Introduction

The Sheanut tree, Vitellaria paradoxa C.F. Gaertn. (Family—Sapotaceae, syn. Butryospermum paradoxum), is a semi-domesticated, slow-growing fruit tree, and a major component of the woody vegetation in the semi-arid Sudan and Guinean zones (Lovett and Haq 2000; Masters 2002). Shea trees grow across 16 African countries in sub-Saharan Africa, from Senegal to Ethiopia (Rousseau et al. 2015) (Fig. 1).

Fig. 1
figure 1

Distribution of shea (Vitellaria paradoxa) in West, East and Central Africa (Lovett 2013 and USAID)

Full size image

In Western sub-Saharan Africa, farmers often keep shea trees as part of their parkland agroforestry systems (Aleza et al. 2015; Elias 2015; Lovett and Haq 2000; Raebild et al. 2012). These areas are sequential farm-fallow systems where crops are cultivated on the same areas as shea trees, followed by temporary fallow periods, where the trees remain, but the areas are left uncultivated to allow for soil regeneration. In addition to this farming approach, shea trees in these regions also grow in uncultivated bushlands (Cronkleton et al. 2021; Lovett and Haq 2000; Poudyal 2011). Bushlands are characterized by mixed, self-established remnant vegetation with minimal human disturbance.

Shea trees have a prominent role in the subsistence of rural households, in particular for the women who are mainly responsible for the collection of kernels of the trees and processing them into shea butter (Chalfin 2004; Schreckenberg 2004; Gausset et al. 2005; Pouliot 2012; Boffa 2015; Hammond et al. 2019). A study on the anthropic selection of shea trees in Northern Ghana shows that farmers describe shea trees as “wild”, since they rarely plant them, meaning that most trees stem from natural regeneration (Lovett and Haq 2000). However, research indicates that the shea tree, due to human selection, is a semi-domesticated species (Hale et al. 2021; Yao et al. 2020a, b). According to local farmers in northern Ghana, shea trees are selected due to their yields but other properties of the trees, such as health, size, growth, age, competition with crops and spacing allowing for mechanical soil preparation, are also important (Baziari et al. 2019; Lovett and Haq 2000).

A recent meta-analysis of agroforestry as a solution for maintaining ecosystem services in Sub-Saharan Africa showed that agroforestry systems generally deliver higher yields and improve soil fertility by providing higher contents of soil nitrogen, organic carbon and available phosphorus, as compared to non-agroforestry systems (Kuyah et al. 2019). This analysis also indicated that agroforestry practices improve the soil structure and increase the soil infiltration rates and moisture content, causing a reduction in run-off and erosion. Further, studies show that agroforestry practices improve livelihood resilience in Sub-Saharan Africa by diversification of farming activities and livelihood strategies (Kuyah et al. 2020; Quandt et al. 2019).

However, agroforestry practices are also known to cause certain trade-offs and losses. Younger trees with abundant roots in the top 30 cm soil compete with crops for nutrients and water, and certain plant species produce below-ground allelopathic compounds that are toxic to other plants and inhibit their growth (review by Atangana et al. 2014). Yet, for shea trees there is no information available on whether they produce such allelopathic compounds. Above ground, modification of the microclimate by trees plays a role, and competition for light may lower growth and yields of understory crops. Shaded and humid below-canopy microclimates promote the growth and thriving of certain plant pathogenic fungi and bacteria (ibid). Despite these risks, the creation of more stable microclimates in agroforestry systems can protect crops from climate and soil-moisture extremes. Therefore, agroforestry systems may provide economically feasible solutions as strategies for farmers to adapt to climate changes (Lin 2007). A recent study of agroforestry as a remedy for climate change adaptation showed that local farmers in northern Ghana are aware of the positive impacts that agroforestry practices may have (Apuri et al. 2018). The farmers perceived that agroforestry contributed to improved nutrient content in the soils, reduced wind and water erosion, conserved soil moisture, and increased the availability of foodstuff for the households.

In dry climate conditions, shea trees can regulate their water status. This was confirmed in a study of shea trees in an agroforestry parkland in Burkina Faso, showing that the trees efficiently controlled the sap flow and could down-regulate the canopy transpiration (Bazié et al. 2018). Maintaining a high leaf water potential is not only important for physiology and biomass growth, it also affects the development of fruits (Ratzmann et al. 2019).

