Can harvesting for timber in tropical forest enhance timber tree regeneration?

Forest Ecology and Management 314 (2014) 26–37

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Can harvesting for timber in tropical forest enhance timber tree regeneration? A. Duah-Gyamfi a,⇑, E.K. Swaine b, K.A. Adam a, M.A. Pinard b, M.D. Swaine b a b

Forestry Research Institute of Ghana, University P.O. Box 63, Kumasi, Ghana Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen AB24 3UU, UK

a r t i c l e

i n f o

Article history: Received 2 September 2013 Received in revised form 10 November 2013 Accepted 19 November 2013 Available online 18 December 2013 Keywords: Functional groups Ghana Logging disturbance Mortality Recruitment Soil seed bank

a b s t r a c t It has been suggested that the disturbance caused by logging of tropical forests impairs timber tree regeneration, whilst foresters view logging as a means to stimulate both the growth of the residual trees and the germination and establishment of new recruits. Whilst substantial increases in tree seedlings after logging have been shown, there have been no direct tests of whether this initial enhancement is maintained. We tested if the initial response to logging in a tropical forest in Ghana resulted in increases in seedling recruitment relative to unlogged parts of the same forest and whether subsequent seedling mortality was greater in areas disturbed by logging. We also compared seedling composition, density, species diversity and height 7 y after logging to see if disturbed areas differed from unlogged areas. Timber tree species seedling recruits were assessed after logging in 160 sample plots randomly located on skid trails, felling gaps and in unlogged parts of the same forest. Recruits were named, tagged, mapped, counted and heights measured on five occasions spanning 7 y after logging had ceased. Seedling recruitment was initially enhanced by logging disturbance, principally by Pioneer species, but after 1 y Non-Pioneers dominated the recruits. Mortality of recruits was initially greater in unlogged areas but declined over the 7 y of the study converging on that of disturbed areas to an annualised rate of 0.2. The resultant densities of surviving recruits declined after the initial increase, especially amongst Pioneers. After 7 y, Pioneers were least abundant. Species diversity was initially enhanced by logging disturbance but declined on skid trails after 3 y. In unlogged forest, diversity of recruits increased steadily, and converged on disturbed samples by 7 y. Species composition of recruits initially differed between unlogged and logged samples, but showed convergence after 3 y. The initial effects of disturbance were, however, still detectable at 7 y. The results confirm the initial enhancement of timber tree recruitment after logging disturbance which suffered no greater mortality than unlogged areas. Within 7 y of logging, however, most measures of seedling dynamics were similar in logged and unlogged forest presumably due to canopy closure. Whilst most of the tallest trees at 7 y were Pioneer timber species, Non-Pioneers were numerically dominant. We conclude that appropriately controlled logging does not impair timber tree regeneration. The benefits appear to be securing the establishment and fast growth of Pioneer timber species whose regeneration might be less successful in unlogged forest. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction An important expectation of the Selection System of natural forest management in the tropics is that harvesting operations will stimulate the growth of residual trees (Maitre, 1987) and enhance the regeneration of tree species, particularly those of commerce (Kuusipalo et al., 1996; Magnusson et al., 1999). These effects are argued to be due to reduced competition caused by gap formation in the canopy and soil, increasing light for growth of the residuals and stimulating the germination and establishment of new recruits from seeds. ⇑ Corresponding author. Tel.: +233 (0)3220 60123; fax: +233 (0)3220 60121. E-mail address: adgyamfi@csir-forig.org.gh (A. Duah-Gyamfi). 0378-1127/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.foreco.2013.11.025

There is good evidence for increased growth of residuals (e.g. Maitre, 1987; Gourlet-Fleury et al., 2013a,b), but this is accompanied by a period of higher mortality rates, largely due to damage to residual trees during extraction (Johns et al., 1996; Sist and Nguyen-The, 2002). There is also evidence for the stimulation of new seedling recruits in those parts of the forest disturbed by logging operations, (especially skid trails and felling gaps), compared with unlogged parts (Kuusipalo et al., 1996; Magnusson et al., 1999; Fredericksen and Mostacedo, 2000; Swaine and Agyeman, 2008). The potential benefits for regeneration are tempered, however, by various negative effects. The relaxation of competition offers benefits for all plants, not just timber tree species, but short-lived, non-commercial Pioneer tree species are the most numerous tree beneficiaries in studies where they have been included (Dickinson

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

et al., 2000; Fredericksen and Mostacedo, 2000; Arets, 2005; Park et al., 2005; Felton et al., 2006; Swaine and Agyeman, 2008; Ouédraogo et al., 2011; Gourlet-Fleury et al., 2013a). The effect is stronger in larger canopy openings where the soil is disturbed because these conditions break dormancy of seeds in the soil seed bank and because such tree species demand high irradiance for establishment and continued growth. Other non-tree Pioneers also benefit and may compete strongly with newly recruited commercial species. These include invasive exotics such as Chromolaena odorata (L.) King & Robinson in West Africa which is known to suppress the growth of tree seedlings in the field (Honu and Dang, 2000) and in a pot experiment with Ceiba pentandra and Pycnanthus angolensis (Gough, 2004) (Nomenclature follows Hawthorne, 1995, unless stated). In Bolivia, native species of Heliconia may also dominate in logging gaps (Felton et al., 2006) and liana proliferation is widely reported to follow logging (Schnitzer et al., 2004). These effects of competitors may be equivocal as germination and seedling growth of many tree species are maximal in less than full irradiance (Agyeman et al., 1999; Kyereh et al., 1999) so that a light canopy of Pioneers may ‘nurse’ tender seedlings of such species. These uncertainties are reflected in a lack of agreement amongst studies. In felling gaps, where soil disturbance is generally less than on skid trails, germination may be less and competition from residual vegetation greater than on skid trails (Fredericksen and Pariona, 2002). However, soil compaction due to skidding may prevent the establishment of seedlings from seeds whose dormancy was broken by the gap microclimate. Removal of topsoil during skidding will also displace the seeds contained within it (Pinard et al., 1996). Timber tree regeneration following logging may also be affected by the way in which the logging is done (Pena-Claros et al., 2008; Putz et al., 2008). The length and distribution of skid trails will affect the area of disturbance in addition to the number of trees felled. Different operators of extraction vehicles may have different impacts on the forest in terms of total area disturbed and the number of relict trees which are damaged. The type of extraction vehicle (tracked or wheeled) and the season of operations (wet or dry season) will affect the impacts on soil conditions (e.g. compaction, topsoil removal, etc.). In Sabah, Pinard et al. (2000) found that skid trails were still impoverished of woody stems 18 years after logging and in Ghana, Hawthorne et al. (1999) reported reduced stocking on skid trails up to 30 years after logging. Gourlet-Fleury et al. (2013b), in central Africa, concluded that disturbance caused by logging and subsequent thinning of non-commercial species and lianes was followed by rapid recovery of above-ground biomass but slow recovery of timber stocking which was not expected to be complete within the 25–30 years felling cycle. In contrast, Dekker and de Graaf (2003) suggested that forests logged under the CELOS management system recover the stocking of commercial species within 20 years whilst Okuda et al. (2003) found similar basal area and stem density in unlogged forest and forest regenerated with the Malayan Uniform System over 41 years after harvest. These management systems, unlike the Selection System, involve considerable intervention to achieve their results. The differential effects of logging practice on forest have been shown, for example, in comparisons of Reduced Impact Logging with traditional logging (Pinard et al., 2000; Sist et al., 2003; Mac Donagh et al., 2010), and we may expect similar contrasts between carefully controlled and uncontrolled traditional logging. In studies which showed poor stocking of trees on old skid trails, the presumed initial enhancement of tree recruitment would have been negated by greater mortality on skids trails in subsequent years. The outcome of the diverse possible effects of logging on tree regeneration should therefore be evident by examining logged forest over the years succeeding the harvest by comparing

