The study was carried out in the Aberdare mountains, located in central parts of Kenya (Bennun and Njoroge 1999). It stretches for 120 km from north to south from latitude 00°08' to 00°42' south, with an expanse of about 40 km across at its widest point between longitude 36°31' to 35°57' east (Butynski 1999) (Fig. 1). The Aberdare ranges are characterized by undulating hills formed through uplift and warping, then later shaped by volcanism and faulting of the earth surface from early Tertiary to the Pleistocene (Peltorinne 2004), giving rise to geographically isolated islands of complex tropical-alpine vegetation (Hedberg 1964). There are two main peaks; Oldonyo Lesatima (4000 m a.s.l.) to the north and Il Kinangop (3906 m a.s.l.) to the south, separated by a long stretch of land above 3000 m elevation (Muiruri 1978; Bennun and Njoroge 1999). Aberdare exhibits a unique topography sloping gradually to the East while, in contrast, the western slopes drop rapidly along imposing fault escarpments towards the Kinangop plateau and finally the Gregory Rift Valley, giving way to a number of torrential waterfalls that cascade into deep u-shaped ravines (KFS 2010; KWS 2010).

Figure 1. 

The location of Aberdare ranges forest, (i) two sections of forest (ii) map of Kenya.

Similar to other East African mountains, vegetation in Aberdare is typical afromontane type characterized by heterogeneous vegetation structure along the elevational gradients (Hedberg 1951; Myers et al. 2000). Three broad vegetation zonations have been described by Schmitt (1991) based on floristic compositions, including the montane forest belt at lower altitudes, the subalpine zone at mid elevation, and the alpine zone at the summit. Administratively, this forest is managed in two sections, i.e., Aberdare national park, size ca. 76700 ha., managed by Kenya Wildlife Service (KWS), and Aberdare forest reserve, size ca. 139500 ha., governed by Kenya Forest Service (KFS). The national park is situated at higher elevations above ca. 2600 m a.s.l. and the regions below this park constitute the forest reserve composed of three forest blocks, Aberdare, Kikuyu escarpment, and Kipipiri forest reserve (KWS 2010; KFS 2010, Lambrechts et al. 2003). Altitude is from 1800 m to 4000 m a.s.l. (Bennun and Njoroge 1999). Climate in Aberdare forest is dominated by the passage of the Inter-Tropical Convergence Zone north and south during its annual cycle, producing a bimodal rainfall distribution (CEPF 2012). Long rains usually come from March to May and short rains from October to November (Lambrechts et al. 2003; KFS 2010). Annual rainfall varies with altitude and exposure to the dominant winds from the Indian Ocean, ranging from 1000 mm on the drier north-western slopes to as much as 3000 mm on the south-eastern slopes (Bennun and Njoroge 1999). The mean maximum temperature is 25.8 °C. while mean minimum temperature is 10.3 °C. (KFS 2010; CEPF 2012).

The floristic surveys were carried out from the year 2016 to 2018, during the optimum flowering and fruiting periods which were mainly after the long and short rains of March-May and November-December respectively. A team of botanists from the National Museums of Kenya and Sino-Africa Joint Research Center undertook the surveys. Fertile voucher plant specimens, with either flower, fruit or both were collected, pressed, and dried. Specimens were identified to species level using varied taxonomic monographs and botanical guide books (FTEA 1952–2012; Blundell 1992; Agnew and Agnew 1994; Beentje et al. 1994; Agnew 2013). Plant species were further categorized into their life forms, including (i) herbs (plants less than 50 cm or less than 100 cm but annual), (ii) shrubs (plants between 50cm to 5 m tall), (iii) climbers (plants with twining herbaceous or woody stems), and (iv) arboreal (plants taller than 5m) (Raunkiaer 1934, Schmitt 1991, Bao et al. 2018). Finally, the dried voucher specimens were deposited in the East African Herbarium (EA), Nairobi, and Wuhan Botanical Garden herbarium (HIB), China.

