Forestry Advance Access originally published online on June 19, 2008
Forestry 2008 81(4):525-541; doi:10.1093/forestry/cpn026
| ||||||||||||||||||||||||||||||||||||||||||||||||||
Seed production and seedling survival in a 50-year-old stand of Corsican pine (Pinus nigra subsp. laricio) in southern Britain
Forestry Commission Research Agency, Alice Holt Lodge, Wrecclesham, Farnham, Surrey, GU10 4LH, UK
* Corresponding author. E-mail: gary.kerr{at}forestry.gsi.gov.uk
 |
Summary |
|---|
 Top Summary IntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
There is much literature on natural regeneration which emphasizes the importance of good seed year, but few authors consider seed input in terms of the combination of seed quantity (i.e. number of seeds) plus seed quality (i.e. percentage of viable seeds). We have considered both aspects and also attempted to identify the proportions of good vs poor quality seeds contributing to natural regeneration via ‘seed rain’ vs ‘cone drop’. In addition to studying seed input, we looked at the effects of vegetation control, ground preparation and protection from small mammals on seedling emergence and survival. Over a 3-year period (February 2001 to March 2004), there was enough seed production and seedling survival to conclude that natural regeneration could be successful beneath a 50-year-old stand of Corsican pine in the south of England. Peaks of pine seed release occurred in March/April in 2002 and 2003, but it was extremely surprising to observe that some seed was trapped in virtually every month of the 3-year study, demonstrating an almost continual release of (at least) small quantities of seeds. In line with this finding, although most pine seedlings were found shortly after peak seed dispersal in May, June and July; new seedlings were found in every month throughout the study except February and October. In general, vegetation control and ground preparation had a positive effect on seedling survival; the probability of a seedling surviving for 300 days was between 50 and 60 per cent.
|
Introduction |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
From the beginning of the 1990s a number of factors, such as the Rio-Helsinki process, the requirements of certification and an international movement favouring more natural forest management, have caused increased interest in the use of natural regeneration in British forests (Mason et al., 1999
; UKWAS, 2000; National Assembly for Wales, 2001; Mason and Kerr, 2004). This is a major challenge for forestry in Britain because there is generally little experience of using natural regeneration (Harmer et al., 1997; Malcolm et al., 2001). In order to meet the information needs of policy and practice, natural regeneration is now a focus of research within both native woodlands (Harmer et al., 2005) and plantations (Mason, 2003; Mason et al., 2004).
One example of the challenges of the wider use of natural regeneration is the widely held view that Corsican pine (Pinus nigra subsp. Laricio (Poir.) Maire) (Farjon, 1998), one of the two main lowland pine species, does not regenerate in Britain (Kerr, 2000). This view prompted us to examine in more detail some of the factors affecting the natural regeneration of Corsican pine, to supplement the earlier literature review by Kerr (2000).
Much literature on pine natural regeneration emphasizes the importance of good seed years (Shelton and Wittwer, 1996; Karlson and Orlander, 2000; Summers and Proctor, 2005), but only a few authors consider seed input in terms of the combination of seed quantity (i.e. number of seeds) plus seed quality (i.e. % germination of seeds). It is also commonly assumed in studies of natural regeneration that seed is dispersed via ‘seed rain’, when seed is released from cones attached to the tree. However, our observations indicated that another important route for seed input was via ‘cone drop’, when cones become unattached from trees and any seed can be released from the resting position of the cone. The literature is also clear that important factors determining the success of natural regeneration are vegetation control (Caccia and Ballare, 1998), ground preparation (Beland et al., 2000) and predation by small mammals (Castro et al., 1999). An experiment was therefore designed with the following objectives:
- 1 to examine the temporal variation of seed fall and seedling emergence;
- 2 to estimate the amount and quality of seed produced via seed rain and cone-drop;
- 3 to examine the influence of seedbed preparation (ground disturbance and vegetation control) and exclusion of small mammals on the establishment and survival of seedlings and
- 4 to investigate the relationships between seed release, seedling survival and the regeneration environment (climate and competing vegetation).
- 2 to estimate the amount and quality of seed produced via seed rain and cone-drop;
In most studies of natural regeneration, a range of species is observed. From the outset of the experiment, we decided to use the same methods of assessment for all the main species regenerating; these were Corsican pine, birch (Betula pendula Roth.) and western hemlock (Tsuga heterophylla (Raf.) Sarg.).
|
Materials and methods |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
The experiment was established between August 2000 and February 2001 in a 1.5-hectare, even-aged stand of Corsican pine planted in 1950. The stand is located within Alice Holt Forest, Hampshire, UK (Longitude 0° 51'W; latitude 51° 11'N) and is at an altitude of 112 m a.s.l. The site is level but on the north-west edge of a plateau and the ground beyond falls off into the valley of the river Wey at 70 m a.s.l. To the immediate south and west the ground is generally flat. The stand is surrounded by conifer woodland in which birch is a component except to the south where there was a young stand of Corsican pine planted in 1992. To the north, there was an adjacent stand of western hemlock planted in 1966. Soils in the area are surface water gleys (classified as 712b) according to the classification of Avery (1980)
; these are slowly permeable seasonally waterlogged clay soils originating from Jurassic and Cretaceous clays. At the start of the experiment, the trees were 50 years old and there were 217 trees ha–1 with a mean d.b.h. of 36 cm and height of 25 m. The stand had a patchy understorey (20 plants ha–1) of hazel (Corylus avellana L.), holly (Ilex aquifolium L.), birch, oak (Quercus robur L.) and sweet chestnut (Castanea sativa L.). The ground flora was dominated by bramble (Rubus fruticosus L. agg.), bracken (Pteridium aquilinum (L.) Kuhn), soft grasses, honeysuckle (Lonicera periclymenum L.), bluebells (Hyacinthoides non-scripta L.) with occasional wood sorrel (Oxalis acetosella L.) and stitchwort (Stellaria holostea L.). The area would be classified as a W10a sub-community (Quercus robur – Pteridium aquilinum – Rubus fruticosus) according to Rodwell (1991).
