научная статья по теме SPATIAL AUTOCORRELATION IN TWO IRIS PUMILA POPULATIONS ESTIMATED ON MORPHOLOGICAL DATA FROM NATURAL CLONES AND THEIR SAMPLES GROWN IN TWO DIFFERENT HABITATS Биология

Текст научной статьи на тему «SPATIAL AUTOCORRELATION IN TWO IRIS PUMILA POPULATIONS ESTIMATED ON MORPHOLOGICAL DATA FROM NATURAL CLONES AND THEIR SAMPLES GROWN IN TWO DIFFERENT HABITATS»

ГЕНЕТИКА, 2006, том 42, № 2, с. 282-285

КРАТКИЕ СООБЩЕНИЯ

УДК 575.17

SPATIAL AUTOCORRELATION IN TWO Iris pumila POPULATIONS ESTIMATED ON MORPHOLOGICAL DATA FROM NATURAL CLONES AND THEIR SAMPLES GROWN IN TWO DIFFERENT HABITATS

© 2006 r. A. Tarasjev1, N. BarisiC KlisariC\ B. StojkoviC2

1 Department of Evolutionary Biology, Institute for Biological Research, University of Belgrade, Belgrade 11000,

Serbia and Montenegro; e-mail: tarasjev@ibiss.bg.ac.yu 2 Department of Genetics and Evolution, Faculty of Biology, University of Belgrade

Received August 15, 2005

Morphological data from two Iris pumila populations (measured on native clones, on their replants into the same habitat, and on their transplants into alternative habitat) were combined with native clones spatial position and spatial autocorrelations (SA) were calculated. Naturally growing I. pumila clones revealed significant SA that were positive on small distances and negative on medium ones in both open Hillock and shaded Woodland populations. No significant SA were detected when calculated with original clone positions, but with morpho-metric data from replants into the experimental plot in the same habitat. Some significant SA were, however, detected when morphometric data from transplants to alternative habitat were used. Detected SA on I. pumila clones were primarilly a consequence of spatial structuring of environmental factors but also, in lesser degree, a result of genetic spatial arrangements (most probably due to patterns of gene flow).

Spatial and geographic aspects of ecological and genetic variation have been traditionally of great interest to evolutionary biology [1, 2]. Spatial variation has been examined on various scales - from clinal variation on geographic scale [3] to variation within few meters in plant populations under high selection pressures [4]. Gene flow and spatially varying natural selection are genetic processes that can produce detectable spatial patterns [5], but phenotypic spatial patterns also may result from the influence of some external environmental variables exibiting spatial structure [6]. While there are methods that can describe amounts of spatial genetic variability (like Sewall Wright's F statistic), actual spatial patterns of that variation can be examined by spatial autocorrelation analysis [2] - method pioneered in ecology and genetics mostly by works of Robert Sokal and his collegeaues [3, 7, 8]. When values of a variable are not independent of the values of the same variable in nearby sites, spatial autocorrelation is said to occur. This means that values for such variable in pairs of locations a certain distance apart are more similar/disimilar than expected for randomly chosen pair of obsevations [9]. Those analyses have high statistical power even when sample sizes are quite small, as long as spatial scale for distance class is smaller than the scale on which spatial change occur [2]. Spatial autocorrelation analysis was applied to both genetic and environmental variability on the largest geographic scales, but it is also applied at within-population level. Within-population studies mostly included spatial analyses on genetic molecular markers [10] and analyses of spatial patterns in environmental variables (including utilization of transplated genotypes as replicated phy-

tometers [11]). Spatial genetic analysis of traits other than molecular markers on within population level are hovewer still sporadic (but see studies of phenology [6] and flower color polymorphism [12]) since for these analyses on phenotypic traits it is necessary to account for both genetic and environmental sources of their variation.

In this work we utilized three-year morphological data from I. pumila native clones from two populations (inhabiting open Hillock and shaded Woodland habitats) as well as morphological data from their replants (to the same habitat) and transplants (into the contrasting habitat), but with distance classes always based on clones original spatial position in their natural habitat. In that way we were able to account for both genetic and environmetal sources of spatial autocorrelations and to adress following questions: Is there any spatial pattern in morphological variation within I. pumila populations? If there is spatial pattern, is it primarilly a consequence of population genetic substructuring or of spatial structure of environmental variability? If there is genetic substructuring, is it more influenced by spatial structure of selection regimes in habitat, or by patterns of gene flow?

