Species composition and stand structure of secondary and plantation forests in a Kenyan rainforest

Forest succession has been reported in forest plantations in tropical forests, but little is known about their successional dynamics because most studies have focused on succession in secondary forests. We assessed changes in species composition and stand structure in secondary and plantation forests in Kakamega rainforest in western Kenya. We used a nested experiment to collect data on tree species types, tree height and stem diameter at breast height from secondary forest stands, mixed indigenous plantations and indigenous and exotic monoculture plantations in three forest blocks. Data were analyzed for variation in species diversity, species similarity to the primary forest, stem density and basal area using analysis of variance in Genstat. The results indicated that species diversity and similarity to the primary forest were not different between secondary and plantation forests. However, successional species occupied all the canopy strata in secondary forests, but they occupied only the shrub and understorey layers of monoculture plantations, and the shrub, understorey and sub-canopy strata of mixed indigenous plantations. Mixed indigenous plantations had become nearly indistinguishable from secondary forests, but monoculture plantations maintained a plantation outlook. Old secondary forest had a significantly lower stem density than plantation forests, but their basal area was not significantly different. Middle-aged and young secondary forests had comparable stem density to plantation forests, but their basal area was significantly lower. The results confirmed that plantation forests are experiencing forest succession in tropical forests, their species composition and stand structure are comparable with secondary forests, but they differ in the emergence pattern of successional species and their distribution in forest canopy strata.


INTRODUCTION
Many tropical rainforests have been felled and thereafter converted to farmland, human settlement or pasture.Over time, as the farmlands, settlements and pasture are abandoned; these areas either regenerate naturally into secondary forests or are planted as plantation forests (Chazdon et al., 2010;Omeja et al., 2012;Chua et al., 2013).The emerging forest stands are important as carbon sinks, biodiversity habitats, and sources of timber and non-wood forest products (Brown and Lugo, 1990;UNDP, 2000;FAO, 2010).Tremendous progress has been made to describe post-disturbance secondary forest succession in abandoned farmlands and pasture (Guariguata and Ostertag, 2001;Chazdon, 2003;Norden et al., 2009).However, little attention has been paid to changes in species composition and stand structure of plantation forests that are established in such abandoned sites (Fawig et al., 2009;Kanowski and Catterall, 2010).Studies in post-disturbance natural forest recovery have demonstrated that woody species emergence pattern during secondary forest succession is determined to a large extent by the light environment and longevity of the species involved (Montgomery and Chazdon, 2001;Pooter et al., 2005;Wright et al., 2010).For instance, the first 100 years of secondary forest succession have been described to comprise about three successional phases (Finegan, 1996;Pena-Claros, 2003;Capers et al., 2005;Norden et al., 2009;Bonner et al., 2013).The first phase has been described to comprise herbs, shrubs and climbers that emerge soon after site disturbance.The second successional phase is characterised by the emergence of early pioneer tree species, which dominate the site for between 10 and 30 years, depending on their life span.The third successional phase is dominated by long-lived pioneers, which emerge as early pioneers die off.The emergence of shade-tolerant species and other late successional species is thought to occur continuously during the establishment of both early pioneers and long-lived pioneer species (Finegan, 1996;Pena-Claros, 2003;Bonner et al., 2013).
A few studies have indicated that there is forest succession in plantation forests, particularly in tropical forests (Parrotta et al., 1997;Brockerhoff et al., 2008;Fawig et al., 2009).Some of the studies have suggested that such plantations may eventually convert to or facilitate the formation of secondary forests (Fawig et al., 2009).One of the tropical forests where forest succession has been reported in plantation forests is the Kakamega Rainforest in western Kenya (Fawig et al., 2009).The discovery of gold in this forest in 1923 led to commercial logging, which commenced in the early 1930s to clear up areas for gold mining (Tsingalia, 1990;Mitchell, 2004;Ouma et al., 2011).The situation necessitated the gazettement of the forest, which was done in 1933 (Glenday, 2006).By 1952, about 15 % of the closed canopy forest had been cleared (Schaab et al., 2010).Commercial logging intensified between 1952 and 1985 leading to a further decrease in the closed canopy forest cover by about 63 % (Wass, 1995;Althof, 2005;Schaab et al., 2010).Whereas some of the logged forest sites regenerated naturally into secondary forest, some were settled in by forest adjoining communities who provided labour for both mining and logging activities (Lung, 2009;Tsingalia and Kassily, 2009).Other logged sites were placed under agriculture and later abandoned, while others were planted with both indigenous and exotic species as plantation forests with a view of providing timber in the future.The plantation forests comprised mixed indigenous species, indigenous monoculture species and exotic monoculture species, which were established between 1937 and 2005 (Lung and Schaab, 2006;KFS, 2010;Schaab et al., 2010 2009) that plantation forests of this rainforest were undergoing forest succession and were therefore likely to convert to secondary forests in the future, we sought to examine the similarities and differences between plantation forests and secondary forests in this forest from a successional perspective.A clear understanding of the successional pathway in these plantation forests and its likely outcome is useful in providing insight on whether to change the prime function of some plantations from timber production to ecosystem functions, such as carbon sequestration and biodiversity conservation.It is also an important decision support tool for forest managers regarding the kind of plantations to establish in the future, how to manage them and the kind of logging operation to apply.Thus, with the disturbed primary forest serving as a control, we carried out a chronosequence study in this forest to understand changes in tree species composition and stand structure in secondary forest stands and plantation forests between 1930s and 2013.The objective of the study was to describe successional pathways of secondary and plantation forests with regard to (i) woody species diversity (ii) similarity in species composition, (iii) distribution of successional species in forest canopy strata, (iv) the fate of planted trees in the successional process, and (v) the stand structure of secondary and plantation forests.