27

sites of logging disturbance with the undisturbed parts of the same forest. The Selection System as operated in Ghana is controlled by the Resource Management Services Centre of the Ghana Forestry Commission which calculates the allowable harvest from 100% stock surveys, identifying individuals to be felled and those to be retained as ‘seed trees’ and completes pre- and post-harvest inspections. Logging operations in the forest follow the advice of the Forestry Commission (Ghana Forest Service, 1998) and are monitored by the relevant Regional Forest Office. Conservationists have claimed that sustainable management of tropical forests for timber is problematic (e.g. Rice et al., 1977) arguing that economic and political drivers cause overexploitation and permanent damage to the forest. In contrast, others argue that sustainable management is possible, if not often achieved (e.g. Dawkins, 1988). A first requirement of sustainable use is that the regeneration of timber tree species after logging is adequate to replace the extracted and damaged trees. In this study, we tested if controlled logging at two different intensities benefits timber tree regeneration in one Forest Reserve in Ghana. We tested three hypotheses: (1) the initial seedling recruits of timber tree species after logging are more numerous in parts of the forest disturbed by logging; (2) recruited seedlings suffer higher mortality in parts of the forest disturbed by logging than in areas unaffected by logging; and (3) within 7 years, the density of timber tree recruits is no greater in disturbed areas than in areas unaffected by logging. 2. Materials and methods The rationale for our methods was as follows. We established permanent sample plots on skid trails, in felling gaps and in unlogged parts of the forest in order to monitor the density, growth and mortality of tree seedling recruits for 7 y after logging. Recruits were identified to species so that the effects of logging on species diversity and composition could be monitored. We measured initial post-logging canopy openness and scored soil conditions in order to distinguish these two influences on seedling dynamics. We measured soil seed bank densities in order to assess its contribution to the recruits relative to the seed rain from current adults. 2.1. Study site The research was conducted in Compartment 4 (c. 134 ha) of Pra Anum Forest Reserve (123 km2, 6°110 to 6°200 N, 1°070 to 1°160 W; Fig. 1) in Ghana, which was reserved and gazetted in 1925. The reserve has been logged several times since 1954 under the control of the Ghana Forestry Commission (formerly the Forest Department). The forest falls within the moist semi-deciduous forest type (SE sub-type) of Hall and Swaine (1981) and is mostly of gentle topography at about 120 m asl rising to about 160 m (Fig. 1). The tallest trees are about 50 m. Mean annual rainfall at Amantia over a 5 y period prior to 1963 was 1620 mm, slightly higher than longerterm means at two other stations within 15 km (Ntronang, 1538 (25 y); Akokoase, 1543 mm (25 y)). The 45 y mean for Akim Oda (40 km S) was 1580 mm prior to 1963 but the mean at Akim Oda for 2000–2005 was less at 1377 mm. It is not clear if this is a reflection of the general decline (Nicholson et al., 2000) in West African forest rainfall. There are 4 months each year with rainfall < 100 mm, with peaks in June and October. Soil conditions, as judged from a nearby single bulked sample (No. A65 in Hall and Swaine (1976)) are: pH 5.1, total exchangeable bases 4.2 m-equiv/100 g, total nitrogen 0.22% (Kjeldhal), available phosphorus 6.6 ppm (fluoride extraction), 33% sand and 21% clay (Boyoucos particle size analysis).

28

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

Fig. 1. Map of Pra Anum Forest Reserve (shaded) with allowed farms in lighter shading. Compartment 4 is shown by a white polygon. Dashed lines: roads and tracks; dotted lines: 50 ft (15 m) contours, thin solid lines: rivers and streams; thick solid lines: reserve boundaries and rivers. Compiled from: 1:50,000 map sheet 0602D2, Ghana Survey Department from 1973 air surveys and Landsat scene 194/56 of 15 January 2002.

Fig. 2. Map of Compartment 4 in Pra Anum F.R. showing division into 4 treatment blocks, logging road, skid trails and sample plot locations for main skid trails, secondary skid trails, felling gaps and unlogged forest.

2.2. Logging Compartment 4 was divided into four approximately equal areas, each 400  800 m (32 ha) designated as Blocks A, B, C, and D. Block A was somewhat larger than the other blocks (ca. 38 ha,

Fig. 2). Each block was subject to different planned logging intensities by the logging concessionaire, Topbell Integrated Company (formerly Atwima Timbers). Prescribed felled tree volumes were 26 and 53 m3 ha1 (the calculated annual allowable cut (AAC) in Blocks C and D, and twice the AAC in Blocks A and B).

29

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37 Table 1 Number of trees harvested in four treatment blocks and the areas disturbed by associated skid trails. Block

Area (ha)

Number of trees felled

Felled trees (ha1)

Volume (M3 ha1)

Skid trail area (ha)

Skid trail area (%)

A B C D

38 32 32 32

154 187 85 102

4.8 5.8 2.7 3.2

53 53 26 26

2.31 1.76 1.35 1.24

7.2 5.5 4.2 3.8

The actual fellings and associated skid trail areas are given in Table 1. Each block was harvested separately by the same concessionaire following instructions to minimise skidding lengths and canopy opening. In Blocks B-D the recommended dendritic pattern of skidding was achieved, but in Block A, skid trails were less well-organised due to an inexperienced skidder operator (Fig. 2). Extraction was mostly by tracked vehicle, though a large-wheeled vehicle was also used. The operations were monitored and controlled by one of us (KAA) as part of a parallel study (Adam, 2003). Logging of the four blocks was completed in July and August 2000, at the end of the main rainy season and into the short dry season. The skid trails and felled trees were then mapped with prismatic compass and tape (Fig. 2).