Most studies on shea trees have focused on the production of butter from the nuts and the related impact on the local economy (Schreckenberg 2004; Boffa 2015; Hammond et al. 2019). Less attention has been paid to the interactions between crops and trees in the agroforestry systems, including the sustainability of shea tree populations on different land-use types around village communities. In this study, we analysed structure, density and water status of shea tree populations in three land types, namely fields, fallows and bushlands during the dry season in the village of Nakpalli in northern Ghana. Further, effects of the trees on the below-canopy microclimate and the co-cultivation of crops beneath the trees were examined. We hypothesise that canopies of shea trees create a microclimate that allows a successful co-cultivation of important produce, such as yams. Additionally, we hypothesize that land-use affects population structures and densities of shea trees.

Material and methods

Study area and community

The study was conducted in Nakpalli (8°58′ 52.1′′ N, 0° 19′ 13.4′′ E) in the north-eastern part of Ghana (Fig. 2a, b) and included 2 months of fieldwork from January 20 to March 2, 2022. Data collection occurred in the dry season and no rainfall occurred during the stay.

Fig. 2
figure 2

Maps of Ghana. The village of Nakpalli in North-Western Ghana is marked with a red pinpoint (a). Satellite image of Nakpalli and the surrounding areas and locations for measurements of shea tree size distributions and densities are marked (b). Maps obtained from Google Maps and Google Earth

Full size image

Nakpalli is a medium-sized village situated in the northern region of Ghana, about 180 km southeast of Tamale and less than 30 km from the border to Togo. The village has about 4000 inhabitants, living in around 318 households. The population is dominated by the Dagomba ethnic group, with a minority of Fulani, and a few Kokomba and Ewe. The main activities in the village include farming, small businesses, petty trading, and shea nut collection. Crop cultivation is carried out by farmers who manage and decide over particular land plots, following a customary land tenure system (Amanor 2009).

Characteristics of the land and climate

Nakpalli is located in the ecoregion of Sudanian Savanna (Liu et al. 2017). Farming seasons are defined by a mono-modal rainfall system with only one rainy season that occurs between July and September. This constitutes the major farming season. A minor farming season occurs in the following drier months. The average annual rainfall is 1000 mm (Ministry of Food and Agriculture in Ghana 2016). According to the International Soil Reference and Information Centre (ISRIC 2023), soils in this region are ferric luvisols, which typically are characterized by an argic horizon overlaid by loamy sand.

The vegetation is characterized by trees and bushes with a herbaceous layer of forbs and annual and perennial grasses (Sawadogo et al. 2010). Agriculture is predominated by rainfed smallholder farming, and most farm holdings do not exceed 2 hectares in size (ibid.). Farmers in Nakpalli typically manage agroforestry parkland systems by seasonal crop rotation with sequential cultivation and fallow periods. Farmers keep their land fallow for an average duration of 2.8 years and cultivate the lands for 3 years on average (Stopponi and Jepsen 2022). Almost every household has one or more farmland plots situated around the village, where they mainly cultivate yam (Dioscorea rotundata (Poir.) J. Miége), maize (Zea mays L.), sorghum (Sorghum bicolor (L.) Moench), cassava (Manihot esculenta Crantz) and millet (Pennisetum glaucum (L.) R. Br). The parklands and surrounding areas in Northern Ghana are dominated by tree species of anthropogenic interest, such as shea (Vitellaria paradoxa Gaertn.), locust bean (Parkia biglobosa (Jacq.) R.Br. ex G. Don), cashew (Anacardium occidentale L.), mango (Mangifera indica L.), neem (Azadirachta indica A. Juss) and teak (Tectona grandis L.f.) trees. Shea trees were by far the most dominant tree species in the agroforestry systems, followed by locust bean trees, which were considerably less prevalent.

Size distributions and densities of shea trees in fields, fallows and bushlands

To obtain knowledge on shea tree population structures and to evaluate size-distributions, densities and regeneration of tree populations in the area, the circumferences and numbers of minor-sized shea trees and seedlings were registered on three different land types as follows: (i) Fields: Shea tree agroforestry systems with currently growing crops; (ii) Fallows: Temporarily uncultivated shea tree agroforestry systems; and (iii) Bushlands: Lands that were minimally influenced by human activity. The three land types are shown in Fig. 3.