Study site
The study was carried out in Kibiri, Yala and Isecheno blocks of Kakamega Forest between April 2012 and December 2013.The forest is a mid-elevation tropical rainforest -an eastern relic of the African equatorial rainforest (Kokwaro, 1988;Fawig et al., 2009).It is located in western Kenya between latitudes 0 ° 10' N and 0° 21' N and longitudes 34° 47' E and 34° 58' E at an elevation of 1,600 m above sea level (Fashing and Gathua, 2004;Musila et al., 2010).The area experiences a hot and wet climate characterised by a mean annual temperature of 25 o C and an annual precipitation of 1,500 to 2,000 mm with a dry season between December and March (Otuoma et al., 2010).The forest is home to over 400 plant species (of which about 112 are tree species), 300 bird species and about seven endemic primate species (Kokwaro, 1988;Althof, 2005).The vegetation comprises a disturbed primary forest, secondary forests in different stages of succession, mixed indigenous forest plantations, indigenous and exotic forest plantations, and natural and man-made glades (Tsingalia and Kassily, 2009).Old-growth closed canopy natural forest stands are dominated by tree species such as Funtumia africana (Benth.)Stapf, Antiaris toxicaria Lesch., Ficus exasperata Vahl, Croton megalocarpus and Celtis mildebradii Engl.(Lung, 2009).The forest supports an adjoining human population of about 280,000 people who are distributed in surrounding farmlands and numerous isolated urban centres (Mutangah, 1996;KEFRI, 2010).The farmlands are fragmented into units measuring approximately 0.2 ha, each holding an average household of 8 to 12 people.Majority of these households are dependent on the forest to supplement their meagre agricultural produce.Some of the resources they obtain from the forest include fuel wood, timber, construction poles, medicine, fibre, pasture for livestock and indigenous fruits and vegetables (Althof, 2005;KEFRI, 2010).

