The Akak forest area is located between latitude 5˚20' - 5˚25'N longitude and 9˚ 12' - 9˚30'E latitude, in the Eyumojock subdivision, Manyu Division, Southwest Region of Cameroon as shown in Figure 1 . Vegetation consists of semi deciduous lowland rainforest of the Guineo-Congolian type ( Kenfack et al., 2006 ). Precipitation is unimodal, with an annual average around 4100 mm ( Nchanji & Plumptre, 2003 ) and a three-month dry season from December to February ( Groenendijk, 2015 ). The topography is relatively flat. Although human intervention through the establishment of large plantations of cash crops (Palm oil, coffee), as well as natural factors such as elephant disturbance and windfalls have created large gaps in these forests. Logged forest sites are located in the “heart” of the MPL (Mukete Plantations Limited) and the forests of this area have undergone logging both formal and informal from 1995 to date.

49 sprouted stump were identified randomly as shown in Figure 2 ( Fuashi et al., 2020 ; Ayamba et al., 2019 ). 20 m × 20 m plots were demarcated along a canopy gaps for each sprouted stump, the plots were established in such a manner that the sprouted stumps will be in the middle. For each of the selected 49 sprouted stump, indirect measurements of canopy cover were performed in the 49 plots of 20 m × 20 m (0.04 ha), giving a total of 1.96 ha of land covered. Galaxy S3 smartphone with an built-in Infinix ZERO 4 fish-eye lens with 198˚ view angle equidistant projection was used to take photos as shown in Figure 3 ( Bianchi et al., 2017 ). The fish-eye lens was mounted on the phone camera and photograph were taken at a fixed height of 1.3 m ( Bianchi et al., 2017 ) as shown in Figure 3 . The smartphone was held by hand, keeping it leveled and pointing upward as best as we could ( Bianchi et al., 2017 ). Pictures were taking using the automatic exposure with good contrast between sky and canopy ( Hale et al., 2009 ). Pictures

were takenly immediately after reaching the point ( Planchais & Pontailler, 1999 ). The smartphone pictures had pixel dimensions of 3.1341 × 1.624. The smart phone was used because smartphone hemispherical photographs is comparable to hemispherical photographs using traditional cameras ( Tichý, 2015 ) and it take only diagonal photographs which is suitable for rectangular or square plots following the definition of ( Schneider et al., 2009 ). Four pictures were taken per plot randomly as shown in Figure 4 , two pictures taken at north-south direction and the other two pictures east-west direction with the aid of a compass ( Bianchi et al., 2017 ). Photographs were taken at each sampled plots (that was average time the team spent on each plot), trying to catch the best backlighting conditions. A total of 196 images (pictures) were taken in Febuary - April 2018, on 49 plots of 20 m × 20 m in size. Hemispherical photos were taken under completely overcast sky conditions and when the sun was under the skyline in order to minimize glare from Sunlight ( Bianchi et al., 2017 ). Fieldwork took place the whole day resulting in different time of hemispherical photographs acquisition ( Bianchi et al., 2017 ).

To calculate canopy chracteristics, 196 images collected were imported in the Gap Light Analyser (GLA). This software transforms the colours of the images to black and white, in order to quantify the pixels before calculation of canopy openness, leaf area index and site openness ( Frazer et al., 2000 ).

Before the pictures were processed, GLA software was downloaded on the desktop. A folder was created on the desktop to store the 196 images collected from the field. The images were uploaded in to the GLA application and process. Before processing began, Geographic north and polar projection distortion on the configuration menu of the GLA software were selected as the geographic parameters of the images. Geographical North was used because our

camera was aligned using a compass northward ( Frazer et al., 2000 ), and polar projection distortion was used because it is the specification of the camera lens ( Herbert, 1987 ).

Still in the configuration menu, the image was registered using “register the image” selection, and the cross-hair cursor was draged from west to east of the image which created a symmetrical circle around the image. During image registration, the box for “fix the registration for subsequent image” was checked since all our images were taken using the same lens.