Fig. 3
figure 3

A field with yam mounds and shea trees (a), shea trees in a 3-year-old fallow (b) shea trees in far bushlands (c)

Full size image

A total of 10 plots of fields, 10 plots of fallow lands and 10 plots of bushlands, each containing 20 shea trees as a minimum, were selected. Plot sizes were measured using the perimeter function on a Garmin eTrex(R) 10 Global Positioning System (GPS) (www.garmin.com).

In the defined plots, circumferences of the trunks at breast height (130 cm) of all shea trees > 130 cm height were measured using measuring tape and converted to diameter at breast height (dbh). Five subplots of 5 × 5 m were placed in corners and centres of the main plots. All shea trees < 130 cm in height, including seedlings and coppices within the sub-plots, were counted and used for estimating the total tree number for the whole plot by extrapolation. A graphic overview of the sampling setup is presented in Fig. S1.

The average duration of land-use in the registered areas was 3.2 ± 0.57 (n = 10) years of fallow and 3.1 ± 0.57 (n = 10) years of cultivation. The spatial distribution of locations of measurement of densities and size-distributions of shea tree populations according to land-use type are shown in Fig. 2b.

To describe and compare the shea tree size distributions in fields, fallows and bushlands, the continuous datasets of shea tree diameters at breast height in the three respective land-use types were fitted to a Weibull distribution, and shape and scale parameters were extracted. Weibull functions are described as follows:

$$f(d) = \frac{\lambda }{\alpha }\left( {\frac{d}{\alpha }} \right)^{\lambda - 1} e^{{\left( { - \left( {\frac{d}{\alpha }} \right)^{\lambda } } \right)}}$$

In the equation, f(d) is the probability density, d is the tree diameter at breast height, λ is the parameter determining the shape of the distribution, and α is the scale parameter determining the width and height of the distribution.

To detect whether the shape and scale parameters provided by the Weibull distribution curves were different for the three land-use types, a one-way ANOVA test with the land-use type as the single factor was applied. Dependent data were the respective scale and shape parameters for each individual land plot (Appendix A). To detect possible effects from the land-use on densities of shea trees larger than 130 cm in height and of minor shea trees and seedlings with height below 130 cm, respectively, a one-way ANOVA test was applied with the land-use type as the single factor and the shea tree (> or < 130 cm) density (expressed in trees/ha) for each individual land plot as the dependent factor (Appendix B).

Below-canopy microclimate

Between January 30 and February 3, 2022, three different climate variables were measured underneath and outside shea tree canopies in the fields to evaluate the microclimate below the shea trees. Soil surface temperature and air temperature were measured using a Laserliner CondenseSpot Plus (082.046A) (www.laserliner.com). A LI-COR Quantum sensor (LI-250A Light Meter) (www.licor.com) was used for measuring the penetration of Photosynthetic Active Radiation (PAR) through the shea tree canopies. Following Tanga et al. (2014), the data collection occurred at midday, in this case at 12.05–12.25 pm, when the sun was at its highest point.

The climate variables were measured under the shea trees in the following zones: next to the trunk and at 1/3, 2/3 and 3/3 of the canopy radius. Five shea trees with similar characteristics were chosen for the microclimate data collection (Table 1). This was done in four different angles, resulting in 16 points of measurement per tree and 80 measurements in total per parameter (Fig. S2). For control, PAR and soil surface temperature were measured in full sunlight outside the canopy, and air temperatures were measured in shaded conditions outside the canopy.

Table 1 Characteristics of the trees used for measuring below canopy microclimate
Full size table

Weather data were registered at midday for the 5 days of data collection using a portable climate station (GMR Strumenti, Italy). The average values of PAR, air temperature and air humidity are shown in Table 2.

Table 2 Mean PAR, air temperature and air humidity at midday on the days for microclimate data collection (January 1 to February 3, 2021)
Full size table

One-way ANOVA tests with random effects were applied to detect the possible effects of shea trees on soil temperature, air temperature and PAR at different distances to the trunk, as compared to outside shea tree canopies (Appendix C).