Endemic species were recorded from published literature and updated by cross-checking with the online occurrence data in Global Biodiversity Information Facility (GBIF) (https://www.gbif.org). In addition to field collections, other plant species in Aberdare ranges forest were obtained from the previously collected specimens in the EA herbarium catalogues, series of Flora of Tropical East Africa (FTEA) and other standard references (e.g., Schmitt 1991). Relative altitude range sizes of plant species in Aberdare i.e., the maximum and minimum altitude where a species has been previously recorded or collected, were determined from our collections, as well as from other specimens in the EA herbarium and the standard botanical references. Based on the contemporary description of the Flora of Kenya, information on the floral species and their geographical distributions was considered adequate for this study.

Both air and soil temperature data were obtained from Schmitt (1991), where he calculated the average air temperature from 28 months’ recordings from four locations at different altitudes in Aberdare ranges. Extrapolation was done based on a calculation from East African Meteorological Department (1970) of a decrease of 1.7 °C. for every 1000 feet of altitude equivalent to 0.56 °C. per 100 m elevation. Regarding the soil temperature, Schmitt (1991) obtained the mean of 9 months recordings at 70 cm depth of soil in eight different elevations in the Aberdare ranges. He calculated an average of 0.52 °C. decrease per 100m elevation between 1900−3200 m a.s.l. and 1.5 °C. decrease per 100 m between 3200−3600 m a.s.l. (Braun 1986; Schmitt 1991). This criterion was applied to calculate soil temperatures at altitude bands where there were no recordings.

The Aberdare mountain rises from 1800−4000 m a.s.l. However, the onset of continuous forest cover differs in elevation at various sites of forest edges. To control the biases of uneven elevation of forest margins and disjunct forest blocks, we analyzed species from 2000−4000 m a.s.l. as this represents relatively the continuous forest cover in the entire ecosystem (Schmitt 1991, Butynski 1999). Elevation gradient was divided into 20 bands of 100 m interval between the 2000−4000 m altitude in the same manner as Vetaas and Grytnes (2002), Bhattarai et al. (2004) and Grau et al. (2007). Recorded plant species were then interpolated at these 20 100 m bands giving an estimate of gamma diversity defined by Lomolino (2001) as total richness of an entire elevation band. The assumption here was that each species was present in all the elevation bands within its altitude range size, ignoring any disjunctions in its distribution along elevation gradient. Interpolation of species facilitates analysis where there is incomplete sampling as in this study. A comparative analysis conducted on empirical and interpolated species richness data of the Baekdudaegan mountains of South Korea showed similar spatial distribution patterns (Lee et al. 2013). Hence interpolation of taxon richness was best suited for our study as it can give reliable results similar to complete sampling. Each plant species, subspecies or varieties were treated as individual taxa. The relationship between all these plant groups and elevation, air temperature and soil temperature were investigated in IBM SPSS statistic 25 software.

Descriptive statistics were explored to interpret data distributions and normality tests (Table 1). Moderate skewness and negative kurtosis were observed indicating nearly normal distributions as they were below 1 and 2.5 respectively (Zhao and Fang 2006). Arboreals were square-root transformed since it failed Kolmogorov-smirnov normality tests (P < 0.023). No transformations were done for other groups as their distributions were normal (P > 0.05). Simple scatter plots were used to observe if non-linear relationship existed between plant groups and combined environmental variables studied. All plant groups showed linear relationship, hence Pearson’s Correlation analysis was done. Multiple regression model was best suited for data analysis, however, there was significant multicollinearity between the predictors (r > 0.07), which would have given unreliable results. Therefore, Partial Least Squares (PLS) regression model was used (Tobias 1995). PLS regression is robust in multicollinear variables and it focuses on components with maximum covariance with dependent variable (Maestre 2004; Davis et al. 2007; Carrascal et al. 2009), hence, an environmental factor with the highest projection on richness patterns of plants groups could be identified through this model. Relative importance of individual predictor was evaluated using Variable Importance in Projection (VIP) values i.e., the higher the VIP the higher the importance in projection. PLS regression with a Python 2.7 extension module was used in IBM SPSS statistic 25 software. In addition to PLS regression, simple linear regression of each environmental factor on each studied plant group was also calculated.

