Savannas are mixed tree-C4 grass systems covering more than 50 percent of the African continent. The trees in this system are scattered and do not form a closed canopy allowing sunlight to reach the herbaceous ground layer (Scholes & Walker, 1993; Suttie et al., 2005). Savannas are home to a diverse community of plant and animal species, including charismatic megafauna that provide opportunities for recreation and tourism thereby supporting local and national economies on the continent (Earnshaw & Emerton, 2000; Grünewald et al., 2016). Additionally, they provide other ecosystem goods and services relevant to the wellbeing of people on the continent including food, fuelwood, medicine, water and storage of carbon (Ryan et al., 2016). However, these ecosystem services are deteriorating due to a multiplicity of factors such as changes in land use and climate change, which degrades African savannas at unprecedented rates (Estes et al., 2016; Hahn & Leßmeister, 2021).
In West African savannas, an age-old traditional land use practice exists where annual food crops replace the herbaceous ground layer in the cropping season while livestock graze in the off-season. This land use, commonly referred to as an agroforestry parkland (i.e., hereafter, parkland), provides ecosystem services comparable to that of savannas including soil fertility improvement and food security (Boffa, 1999). However, unsustainable agricultural practices such as conventional agriculture intensification and expansion are quickly degrading parklands. For instance, recent estimates indicate that tree species composition (i.e., density and diversity) in savannas as well as parklands have declined significantly(ELD-UNEP, 2015; M’Woueni et al., 2019; Mbow et al., 2015). Beyond the increased demand for arable land, excessive harvesting of fuelwood by local communities in parklands is responsible for the worrying trend and its associated effect on the disruption of ecosystem services (Lovett & Phillips, 2018).
Monitoring changes in parklands in West African savannas is critical for researchers and resource managers engaged in restoration and adaptive management for resilient ecosystems in the face of climate change. The northern savanna ecological zone of Ghana has experienced high levels of degradation in the last couple of decades especially in Parkia biglobosa and Vitellaria paradoxa populations (Amoako et al., 2015; O’Higgins, 2007; Poudyal, 2011). In an attempt to reverse this trend, Ghana has ratified a number of initiatives with huge financial commitments made towards addressing degradation, the most recent being the Ghana Shea Landscape Emissions Reduction Project (GSLERP, 2020; World Bank, 2011). Recently, farmer managed natural regeneration has been widely advocated as a strategy to restore degraded savannas (Magrath et al., 2020). This practice, which allows for the deliberate retention of trees on agricultural land, is receiving great attention in Ghana’s northern savanna ecological zone. Hence, the need to monitor woody plants in parklands is vital to enhancing our understanding of preservation and restoration efforts in savannas and the strides made so far.
Plant density, a quantitative vegetation metric, can be determined using either plot or plotless sampling methods for ground truthing at local or site scale. These sampling methods can be used to assess and detect shifts in woody plant communities in savannas monitored over time under varying spatial patterns. In this study, I used plotless sampling rather than plot sampling because they are faster to apply and require less labor (Cottam & Curtis, 1956; Engeman et al., 1994; White et al., 2008). Specifically, I compared the performance of five plotless sampling methods (Table 1) that are easy-to-apply and are relatively robust to deviations from random spatial patterns, which may occur in parklands (Engeman et al., 1994; Krebs, 2014). Thus, this study aims to provide researchers and resource managers with information concerning which plotless sampling methods produce reasonably accurate density estimates in parklands while considering the spatial pattern of woody plants.
Table 1
Summary of plotless sampling methods used in this study
Method and description | Equation | Reference |
Nearest neighbor (NN): distance between the first and second closest woody plant from a sampling point | \(\widehat{D}=1/\left[2.778{\times \left(\sum _{i=1}^{n}{r}_{i}/n\right)}^{2}\right]\) | Cottam and Curtis (1956) |
Closest individual (CI): distance between a sampling point and the closest woody plant | \(\widehat{D}=1/\left[4\times {\left(\sum _{i=1}^{n}{r}_{i}/n\right)}^{2}\right]\) | Cottam and Curtis (1956) |
Ordered distance (OD): distance between a sampling point and the third closest woody plant | \(\widehat{D}=\left(3n-1\right)/\left[\pi \times \sum _{i=1}^{n}{r}_{i3}^{2}\right]\) | Pollard (1971) |
Point-centered quarter (PCQ): distance between a sampling point and the first closest woody plant in each quadrant. Correction factor is applied if vacant quarters exist | \(\widehat{D}=4\left(4n-1\right)/\left[\pi \times \sum _{i=1}^{n}\sum _{j=1}^{4}{r}_{ij}^{2}\right]\) | Cottam and Curtis (1956), Pollard (1971), Warde and Petranka (1981), Mitchell (2015) |
Variable area transect (VAT): distance from a sampling point to the third woody plant in a specified direction with fixed transect width | \(\widehat{D}=\left(3n-1\right)/\left[w\times \sum _{i=1}^{n}{l}_{i}\right]\) | Parker (1979) |
\(\widehat{D}\) = estimated density, \(i\) = sampling point, \(j\)= number of quadrant, \(l\) = length of transect, \(n\) = number of sampling points, \(r\) = distance measured \(w\) = width of transect.