1 See discussions, stats, and author profiles for this publication at: Garaway, C. J. (2005). Fish, fishing and the rural poor. A case study of the household importance of small-scale fisheries in the Lao PDR. Aquatic Resources, Culture and Developmen... Article January 2005 CITATION 1 READS author: Caroline Garaway University of London 27 PUBLICATIONS 409 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Assessment and management of inland fisheries View project Impacts of irrigation development on fisheries View project All content following this page was uploaded by Caroline Garaway on 22 September The user has requested enhancement of the downloaded file.
2 g CAB International 2005 Aquatic Resources, Culture and Development 1(2), ISSN X DOI: /ARC20059 Fish, fishing and the rural poor. A case study of the household importance of small-scale fisheries in the Lao PDR Caroline Garaway* Department of Anthropology, University College London, Gower Street, London WC1E 6BT, UK * Corresponding author. Submitted 19 May 2004; Accepted 22 February 2005 Abstract Small-scale fisheries in developing countries are often considered to be of particular importance to poorer groups, with fishing often characterised in the literature as an activity of last resort. Despite this assertion, there are few quantitative studies that compare the role of these fisheries in households from different socioeconomic groups. This is particularly true of many agricultural areas in countries of the Lower Mekong Basin, where fishing is, frequently, one of a range of livelihood activities carried out for either subsistence or income generation. Through a detailed household survey, this study quantifies the importance of small-scale fisheries to households from different socioeconomic groups in Savannakhet Province, Lao PDR. It looks at three potential axes of importance : catches from small-scale fisheries; utilisation of fish caught; and reliance on fishing as a means of bringing fish into the household. Results suggest that along all three axes of importance, the role of small-scale fisheries, and fishing, whilst substantial, is not substantially different between the socioeconomic groups. Fishing therefore cannot be described as important only for the poorest of the poor, but as an essential component of all these households livelihoods. Keywords Small-scale fisheries, Lao PDR, Poverty, Fishing practices, Lower Mekong Basin, Southeast Asia, Household survey Introduction In southeast Asia, it is often stated that small-scale fisheries in rural areas are particularly important for poorer groups and, it is sometimes suggested, the poorest of the poor (e.g. 1 5). Such arguments are in agreement with the literature on common pool resources in general, which has demonstrated that resources that are under systems of either common property or open access have been found to be of particular importance to resource-poor, poorer groups (e.g. 6, 7). As stated by Bene 8, there is much reference in the literature to fishing as an activity of last resort. At the same time, there is a growing recognition that development initiatives in rural areas, such as fisheries enhancement or agricultural development, can bring about changes either in the nature of these resources, or rights of access to them, or both 1,8 15. Partly as a result of the common perception of small-scale fisheries as being of greater significance to the poor, there is growing concern amongst some that such changes redistribute benefits in a manner that may adversely affect these poorer groups. However, despite the fact that the increased importance of rural small-scale fisheries to the poorest groups is widely perceived and intuitively plausible, there are few detailed assessments quantifying and comparing its role here, with its role in other socioeconomic groups. This is particularly true of those agricultural communities, which characterise large parts of the rural Lower Mekong Basin region, where fishing is one of a range of livelihood activities carried out for either subsistence or income generation. In addition, whilst changes brought about by the aforementioned development are increasingly acknowledged and taken into consideration, there have been few quantitative
3 132 Caroline Garaway assessments of the impact they have specifically had on poorer groups in these areas. In fact, in the case of enhancement, specific issue has been made of lack of evaluation of initiatives per se 16. Nowhere are these information gaps more apparent, or the issues and concerns raised by agricultural development and fisheries enhancement initiatives more current and relevant, than in the Lao PDR. Fish is widely considered to be the major source of animal protein for the majority of people in the Lao PDR. Subsistence fishing is carried out by almost everyone who has convenient access to water 17. The country is characterised by semi-independent rural villages engaged in predominantly subsistence orientated agricultural production. According to Claridge 18, small perennial waterbodies have traditionally played an important role as a source of fish, particularly in the dry season when other seasonal water resources disappear. Even in areas with few perennial waterbodies, rice-field fisheries are an important seasonal fishery resource 19. Other research has pointed to the importance of small-scale fisheries to rural households in various parts of the country 13,15,20. At the same time, agricultural (and irrigation) development is a top priority in the country and fisheries enhancement in the Lao PDR is also gaining popularity with governments and communities alike 10,21. As elsewhere, Lao specific studies have shown that enhancement initiatives bring about changes in either the nature of benefits from resources and/or access to them 8,10,14, and concern has been raised that this may redistribute benefits in a manner that may adversely