Gene diversity in grevillea populations... GENE DIVERSITY IN GREVILLEA POPULATIONS INTRODUCED IN BRAZIL AND ITS IMPLICATION ON MANAGEMENT OF GENETIC RESOURCES

http://dx.doi.org/10.1590/1806-90882018000200005 ABSTRACT – We describe isoenzymes variability in six populations of Grevillea robusta from a provenances and progenies test established in a randomized block design with five plants per replication in Southern Brazil. The population genetic structure was examined by using biochemical markers in 5-year old trees, specifically at MDH-3, PGM-2, DIA-2, PO-1, PO-2, SOD-1, and SKDH-1 loci. The northern provenances (Rathdowney and Woodenbong) showed a strong divergence in relation to the average of provenances when alleles per locus (Ap), allele richness (Rs), Nei’s gene diversity (H), and inbreeding coefficient (f) were considered. Inbreeding in varying degrees was detected. The commercial control showed the highest inbreeding coefficient, (f = 0.4448), whereas the provenance average was f = 0.2306, possibly due to insufficient sampling of populations in their origin (Australia). Despite its restricted natural range, a positive correlation between genetic divergence and geographic distance among original populations was detected. Genetic distance and cluster analyses based on the Bayesian model revealed three distinct provenance groups: 1) Rathdowney-QLD and Woodenbong-QLD; 2) Paddy’s Flat-NSW; and 3) Mann River-NSW, Boyd River-NSW and the commercial control (material used in Brazil). The grouping of the control to both Mann River-NSW and Boyd River-NSW provenances suggests that the northern provenances have the highest potential for genetic improvement of wood productivity in Brazil, due to their high genetic diversity and low inbreeding coefficient.


INTRODUCTION
Grevillea (Grevillea robusta A. Cunn.) is a tree species of Proteaceae family occurring naturally in Queensland, Australia, in areas extending from northern Gympie (latitude 25 o 50' S) to Guy Fawkes and Orara river areas (30 o 10' S). Orara River is a tributary of Clarence River that runs mostly through New South Wales at altitudes varying from sea level to 1,100 m at Bunya Mountains in Queensland (Harwood et al., 1997). This species occurs in two main habitats: in narrow bands associated with Eucalyptus along Richmond and Clarence rivers, and mixed with Araucaria cunninghamii Aiton ex. Don forests (Webb, 1959).
Grevillea populations are geographically discontinuous and separated by Sclerophyllous forests that are prone to fire, or by climax forests where they struggle to survive forest fires and the disadvantage of their typically small population size (Harwood et al., 1997). In general, there is a small number of effectively reproducing individuals. In Emu Creek, Queensland (28 o 58' S; 152 o 10' E) for example, only 90 trees were breeding in an elongated population stretching 2.4 km next to a stream. Population density of less than one individual per hectare is observed in association with Araucaria cunninghamii forests (Webb et al., 1967). Also, according to a survey by Harwood (1992), grevillea is naturally distributed in a number of small populations distant two to 10 km apart. This situation leads to genetic structures with strong differentiation among and weak within populations (Moran andHoper, 1983, 1992). Natural grevillea stands have been continuously depleted through exploitation for wood (Maiden, 1911) and deforestation for agricultural and pasture lands since the early days of the European colonization. Remnant populations are represented by young dispersed trees in degraded sites as well as in areas where Araucaria forests were removed for agriculture practice.
Grevillea has been introduced into many countries in South America, Africa and South Asia, especially for shade trees in tea and coffee plantations. It is also an important source of firewood and timber in small upland farms in tropical regions of eastern and central Africa (Harwood, 1989). It was first brought to Brazil in the late 19 th Century (Instituto Brasileiro do Café, 1981) and disseminated throughout the southern and southeastern regions where climatic conditions are favorable for the species. Other uses of grevillea in Brazil include wood production for timber (Iwakiri et al., 2004), and as windbreak around croplands and pasture (Baggio, 1983;Durigan and Simões, 1987;Campanha et al., 2007) as well as in urban forestry (Beloto and De Angelis, 2003;Kramer and Krupek, 2012). According to Ribaski (2006), grevillea plays an important role in agroforestry systems. It contributes to improve environmental conditions in the field and provide comfort to animals and better pasture development. Moreover, it produces valuable logs for processing by small landowners (Radomski and Ribaski, 2011). Grevillea is known not to compete for resources with agricultural crops, while maintaining fast growth even in low fertility soils (Pinto Júnior et al., 2013) and is one of most popular tree species planted in pasture lands in Brazil (Almeida, 2011).
Efforts in genetic improvement of grevillea in Brazil started with provenance/progeny tests in the early 1980's. Important variations among provenances were reported by Shimizu et al. (1998) and Martins et al. (2000). Thereafter new provenances were reintroduced in Brazil and institutions as Instituto Florestal de São Paulo (IF-SP), Embrapa Florestas (CNPF) and Instituto Agronômico do Paraná (IAPAR) which demonstrated it potential Paraná, São Paulo and Mato Grosso do Sul States. Differences between them have been reported (Sebbenn et al., 1999;Martins et al., 2005;Martins et al., 2006;Leal and Ramos, 2011).
Population genetics studies are important elements in supporting genetic improvement and conservation programs, especially in the case of grevillea. These are vital in understanding the genetic structure in the remnant populations in Australia as well as in populations introduced into different countries. Studies have revealed high outcrossing rates (0.92; Hadwood et al., 1990) and evidence of self-incompatible breeding system in natural populations (Hadwood, 1992) and also in Kenya (Kalinganire et al., 1996(Kalinganire et al., , 2000.
So far, no study on population genetics had been carried out in grevillea populations introduced into Brazil. The present work addresses this issue in order to generate elements in support to the species breeding and conservation programs.