Study design
The study design followed the understanding that data were collected from three forest blocks, which were managed as distinct entities within the same forest ecosystem.The forest blocks comprised nine different forest vegetation types, which were the treatments in the study.The forest vegetation types were disturbed primary forest, old-growth secondary forest, middle-aged secondary forest, young secondary forest, mixed indigenous plantation, Maesopsis eminii indigenous monoculture plantation, and Bischofia javanica, Cupressus lusitanica and Pinus patula exotic monoculture plantations.They were treated as sub-blocks nested within each of the three forest blocks.The forest vegetation types were delineated from forest compartment registers (KFS, 2010) and geo-referenced with the aid of existing base maps (Schaab et al., 2010) in Geovis remote sensing software (FAO, 2003).The computer generated maps were validated through field assessment.Data were collected from the sub-blocks using a variable area technique, which ensured that trees of different stem sizes were assessed in transects of different sizes to enhance the probability of obtaining tree data in equal proportions (NAFORMA, 2010;Nath et al., 2010).The sampling unit comprised a concentric sample plot of 30 m radius with stratified sub-plots of 15 m, 10 m, 5 m and 2 m radius from the center of the sampling unit.The sub-plots were nested within sample plots, which were also nested within the nine sub-blocks.The observational unit of the study was an individual tree.The study employed a nested experimental design (Kuehl, 2000;Onwuegbuzie and Leech, 2007).

Data collection
Stratified random sampling was employed to collect data on tree species types, tree height and stem diameter at breast height (DBH) from the sub-blocks (Gregoire and Valentine, 2007;Coe, 2008).Sample plots were randomly located in sub-blocks, but tree assessment was stratified on the basis of stem DBH.The 30 m radius plot was used to measure the height and DBH of trees > 50 cm DBH.The 15 m radius sub-plot was used to measure the height and DBH of trees of 20.1 -50 cm DBH.The 10 m radius sub-plot was used to measure the height and DBH of trees of 10.1 -20 cm DBH.The 5 m radius sub-plot was used to measure the height and DBH of trees of 5.1 -10 cm DBH; while the 2 m radius plot was used to measure the height and DBH of saplings ≥ 1.3 m in height, but with less than 5 cm DBH.Tree species were identified by their botanic names.Data on tree species were collected with the assistance of a plant taxonomist.Tree species that could not be identified in the field had their specimens collected and taken to the National Museums of Kenya herbarium for identification using previous collections.Data on tree DBH were obtained by measuring tree diameter in centimeters at 1.3 m above the ground using a diameter tape.The DBH of trees with a buttress was measured above the buttress.Tree height was measured in meters using a clinometer.Information on the age of disturbed sites, causes of disturbance and post-disturbance land use was obtained from forest compartment registers at the Kenya Forest Service office in Kakamega Forest (KFS, 2010).The information was corroborated through witness accounts by elderly members of forest adjoining communities who worked as casual labourers during commercial logging or forest plantation establishment.
Data on stem density, tree height and basal area among forest vegetation types were analyzed using analysis of variance in Genstat at 5% significance level (Sokal and Rohlf, 2012;VSN International, 2013) to determine possible variations between secondary and plantation forests.In situations where statistical significance was recorded, means were separated using the Ryan-Einot-Gabriel-Welsch Multiple Range Test (REGWQ) at 5% significance level (Krull and Craft, 2009;Sokal and Rohlf, 2012;Holt et al., 2013).Regression analyses were used to determine the relationship between tree species richness and stand age (Sykes, 1993).

