David Boshier1
1. Introduction
Sound decisions in genetic conservation and progress in long term breeding require detailed knowledge of the taxonomy, population structure, reproductive biology, mating systems, etc of the species concerned. For many tropical forest tree species, little information is available and managers are faced with problems in developing effective strategies. For example, questions that need to be answered are:- Are neighbouring trees in natural stands highly related and is there inbreeding? What, if any, is the incompatibility mechanism to stop inbreeding? What agents effect pollination and how do they affect gene flow and population size? What degree of inbreeding is acceptable?
The case study presented here describes the methodology used in ongoing research carried out for one particular species - Cordia alliodora. It is not intended that the methodology described here be used as a recipe for studying any tropical tree species, but that it may provide ideas on how to approach the problem. Techniques used will always depend on the type of tree, its distribution, size of flowers, pollinators etc. Previous work with C. alliodora, relevant to the study, is generally of a preliminary nature but provides useful information.
C. alliodora is an important neotropical tree, which combines high quality timber and value with fast growth on good quality soils. It is used extensively throughout its natural range and its light crown and self pruning habit make it particularly suitable for use in various agroforestry systems, providing valuable timber and income to small farmers. The species has a wide distribution occurring from northern Mexico through Central and South America as far south as Bolivia, Paraguay, southern Brazil and northern Argentina. In Paraguay, northern Argentina and parts of southern Brazil the closely related C. trichotoma (also classified as C. alliodora var tomentosa) predominates over C. alliodora (Gibbs and Taroda 1983). C. alliodora is found on most of the Caribbean Islands from Cuba to Trinidad, but is almost certainly not native to Jamaica. Through this geographical range the species occurs under a wide variety of ecological conditions, varying from very wet (as much as 6 000 mm per year) to seasonally dry (as low as 600 mm per year); and from sea level to as high as 1 200 m a.s.l. in Central America and 2 000 m a.s.l. at lower latitudes in Colombia. In lowland humid tropical regions it is generally a tall, thin, lightly crowned tree, reaching heights over 40 m and d.b.h. up to 1 m in mature trees, although in the region of 50 cm d.b.h. is more normal. In seasonally dry areas it is smaller, and more poorly formed rarely reaching more than 20 m in height.
The overall objective of the present study was to gain an understanding of the reproductive biology of C. alliodora that will give adequate knowledge to make sound long term decisions for in-situ/ex-situ conservation and breeding of the species.
1 Oxford Forestry Institute, South Parks Road, Oxford, OX1 3RB, U.K.
The specific objectives were to:
2. Flowering and Seed Phenology
Flowers are hermaphrodite, unspecialized, about 1 cm in length and occur in large panicles. The size of panicles varies, with as few as 50 flowers to as many as 2 000. The petals are white and persistent turning to brown and acting as a parachute in wind dispersal of the seed. The aim of the first phase of the phenology study was to detect any patterns of flowering that may exist within the tree, branch and inflorescence and provide data on length of the flowering and fruiting periods and the proportions of seed reaching maturation. It is important to know how a tree flowers and what effect that might have on pollinator movements. For example: sequential flower opening from the top to the bottom of the crown may mean that early flowers are pollinated by one group of pollinators and later flowering by a completely different group of pollinators. Conversely, random flowering within the crown could stimulate greater movement of pollinators between panicles and different trees.
2.1 Within inflorescences: individual groups of flowers were identified and the stage of flowering observed every three days from flowering to seed fall. Classification of the stages of flowering was based on Mendoza (1965), distinguishing:
1 - start of formation of flower buds
2 - individual flower buds visible, but not open
3 - individual flower buds full
4 - white petals emerge from flower buds, but not open
5a- petals open, stigma receptive
5b- petals still white, stigma withered
6 - petals turn brown
7 - embryo starts to swell.
It was thus possible to observe at which stage of flowering most losses occurred, the length of each stage, any sequential pattern of flowering or losses within a panicle, and the effect certain weather conditions might have on flowering and fruiting.
2.2 Within branches and trees: individual inflorescences/branches within a number of branches/trees were observed periodically from flowering to seed production. The same flowering categories were used as in 2.1 (categories 5a and 5b combined) and each panicle/branch classified on the predominant category of flowers.
2.3 Within populations: flowering phenology within a population of trees of the same species is of fundamental importance to understanding gene flow, genetic structure, seed production and yet few studies exist of individual tropical tree species based on large samples. Similarly few studies have looked at the flowering of the same population for more than one year. It is important to know about periodicity of flowering for the species and how individual species flower with respect to one another from year to year. For example, because of asynchrony of flowering two adjacent trees may be unable to mate, which may be the case in one year but not in the following. Similarly some trees may be heavy flowerers in one year and totally sterile the following year.
