Effect of light on the germination of forest trees in Ghana

Authors


M.D. Swaine (fax 01224 272703; e-mail m.swaine@abdn.ac.uk).

Summary

1  Seed germination in light and dark, and responses to irradiance and light quality, were tested in shadehouse experiments for 19 West African tropical forest tree species representing a wide range of ecological types. Germination in forest gaps of different size was tested for 11 species.

2  Percentage germination was reduced in the dark only for three small-seeded species that are common in forest soil seed banks: Musanga cecropioides, Nauclea diderrichii and Milicia excelsa. Percentage germination of the other 16 species, including four widely regarded as ‘pioneers’, was unaffected.

3  Effects of different irradiances in shadehouses, where the seeds were watered, were significant for some species, but there was no consistent pattern. Irradiance effects in forest gaps, where the seeds received only natural wet season rainfall, were more widespread and substantial, and were most commonly shown as a depression of percentage germination at high irradiance.

4  Effects of light quality (neutral vs. green shade; red : far-red = 0.43) were insignificant at 5% irradiance in shadehouses for all species except Nauclea diderrichii. In growth chamber experiments, the low energy response was only evident at 1.0 µmol m–2 s–1 (< 1% of unshaded forest irradiance) in Musanga and Nauclea.

5  The speed of germination was affected by irradiance in many species, but the effect was small compared with differences between species, in which time to complete germination varied between 3 weeks and over 6 months.

6  Seeds of Ceiba pentandra and Pericopsis elata planted in deep forest shade (2% irradiance) and in a small gap (30% irradiance) germinated well in both sites, showed exponential biomass growth in the gap but a linear decline in mean seedling biomass and subsequent death in deep shade.

7  Light-mediated germination is relatively rare among these forest trees, even among pioneers, so that the working definition of a pioneer should be seen to depend more on a species’ ability to survive in forest shade. The effects of canopy opening on seed germination are small except in the largest openings, which severely depress germination in a number of species, including some species with strongly light-demanding seedlings.

Introduction

Environmental influences on the germination of tropical forest species have received much less attention than temperate species, although significant contributions have been made by Vázquez-Yánes and co-workers in Mexico (Vázquez-Yanes 1977, 1980; Vázquez-Yanes & Orozco-Segovia 1982a, b, 1984, 1990, 1993; Vázquez-Yanes et al. 1990). This may, in part, be due to the perception that most timber tree species from tropical rain forest have recalcitrant seeds (sensuRoberts 1973; Corbineau & Comb 1989), which have no dormancy, rapid germination and short viability. With typically high seed moisture content, recalcitrant seeds are difficult to store, but if collected fresh they are easy to germinate and establish in nurseries and germinate abundantly in natural forest. The contrasting seed type, known as orthodox because their seeds can be readily stored in artificial conditions, have low seed moisture and generally show various degrees of dormancy. Recalcitrant seeds are characteristic of non-pioneer (climax) species, while orthodox seeds are associated with pioneer species (sensuSwaine & Whitmore 1988), which include relatively few timber species (Whitmore 1983).

This dichotomy among tropical forest tree species is a useful generalization, but conceals considerable diversity of species’ germination response. Garwood (1983), for example, has shown considerable variation both between and within species in the period of seed dormancy following dispersal, and suggested that this often allows the seed to germinate at the most favourable season for seedling establishment. The speed of germination is also very variable: in 335 Malaysian tree species, Ng (1980) showed that the time taken to complete germination varied between 1 week and more than 1 year.

Swaine & Whitmore (1988) proposed that tropical forest tree species could be divided into two guilds (functional groups; sensuGitay & Noble 1997), the pioneers and non-pioneers, distinguished by the dependence of pioneers on canopy gaps for seed germination and seedling establishment. The implication was that pioneers had seed dormancy that was broken by gap conditions (increased irradiance and red : far-red ratio, elevated or fluctuating temperatures), associated with a requirement of their seedlings for high irradiance, so that the guilds were naturally distinct. The hypothesis was based on the circumstantial evidence that pioneer species were never found as young seedlings in forest shade, and only occurred as young plants in canopy gaps or other disturbed areas such as roadsides and farms. Various studies have shown that some pioneer species require light for germination (photoblasticity) (Valio & Joly 1979; Vázquez-Yanes & Smith 1982; Orozco-Segovia & Vázquez-Yanes 1989; Vázquez-Yanes et al. 1990), which lends support for pioneer gap-dependence determined by seed germination.