Aberdare ranges forest harbors a number of globally important plant species. A total of 73 species have been assessed globally to be threatened and 30 species are possibly threatened according to Botanic Gardens Conservation International threat search 2019 (https://tools.bgci.org/threat_search.php). This is a clear indication of the global importance of Aberdare ranges forest as a biodiversity hotspot and the urgent need for effective conservation measures to protect the threatened species and the rich plant diversity in general. The majority of the threatened species were herbs with 70 species, shrubs 12, climbers 11 and the arboreal numbered 10 (see Appendix 1).

Plants diversity in Aberdare ranges is high, based on the total taxa recorded in our study. This was higher than the previous survey done by Schmitt (1991) which documented 778 species in 128 families and 421 genera. Altitude was found to have relatively higher influence on distribution and richness patterns of plant species in this forest. This suggests that the heterogeneous vegetation structure exhibited in the Aberdare ranges is a manifestation of altitude increase. This phenomenon has attracted a lot of debate with several factors mentioned as explanatories for habitat heterogeneity, e.g., energy, water, soils, and area. However, these factors are known to have direct influence on plant physiological processes which in turn control plant growth and spatial richness; at the same time these factors are directly influenced by altitude (Whittaker 1967; Rahbek 1997; Schmitt et al. 2013; Berhanu et al. 2017). Therefore, altitudinal gradient can be regarded as an overall fundamental factor with indirect effects on species richness patterns. Dyakov (2014) argued that altitude can be used to provide insightful information on plants richness and distribution patterns, especially in ecosystems where other abiotic parameters are missing and this notion has been supported by this study. Total plant species in Aberdare showed a monotonic declining relationship with optimum altitude ranging between 2000–2100 m a.s.l., which was simply the foot of the Aberdare ranges. The aggregation of species at lower elevation indicated mass effect which coincided with Rahbek’s (1997) observation that lowlands represent sink habitats with higher species richness than other higher elevation zones. This higher richness is likely due to the infusion of species from surrounding lowland vegetation which could not survive at higher elevations (Rahbek 1997). It is also probable that human transferred species might have been perfectly integrated and naturalized during early historic community settlements in and around Aberdare mountains (Trigas et al. 2013). The known harsh environmental constraints at higher elevations such as cooler temperatures, low energy, shorter growing seasons, solifluction, isolation, etc., could have deterred to an extent any immigration, dispersal and invasiveness of new plant species at these elevated areas (Hedberg 1964; Vetaas and Grytnes 2002; Dyakov 2014). Furthermore, monotonic decline of species richness could be due to the absence of true montane flora in Aberdare, where the current richness patterns are a result of the spread of lowland species that could withstand and adapt to the mentioned harsh conditions associated with higher elevations (Hedberg and Hedberg 1979; Trigas et al. 2013). Numerous studies in similar montane ecosystems have reported monotonic decline of floristic richness. In Ethiopian afromontane vegetation, species richness peaked between 1600–1700 m a.s.l. which was lower than Aberdare despite both mountains flanking the Great Rift Valley (Berhanu et al. 2017). Vetaas and Grytnes (2002), using interpolated data of Nepal Himalayas, found a hump-shaped unimodal relationship with maxima species richness between 1500–2500 m a.s.l., which was similar to this study.

There was a slight increase in richness of herbs between 2000–2100 m a.s.l. then a monotonic decline as altitude increases. Herbs showed higher gamma diversity i.e., total species in each altitude band, compared to other studied plant groups in entire altitude gradient of Aberdare ranges. Regarding this observation, this study contrasted with Schmitt et al. (2010) generalization that herbs are more sensitive to small-scale changes in environmental conditions while woody plants are affected by environmental changes at larger scale. If this was the case then woody species richness i.e., shrubs and arboreal would be more abundant at higher altitudes which was not observed in this study. As floral richness is known to be maximum at altitudes associated with their optimal climatic gradients (Lomolino 2001; Dyakov 2014), based on this study 2000–2200 m a.s.l. seems to be the focal altitude for total plants in Aberdare ranges forest. The effects of heavy deforestation at lower altitudes in close proximity to local communities around the forest (Butynski 1999; Lambrechts et al. 2003; KWS 2010; KFS 2010) were not shown in our study. It was expected that richness in shrubs and arboreal would be low at the foot of the range as a result of overexploitation then a slow or even absent succession by arboreal species afterwards.