To look at flowering within natural populations, in a 26 hectare plot all trees (216) of a flowering age/size surrounding three C. alliodora selected trees up to a radius of approximately 500 meters were mapped, and labelled. The stand, which was also to be used for the mating system and gene flow study (see 5), was chosen taking into account various factors: (a) the presence of selected plus trees would provide direct information on the degree of diversity sampled when collecting open-pollinated seed from the plus trees, which are part of a breeding population (Boshier and Mesen, 1989); (b) the possibility of easily defining a population, there being a natural break between the trees under study and nearest trees of the same species; (c) ease of access to allow frequent visits.
The crown of each tree was observed every three days to determine the start, peak and end of flowering and the percentage of the total number of flowers open by that date. Subjective assessments of flower (scale of 0–5) and seed production (scale of 0–3) were also made. The same observations were made over three successive flowering seasons (Jan-April) to study year to year variation. An index for synchronicity of flowering was calculated, for the whole stand as well as for the plus trees in relation to the surrounding trees (Augsberger, 1983). To look at the effect population substructure may have on flowering synchrony the individual tree and stand index was re-calculated on the basis of increasing population size from a base tree (a plus tree). Two other measures of population synchrony were made: (1) synchrony of the first day of flowering, calculated as one standard deviation around the mean of the first day of flowering; and (2) synchrony of the median day of flowering, calculated as one standard deviation around the mean of the median flowering day. For one flowering season similar phenological observations were carried out in a stand in the seasonally dry Pacific region to allow comparison of flowering under differing climates.
2.4 Between populations
Much information on the general phenology of a species can be gained from the examination of herbarium specimens. Even for species for which full collections are not available, studying existing specimens can reveal much about phenology, distribution and likely regions where the species may occur. Herbarium collections of C. alliodora and C. trichotoma were studied and full details noted form specimen labels as well as the stage of flowering. It was possible to classify well-preserved specimens under one of the flowering categories used in 2.2. Logging of the information onto the OFI herbarium database, BRAHMS (Filer, 1991) allows mapping of the distribution of the species as well as studying gross variation in flowering time over this range.
3. Breeding System
Many species of Cordia are heterostylous, indeed in one case failure to recognize this originally led to one species being classified as two (C. thaisiana and C. apurensis, Agostini, 1983). Gibbs and Taroda (1983) studied the C. alliodora - C. trichotoma complex based on herbarium specimens from South America and distinguished the two species on the basis of the pattern of heterostyly. In C. alliodora heterostyly appears to have broken down although there is variation in style length. Opler et al (1975) studied the pollination biology of a number of species of Cordia and found varying degrees of self-incompatibility in the trees of C. alliodora they sampled.
A combination of field and microscope work were used to study stylar patterns, both within and between families and their relationship to the incompatibility mechanism. On freshly cut inflorescences from an open pollinated progeny trial, flowers were emasculated as they opened and before anther dehiscence had occurred. Controlled cross, self and related pollinations between half-sibs were carried out and the flowers fixed at varying times after pollination to allow study of pollen tube development on the stigma and through the style.
The same controlled crosses were carried out in the field to allow comparison with the results from the microscope study, checking that crosses deemed compatible produce seed and those deemed incompatible failed to do so. The flowers on any one panicle open during a period of 4–6 days. The numbers of seed produced per cross and empty flowers were counted to compare fertility from related and unrelated crosses. The seed was used to mount a nursery trial to look at the effects of inbreeding on initial growth rate. Controlled pollinations were also carried out in the field to look at length of stigma receptivity, and to exclude any possibility of parthenocarpy. It was noticeable that in the dry zone the stigma was receptive for only one day at the end of which it was withered whereas in the wet zone the stigma could be receptive for as long as three days, emphasising the importance of caution in extrapolating results from one region to a climatically different one.
4 Pollinators
Opler et al (1975) looked at potential pollinators and concluded that noctuid and geometrid moths and anthophorid bees were the most likely pollinators. Their work was however restricted to the dry Guanacaste region of Costa Rica and it is likely that in the wet lowland regions other pollinators are of importance. As many insects are seasonal, it is likely that different insects act as pollinators at different times during a prolonged flowering season. Collections were made of insects visiting flowers during the day and night at various times in the flowering season. The insects were mounted, the identified to genus and stored for future use as a reference collection. Observations were made on several occasions of numbers and types of insects visiting flowers, to see which were the more important ones. Other studies such as those on the effectiveness of different pollinators in depositing pollen on the stigma, or marking and recapturing to look at fidelity to food source could also be carried out (Frankie et al 1976).
Flowers were collected and fixed at different times of the day following anthesis (dawn, midday, dusk) from a number of trees. Microscopic observation of these flowers for the quantity of pollen on stigma and pollen tube growth can provide supportive evidence on the time of pollination and the relative importance of certain pollinators. Similar collections were made in the Pacific region to compare pollination for the species in differing climates.