Yields of yams cultivated underneath and outside shea tree canopies

Yam is an essential subsistence and cash crop in the village and is widely cultivated in the agroforestry systems. Yams were planted in approximately 50 cm tall and manually prepared mounds at 1.5 m distance as can be seen in Fig. 3a. To evaluate whether yields of yams were affected by shea trees, weights of yams cultivated in mounds underneath and outside shea tree canopies were determined. Yams were weighed from 25 mounds located outside shea tree canopies and from 25 mounds underneath shea trees using a SALTER Hanging Scale Model 235 6S. The mounds were located in the following zones under shea trees: at 1/3 canopy radius (n = 8), 2/3 canopy radius (n = 8), 3/3 canopy radius (n = 9) and on open land (n = 25).

To obtain information on yam yields per area, the number of mounds per hectare was estimated by extrapolating the number of yam mounds per 5 × 5 m2 plots (n = 5). To detect possible effects of shea trees on yam yields at different distances to the trunk (expressed in zones), compared to yields outside shea tree canopies, a linear mixed model, i.e., one-way ANOVA test with random effects, was applied to the data (Appendix D).

Pre-dawn water potentials of shea tree leaves in fields, fallows and bushlands

To evaluate the water status of shea trees during the dry season and determine whether this differs between shea trees located in fields, fallows and bushlands, the predawn leaf water potentials (ψp) were measured in all three land-use types. ψp was measured in 2 leaves from the lower canopies of 10 shea trees per land-use type, resulting in 20 measurements for each land-use type and 60 measurements in total. Measurements were made using a Pump-up Pressure Chamber (PMS Instrument Co., OR) before dawn (6.15 am), i.e., between 4.28 am at earliest and 5.42 am at latest.

The precautions for data collection stated by Turner (1988) were considered. Thus, right before harvesting a leaf, it was sealed loosely by a zip-lock aluminium bag to avoid evaporative water loss. The leaf stem was cut with a razor blade, ensuring that the length of the petiole external to the sealing was as short as possible. The leaf was placed in the pressure chamber within 30 s. The pressure chamber was filled slowly, to avoid heating and to capture the exact endpoints.

Locations for measurements of ψp are shown in Fig. S3. The fallow land was located 166 m above sea level (m asl), the field was located at 147 m asl, and the bushland was located at 176 m asl. To detect the effects that land-use may have had on shea tree predawn leaf water potentials, a linear mixed model (one-way ANOVA test) with random effects was applied (Appendix E).

Statistical analyses

The statistical analyses were performed using the software RStudio (version 4.2.0). Figures were generated using the R package ggplot2 (https://www.R-project.org/). Weibull curves for distribution of tree size classes were estimated in the R-software using the package fitdistrplus. Weibull curves were fitted to size distribution histograms (trees/ha) by overlay in Microsoft Excel version 16.59.

Results

Densities of larger shea trees in fields, fallows and bushlands

The density of larger shea trees (> 130 cm height) varied between bushland, fallows, and fields. In the natural and non-cultivated bushland, the mean density of shea trees taller than 130 cm was significantly higher than in fields (p = 0.0007) and fallows (p = 0.007) (Fig. 4a). The mean shea tree densities in fields and fallows were not significantly different from each other (p = 0.38). The median density of larger shea trees was lowest in the fields, marginally higher in fallows, and clearly highest in the bushlands. An overview of the average values of the shea tree densities in the different land-use types is given in Supplementary Information (Appendix A).

Fig. 4
figure 4

Boxplots of densities of larger shea trees (a) and minor trees (b) according to land-use. Significant difference between the mean density in bushlands, as compared to fields and fallows, respectively, is marked (**p < 0.01, ***p < 0.001). The boxplots show 25th and 75th percentiles (end of boxes), with a line at the median and error bars defining the 10th and 90th percentiles. Filled circles indicate outliers. Trees in 30 fields, fallows, and bushlands were registered

Full size image

Densities of minor shea trees and seedlings in fields, fallows and bushlands

The number of shea tree seedlings and saplings (< 130 cm in height) varied significantly between bushland, fallows and fields (Fig. 4b). Higher mean densities of seedlings and small shea trees were found in bushlands as compared to fields (p = 0.005) and fallows (p = 0.0006), while no difference was found between densities in fields and fallows (p = 0.42). The median density of seedlings and minor shea trees was four times higher in bushlands than in the other land-use types. An overview of the average values of the shea tree densities in the different land-use types is given in Appendix B.