Hawthorne (1993, 1995) provided a subdivision of the more speciose non-pioneer guild on the basis of seedling ability to survive and grow in deep forest shade. In such conditions, seedlings of non-pioneer light demanders (NPLD) have higher mortality, most dying before exceeding a metre or so in height, while non-pioneer shade bearers (NPSH) have lower mortality and continue to grow in deep shade, eventually becoming established trees.

The pioneer/non-pioneer dichotomy has the advantage that the groups are readily distinguishable from field experience in any tropical forest region, but the defining characteristics have not been widely tested by experiment. Raich & Gong (1990) planted seeds of 43 Malaysian tree species in forest shade, as well as in small and large canopy gaps, and recorded percentage germination. They recognized three species groups: those that showed higher germination in gaps, those with higher germination in forest shade and those that germinated well both in shade and in gaps. There were, however, no clear discontinuities between the three categories, and in some cases there were marked differences in response to canopy cover between different seed collections of the same species.

However, there are several reports of pioneers germinating in deep forest shade (e.g. Endospermum peltatum in Sabah; Kennedy & Swaine 1992). Ceiba pentandra was observed by M. Dike and D. U. U. Okali (personal communication) to germinate in forest shade in Nigeria; and Cecropia spp. were the most abundant germinants in the understorey of a Costa Rican rain forest (Li et al. 1996). Such observations cast doubt on the gap-dependence of pioneers for germination, and thus on the naturalness of the dichotomy between them and non-pioneers. If this widely recognized and frequently applied dichotomy is to be preserved, it becomes more dependent on the second of the Swaine & Whitmore (1988) criteria for pioneers, the seedling's requirement for high irradiance. Thus, although pioneers may germinate in deep shade, they will usually be overlooked and very small seedlings die swiftly in low irradiance (Kennedy & Swaine 1992).

The environmental conditions that break seed dormancy in tropical forest tree seeds may be a high red : far-red ratio (Vázquez-Yánes 1977, 1980; Orozco-Segovia & Vázquez-Yánes 1989; Valio & Joly 1979), high maximum temperature or temperature range (Vázquez-Yánes & Orozco-Segovia 1982a), fluctuating soil moisture (e.g. Terminalia sp.; Taylor 1960) or some combination of these. All of these changes in microclimate can occur following canopy opening, and because they are substantial may also reduce seed germination in non-dormant (non-pioneer) species by causing desiccation or heating (insolation). Such environmental effects have implications for forest management: exploitation of tropical forest for timber commonly creates large gaps that appear to favour pioneer species over non-pioneers, because of the differential effect on seed germination and on seedling growth and survival.

The present study set out to test if West African pioneer trees have gap-dependent germination as proposed by Swaine & Whitmore (1988), or if their failure to establish is due to negative carbon balance in young seedlings. Experimental testing is necessary because existing allocation to the pioneer guild has been based on the subjectively perceived distribution of young plants in forest habitats. We used an experimental approach in shadehouses, growth chambers and forest gaps to determine germination responses to light in West African tree species, chosen to include both pioneers and a range of non-pioneers as previously categorized by Hall & Swaine (1981) and Hawthorne (1995). Our strategy was to find out first if the species were photoblastic, in order to determine if further experiments were needed to describe the nature of the light effect, and then to examine the effect of irradiance on germination because of its influence on both seed temperature and moisture content. The applicability of these shadehouse results for forest conditions were then tested by germination trials in artificial forest gaps of different size. A final experiment compared biomass accumulation of young seedlings in forest shade and a canopy gap to test the shade tolerance of seedlings of pioneers.

Species nomenclature follows Hawthorne (1995). Except where ambiguity may arise, species are referred to by genus only after the first mention.

Methods

Species selection

A wide range of tree species (Table 1) was chosen for experimental work, representing considerable variation in seed mass (0.3–2940 mg fresh mass) in fresh seed moisture content (4–54% fresh mass) and ecological guild. The species included the fast-growing short-lived pioneer Musanga cecropioides, which has a large and persistent soil seed bank (Hall & Swaine 1980) of very small seeds and is an abundant colonizer of abandoned farmland in wet forest. Terminalia ivorensis and Terminalia superba are also fast-growing pioneers but achieve larger size and their larger seeds (in winged fruits) are uncommon in soil seed banks. Like Musanga, their seedlings are intolerant of shade. Entandrophragma utile has winged seeds, is absent from soil seed banks (Hall & Swaine 1980), and is a non-pioneer light-demander (sensuHawthorne 1995) somewhat tolerant of shade but growing most rapidly in moderate irradiance. Guarea cedrata is an example of the most shade tolerant of West Africa's timber species (a non-pioneer shade-bearer; Hawthorne 1995), with a large seed of short viability. Its seedlings are commonly abundant in deep shade, with slow growth, even in non-limiting irradiance (Swaine et al. 1997). All species used in the study are important timber trees, with the exception of Musanga which is of small stature and has low-density wood.