Utilized morphometric data were obtained from transplant/replant experiment for which comprehensive decsription of analyzed habitats (open Hillock and shaded Woodland), experimental design and data collection is given in Tarasjev [13, 14]. During that study 60 naturally growing clones were marked with plastic peg and spatial position of every utilized clone was determined. In this work we utilized constructed site maps to calculate distances between clones and also

0.3 0.1 -0.1

-0.3 -0.5 0.3 ^ 0.1 ¡-0.1 -0.3

-0.5 0.3

0.1 -0.1 -0.3 -0.5

Native

Replant

— 1992

— 1993 --- 1994

Transplant

• p < 0.01 p < 0.05

15 25 35 45 Distance, m

Fig. 1. Moran's I correlograms based on morphometric data from I. pumila clones from Hillock population. Data was collected on naturally growing clones (upper graph), on replants to the same open habitat (middle graph) and on transplants to contrasting shaded habitat (lower graph) in three years. Distance classes were constructed on the basis of clone original positions.

^ 0.1

i-0.1

o

S -0.3 -0.5 -0.7 0.5 0.3 0.1 -0.1 -0.3 -0.5 -0.7

- 1992

--1993

Replant — 1994

\

Transplant

• P < 0.01 ° P < 0.05

5

20

35 50 Distance, m

65

Fig. 2. Moran's I correlograms based on morphometric data from I. pumila clones from Woodland population. Data was collected on naturally growing clones (upper graph), on replants to the same shaded habitat (middle graph) and on transplants to contrasting open habitat (lower graph) in three years. Distance classes were constructed on the basis of clone original positions.

added morphometric data measured directly on naturally growing donor clones. Following morphological traits were measured in all samples: leaf length, leaf width, stem length, length of the first spathe, length of the second spathe, ovary length, ovary width, tube length, tube radius, fall length, fall width, beard length, standard length, standard width, stigma length, crest length, crest width, stamen length, anther length, number of leaves, vegetative weight, and flower weight. Principal component analysis (PCA) was performed on morphological data for all population/habitat combinations as well as on morphological data from naturally growing donor clones. Individual clone scores on First component from PCA (as component explaining most of the morphological variability in a sample and a measure of general size) were utilized for spatial autocorrelation analysis. Due to less succesfull flowering of Woodland replants in 1992 [14] and consequently fewer data available, that data set was ommited from further analysis. We followed recommendation of Du-tilleul and Legendre [15] and used equal frequency distance classes since they give even precision in

estimation and testing. Distance classes determined from spatial position of naturaly growing clones were always used in all analyses. They were 5 equal frequency classes formed separately for each of two analyzed populations. Moran's I spatial autocorrelation coefficient was then calculated by SAAP computer program [16]. Values of Moran's I under isolation by distance processes are very well characterized through a series of simulation studies showing that the values of correlations stabilize fastest for short distances and are highly statistically powerful [2, 17]. Moran's I coefficients were presented in a form of correlograms-diagrams showing coefficient as a function of distance between pair of clones [7].

Spatial autocorrelations estimated on morphometric data from naturally growing I. pumila clones were significant and positive on small distances and negative on medium ones in both analyzed populations (Figures 1 and 2, upper graphs). They can in principle be a result of both genetic similarities between neighbours as well as of similarities in environmental conditions that neighbours are experiencing. Spatial autocorrelations

rEHETHKA tom 42 < 2 2006

284

TARASJEV ë ap.

between I. pumila clones were also calculated with original clone positions but with morphometric data from replants of those clones to the experimental plot in the same habitat, as well as with morphometric data from transplants of those clones to the experimental plot in the alternative habitat. Environmental spatial pattern cannot contribute to autocorrelations estimated in this way since samples were not grown in their original positions but were randomly assigned to position in experimental plot, so any significant correlation could be a result of genetic structuring only. However, no significant spatial autocorrelations on replant data were detected (Figures 1 and 2, middle graphs). Therefore, this result would suggest that almost all spatial structure detected on morphometric data from naturally growing clones was a consequence of spatial patterns in environmental variables only. However, this picture changes when correlations were calculated with classes determined on the basis of original clone positions but with morphometric data from transplants to alternative habitat (Figures 1 and 2, lower graphs). Here, some significant negative autocorrelations emerged in middle distance classes. Their appearance in transplant but not in replant analysis indicate greater expression of genetic differences between clones grown in less common environment - situation that can be described as release of so called hidden variance [18] that is unavailable to s

Для дальнейшего прочтения статьи необходимо приобрести полный текст. Статьи высылаются в формате PDF на указанную при оплате почту. Время доставки составляет менее 10 минут. Стоимость одной статьи — 150 рублей.

Показать целиком