Shea tree size distribution according to land-use type

Farmers manage the fields by mechanical soil preparation by tractor. In addition to this, the farmers protect shea trees from fires, cutting and other damages. These practices are not undertaken in the unmanaged bushlands.

The circumferences at breast height of trunks of a total of 682 shea trees were registered on 18.74 ha to determine distribution and tree densities in the three land-use types (Table S1). Histograms of the shea trees showed that the tree populations in the bushlands were characterised by a markedly higher density of trees in the lower diameter class (saplings with diameter less than 10 cm), while the abundance of trees in higher size classes showed a rather steep decline (Fig. 5c). In the fields and fallows, trees were more evenly distributed over the different size classes, dominated by trees in size classes 10– < 20 cm and 20– < 30 cm (Fig. 5a and b). These differences in the size class distributions were reflected in the scale parameters of the Weibull curves causing the “stretch” of the curves. The mean scale parameter for bushlands was lower than the mean scale parameters for fallows (p = 0.01) and fields (p = 0.001). The similarities in the size distributions of fields and fallows were underlined by the fact that the mean scale parameters in these land-use types did not differ from each other (p = 0.47).

Fig. 5
figure 5

Histograms of shea tree populations in fields (a), fallows (b) and bushlands (c) and the fitted Weibull curves. λ is the shape parameter (1 < λ < 3.6 indicates right skewed distributions), and α is the scale parameter (determines the width and the height of the distribution)

Full size image

Pre-dawn leaf water potentials

Whether the trees were growing in the fields, fallows or bushlands, similar pre-dawn leaf water potentials were found (p = 0.085) (Table 3). The mean values of ψp for shea trees in both fields, fallows and bushlands ranged from − 1.03 to − 0.88. These close-to-zero values indicate that the leaves were not experiencing severe drought stress in any of the three land-use types.

Table 3 Average predawn leaf water potentials for the three land-use types and corresponding 95% confidence intervals
Full size table

Effects of shea tree canopies on microclimate

The shea trees had a significant effect on soil temperature beneath the canopies, relative to the soil temperature outside the canopies. The mean soil surface temperature outside shea tree canopies was high; 57.5 °C as compared to mean temperatures of 33.0 °C next to the trunks and 39.6 °C at the periphery of the canopy (Fig. 6a). Thus, the soil surface temperatures were reduced by 31.2–42.5% underneath shea tree canopies, as compared to the open fields. The temperature difference at various distances to the trunk and the open land was statistically significant (p = 0.0006). The random effect of trees accounted for 51.5% of the total variance, while significant effects of the site could not be found. The residual variance, corresponding to the within-group variance, accounted for 48.5% of the total variance.

Fig. 6
figure 6

Boxplots of soil surface (a) and air temperatures (b) in the different zones relative to the shea tree canopy (CR = canopy radius) and outside shea tree canopies. Significant differences between mean soil surface temperature in zones underneath canopies and temperatures outside shea trees are shown (*p < 0.05, **p < 0.01). The difference between mean air temperatures in different zones underneath canopies and temperatures outside shea trees was not significant (ns = not significant). The boxplots show 25th and 75th percentiles (end of boxes), with a line at the median and error bars defining the 10th and 90th percentiles. The filled circles indicate outliers

Full size image

The canopies also affected the air temperature beneath the trees. Below the canopies, the median air temperature was stable at 37–38 °C but was 41 °C outside the canopies (Fig. 6b). This corresponds to a 9–10% reduction in air temperatures underneath shea tree canopies, as compared to outside on open land. Those differences were borderline significant (p = 0.06).

Effect of shea trees on PAR beneath canopies

Canopies caused a significant, gradual reduction in photosynthetically active radiation (PAR) in the different canopy zones, relative to outside the canopies. Below the canopies, the mean PAR was 421 µmol/s, while 1611 µmol/s were measured outside the canopies (p = 1.69 × 10–5), corresponding to a 73.9% PAR reduction relative to outside canopies (Fig. 7). The random effect of trees accounted for 34.5% of the total variance, while the random effect of the site did not exist in this case either. The residual variance therefore accounted for most of the total variance.