Table 1.  Mean mass and mean water content of freshly collected seeds, dispersal morphology and ecological guild of tree species used in germination experiments arranged in order of fresh seed mass. Ecological guilds after Hawthorne (1995): P = pioneer; NPLD = non-pioneer light-demander; NPSH = non-pioneer shade-bearer
SpeciesCodeFamilyEcological guildDispersal morphologyFresh seed mass (mg)Water content (% fresh mass)
Blighia sapidaBsSapind.NPLDArillate, black seed294046
Ricinodendron heudelotiiRhEuphorbiPFleshy fruit159010
Pterygota macrocarpaPmSterculi.NPLDWinged seed146616
Guarea cedrataGcMeli.NPSHFleshy capsule102127
Entandrophragma utileEuMeli.NPLDWinged seed5957
Sterculia rhinopetalaSrStreculi.NPLDArillate56330
Khaya anthothecaKaMeli.NPLDWinged seed2494
Pericopsis elataPePapilio.NPLDWinged fruit22014
Mansonia altissimaMaSterculi.NPLDWinged fruit19112
Lovoa trichilioidesLtMeli.NPLDWinged seed18654
Khaya ivorensisKiMeli.NPLDWinged seed1659
Terminalia ivorensisTiCombretPWinged fruit11710
Terminalia superbaTsCombretPWinged fruit1129
Albizia ferrugineaAfMimos.NPLDWinged fruit1058
Ceiba pentandraCpBombacPSeed in kapok6017
Celtis mildbraediiCmUlm.NPSHFleshy fruit4914
Funtumia elasticaFeApocyn.NPLDPlumed seed347
Milicia excelsaMeMor.PFleshy fruit37
Musanga cecropioidesMcMor.PFleshy fruit0.89
Nauclea diderrichiiNdRubi.PFleshy fruit0.37

Shadehouse experiments

A series of experiments was conducted in seed trays placed in screened shadehouses at the Forestry Research Institute of Ghana, Kumasi. The shadehouses were constructed for experiments on the growth rate of tree seedlings (Agyeman 1994; Swaine et al. 1997) under different degrees of neutral shading provided by bamboo slats (to provide forest-like sunflecks) and hessian shade cloth, but were also used in the present experiment to provide a range of irradiances and for protection from seed predators and herbivores.

Kumasi (6°43′N 1°38′W) lies in the forest zone of Ghana. It has a bimodal rainfall pattern consisting of a major rainy season (March–July) and a minor one (September–November). The mean annual rainfall is about 1500 mm, and the mean annual temperature is 25.6 °C with an average daily range of 8–9 °C. Relative humidity at mid-day is lowest in January, about 47%, and above 85% in the rainy season.

Experiments to test for the effect of light on germination used the 42% irradiance (= 42% of unshaded irradiance) shadehouse and shallow wooden seed trays filled with forest topsoil and closed with a wooden-framed lid that permitted air circulation. The lids were covered by two layers of black polythene (dark treatment, 0% irradiance), one or more layers of clear polythene (neutral shade, to give 30% and 5% of unshaded irradiance) or by a green filter [Strand Lighting, Middlesex, UK, pea-green filter no. 421, reduced red : far-red ratio (0.43), 5% unshaded irradiance]. Irradiances within the treatment boxes were measured with Didcot Instruments, Abingdon, Oxford, UK, integrating PAR sensors (DRP2) and expressed as a percentage of the irradiance outside the shadehouse.

Tests of the effects of irradiance on a subset of species used 2%, 30% and 65% irradiance shadehouses with open plastic bowls filled with forest topsoil. Shadehouse irradiances were measured in the same way as the enclosed germination boxes.

Freshly collected fruits and seeds were used in all experiments. These were mostly collected opportunistically, so that we did not know how many parent trees were included in each species collection. It is likely, however, that some collections came from a single parent. Some collections were made into black plastic bags (e.g. Ceiba, Blighia), but none was exposed to high irradiance before the experiments. As many seeds are exposed on the forest floor after dispersal it is unrealistic to store them in total darkness. Wings, kapok, pappuses and flesh were removed from seeds before they were pressed half-way into the soil. Each irradiance treatment was shared among four bowls (to allow randomization on the bench), each with 25 seeds, except for Celtis mildbraedii, for which 20 seeds were used. The small-seeded species, Musanga, Nauclea diderrichii and Milicia excelsa (Table 1), received the same treatments, but were planted in Petri dishes with two layers of Whatman no. 1 filter paper. The arrangement for the small-seeded species was the same, but Musanga had 40 seeds per dish (n = 160) and Nauclea had only three dishes (n = 75).