Within the stand, five rectangular enclosures were created each by the erection of a deer and rabbit proof fence measuring 24 x 10 m (Trout and Pepper, 2006). In each enclosure, there were three plots measuring 6 x 8 m with an inner assessment plot of 2 x 2 m (Figure 1). These seedling survival plots were randomly assigned to one of the treatments described in Table 1.
|
|
Within each enclosure, in a parallel area of 4 x 24 m, in which there was no vegetation control, there were four locations randomly assigned to two seed traps and two cone collection quadrats (Figure 1). The cone collection quadrats were marked areas 1 x 1 m from which any fallen cones, with >50 per cent of cone length in the quadrat, were collected. The seed traps (1 x 1 m) were based on a seed trap design recommended by David et al. (1997)
. Four prototypes were trialled during 2000 and a number of modifications made; these were mainly to stop grey squirrels (Sciurus carolinensis Gmelin) using the traps as vantage points from which to strip pine cones of seed. The final seed trap design is shown in Figure 2; each trap was 1 m above the ground. During the trial in 2000, only occasional Corsican pine seedlings were observed and there were no saplings (height >1.3 m and d.b.h. <7 cm) present.
|
The following assessments took place between 2 February 2001 and 1 March 2004.
Seedling survival plots
At the start of the experiment, only one holly seedling was recorded in one of the 15 plots. Thereafter, at regular intervals (March to August, 2–4 weeks; September to February, 6–8 weeks), the presence of new seedlings was monitored. Each new seedling was recorded by the careful insertion of a small plastic tag (1.0 x 8 cm) which was given a unique number. The survival of all seedlings was then recorded at each subsequent assessment. The presence of seedlings was recorded using the following scores: 1 – present; 2 – missing and 3 – dead. The missing score was a precaution to ensure that seedlings did not regenerate from below ground. When the seedling had been dead for a period of 2 months, the plastic tag was carefully removed from the soil to minimize disturbance at the surface.
Seed traps
Cones and seeds from each seed trap were carefully collected at least weekly using a small dustpan and brush and placed in a sealed polythene container for transfer to the seed storage area. All material on the ‘top’ and in the tray was collected. Material was air dried in open wooden pallets each with four trays made from thick filter paper; each of these paper trays was used to collect the monthly material from one seed trap. The seed storage area was a section of an unheated building used to store seed in the Forestry Commission's seed extractory. This is located on the same campus as Alice Holt Research Station, Hampshire, UK, which is 
1 kilometre to the south of the experimental site.
At regular intervals, material from each seed trap was amalgamated into monthly collections, sorted and seeds identified and counted. Where appropriate, seeds were extracted from intact cones. The most numerous seeds were birch, Corsican pine and western hemlock – the quality and performance of these seeds were assessed in accordance with the International Rules for Seed Testing (ISTA, 1999). Unfortunately material for July, August and September 2001 was combined. It was possible to compensate for this in the analysis; however, when the data are presented, the seed count for the 3-month period has been divided by three and presented as a monthly total. Seed collection was continued until the end of April 2004.
Cone collection quadrats
Cones were collected at the same time as material from the seed traps and were handled and stored separately using similar procedures. Cones that had been chewed by grey squirrels were classified as cores and the undamaged ones were cones. Seed was extracted from the cones and tested as above. At the start of the experiment in February 2001, a total of 94 (mean = 9.4; SD = 8.1) old cones were removed from the quadrats. All except six of these old cones were judged likely to have fallen off the Corsican pine trees during 2000.
Germination tests
Birch, Corsican pine and western hemlock seeds were incubated on a substrate of moist filter paper for up to 28 days at a daily alternating temperature of 20/30°C (16 h at 20°C in the dark, followed by 8 h at 30°C illuminated by 11 Wm–2 warm-white fluorescent tubes). Germination tests were assessed weekly and at the end of the germination period, all seedlings were classified as either ‘normal’ or ‘abnormal’ germinants, and wherever possible, ungerminated seeds identified as ‘fresh’, ‘dead’ or ‘empty’ according to ISTA (1999). The proportion of viable seeds (P) was calculated as:
 |
where n is the number of normal seeds; ab is the number of abnormal seeds; f is the number of fresh seeds and T is the total number of seeds. The contribution of abnormal and fresh seed for each of the species studied was 3.6 per cent for Corsican pine, 7.7 per cent for western hemlock, 3.6 per cent for birch and 12.5 per cent for pine from the cone collection quadrats.
Vegetation
The vegetation in each of the assessment plots was recorded in April 2001, April 2002 and May 2003 by estimating cover in 5 per cent classes of grasses, bramble, bracken and ferns, other herbaceous and broadleaved, woody shrubs, bare ground, leaf and needle litter, mosses and deadwood according to the method of Kerr et al. (2002).
Meteorological data
Data from the meteorological station at Alice Holt Lodge were obtained; the station is 750 m to the south of the experiment site and at the same altitude. Daily averages or totals of the following were obtained for the period of the experiment: wet and dry bulb temperature, maximum temperature, minimum temperature, grass minimum temperature, rainfall and sunshine hours. A standard formula was used to calculate relative humidity as shown at: http://members.nuvox.net/on.jwclymer/wet.html.
Data analysis
The seed counts were analysed using a Poisson generalized linear model with log link (McCullagh and Nelder, 1989; Payne, 2005). The log of the collection time (in months) was used as an offset to allow for the one occasion when the counts were for a period other than a month. The equivalent generalized linear mixed model (Schall, 1991) was used to compare the variation between blocks and positions (seed traps or quadrats) with the residual variation.