2.3. Seedling sample plots Post-logging seedling recruitment was enumerated by 10 randomly-located sample plots in each block and in each of four categories of forest: main skid trail, secondary skid trails, felling gaps and locations within the areas of the forest unaffected by logging, giving a total of 160 permanent sample plots. Plot corners were marked with 1 m long, 5 mm diameter steel rods. On skid trails, plot width was determined by the width of the skid trail, spanning the disturbed soil and including the berms on either side. Plot length was 10 m so that plot area varied somewhat, but 89% were between 40 and 60 m2. For felling gaps, sample plots were 2  30 m running the length of the felling gap from the stump to the crown zone. Samples in unlogged areas of the forest were also

Table 2 Species recorded as recruits during five enumerations over 7 years in 160 sample plots (0.864 ha). Species

Family

Timber class (pre-1981)

Timber class 1989a

Number of seedlings

Albizia ferruginea Albizia zygia Alstonia boonei Amphimas pterocarpoides Aningeria robusta Antiaris toxicaria Bombax brevicuspe Canarium schweinfurthii Ceiba pentandra Celtis mildbraedii Celtis zenkeri Cylicodiscus gabunensis Daniellia ogea Dialium aubrevillei Distemonanthus benthamianus Entandrophragma angolensis Entandrophragma candolei Guarea cedrata Khaya ivorensis Mansonia altissima Milicia excelsa Nauclea diderrichii Nesogordonia papaverifera Petersianthus macrocarpus Piptadeniastrum africanum Pterygota macrocarpa Pycnanthus angolensis Ricinodendron heudelotii Sterculia rhinopetala Strombosia glaucescens Terminalia superba Triplochiton scleroxylon Turreanthus africanus

Fabaceae Fabaceae Apocynaceae Fabaceae Sapotaceae Moraceae Bombacaceae Burseraceae Bombacaceae Ulmaceae Ulmaceae Fabaceae Fabaceae Fabaceae Fabaceae Meliaceae Meliaceae Meliaceae Meliaceae Sterculiaceae Moraceae Rubiaceae Sterculiaceae Lecythidaceae Fabaceae Sterculiaceae Myristicaceae Euphorbiaceae Sterculiaceae Olacaceae Combretaceae Sterculiaceae Meliaceae

3 3

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 2580 166 4 78 222 2 6 3317 837 1241 178 2 10 136 189 1 35 55 252 843 137 1755 8 55 69 3 1025 416 74 81 531 2302

Albizia adianthifolia Blighia sapida Cleistopholis patens Cola gigantea Sterculia oblonga Sterculia tragacantha

Fabaceae Sapindaceae Annonaceae Sterculiaceae Sterculiaceae Sterculiaceae

2 2 2 2 2 2

374 69 239 100 37 2

Cedrela odorata Musanga cecropioides

Meliaceae Moraceae

Invasive exotic Non-timber

705 4714

2b

3 3 3

3 1a 2a 2a 1a 2b 1a 1a 2b 3 2a 3 3 3 3 1b 2b 3

a Class 1 species are defined as those registered as exported from Ghana between 1973 and 1988. The Class 2 species included here have been sold for timber since the classification was defined in 1989 (Wong, 1989: pp. 40–41).

30

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

2  30 m, with a starting location and direction of sample determined at random, but avoiding logging disturbance. Natural gaps were not excluded. Where necessary for analysis, seedling counts were standardised to density (m2). Each plot was given a score on a 5-point scale for soil condition as follows: (1) topsoil undisturbed; (2) topsoil disturbed but not removed; (3) topsoil removed; (4) topsoil removed and subsoil rutted; and (5) topsoil removed, subsoil rutted and puddled. Few samples were allocated to category 3 and were pooled with category 4 in analyses. The number of logs which had been skidded over each sample plot was estimated from the map in Fig. 2 by assuming that each felled tree was removed from the forest by the shortest skidding route. Felling gap plots were assigned 1 log. Canopy openness was measured for each plot during the first enumeration in August 2000 (the short dry season) when the sky was overcast, using hemisphere photographs taken with an 8 mm Sigma fisheye lens on a Nikon F30 camera body. Images were taken at 1 m height with the lens vertical on a tripod. One image was taken at the centre of each skid trail sample plot and two along each transect in felling gaps and in unlogged forest samples. Canopy openness was estimated with Winphot 5.0 (ter Steege, 1996) using images converted to B&W in Photoshop 7 and a standard image to define the hemisphere circle. Post-logging seedling recruits were enumerated on five occasions at approximately 0.5, 4.2, 10, 32 and 85 months (0.04, 0.35, 0.83, 2.7 and 7.1 years) after logging ceased in each sub-compartment. Each tree seedling was identified, mapped within the sample plot, tagged and its height measured. Timber Class 1 species were recorded as well as five Class 2 species, the abundant short-lived Pioneer Musanga cecropioides, and the invasive exotic Cedrela odorata L. which had been introduced in the 1950s and had become so common that it was included amongst the trees harvested (Table 2). During the re-enumerations, surviving seedlings were re-measured for height; searches were made for missing seedlings and their tags to confirm mortality. New seedlings were identified, tagged, mapped and their heights measured. For plants which exceeded about 2 m in height, a 5 m telescopic measuring pole was used (Stanley Mabo). Species names and their functional group follow Hawthorne (1995): Pioneers (P) cannot establish in forest shade in contrast with Non-Pioneers, which are of two kinds - Non-Pioneer Light-Demanders (LD) establish in shade but require increased irradiance to survive, whilst Non-Pioneer Shade-Bearers (SH) also establish in shade but can maintain growth without further increase in irradiance and their seedlings are often at high densities in shade (Hawthorne, 1995). The soil seed bank was assessed by a random sample in each of the four blocks and on two occasions in an attempt to include any seasonal differences. Samples were located in areas undisturbed by logging activities in order to capture the pre-logging soil stock. On the first occasion (March 2003), there were 52 samples (13 per block) of 5  5 cm and 5 cm deep. On the second (July 2004), 48 samples (12 per block) were each 5  10 cm and 5 cm deep. Each sample was spread 5 mm deep (Dalling et al., 1994) on a 3 cm deep bed of seed-free builders’ sand in wooden trays and placed in a screened shade-house with about 30% of full irradiance, watered regularly and germination recorded at 3-day intervals until no further germination occurred. Seedlings were counted and removed after identification. Tree species and other common species were identified; other species were allocated to life-form. 2.4. Analyses Response variables were tested for normality with the Kolmogorov–Smirnov procedure. Normally distributed variables (loge transformed where necessary) were analysed with ANOVA; others with the Kruskal–Wallis H-test or with X2 as indicated in

