Changes in Seed Quality during Seed Development and Maturation of Bambara Nut ( Vigna subterranea (L.) Verdc.) Landraces

,


Introduction
Bambara nut (BN) (Vigna subterranea (L.) Verdc.) is an important grain legume grown mostly by subsistence farmers in the dry land regions of sub-Saharan Africa (SSA) [1].In many countries where BN is cultivated including Uganda, it is considered a "minor" crop and as a result often neglected and underutilised, despite its huge potential for food and nutrition security [2][3][4].Consequently, the majority of the resourceconstrained BN farmers rely on traditional knowledge to carry out crop production and seed management ( [2,5];) and use their own seeds saved from previous harvests, seeds obtained from neighbors, or in the local market to sustain production [4][5][6].An important aspect of seed management is the timing of harvest.It is critical to the seed quality of most species including BN, with too early or delayed harvest seriously compromising seed quality [7][8][9].For most BN farmers in Uganda, timing of seed harvest is tied to such physical indicators as yellowing and browning of leaves, drying of leaves and stems, and characteristic colour development ( [5,10]).While this can be a good guide for timing of harvest, soil and environmental conditions can potentially alter these indicators [11,12], thereby compromising their reliability.A robust method based on physiological developmental indicators during seed development and maturation would provide a more stable and reliable method of timing seed harvest.Te process of seed development and maturation of seeds such as BN begins with ovule fertilisation and typically follows three stages.Histodiferentiation is a process in which ordered cell division and diferentiation results in formation of the embryo, seed flling involving cell expansion and storage reserves accumulation, and maturation drying marked by seed moisture loss and reduced metabolism occurs [13][14][15].Seed flling ceases at physiological maturity, with subsequent maturation drying [14].Te length of each of the phases usually varies from several days to many months depending on the species and prevailing environmental conditions [14].Te whole process is under genetic control of quite complex and distinct biosynthetic pathways and physiological processes ( [13,14,16]).Seed development and maturation in relation to seed quality of BN has only previously been partially described for any landrace.Embryo development in a bambara nut landrace was observed to proceed up to at least 41 days after anthesis and was photoperiod sensitive whereby ≥14 hours photoperiod terminated growth altogether [17].Likewise, grain flling was delayed and reduced at this photoperiod [17].A report by Linnemann [18] had determined the mass maturity of a BN landrace named "Tiga Nicuru" from Mali to be approximately 103 days after sowing (DAS) under a 12 hr photoperiod regime.More recently, Mandizvo and Odindo [19] estimated that BN landraces from a feld experiment in South Africa reached physiological maturity at about 65 days after fowering (approximately 100 DAS).In Uganda, however, the seed developmental processes of the country's landraces and their physiological and harvest maturity remain to be understood.A better understanding of the seed development and maturation process at critical stages could provide a scientifc basis for the timing of harvest for landraces and therefore "fne tune" the harvest stages of BN landraces in Uganda.Tis will result in better quality seeds and higher yields for farmers, hence is the focus of this study.

Field Experimental Sites.
Te feld experiments were set up at two on-farm sites in Ocettoke, Kitgum district (2020/2021), and Koro, Omoro district (2021/2022), both in the northern region of Uganda.Both Ocettoke (03 °22′33.071″N/032 °51′55.69″E,1020 masl) and Koro (02 °39′22.92N/032 °18′44.27E,1085 masl) are found within the warm subhumid zone of the northern moist farmlands agroecological zone (AEZ) [20].Tis AEZ has a unimodal rainfall pattern stretching from April to October (average annual temperature is >20 °C and annual rainfall >1200 mm) and a long dry season from November to March [20,21].Te landscape is mostly of gentle rolling plains punctuated with narrow valleys, while the soils are generally sandy and sandy clay, having low organic matter and low nutrient availability [20].Dryland food crops such as pigeon pea, cowpea, millet, sorghum, and cash crops such as cotton, sesame, and tobacco are the most predominantly cultivated in this AEZ [20].Te zone also covers one of the regions in Uganda where bambara nut is commonly grown [5].