For the viability data, a binomial generalized linear model with a logit link was used. For both Poisson and binomial generalized linear models, over dispersion was accommodated by estimating the dispersion parameter and testing changes in deviance against the F-distribution.
To examine the pattern of viability across time, a smoothing spline with 8 degrees of freedom was fitted. This provides a flexible smooth function that highlights the main features of changes over time. The splines were fitted as part of a generalized additive model (Hastie and Tibshirani, 1990; Payne, 2005). The choice of degrees of freedom was based on visual inspection to achieve the best fit of the curve.
From the data recorded for each seedling, the survival times were computed. These are the times (in weeks) between the first recorded value and the last recorded value of the seedling being alive. Those seedlings still alive at the end of the study are considered to be censored, that is their lifetime is known to exceed the time for which they were observed. The minimum observation period during the study was 2 weeks, an observed seedling may have been present for up to 14 days before being recorded; similarly, when it was observed to have died, it could have been alive for up to 14 days after the last recording. As a result of this, the observed time represents a minimum lifetime. As it was long-term survival that was of interest, this difference between recorded and actual lifetime was not important. However, it can lead to seedlings with a zero lifetime; so when fitting models, an adjustment of 3.5 days was added to the lifetimes.
To examine the distribution of lifetimes, the Kaplan–Meier estimate of the survival function was computed. The survival function S(t) for a seedling is the probability that a seedling will survive to time t (see Gross and Clark, 1975; Kalbleisch and Prentice, 2002). If no seedlings had censored survival times, then a simple estimate of S(t) would be the proportion with a survival time greater or equal to t. However, an adjustment had to be made for the seedlings that had survived to time t at the end of the experiment and so would have had survival times greater than t if they had been observed beyond the end of the experiment. This was achieved using the Kaplan–Meier estimator:
 |
where nt is the number of seedlings that survived to time t and dt is the number of seedlings that died at time t.
To produce a parametric model of seedling survival, the form of the function S(t) had to be identified. To identify a suitable model, a useful tool is the cumulative hazard plot. The hazard rate is the probability of failure at time t, given survival to time t and is denoted by 
(t). The cumulative hazard rate is the integral of (t) and is denoted by 
(t). The survival function can then be written as S(t) = exp(–(t)), so an estimate of the cumulative hazard rate can be computed as the minus log of the Kaplan–Meier survival function. Plotting the cumulative hazard (or its log) against time (or the log of time) gives a linear plot for some common survival functions. In particular, for the Weibull distribution, the cumulative hazard rate is
 |
 |
A plot of log(–log(S(t)) vs time should be approximately linear if the distribution is Weibull.
The Weibull distribution can be fitted by the algorithm given by Aitkin and Clayton (1980) and Aitkin et al. (1989). This has been implemented in the GenStat procedure RSURVIVAL (Payne et al., 1993). Models can be compared using the difference in deviance (–twice the log likelihood ratio) with a chi-squared distribution; approximate standard errors of model parameters are also given and these were used to test for significant effects.
In examining the survival of seedlings, there are two time scales that need to be considered.
- 1 Calendar time, T, which reflects the usually cyclic changes in the environment and indicates differences in survival that are environmentally dependent.
- 2 Lifetime of the seedlings, t, which reflects differences in survival that are age dependent.
- 2 Lifetime of the seedlings, t, which reflects differences in survival that are age dependent.
As the seedlings germinated at different times, the relationship between the two time scales is not the same for all seedlings. So the hazard rate for an individual will be
 |
where xT is a vector of calendar time-dependent covariates such as temperature and vegetation cover.
There is also the issue of the possible lag in the response between a time-dependent covariate and failure. For example, a frost may not immediately kill a seedling, but weaken it so that it is killed by a combination of effects later in its life.
In this study, there were two types of time-dependent covariates considered: (1) yearly measurements of vegetation cover and (2) daily measurements of meteorological data. In the case of the yearly measurements, the data can be expanded so that a seedling occurs once for each year that it is alive. For example, if the seedling appears in year 1 and dies in year 2, it will appear twice in the dataset. In the first instance, it will have the covariates for year 1 and will be censored at the end of the year. In the second instance, it will have the covariates for year 2 and include the actual time of death. An approximation to the Weibull distribution was obtained by transforming the time scale by the Weibull parameter and fitting an exponential distribution. The vegetation variables were selected for inclusion in the models by using forward selection and backward elimination procedures.
The meteorological data were summarized for the intervals between recordings for seedling survival. As there were 65 recording dates, rather than use the approach that was used above for the vegetation data, the data were considered as a sequence of binomial observations at each interval. The binomial observation being the number that died in the period out of the number that were alive at the start of the period. As the live seedlings for each interval have different lifetimes up to the interval, the average log lifetime of the live seedlings was included in the model. By using a binomial generalized linear model with a log link, this is an approximation to the Weibull model used on the lifetime data. The general seasonal variations were modelled with a smoothing spline with 8 degrees of freedom. The meteorological variables were then added to the model and tested for significance.
