The idea of alien species had always been important to me. In Colombia, I visited several national parks which were concerned with the threat invasive species posed on their endemic vegetation. I decided to investigate this topic further after having seen an announcement in the Dainfern magazine by the Dainfern Nature Association. The announcement contained a list of alien invasive plants, urging the residents not to plant them. Simultaneously in my IBDP Biology class, I was learning about competitiveness and ecological niches in an ecosystem, something I was aware was relevant in the context of South Africa. Following this, I contacted Olivia Denney, who works in the Dainfern Nature Association. After a few conversations, she mentioned that the small forest in Dainfern had big White Poplar trees, which are an invasive species to South Africa, and some indigenous Celtis, which were trying to grow through. Interestingly, she had gone to the area before and removed some of the alien species in hopes of the Celtis growing. I went to this forest area within the estate and identified the White Poplar trees. These trees were quite large, and when looking up, I could tell their leaf canopy cover was not allowing much light to get through the smaller endemic plants. Following this observation, I was interested in exploring the effect that canopy coverage has on the height of the Celtis trees. Regardless of other variables affecting growth, I was curious to investigate to what degree there is a correlation between increasing canopy percentages and the height of the native species. Considering this is a small forest, it took hours of planning to find the Celtis species that were under the canopy of the Poplar trees and collect the data with the correct tools.
Research question - How does the canopy cover (36, 57, 59, 68, 75 %) of invasive white poplar (Populus alba) affect the growth of native White stinkwood (Celtis Africana) as indicated by its height measured in meters?
Populus Alba, known as White poplar or Silver Poplar tree, is native to central and southern Europe, Morocco, and the Iberian Peninsula. (Poplar, White). P. Alba is fast-growing and drought resistant. It can reach heights of 30 meters and live for 300 - 400 years. (Populus Alba). It adapts well to warm conditions, and it establishes in sunlight, shading out native vegetation. Although it is unknown when it first appeared in South Africa, it's known that it has been introduced into North America, South America, Australia, South Africa, and Korea, mostly as an ornamental tree”(Populus Alba) and now appears to have become invasive in South Africa. Celtis Africana known as white stinkwood is a tree indigenous to South Africa and can grow up to 30 meters. Regarding biological niches, P Alba has a tolerance to drought and salt. However, it prefers high temperatures and is light-demanding. (Populus Alba). C. Africana grows in almost all of South Africa, growing anywhere from “koppies, coastal dunes, river banks, and dense high forests” (Celtis africana – Tree SA). African Snout butterfly and Blue-spotted Emperor butterfly feed on these trees. Native birds also help in the distribution of seeds. (Celtis africana – Tree SA). Similarly to P. Alba, C. Africana can withstand long periods of drought. It can tolerate mild frosts, but extremely cold conditions are intolerable. When young, C. Africana should be able to grow 800 - 1500 mm per year. Its preferred niche for lighting is the full sun. (Celtis africana – CJM Growers). So, these two species share preferred niches, thereby creating competition for limited resources. In the case of this investigation, the resource is sunlight; both species prefer full sun, and “inevitably, the less well-adapted species will struggle to survive and reproduce” (Ecological Niche). The danger of competitive exclusion is that it can drive indigenous species to extinction. For this reason, exploring C. Africana’s realised niche with P. Alba is important to determine the interactions they have when sharing resources and the threats endemic species face in South Africa.
The study was conducted in Dainfern Golf Estate. Dainfern is a residential estate with both planned (gardens) and indigenous nature. The Dainfern Nature Association is aware of the invasive and native plants within the estate, including the presence of the White Poplar tree and native Celtis Africana trees in a small forest area behind “The Glades Village.” (Denney). The forest area has a small body of water and a stream, and it is located between two rows of houses. There is also a walking path crossing the forest. However, the actual forest area is mainly untouched by humans, and interference only occurs in cases of significant invasiveness.