affect poorer groups. Such concern, whilst important, is based on an assumption that either poorer groups are more dependent on these resources in the first instance and/or that they will have less access to benefits from them as a result of changes catalysed by fisheries or agricultural development. There are no other studies in the country that have explicitly investigated these assumptions in a detailed and quantitative manner. This paper addresses this information gap by comparing the importance and role of small-scale fisheries in rural households particularly with respect to socioeconomic status, thereby investigating the assumption that poorer groups are more dependent on these resources. A description and quantification of how benefits change and are re-distributed in these rural households, following enhancement initiatives, is provided by the author elsewhere 22,23. This information is vital to ensure that future development, and in particular, enhancement policy, addresses the particular situations faced by rural households in the Lao PDR, as opposed to the development narratives common in the academic and development literature. The paper presents results from a household survey, conducted between May 1996 and April 1997, into the role and importance of small-scale fisheries to different socioeconomic groups in Savannakhet Province, Southern Lao PDR. For comparative purposes it also looks at how such differences compare with those at a between-village level. Three different aspects of the role and importance of fisheries are investigated: catches from small-scale fisheries, utilisation of fish caught, and reliance on fishing as a means of bringing fish into the household. Combined, these different aspects give a fuller description of the resources role in rural livelihoods. Methods The study was carried out in Savannakhet Province, the largest Province in the country and one of the 17 that, together with the Prefecture of Vientiane and the Special Zone of Xaisomboun, constitute the Lao People s Democratic Republic. Village selection The districts selected for this study are situated in one of the three wetland areas in the Province (the Xe Champon wetlands covering 240 km 2 ) 17 and, hence, were relatively water-resource rich, particularly compared with the upland districts in Savannakhet Province. To get adequate representation of households from different socioeconomic groups within villages, the total number of villages selected was kept low, being restricted to four, three in Sonbuli district (Kong Knak, Xieng Hom and Bung Xiang) and one in Champon (Nong Hong). The villages had already been pre-selected as part of a wider research project investigating the impact of enhancement initiatives on rural livelihoods 10. However, evidence suggested they were not atypical of other villages in the districts concerned. First, the selected villages represented a wide and characteristic range of aquatic resource availability scenarios. Secondly, exploratory research 10 showed that households perceived that the presence of the new enhancement initiatives, the reasons why these villages were initially selected, had not significantly affected their fishing activity, or their dependence on fishing. This lent support to the assumption that the presence of an enhanced waterbody in their village did not make the overall dependence on their remaining small waterbody resources significantly different from other villages, with similar resource availability, in the district. Finally, a contemporaneously carried out Province-wide census, documenting village assets and amenities, had shown that the villages were typical
4 Fish, fishing and the rural poor 133 in all respects bar one 24,25. These villages were above average in terms of accessibility, which probably had knock-on effects in terms of their increased involvement with government and other advisory bodies, as well as their opportunities to participate in a more market-orientated economy (as opposed to a subsistence one). However, at the time of this research, this had not yet opened up different economic opportunities that would cause a move away from the rural lifestyle in which fishing played an important role. Socioeconomic classification Prior to selection of households for the main survey, the socioeconomic status of households in the participating villages was ascertained. In the original research two methods were used to classify households; wealth ranking and the classification of households based on a socioeconomic index developed from variables collected in a household socioeconomic baseline survey 23. In subsequent analysis, results were not significantly different between the two methods 23, hence only results from the former (wealth ranking) are described in this paper. However, the socioeconomic baseline survey was used to overcome problems of the relative nature of wealth-ranking data, and its role in this respect is described. Wealth ranking The method of wealth ranking used in this research was as described by Guijt 26. In each village, the same households (50 100% of the total village population depending on village size) were ranked on three separate occasions by three separate groups of people from the village concerned. Villagers were first asked to describe what, to them, characterised wealth or well-being and then asked to group households based on their relative wealth. Groups chosen to perform the wealth-ranking exercises were generally mixed (sex and age), in order to involve a wide range of perspectives*. Using the results of the three individual wealthranking exercises, each household was given an average ranking. The few households that had been very inconsistently ranked (e.g. assigned to nonadjacent wealth groups in the different exercises) were *Whilst acknowledging the fact that power relations amongst villagers may have left some villagers unable or