The differences between years for vegetation cover data were approximately normal and analysed using an analysis of variance.
|
Results |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
Seed count data
The quantity and pattern of seed dispersed was very different for Corsican pine compared with the other two species (Figure 3). In 2001, there was little variation in the amount of pine seed dispersed between March and November; in 2002 there was a distinct peak in March and April and in 2003 there were two lower peaks in March/April and September. For pine, the range of monthly values varied between 2 (a hectare equivalent of 2000) and 223 (223 000 ha–1): some pine seed was trapped in every month of the study. The patterns of seed production for birch and western hemlock were similar to each other, with distinct peaks in the autumn of each year (Figure 3). The range of monthly values varied between 0 (0 ha–1) and 2208 (2 208 000 ha–1) for western hemlock and 0 (0 ha–1) and 2510 (2 510 000 ha–1) for birch. For hemlock and birch, there were only three occasions when no seed was observed in any one month, in April and June 2001 for birch and June 2002 for western hemlock. The total number of seeds collected for each species was Corsican pine 1342, western hemlock 5850 and birch 8788.
|
Analysis of the data showed that there were significant effects of block, year, month and the interaction between year and month, confirming that there were variations in the amount of seed produced between months and between years (Table 2). The block effect was significant for all three species (P < 0.001) and the effect of seed trap position was significant for Corsican pine and western hemlock (P < 0.05); however, these latter effects, although significant, were small compared with the temporal variation.
|
In total, 38 cores (26 in 2003) and two cones were collected in the seed traps during the course of the experiment.
Cone and core data
The total number of pine cones collected in the quadrats was 107 and the number of cores was 37 (Figure 4). Both sets of data indicate that, with the exception of 2001, most cones are lost from the trees in March to May, a narrower range compared with the Corsican pine seed data. Very few cones were lost from trees outside this period. Statistical analysis showed that months and years were significant for cones and cores and the interaction between months and years was significant for cones (all P < 0.001).
|
The timing of seed dispersed via the cones and cores (Figure 3) follows very closely that for the release of cones and cores from the trees (Figure 4). The total amount of seed from cones and cores (1141 seeds) was similar to that from the seed traps (1342 seeds). Analysis of the data showed that block, month and the interaction of year and month were all significant (all P < 0.001), but not year.
Seed viability
The distribution of viable seeds for the three species for all dates combined is shown in Figure 5 (for observations with 10 or more seed). In general, western hemlock and Corsican pine showed greater variability than does birch, while western hemlock had a lower proportion of seeds with very low viability. Analysis of the pine data showed that block, year, month and quadrat were all significant (P < 0.001). An interesting result was that seed from traps had higher viability than from quadrats (P < 0.001) (Figure 6); this can probably be explained by the fact that there was three times as much fresh and abnormal seeds in the quadrats compared with trays. Figure 6 also shows the variation of viability between a maximum of over 35 per cent to a minimum of under 5 per cent. The patterns of viability were similar for hemlock and birch, although analysis showed that the influence of all main factors and two-way interactions were significant (P < 0.001) with the exception of the table/quadrat comparison. Arguably, the most important result was that having adjusted for year and month there was still a significant positive relationship (P < 0.001) between the number of pine seeds and viability.
|
|
Seedling emergence
During the course of the 38-month experiment, a total of 205 (34 235 ha–1) Corsican pine seedlings were observed along with 418 (69 806 ha–1) birch and 2514 (419 838 ha–1) western hemlock seedlings. Very few seedlings were found in the control treatment: 3 per cent of the total number for western hemlock and birch and 8 per cent for Corsican pine (Table 4).
The pattern of emergence of the seedlings showed variation between species, years and months (Figure 7). A surprising result was that at least one seedling emerged in every month over the 3-year period for western hemlock; for birch and Corsican pine, there was only 1 month with no new seedlings, November and October, respectively. As shown in Figure 7, the circular mean for seedling emergence for Corsican pine was May; for birch, it was late June, and it was April for western hemlock.
|
Seedling survival
Examination of the estimated survivor function showed that western hemlock had a higher survival rate than either birch or Corsican pine (Figure 8a). Across all species, the control treatment had much lower survival compared with the other two treatments (Figure 8b).
|
The Weibull model with terms for blocks, species, treatment and species by treatment interaction had an estimated value for
of 0.3554. The interaction between species and treatment was significant (difference in deviance of 17 with 4 degrees of freedom, P < 0.01). Examination of the t probabilities of the pairwise differences indicated that there were no significant differences in survival between birch, Corsican pine and the control treatment for western hemlock. In general, the probability of a seedling of birch, Corsican pine or a hemlock in the control treatment surviving for 300 days was 0.50, whereas for hemlock in the two other treatments were higher at 0.60 (Table 5). The maximum and minimum temperatures were significantly (P < 0.001) related to the death rate for seedlings in the interval. Rain and temperatures from previous intervals were not significant. The coefficient for maximum temperature was positive while the minimum was negative, indicating high maximum temperatures (>25°C) and low minimum temperatures (<0°C) led to increased seedling mortality.
Vegetation cover
The initial changes in vegetation in the ‘S’ and ‘P’ treatments were linked to the nature of treatments (Table 6). The vegetation in the control treatment was dominated by bramble, bracken and leaf litter. In general, the assessments were early in the year and probably caused the cover of bracken to be underestimated compared with the situation in July. The control treatment generally showed little change throughout the experiment, although an increase in bramble was noted in May 2003. Large changes occurred in treatments S and P with the bare conditions changing rapidly to moss, grass and bramble. The only significant change (P < 0.05) between years 2 and 3 was the increase in the proportion of grass. Attempts to examine the relationship between seedling survival and vegetation cover were inconclusive.