A total of 63 species were endemic in Aberdare ranges forest. Most of these endemics were herbs - with 35 species, then shrubs 23, climbers 3 and the lowest were arboreal with 2 species (see Appendix 2). The altitudinal range sizes, i.e., between the minimum and maximum altitude, of these endemic species varied from as low as 100 m for Cissampelos friesiorum Diels, 150 m for Senecio margaritae C.Jeffrey to 2500 m for Adenocarpus mannii (Hook.f.) Hook.f. and Erica sylvatica (Welw. ex Engl.) Beentje. Endemic species had positive relations with elevation and negative relations with both air and soil temperatures. There was a continuous spread of endemic species in the entire forest; however, higher richness was observed at higher altitude. Endemism richness maxima ranged between 3200–3900 m a.s.l., which overlapped with the endemic species of the Himalayas mountain which peaked between 3500–4500 m a.s.l. (Vetaas and Grytnes 2002). This study supported the generalization that endemic species as restricted range taxa were manifested by the survival of species in refugia during Pleistocene climatic fluctuations and/or their resilience and adaptation to unique long-term abiotic conditions that promoted morphological differentiation of relict taxa (Lovett and Friis 1994; Lovett et al. 2000). Higher endemism towards the top of Aberdare ranges might be a result of temporal individual responses of plant species to climatic vicissitudes leading to novel assemblages at high altitudes as different species move individually up the slope (Lovett et al. 2005). There was a notably rapid increase in endemic richness at 2700–3000 m a.s.l. which coincided also with the rapid decrease in total plant species richness. This striking feature was also observed among the endemics of the Himalayas in Nepal although at higher altitudes above 4000 m a.s.l. (Vetaas and Grytnes 2002). The authors at Himalayas Nepal argued that timberline and glaciation limits are responsible for the inverse relationship between endemism and total plant richness. However, this view was only partially supported in our study due to the current absence of glaciers in the Aberdare ranges which are thought to have disappeared during early Holocene glaciation (Young 1980; Rosqvist 1990; Mark and Osmaston 2008). This study therefore suggests that perhaps localized competition among plant species has an effect on endemic richness and diversity although this largely remains a speculative concept. Increasing isolation and decreasing surface area at higher elevation which are responsible for fragmentation of species population, have been argued to facilitate endemism because of the small population’s vulnerability to speciation (Trigas et al. 2013 and references there in). The formation of the Aberdare mountains during the break-up of Gondwanaland millions of years ago led to isolation of plant species that existed then. Over time these species have undergone speciation, resulting in numerous endemic species (Butynski 1999).

In the wake of elevated anthropogenic threats and global climatic vicissitudes, conservation and management of Afromontane ecosystems should be prioritized, particularly for future biodiversity and sustainable provision of ecological services (Muiruri 1978; Myers et al. 2000; Hannah et al. 2002; Kiringe et al. 2007; KFS 2010; Schmitt et al. 2010; Di Minin et al. 2013; Anderson-Teixeira 2018). Afromontane forests are thought to provide survival options for plants in a changing climate through the close physical proximity of a wide range of habitats with varied biotic and abiotic factors (Lovett et al. 2005). Thus, the Aberdare ranges with its proven diverse flora and unique physical properties calls for more effective management interventions. Floristic composition and spatial patterns studied in this forest can steer conservation strategies by pinpointing exceptional rich species and ecologically unique zones (Platts et al. 2010). Thence, the type of management regime to be implemented should target the causal factors of observed richness patterns since species richness is an indicator of environment history (Lovett et al. 2000; Platts et al. 2010). This implies that any conservation intervention to be undertaken should focus on maintaining the forms and magnitudes of disturbances the existent vegetation has adapted to. For the case in Aberdare, higher endemic richness at higher elevations indicates the existence of undisturbed long-term environmental stability as argued by Lovett and Friis (1994), Sosef (1994) and Lovett et al. (2000). Therefore, disturbances should be minimized at this zone because the vegetation might not be resilient to perturbation. Networks of roads and camp sites should be reduced, if not totally eliminated, in upper parts of the forest (KFS 2010; KWS 2010). According to Connell (1978), Phillips et al. (1994) and Lovett et al. (2000) high species diversity is a result of intermediate levels of disturbance that enhances recruitment and regeneration of new species. If this concept of disturbance-driven species richness is applied at lower elevations of Aberdare, where there is a history of exploitation by local community (Butynski 1999), then it means that the great species richness observed at this region could be due to disturbance. Therefore, a suitable management intervention should complement the previous disturbances at appropriate levels and magnitude so as to maintain plant species complementation. This will necessitate some controlled small-scale extraction and utilization of forest resources around the foot of Aberdare ranges. Regulated extraction and utilization of forest products could be the best conservation strategy at species level as this can be an intermediate disturbance if well managed. The carrying capacity of the vegetation area to be grazed should be established prior to grazing.