unwilling to express their views by using this method 36, the problem was not felt to be as intractable as in other countries. Except in the specific instance of talking to village elders and the administration, women were generally ready to speak up. Compared to elsewhere, Lao women are thought to rank very high in terms of status, power and autonomy 37 (p. iv) with matrilinearity and matrilocality being thought to be important causes of this. re-examined through discussions with the original participants or the village headman. Analysis of wealth-ranking data The wealth-ranking technique produced a relative classification of households. However, because the criteria for doing so were not explicit or quantified for each case, or each exercise, comparability between sites was an issue, as was internal (within village) consistency between individual exercises. The internal consistency of rankings was tested using a non-parametric test that measured the degrees of association between the three individual rankings of the same households (Spearman s rank correlation coefficient). To check whether household status within groups was the same between villages (i.e. group three in village one represented a similar class of people to group three in village two), those households subject to the fishing survey were asked additional socioeconomic questions in a socioeconomic baseline survey. Variables selected for this survey were based on those that had been considered key indicators of well-being during the wealth-ranking exercises and that could be subject to quantitative analysis. Comparative analysis of the values of these individual variables (using a one-way ANOVA) gave some check on whether there were significant differences between similarly defined socioeconomic groups in each village. Household fishing and fish acquisition survey Household selection and sample size Wealth-ranking data had produced three/four wealth groups per village. To ensure a minimum sample size of three within each wealth group in each village, a constant sampling fraction of 20% of households from each village, chosen from those households that had already been wealth ranked, was chosen. This resulted in a maximum possible number of households (103), given logistic constraints, being selected. The households ranked in the wealth-ranking exercise were placed in a list from richest to poorest for each village (those with the same score were placed in a random order) and linear systematic sampling (LSS) used to ensure adequate representation of households from each socioeconomic group. Survey design The survey measured the quantities of fish coming into households from different sources, and what it was
5 134 Caroline Garaway Table 1. Local classification of types of fish, the forms they are acquired in, and conversion rates Method of acquisition Types of fish and forms in which acquired Common units of measurement/method of description Converted weight Caught Small (<5 cm Noodle bowls (four Bowl A: 150 g fish and shrimp) sizes*) Bowl B: 225 g Bowl C: 300 g Bowl D: 488 g Large (>5 cm) Width, length Fish sticks # Bought Small wet Noodle bowls, kg s As bowls above from price Large wet Width, length, kg s Fish sticks (including live) Fermented (ba dek $ ) Dried from price Noodle bowls, from price No standard measurement, estimate from price only Bowl A: 67.5 g Bowl B: 105 g Bowl C: 135 g Bowl D: 219 g Received as a gift All of the above As above As above Fish soup Noodle bowls Bowl A: 76 g Bowl B: 121 g Bowl C: 172 g Bowl D: 250 g Payment in kind Large wet (including live) As above Fish sticks * Noodle bowls were the most basic item of household crockery and there were four sizes that were commonly recognised and used by households. Weights for each bowl size for small fish and shrimp were calculated via direct measurement. # Whilst field-testing had showed that, generally, fishermen were not able to adequately estimate the weights of fish they had caught, they and fish buyers could accurately estimate, via recall, the length of fish caught or bought to within 5 cm in a 5 25 cm range. This range covered 95% of fish caught 23. Villagers estimates of the lengths of their fish (using different length sticks as a guide) were therefore used to estimate weights using length/weight relationships. Firstly, length/weight relationships were calculated for some of the most common species of fish in these districts using the following relationship: Weight = a(length) b, where a is constant and bffi3. Data for these calculations came from the project fish sampling and other literature sources 38. Next, on the basis of similarity of coefficients, the common fish species were separated into three groups: 1. Anguilla spp. and Macrognathus spp.; 2. Anabas testudineus and Oreochromis niloticus; 3. All other species. The coefficients were recalculated so that they all referred to an exponent of three and the average weight for each group of fish for differing lengths was calculated. Lastly, fish sticks were specially designed to provide an appropriate tool for using the length/weight information in the field. $ Common form of processed fish, frequently used daily. It was made from small fish and shrimp, combined with rice-bran, salt and water, and was the most common way of utilising and storing surplus fish caught at relatively abundant times of year. Different grades of ba dek were commonly recognised and figures here relate to average quality. Measurement of the quantity of fish in ba dek was calculated by asking women from different socioeconomic groups their recipes. used for, at regular intervals, over a period of one year. Additional information on who was fishing, hours spent fishing and fishing gears used was also collected. The exploratory research 10 had identified several issues that had to be incorporated into the survey design. These included: Seasonal variation in fishing