• No results found

This chapter is published as:

6.2. Possibilities for further research

Improving seed sourcing guidelines requires more information on patterns of adaptive genetic variation than currently exists within GB (Boshier and Stewart, 2005; Cavers and Cottrell, 2014). The approach applied in chapter two using climatic and ecological information forms a basis for improving the ability to match seed sources to planting sites although remains couched in the assumption that local adaptation exists. To define transfer limits, well replicated provenance tests for a range of species, representing a large a range of source and trial environments are possible. Work is underway with trial series’ of Betula pendula, Fraxinus excelsior and Sorbus aucuparia which will contribute greatly towards our understanding of adaptive variation in these species (C. Rosique et al., unpublished data).

Given the highly resource intensive nature of such experiments and the influence that early decisions can have on results, it will be necessary to set priorities in terms of species choice and to make very clear the purpose of conducting such experiments. If they are to identify seed sources which will be most productive in certain environments the goal may be to sample from high quality trees in the field and to focus on measuring and reporting variation in economically important traits so that these can contribute to improvement programmes in the long term. However, if the aim is to establish more generally whether genotype by environment interactions exist then initial planning may be less selective (White et al., 2007). Collaboration with the forest industry would be very helpful in this regard, for instance by gaining semi-quantitative data on establishment mortality rates of different species in different environments by means of a survey of forest managers. The survey could also target the forest nursery sector and attempt to gain information on relative volumes of stock for different species sold as ‘beat-up’ (planting stock resold to replace failed trees in the first year), as well as canvassing opinion on seed sourcing throughout the sector as a whole.

Future research could endeavour to determine the influence of edaphic or biotic variation on phenotypic variation among populations of Scots pine and other species. This would help to understand whether it is necessary to consider such variation within seed sourcing

guidelines. A plausible experiment would be to reciprocally grow genotypes from different environments in soils collected from their home sites in short term seedling tests under common conditions, with a controlled replicate growing in sterilised soil. Although it would difficult to fully replicate the edaphic environment of any given site in artificial conditions; it would be impossible to control for the effect of climate in field conditions and so sowing

131

seed in pots in controlled conditions would be desirable. Such an experiment would help indicate whether populations are differentially adapted to their edaphic environments. However, caution would be required in choice of fitness measures, as traditional

‘performance’ indicators such as height growth may reflect enemy release, rather than local adaptation (Dostál et al., 2013; Gundale et al., 2014). A sample of the biotic component of the different soils could be described using molecular approaches prior to sowing seed and at the end of the experiment to determine whether the different tree genotypes influence

recruitment or turnover of different species (e.g. ectomycorrhizae).

The aim of chapter three was to investigate and demonstrate, using a simple methodology, whether timing of pollen production is synchronous amongst populations. Differences among distant sites were found to be as many as 15 days, with anthesis taking place earliest in the warmer west of the country. Due to protogyny and prevailing westerly or south- westerly winds, a hypotheses emerging from the study was that the directional bias in pollen transfer among populations is from the west to the east. However, there is a large conceptual difference between the presence of pollen in one location and the effective dispersal of pollen among locations (i.e. successful dispersal, fertilisation, germination and

establishment). There may be several other pre-zygotic barriers to gene flow among

populations which may be temporal, spatial, or spatio-temporal. Future studies could aim to investigate the timing of female strobilus receptivity among sites and more clearly test hypotheses surrounding the cues of spring reproductive phenology in pines. A relationship with temperature seems highly plausible, but would require higher resolution temperature data than were available in this study for confirmation. Continuing to visit the same trees into the future would generate valuable data, and the mechanisms underpinning variation in reproductive phenology become clearer under further investigation.

The next steps in developing a clear picture of patterns of pollen dispersal in Scottish pine populations would be to parameterise pollen dispersal kernel using molecular markers, accepting that it will be very difficult to capture long-distance events (Kremer et al., 2012). Pollen dispersal kernels could be made spatially explicit/coherent with modelling studies of wind patterns at the time of predicted pollen shedding, as well as elucidating whether certain landscape features (e.g. mountain ranges, conifer plantations) would act against pollen dispersal. Future studies may also endeavour to investigate whether interannual variation in reproductive phenology has consequences for seed production (Koenig et al., 2015).

132

The model presented in chapter four was highly abstracted and necessarily made a series of simplifying assumptions to generalise adaptive responses. Future efforts to use simulations to investigate seed sourcing strategies could endeavour to incorporate more realistic

processes, for instance by considering clinal phenotypic variation in a trait linked closely to fitness, e.g. timing of bud burst (Aitken and Bemmels, 2016), alongside a more spatially explicit landscape and a more sophisticated basis for genetic adaptation to a realistic climate. A further improvement would be to explicitly incorporate heritable variation in phenotypic plasticity into the model (Chevin et al., 2010; Oddou Muratorio and Davi, 2014), to assess the relative contribution of plasticity to the adaptive responses of populations established under the different seed sourcing strategies and identify whether there are evolutionary tipping points in plastic responses (Botero et al., 2015).

An obvious flaw in the model is that there were no implicit consequences for low population size and evolutionary recue was not constrained by any ecological interactions or Allee effects. This was simply a matter of interpretation, as the population size at year 5 was considered to be the main response variable but it is conceivable that, had a competing species with a faster life cycle been included in the model, evolutionary rescue from a very small population size would have been limited.

However, gaining experimental data would be more informative and more valuable than more modelling (Bucharova, 2016). Experiments could help to fill two of the major knowledge gaps the model identified. Gaining information on the extent to which

populations are adaptively differentiated in terms of non-climatic aspects of the environment, as discussed in short term seedling tests would help to inform how likely phenotypic

mismatch leading to early mortality could be among provenances. Longer term provenance tests in realistic field settings will be more useful to determine whether genotypes

translocated from currently more benign climates can survive in the field. Confirming whether there are any differences in the extent to which any of the seed sourcing strategies will help populations adapt to climate change will remain very difficult with long-lived trees. Virtually all aspects of the simulation experiments could realistically be replicated in a microbial system (e.g. Chlamydomonas reinhardtii P.A. Dang.). Although such an

experiment would be biologically interesting, it would not contribute very useful information for forest management and would be little more biologically realistic, and much more difficult than using simulated data.

133