Plant Material.
Bambara nut seeds of two landraces (LocalBam and TVSU544) were obtained from the market in Arua (north western region), while a third landrace (Abi-Bam001) was bought in Lira market (northern region) (Figure 1).Seeds for subsequent plantings were obtained from the previous experiments.Landraces were selected on the basis of their seed coat colour (dark, intermediate, and light), seed size (small, medium, and large), and frequency of occurrence ("preference") among the BN farming communities.TVSU544 has medium-sized cream seeds with "black eyes" and is of light colour (Figure 1).AbiBam001 has small round seeds that are cream with black stripes and of intermediate or dark colour.LocalBam has large oblong mottled seeds with brown or purplish specs and of dark colour (Figure 1).Both AbiBam001 and LocalBam were encountered at several locations during a baseline survey, while TVSU544 was only found at a single location [10].

Field Experimental Setup and Sampling.
Te three bambara nut landraces were planted in randomised complete blocks and replicated three times at each site.Te Kitgum experiment was established on 25 th October, 2020, while the Omoro experiment was achieved on 19 th October, 2021.Standard crop management practices were applied at all stages of the crop growth.Te crop was irrigated twice a week during periods of extended dry spells, especially in December and January.Pods were harvested from randomly selected rows in each plot on 16 th and 26 th January and 5 th , 15 th , and 25 th February, 2021, (at the Ocettoke site) and on 19 th and 26 th February and 6 th March, 2022 (at the Koro site).Tese predetermined days corresponded to 83, 93, 103, 113, and 123 days after sowing (DAS) for Ocettoke and 123, 130, and 138 DAS for Koro.Pods were sun-dried and temporarily kept at room temperature and then later transferred to a −20 °C freezer until use.Pods were hand threshed and used for seed quality analyses.Daily weather data (temperature, RH) were recorded with a Tinytag ® data logger (Gemini Data Loggers, UK) suitably located in the feld.Soil samples were collected from each experimental site and analysed for pH, organic carbon, total N, exchangeable K, available P, Ca, Mg, Na, and S.

Seed Size Measurements.
Seeds from each replicate were visually size graded as small, medium, and large and a proportionate number of each size grade was included in subsequent measurements.Te seed length and width were measured with a digital Vernier calliper.Ten seeds were representatively selected from each replicate and measured.Seed dry weight was measured with a weighing balance.Twenty separately drawn seeds from each replicate (40 per replicate) were weighed, and an average of the 2 measurements was recorded as 20 seeds weight.Te result of which was expressed as 100 seed weight by multiplying by 100.® (Biobase Biodustry, Shandong, China) was used for the second experiment.Light in both cases was provided by fuorescent bulbs inside the incubation chambers.Te sand medium was monitored daily and hydrated accordingly to keep it moist.Te number of seeds germinated was recorded on a daily basis for 16 days.A seed was considered germinated when the plumule had emerged from the sand surface.At 16 days, seedlings were evaluated and the percentage of normal and abnormal seedlings was determined.Five normal seedlings from each tray (each replication) were air-oven dried at 95 °C for 24 hr and weighed to determine seedling dry weight (SDW).Te fnal germination percentage (FGP) and germination velocity index (GVI) were calculated as follows.Te fnal germination percentage (FGP) was calculated using the following formula:

Germination
where N g is the number of germinated seeds and N T is the total number of seeds sown [22].Te germination velocity index (GVI) was calculated as proposed by Maguire [23] as follows: where G 1 , G 2 , • • • .G n are the number of seeds germinated on frst, second, and last count.N 1 , N 2 , • • • , N n are the number of days at frst, second, and last count from the day of sowing.

Data Processing and Analysis. Data were entered in
Excel from where germination parameters were calculated, and germination and seed size-related graphs were generated.A seven-day fxed average of both temperature and relative humidity were calculated in Excel and used to generate a time series profle at the two experimental sites, while descriptive statistics were used to identify similarities and diferences between the two environments.Non-normal percentage data (FGP) were arcsine transformed, and analysis of variance (ANOVA) was performed on germination and seed size data in GenStat ® 14 th Edition (VSN International Ltd, Hemel Hempstead, UK).Means were compared using Tukey's procedure at the 0.05 level.To describe relationships between the diferent germination and seed size variables, Kendall's rank correlation coefcient (τ) (Kendall's tau-b) was calculated in IBM ® SPSS ® Statistics (Version 20) statistical software (IBM Corporation, Armonk, New York, USA) for paired variables and subjected to signifcance tests at both the 0.05 and 0.01 levels.For both LocalBam and TVSU544 at Ocettoke, however, τ values were calculated excluding the developmental stage of 123 days after sowing to ensure linearity of the data points.