After the image was registered, a second working image opened. The threshold button in the image menu was clicked to slide the cursor till we had an accurate representation of the sky and shadow for both first and second working image. This was done to ensure that both dark and light areas of the original image had similar representation in either directions ( Frazer et al., 2000 ).

In the menu bar the registered image was calculated for canopy only since the research is focused on canopy characteristics.

Canopy characteristics included:

1) Leaf area index (LAI) that can be simply defined as the amount of leaf surface area per unit ground area. It describes the photosynthetic and transpirational surface of plant canopies. Here, calculated effective leaf area index integrated over the zenith angles 0˚ - 60˚ (referred as LAI 4 in the GLA software) were used.

LAI = leaf area/ground area, m²/m².

2) Canopy openness that is a percentage of open sky seen from beneath a forest canopy, or called “a frog per-spective”.

$\text{CO}={\displaystyle {\sum }_Z\left(G_Z{W}_Z\right)\times 100}$

where CO represents canopy openness, Z represents the different intervals of ze-nithangle, G_Z is the gap fraction of interval Z, and W_Z the weighting factor for interval Z.

3) Site Openness is the percentage of the total sky area that is found in over-lapping gaps in the canopy and mask for each sky region.

$\text{SO}={\displaystyle {\sum }_Z\left(G_Z{W}_Z\right)\times 100}$

where SO represent canopy openness, Z represents the different intervals of ze-nithangle, G_Z is the gap fraction of interval Z, and W_Z the weighting factor for interval Z.

The result was transfer to minitap for further analysis.

All statistical analyses were done using Minitab Version 17 Statistical Package (Minitab Inc., PA, USA). The significance level (α) was set at 0.05.

The field data were imported into QGIS 2.8 to digitise the coordinates of the sprouted stumps. Baseline information from the Cameroon official forest data and the Cameroon atlas were collected before any data manipulation commenced. After digitizing was completed, a quality check was performed to ensure that no points or lines were missing. To ensure correct display of data, all of the shape files were projected to UTM 84, zone 32 N.

Multivariate factor analysis without rotation was done to explore relationships among the variables. Two factors were extracted and their contribution to explaining the resulting data patterns was determined by the percent variation and communality. This was followed by applying the Spearman rank correlation.

The relationship between canopy characteristics and LAI was determined using Spearman Rank correlation and α = 0.05. The relationship between all parameters was established through Multivariate Factor analysis based on the correlation matrix.

The combine Principal Component Factor Analysis (2013, 2015 and 2017) of the Correlation Matrix for Sprout, Years, LAI 4%, LAI 5%, Canopy and Site openness, shows that factor 1 explained 62.6% of total variance while factor 2 explained 17.9% ( Table 1 ).

Factor 1 and factor2 together explain 80.05% Communalities. This means that natural forest recovery in the Akak forest area can be explained 80.05% by the variables listed ( Table 1 ).

Factors 1 and 2 together explain 96.5% Communalities ( Table 2 ) in 2013. This means that natural forest recovery in the forest compartment logged in the year 2013 in the Akak forest area can be explained 96.5% by the variables listed ( Table 2 ).

Factors 1 and 2 for 2015 together explain 96.2% Communalities. This means that natural forest recovery in the forest compartment logged in the year 2015 in the Akak forest area can be explained 96.2% by the variables listed above ( Table 3 ).

Factors 1 and 2 in 2017 together explain 96.1% Communalities. This means that natural forest recovery in the forest compartment logged in the year 2017 in the Akak forest area can be explained 96.1% by the variables listed above ( Table 4 ).

For the year 2013, 2015 and 2017 respectively shows that there is a very strong correlation between LAI (4 and 5) and (canopy and site openness) ( Table 5 ).

Very significant***, Very significant**, significant*.

There is a very strong highly significant interaction (p < 0.0005) between LAI4 and LAI5. This means that as LAI4 increases so to LAI5 also increases ( Table 6 ). There is no significant correlation between sprouting with other variables (LAI, canopy and site openness) ( Table 6 ).