Populations and sampling
The study was based on a five-year old combined provenance and progeny trial involving Australian sources (Table 1) and a control from commercial populations disseminated in Brazil, presumably of NWS origin. The trial was established in Ponta Grossa, State of Paraná, Brazil (latitude 25° 05' 42" S, longitude 50°0 9' 43" W, and elevation 900 m above sea level). The local climate is classified according Koeppen's system as Cfb (mild summers). Average summer temperatures remain below 22 °C and winters are mild, although severe frosts can sometimes occur. Annual rainfall exceeds 1,500 mm and no soil water deficit is observed. The local soil is classified as Cambissol (equivalent to Inceptisol) and the relief varies from flat to a slightly rolling topography.
The trial was established in a randomized blocks design with seven replications of 5-plant plots. The treatments were represented by five Australian provenances and one local commercial control of unknown original provenance (Table 1).
Natural populations were sampled in areas extending from latitudes 28°13' 00"S to 29° 53 00"S, longitudes 152°26'00"E to 152°52'01"E, and altitudes 15 m to 100 m above sea level. The maximum distance (based on Ersts, 2017) among sampled populations was 171.21 km between Rathdowney and Boyd River, and the minimum distance of 25.08 km between Rathdowney and Woodenbong.

Isoenzyme electrophoresis
Isoenzyme technique played an important hole in characterizing plants, in general, and also forest species studies since the years 1970 have been largely replaced by molecular techniques, they are still informative and maintain their applicability, especially in studies of population genetics. Nowadays articles using isoenzymes as main technique (Litkowiec et al., 2013;Mathiasen and Premoli, 2013;Harter et al.;Medoukali et al., 2015;Chlumský et al., 2016;Kutlunina et al., 2016;Luo et al., 2016;El-Esawi et al., 2017;Konzen and Martins, 2017) as well in association with molecular markers (Streiff et al., 1998;De Woody et al., 2009;Ekart et al., 2016;Reichmann et al., 2017;Quipildor et al., 2018;) are still published worldwide on population genetic studies. Although molecular markers are more informative and currently used, in some cases isozymes is still helpful to screen population and individuals  (Nei, 1987); f = inbreeding coefficient; F ST = fixation index (Wright, 1965).
generating results to support decisions on genetic improvement and conservation.
Genetic structure of populations was performed by using Admixture analysis using Bayesian assignment procedures (Structure 2.2.3 (Pritchard et al., 2000) using 500.000 burn-in periods and Evanno method (Evanno, 2005) implemented by Earl (2012), a website program, for estimating K partitions (population groups).

Genetic diversity
Genetic diversity parameters obtained from isoenzymes analysis are summarized in Table 2. Measures of genetic variation within populations (Table 2) indicated an average of 2.93 alleles per locus (Ap) across loci. Rathdowney and Woodenbong populations showed higher numbers of alleles per locus (3.14 and 3.00, respectively) than the average of other populations (Ap = 2.86). These provenances are geographically closer than the distances between other provenances.
The expected heterozygosity (H e ) varied from 0.4849 to 0.5441 with an average of 0.5140. They were higher from those reported by Harwood et al. (1997)  Private alleles (number four) were found both in the commercial control at locus MHD3 and in Rathdowney at locus PO1, suggesting that the local population contains other than the Australian northernmost provenances.
By comparing Rathdowney, the northernmost provenance, with Woodenbong the southernmost provenance, it becomes evident that they represent two distinct groups, as reported by Harwood et al. (1997). The local commercial control showed a high inbreeding level. Therefore, in spite of their common presence in small farms in Brazil, they must be used with caution, due to inbreeding depression.