Germination media were kept moist by regular watering, and germination was recorded at 3-day intervals. The dark and low red : far-red treatments were examined only at night with the aid of a dim, green-filtered torchlight. A seed was recorded as germinated once the radicle had emerged. Monitoring of germination ceased when no further seeds had germinated for at least 1 week. At this stage, all ungerminated seeds appeared decayed, with the exception of Ricinodendron. This species has extended germination and the monitoring period was therefore extended.

Growth chamber experiments

For two of the species that showed a photoblastic response (Musanga and Nauclea) it was thought necessary to conduct further experiments in controlled environment growth chambers at the University of Aberdeen, UK, because of the difficulty of controlling temperature in shadehouse conditions. The Sanyo-Gallenkamp growth chambers were operated at a constant 27 °C. Light conditions were varied by the use of black or clear polythene, a no. 421 pea-green filter and another filter (Garcia & Smith 1993) which provided a lower red : far-red ratio light. Combinations of these provided irradiances of 0, 0.1, 1, 5 and 30 µmol m–2 s–1, and red : far-red ratios of 0.83 and 0.43–0.47. Irradiances were measured by integration over 1 week using a factory-calibrated Skye Instruments, Llandrindod Wells, Powys, UK, DataHog. Thirty seeds were sown on two layers of Whatman no. 1 filter paper within Petri dishes in each of four replicates

Experiments in forest

The relevance of the results from the preceding experiments was tested for forest conditions by germinating seeds in a range of artificial gaps in Tinte Bepo Forest Reserve, Ghana, which has a somewhat drier climate than Kumasi with a mean annual rainfall of about 1300 mm. Gaps were created by felling small trees and clearing undergrowth beneath pre-existing openings in the upper canopy. This method avoided the felling of large trees. Irradiance in gaps was measured by Didcot Instruments integrating PAR sensors (DRP2). Several sensors were rotated among the gaps and there was a fixed sensor in an open area outside the Forest Reserve. Gap size was adjusted to achieve approximately 15%, 30% and 50% of unshaded irradiance. Four replicate gaps of each irradiance and four areas of closed canopy forest (mean irradiance 2%) were used for the experiment, as well as the area outside the forest (100% irradiance).

Eleven species were tested in two batches, in the rainy seasons of 1992 and 1993. In each replicate, 25 seeds of each species were planted in forest topsoil in bowls protected with 1-cm mesh chicken wire. Freshly collected seeds were pressed half-way into the soil and subsequent germination monitored at weekly intervals between mid-May and early October in 1992 and April–June in 1993. Water was provided only by natural wet season rainfall.

Seed germination and seedling biomass accumulation in forest shade (2% irradiance) and in a gap (30% irradiance) was assessed in two species, the pioneer Ceiba pentandra and the non-pioneer light demander Pericopsis elata, by planting seeds in bowls as in the previous experiment and taking weekly seedling harvests for 7 weeks after germination. Harvested seedlings were dried to constant weight and dry mass of the whole seedling recorded.

Statistical analysis

Testing for differences in percentage germination employed chi-squared tests on raw counts with results pooled for bowls/replicates within a treatment. Mean days to germinate were compared with t-tests (the 3-day period over which each seed germinated was recorded).

Results

Seed characteristics

Relationships among seed characters in Table 1 were not strong. The smallest-seeded species had seeds with low water content and the bigger seeds included some of the higher water contents, but the correlation was not significant. Wind-dispersed species were generally of moderate mass, and pioneers were more common among small-seeded species, with the notable exception of Ricinodendron heudelotii, which had the second most massive seed at 1.6 g. Larger seeds are known in West African rain forest tree species [e.g. Okoubaka aubrevillei (c. 100 g fresh weight), Balanites wilsoniana (c. 50 g), Mammea africana (38 g) and Tieghemella heckelii (19 g)] but few are smaller than those of Nauclea (0.0003 g).

Germination in light and dark, and in neutral and green shade

Of 17 species tested in the shadehouse experiments, only the three with the smallest seeds showed significant differences in percentage germination between light and dark: germination was very low or zero in darkness but appreciable in 5% irradiance (Table 2 and Fig. 1). Musanga, Nauclea and Milicia are common in forest soil seed banks (Bouharmont 1954; Hall & Swaine 1980) and their seeds develop in large numbers in fleshy, scented fruits, and are probably dispersed by bats. Several of the species that are widely regarded as pioneers (Hawthorne 1995), notably the two forest species of Terminalia, Ceiba pentandra and Ricinodendron heudelotii, were not responsive to light, which seems to confirm that photoblasticity is characteristic only of a subset of the pioneer guild.