In view of envisaged climatic changes, it has been suggested that ecosystems migrate to new regions as a result of climatic fluctuations (Hannah et al. 2002; Lovett et al. 2005). In fragmented ecosystems characterized by isolated habitats that are surrounded by agricultural or other human activities, such ecosystems will have no room for migration resulting in total collapse (Hannah et al. 2002). This scenario resonates with the present situation in the Aberdare ranges with agricultural activities surrounding the forest almost entirely. To overcome the impending predicament of ecosystem collapse, creation of corridors seems to be the best strategy, perhaps the only way, to facilitate ecosystem migration and enhance survival of species (Lovett et al. 2005).

Completion of a perimeter electric fence around the Aberdare ranges almost a decade ago has impacted significantly on both ecological and economic aspects of this region (Ark and Group 2011). However, the long-term effects of this fence regarding migration and population increase of wildlife, particularly the large herbivores, have not been adequately addressed (Butynski 1999; Ark and Group 2011). A complete barrier to the migration of elephants has increased damage to vegetation mostly around their traditional migratory routes to the lower Laikipia savanna and Mount Kenya (Ark and Group 2011). Prolonged damage to vegetation in such areas will result in larger open grounds with reduced plant diversity which will lead to minimal utilization of nutrients by plants since the full range of plant niches is not covered (Lovett et al. 2005). This unbalanced destruction of vegetation in the Aberdare ranges will disrupt its ecological equilibrium and threaten the ecosystem functioning and sustainability in the long run (Tilman et al. 1996, 2006). In addition, the expanding populations of range-restricted elephants and other herbivores will increase vegetation damage in the near future. A management intervention in this respect should be at vegetation level and must aim at balancing utilization and regeneration of resources to maintain adequate vegetation cover and rich species diversity, thus maintaining ecological balance and ecosystem productivity of the Aberdare range forest. Migratory corridors should be established, perhaps along the traditional migratory routes, to facilitate seasonal migration of large herbivores in an out of Aberdare ranges.

Wildfires in the Aberdare ecosystem have been common incidences for past decades, occurring mostly during dry seasons in the months of January, February or March and a few cases in September (KFS 2010; KWS 2010). Causes of these fires vary from arsonist ferrying poles, charcoal burning, nearby litter burning during farmland preparations, camp fires and improper cigarette butt disposal by tourists (KFS 2010; KWS 2010; Njeri et al. 2018). Vegetation damages caused by fire are higher at the upper parts of Aberdare including the established plantations due to higher fuel biomass (Njeri et al. 2018). As a result, fragile habitats like the moorland which harbors high endemic and threatened species are negatively impacted. However, fires have also been found to facilitate regeneration of Juniperus procera Hochst. ex Endl., Bambusa vulgaris Schrad, Hagenia abyssinica (Bruce ex Steud.) J.F.Gmel. including some Pines, Cypress and Eucalyptus trees (KFS 2010; KWS 2010; Nyongesa and Vacik 2018). Therefore, the current management approach of prevention and suppression of fire implemented by the KWS and KFS in Aberdare is feasible in the moorland but misplaced in areas with fire-dependent plant species. This study recommends the implementation of an Integrated Fire Management (IFM) framework proposed by Nyongesa and Vacik (2018) in Mount Kenya forest as they are similar ecosystems. In IFM framework, fire-sensitive sites are protected while prescribed-burning is undertaken in fire-dependent ecosystems.