Species composition
A total of 4,261 trees representing 85 woody species, 75 genera and 34 families were identified from an area of 36.82 ha.Seven woody species could not be identified.Among families, Euphorbiaceae and Moraceae had the highest number of species with each representing 10.6% of all the tree species.Rubiaceae and Sapotaceae had the second largest number of species with each accounting for 7.1 % of the woody species.Rutaceae and Ulmaceae tied for the third position with 5.9 % of the tree species.Approximately 54% of the tree species were represented in the main forest canopy, 62% were located in the sub-canopy, while 55% and 66 % were located in the understorey and shrub canopy layers, respectively.There was a significant variation in the representation of the 85 tree species among the nine forest vegetation types (F (1,8) = 12.30; p ˂ 0.001) (Table 1).A comparison of means indicated that the variation in species representation was caused by a significantly low number of tree species in Bischofia javanica and Cupressus lusitanica monoculture plantations, and the young secondary forest.Apart from the young secondary forest and Bischofia javanica and Cupressus lusitanica plantations, the number of tree species did not vary significantly between natural forest stands (disturbed primary forest, old-growth secondary forest and middleaged secondary forest) and other plantation forests (mixed indigenous plantation, and Maesopsis eminii indigenous monoculture plantation and Pinus patula exotic monoculture plantation) (F (1,5) = 0.96; p = 0.463).Apart from the young secondary forest, natural forest stands had higher Shannon diversity indices than plantation forests, which suggested that they were relatively more species-rich than the latter (F (1,8) = 5.46; p = 0.002) (Table 1).Among plantation forests, mixed indigenous, Maesopsis eminii and Pinus patula plantations had higher Shannon indices than Cupressus and Bischofia plantations.
Although the number of tree species did not vary significantly between natural forest stands and plantation forests except for Bischofia javanica and Cupressus lusitanica plantations and young secondary forest, a comparison of the distribution of tree species in the main canopy, sub-canopy, understorey and shrub canopy strata of the nine forest vegetation types indicated significant differences in species representation (F (1,3) = 3.38; p = 0.021).The number of tree species in the four forest canopy layers was significantly higher for oldgrowth secondary forest than all the plantation forests (Table 2).Middle-aged secondary forest had more tree species in the main canopy and shrub layer than all plantation forests, but it had comparable species representation with mixed indigenous, Maesopsis eminii and Pinus patula plantations in the sub-canopy and understorey layers.The disturbed primary forest had the largest number of tree species representation in the understorey and main canopy layers.The distribution of woody species in the young secondary forest, Bischofia javanica and Cupressus lusitanica forest plantations was significantly lower in all the canopy strata.Mixed indigenous plantation had significantly more tree species representation in the main and sub-canopy layers than other plantation forests.

Similarity of species composition to the primary forest
There was a significant variation among forest vegetation types in tree species similarity to the disturbed primary forest (F (1,7) = 8.08; p < 0.001).Tree species similarity to the disturbed primary forest ranged between 3.1% and 36.8%.The similarity of secondary forest stands to the disturbed primary forest increased with stand age.Oldgrowth secondary forest was more similar in species composition to the disturbed primary forest (36.8%) than middle-aged secondary forest (30.7%) and young secondary forest (3.1%) (Figure 1).Among plantation forests, mixed indigenous plantation had higher similarity to the disturbed primary forest (35.4 %) than Maesopsis indigenous monoculture plantation (26.3%), and Pinus (22.7%),Bischofia (14.7%) and Cupressus (14.2%) exotic monoculture plantations.