Fig. 7
figure 7

Boxplots for visualizing the PAR in the different zones relative to the shea tree canopy (CR = Canopy radius) and outside shea tree canopies. The significance of the difference between mean PAR in different zones underneath canopies and PAR outside the trees is shown (*p < 0.05, **p < 0.01, ns = not significant). The boxplots show 25th and 75th percentiles (end of boxes), with a line at the median and error bars defining the 10th and 90th percentiles. Filled circles indicate outliers

Full size image

Co-cultivation of yam and shea trees

Across all four zones below the canopies, the average yield was 2.3 kg per mound (corresponding to 11.5 t/ha), which is 44.7% lower than the yam production of 4.13 kg per mound (or 20.82 t/ha) outside the canopies. There was a non-significant tendency of reduced median values of yam yields closer to the tree trunk (Fig. 8). The random effects of tree and site were negligible, and the residual variance accounted for 90.33% of the total variance.

Fig. 8
figure 8

Boxplots of yam yields in the different zones below the shea tree canopies (CR = canopy radius) and outside shea tree canopies. The difference between yam yields of mounds in the different zones was not statistically significant (ns = not significant). The boxplots show 25th and 75th percentiles (end of boxes), with a line at the median and error bars defining the 10th and 90th percentiles

Full size image

Results of the average yam yields per mound and per hectare for the different zones, determined with the linear mixed model, are given in Supplementary Data (Appendix B).

Discussion

Influence of agroforestry practices on shea tree densities

This study on agroforestry systems with shea trees in the savannah region of northern Ghana demonstrated that the abundance of trees is affected by human practices and management of the land. In natural, non-cultivated bushland there was a higher density of both large and small shea trees, as well as seedlings, as compared to fields and fallows.

In several western Sub-Saharan African countries, farming activities lead to mechanical removal of shea tree seedlings (Djossa et al. 2008; Kelly et al. 2004; Lovett and Haq 2000). To ensure the regeneration of shea trees, mechanical soil preparation, controlled burning of farmlands and other human disturbances should be reduced by keeping agroforestry systems fallow for shorter or longer periods (Aleza et al. 2015; Elias 2013). Supporting this, studies show that higher densities of shea tree seedlings may occur in fallows than in fields (Aleza et al. 2015; Raebild et al. 2012). In our study, the densities of seedlings did not differ in fields compared to fallows, and the densities of seedlings were four times lower than in bushlands. Systematic collection of nuts in fields and fallows may partly explain the lower seedling densities on those land types, relative to bushlands located outside the village. The similar densities of seedlings in fields and fallows may have been caused by shortened fallow periods. According to Djossa et al. (2008), shorter fallow periods increase human disturbance of the tree seedlings and reduce tree regeneration in the agroforestry systems. At the beginning of a subsequent cultivation period, seedlings of shea trees and other tree species will still be too small to be distinguished from weeds, and they may therefore be removed unintentionally by mechanical soil preparation.

A tendency of shortened fallow periods is also seen elsewhere in western Sub-Saharan Africa due to population expansion and a higher need for land (Djossa et al. 2008; Lovett et al. 2018). Population expansion was also confirmed in the Nakpalli area, causing an elevated demand for farmlands and conversion of bushlands into farmlands.

Shea tree populations in all land types at Nakpalli were dominated by trees with lower diameters as indicated by the positive asymmetry (1 < λ < 3.6) of the Weibull curves, although this tendency of distribution was more prominent for bushlands. Positive asymmetry is characteristic of populations with good rejuvenation and stable densities (Condit et al. 1998). Given that shea trees can become 200–300 years old, a reduction in density of shea tree seedlings and saplings in parklands does not cause an immediate reduction of the adult tree population (Djossa et al. 2008). Yet, we raise concern that a continued removal of seedlings poses a potential risk to the replacement of aging shea trees in these agroforestry systems within a few decades. A potential solution for the farmers to address this challenge might be to mark a few shea tree seedlings in the fields before soil preparation to make them visible from inside the tractors, enabling the diver to steer clear of them.

Protection of shea trees towards other threats

Fire poses a potential risk to the survival of shea trees around Nakpalli, either caused by lightning or by deliberate burning of farmland for preparation of land before a new cultivation period. The burning of farmlands is intended to be controlled but according to farmers, these burnings sometimes turn into wildfires. Around Nakpalli, there were visible indications that fire had occurred in all land types, but the landowners stated that they protect shea trees in fields and fallows against fires, mainly by weeding around them (Stopponi and Jepsen 2022). However, this act of protection was not reflected in the tree population structure in farmland relative to bushland in our study. Yet, fire may not be critical to the viability of well-established shea trees. After surviving to the 3–5 years after germination, shea trees become highly fire-resistant (Hatskevich et al. 2011). The high densities of shea tree seedlings found in the bushlands regardless of wildfires, suggests that high recruitment of seedlings may have counteracted a loss of rejuvenation in these lands.