Temperature, Relative Humidity (RH), and Soil Conditions at the Field Experimental
Sites.Temperatures at both the Ocettoke (Kitgum) and Koro (Omoro) feld experimental sites were quite similar in terms of average daily/ weekly values during the respective periods they were measured (Figure 2).Average temperature at Ocettoke for the entire period of 13 weeks was 28.16 °C with a range of 10.43 °C.Tat for Koro for the 12 weeks period was 26.8 °C with a range of 9.103 °C.Ocettoke was therefore slightly hotter than Koro by an average temperature of 1.36 °C.
Essentially, the overall temperature fuctuations during the measurement periods for both sites were similar (Figure 2).Tere was, however, a slight diference in the average RH between the sites (Figure 2), with Ocettoke having a higher average RH of 54.59% compared to 47.74% at Koro.Tis made Ocettoke more humid than Koro by an average of International Journal of Agronomy 9.51% RH.Notably, for both sites, towards the end of the experiment, temperature was consistently rising, while RH was declining in the same time period (Figure 2).Nevertheless, the within site variation in relative humidity was similar for both sites.Generally, during the growing seasons at both sites (April to June for the frst season and July to October for the second season), temperatures were comparatively low in most months with the corresponding higher RH, except in June, July, and October [24].Soil conditions at the two sites were generally similar in characteristics, though with slight variations for certain essential macro and micronutrients (Table 1).

Seed Size Characteristics.
All landraces showed consistent increase in the seed length, width, and weight from the early to the latter stages of sampling/development in the frst experiment (Table 2).Each landrace difered signifcantly in terms of its seed length, width, and weight (P < 0.01) at every stage of development except between 113 and 123 DAS, where the increases in all three parameters were not signifcant for LocalBam and TVSU544 and for both seed length and width in AbiBam001 (Table 2).Interestingly, however, AbiBam001 showed a signifcant increase in seed weight beyond 113 DAS (P < 0.01) compared to the other two landraces (Table 2).Also, as expected, there were signifcant diferences in the seed length, width, and weight among landraces at all stages of development (P < 0.01).
LocalBam was consistently the largest seed for all three parameters followed by TVSU544, while AbiBam001 was the smallest seed (Table 2), thereby conforming to the size categorisation of large, medium, and small seeds of Local-Bam, TVSU544, and AbiBam001, respectively.Sampling in the second experiment at 123, 130, and 138 DAS was able to validate the trend of nonsignifcant increase in the seed size between 113 and 123.DAS exhibited by LocalBam and TVSU544 in the frst experiment.Both LocalBam and TVSU544 showed no diference in seed length, width, and weight between 123 and 138 DAS (Table 2).Equally, AbiBam001 did not difer in seed length and width between 123 and 138 DAS (P > 0.05).However, it showed a clear diference in seed weight between 123 and 130 DAS but not between 130 and 138 DAS (Table 2).Tis therefore establishes a point of diminished increase in seed weight for AbiBam001.Collectively, these patterns strongly suggest that the stage of mass maturity (highest seed mass) for the landraces is approximately 113 DAS for both LocalBam and TVSU544 and 130 DAS for Abibam001.Tis also means that both LocalBam and TVSU544 mature earlier than AbiBam001.