Effect of stand age on woody species richness
There was a strong relationship between stand age and woody species richness among secondary forest stands (y = 0.4107x + 2.7907; R 2 = 0.9722) (Figure 2).For instance, ten year-old young secondary forest stands had between four and five early successional woody pioneers, such as Psidium guajava L., Bridelia micrantha (Hochst.)Baill.and Harungana madagascariensis Poir.The species composition of 30 year-old middle-aged secondary forest stands comprised an average of 18 tree species per ha which consisted of both early successional pioneers and long-lived pioneers, such as Polyscias fulva (Hiern) Harms and Sapium ellipticum (Hochst.)Pax.together with one or two shade-tolerant species, such as Funtumia africana.In middle-aged secondary forest stands of about 50 years in age, most early successional pioneers had disappeared, the number of long-lived pioneers increased, while more shade-tolerant species, such as Ficus sur Forssk., Antiaris toxicaria and Albizia gummifera (J.F.Gmel.)C.A. Sm. were represented.For instance, Bridelia micrantha and Psidium guajava that were represented in ten and 30 year-old secondary forest stands had completely disappeared.The only early secondary pioneer species that was left was Harungana madagascariensis, which was present in the two earlier stands, but its abundance had decreased from 36 % in the ten year-old stand to 23 % in the 30 year-old stand to 14 % in the 50 year-old stand.The 50 year-old secondary forest stand had an average of 23 tree species per ha.In old-growth secondary forest stands of about 60 years in age, there were no early secondary pioneer species.The number of long-lived species had reduced, while the number of shade-tolerant species increased.For instance, Funtumia africana, Antiaris toxicaria, Trilepsium madagascariense, Croton megalocarpus, Ficus exasperata and Aningeria altissima (A.Chev.)Baehni., which were among the most abundant species in the disturbed primary forest, were represented among the ten most abundant species in 60 year-old secondary forest stands.The stand had approximately 28 tree species per ha.In 78 year-old secondary forest stands, the number of long-lived species had reduced further, while the number of shade-tolerant species increased.Comparing these stands with the 60 year-old secondary forest stands, the number of ten most abundant species in the disturbed primary forest increased from six to seven.The 78 year-old secondary forest stands had approximately 34 tree species per ha.In indigenous plantation forests, the relationship between stand age and woody species richness was strong, but relatively weaker than that of secondary forest stands (y = 0.3557x + 1.7687; R 2 = 0.6674) (Figure 3).
For instance, a 61 year-old mixed indigenous plantation lacked early secondary pioneers, but had more long-lived pioneers and fewer shade-tolerant species than a 76 year-old stand, while the latter had more shade-tolerant species and fewer long-lived pioneers.Apart from Prunus africana, Olea capensis, Trilepsium madagascariense and Zanthoxylum gilletii which were planted, the 61 yearold stand had another 21 woody species, which recruited through forest succession.The species composition of a 61 year-old mixed indigenous plantation forest resembled that of a 50 year-old middle-aged secondary forest stand.A 76 year-old mixed indigenous forest plantation had about 30 other woody species apart from planted species and resembled a 60 year-old secondary forest stand.In Maesopsis indigenous monoculture plantations, a 36 year-old plantation had 12 woody species; a 49 year-old plantation had 25 woody species, while a 68 year-old plantation had 22 woody species apart from Maesopsis eminii, which was planted.
In exotic plantation forests, the relationship between stand age and woody species richness was much weaker than that of secondary forest stands and indigenous plantation forests (y = 0.1419x + 6.7583; R 2 = 0.3326) (Figure 4).Increase in stand age did not appear to have a significant influence on the number of wood species.For instance, in Pinus patula plantations, a 43 year-old stand had 16 tree species, while 53, 59 and 68 year-old stands had 18, 15 and 16 woody species, respectively.The situation was similar in Bischofia and Cupressus plantations.In Bischofia plantations, 37 year-old stands had approximately 11 woody species, while 43 year-old stands had 10 species.In Cupressus plantations, 37 year-old stands had about 7 tree species, while 19 yearold stands had 12 species.

Stem density
Among families, Phyllanthaceae had the highest number of stems with 16.7% of all the tree stems.Moraceae was second with 11.4% of the stems, while Rhamnaceae was third with 8.2%.Apocynacea and Euphorbiaceae were fourth and fifth with 8.1 and 7.5%, respectively.Families with the least representation were Papilionaceae, Rhizophoraceae, Meliaceae, Guttiferae, Alangiaceae, Violaceae, Ebenaceae, Verbenaceae, Annonaceae, Loganiaceae, Boraginaceae, Pittosporaceae, Myrsinaceae and Melianthaceae with less than one percent of tree stems.Among tree species, Bischofia javanica had the largest number of stems at 16.7% of all the stems; Maesopsis eminii was second with 8.2%, while Funtumia africana was third with 8.1%.Rothmania longiflora, Synsepalum cerasiferum and Vitex keniensis were the least represented tree species with 0.29% of tree stems each.Considering that some of the most abundant species were planted and this certainly increased their abundance, we considered also nonplanted species separately and Funtumia africana emerged the most represented with 14.2% of the tree stems.Antiaris toxicaria was second with 6.9%, Psidium guajava was third with 5.3%, Trilepisium madagascariense was fourth with 5.2%, while Bridelia micrantha was fifth with 4.2% of all the stems.However, the list of least represented tree species did not change.
There was no significant variation in stem density for stems ≥ 0.1 cm DBH among the nine forest vegetation types (F (1,8) = 2.0; p = 0.113).Young secondary forest had the highest number of stems ≥ 0.1 DBH with 6,266.8 ± 140.4 stems ha -1 .It was closely followed by Bischofia plantations with 6,032.6 ± 304.2 stems ha -1 .Old-growth secondary forest had the least number of tree stems ≥ 0.1 DBH with 3,176.5 ± 563.3 stems ha -1 .Among tree stems ≥ 10 cm DBH, there was a significant variation in stem density among the nine forest vegetation types (F (1,8) = 5.63; p = 0.002).The variation was attributed to significantly fewer tree stems in old-growth secondary forest stands with 252.9 ± 36.51 stems ha -1 and significantly more stems in Bischofia and Maesopsis monoculture plantations with 542.1 ± 101.6 stems ha -1 and 447.0 ± 23.97 stems ha -1 , respectively (Table 3).