the text. Between-category comparisons for non-parametric analyses were tested with the multiple comparisons macro KrusMC in Minitab 14. Species diversity was measured by 1/Simpson’s Index following Hill (1974). Seedling composition was analysed with Detrended Correspondence Analysis (DCA) in PCord 5 (McCune and Mefford, 1999) in preference to the un-detrended version (Reciprocal Averaging) for the benefit of improved axis scaling. 3. Results 3.1. Block effects (logging intensities) Because the extraction rates differed amongst the four blocks (Table 1), the disturbed area differed amongst the blocks. The higher extraction rates (Blocks A and B) created an average of 2.03 ha of skid trail area per block compared with 1.3 ha for Blocks C and D (Table 1). Despite these differences there was no significant difference amongst blocks for sample canopy openness (1-way ANOVA, F3,156 = 1.06, P = 0.369). Total recruitment for each block at each of the five enumerations did not differ amongst blocks (Kruskal–Wallis test: H = 0.93, DF = 3, P = 0.817). Mortality rates for each cohort of seedlings in each block were calculated following Sheil et al. (1995) over the full period for which each cohort was recorded and there were no significant differences amongst blocks (1-way ANOVA, F3,12 = 0.04, P = 0.988). Species diversity (1/Simpson’s Index, all enumerations pooled, differed amongst blocks (Kruskal– Wallis: H = 26.29, DF = 3, P = 0.000). Blocks A and B had slightly higher diversity (3.25 and 3.14) than Blocks B and C (2.49 and 2.61). Soil seed bank densities of all species in the second survey in unlogged areas of the forest did not differ significantly amongst blocks (1-way ANOVA of loge[number of seeds + 1]: F3,44 = 0.21, P = 0.891). Species composition as measured by the median Axis 1 scores of the sample plots for the four blocks for three of the five enumerations of a Detrended Correspondence Analysis did not differ significantly amongst blocks (Kruskal–Wallis tests, (DF = 3), H = 5.13, 1.31, 7.02, P = 0.162, 0.072, 0.071 respectively). The fifth enumeration had marginally different median axis values amongst blocks (H = 8.91; P = 0.031). In subsequent analyses, the data for the four blocks are pooled. 3.2. Environmental conditions Canopy openness at 1 m above ground was increased by tree felling and log skidding, but did not differ significantly amongst skid trails and felling gaps (mean 14.4 ± 0.54%) but was significantly higher (P < 0.05) than unlogged forest (mean 4.9 ± 0.44%). Greatest openness (>20%) was evenly distributed amongst main and secondary skid trails and felling gaps. Openness did not exceed 10% in unlogged forest where sampling included a few natural canopy gaps. The effect of log skidding on soil conditions (categories 1, undisturbed, to 5, rutted and puddled) were significantly related to the number of logs passing over the sample plot (Spearman’s rank correlation = 0.824, DF = 158, P = 0.000), suggesting that the subjective measure of soil condition captured the main effects. Greatest disturbance (category 5) was almost confined to the main skid trail (Fig. 3). All unlogged forest samples were ‘undisturbed’ as well as 15 of the 40 felling gap samples (Fig. 3). These results show that unlogged forest is more shaded with undisturbed soil, whilst the felling gaps differ from skid trails in having less soil damage. 3.3. Seedling recruitment Total seedling recruitment was recorded at each of the five enumerations, but rates of recruitment over different periods cannot

31

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

40

Number of plots

35 30

100

Undisturbed Topsoil disturbed Topsoil removed soil rutted Ruts & puddles

Cohort 1 (0.5 mo)

90 80 70 60

25

50

20

40 30

15 10

20 10

5

0 Felling gaps

0

MST

SST

FG

unL

Skid trails

unlogged

100

Disturbance category

90

Cohort 2 (4 mo)

80 Fig. 3. Frequency of soil disturbance categories amongst logging disturbance categories.

3.4. Seedling mortality

60 50 40 30 20 10 0 Felling gaps

90 80 70 60

P LD SH

unlogged

Cohort 3 (10 mo)

50 40 30 20 10 0

Felling gaps

Skid trails

unlogged

100 Cohort 4 (32 mo)

90 80 70 60 50 40 30 20 10 0 Felling gaps 100 90 80 70 60 50 40 30 20 10 0

Skid trails

unlogged

Cohort 5 (85 mo)

Felling gaps

Pioneer seedling mortality exceeded that of other functional groups for all cohorts, in all disturbance categories and for all enumeration intervals (Fig. 5). Mortality rates in unlogged forest were significantly greater than in disturbed forest (skid trails and felling gaps) for the first cohort in the three periods up to 2.6 y (X 23 ¼ 22:6,

Skid trails

100

% in disturbance class

be compared because of differences in the interval between enumerations. The distribution of recruits amongst disturbance categories within an enumeration can, however, be usefully compared. Initial seedling densities (Enumeration 1) in skid trail, felling gap and unlogged forest samples were 0.38, 0.35 and 0.07 m2, respectively. In Fig. 4 the percentage of recruits within a disturbance category are shown for the three functional groups of species. It is evident that for the first two enumerations, covering the first 4 months after logging, Pioneer species dominate the recruitment on skid trails and in felling gaps. By enumeration 4 (ca. 3 years after logging), Pioneers represented a small percentage of the surviving recruits. At 7 y, Non-Pioneer Shade-Bearers were the leading recruits in all disturbance categories. Total seedlings (density) recruited over the 7-year period were significantly fewer in soil condition Class 1 (undisturbed; 1-way ANOVA of loge-transformed data: F3,156 = 8.16, P = 0.000) than in soil conditions 2–5 between which there were no significant differences. Total recruitment was also significantly lower (1-way ANOVA: F4,155 = 4.19, P = 0.003) in the most shaded canopy openness classes (0–5%) than in more open canopies which did not differ significantly. Since there were no differences amongst disturbed soil categories, these results show that although soil disturbance increased recruitment, it was unaffected by the degree of disturbance. The great majority of these new seedlings were recruited from the seed rain because the soil seed bank contained seeds of only three (Pioneer) timber species (Table 3). The total number of seedlings raised from the two surveys pooled was 352, of which 66 (18.8%) were tree species (1.78  106 ha1), but of these, only 10 (2.8%, 2.7  105 ha1) were timber species (bold in Table 3). Of the non-tree species, the invasive exotic, C. odorata was very abundant (75% of all seedlings in the second survey). Seedling densities observed in the sample plots were substantially lower than those estimated for the soil seed bank (Table 3). This may be expected because of high initial mortality in the forest and the unknown length of time after germination when the seedlings were recorded, but the soil stock is evidently sufficient by itself to account for the seedlings of Ceiba, Milicia and Nauclea that were recruited.

70

Skid trails

unlogged

Disturbance category Fig. 4. Surviving timber species seedling recruits as recorded at the five enumerations expressed as % in forest disturbance category. Note that the intervals between enumerations are not equal.

32

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

Table 3 Numbers and densities of seeds germinated from 5 cm deep random soil samples on two occasions in unlogged parts of the forest. Seedling density (from Table 2) was based on the 120 sample plots disturbed by logging (6240 m2). Species

a

Number of seeds germinating

Mean density of seeds (m2)

Mean seedling density (m2)

% Frequency (2nd sample)

3 2 1 1 0 24 24 3 245a

16.2 8.1 2.7 8.1 5.4 64.8 64.8 35.1 772.9

0.53 0.13 0.02 0.74

4 2 2 2 0 13 21 2

303 48 0.24

978.1 100 0.37

1st Sample

2nd Sample

Ceiba pentandra Milicia excelsa Nauclea diderrichii Musanga cecropioides Ficus spp. Solanum erianthum Trema orientalis Rauvolfia vomitoria Other species

3 1 0 2 2 0 0 0 41

Total Number of samples Total sample area (m2)

49 52 0.13

48

93% Chromolaena odorata.