Table 2.  Effects of light (30% irradiance) vs. dark and neutral (R:FR = 0.83) vs. green (R:FR = 0.43) shade on the germination of tree species in shadehouses. Differences between pairs of treatments are significant (χ2) at: **P < 0.01; ***P < 0.001; or not significant (NS)
Mean percentage germinationMean days to germinate
SpeciesIrradiance (%)Light
30
 Dark
0
Neutral shade
5
 Green shade
5
Light
30
 Dark
0
  1. + 5% irradiance.

  2. – Not recorded or not tested.

Blighia sapida100NS9893NS988 * 7 
Ricinodendron heudelotii33NS4232NS2842 * * 35 
Pterygota macrocarpa71NS7365NS7056NS51 
Guarea cedrata71NS7074NS71  
Entandrophragma utile77NS8377NS7527 * * * 23 
Sterculia rhinopetala52NS57 10NS10 
Khaya anthotheca51NS47 17NS17 
Pericopsis elata69NS5969NS6616 * * * 9 
Mansonia altissima80NS8176NS7921 * * * 16 
Lovoa trichilioides89NS9394NS92  
Khaya ivorensis68NS7364NS7426 * * * 22 
Terminalia ivorensis43NS3734NS3366NS64 
Terminalia superba77NS7769NS7225 * * 23 
Ceiba pentandra56NS5353NS5225 * * * 19 
Celtis mildbraedii71NS7067NS6521 * * 19 
Funtumia elastica63NS6666NS6028NS28 
Milicia excelsa33 + * * * 333NS30 20 
Musanga cecropioides78 + * * * 078NS77  
Nauclea diderrichii37 + * * 737 * * * 15 24 
Figure 1.

Germination (%) of freshly collected seeds of 19 Ghanaian rain forest tree species in light [30% irradiance, except Milicia, Musanga and Nauclea (5%)] and dark treatments of the shadehouse experiment. The diagonal line shows equality of germination in light and dark. Species classified a priori as pioneers (Hawthorne 1995) are shown as open circles and named in bold face. Full species names as in Table 1.

In the shadehouse experiment, there was no effect of low red : far-red ratio (R:FR) at low irradiance (5%) for any species, including those shown to be photoblastic above, with the exception of Nauclea diderrichii in which germination was reduced by more than 50% (Table 2). The R:FR quotient used was 0.43, a little higher than would be typical of closed forest understorey. Thus it is possible that a lower value might have elicited more differences among the species (Vázquez-Yánes & Smith 1982).

In the growth chamber tests on two of the photoblastic species, the interaction between irradiance and R:FR ratio was examined at constant temperature. In both Nauclea and Musanga, germination increased with irradiance (Table 3). Low R:FR light had no effect on germination at 5 or 30 µmol m–2 s–1 irradiance, but significantly depressed germination in both species at 1 µmol m–2 s–1 irradiance. Thus the effects of the low energy reaction were only evident when irradiance was low and the high irradiance response, which may inhibit germination (Salisbury & Ross 1978), was muted. The implication is that in forest shade, where irradiance is low, the seeds of these pioneers are more likely to remain dormant, but will germinate in even small canopy gaps with only 5 µmol m–2 s–1. In Musanga, germination at 0.1 µmol m–2 s–1 only occurred at a high R:FR ratio, but even then was very low (three seeds of 120).

Table 3.  Effect of irradiance and R:FR ratio on percentage germination (n = 120 per treatment) in two photoblastic tree species at constant temperature (27 °C). –, not tested. Significance (χ2) between rows and between columns are: **P < 0.01; ***P < 0.001; not significant (NS)
Irradiance (µmol m–2 s–1)
 R:FR30510.1
Musanga cecropioides0.8387 NS93 ***71 ***2
 0.4389 NS87 ***35 ***0
P NSNS * * *  
Nauclea diderrichii0.8396 **85
 0.4396 ***68
P  NS * *  

Effects of irradiance in shadehouse experiments

Five species representing a range of guilds were tested for the effect of irradiance on percentage germination by planting seeds in shadehouses with 2%, 30% and 65% irradiance. Two of the non-pioneers (Guarea cedrata and Pterygota macrocarpa) were unaffected by differences in irradiance, but the other three species showed contrasting responses to increased irradiance (Table 4). Germination was enhanced by high irradiance in the non-pioneer Pericopsis elata and in Ricinodendron, and depressed in Terminalia ivorensis. In Ricinodendron, the species with extended germination, ungerminated seeds were tested for viability using tetrazolium chloride. A high proportion (91%) of the seeds that failed to germinate in 2% irradiance remained viable, suggesting that the ungerminated seeds had some requirement for radiant energy for germination, although no such effect had been suggested by the results of the previous light/dark experiment.