Arboreal Tree Height is an AR application that lets you measure the height of a tree. The app creates a 3D environment with the camera and the internal sensors. “The user marks the position of a tree and walks some distance from the tree, and the inclinometer (angle) is used to compute the height of the actual tree.” (Questions and Answers). This tool was chosen as some of the trees were too high to be measured with a measuring tape.
Percentage Cover is an app that measures canopy cover. “It utilises standard smart phone technology (such as camera, GPS, date/time) to apply existing assessment techniques, used by environmental professionals, students, and amateurs, in the ecological assessment of ... leaf cover versus leaf gap, above a nominated point along a transect or under an individual tree” (Percentage Cover). The value is expressed as a percentage canopy filled (leaf cover). For instance, 68% indicated 68% filled and 32% unfilled. The app has a detection level which is a slider bar used to adjust the contrast to a level that best reflects the overlying canopy silhouette. This detection level is chosen automatically by the app.
The independent variable being changed is the canopy cover. The area where this investigation was conducted is relatively small. Therefore the canopy cover percentages depended on the number of invasive P. Alba trees. The percentages used ranged from 36 - 75%. There were not enough invasive trees to choose percentages that increased in constant intervals. Nevertheless, the percentages found (36, 57, 60, 68, 75%) had enough variation; with a minimum of 3% between 57 and 60, and a maximum of 21% between 36 and 57. Moreover, the difference between the maximum value and the minimum (76 – 23) is 40%, a significant difference in leaf cover filled.
According to a study by the University of Venda, South Africa, “Populus alba (white poplar) outcompetes numerous indigenous tree and shrub species in mostly sunny areas... Dense stands of white poplar prevent other plant species from coexisting by reducing the amount of sunlight.” The same study also states, “white poplar can develop dense stands which can crowd and shade native vegetation and reduce species diversity.” (Mbezi et al. 1674 - 1675)
If the independent variable is increased, then the dependent variable will decrease. Meaning as the percentage cover of Poplar Alba increases, then the height of Celtis Africana will decrease. White poplar reduces the amount of sunlight and shade native vegetation. (Mbezi et al. 1674 - 1675). Accordingly, if C. Africana’s preferred niche for lighting is the full sun (Celtis africana – CJM Growers), there will be competitive exclusion between the species as P. Alba uses light resources more efficiently.
Canopy leaf cover of P. Alba expressed as a percentage of leaf filled. (36, 57, 60, 68, 75%)
The existing field practice for estimating the percentage of canopy cover is to visually assess the canopy cover using a hard copy reference sheet, with black and white images showing incremental area increases of 10%, from entirely white (0%) to entirely black (100%). The range and variation of the independent variables are convenience sampling, the trees investigated displayed these different percentages.
Height of C. Africana measured in meters. (± 0.005 mm)
Measured with Arboreal Tree Height App.
Controlled variable | Impact of not controlling variable | How will it be controlled |
---|---|---|
Distance of native C. Africana from invasive P. Alba | If P. Alba is too far from a specific C. Africana then its canopy won’t cover the native. If there is no canopy coverage the dependent variable, simply won’t depend on the independent | Established a maximum radius from the base of the P. Alba tree. The biggest foliage extension (how far the leaves' shadow extended to) was 7.8meters. So, every native species had to be underneath the foliage of P. Alba, thus all the C. Africana were within this radius (7.8m) |
Time of day | The app measures the filled leaf cover by assessing the light that goes through versus the canopy cover. If the sun is setting or rising it can alter the results as the app works best with a high contrast between shadows and light. | All data was taken on the same day in the span of 2 hours. The sun remained above the trees and did not move considerably. This allowed for a more accurate reading of the light and the canopy percentage. |