Another threat to shea trees is cutting for production of charcoal, firewood and other purposes as is locally practised in Ghana, affecting both density and size of the trees (Asare et al. 2022; Cronkleton et al. 2021; Dapilah et al. 2019). However, cutting of shea trees is uncommon and considered unacceptable in the Nakpalli area, as also reported among other rural farmers and community members in Northern Ghana (Hansen et al. 2012).

An inherent property that appears to protect shea trees is their ability to tolerate water stress. In the middle of the dry season, after at least 2 months without any precipitation, our observations of close-to-zero pre-dawn water potentials of the leaves indicated that the shea trees were not under severe drought stress. This does most likely reflect a high drought resistance of shea trees in the region. The capacity of shea trees to regulate their water status was also confirmed in a study of shea trees in an agroforestry parkland in Burkina Faso (Bazié et al. 2018).

Below canopy microclimate and effects on yields of understorey crops

Shea trees in the Nakpalli area created a microclimate below the canopies, causing a reduced soil and air temperature, and lower PAR. Changed microclimates under shea trees have previously been assessed in Burkina Faso (Bayala et al. 2002), but this is the first study of below-canopy-microclimates of shea trees in Ghana.

The lowered PAR beneath the shea tree canopies may affect the yields of understorey crops negatively, as indicated for a range of different crops in shea tree agroforestry systems (Baziari et al. 2019; Gnanglè et al. 2013; Seghieri 2019). In Nakpalli, the observed co-cultivation of yam beneath the shea canopies yielded on average 11.5 t/ha as compared to 20.8 t/ha outside canopies. The overall yam yields found in this study resemble those found in other districts in Ghana, averaging 18 t/ha (Danquah et al. 2022). Even if crop yields may be affected negatively right below shea tree canopies, the effect may not occur on field scale. Boffa et al. (2000) found that adequate balancing of shea tree densities can even increase the yields of sorghum on field scale. However, if yields of understorey crops are reduced, this may not necessarily be considered an overall loss, since income from shea nuts may outweigh the downsides of keeping shea trees in the farmlands (Baziari et al. 2019; Gnanglè et al. 2013; Seghieri 2019). Co-cultivation of crops and trees also provides the benefit of diversification of subsistence and income sources, an approach that is deeply rooted in the local historical and cultural contexts (Kuyah et al. 2020; Quandt et al. 2019). Furthermore, stabilised microclimates in agroforestry systems may contribute as strategies for farmers to adapt to climate extremes (Lin 2007).

Conclusion

Collection of shea nuts is an important supplementary household income to villagers in Nakpalli. In addition, the microclimate created below the shea tree canopies allows the cultivation of crops, but the yields may be reduced due to lower penetration of light, as suggested for yams in this study and also observed for other crops in similar studies. To optimize the agroforestry systems, cultivation of adapted crops, including shade-tolerant species, is an option. Alternatively, proper balancing of the density of shea trees in these systems is advantageous since it may permit co-cultivation of crops, whilst still benefitting from shea tree income and subsistence. Among benefits of trees in cropping systems are the establishment of agroforestry microclimates that help protect crops from climate and soil-moisture extremes and that may provide economically feasible solutions as strategies for farmers to adapt to climate extremes.

Even in the middle of the dry season, the shea trees in the Nakpalli area appeared not to be affected by water stress, as indicated by the close-to-zero predawn leaf water potentials, irrespective of the location of the trees. The thriving of shea trees underlines the benefit and importance of conserving this well-adapted indigenous tree species in agroforestry systems in western Africa. However, we raise a concern that increasing population pressure in this area is likely to reduce the length of fallow periods, threatening the long-term regeneration of shea tree populations. Yet, the shea tree is a highly valued species in this region, and if care is taken to protect seedlings from removal by mechanical soil preparation, or deliberate burning of land for soil preparation, this will increase the sustainability of shea tree populations in agroforestry systems in Northern Ghana.