Table 4.  Percentage germination in three shadehouses with differing irradiance. Pioneer species in bold. –, not tested
Irradiance (%)
Species23065Probability (χ2)
Guarea cedrata566761NS
Pterygota macrocarpa79 –   84NS
Pericopsis elata69 –   1000.001
Terminalia ivorensis4242190.001
Ricinodendron heudelotii011340.001
Viability of ungerminated Ricinodendron seeds916730 

Germination in forest conditions

The small-seeded photoblastic species were not tested in the forest experiments, but three pioneers according to Hawthorne (1995) were included (Table 5). Germination in the forest understorey (mean irradiance, 2%) was less in these three pioneers than in any of the non-pioneers. Terminalia ivorensis showed only 12% germination in deep shade, about a third of its maximum (in 30% irradiance), while for the other two pioneers germination in shade was near the maximum value for the species.

Table 5.  Percentage germination and rate of germination (mean day of germination) in forest understorey (2% irradiance), canopy gaps of different size (15%, 30%, 50% irradiance) and outside the forest (100%). Species are ordered by percentage germination in 2% irradiance; pioneers (Hawthorne 1995) in bold. Significance of differences within species are: *P < 0.05; **P < 0.01; ***P < 0.001; not significant (NS)
Percentage germination
Irradiance (%)
Mean days to germinate
Irradiance (%)
Species2153050100 χ 22153050100Mean
Terminalia ivorensis122337331922.3 ***795541373361.3
Ceiba pentandra574055604213.4 **191915161320.5
Terminalia superba68667563559.4 *252020305838.3
Khaya anthotheca7366685710106.7 ***242730347046.3
Pterygota macrocarpa758087875734.8 ***90851008681110.5
Entandrophragma utile7785888818176.5 ***292930336646.8
Celtis mildbraedii797583794450.2 ***232322213430.8
Albizia ferruginea80728585748.9 NS232726262732.3
Pericopsis elata834375807249.3 ***161217152020.0
Blighia sapida94969798983.4 NS181718161320.5
Mansonia altissima9596928217251.4 ***191918244832.0

All but two species showed significant differences in percentage germination among the forest sites. In most, these differences were substantial and largely due to depression of germination in 100% irradiance relative to that in 2% (Fig. 2). Terminalia ivorensis was notable for significantly greater germination in intermediate irradiances, while Blighia sapida, which has a large, black seeds with high water content, was unaffected by gap size. These results did not agree in all cases between shadehouse (Table 4) and forest (Table 5) experiments. Terminalia ivorensis showed significant depression of germination in high irradiance in both environments, but the reduced germination observed in forest shade had not been recorded in the 2% shadehouse. In the forest, Pericopsis did not show the increased germination at high irradiance that was recorded in the shadehouse, and Pterygota showed reduced germination outside the forest but not in the 65% shadehouse. Some of these discrepancies may be attributed to desiccation in the forest, where seeds were not watered, compared with the shadehouse experiment, in which the seeds were kept moist (Orozco-Segovia & Vázquez-Yánes 1990).

Figure 2.

Germination in forest environments of differing irradiance of 11 tree species (codes as in Table 1) expressed as a proportion of germination in deep forest shade (2% irradiance). Species are ordered by germination in 100% irradiance. Significance of difference among irradiances (χ2) are: NS, not significant; *P < 0.05; **P < 0.001; ***P < 0.001.

Speed of germination

The distribution of germination day was unimodal for all species (except for Ricinodendron in light), so that the mean day of germination could be used as a measure of rate. The mean day was generally less in darkness than in the light in the shadehouse experiments (Table 2), but the effect was generally small compared with differences among species. In the forest experiments, some species were considerably slower to germinate in 100% irradiance than in shade, but the effect was not consistent among species (Table 5). These effects were probably due in part to slower imbibition at higher irradiance, especially in unwatered treatments, interacting with differences in species’ abilities to remain viable.

The mean day of germination and the number of days to complete germination were highly correlated in both shadehouse experiments (r = 0.973, d.f. = 15, P < 0.001) and in the forest (r = 0.976, d.f. = 9, P < 0.001), suggesting that either measure is a robust indication of a species’ rate of germination.