1.0

FELLING GAPS

SKID TRAILS

UNLOGGED FOREST

0.8 0.6 0.4 0.2 0.0 1.0 0.8

Annualised mortality

0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8

P

0.6

LD

0.4

SH

0.2 0.0 0.5-4 mo 4-10 mo 10-32 mo 32-85 mo

0.5-4 mo 4-10 mo 10-32 mo 32-85 mo

0.5-4 mo 4-10 mo 10-32 mo 32-85 mo

Enumeration interval Fig. 5. Annualised seedling mortality rates (Sheil et al., 1995) for species guilds in three categories of forest disturbance over four cohorts of seedling recruitment: top row – cohort 1 (0.5–4 months), second row – cohort 2 (4–10 months), third row – cohort 3 (10–32 months), fourth row – cohort 4 (32–85 months).

11.1 and 19.0, P < 0.001, P < 0.02 and P < 0.001, respectively for the three periods) and for the second cohort in the period between 4 and 10 months after logging (X 23 ¼ 135:6, P  0.001). Thereafter, mortality rates did not differ clearly between disturbance categories, presumably because of canopy closure. Annualised mortality rates for all species pooled declined significantly with time for each cohort (r = 0.773, DF = 25, P < 0.001) (Fig. 6) converging to a mean of 0.16 ± 0.1 for the 3–7 years interval. These results show that the high recruitment due to logging disturbance initially suffers less mortality than recruitment in unlogged areas of the forest. Mortality rates for the four seedling cohorts (all species pooled) did not differ significantly amongst the four soil condition classes

(1-way ANOVA: F3,12 = 0.38, P = 0.771) nor with canopy openness (1-way ANOVA: F4,19 = 0.13, P = 0.969). The higher mortality in unlogged forest must therefore be due to some other feature of the unlogged parts of the forest. 3.5. Seedling density The net result of mortality and recruitment amongst functional groups over time for the four disturbance categories is shown in Fig. 7. Net seedling density shows an initial rapid increase, followed by a decrease between 1 and 3 y, due to the high mortality of Pioneers and of young seedlings following heavy mast years,

33

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

Annual mortality

1.0

Cohort 2

Cohort 1

0.8

Felling gaps

Cohort 3

Skid trails

0.6

Unlogged

0.4 0.2 0.0 0

1

2

3

4

5

0

1

2

3

4

5

0

1

2

3

4

5

Time since logging (years) Fig. 6. Decline in annualised mortality rates over time after logging for three seedling cohorts for each species guild. The time axis is the enumeration mid-point.

45

1800

(a) Pioneers

40

Number of trees >5 m tall

1600 1400

FG 1200

MST

1000

SST

800

unL

600 400

P

35

LD 30

SH

25 20 15 10 5

200

0 0 0

2

4

6

8

Main skid trails

Felling gaps

Unlogged forest

Disturbance category

1800

Seedling density (40 plots)

Secondary skid trails

(b) Non-Pioneer Light-Demanders

1600

Fig. 8. Distribution of timber species recruits >5 m tall (total over all 160 samples) at 7 y after logging amongst disturbance categories, by functional group.

1400 1200 1000 800 600 400 200 0 0

2

4

6

8

1800

notably in Albizia zygia, Milicia excelsa and the introduced exotic, C. odorata. Subsequently, further recruitment, especially of NonPioneer Shade Bearers on skid trails (Fig. 7c), maintained seedling densities. After 3 years, Non-Pioneer species were the most numerous seedlings in both logged and unlogged forest and Pioneers least numerous, though maintaining higher densities in the logged forest samples (Fig. 7). In terms of height, however, the leading trees 7 y after logging were Pioneers: of 104 trees exceeding 5 m, 79% were Pioneers, 13% were LD and only 8% SH. Most of the trees >5 m tall (84%) were on skid trails or in felling gaps (Fig. 8). 3.6. Species composition

(c) Non-Pioneer Shade-Bearers

1600 1400 1200 1000 800 600 400 200 0 0

2

4

6

8

Time since logging (y) Fig. 7. Changes in mean seedling density over 7 years since logging in the four forest disturbance categories and for each functional group. Error bars (±2 s.e.) are too small to be visible except for shade bearers on skid trails at Enumeration 2 and LDs on skid trails and unlogged logged forest at Enumeration 3 (not shown).

Of tree species recorded in the enumerations, thirty-eight are used in these analyses as of timber value, including the exotic C. odorata which was introduced in the 1950s and has since invaded many parts of the reserve (Table 2). The species with the most numerous seedlings, M. cecropioides (25% of all seedlings), is an extreme Pioneer with very low density wood, and of no commercial value and was excluded from these analyses. Over 3000 seedlings of Musanga were recruited between enumerations 1 and 2 in 68% of the logged forest sample plots; none were recruited in the unlogged parts of the forest, where there was no soil disturbance, suggesting that the species was recruited from the soil seed bank rather than the seed rain (Table 3). Most species were recorded in small numbers at the first enumeration, but several did not appear in significant numbers until later enumerations, reflecting annual differences in seed dispersal. Notable amongst them was Triplochiton scleroxylon which was

34

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

absent from the first two enumerations and only produced a substantial number (495) of seedlings between 3 and 7 years (Fig. 9). This species is notorious for its erratic seed production and short seed viability (Taylor, 1960; Hawthorne, 1995). Species diversity (1/Simpson’s Index) in the disturbed samples increased during the first three years, and subsequently showed little change but the increase was initially slower in unlogged forest. At 7 years, however, species diversity in the unlogged samples had converged on those of the felling gaps and skid (Fig. 10). Changes in timber species composition were described with DCA. Fig. 11a shows the change in mean Axis 1 values for each disturbance category in the five enumerations. Skid trail and felling gap samples follow a closely similar change in species composition converging on unlogged forest samples. In Fig. 11b, the species Axis 1 values show the early predominance of Pioneer species gradually superseded by Non-Pioneers, a classic secondary succession trend. Mean Axis 1 values for the 13 Pioneer species were significantly less than for the 18 LD and 9 SH species (1-way ANOVA, F2,37 = 9.21, P = 0.001). The second axis of this analysis (not shown)

4. Discussion

3000 Ceiba pentandra

The areas of skid trails created by the different intensities of logging differed by a factor of 1.6 and of number of trees felled by 1.8 (Table 1, data of Adam (2003)), indicating that a higher intensity of harvesting does not necessarily lead to a proportionally larger disturbed area. The seedling data, however, would not be expected to differ between the two logging intensities because they are based on densities. However, the larger area of skid trails and greater number of felling gaps would initially increase the total number of new recruits in the forest as a whole.