Seed release and meteorological data
Initial investigation of the relationship between seed release and meteorological data suggested that relative humidity was worthy of further analysis. The models were adjusted for both year and month and this showed that relative humidity was significantly related to the Corsican pine seed counts on both quadrats and seed traps (P < 0.05) along with the relative humidity from the previous month (P < 0.001). The coefficients in all cases were negative, confirming the relationship between dryness and seed release.
|
Discussion |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
In any stand of trees, the recruitment of new seedlings is limited by both the supply of seeds and the availability of suitable sites for establishment (Caspersen and Saprunoff, 2005
). In this experiment, we attempted to investigate both these factors, firstly, by detailed examination of the seed supply, and secondly, by manipulation of the ground flora. We attempted to produce conditions that had been shown in other studies to improve the germination and survival of pine seedlings (Cucchi, 1965; Beland et al., 2000; Karlson and Orlander, 2000). In addition, we attempted to reduce the predation of seeds and seedlings by small mammals, which Brown and Neustein (1972) considered as a particular problem for the regeneration of Corsican pine in Britain. The results will be discussed in terms of seed production and viability and seedling emergence and survival. The focus of the study was on Corsican pine; however, the fact that western hemlock and birch seed were present allows the reproductive strategies of the three species to be compared and contrasted. Seed production and viability
The frequency of seed collections in the study has allowed detailed consideration of the temporal aspects of seed supply. Guidance of the main periods of conifer seed dispersal in Britain is given by Nixon and Worrell (1999) as being March to June for Corsican pine and September to April for western hemlock. For birch, the main period is generally considered to be August to January (Kerr and Evans, 1993). In general, the results from this study support the recommendations for western hemlock and birch; however, the results for Corsican pine are different. In 2001, 2003 and the early part of 2004, there was a reasonable amount of Corsican pine seed produced after the end of June and before March. Analysis of the relationship between seed dispersal of Corsican pine and meteorological data showed a clear link with humidity, i.e. when humidity was lower cones released seed. Meteorological data for Corsica and Alice Holt Lodge has been compared and contrasted by Brown (1960a, b). It is clear from the information presented that summers are warmer and drier in Corsica compared with southern Britain and so most seed in Corsica is probably released between March and June. However, the situation is different in Britain; the summer is generally cooler and moister and the results of this study indicate that this leads to a longer season of seed release for Corsican pine.
An important point here is how information on the temporal aspects of seed supply is presented to forest managers. Nixon and Worrell (1999) do this by indicating distinct dates within which most seed will be dispersed. The origin of this information is from earlier publications by Seal et al. (1962) giving recommended times for cones to be collected from standing trees. They recommended collecting Corsican pine cones between December and February to maximize the proportion of ripe cones and minimize the possibility of earlier seed shed. The results of this study are not at odds with the recommendations of Seal et al. (1962), but do indicate that that the dates of seed release are more variable than guidance given in Nixon and Worrell (1999).
One of the main questions that the study was seeking to answer was ‘is the supply of seed a constraint to regeneration?’ Other studies on Pinus nigra spp. consider other aspects of the process of regeneration (Ordonez and Retana, 2004; Thibault and Prodon, 2006) and therefore it is necessary to consider this question with reference to work on other pine species. A useful piece of work by Shelton and Wittwer (1996) working with shortleaf pine (Pinus echinata Mill.) drew on earlier work by Haney (1962) to suggest different classes for the potential adequacy of seed supply for establishing natural regeneration. The classes are failure, <25 K viable seeds ha–1; poor, 25 to <75 K; fair, 75 to <200 K; good, 200 to <625 K and bumper, 
625 K. This classification is an attempt to show the seed crop's general potential for regeneration if other factors are favourable (Lynch et al., 2003). Shortleaf pine has very similar seed yields per cone compared with Corsican pine, although the average seed weight is 50 per cent less than for Corsican pine (USDA Forest Service, 1974). Application of Shelton and Wittwer's classification to Corsican pine therefore seems to have a reasonable basis. For seed intercepted in seed traps in this study, the classification would judge 1 year to be a ‘failure’ and the other two as ‘poor’ (Table 3). However, seed was also dispersed via cone drop and considering that seedlings were produced, it would be difficult to conclude that that seed production was a constraint to regeneration in this experiment. Seed production is likely to increase in the future as the stand is younger than the period of maximum seed production of 60–90 years (Gordon and Faulkner, 1992) for Corsican pine in Britain.
|
|
|
|
An interesting finding from the study was that 46 per cent of the total amount of pine seed reaching the forest floor was still in the cone. This contrasts with the findings of Summers and Proctor (2005)
for Scots pine who found that few residual seeds remained in cones after shedding. However, few other studies have been located where this route of seed dispersal has been quantified. The fact that this is a major input of seed challenges the classifications used above to judge whether or not seed production was adequate as they ‘did not’ include seed dispersed in the cone. An interesting question is how does this source of seed contribute towards the regenerative capacity of the stand? If it makes a major contribution, then secondary dispersal would be involved (Castro et al., 1999) or if low, then most of it may be consumed by birds and animals.
The finding that seed supply and viable seed supply are directly related is an important finding from this experiment; few other studies have been able to make this link. This is because many studies on pine regeneration do not consider both aspects, for example Beland et al. (2000), Nilsson et al. (2002) and Summers and Proctor (2005). However, in other studies such as Karlson and Orlander (2000) and Cain and Shelton (2000), seed quality is considered, although the methods to determine this vary. In general, seed viability for the three species was low and for the two conifers was below the figures published by Nixon and Worrell (1999). They quote seed viability ranges of 70–80 per cent for Corsican pine and 50–65 per cent for western hemlock; however, the basis of these figures is tested seed samples from which much of the obviously non-viable seed has already been removed. Improved and more realistic seed viability data are required for better guidance to forest managers on natural regeneration.
Although the study was only 38 months duration, it has allowed some information to be collected on the variation of seed quantity and quality between years for Corsican pine and the other two species. Kerr (2000) stated that Corsican pine has the capacity to produce some seed every year, but good seed years only occur every 3–5 years, citing Seal et al. (1962) as evidence. If the earlier classification of Shelton and Wittwer (1996) for P. echinata is accepted, then 1 year was a failure and two were poor, but overall enough viable seed was produced in the 3-year period for successful regeneration. Very little other data are available for Pinus nigra spp., but Sarvas (1962), Bergman et al. (1981) and Nixon and Worrell (1999) consider that Scots pine often exhibits a more uniform level of seed production compared with the spruces. Nixon and Worrell (1999) cite data from Forestry Commission (1948) that show that in the period 1930 to 1948, Scots pine had 6 ‘abundant’ seed years, 7 ‘good’ years and 5 ‘poor’ years. Evidence from other studies support the fact that pine species produce seed in most years, but occasionally have years when random events, such as fluctuations in weather or pest populations, have a controlling influence and very little seed is produced (Karlson and Orlander, 2000; Lee et al., 2004). In much literature on natural regeneration, emphasis is placed on the frequency of good seed years (Philipson, 1997; Cain and Shelton, 2000). However, a greater constraint on successful regeneration is when there is little or no seed production: perhaps this should be given greater emphasis in guidance to forest managers?
