PLAN DE DESARROLLO GUADALUPE 2012 –
Gráfica 12. Nivel Educativo Fuente: DANE 2010.
Functional trait data were both collected from the field and collated from a number of existing sources.
2.3.1. Identifying potentially useful functional traits
The attributes desirable for the Framework Species Approach were considered, and were specified as the following eight characteristics: (1) Fast germination, (2) Reliable germination, (3) High Field Survival, (4) Fast Field Growth, (5) Dense crown, (6) Early flowering/fruiting, (7) Reliable flowering/fruiting and (8) Abundant
flowering/fruiting. Previous studies were also found which identified key functional traits both generally (Cornelissen et al. 2003) and within seasonally-dry tropical forest (Chaturvedi, Raghubanshi & Singh 2011).
As part of this process, common trade-offs between traits (Grime 1977; Swaine & Whitmore 1988; Chave et al. 2009) were identified as well as collections of species traits linked to particular strategies such as shade and drought tolerance and avoidance (Levitt 1972; Abrams 1994; Kitajima 1994; Bloor & Grubb 2003; Poorter
et al. 2010; Wright et al. 2010). Appropriate ecological models and species strategy groups were also identified (Grime 1977; Westoby 1998).
Traits were then assessed based on their potential as good indicators, their ease and practicality of collection and their unique value. I.e. whether another, more easily collected trait may be an appropriate proxy for them. Where data required field collection, an appropriate method was identified. A literature search was carried out to identify any existing sources of trait data (Table 2.2) for the species of interest.
As part of this process, a full list of species synonyms was created and stored within the database (see database model for details) to allow the easy extraction of
2.3.2. Existing data sources
A number of key data sources were identified (Table 2.2). However, these primary sources did not provide coverage of all traits and species. Where necessary, the dataset was augmented from other existing sources (such as published papers on specific species). Samples from both the CMU Herbarium and directly from the field were also used to fill gaps in the database. Certain traits, such as leaf strength, were entirely unavailable from existing sources; so necessitating both observations and collection in the field (see method sections of chapters 3, 4 and 6 for details).
Table 2.2: Existing primary data sources used as the basis of this project along with a brief description of their content and scope.
Data Source Description of data
FORRU (FORRU,2011) and Chiang Mai University Herbarium Databases
(CMU, 2011)
Contains phenological information, performance data and directly measured fruit and seed trait data. Most other trait data have been collated from existing sources and are therefore duplicates.
Royal Botanic Gardens Kew Seed Information Database
(RoyalPBotanicPGardensPKew 2008)
Contains excellent fruit and seed functional trait data. However, limited species coverage. Valuable indication of traits by genus and family to act as confirmation
The Flora of Thailand
(Flora.of.Thailand.Editorial.Board 1970+)
Only 30 of the 54 species are included within this flora
The Flora of China (Zhengyi, Raven & Deyuan 1994+)
Whilst species and trait coverage is good, some of the traits differ significantly from those recorded in the Flora of Thailand, suggesting either geographical variation or phenotypic plasticity. Is therefore appropriate mainly as a secondary source.
2.3.3. Functional trait data collation and reorganisation
Traits available, either entirely or partly, from existing sources were collated into a single MS Access database. Due to possible discrepancies between data sources, in all cases, matching records from at least two sources were required. Where only a single dataset was available, different sources were in conflict or data were not available in existing sources, data were collected from the Chiang Mai University herbarium and field observations within Doi Suthep-Pui National Park. Where field observations were made, at least five replicates were collected and compared.
Much of the information from existing data sources was highly descriptive and inappropriate for analysis. For instance, leaf trichomes are linked to drought tolerance and therefore, an important trait to consider in the analysis. However, there are many descriptive terms for the quantity, colour, shape, amount and duration of trichomes on a leaf. Using a full description of leaf hairiness resulted in as many categories as there were species in the dataset. However, identifying leaves as either “hairy” or “non-hairy” grouped clearly “very different” leaves into the same category. Quantifying data of this type was a vital stage prior to analysis. However, it was important to ensure that important information was not lost. The balance between retaining too much irrelevant data and discarding potentially important information was difficult to maintain but was assisted by extensive consideration of the role and function of specific traits when considering their categorisation.
Where possible, a trait was expressed as an objective numerical value, e.g. leaf shape was broadly expressed through the ratio of leaf length to leaf width. Where potential trait relationships had been identified in the literature and were
appropriate, the established grouping used within that originating study was used. Natural Groupings were used to assign different species to categories. Functional categories were assigned to traits with the aim of having no more than ten categories for each trait, with no category containing fewer than two species. Where there was no natural grouping or where it was unclear where category distinctions should fall, multiple categorisations of the same data may have been used. This was particularly true of data that appeared hierarchical in nature. For instance, species such as Melia toosendem could be described as having either compound or pinnate leaves. Both a two category system (simple and compound) and a 5 category system (Simple-unlobed, Simple-lobed, Trifoliate, pinnate, bi- pinnate) system may be appropriate. Therefore, both the detailed and grouped term were used in the analysis but care taken that these were not mistaken for independent variables.
2.3.4. The Database
The database and the functional relationships between the tables are described in Fig. 2.3. Two versions of this database are available, the first containing only collated and cleaned data (Electronic Appendix A), and the other containing raw data prior to cleaning. A general description of the database model is provided below;
The Species table contains the name and family of every species within the database. The primary key of this table (Species_ID) is the primary key of all data tables.
Reference tables: Reference tables provide lists of terms referenced in other tables (such as list of habitat types). In addition to the lookup tables, there are two other references tables (Data Sources and Measurements) which provide useful metadata and are not included in the referential database structure.
Species Trait Data: Data collated from a number of sources, such as flora,
herbarium, field observation etc. as well as collected in the field. Includes derived fields that are either calculated from other fields (e.g. SLA is leaf area over leaf mass) or fields where there has been grouping or reallocation of traits. All tables have been summarised and contain a single row for each species with species code as a primary key. For ease of access, data have been divided into a number of tables relating to a specific part or aspect of the tree, such as seed traits; leaf traits etc. Performance data: Data from these tables has been primarily extracted from the FORRU database (FORRU, 2011) and contains germination, nursery and field performance data.
Figure 2:3: Project database structure showing all 27 tables, with primary and secondary keys. The SpecIes_ID key , as provided in the Species table is used as a secondary key for 21 tables
containing trait, phenology, distribution and performance data. Additionally, there are lookups for the habitat and distribution tables and two tables containing metadata on data sources and calculations used to derive measurements used elsewhere.
3. Co-ordination and trade-offs of leaf traits in seasonally-dry tropical forest