Ricinodendron

Total seedlings (160 plots)

2500

Triplochiton scleroxlyon Albizia zygia Celtis zenkeri

2000

Sterculia rhinopetala Nesogordonia papaverifera Turreanthus africanus

1500

1000

4.1. Seedling recruitment 500

0 0

1

2

3

4

5

6

7

8

Time (years since logging) Fig. 9. Change in total number of selected timber species seedlings over 7 years since logging; all 160 sample plots pooled. Open symbols, Pioneers, closed symbols Non-Pioneer Shade-Bearers, other symbols: Non-Pioneer Light-Demanders.

5

Species Diversity (1/Simpson's index)

described the somewhat erratic and temporary diversions of the unlogged samples due to prolific seed production in some species, notably Albizia species, Turreanthus and Cylicodiscus. To test if seedling composition after 7 years was related to initial environmental conditions caused by logging, a further DCA was done using only the seedlings surviving at enumeration 5 (7 y after logging). Axis 1 was negatively correlated with initial soil disturbance (r = 0.364), with initial canopy openness (r = 0.474) and with percentage of Pioneer seedlings (r = 0.611), and positively correlated with percentage of SH seedlings (r = 0.738). These relationships suggest that the initial environmental effects of logging are still evident in the main differences (Axis 1) in species composition of recruited seedlings after 7 years. Comparing axis values amongst disturbance categories, there were no significant differences amongst the four disturbance categories (treating main and secondary skid trails separately), but unlogged forest axis values were significantly higher than all other categories (1-way ANOVA: F3,156 = 13.01, P = 0.000).

4

Initial recruitment of tree seedlings was substantially greater in the parts of the forest disturbed by logging than in unlogged areas, confirming previous studies (Kuusipalo et al., 1996; Magnusson et al., 1999; Fredericksen and Mostacedo, 2000; Duah-Gyamfi, 2007; Swaine and Agyeman, 2008) and supporting our first hypothesis. Recruitment for the period 10–32 months, however, was similar amongst all three disturbance categories and remained so for the two succeeding periods up to 7 y (Fig. 4). Evidently, the enhancement due to logging disturbance is temporary. On skid trails and felling gaps the initial enhancement was due principally to Pioneer species (75–97%) as has been reported in other studies (Dickinson et al., 2000; Fredericksen and Mostacedo, 2000; Arets, 2005; Park et al., 2005; Felton et al., 2006; Duah-Gyamfi, 2007; Swaine and Agyeman, 2008), however, between 32 and 85 months Pioneer recruitment was notably less than Non-Pioneers (<20% of recruits within a disturbance category; Fig. 4). This change in representation of functional groups may be attributed to the closure of the canopy in the disturbed areas and to the closure of the few natural gaps in the unlogged samples.

3

4.2. Seedling mortality Felling gaps 2

Main skid trail Secondary skid trails Unlogged forest

1 0

1

2

3

4

5

6

7

8

Time (years since logging) Fig. 10. Changes in species diversity (1/Simpson’s Index) over 7 years since logging. Error bars are ± 1 s.e.

The mortality rates of these recruits were greatest for Pioneers amongst all cohorts and for all periods and in all disturbance categories (Fig. 5), confirming this as a characteristic of the species. Mortality rates declined over the seven post-logging years to less than 0.2 (Figs. 5 and 6) but were initially greater in the unlogged samples compared with the disturbed areas. This result denies the our hypothesis that mortality would be greater in disturbed areas and casts doubt on it being the cause of the poor stocking reported by Pinard et al. (2000) and Hawthorne et al. (1999). Mortality appeared to be unrelated to either soil conditions or canopy openness, so that

35

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

400

(b) Species axis 1 score

(a) Sample axis 1 scores

Turreanthus africanus Amphimas pterocarpoides Distemonanthus benthamianus Aningeria robusta Albizia adianthifolia Pterygota macrocarpa Strombosia glaucescens Triplochiton scleroxylon Cylicodicodiscus gabunensis Cola gigantea Sterculia rhinopetala Khaya ivorensis Entandrophragma angolensis Nesogordonia papaverifera Pycnanthus angolensis Guarea cedrata Albizia ferruginea Antiaris toxicaria Daniellia ogea Sterculia oblonga Celtis zenkeri Albizia zygia Mansonia altissima Blighia sapida Bombax brevicuspe Dialium aubrevillei Cedrela odorata Piptadeniastrum africanum Celtis mildbraedii Hymenostegia afzelii Petersianthus macrocarpus Terminalia superba Nauclea diderrichii Alstonia boonei Milicia excelsa Musanga cecropioides Cleistopholis patens Ceiba pentandra Ricinodendron heudelotii Canarium schweinfurthii

350

DCA Axis 1

300

250

200

FG

150

MST SST unL

100 0

1

2

3

4

5

6

7

8

-100

Time since logging (yrs)

0

100

200

300

400

500

DCA axis 1

Fig. 11. (a) Change with time of mean DCA Axis 1 scores (±1 s.e.; N = 10) for the seedling samples in each disturbance category (FG, felling gaps; MST, main skid trail; SST, secondary skid trails; unL, unlogged forest). (b) Axis 1 scores for the species: open bars, Pioneers; grey bars, LD; solid bars, SH. Axis 1 eigenvalue = 0.530.

the initially higher mortality in unlogged forest must be due to another effect, perhaps competition for soil resources in the denser vegetation. The general decline in rates and loss of differential mortality rates amongst disturbance categories implies a convergence, most probably due to canopy closure. The effect of higher mortality rates in unlogged forest is therefore temporary.

nance by Pioneers has disappeared and there is some convergence with the undisturbed forest. This also seems to be the case for species diversity where the initial increase in disturbed areas halted between 1 and 3 y whereas the unlogged samples showed a slower but continuing increase, to converge on the disturbed samples by year 7 (Fig. 10).

4.3. Seedling density

4.4. Seedling height growth

The effects of these differences in recruitment and mortality on net densities of seedlings over the 7-year period show a rapid increase during the first year, followed by a decline (Fig. 7) probably due to self-thinning as was recorded by Swaine and Hall (1983) for colonisation of an abandoned bauxite mine. The actual numbers of seedlings also responded to fruiting events in parent trees so that periodic increases may be expected, mostly of short duration due to heavy mortality of young seedlings (Fig. 9). The best stocking of new recruits in these data were on skid trails with more than twice the density after 3 y than in felling gaps or in unlogged areas. This difference may be due to the greater competition from preexisting plants in the unlogged areas and to greater competition from weeds in the felling gaps (Fredericksen and Mostacedo, 2000; Felton et al., 2006; Schnitzer and Carson, 2010). The result partly disproves our third hypothesis in that skid trail densities are higher than other parts of the forest after 7 y. In terms of species functional groups, the more successful (in numbers) after year 3 are the Non-Pioneers, with three times the density of the Pioneers (Fig. 7). Thus, after 3 y, the initial domi-