Tree height
There was a significant variation in tree height among the nine vegetation types (F (1,8) = 21.46;p ≤ 0.001).The variation was attributed to low tree height in middle-aged secondary forest, young secondary forest and Bischofia javanica monoculture plantation (Table 3).Apart from young secondary forest, all vegetation types had shrub, understorey, sub-canopy and main forest canopy layers.The young secondary forest lacked the main canopy and sub-canopy layers.

Basal area
There was a significant variation in basal area among the nine forest vegetation types (F (1,8) = 10.51;p < 0.001).The variation was attributed to lower basal area in young secondary and middle-aged secondary forests than the other forest vegetation types (Table 3).Although there was no significant difference in basal area between oldgrowth secondary forest and plantation forests, the former had a relatively lower basal area than mixed indigenous plantation.This suggests that forest plantations had relatively higher basal area than secondary forests of comparable age.The disturbed primary forest had a significantly higher basal area than all forest vegetation types.Analysis of the relationship between basal area, stem density and mean stem DBH indicated that stem DBH had a stronger influence on stand basal area than stem density (Table 3).For instance, Bischofia plantation had the highest stem density (stems ≥ 10 cm DBH), but it ranked fourth in basal area because its mean stem DBH was second lowest after young secondary forest.Similarly, young secondary forest was ranked third in stem density, but it had the lowest basal area because it had the lowest mean stem DBH.The disturbed primary forest had a significantly larger mean stem DBH than all forest vegetation types and this is likely to have contributed to its high basal area given that its stem density was not significantly different from other forest vegetation types.