Detection level (analysis of canopy cover) | When analyzing the canopy, the app automatically chooses a detection level. The detection level is the recognition of leaves versus light according to contrast. However, the leaves of P. Alba are light green, and sometimes the detection level did not recognize all the leaves. The detection could be adjusted to contrast level that best reflects the overlying canopy silhouette using the slider bar. | Adjusted the detection level for one of the canopy coverage measurements (Tree #3 – 75%) because the app did not properly recognize the overlaying canopy leaves. To ensure an accurate detection level I turned the picture to black and white and increased the contrast. Next, I compared this picture with the detection level analysis until all leaves of P. Alba were recognized. The detection level was 135.7 *Refer to Figure 1and 2 |
Distance from Houses and path | The houses and paths (human intervention) can disturb the environment affecting biodiversity and resources. Also, shade from houses can affect the growth of the Celtis tree independently to the canopy of the Poplar tree. | Nearest house was around 10 meters. None of the trees touching the path were considered. |
Angle when measuring height when using Arboreal app | The arboreal app calculates the height by marking the position of a tree and walking some distance from the tree and the inclinometer (angle) is used to compute the height of the actual tree. If the angle is changed when walking backwards from the tree the result is not calculated accurately, directly affecting the result. | In order to get the most accurate results, the camera was held in an upright position when walking away from the tree. Finally when measuring, stood at 4 meters distance from the Celtis trees every time (app displays distance while walking away from the tree) to have a full (top to bottom) view of the tree. |
Figure 1 - Table On Table Showing Assessment Of Controlled Variables
The sampling strategy used was convenience sampling. This was due to the small area, and the species evenness in the zone studied.
Apparatus Description | Number used | Instrumental Precision |
---|---|---|
Notebook | 1 | N/A |
Pen | 1 | N/A |
Arboreal App | - | ± 0.005 mm |
Percentage Cover App | - | 1 |
Measuring tape | 1 | ± 0.005 cm |
Figure 4 - Table On Table Showing Apparatus List
Safety, Ethical and environmental consideration
Canopy coverage Populus Alba (% filled) | Mean height Celtis Africana (m) |
---|---|
36 | 8.74 |
57 | 2.47 |
60 | 1.98 |
68 | 1.98 |
75 | 1.20 |
Figure 6 - Table On Table Showing Canopy Coverage Percentage Of Populus Alba And Mean Height Of Respective Celtis Africana In Meters.
Row | Value | Z | Significant Outlier |
---|---|---|---|
1 | 8.74 | 1.7415 | Significant outlier. P < 0.05 |
2 | 2.47 | 0.3399 | |
3 | 3.08 | 0.1374 | |
4 | 1.98 | 0.5026 | |
5 | 1.20 | 0.7616 |
Figure 9 - Table On Grubb’s Outlier Result For Dependent Variables
To perform a Pearson’s r test, there cannot be any outliers. GraphPad outlier calculator showed 8.74 is a significant outlier P<0.05. however, because the intervals of independent variables are not consistent, it was not removed. In other words, the y value (36%) for this tree was appreciably lower than the rest of the y values due to the nature of the sampling method.
A Pearson's correlation coefficient test was carried out using Social Science Statistics Pearson Correlation Coefficient Calculator.
R Calculation breakdown -
X Values -
Σ = 296
Mean (Mx) = 59.2
Σ(X - Mx)2 = SSx = 870.8
Y Values -
Σ = 17.47
Mean = 3.494
Σ(Y - My)2 = SSy = 36.295
X and Y Combined -
N = 5
Σ(X - Mx) (Y - My) = -169.354
R Calculation -
r = Σ((X - My) (Y - Mx)) / √((SSx)(SSy)) = -0.953
r = -169.354 / √((870.8)(36.295)) = -0.953
r = -0.953
Pearson’s r measures the strength and direction of the correlation between independent and dependent variables. Where r is a value from -1 & 1. Between 0 and -1 indicates a negative correlation. 0 indicates that there is no association between the two variables, and 1 or -1 means a perfect correlation. R equaled -0.953; this indicates a strong negative correlation. As the percentage cover increases, the height decreases.
R2
R2 of the linear graph is 0.908 (to three significant figures) Calculated by excel built-in R-squared value function
R2 determines the proportion of variance in the dependent variable explained by the independent variable. 1 indicates that all movements in the y - axis (dependent variable) are completely explained by movements in the x - axis (independent variable). The result 0.908 for the linear model in Graph 1 essentially means that 90% of the variance of the dependent variable is explained by the independent variable.