Emergence and survival of seedlings
One of the main reasons for this experiment was that many foresters in Britain believed that Corsican pine does not regenerate in Britain (Kerr, 2000). This perception is caused partly because Corsican pine has generally been managed on a relatively short rotation of 55–65 years – before significant seed production begins – and partly because the other lowland pine (Scots pine, Pinus sylvestris L.) regenerates freely. However, Johnson (1976) clearly demonstrated that natural regeneration of Corsican pine takes place in Britain and this study corroborates his findings. The fact that during this 38-month study 205 new Corsican pine seedlings were recorded (equivalent to 34 235 ha–1) is good evidence that it regenerates in Britain and produces enough viable seed for successful regeneration. However, many other factors affecting regeneration, such as competition from vegetation and browsing mammals (Harmer et al., 1997), were eliminated from the experimental blocks but these were not studied in enough detail to make a judgement about the likelihood that seedlings would develop into saplings. The number of pine seedlings was relatively small compared with western hemlock, a species that is well known to be a successful regenerator (Nixon and Worrell, 1999). In summary, Corsican pine in the stand studied has produced enough seedlings to regenerate the stand under optimal conditions.
An interesting result was that at least one seedling emerged in every month over the 3-year period for western hemlock; for birch and Corsican pine, there was only 1 month with no new seedlings, November and October, respectively. For Corsican pine, most seedlings emerge between March and July, which is within the period of seed production. However, for western hemlock (January to August) and birch (April to September), the main period of seedling emergence occurs a number of months after the main period of seed production.
In terms of seedling emergence, there were clear differences between the control and other two treatments. This was of no real surprise as many previous studies have shown that vegetation control and seedbed preparation can have a positive effect on the number of seedlings produced in natural regeneration (Winsa, 1995; Beland et al. 2000; Karlson and Orlander, 2000; Nilsson et al. 2002; Karlsson and Nilsson, 2004; Lee et al. 2004; Wang and Kemball, 2005). The relatively small number of seedlings that were found in the control plot explains the observation that very few new seedlings were observed in the stand during 2000, when the seed trap design was being trialled. However, there were no clear differences between the ‘seedbed preparation’ and the ‘seedbed preparation with small mammal protection’. This was not expected as there were clear signs in the ground flora that a range of small mammals actively used the area; however, these results indicate that feeding was non-limiting in contrast to some other studies (Nystrand and Granstrom, 2000).
A unique aspect of this study was the survival modelling of the seedlings of the three species that regenerated. Detailed studies on the demography of seedlings have been reported (Sacchi and Price, 1992; Shibata and Nakashizuka, 1995; Duchesneau and Morin, 1999; Forbis, 2003), but only one by Daskalakou and Thanos (2004) has reported monthly survival of pine seedlings (Pinus halepensis Mill.). Other studies on the survival of pine seedlings such as those above have used annual assessments of seedlings to derive figures for seedling mortality; for example Karlson and Orlander (2000). The demography of seedlings found in this study would suggest that annual assessments could miss the transient appearance and mortality of the majority of seedlings.
The main differences in survival were between the western hemlock in the two seedbed preparation treatments and all the other species/treatment combinations. The fact that the survival of Corsican pine and birch were similar is probably explained by the fact that both have shade intolerant seedlings (Savill, 1991). However, shade tolerance is not the only factor involved as shown by the fact that western hemlock, which has shade tolerant seedlings, had relatively small differences in survival over 300 days compared with the other species.
If forest managers want to naturally regenerate this stand of trees, then the results of this study show that
- 1 The regenerative capacity of the Corsican pine is adequate.
- 2 Many of the negative factors affecting regeneration, particularly competing vegetation and browsing, would need to be controlled. The effects of browsing were not studied in this experiment, but it was clear from observations in the stand that it was important.
- 3 Any improved conditions for regeneration would also favour birch and western hemlock; the latter is likely to dominate unless the seed source is removed.
- 4 Under current conditions, 50 per cent of pine seedlings would survive for 300 days; this figure could be used to judge the correct time for thinning to improve growth conditions for the seedlings. Any canopy removal would lead to an immediate reduction in seed supply because of the 3-year reproduction cycle in Corsican pine (Philipson, 1997
).
- 2 Many of the negative factors affecting regeneration, particularly competing vegetation and browsing, would need to be controlled. The effects of browsing were not studied in this experiment, but it was clear from observations in the stand that it was important.
An unfortunate endnote to this research is that between the start of the project and publication of results, the incidence of red band needle blight (Dothistroma pini (Hulbary)) has increased on Corsican pine in Britain and the species faces an uncertain future (Brown et al., 2003).
|
Conflict of Interest Statement |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
None declared.
|
Acknowledgements |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
The authors would like to acknowledge the help of a large number of staff in the Technical Support Unit of Forest Research who established, maintained and assessed the trial to a high standard. Lorelie Ives also gave valuable support for the study in the seed laboratory. Anna Brown, Bill Mason, Helen McKay and two unknown reviewers who gave useful comments on drafts of the paper.
|
References |
|---|
|
TopSummaryIntroductionMaterials and methodsResultsDiscussionConflict of Interest StatementAcknowledgementsReferences |
|---|
Aitkin M, Clayton D. The fitting of exponential, Weibull and extreme value distributions to complex censored survival data using GLIM. Appl. Stat. (1980) 29:156–163.[CrossRef]
Aitkin M, Anderson D, Francis B, Hinde J. Statistical Modelling in GLIM (1989) Oxford: Clarendon Press.
