## Introduction

The keystone species concept is an important aspect of population ecology, community ecology, and conservation biology1,2, and its application is likely to be critical with ongoing climate change3. Keystone species can be identified because they have a larger effect on communities and ecosystems than would be predicted based on their abundance or dominance. Loss of keystone species within communities and ecosystems is likely to result in secondary extinction events, and in extreme cases these events can lead to community and ecosystem collapse4. The critical importance of keystone species is derived from the wide range of biotic interactions they engage in with other community members (predation, competition, herbivory, mutualism, facilitation, etc.) and their influence on abiotic environmental conditions2. Keystone species have been described in a range of ecosystems (e.g., marine, fresh water, terrestrial, etc.) and have included a variety of taxa (e.g., fungi, animals, and plants)1,3,5.

Plant communities consisting of isolated or scattered trees occur across the globe, and such trees have been described as keystone species, or “keystone structures”6. This certainly applies to trees and shrubs that are members of plant communities in arid and semi-arid habitat7. Many members of Acacia s.l. (Fabaceae: Mimosoideae8), which are broadly distributed around the world, are considered keystone species within the communities they reside. For example, they are considered keystone species in parts of Australia9, Pakistan10, the Kalahari Desert, Botswana11, Tunisia12,13,14, the Sinai Desert, Egypt15,16, and south-eastern Egypt16,17. As pointed out by Abdallah et al.12, isolated trees in arid habitats, including Vachellia species., have several characteristics that contribute to their keystone status: (1) shade from their canopies prevents extreme temperature fluctuations, increases soil moisture levels, and provides shelter for wildlife, (2) they improve soil conditions through biological nitrogen fixation and litter fall by increasing soil nitrogen content, organic carbon, and water-holding capacity, (3) they increase plant and animal biodiversity as a consequence of characteristics one and two, (4) they provide a source of food for wildlife, and (5) they provide a source of fuel, fodder, and medicines for local people and their domesticated animals. Because of their critical importance, a full characterization of keystone species and the roles they play within communities and ecosystems is urgently needed; especially as they are adversely impacted by various human activities.

The Gebel Elba mountain range is an extension of the Afromontane “biodiversity hotspot” and is at the northern limit of the Eritreo-Arabian province and the Sahel regional transition zone18. The relatively high abundance of moisture of this mountain range leads to higher plant biodiversity than reported elsewhere in Egypt, it consists of 458 species, which constitutes approximately 21% of the Egyptian flora19,20. According to the plant checklist provided by Boulos21, the flora of Egypt consists of 2100 taxa belonging to 755 genera and 129 families; including 45 genera and 228 taxa in the Fabaceae. Gebel Elba is one of the seven main phytogeographical regions in Egypt21. Additionally, the region’s tree and shrub species diversity is higher than in any other regions in Egypt19, with some Sahelian woody elements restricted to the Gebel Elba region and not reported elsewhere in Egypt. Of the 10 Vachellia (synonym: Acacia8) species reported in Egypt, seven are known to occur in the Gebel Elba region, with Vachellia asak (synonym: Acacia asak) and Vachellia oerfota subsp. oerfota (synonym: Acacia oerfota subsp. oerfota) restricted to this region.

An analysis of the plant communities of wadi Yahmib and three of its tributaries, on the north-western slopes of Gebel Elba, revealed the presence of seven plant communities, with these communities being arrayed across an elevational (environmental) gradient17. The Vachellia tortilis subsp. tortilis (synonym: Acacia tortilis subsp. tortilis) community was the main vegetation type on Gebel Elba. This community type occurred commonly in the water channels of wadis and gravel terraces from low to mid elevations (130–383 m), and the species was a member of all of the other six communities in the study area17. In addition, Vachellia tortilis subsp. raddiana (synonym: Acacia tortilis subsp. raddiana) was an overstory co-dominant species in another community on Gebel Elba. Finally, a third acacia species, Vachellia etbaica (synonym: Acacia etbaica), was also detected in this study.