The percentage difference between 36% and 75% canopy cover is 152%
Calculation -
8.74 - 1.20 = 7.54
Average = (8.74 + 1.20) / 2 = 4.97
Ratio = 7.54 / 4.97 = 1.1517102516
1.1517102516 × 100 = 152%
A difference of 39% between independent variables shows a high percentage difference (151%) in the results of the dependent variable.
The degree of accuracy, according to the app developer, was ± 0.4 m (Arboreal-height). A study assessing the uncertainty of measuring tree height using found that “Results proved that σH could vary between 0.5 m and up to 20 m (worst case) ...height measurement uncertainty (σH)” (De Petris et al. Pg 1). This suggests that the application used in the experiment is mostly accurate.
A chi-squared test was also carried out to further reveal if the presence of P. Alba had an effect on the presence of C. Africana upon the competitiveness between them. The test was done in the same area as the data was collected. The quadrats were 3m x 3m since measuring trees requires a larger quadrat sampling. The quadrats were placed randomly within the area, and the total of 11 quadrats was convenience sampling as the area is not large and did not allow for more relevant quadrat samplings.
The calculated value for chi-squared = 3.833 and the degree of freedom (Df) = 1. The critical value in the P values for the chi-squared distribution table at the 0.05 level of significance the X2 value has to be higher than 3.841 to be considered statistically significant. 3.833 < 3.841 thus, the null hypothesis is accepted and concludes that there is no significant association between the variables. However, 3.833 and 3.841 have a difference of 0.008 meaning 0.8%. A value is considered significant if there is less than a 5% probability (p < 0.05) so in this case there is less than 5.8% chance that the results are due to chance.
The line of best fit shows a steep negative slope, indicating a negative correlation. The scatter points are close to the trend line suggesting a stronger correlation. The maximum value is 8.74m, and the minimum is 1.20m. The percentage difference between the highest and lowest value was 152% This is a high percentage difference that indicates the dependent variable can decrease greatly as the independent variable increases. Meaning that a canopy cover difference of 39% has a big difference between the heights of the endemic tree. Pearson’s correlation coefficient was - 0.953. This is a significant value, and it is really close to -1, which is close to a perfect negative linear relationship. Thereby, a strong association between invasive P. Alba canopy and C. Africana height. R2 of the model was 0.908, or 90% of the observed variation is explained by the inputs of the different canopies. In general, a high coefficient of determination results from a model (linear) that fits the data well, but this depends on the context. Another point to consider in relation to the results is that C. Africana should grow 800 - 1500 mm per year (Celtis africana – CJM Growers). So if the C. Africana is “about 20 years” (Denney) old, they should have a minimum height of 16 meters. When comparing the maximum height result (8.74m) against the minimum height, a C. Africana should grow in 20 years (16m), there is a vast difference. This perhaps suggests the canopy cover of invasive White Poplar impedes the growth of the native Celtis by a half.
The chi-squared test had a value that accepted the null hypothesis and concludes that there is no significant association between the variables. However, 3.833 and 3.841 have a difference of 0.008 meaning 0.8%. there is less than a 5.8% chance that the results are due to chance. Although only under 5% is considered statistically significant 5.8% is still a low percentage accredited to chance it is, therefore, an important value to consider. Consequently showing that there is still a negative relationship between the abundance of C. Africana vs. P. Alba. Also seen in table 6 where there were 8 quadrats where P. Alba was present and C. Africana was absent in comparison to 1 quadrat where C. Africana was present and P. Alba was absent.