Whilst Pioneers become subsidiary in terms of numbers, they are often the tallest trees so that the surviving individuals were the leading trees in the recovery of forest structure and basal area. This matches their distribution in natural forest where they are often characterised by a bias towards large, isolated trees (e.g. Whitmore, 1974). Since Pioneers have intrinsically high growth rates in high irradiance, we may expect these trees to be the first to attain harvestable size. For an average growth rate of 5 m in 7 y, such trees would reach the upper canopy at 30 m in 42 y. For individuals with faster growth (22 of those noted above exceeded 8 m in 7 y) the same stature would be achieved in only 26 y. Note, however, that these represent mean annual height growth rates of less than 1.2 m y1, substantially less than can be achieved in plantations. 4.5. Seedling species composition The changes in tree species composition over the 7 y were summarised by ordination in Fig. 11. All three kinds of disturbance

36

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37

(main and secondary skid trails, and felling gaps) followed closely similar trends in species composition, initially dominated by Pioneers and later by Non-Pioneers as greater mortality and reduced recruitment amongst Pioneers allowed Non-Pioneers to become numerically dominant (left to right on Axis 1 in Fig. 11). This pattern of change is consistent with changes in species composition reported for secondary succession in tropical forests (e.g. Swaine and Hall, 1983) and recalls recommendations that management should mimic natural gap dynamics (Brown and Press, 1992; Sist and Brown, 2004). The unlogged samples, with few Pioneers, responded more strongly to parent tree fruiting events but it is evident that by 7 y the composition of recruits in logged and unlogged forest are convergent though differing significantly in median Axis 1 scores. 4.6. Conclusions We conclude that for this example of harvesting under the Selection System, with careful controls over logging practice, there is no detriment to timber tree regeneration. At least for the first seven years after logging, skid trails seem beneficial because of the superior densities of recruits, though felling gaps exceed skid trails in the number exceeding 5 m in height. This conclusion is in part dependent on the numbers of species which are currently commercial in Ghana. The commercial species list prior to 1989 included only 25 species of which only 12 were recorded in our data, accounting for 37% of the individuals of the timber species enumerated (Table 2). The earlier classification included representatives of all three functional groups with some bias towards slowgrowing, high value Non-Pioneer Shade-Bearers and if used in place of the current classification would not have resulted in significantly different conclusions. In forests where only a few species are marketable (e.g. in some areas within Central Africa where harvesting has been dominated by the extraction of African mahoganies of the genus Entandrophragma (Hall et al., 2003; ITTO, 2004) or where the successful Pioneers are not marketed, conclusions might be different. Federicksen and Putz (2003), for example, argue that greater disturbance may be needed for some forests where the commercial species, such as Swietenia macrophylla and Cedrela spp., require such openings for successful regeneration. Sist and Brown (2004) warned against more intense logging disturbance, arguing that few species demand such large openings and that other species might be impaired by them. The current trend in natural forest management in the tropics is to expand the number of species to be exploited, both Pioneers and Non-Pioneers, to allow a greater harvest. On the evidence of the present study, the initial success of the Pioneers would provide the first restocking of commercialsize individuals, whilst the Non-Pioneers would be not be adversely affected, proceeding more slowly to commercial size, much as they do in natural forest. Acknowledgements We thank Peter Amoako, Kwaku Asumadu, Vincent Berko and Dominic Bosompra for assistance with fieldwork and the Forestry Research Institute of Ghana for logistic support. The research was funded by the University of Aberdeen, the British Ecological Society and Tropenbos International (Tropenbos Ghana Project). References Adam, K.A., 2003. Tree selection in selective logging: ecological and silvicultural considerations for natural forest management in Ghana. Ph.D. Thesis, University of Aberdeen, UK.

Agyeman, V.K., Swaine, M.D., Thompson, J., 1999. Responses of tropical forest tree seedlings to irradiance and the derivation of a light response index. J. Ecol. 87, 815–827. Arets, E.J.M.M., 2005. Long-term responses of populations and communities of trees to selective logging in tropical rain forests in Guyana. Tropenbos-Guyana Series 13, Tropenbos-Guyana Programme, Georgetown. Brown, N., Press, M.C., 1992. Logging rainforests the natural way? New Sci. 133 (1812), 25–29. Dalling, J.W., Swaine, M.D., Garwood, N.C., 1994. Effect of soil depth on seedling emergence in tropical soil seed-bank investigations. Funct. Ecol. 9, 119–121. Dawkins, H.C., 1988. The first century of tropical silviculture: success forgotten and failures misunderstood. In: McDermott, M.J. (Ed.), The Future of the Tropical Rain Forest. Proceedings of an International Conference. Oxford Forestry Institute, Oxford, pp. 9–12. Dekker, M., de Graaf, N.R., 2003. Pioneer and climax tree regeneration following selective logging with silviculture in Suriname. Forest Ecol. Manage. 172, 183– 190. Dickinson, M.B., Wigham, D.F., Hermann, S.M., 2000. Tree regeneration in felling and natural treefall disturbance in a semideciduous tropical forest in Mexico. Forest Ecol. Manage. 134, 137–151. Duah-Gyamfi, A., 2007. Tree seedling demography after selective logging in a tropical moist semi-deciduous forest in Ghana. MPhil. Thesis, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana. Federicksen, T.S., Putz, F.E., 2003. Silvicultural intensification for tropical forest conservation. Biodivers. Conserv. 12, 1445–1453. Felton, A., Felton, A.M., Wood, J., Lindenmayer, D.B., 2006. Vegetation structure, phenology, and regeneration in the natural and anthropogenic tree fall-gaps of a reduced-impact logged subtropical Bolivian forest. Forest Ecol. Manage. 235, 186–193. Fredericksen, T.S., Mostacedo, B., 2000. Regeneration of timber species following selection logging in a Bolivian tropical dry forest. Forest Ecol. Manage. 131, 47– 55. Fredericksen, T.S., Pariona, W., 2002. Effect of skidder disturbance on commercial tree regeneration in logging gaps in a Bolivian tropical forest. Forest Ecol. Manage. 171, 223–230. Ghana Forest Service, 1998. A Logging Manual for Ghana. Guidance to Companies Operating Timber Utilisation Contracts in the High Forests of Ghana. Ghana Forestry Commission, Accra, p. 26. Gough, I.J., 2004. Nutrient and light competition between the invasive exotic, Chromolaena odorata, and tropical tree seedlings. B.Sc. Thesis, University of Aberdeen, Aberdeen. Gourlet-Fleury, S., Beina, D., Fayolle, A., Ouédraogo, D.-Y., Mortier, F., Bénédet, F., Closset-Kopp, D., Decocq, G., 2013a. Silvicultural disturbance has little impact on tree species diversity in a Central African moist forest. Forest Ecol. Manage. 304, 322–332. Gourlet-Fleury, S., Mortier, F., Fayolle, A., Baya, F., Ouédraogo, D., Bénédet, F., Picard, N., 2013b. Tropical forest recovery from logging: a 24 year silvicultural experiment from Central Africa. Philos. Trans. Roy. Soc. B 368, 1–10. Hall, J.B., Swaine, M.D., 1976. Classification and ecology of closed-canopy forest in Ghana. J. Ecol. 64, 913–952. Hall, J.B., Swaine, M.D., 1981. Distribution and Ecology of Vascular Plants in a Tropical Rain Forest. Forest Vegetation in Ghana. W. Junk, The Hague, p. 383. Hall, J.S., Harris, D.J., Medjibe, V., Ashton, P.M.S., 2003. The effects of selective logging on forest structure and tree species composition in a Central African forest: implications for management of conservation areas. Forest Ecol. Manage. 183, 249–264. Hawthorne, W.D., 1995. Ecological Profiles of Ghanaian Forest Trees. Tropical Forestry Papers 29. Oxford Forestry Institute, Oxford. Hawthorne, W.D., Agyeman, V.K., Abu-Juam, M., Foli, E.G., 1999. Taking stock. An annotated bibliography of logging damage and recovery in tropical forests, and the results of new research in Ghana. Unpublished report to DFID, London. Hill, M.O., 1974. Diversity and evenness: a unifying notation and its consequences. Ecology 54, 427–432. Honu, Y.A.K., Dang, Q.L., 2000. Responses of tree seedlings to the removal of Chromolaena odorata Linn. In a degraded forest in Ghana. Forest Ecol. Manage. 137, 75–82. ITTO, 2004. Annual Review and Assessment of the World Timber Situation. International Tropical Timber Organization, Yokohama, Japan. Johns, J.S., Barreto, P., Uhl, C., 1996. Logging damage during planned and unplanned logging operations in the eastern Amazon. Forest Ecol. Manage. 89, 59–77. Kuusipalo, J., Jafarsidik, Y., Ådjers, Göran, Tuomela, K., 1996. Population dynamics of tree seedlings in a mixed dipterocarp rainforest before and after logging and crown liberation. Forest Ecol. Manage. 81, 85–94. Kyereh, B., Swaine, M.D., Thompson, J., 1999. Effect of light on the germination of forest trees in Ghana. J. Ecol. 87, 772–783. Mac Donagh, P., Rivero, L., Garibaldi, J., Alvez, M., Cortez, P., Marek, M., Erbetta, D., Fernandez, R., Phar, N., Toma, T., 2010. Effects of selective harvesting on traffic pattern and soil compaction in a subtropical forest in Guarani, Misiones, Argentine. Sci. Forestali, Piracicaba 38 (85), 115–124. Magnusson, W.E., de Lima, O.P., Reis, F.Q., Higuchi, N., Ramos, J.F., 1999. Logging activity and tree regeneration in an Amazonian forest. Forest Ecol. Manage. 113, 67–74. Maitre, 1987. Natural forest management in Côte d’Ivoire. Unasylva 30, 53–60. McCune, B., Mefford, M.J., 1999. PC-ORD. Multivariate Analysis of Ecological Data. Version 5.01. MjM Software, Gleneden Beach, Oregon, USA.