Within arid and semi-arid ecosystems across north Africa and the Arabian Peninsula, plant ecologists have focused their attention on describing the vegetation of wadis that drain to the Red Sea, with these studies focusing on keystone Vachellia species12,13,14,15,16,17,22,23. The present study aimed to contribute to this body of knowledge by determining the distribution, abundance, and describing the growth characteristics of three Vachellia tree taxa in wadi Khoda and wadi Rahaba, in Gebel Elba National Park, south-eastern Egypt. These data will allow us to provide detailed descriptions of the characteristics of these three taxa. This study is essential at this moment because these tree taxa are keystone species within these ecosystems, and their presence and conservation are likely to be threatened by human activities and ongoing climate change.

## Materials and methods

### Study sites

The Gebel Elba National Park (35° 00′ E–37° 00′ E, 22° 00′ N–23° 50′ N) is located in the southeastern corner of Egypt and comprises an area of approximately 35,600 km2 (Fig. 1). The boundaries of this protected area extend more than 50 km north of Shalatein, Egypt, eastward to the Red Sea, south to the border with Sudan, and westward into the Eastern Desert24.

This study was undertaken in 2017 and 2018. Over these two years, the average daily air temperature ranged between 25 and 29 °C, the maximum air temperatures were 40.8 °C in 2017 and 41.9 °C in 2018, the minimum air temperatures were 15.3 °C in 2017 and 14.3 °C in 2018. Annual precipitation was 31.1 mm in 2017 and 17.6 mm in 2018 (Fig. 2). Mean daily relative humidity ranged between 39 and 45% in 2017 and 38 and 44% in 2018, with a range of 28% to 61% relative humidity across the two years.

### Data collection

For each macroplot, all tree taxa were identified according to Boulos19. Because this study was conducted during a period of drought, saplings and seedlings were only rarely observed in the two wadis. Consequently, we only encountered mature trees with a DBH > 10 cm in the macroplots we sampled. The tree characteristic parameters we measured included crown depth, crown diameter, crown volume, tree height, Diameter at Breast Height, DBH (tree diameter), tree radius, and the basal area of each tree25. The crown-to height ratio (C/H ratio) was calculated by dividing the crown volume by tree height. Crown diameter was measured by projecting several edges of the crown to the ground and measuring the length along an axis that extended from crown edge to crown edge. The crown diameter can be used to estimate the area of a tree’s crown (which is used in the crown surface area and volume calculations) by obtaining the average of two axes26. In this study, the diameters of two axes perpendicular to each other (N–S and E–W) were measured and averaged. Crown depth was measured as the distance between the top of the tree and the base of the crown. The base of the crown was identified by the lowest complete whorl of branches or the lowest single branch that formed the canopy of a tree. Crown depth was expressed as the "crown length ratio", which is calculated as the crown length divided by the tree height. Crown volume (Cv) was estimated from crown diameter (D) and crown depth (L) using the equation described by West26:

$${\text{Cv}} = \pi {\text{D2L}}/{12}$$
(1)

The trees in each wadi were numbered sequentially, and their latitude/longitude coordinates were recorded using a 12XL Garmin GPS unit. The nearest-neighbor distance for each pair of trees was measured and recorded. As noted by Cottam and Curtis27, many of these nearest-neighbor distance values are duplicates, since paired neighbors which have each other as nearest neighbors make up a large portion of the total dataset.

To determine the distribution pattern of the three Vachellia tree taxa we detected in this study, we used the method of Clark and Evans28. Briefly, in a population of N individuals with known density (d) and distance from each nearest-neighbor pair (r), the mean observed distance is calculated as follows:

$${\text{r}}0 = \Sigma {\text{r}}/{\text{N}}$$
(2)

The mean distance that would be expected if members of a population were randomly distributed (re) is calculated as:

$${\text{re}} = {1}/{2}\surd {\text{d}}$$
(3)

The degree to which the observed distribution for the distance to the nearest neighbor approaches or departs from random expectation was expressed as the ratio (R) as follows:

$${\text{R}} = {\text{r}}0/{\text{re}}$$
(4)

According to Clark and Evans28 and Petrere29, R has a defined range: 0.0 < R < 2.1491. If R = 0.0, individuals are highly aggregated, when R = 2.1491, there is a completely uniform distribution pattern, and when R = 1.0, the distribution pattern of individuals is random.

