reported to mainly dominate the shade canopy layer in the traditional coffee agroforestry of southern Ethiopia (Moges 2009).

As the name of the agroforestry system implies, ensete was common in all of the elevation categories. It is preferred by most farmers as it is drought resistance, and probably due its role in erosion control, soil fertility enhancement, income generation and staple food source (Abebe et al. 2013). The majority of the woody species were in the families of Fabaceae, Rutaceae and Euphorbiaceae. According to Bajigo and Tadesse (2015), these families are also dominant in the homegarden agroforestry in Gununo of the Wolayitta Zone, southern Ethiopia.

Woody species diversity along the elevational gradient

The highest woody species diversity was recorded in the lower elevation followed by upper and middle elevation categories. This might be due to the tendency of the farmers to grow plant species, mainly ensete (the most important food crop) and Cordia africana on their garden in middle elevation category than lower elevation. This means that woody species are more equally represented in lower elevation category. The result was contrary to the notion that Shannon and evenness indices in the homegarden practice decreased with increasing elevation gradient.

Important value index (IVI)

IVI is a cumulative index that summarizes the density, abundance, and distribution of a species. In general, Ensete ventricosum, Juniperus procera and Cordia africana were dominant species with high IVI in the tree-ensetee-dominated homegarden agroforestry system. The highest IVI of these species might be associated with their higher uses for the farmers. Linger et al. (2014) reported that species with higher IVI had relatively several uses to the farmers. The species with highest IVI recorded were Ensete ventricosum and Hagenia abyssinica in the upper elevation gradient, and Ensete ventricosum and Cordia africana in both middle and lower elevation gradients (Table 4), presumably because most farmers prefer these species for their food (ensete), environmental and market values. High IVI value of woody species in tree-ensete dominated agroforestry relates to the farmers’ conscious management of the species for its various benefits.

Biomass carbon stock along elevation gradient

The C stock varied significantly along the elevation gradients. The mean AGB and BGB C stock increased with increasing elevation. However this increase was related to the higher frequency of individuals in the higher DBH classes, given that trees in the homegardens at the middle and lower elevation are frequently exploited for different purposes.

Soil organic carbon stock along elevation gradient

The overall trend of total means SOC and biomass C stock in the study area increased along elevation gradient. This is likely attributed to the practice of cutting-off of old ensete leaves left to mulch on site and lower mineralization rates due to the higher elevation (lower temperature) of the Ensete system (Negash and Starr 2015). Besides, Negash and Starr (2013) reported in southern Ethiopia that litter inputs form C. arabica, C. africana and M. ferruginea accounted for 93% of the total annual litterfall production in the Enset-Coffee system, while C. macrostachyus and C. africana accounted for 66% of the total annual litterfall production in the Enset system. For SOC stock, higher levels at higher elevation are also consistent with previous studies. For example, Lemenih and Itanna (2004) found a strong negative correlation between temperature and carbon stocks in successional vegetation in Ethiopia, so higher stocks at the cooler higher altitudes. Our results are also in line with a recent global meta-analysis showing that elevation accounts for 71% on variation in soil C (Tashi et al. 2016). Litterfall decomposition rates are reduced at the lower temperatures prevalent at higher elevation, thus soil C retention is higher (Sheikh et al. 2009). The higher plant canopy density in the upper elevation gradient resulted in higher input of litter fall and belowground root activity (Lemma et al. 2007). However, the result of this study was opposite to other study in which SOC was reported to decrease with increasing elevation gradient (e.g. Sheikh et al. 2009).

Correlation between C stock and woody species diversity

We did not find any significant relationship between woody species diversity and carbon stocks, which was surprising as we were expecting a positive relationship between diversity and productivity. This implies that other factors has a stronger influence and that interventions that enhance woody species diversity would not necessarily lead to improvements in C stocks. Greater species diversity might not associated with greater standing C stock due to the agroforestry practices are human-managed systems with high disturbance in the system (Richards and Mendez 2014). According to Thompson et al. (2012), C stock in a system depends not only on woody species diversity, but also depends on functional diversity. These results are consistent with several other studies were C stocks depends on the increasing plant species biomass (e.g. Henry et al. 2009; Mandal et al. 2013).

Conclusions

Tree-ensete dominated homegarden agroforestry practices are the main agroforestry land use system in Hawassa Zuria District. These agroforestry practices play an important role in conserving different woody species and providing storage for biomass and SOC. The diversity index, structure and composition of woody species significantly vary among elevation gradient. The lower elevation gradient had a higher species diversity index followed by the upper and middle elevation. Above and belowground biomass C and SOC stock also varied significantly along the elevation gradient. Upper elevation category resulted in higher biomass C and SOC than lower elevation. The result of this study did not show a significant correlation between woody species diversity and C stocks in tree-ensete dominated agroforestry practices. This means that variation of C stock does not depend on the species diversity. Good management practices in the tree-ensete based agroforestry practices at all the elevation gradients is needed for conservation of the species diversity while increasing more SOC and biomass C stock. Further studies should be conducted to understand which species make a higher contribution to C stock at different parts of the elevational gradient and how management alters woody species diversity.