Species composition
The results of this study confirm earlier indications that plantation forests were undergoing secondary forest succession in tropical forests (Parrotta et al., 1997;Fawig et al., 2009).The situation is exemplified by the fact that there was no significant different in the number of tree species between secondary forest stands and most plantation forests.Even the few monoculture plantation forests, such as Bischofia javanica and Cupressus lusitanica plantations, which had significantly fewer tree species than most secondary forests, had between 10 and 15 naturally recruited indigenous species.Thus, it is reasonable to argue that plantation forests in this rainforest do not differ from secondary forests in tree species composition.
Despite a pattern of tree species convergence between secondary forests and plantation forests, there were four key areas where the two forest types were dissimilar, namely: (i) secondary forest successional pathway, (ii) representation of successional species in canopy strata, (iii) persistence of planted trees in mature plantation forests, and (iv) effect of stand age on species composition.Whereas forest succession often began with early successional pioneers in secondary forest stands, the results of this study suggest that forest succession in plantation forests tended to bypass this light demanding early pioneers' stage.The process appeared to commence with the recruitment of long-lived pioneers and shade-tolerant species in plantation forests.As illustrated by Finegan (1996), Chazdon (2003), Pena-Claros (2003) and Bonner et al. (2013) on changes in species assembly in forest stands, it is likely that the formation of a canopy by planted trees in plantation forests facilitated the recruitment of long-lived pioneer and shade-tolerant species instead of early successional pioneers.This line of reasoning is supported by the fact that natural recruits were represented mostly in the shrub and understorey layers of plantation forests, which suggests that they often recruited when plantation trees had closed the forest canopy.Thus, whereas it took a fairly long period of time for long-lived pioneers and shade-tolerant species to recruit in secondary forests, it took a relatively shorter duration for the same tree species to recruit in plantation forests.The net effect is that secondary forests ended up having fairly similar species composition with plantation forests of comparable or younger age.
The other issue that this study considered was whether similarity in tree species composition would lead to the conversion of plantation forests to secondary forests as reported by Fawig et al. (2009).We looked at this observation from three perspectives, that is the representation of successional species in canopy strata, persistence of planted trees in mature plantation forests, and the effect of stand age on species composition.Our results indicated that successional species were represented in all the forest canopy layers in secondary forests, while in plantation forests, they were located in the shrub and understorey layers of monoculture plantations, and the shrub, understorey and sub-canopy layers of mixed indigenous plantations.The results indicated also that some mixed indigenous plantations were nearly indistinguishable from secondary forest while monoculture plantations had maintained a plantation physiognomy.Unlike secondary forests where late successional species had replaced early successional pioneers, planted trees persisted in plantation forests, which suggests that chances of shrub and understorey species reaching the main canopy remain slim at this stage.Moreover, the results illustrated that species composition changed with stand age in secondary forests and mixed indigenous plantation, but not in monoculture plantations.Thus, in the absence of logging operations or some major disturbance events that would remove trees in the main canopies of monoculture plantations as suggested by Parrotta et al. (1997) and Hardiman et al. (2013), it is unlikely that natural recruits would dislodge planted trees from the main canopies of monoculture plantations and convert the plantations to secondary forests any time soon.Consequently, the suggestion by Fawig et al. ( 2009) that plantation forests in Kakamega Forest are likely to develop into old-growth secondary forests in the near future may only apply to mixed indigenous plantations.The phenomenon is akin to findings by Pena-Claros (2003) and Van Breugel et al. (2007) that species changes during forest succession may occur slowly and it may take several decades for understorey and sub-canopy species to replace existing canopy species.In Kakamega Forest, the spacing of planted trees at about three meters may delay the process much longer because this spacing presents natural recruits with a great challenge in reaching the main canopy of mature forest stands.We suspect that it was easier for successional species to reach the subcanopy in mixed indigenous plantation forests because planted indigenous tree species grew at different rates, which provided recruits with a relatively better chance of competing with slow growers.

Stand structure
The results of the study indicated that the stem density of young and middle-aged secondary forests was not significantly different from that of plantation forests.However, the stem density of old-growth secondary forest stands was significantly lower than that of plantation forests.Although these results support findings by Fawig et al. (2009) in this forest that stem density can be as high in plantation forests as in secondary forest stands, they create a new phenomenon regarding stand structure in old-growth secondary forest.Since the stem density of disturbed primary forest, young secondary forest and middle-aged secondary forest was not significantly different from all the plantation forests, a significantly lower stem density in all old-growth secondary forest stands can be looked at from two perspectives: (i) old-growth secondary forest stands were undergoing successional transition after which their stem density would increase or (ii) the stem density of the disturbed primary forest should have been in the same range as old-growth secondary forest, but it was higher as a result of recovery from selective logging operations that it was subjected to between 1960s and early 1980s, which made its stem density to resemble that of young and middle-aged secondary forests.These suggestions notwithstanding, we do not have a conclusive explanation for the low stem density in old-growth secondary forests in this rainforest.
The observation that young and middle-aged secondary forests had significantly lower mean tree height than most plantation forests was consistent with those of Fawig et al. ( 2009) that there were variations in tree height between plantation forests and secondary forests.Given that the stand age of middle-aged secondary forests was comparable to that of most plantation forests and yet the former were relatively shorter, the results agree with those of Lung (2009) that plantation forests grow faster than secondary forests.However, the fact that the mean tree height of old-growth secondary was not significantly different most plantation forests of comparable age, supports findings by Pena-Claros (2003) and Ruiz et al. (2005) that tree height and basal area of secondary forests are positively correlated with stand age.
Our results indicated that plantation forests had relatively higher stand basal area than secondary forests of comparable age.The basal area of old-growth secondary forest stands was not significantly different from that of most plantation forests, but middle-aged and young secondary forests had significantly lower basal area than all the plantation forests.As illustrated by Montgomery andChazdon (2001), McElhinny et al. (2005) and Da Silva et al. (2012) on changes in structural complexity of forest stands, we suspect that the difference in basal area between secondary forest stands and plantation forests can be explained by the results on stem density and mean stem DBH.Generally, it takes natural forest stands a relatively longer duration to attain structural complexity similar to plantation forests.This phenomenon arises from the fact that early successional pioneers, which occupy a secondary forest for the first two to three decades, disappear as long-lived species take over.The emergence of long-lived pioneers and the subsequent disappearance of early successional pioneers (Finegan, 1996;Pena-Claros, 2003;Norden et al., 2009;Bonner et al., 2013) suggest that middle-aged forests, in which they are commonly found, may be aged between 30 and 50 years but are much younger in structural complexity with regard to tree height and mean stem DBH.In old-growth secondary forests, which mostly comprise late successional species, a majority of the trees have persisted long enough to enable them attain tree height and mean stem DBH that is either equal or greater than that of trees in mature plantation forests.In plantation forests, most planted trees persists to maturity and this gives them a competitive edge in basal area, stem DBH and tree height over secondary forests of comparable age.