activities; the active role of women and children as well as men in fishing activities; the potentially important role of being given fish as a wage proxy (payment in kind) or receiving it as a gift, as sources of household fish; the different forms of processed fish; the need to use widely recognised units of measurement for both wet and processed fish; and the need to estimate the weights of fish from all of these forms and units of measurement. Forms of fish, units of measurement and corresponding weights that were subsequently used in the survey are presented in Table 1. Data collection The survey was conducted over a 12-month period from the beginning of April 1996 to the end of March Staff availability limited the number of potential rounds to six, undertaken at two-monthly intervals. The survey was designed to collect information on the previous week s fish acquisition activities (caught, bought, earned and received), allowing for a flexible recall period of one day to one week depending on the frequency of the fish acquisition activities and capabilities of the household respondent(s). Questions began with the previous day s activities and went backwards, where appropriate, from there. Each time fish was bought, caught, earned or received was recorded as one entry on the form. This procedure was used to collect specific data on fish quantities,
6 Fish, fishing and the rural poor 135 origin of fish and subsequent use. Recall as opposed to direct measurement was chosen as, despite the advantages of the latter, households rarely had means of weighing and recording fish quantities and, being a substantially subsistence orientated fishery, there were no locally recognised landing sites where fish could be weighed. Scaling up the data When present, records of the most recent quantities of fish acquired in the week the survey was conducted were averaged, and then multiplied by the average number of times such an activity was carried out to give a weekly estimate. To provide estimates for the first four weeks, this weekly estimate was multiplied by four. Households were then asked to estimate how this compared relatively (as a percentage) with the four weeks prior to this. This percentage was used to obtain an estimate for this first month of the monitoring period. This method was preferred over multiplying the weekly estimate by eight, as at some periods of the year (particularly as water levels rise or fall), there could have been large differences over a two-month period. Some events were rarer than others (such as buying ba dek or earning fish) and where there were no records for the previous week, estimates were based on recall from other weeks, and scaled up accordingly. One monitoring period (August/September) was completely missed due to large floods and the inaccessibility of villages at this time. After consultation with fishers about this fishing period* estimates for this period were calculated by averaging values for the round immediately preceding and following this period. There was an exception to this. A few households recorded exceptionally high catches in the October/November period. This was due to a specific type of intensive fishing with dragnets in the rice fields in a very short period (less than two weeks) when the floods were receding. In these instances therefore, the average fish catch for the October/ November period over all households was used to provide the value leading to the averaged estimate for the August/September period. *Fishers indicated that catches rose from the end of the dry season (April/May) peaking when the annual floods receded in approximately October/November, and then declining again. The exceptionally heavy floods in this year could have decreased fish catchability in some areas and increased it in others and therefore it was impossible to estimate what overall impact such flooding would have had on fishing at this time of year. Under the circumstances, it was appropriate to base estimates on information collected from fishers. Table 2. Wealth criteria generated from wealth-ranking exercises (Criteria are listed according to the frequency with which they were cited. Those with a* next to them were cited in each wealth-ranking exercise) Richer groups Productive assets Rice mill* Many livestock (>5)* Surplus rice production in some years* Small business (e.g. shop)* Fertile land* Boat Hand tractor Consumer goods Gold Big house Motorbike Television Several bicycles Characteristics of members Educated Innovative Hard working Analysis of household fishing survey data Data were scaled up and comparative analysis of differences between villages and/or socioeconomic groups was conducted using standard statistical techniques (e.g. ANOVA, t tests), which are referenced, as used, in the results section. Results Socioeconomic classification Poorer groups Productive assets Few (<3) or no livestock* Not enough rice production in all or some years* Less productive land or no land* Other Large number of dependants Not enough labour Indebtedness Characteristics of members Poor work ethic Wealth-ranking criteria Villagers used a wide range of criteria to classify socioeconomic groups and these are presented in Table 2. Whilst none were believed to be rich per se, there was recognised variation between households. As can be seen, the criteria for poorest were not just asset based (though these were most frequently cited) but also covered other dimensions of disadvantage 27 such as being physically weak and vulnerable. The emphasis on productive asset criteria as opposed to consumption and expenditure, reflects the predominantly subsistence economy of the rural Lao. In addition, it can be seen that differences between wealth groups (on commonly cited criteria) are based on more or less of any one criterion as opposed to different criteria, reflecting the narrow range of livelihood strategies open to Lao villagers.