Based on the results of the investigation, it can be said that the canopy cover of Populus Alba is detrimental to the growth of indigenous Celtis Africana. Pearson’s correlation test (r = - 0.953) showed a strong association between the independent variable and the dependent variable. Moreover, the model of regression was shown to be a strong negative correlation by the R2 . Also, as stated above, the scatter plots are close to the trend line, further reinforcing the strength of the model. The tests indicate that as the percentage of leaf cover increases, the height decreases. These results are congruent with previous studies: Populus Alba outcompetes indigenous trees in sunny areas, as it can “shade native vegetation” (Mbezi et al. 1674 - 1675). This was the case for this investigation, especially since Celtis Africana’s preferred niche for lighting is the full sun. (Celtis africana – CJM Growers). Their shared niches result in competition, in which, according to the data, Populus Alba thrived over Celtis Africana in the micro-ecosystem studied. The percentage difference between the highest and lowest results was 152%, revealing the extent of the impact of canopy cover on tree height as this is a large percentage. The qualitative data also corroborates this. The more the foliage extended, the shorter and smaller the Celtis Africana was. Also, the leaves appeared darker for the three highest canopy covers and lighter for (36 and 57) The shade of the color suggests the leaves with the most shading are darker as they require more chlorophyll. Finally, the chi-squared test showed a negative association (C. Africana species tends to not occur as often within the same habitat as P. Alba). Species typically present a negative association if there is competition for the same resources – in this case sunlight. Thus it suggests P. Alba utilizes sunlight more efficiently, leading to competitive exclusion in the area studied.
Nonetheless, the hypothesis cannot be concluded as true due to limitations in the investigation. While the R2 test indicated a strong correlation, R-squared does not tell whether the model chosen is good or bad, not if the data is biased. Furthermore, the method and range chosen for the independent variables are frail. The number of invasive trees sampled was 5, thus 5 range of independent variables; this is a low value to be able to generalize the trend to the population as a whole. Also, the distance of the ranges can cause outliers, as seen with a canopy of 36% and a height of 8.74. Uncontrolled variables that are key in leading to a clear conclusion are the age of the Celtis Tress and the disturbed area. During the interview with Olivia Denney, she stated the Celtis in the area had been there “for about 20 years”. However, this isn’t an exact estimate; the age of each Celtis tree has a direct impact on its height, hence an important variable to consider. Finally, the micro-ecosystem within Dainfern is inevitably impacted by human interference in some way, and thus, the results of this experiment do not fully reflect the effect of this invasive species on the generalized ecosystems within South Africa.
Based on the results of the investigation, it can be said that the canopy cover of Populus Alba is detrimental to the growth of indigenous Celtis Africana. Pearson’s correlation test (r = - 0.953) showed a strong association between the independent variable and the dependent variable. Moreover, the model of regression was shown to be a strong negative correlation by the R2 . Also, as stated above, the scatter plots are close to the trend line, further reinforcing the strength of the model. The tests indicate that as the percentage of leaf cover increases, the height decreases. These results are congruent with previous studies: Populus Alba outcompetes indigenous trees in sunny areas, as it can “shade native vegetation” (Mbezi et al. 1674 - 1675). This was the case for this investigation, especially since Celtis Africana’s preferred niche for lighting is the full sun. (Celtis africana – CJM Growers). Their shared niches result in competition, in which, according to the data, Populus Alba thrived over Celtis Africana in the micro-ecosystem studied. The percentage difference between the highest and lowest results was 152%, revealing the extent of the impact of canopy cover on tree height as this is a large percentage. The qualitative data also corroborates this. The more the foliage extended, the shorter and smaller the Celtis Africana was. Also, the leaves appeared darker for the three highest canopy covers and lighter for (36 and 57) The shade of the color suggests the leaves with the most shading are darker as they require more chlorophyll. Finally, the chi-squared test showed a negative association (C. Africana species tends to not occur as often within the same habitat as P. Alba). Species typically present a negative association if there is competition for the same resources – in this case sunlight. Thus it suggests P. Alba utilizes sunlight more efficiently, leading to competitive exclusion in the area studied.