In addition, the mean day of germination (and days to complete germination) was correlated between shadehouse and forest experiments (r = 0.817 and 0.788, respectively; d.f. = 8, P < 0.01 for each), but rates in the forest were consistently slower than those in the shadehouse. On average, seeds planted in the forest took about 2 weeks longer to complete germination than in the shadehouse, presumably due to slower imbibition. These relationships were used to estimate forest germination rates for those species that were tested only in the shadehouse. When species were ordered by the mean number of days (observed or predicted) to complete germination in the forest, as in Fig. 3, the ranking was very similar to that for mean day of germination. There was considerable variation among species, germination being complete in Blighia in about 3 weeks but taking nearly 6 months in Pterygota. The rate of germination showed no significant relationship with seed size, moisture content or photoblasticity (Tables 1 and 2).

Figure 3.

Rate of germination (mean day of germination, shaded columns; days to complete germination, open columns) in forest conditions for 18 tree species (codes in Table 1). The values are the means of measurements in several treatments, and those marked ‘e’ are estimated for the forest from shadehouse results. Species are ordered by days to complete germination.

Ricinodendron was the only species to show bimodal distribution of germination, and then only in lighted treatments. In 30% neutral irradiance in the shadehouse, the first phase of germination started 27 days after planting and ended 21 days later. Germination resumed after an interval of 28 days and ceased 90 days after planting. A similar pattern was recorded in the low R:FR treatment, but not in darkness where germination was continuous and completed after 41 days, achieving a higher final germination. Ungerminated seeds in the dark treatment showed high viability with the tetrazolium chloride test, and germinated later in light (Table 4).

Biomass accumulation of young seedlings in forest conditions

To test the hypothesis that some pioneers fail to establish in forest shade because of a negative carbon balance, rather than an inability to germinate in shade, two species were planted in forest shade (2% irradiance) and in a small canopy gap (30% irradiance). Ceiba pentandra is a well-known pioneer, and although Hawthorne (1995) calls Pericopsis elata a non-pioneer light-demander, it was regarded by Hall & Swaine (1981) as a pioneer.

In the small gap, both species showed positive biomass accumulation, following exponential trends (Fig. 4), with the pioneer, Ceiba, growing fastest. In deep shade, however, both Ceiba and Pericopsis lost biomass, following linear trends. After 7 weeks, all seedlings of Ceiba in shade had collapsed due to failure of the stem above the hypocotyl. At this time seedling dry mass had fallen to 60% of the value 1 week after germination.

Figure 4.

Changes in seedling biomass over 7 weeks after germination in forest shade (2% irradiance, filled symbols) and in a small canopy gap (30%, open symbols) for the pioneer Ceiba pentandra (circles) and the non-pioneer light-demander, Pericopsis elata (squares). Fitted lines are exponential (in gaps) and linear (in shade).

Discussion

Substantial differences exist among the seed characteristics of the West African tree species tested in these experiments. Seed fresh mass varies by an order of four, seed water content by a factor of 12 and days to complete germination in forest conditions by a factor of six. None of these are correlated among themselves, nor with percentage germination in the forest.

Percentage germination varies among species, but it is difficult to propose species-specific rates because of the unknown quality of the seed collections and because variation in incident radiation has a marked influence on percentage germination for several species. The relationships with irradiance are thought generally to be due to indirect effects on seed temperature, or more probably seed moisture content, rather than a light effect because only three species showed a photoblastic response in the light/dark experiment, and only one showed an effect of low R:FR ratio at low irradiance in the shadehouse (Table 2).

The growth chamber experiment with Musanga and Nauclea (Table 3) showed that an effect of a low R:FR ratio was only detectable at irradiances less than 5 µmol m–2 s–1 (which is < 0.25% of a conservative value for full irradiance of 2000 µmol m–2 s–1). As the shadehouse light quality experiment was conducted at 5% irradiance we should not be surprised to see no effect in Musanga or Milicia. For Musanga at least, the low energy reaction appears to be irrelevant at irradiances < 1 µmol m–2 s–1, suggesting that germination in the forest in these photoblastic species will be depressed only in the darkest corners, or if the seed is buried in the soil.

The very low threshold for light-mediated influences on germination leads us to expect that germination in the forest will be at or near maximum in the deep shade treatments (2% irradiance), as is generally the case in Table 5. For the photoblastic species, none of which was tested in forest conditions, we predict that dispersed seed will germinate in deep forest shade providing it is not buried in the soil or covered by litter. The effects of increasing canopy opening on seed germination are somewhat varied, but the most common response is a depression of germination in open conditions (Fig. 2). This effect does not appear to be due to desiccation as it is not stronger in seeds with high water content, which can be killed by relatively small losses of moisture (Baskin & Baskin 1998). Blighia, the largest seeded-species, has the second highest water content and a black seed coat and showed no response to irradiance, maintaining high germination (> 94%) in all sites. Two factors may assist Blighia: it has the fastest rate of germination, reducing the opportunity for desiccation, and is unusual among forest species in dispersing its seeds in the relatively cloudy and cool wet season. Pioneers were no more resistant to the effect of high irradiance than non-pioneers: Ceiba and Terminalia superba both showed depression of germination at high irradiance, as did Terminalia ivorensis, although germination in this species was even more reduced by the lowest irradiance (Table 5).