Seed Germination. Both AbiBam001 and LocalBam
exhibited relatively high germination percentage (FGP) at the early developmental stage of 83 (71% and 82%, respectively), while TVSU544 had comparatively lower germination (58%) at this stage in the frst experiment (Figure 3).Te fnal germination percentage of the landraces steadily increased from 83 DAS onwards, with the highest FGP occurring at 103, 113, and 123 DAS for TVSU544, LocalBam, and AbiBam001, respectively (Figure 3).Tis is similarly depicted by the cumulative germination curves of the landraces (Figure 4).Beyond these peaks, the FGP begin to decline for both LocalBam and TVSU544, but for Abi-Bam001, there was still a steady increase between 103 and 123 DAS (Figure 3).At Koro, seed germinability remained relatively uniform at all sampling stages, albeit with a general declining trend for all landraces (Figure 3).Tere were no diferences among the germination speed of seeds at different developmental stages for all landraces at both sites (P > 0.05).Nevertheless, the fastest germinating seeds (highest GVI) were at 113 DAS for both AbiBam001 and LocalBam and 103 DAS for TVSU544 at the Ocettoke site, while at Koro, they were at 123 for both AbiBam001 and TVSU544 and 130 DAS for LocalBam.For both AbiBam001 and TVSU544, the seedling dry weight increased from 83 to the highest recorded value at 103 DAS and then declined or steadied afterwards (Figure 5).For LocalBam, however, the seedling dry weight steadily increased throughout the seed developmental period at Ocettoke (Figure 5).At the Koro site, the seedling dry weight remained relatively uniform for all landraces (Figure 5).Accordingly, cumulative germination curves suggest that by speed (GVI) and extent of germination (FGP), seeds of the highest quality from Ocettoke are obtained at 103, 113, and 123 DAS for TVSU544, LocalBam, and AbiBam001, respectively (Figure 4; Table 3).At Koro, the FGP was highest at 123 DAS for both AbiBam001 and TVSU544 and then declined from 123 to 138 DAS for both landraces (Figure 3).For LocalBam, however, it was marginally highest at 130 DAS, although this did not seem to difer much from that at 123 DAS (Figures 3  and 4).Tis trend is confrmed by a combined ranking of sampling days (DASs) for the seed quality parameters of FGP, GVI, and SDW, which shows that the best ranked DAS for each landrace (i.e., the developmental stage when the seed quality is highest) is 103 for TVSU544 and 113 DAS for both LocalBam and AbiBam001 (ranking data not shown).Conversely, the poorest quality seeds from Ocettoke were obtained at 93 DAS for both AbiBam001 and TVSU544 and at 123 DAS for LocalBam (Figure 4).For Koro, the poorest quality seeds were at 138 DAS for all landraces (Figure 4).However, the ranking of DAS shows that the poorest quality seeds from Occetooke occur at 83 DAS for both AbiBam001 and TVSU554, while for LocalBam, it is maintained at 123 DAS.Similarly, at Koro, the poorest quality seeds for ranked DAS were obtained at 138 DAS for all landraces.Tis means that for both AbiBam001 and TVSU544, seeds with the highest quality are obtained before mass maturity, while for LocalBam, they are obtained at approximately mass maturity.Diferences between the two experiments for the frst day of germination, last day of germination, time spread of germination, and speed of germination (Figure 5) were largely attributed to diferences in incubators' conditions.

Relationships among Seed Germination and Seed Size
Parameters.All germination parameters had positive relationships with all the seed size characteristics of seed weight, seed length, and seed width at Ocettoke with both 4 International Journal of Agronomy FGP and GVI having quite strong relationships with the seed size characteristics (Table 4).At Koro, FGP and GVI had negative relationships with the seed size characteristics for all landraces, with those for AbiBam001 being very strong (Table 5).Conversely, at Koro, SDW had positive relationships with all the seed size characteristics for all landraces, except for TVSU544, which were all negative relationships (Table 5).A positive relationship was observed between the FGP and GVI for all landraces at both sites with that of both AbiBam001 and TVSU544 being very strong at Koro (Tables 4 and 5).As would be expected, Kendall's τ values were strongly positive for seed weight and seed length, seed weight and seed width, and seed length and seed width for all landraces at both sites (Tables 4 and 5).