Avery BW. Soil Classification for England and Wales (Higher Categories). In: Soil Survey Technical Monograph No. 14 (1980) Harpenden, UK: Soil Survey of England and Wales.
Beland M, Agestam E, Ek PM, Gemmel P, Nilsson U. Scarification and seedfall affects natural regeneration of Scots pine under two shelterwood densities and a clear-cut in southern Sweden. Scand. J. For. Res. (2000) 15:247–255.[CrossRef]
Bergman F, Borset O, Leikola M. Natural regeneration of conifers in Fennoscandia. Proc. XVII IUFRO Congr. Div. (1981) 1:381–401.
Brown JMB. The Corsican pine in its native island, Part 1. Emp. For. Rev. (1960a) 39:294–318.
Brown JMB. The Corsican pine in its native island, Part 2. Emp. For. Rev. (1960b) 39:422–436.
Brown JMB, Neustein SA. Natural regeneration of conifers in the British Isles. In: In Conifers in the British Isles: Proceedings of the Third Conifer Conference, Royal Horticultural Society, London, 5–8 October 1970 (1972) London: Royal Horticultural Society. 29–39.
Brown A, Rose D, Webber J. Red Band Needle Blight of Pine. In: Forestry Commission Information Note 49 (2003) Edinburgh, UK: Forestry Commission.
Caccia FD, Ballare CL. Effects of tree cover, understorey vegetation, and litter on regeneration of Douglas-fir (Pseudotsuga menziesii) in southwestern Argentina. Can. J. For. Res. (1998) 28:683–692.[CrossRef]
Cain MD, Shelton MG. Revisiting the relationship between common weather variables and loblolly-shortleaf pine seed crops in natural stands. New For. (2000) 19:187–204.
Caspersen JP, Saprunoff M. Seedling recruitment in a northern temperate forest: the relative importance of supply and establishment limitation. Can. J. For. Res. (2005) 35:978–989.[CrossRef]
Castro J, Gomez JM, Garcia D, Zamora R, Hader JA. Seed predation and dispersal in relict Scots pine forests in southern Spain. Plant Ecol. (1999) 145:115–123.[CrossRef]
Cucchi CC. Studies on natural regeneration of plantations of Pinus nigra subsp. in the Northern Apennines. Ann. Accad. Ital. Sci. For (1965) 14:57–118.
Daskalakou EN, Thanos CA. Postfire regeneration of Aleppo pine – density, survival and early growth of Pinus halepensis seedlings. In: Proceedings 10th MEDECOS Conference, 25 April–1 May 2004.—Arianoutsou M, Papanastasis VP, eds. (2004).
David A, Wender B, Weis P, Stringer J, Wagner D. A new seed trap design. Tree Planters’ Notes (1997) 48(1–2):35–37.
Duchesneau R, Morin H. Early seedling demography in balsam fir seed banks. Can. J. For. Res. (1999) 29:1502–1509.[CrossRef]
Farjon A. World Checklist and Bibliography of Conifers (1998) Kew, London: The Royal Botanic Gardens.
Forbis TA. Seedling demography in an alpine ecosystem. Am. J. Bot. (2003) 90:1197–1206.
Forestry Commission. Forestry Commission Annual Research Report 1947 (1948) London: HMSO.
Gordon AG, Faulkner R. Identification and assessment of cone and seed crops. In: 1992 Seed Manual for Forest Trees. Forestry Commission Bulletin 83—Gordon AG, ed. (1992) London: HMSO.
Gross AJ, Clark VA. Survival Distributions: Reliability Applications in the Biomedical Sciences (1975) New York: Wiley.
Haney GP. Seedbed scarification aids regeneration of shortleaf pine. J. For. (1962) 60:400–402.
Harmer R, Kerr G, Boswell R. Characteristics of lowland broadleaved woodland being restocked using natural regeneration. Forestry (1997) 70:199–210.
Harmer R, Boswell R, Robertson M. Survival and growth of tree seedlings in relation to changes in the ground flora during natural regeneration of an oak shelterwood. Forestry (2005) 78(1):21–32.
Hastie T, Tibshirani R. Generalized Additive Models (1990) New York: Chapman and Hall.
ISTA. International rules for seed testing. Seed Sci. Technol. (1999) 27((Suppl.)).
Johnson CL. Factors affecting the establishment and distribution of Corsican pine natural regeneration at Holkham National Nature Reserve. Q. J. For. (1976) 70(2):95–102.
Kalbleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data (2002) New York: Wiley.
Karlson C, Orlander G. Soil scarification shortly before a rich seed fall improves seedling establishment in seed tree stands of Pinus sylvestris. Scand. J. For. Res. (2000) 15:256–266.[CrossRef]
Karlsson M, Nilsson U. The effects of scarification and shelterwood treatments on naturally regenerated seedlings in southern Sweden. For. Ecol. Manage. (2004) 205:183–197.[CrossRef]
Kerr G. Natural regeneration of Corsican pine (Pinus nigra subsp. laricio) in Great Britain. Forestry (2000) 73:479–488.
Kerr G, Evans J. Growing Broadleaves for Timber. In: Forestry Commission Handbook 9 (1993) London: HMSO.
Kerr G, Mason B, Boswell R, Pommerening A. Monitoring the Transformation to Continuous Cover Management. In: Forestry Commission Information Note 45 (2002) Edinburgh, UK: Forestry Commission.