It is important to study variation of nectar flow with time as this can give clues as to which vectors are active pollination and when. A number of trees should be studied as there can be variation from tree to tree both in time of production, quantity, and quality (sugar concentration). Even within an inflorescence patterns of nectar production may exist which will tend to move pollinators in certain directions. To characterize the availability of rewards to pollinators, variation in nectar production was studied on a number of trees at intervals during a three day period after flower opening. The process was repeated on a number of occasions during the flowering season.
5. Mating System, Gene Flow and Neighbourhood Area
Isozymes may be used a genetic markers to provide estimates of the mating system, gene flow, neighbourhood area and paternity. The interpretation of such data is made more significant when it can be combined with field observations of the spatial distribution of the trees and flowering phenology. For this reason the same population was chosen for studying mating system as for the within population phenology study. Seed was collected during the 1989 season from the trees surrounding two plus trees; during 1990 and 1991 seed was re-collected from the plus trees to look at year to year variation in outcrossing rates and paternity for these two trees. Although sufficient seed could be collected from one panicle, seed was collected from the whole crown, to ensure that the sample was not biased due to particular pollination events. To look at variation of pollination within a tree, at the time of collection, seed from one plus tree was divided into three lots representing the upper, middle and lower parts of the crown. Seed from three panicles on the same branch on the plus tree was also separated as individual lots.
A total of 163 trees occurred within a 250 m radius of the plus trees, a number prohibitively large to allow all to be studied for isozymes. A sub-sample of 52 trees was therefore selected on the basis of distance from the plus tree and phenological data obtained during field work (see 2.3). The following trees were eliminated from the sample: (a) non-flowerers (36 trees); (b) trees that flowered completely before or after the plus tree (5); (c) trees classified as 1 for quantity of flowering (i.e. six or less panicles) (12); (d) trees in which less than six days synchrony in flowering was observed (13); (e) trees with scarce flowering (category 2) (21); (f) trees in category 3 for flowering and greater than 200 m from the plus tree (24). To aid in determining paternity for progeny arrays of specific trees, genotypes were assigned to un-sampled trees by assaying leaf material. The leaf material was collected and immediately frozen in the field in liquid nitrogen.
To complement estimates of pollen flow from the isozyme work and allow estimates of neighbourhood size and area, seed dispersal was studied for four trees in the same population used for phenology and isozymes. A UV fluorescent dye was used to spray seed in the canopy and so enable tracing distance of seed dispersal from the trees. Seed was collected in 1 m2 traps, placed at intervals along transects, laid out in the direction and reverse of the prevailing wind. The dye was still easily visible on the seed after six weeks and probably persists much longer.
6. Variation between Populations
There are two complementary approaches to studying genetic variation between populations. The first, provenance testing, is generally used in forestry and studies the performance under uniform environmental conditions in one or more planting sites of trees grown from seed collected from different populations. Traits studied are usually of a continuously variable nature and the observed variation is split into genetic and environmental components. Provenance variation for Cordia alliodora has been studied in an international trial coordinated by the Oxford Forestry Institute based on collections made in the late seventies principally in Central America, with assistance from FAO and ODA (Stead, 1980). Over a wide range of sites, provenances from the humid Atlantic region of Central America and in particular Honduras and Costa Rica showed superior growth and form to Pacific region provenances, even when grown in seasonally dry areas (McCarter 1988). The study thus suggested ecotypic differentiation of the Pacific and Atlantic populations, but also indicated that within these broad bands variation within provenances was greater than between. There appears to be great potential for genetic improvement through individual tree selection, the provenance assessments showing a 200–300% difference within provenances between the best 10% of trees and the mean.
The second approach is to study genetic diversity shown as variation in enzymes (and more recently DNA) for specific gene loci. The enzyme survey carried out in (5) above showed differences in the number of loci staining; and variability within loci, between the population under study and a Pacific zone population previously studied. Further studies of isoenzyme variation in the original OFI provenance collections of C. alliodora, being carried out at University of Massachusetts, Boston (U.S.A.), indicate interesting differences between provenances.
7. Chromosome Number
More information on variation within and between closely related species can sometimes be revealed by looking at chromosome variation, both in terms of numbers and overall quantities of DNA. The two existing reports of chromosome number in C. alliodora: n=15 (Bawa 1973) and 2n=72 (Britton 1951), differ greatly. The possibility of gene duplication for a number of enzyme systems in the Atlantic population also suggests that some benefit may be gained from looking at chromosome number to see if intra-specific variation does exist in this trait, and if this bears any relation to the Pacific/Atlantic provenance differences encountered in the international provenance trials.
Acknowledgements
The above study was financed by the Overseas Development Administration of the United Kingdom (R4484 & R4724). The author is grateful to the following people; Ing. F. Mesen, CATIE; Ing. F. Lega, Scott Paper, for facilitating the field work in Costa Rica; to Dr. K.S. Bawa and Mr. M. Chase, Univ. Mass., Boston for collaboration with the isozyme work and Dr. G. Frankie, Univ. California, Berkeley, for entomological advice.