Discussion
4.1.Seed Size Characteristics.Landraces showed clear differences in the seed size (weight, length, and width) at diferent stages of development, attaining mass maturity at approximately 113-123 DAS for both LocalBam and TVSU544 and between 130-138 DAS for AbiBam001.Tese trends are well aligned with the result of a survey preceding and linked to the current study, which showed that farmers harvest their BN crops 4-5 months after sowing (roughly 120−150 DAS) [10].Typically, farmers harvest the crop at harvest maturity, usually some days to a few weeks after mass maturity and the survey was able to highlight the fact that harvesting lasts nearly a whole month [10].A recent study by Mandizvo and Odindo [19] estimated that BN landraces from a feld experiment in South Africa reached physiological maturity at about 65 days after fowering (approximately 100 DAS).Earlier, Linnemann [18] had determined the mass maturity of a BN landrace named "Tiga Nicuru" from Mali to be approximately 103 DAS under a 12 hr photoperiod regime.Further reports from other parts of SSA indicate that the maturity period for most landraces is about 4-5 months [2,26,27].Te maturity periods of landraces reported elsewhere closely relate to the fndings of the present study.Diferences in the maturation of landraces are basically determined by the genotypic characteristics of a given landrace and the environmental conditions under which the landrace is cultivated [28][29][30].In general, after ovule fertilization, orthodox seeds such as BN undergo histodiferentiation, in which ordered cell division and diferentiation result in formation of the embryo, seed flling involving cell expansion and storage reserves accumulation, and maturation drying marked by seed moisture loss and reduced metabolism [13][14][15].Seed flling ceases at physiological maturity, with subsequent maturation drying  International Journal of Agronomy 6 International Journal of Agronomy [14,15].In the present investigation, from the start of sampling at 83 DAS-113 DAS at Ocettoke, there was a steady and steep rise in seed weight, suggesting that these developmental stages fall within the seed flling stage.It is also very likely that the seed flling stage for all landraces start 10 or more days earlier than the frst date of sampling.Beyond 113 DAS, the slowed increase in the seed dry weight suggests the drawing to an end of the seed flling period and the onset of maturation drying for both LocalBam and TVSU544.For AbiBam001, however, seed flling continued   International Journal of Agronomy beyond 113 DAS which was strongly alluded to by the trend at the Koro site where there was a diminished increase in the seed weight between 130 and 138 DAS.Variation in the length of each of the phases is known to be dependent on the species and prevailing environmental conditions [14].An earlier study of seed development in a BN landrace ("Ankpa") from Nigeria indicated that embryo development proceeds up to at least 41 days after anthesis (roughly 70-75   8 International Journal of Agronomy DAS) and that longer photoperiod of 16 hr delayed fowering by as much as 7 days compared to 10 hr photoperiod.[17].
While the precise efect of environmental conditions (temperature, RH, and day length) during the experiments was difcult to ascertain, it is very likely that the relatively high temperatures, low RH, and more or less constant photoperiod of about 12 hr shortened each of the development stages for a landrace [30,31].High temperatures and low RH are known to infuence photoheterotrophic metabolism, which in turn provide cues for auxins, cytokinins, abscisic acid (ABA), and gibberellic acid (GA)-mediated seed growth, development, and maturation [13,14,16,32,33].Given that the landraces in the present study were cultivated in a relatively uniform environment, it is plausible that diferences in maturity periods are almost entirely due to their genotypic variations.One likely efect of the environment could arise from the soil factors of Ca levels and associated acidity of the two sites, with a much higher Ca value for Ocettoke than Koro and lower acidity at Ocettoke than Koro.Tis is due to the fact that Ca is perhaps the most limiting nutrient element in BN seed development [34].

Seed Germination.
Seeds of landraces exhibited relatively high germinability when harvested at 83 DAS which strongly suggests that a proportionate number of seeds of all landraces acquire germination capacity earlier than this developmental stage.However, seeds at a developmental stage before physiological (mass) maturity have low physiological potential [35,36], which is demonstrated in this study by the relatively low seedling dry weight and germination capacity.While the moisture content of the seeds was not determined at the diferent stages of development, it is expected to be relatively high before and at physiological maturity [14,15,37], thus further impacting negatively on the seed quality at this stage.Te more mature seeds at latter stages of development showed higher germination capacity, faster speed of germination, and higher seedling dry weight until a point when all these parameters steadied or began to decline except for AbiBam001.Te increase in seed germinability at latter stages of development can be attributed to temporal changes in seed maturation, whereby the lapse in time in the maturation process results in more seeds acquiring germination capacity [15,38,39].In a normal process of ageing, seeds begin to lose their vigour and germination capacity soon after physiological maturity, a phenomenon best explained by the seed survival curve [40,41].Te seed ageing process is accelerated by unfavorable environmental conditions during and after maturation while the seeds are still attached to the mother plant, during harvest, and in storage [40,42,43].Tis is a plausible explanation for the decline in most of the seed germination characteristics after physiological maturity.It is also possible that the lowering of seed quality is the efect of hardseededness [44][45][46][47].Many seeds including dry legumes develop hard-seededness as a dormancy mechanism to defer germination until favourable conditions exist after maturation drying and as a means of seed survival during seed dispersal and prior to seed germination [44,46,47].Hardseedeness in many seeds is proposed to start almost around the same time seeds attain physiological maturity and that it is enhanced by dry and hot conditions during the seed ripening process [45].Te major regulators of dormancy and germination are ABA and GA and their interaction, with high levels of ABA relative to GA towards physiological maturity and during maturation drying, ensuring that seeds go into a quiescent state [13-16, 32, 33].It is likely therefore that the timing of harvest of seeds for this study (in the dry and hot months of January-March) could have to some extent contributed to the seeds becoming harder than usual.Bambara nut seeds have been reported to show extended period to emergence under feld conditions (up to 35 DAS) and hard-seedednees has been suggested as a likely cause of this phenomenon [29].Nevertheless, a combined assessment International Journal of Agronomy of three seed quality parameters (FGP, GVI, and seedling dry weight) revealed that seeds of the highest quality were obtained at 103, 113, and 123 DAS for TVSU544, LocalBam, and AbiBam001, respectively.Accordingly, for both TVSU544 and AbiBam001, the best quality seeds are obtained before mass maturity, while for LocalBam, they are obtained at approximately mass maturity.