The pioneer Euphorb Ricinodendron heudelotii stands out among the species tested for its rather inconsistent responses. It was not photoblastic in the light/dark experiment, was unaffected by reduced R:FR light, but responded strongly to differences in irradiance (Table 4). We may ask how germination can be so high in darkness (42%) if it is zero in the 2% shadehouse? The main difference between the treatments is that the irradiance tests were done in open bowls rather than enclosed seed trays, so that a temperature effect is a strong possibility. Temperatures within the enclosed seed trays were 4 °C higher than in the 2% shadehouse. Temperature commonly interacts with other environmental variables to affect germination (Mayer & Poljakoff-Mayber 1975; Baskin & Baskin 1998). For example, Chamshama & Downs (1982; in Baskin & Baskin 1998) described the interaction between light and temperature for the photoblastic Chlorophora excelsa (= Milicia excelsa;Table 2). They reported that at a constant temperature of 30 °C, light reduced germination to 3% compared with 37% in the dark, but at alternating temperatures of 30/20 °C, germination in light and dark was similar (45–50%). The interaction of temperature and light for our West African tree species needs further study, particularly because temperature was not fully controlled in the shadehouse experiments.

All seeds used in these experiments were freshly collected and generally showed good viability. Vázquez-Yánes & Orozco-Segovia (1990) report that most tropical tree species germinate freely when freshly collected, including most pioneers, but dormancy may be subsequently imposed or induced (or viability lost) if germination is prevented by a lack of water, by storage or by dispersal to forest shade with a low R:FR light. We have no direct information on these processes for our species but our differences in rates of germination are relevant for successful establishment of seedlings in forest. Most species seeds are dispersed in the dry season, and must wait for variable periods before rainfall is sufficient to permit imbibition (Garwood 1983). It is evident from the duration of germination in Fig. 4 that most species have the capacity to remain viable for at least 1 month, and, in 10 species, for 2 or more months.

The results presented here argue against the idea that the pioneer guild, as presently recognized, may be defined solely by gap-dependent germination. The two species tested showed a negative carbon balance in forest shade, suggesting that this limitation may be the more ubiquitous cause of the failure of pioneers to establish in forest shade. Only three species showed a photoblastic response, and growth chamber tests indicated that this is only affected at irradiances less than those normally found in forest shade. Daily temperature range and maximum also change markedly with canopy opening, as well as the availability of water at the soil surface, and the interaction of these factors with irradiance needs to be investigated. Evidently photoblasticity is a characteristic of a subset of pioneers, and is associated here with small-seeded, fleshy-fruited species that are found in the soil seed bank. This is a useful subdivision of the guild, but it would seem impractical to redefine pioneers on this basis because few species have been tested experimentally and because the existing definition is useful and widely applicable. It seems probable that the photoblastic group will be relatively small, and include particularly the small stature genera such as Musanga, some Ficus, Cecropia, Trema, and Harungana.

Preservation of the current suite of pioneer species thus depends in most cases on their inability to grow in forest shade, as has been confirmed here for two species, Pericopsis and Ceiba. Experiments have shown that other West African species also have negative growth rates in low irradiance. Agyeman et al. (1999); Swaine et al. (1997) showed that relative growth rate in established seedlings was negative in a 2% irradiance shadehouse in Mansonia altissima, Milicia excelsa, Ricinodendron heudelotii, Ceiba pentandra and Sterculia rhinopetala. This was confirmed in forest shade (2% irradiance) for Ceiba and Mansonia (Swaine et al. 1997). To place the definition of pioneers on a more objective basis by such means is, however, difficult because it requires the selection of an arbitrary threshold of irradiance, with attendant problems of measurement, and the use of consistent experimental conditions across different continents.

Acknowledgements

We thank the Director of the Forestry Research Institute of Ghana, Dr A. Ofusu Asiedu, for his encouragement and financial support of this research. The University of Science and Technology Institute of Renewable Natural Resources, directed by Mr J.G.K. Owusu, also provided support. The research was also funded by the Overseas Development Authority (now the Department for International Development) through its bilateral aid programme. We are grateful to Mrs A. Gyimah for her advice and the support of her Seed Technology group. Dr E.M. Veenendaal was always interested and helpful.

Received 2 July 1998revision accepted 18 March 1999

Ancillary