Lee CS, Kim JH, Yi H, You YH. Seedling establishment and regeneration of Korean red pine (pinus densiflora S. et Z.) forests in Korea in relation to soil moisture. For. Ecol. Manage. (2004) 199:423–432.
Lynch TB, Nkouka J, Huebschmann MM, Guldin JM. Maximum likelihood estimation for predicting the probability of obtaining variable shortleaf pine regeneration densities. For. Sci. (2003) 49:577–584.
Malcolm DC, Mason WL, Clarke GC. The transformation of conifer forests in Great Britain – regeneration, gap size and silvicultural system. For. Ecol. Manage. (2001) 151:7–23.[CrossRef]
Mason WL. Continuous cover forestry: developing close to nature forest management in conifer plantations in upland Britain. Scott. For. (2003) 57(3):141–149.
Mason B, Kerr G. Transforming Even-Aged Conifer Stands to Continuous Cover Management. In: Forestry Commission Information Note 40 (Revised) (2004) Edinburgh, UK: Forestry Commission.
Mason B, Kerr G, Simpson J. What is Continuous Cover Forestry? In: Forestry Commission Information Note 29 (1999) Edinburgh, UK: Forestry Commission.
Mason WL, Edwards C, Hale SE. Survival and early seedling growth of conifers with different shade tolerance in a Sitka spruce spacing trial and relationship to understorey ligth climate. Silva Fenn. (2004) 38:357–370.
McCullagh P, Nelder JA. Generalized Linear Models (1989) 2nd edn. New York: Chapman and Hall.
National Assembly for Wales. Woodlands for Wales: The National Assembly for Wales strategy for Trees and Woodlands (2001) Aberystwyth, UK: Forestry Commission.
Nilsson U, Gemmel P, Johansson U, Karlsson M, Welander T. Natural regeneration of Norway spruce, Scots pine and birch under Norway spruce shelterwoods or varying sensities on a mesic-dry site in southern Sweden. For. Ecol. Manage. (2002) 161:133–145.[CrossRef]
Nixon CJ, Worrell R. The Potential for the Natural Regeneration of Conifers in Britain. In: Forestry Commission Bulletin 120 (1999) Edinburgh, UK: Forestry Commission.
Nystrand O, Granstrom A. Predation on Pinus sylestris seeds and juvenille seedlings in Swedish boreal forest in relation to stand disturbance by logging. J. Appl. Ecol. (2000) 37:449–463.[CrossRef]
Ordonez JL, Retana J. Early reduction is post-fire recruitment of Pinus nigra by post dispersal seed predation in different time since-fire habitats. Ecography (2004) 27:449–458.[Medline]
Payne RW, ed. The Guide to GenStat Release 8, Part 2: Statistics (2005) Oxford: VSN International.
Payne RW, Lane PW, Digby PGN, Harding SA, Leech PK, Morgan GW, et al. Genstat 5: Release 3 Reference Manual (1993) Oxford: Clarendon Press.
Pepper HW, Kerr G. Preliminary studies on collars to protect trees from grey squirrel bark-stripping damage. Q. J. For. (2005) 99(2):105–112.
Philipson JJ. Predicting cone crop potential in conifers by assessment of developing cone buds and cones. Forestry (1997) 70(1):87–96.
Rodwell J. British Plant Communities Volume 1, Woodlands and Scrub (1991) Cambridge: Cambridge University Press.
Sacchi CF, Price PW. The relative roles of abiotic and biotic factors in seedling demography of arroyo willow (Salix lasiolepis: Salicaceae). Am. J. Bot. (1992) 79:395–405.[CrossRef][Web of Science]
Sarvas R. Investigations on flowering and seed crop of Pinus sylvestris. Commun. Inst. For. Fenn. (1962) 67:1–198.
Savill PS. The Silviculture of Trees Used in British Forestry (1991) Wallingford, UK: CAB International.
Schall R. Estimation in generalized linear models with random effects. Biometrika (1991) 78:719–727.
Seal DT, Matthews JD, Wheeler RT. Collection of Cones from Standing Trees. In: Forestry Commission Forest Record 39 (1962) London: HMSO.
Shelton MG, Wittwer RF. Shortleaf pine seed production in natural stands in the Ouachita and Ozark Mountains. South. J. Appl. For. (1996) 20(2):74–80.
Shibata M, Nakashizuka T. Seed and seedling demography of four co-occuring Carpinus species in a temperate deciduous forest. Ecology (1995) 76:1099–1108.[CrossRef][Web of Science]
Summers RW, Proctor R. Timing of shedding seeds and cones, and production in different stands of Scots pines at Abernethy Forest, Scotland. Forestry (2005) 78:541–549.
Thibault JC, Prodon R. Seed predation by birds shortly after a wildfire in a Corsican pine forest. Int. J. Wildl. Fire (2006) 15(1):81–86.[CrossRef]
Trout RC, Pepper HW. Forest Fencing. In: Forestry Commission Technical Guide (2006) Edinburgh, UK: Forestry Commission.
UKWAS. Certification Standard for the UK Woodland Assurance Scheme (2000) Edinburgh: UKWAS Steering Group. 42p.
USDA Forest Service. Seeds of Woody Plants in the United States. In: USDA Agricultural Handbook 450 (1974) Washington, DC: US Department of Agriculture.
Wang GG, Kemball KJ. Balsam fir and white spruce seedling recruitment in response to understorey release, seedbed type, and litter exclusion in trembling aspen stands. Can. J. For. Res. (2005) 35:667–673.[CrossRef]
Willoughby I, Dewar J. The Use of Herbicides in the Forest 1995. In: Forestry Commission Handbook 8 (1995) London: HMSO.
Winsa H. Influence of rain shelter and site preparation on seedling emergence of Pinus sylvestris L. after direct seeding. Scand. J. For. Res. (1995) 10:167–175.
Received 23 July 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||