Relationships among Seed Germination and Seed Size
Parameters.For the Koro site with mostly mature seeds of all the landraces, seed germinability (FGP) and the GVI exhibited negative relationships with all the seed size characteristics, while SDW had positive relationships with all seed size characteristics for two landraces.It is likely that for all landraces, the smaller the seed, the higher the extent of and the faster the germination and vice versa.Tis trend has been observed for soybeans, whereby medium sized seeds had better physiological potential (germinability and vigour) than larger or smaller seeds [36].It would therefore seem that medium-sized seeds for all the landraces are the best for sowing.Te positive relationships between germination parameters and seed size characteristics at Ocettoke is attributed to temporal changes in the process of seed maturation, whereby as time passes, seeds become larger in sizes (weight, length, and width) with more and more of them acquiring germination capacity as they pass through stages of maturation [14,15,38,39].Te weak relationship between seed germination and seedling dry weight observed in this study seems to suggest that BN seed lots with high germination capacity do not produce the most vigorous seedlings or even at worst produce seedlings with the least vigour as was the case for AbiBam001 and LocalBam at Koro.Tis trend was alluded to by Mandizvo and Odondo [48], who observed that BN seed lots with higher viability did not necessarily produce seedlings of higher vigour.Strong positive relationships among the seed size characteristics of weight, length, and width is expected since they normally tend to have a pattern of concomitant increase as a result of cell expansion, especially during seed flling, thus proportionately maintained at all stages of development and maturation [14,15].

Conclusion
Landraces exhibited typical patterns in seed development and maturation with both AbiBam001 and TVSU attaining highest seed quality before mass maturity, while LocalBam seeds had the highest quality at approximately mass maturity.Tere was reduced germinability and generally lower seed quality beyond mass maturity for all landraces, suggestive of the infuence of physiological and physical factors that reduce germinability such as normal seed deterioration and hardseededness.For better quality seeds, harvest should be timed at 103−113, 113−123, and 120−130 DAS for TVSU544, LocalBam, and AbiBam001, respectively.Tis would result in medium-sized seeds of relatively high seed quality.A germination test involving properly size graded seeds from the respective stages of highest seed quality, i.e., 103, 113, and 123 for TVSU544, LocalBam, and Abi-Bam001, could also be performed to validate the size vs. extent vs. speed of germination relationship to ensure more precise criteria for future seed selection for the landraces.A broader spectrum and higher frequency of sampling including maturity stages outside what was possible in the present study, i.e., before 83 and beyond 138 DAS will provide a more complete picture of the developmental patterns of the landraces.Tis should target the main growing season, i.e., April-August/September to synchronise growing conditions with farmers' practices and ensure better seed yield.[49].

Figure 2 :
Figure 2: Temperature and relative humidity (RH) at the experimental sites during the crop growth and sampling periods.(a) Ocettoke (Kitgum) and (b) Koro (Omoro) readings.

Figure 3 :
Figure 3: Final germination percentage (FGP) of seeds of the landraces at diferent stages of maturity: (a) at Ocettoke and (b) at Koro.Error bars are the means ± standard deviation.

Figure 5 :
Figure 5: Seedling dry weight of seeds of the landraces harvested at diferent stages of maturity: (a) at Ocettoke and (b) at Koro.Means followed by the same letter are not signifcantly diferent at the 5% level.Error bars are the means ± standard deviation.

Table 5 :
Values of Kendall's rank correlation coefcient (τ) among the various seed and germination parameters for the landraces at Koro. * * 1.0 * * Correlation is signifcant at the 0.01 level.

Table 1 :
Soil characteristics of the two felds' experimental sites.

Table 2 :
Size characteristics of seeds of landraces harvested at diferent stages of maturation from the Ocettoke and Koro sites.
Means followed by the same letter within a column are not signifcantly diferent at the 5% level.

Table 3 :
Stages at which seeds of the diferent landraces attain key indicators of maturity.

Table 4 :
Values of Kendall's rank correlation coefcient (τ) among the various seed and germination parameters for the landraces at Ocettoke.Correlation is signifcant at the 0.05 level; * * correlation is signifcant at the 0.01 level. *