Conclusion
Natural forest regeneration is on-going in both secondary and plantation forests in Kakamega Forest.There is presently no difference in tree species composition between the two forest types.However, natural recruits occupy all the forest canopy strata in secondary forests, but are represented mainly in the shrub and understorey layers in monoculture plantation forests, and the shrub, understorey and sub-canopy layers of mixed indigenous plantations.This suggests that forest succession in plantation forests commences after canopy closure and hence bypasses the light-demanding early pioneers' stage.Mixed indigenous plantations have become nearly similar to old-growth secondary forests, but monoculture plantations have maintained a plantation physiognomy.Structurally, younger secondary forest stands do not differ from plantation forests in stem density, but oldgrowth secondary forests have significantly lower stem density.The mean tree height and basal area of oldgrowth secondary forest stands do not differ from those of plantation forests of comparable age, but young secondary and middle-aged secondary forest stands have significantly lower mean tree height and basal area than plantation forests of comparable age.Overall, forest succession has progressed in plantation forests to an extent that mixed indigenous plantations are likely to perform ecosystem functions, such as carbon sequestration, biodiversity conservation and water catchment protection as effectively as secondary forests.Monoculture forest plantations, on the other hand, may be less effective in performing these functions, but more effective in timber provision.

Figure 1 .
Figure1.The similarity of different forest vegetation types to the disturbed primary forest in tree species composition in Kakamega Forest in western Kenya.

Figure 3 .Figure 4 .
Figure 3.A linear regression relationship between stand age and species richness in indigenous forest plantations in Kakamega Forest in western Kenya.
(KFS, 2010)digenous plantation forests comprised Olea capensis L., Croton megalocarpus Hutch., Zanthoxylum gilletii (De Wild.)Waterm.andPrunusafricana (Hook.f.) Kalkm.(KFS,2010).Indigenous monoculture plantations consisted of Maesopsis eminii Engl., Zanthoxylum gilletii and Prunus africana, while exotic monoculture plantations comprised Pinus patula Schlechtend.& Thonn., Bischofia javanica Blume and Cupressus lusitanica Mill.One common feature of these forest plantations is that they were planted at regular spacing of about 3 m between trees.The forest retained primary forest stands that were not subjected to commercial logging and were set aside as nature reserves.However, the nature reserves were exposed to selective logging by local communities between 1960s and early 1980s, which is estimated to have removed between 20 to 30% of the stems in the sub-canopy and main canopy layers (KEFRI, 2010).
Thus, the nature reserves are presently commonly referred to as near-natural forest (but to avoid confusion with old-growth secondary forest, we refer to them in this paper as disturbed primary forest).Following observations by Fawig et al. (

Table 1 .
Floristic composition and species diversity of different forest vegetation types in Kakamega Forest in western Kenya.

Table 2 .
Woody species representation in different forest canopy strata of nine forest vegetation types in Kakamega Forest in western Kenya.
Different superscripts in the same column denote significant difference.

Table 3 .
Relationship between basal area, stem density and mean stem DBH among different forest vegetation types in Kakamega Forest in western Kenya.