A. Duah-Gyamfi et al. / Forest Ecology and Management 314 (2014) 26–37 Nicholson, S.E., Some, B., Kone, B., 2000. An analysis of recent rainfall conditions in West Africa, including the rainy seasons of the 1997 El Niño and the 1998 La Niña years. J. Clim. 13, 2628–2640. Okuda, T., Susuki, M., Adachi, N., Quah, E.S., Hussein, N.A., Manokaran, N., 2003. Effect of selective logging on canopy and stand structure in a lowland dipterocarp forest in peninsular Malaysia. Forest Ecol. Manage. 175, 297–320. Ouédraogo, D.-Y., Beina, D., Picard, N., Mortier, F., Baya, F., Gourlet-Fleury, S., 2011. Thinning after selective logging facilitates floristic composition recovery in a tropical rain forest of Central Africa. Forest Ecol. Manage. 262, 2176–2186. Park, A., Justiniano, M.J., Fredericksen, T.S., 2005. Natural regeneration and environmental relationships of tree species in logging gaps in a Bolivian tropical forest. Forest Ecol. Manage. 217, 147–157. Pena-Claros, M., Peters, E.M., Justiniano, M.J., Bongers, F., Blates, G.M., Fredericksen, T.S., Putz, F.E., 2008. Regeneration of commercial tree species following silvicultural treatments in a moist tropical forest. Forest Ecol. Manage. 255, 1283–1293. Pinard, M.A., Howlett, B., Davidson, D., 1996. Site conditions limit pioneer tree recruitment after logging of dipterocarp forests in Sabah, Malaysia. Biotropica 28, 2–12. Pinard, M.A., Barker, M.G., Tay, J., 2000. Soil disturbance and post-logging forest recovery on bulldozer paths in Sabah, Malaysia. Forest Ecol. Manage. 130, 213– 225. Putz, F.E., Sist, P., Fredericksen, T., Dykstra, D., 2008. Reduced impact logging: challenges and opportunities. Forest Ecol. Manage. 256, 1427–1433. Rice, R.E., Gullison, R.E., Reid, J.W., 1977. Can sustainable management save tropical forests? Sci. Am., 44–49. Schnitzer, S.A., Carson, W.P., 2010. Lianas suppress tree regeneration and diversity in treefall gaps. Ecol. Lett. 13, 849–857.

37

Schnitzer, S.A., Parren, M.P.E., Bongers, F., 2004. Recruitment of lianas into logging gaps and the effects of pre-harvest climber cutting in a lowland forest in Cameroon. Forest Ecol. Manage. 190, 87–98. Sheil, D., Burslem, D.F.R.P., Alder, D., 1995. The interpretation and misinterpretation of mortality rate measures. J. Ecol. 83, 331–333. Sist, P., Brown, N., 2004. Silvicultural intensification for tropical forest conservation: a response to Fredericksen and Putz. Biodivers. Conserv. 13, 2381–2385. Sist, P., Nguyen-The, N., 2002. Logging damage and the subsequent dynamics of a dipterocarp forest in East Kalimantan (1990–1996). Forest Ecol. Manage. 165, 85–103. Sist, P., Sheil, D., Kartawinata, K., Priyadi, H., 2003. Reduced-impact logging in Indonesian Borneo: some results confirming the need for new silvicultural prescriptions. Forest Ecol. Manage. 179, 415–427. Swaine, M.D., Agyeman, V.K., 2008. Early post-logging regeneration in two forest reserves in Ghana. Biotropica 40, 370–374. Swaine, M.D., Hall, J.B., 1983. Early succession on cleared forest land in Ghana. J. Ecol. 71, 601–627. Taylor, C.J., 1960. Synecology and Silviculture in Ghana. Thomas Nelson and Sons Ltd., Edinburgh, UK, p. 418. ter Steege, H., 1996. Hemiphot—a programme to analyze vegetation indices, light and light quality from hemispherical photographs. Tropenbos Documents 3, Wageningen, The Netherlands. Whitmore, T.C., 1974. Change With Time and the Role of Cyclones in Tropical Rain Forest on Kolombangara, Solomon Islands. Commonwealth Forestry Institute, Paper 46, Oxford. Wong, J.L.G., 1989. Ghana Forest Inventory Project Seminar Proceedings, 29–30 March 1989. ODA and Ghana Forestry Department, Accra, p. 101.