Genetic structure and variation among populations
Two most distinct groups are shown in Figure  1 by using Nei's (1978) genetic distance. The first includes populations from the northern region of the species range (Rathdowney and Woodenbong); the second is formed by all other Australian provenances. Within the second group, Paddy's Flat differs from Mann River and Boyd River. Also, a clear difference was observed between the second group and the commercial control. In general the grouping of provenances followed a pattern close to their geographical distribution. This evidence is confirmed in Figure 2. There was an indication of a close association (Rousset's distance) between the population variation and their geographic distances (r 2 = 0.8316).

Admixture analysis
Three groups were shown by K value. These groups are represented in the admixture analysis (Figure 3). Populations from the northern distribution range (Rathdowney, Woodenbong and Paddy´s Flat) showed similar mixture patterns. Within the group including Mann River and Boyd River also similar admixture patterns were observed. The commercial control showed a high degree of admixture, in similar pattern to that observed for the southern provenances.

DISCUSSION
The present study demonstrated clear variations among provenances. Also, correlation of provenance variation with their populations' geographic distribution became clear. The evidence of a high inbreeding rate and other aspects in variation patterns in the local commercial control raises concerns about continued use of this material. The main issue seems to be the restricted genetic variation in this material because of the small number of trees initially sampled for introduction into Brazil (Shimizu et al., 1998). This fact, over the years, has promoted damaging effects caused by inbreeding, such as the reduction of germination rate and loss of vigor and malformation of planted trees (Ferreira and Martins, 1998). This explains the higher inbreeding coefficient observed for ancient Grevillea germplasm introduction (in the mid-19th century) in relation to new populations introduced after 1990 (late 20th century).
For this reason, Embrapa Forest, in partnership with CSIRO, from Australia, introduced a new Australian germplasm base considering a better sampling including more provenances and individuals within provenances. This material were established as several trials and has been showing a higher potential for wood production, with genetic gains from 23 % to 33 % compared to original (commercial) introduced material .
SOUSA VA et al. et al., 2001). It was still observed by Kalinganire et al., (2001) a non-efficient seed production derivate from absence of a higher pollinator activity in Kenya. Thus a desirable seed production depends not only from environmental suitability to produce flowers but also from adequate and abundant pollinators to reach a higher seed production. In absence of specific pollinators in adequate amounts seed production can be increased through controlled pollination. Kalinganire et al. (2001) has observed experimentally, in Kenya, that controlled pollination produced more seeds than open pollination. Therefore in Brazil we have to be aware about pollinators issue and in some cases to consider controlled pollination as a tool to increase seed production aiming the advance in grevillea breeding programs.
Data from a provenance tests carried out in northern State of Paraná and elsewhere in Brazil (Sebbenn et al., 1999;Martins et al., 2006;Leal and Ramos, 2011) have demonstrated higher performances of recently introduced provenances in comparison to previously introduced material.
The pioneer work on population genetics in grevillea as described by Harwood et al. (1997) offered the basic elements to establish genetically sound breeding programs for a more efficient use and conservation of this valuable species. The multitude of knowledge that is being gradually amassed on the species breeding system, geographic variation, genetic parameters, silvicultural requirements, wood quality, and its role in agroforestry systems is a key to a better use and efficient conservation measures for this species.

CONCLUSIONS
The information gathered on different aspects of population genetics in grevillea introduced into Brazil represents an asset of vital importance for the best use of this forest resource. Pioneer introductions made in the early 20 th century raised interest in this species for its value as shade tree and wind break around agricultural crops. However, low wood production and bad stem form discouraged any other form of utilization. Genetic studies have demonstrated that undesirable performances were basically due to inbreeding depression because of a very narrow genetic base.
Further introductions of well sampled populations from the northern areas of the species range are seen as means to place this species in focus as a multiple use alternative, especially in small farms. If ample genetic variation is preserved, there will always be opportunities for genetic improvement toward different end uses and suited for production in different ecologic sites.

ACKNOWLEDGMENT
We thank to Mrs. Simone Aparecida Sopchaki the valuable collaboration in populations' genotyping.