Nonetheless, the hypothesis cannot be concluded as true due to limitations in the investigation. While the R2 test indicated a strong correlation, R-squared does not tell whether the model chosen is good or bad, not if the data is biased. Furthermore, the method and range chosen for the independent variables are frail. The number of invasive trees sampled was 5, thus 5 range of independent variables; this is a low value to be able to generalize the trend to the population as a whole. Also, the distance of the ranges can cause outliers, as seen with a canopy of 36% and a height of 8.74. Uncontrolled variables that are key in leading to a clear conclusion are the age of the Celtis Tress and the disturbed area. During the interview with Olivia Denney, she stated the Celtis in the area had been there “for about 20 years”. However, this isn’t an exact estimate; the age of each Celtis tree has a direct impact on its height, hence an important variable to consider. Finally, the micro-ecosystem within Dainfern is inevitably impacted by human interference in some way, and thus, the results of this experiment do not fully reflect the effect of this invasive species on the generalized ecosystems within South Africa.
An extension to this investigation could be an analysis of the population of African Snout butterflies and Blue-spotted Emperor butterflies in South African ecosystems. As these species feed on Celtis Africana trees and thus their abundance can be linked to the abundance of C. Africana and vice-versa. Although White Poplar “are problematic on one hand, but in the area, they are stopping soil erosion” (Denney). This is an interesting topic to investigate further -areas where the invasive species is more beneficial than detrimental to the resources and health of South African ecosystems. Finally, White Poplars “can be propagated by bare root, as it is very good in root suckering” (Mbezi, Meldford, et al, Pg 1675). In the study by Mbezi, Meldford, et al study, the alien species came back in large numbers after removal due to root suckers left behind. An investigation on the rate of growth and variables boosting the development of the sucker roots can reveal how to control a Populus Alba invasive population.
"Arboreal - Height of Tree Accuracy." Youtube, uploaded by Arboreal, 17 May 2019, www.youtube.com/watch?v=bfHtQI2R3sI. Accessed 14 Sept. 2022.
Caudullo,, G., and D. De Rigo. Populus alba in Europe: distribution, habitat, usage and threats. European Atlas of Forest Tree Species, 2016, ies- ows.jrc.ec.europa.eu/efdac/download/Atlas/pdf/Populus_alba.pdf. Accessed 11 Sept. 2022.
"Celtis africana." CJM Growers, cjmgrowers.co.za/celtis-africana-white-stinkwood/. Accessed 11 Sept. 2022.
"Celtis africana." Tree SA, treesa.org/celtis-africana/. Accessed 11 Sept. 2022.
Denney, Olivia. Telephone interview. 4 Sept. 2022.
"Ecological Niche." BioNinja, ib.bioninja.com.au/options/option-c-ecology-and-conser/c1- species-and-communities/ecological-niche.html. Accessed 11 Sept. 2022.
Mbezi, Meldford, et al. The Rate of Recolonization by Native Plant Species after the Eradication of the Invasive Alien Plant Species(Populus alba L.) in the Limpopo Province, SouthAfrica. EM Internatinal, 27 Feb. 2019. Research Gate, www.researchgate.net/publication/338548905_The_rate_of_recolonization_by_native_pl ant_species_after_the_Eradication_of_the_invasive_alien_plant_species_Populus_alba_L _in_the_Limpopo_Province_South_Africa. Accessed 11 Sept. 2022.
" Pearson Correlation Coefficient Calculator ". Socscistatistics.Com, 2022, https://www.socscistatistics.com/tests/pearson/. Accessed 13 Sept 2022.
"Percentage Cover – Ecological Survey App". Percentagecover.Com, 2022, https://percentagecover.com/. Accessed 11 Sept 2022.
"Poplar, White." Woodland Trust, www.woodlandtrust.org.uk/trees-woods-and-wildlife/british- trees/a-z-of-british-trees/white-poplar/. Accessed 11 Sept. 2022. "Populus alba (silver-leaf poplar)." Invasive Species Compendium, CAB International, www.cabi.org/isc/datasheet/43426. Accessed 11 Sept. 2022.
"Questions and Answers." Arboreal, www.arboreal.se/en/questions-answers/. Accessed 11 Sept. 2022.