### Statistical analysis

Parametric statistical tests were performed on key variables after checking for normality and equality of variance. One-way analysis of variance (ANOVA) was used to evaluate statistical differences among the parameters used to characterize the three Vachellia tree taxa at wadi Rahaba. With only two tree taxa detected at wadi Khoda, t-tests were used to analyze significance among the mean values of these parameters. All statistical analyses were carried out using JMP (ver 4. SAS Institute, Cary, NC, USA).

## Results

The elevation of macroplots were significantly different (P < 0.0001), and the elevation of macroplots increased with distance from the Red Sea (Fig. 4). For wadi Khoda, the macroplot with the lowest elevation was located 29 m above sea level and the macroplot at the highest elevation was 224 m. For wadi Rahada, the elevation of the macroplots ranged from 11 to 418 m above sea level.

### Variation in tree characteristic parameters

Values for these tree characteristic parameters varied among the three Vachellia tree taxa and between the two wadis where this study was conducted (Fig. 5). A comparison of V. tortilis subsp. raddiana and A. tortilis subsp. tortilis in the two wadis did not reveal significant differences between the two taxa for crown diameter and tree height (Fig. 5). However, these two Vachellia trees exhibited significantly higher values for crown depth, crown volume, DBH, tree radius, basal area, and C/H ratio in wadi Khoda.

In wadi Khoda, were only two Vachellia tree taxa were detected (V. tortilis subsp. raddiana and V. tortilis subsp. tortilis), V. tortilis subsp. raddiana had significantly higher values for crown depth (d.f. = 1; t-test 5.021; P < 0.0001), crown diameter (d.f. = 1; t-test 4.064; P = 0.0002), crown volume (d.f. = 1; t-test 3.419; P = 0.0013), tree height (d.f. = 1; t-test 6.402; P < 0.0001), and C/H ratio (d.f. = 1; t-test 2.868; P < 0.0063) compared to V. tortilis subsp. tortilis. There were no significant differences for DBH, tree radius, and basal area for these two species at wadi Khoda (Fig. 5).

In wadi Rahaba, three Vachellia tree taxa were encountered, and they had significant differences for the growth characteristics measured in this study. Vachellia tortilis subsp. raddiana had significantly greater values compared to V. tortilis subsp. tortilis and V. ehrenbergiana for crown depth (F = 40.00; P < 0.0001), crown diameter (F = 41.38; P < 0.0001), crown volume (F = 22.17; P < 0.0001), tree height (F = 71.22; P < 0.0001), and C/H ratio (F = 17.20; P < 0.0001). Whereas, V. tortilis subsp. raddiana and V. ehrenbergiana had greater values than V. tortilis subsp. tortilis for DBH (F = 11.97; P < 0.0001), tree radius (F = 11.97; P < 0.0001), and basal area (F = 4.48; P < 0.0085) (Fig. 5).

Correlation coefficients of elevation with all tree characteristic parameters, as well as the correlation among tree characteristics were provided in Fig. 6. These results showed that crown depth and tree height are significantly associated with elevation (r = 0.44 and 0.48, respectively), while elevation was not significantly correlated with other tree characteristic parameters (see Fig. 6). Crown diameter, crown volume, DBH, and tress radius showed significantly positive correlation patterns. On the other hand, crown depth and tree height were significantly correlated with each other and other traits, but not basal area (Fig. 6).

### Mean observed nearest neighbor distance

The mean observed nearest neighbor distance (r0) was 60.1 ± 6.2 m in wadi Khoda macroplots, compared to 61.6 ± 6.8 m in wadi Rahaba plots (Table 2). The mean expected distance (re) was shorter in the wadi Khoda macroplots (29.0 ± 2.7 m), compared with the value of re for trees in the wadi Rahaba macroplots (34.8 ± 2.6 m) (t-test, P < 0.05; Table 2). R values indicated that the distribution pattern of Vachellia tree taxa varied between random and completely uniform in both wadis. The R values were 1.6 and 1.7 in wadi Rahba and wadi Khoda, respectively (Table 2), and these values were significantly different (t-test, P < 0.05).

## Discussion

Variation between and among plant populations arises due to differences in genetic diversity, heterogeneity of resource availability, competition, herbivory, and pathogen attack, all of which effect the growth rates of individuals in a population30. Topographic and edaphic factors, especially altitude, play a pivotal role in the distribution of plant communities. Understanding the relationships between vegetation cover and these environmental factors contributes to developing sustainable management strategies for such communities31,32.