7 136 Caroline Garaway Households 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Richer Middle Poorer Bung Xiang Kong Knak Nong Hong Xieng Hom Village Fig. 1. Distribution of households between socioeconomic groups in the household survey sample Internal consistency of wealth-ranking data Spearman s rank correlation coefficient was calculated to assess the consistency of the three household rankings in each village. In all cases correlation was strong (average values for each village: r = 0.64, 0.70, 0.73 and 0.89 respectively) and significant (P< 0.001), showing that villagers had a similar perception of where households in their village stood on the relative wealth scale. This consistency suggested that villagers were measuring the same underlying phenomena and that they had considerable knowledge of their own and other households situations. Wealth groups During the wealth rankings, households were put into three groups in all cases but one (four groups in exercise in Kong Knak). Individual households were reassigned to socioeconomic groups on the basis of where the majority of villagers placed them. The largest numbers of households in each village, representing approximately 50% of the population, were in the middle group. The exception was Kong Knak village, which showed a more equal distribution for each group. This was probably due to the fact that the original four wealth-ranking groups had been reassigned to three in subsequent analysis to make inter-village comparison easier. Results are shown in Fig. 1. Overall, more households in the villages were considered to be in the poorer than in the richer group. Comparing relative wealth across villages The between-village consistency of the different wealth groups produced from the wealth-ranking exercises was investigated through an ANOVA test of key socioeconomic indicator variables. To support the hypothesis that the socioeconomic groups were similar across villages but significantly different to each other, it was expected that significant differences would not be found within the same socioeconomic groups but would be found between them. Results are presented in Table 3. Some variables did not meet the assumption of normality and therefore could not be included in the tests. The variables (rice index rich and middle wealth groups and buffalo in the poor group) did not meet the assumption of equal variance (Levene test) and in these cases the conservative non-parametric test, Tamhane s T2, was used. The descriptive statistics show that across all villages, variation between households for most variables was low. For the majority of variables (exceptions being adult equivalent units (AEU), for the middle group; dependence ratio, for the middle and poor groups; and, land index for the rich groups), the null hypothesis, that there is no difference between means, could not be rejected. That is to say, the same socioeconomic groups between villages were not found to be different. The same tests were carried out comparing values between socioeconomic groups. Results are presented in Table 4. In contrast to comparisons within the same socioeconomic groups, those between, whilst small, were significant (P<0.05) for AEU, buffalo, cattle, land index and rice index. With regard to the other variables, results suggest that their relationship with socioeconomic status was more complex. Combined, these two sets of results lent support to the hypothesis that differences existed between socioeconomic groups but not within the same socioeconomic groups across villages. This further suggested that comparability between villages was feasible based on these groupings, partly addressing the problems of relativity brought about by the wealthranking technique. Fishing practices and total catches Fishers Household participation in the fishery was almost universal. Only two households (less than 2%) never participated in the fishery over the year monitoring period. Adults, both men and women, and children from the age of 10, fished. Women were strong participators in the fishery, particularly with respect to certain gears such as the drag net and, to a lesser extent, the scoop net. Seasonality of fishing Figure 2 shows how average household catches varied across monitoring periods. The highest average catches occurred as the floods receded (October/November). Fish were caught as
8 Fish, fishing and the rural poor 137 Table 3. Socioeconomic assessment of household characteristics by wealth group Mean Mean Mean Mean ANOVA Bung Xiang Kong Knak Nong Hong Xieng Hom **(P< 0.05) Group 1 N = 25 (rich) AEU* Bicycles Bicycle/member Buffalo Cattle Dependence ratio # Land (ha) Land $ ** Rice index Group 2 N = 49 (middle) AEU ** Bicycles Bicycle/member Buffalo Cattle Dependence ratio ** Land (ha) Land index Rice index Group 3 N = 29 (poor) AEU Bicycles Bicycle/member Buffalo Cattle Dependence ratio ** Land (ha) Land index Rice index * AEU (adult equivalent units) a measure of household size produced by weighting measure for individual household members based on consumption levels of different ages. Weighting in this research is the same as used by the Lao National Statistical centre (i.e. the first adult is weighted as 1 and following adults by 0.9. Small children below age 7 get 0.1 and children aged 7 15, 0.5) 28 Also weighted by presence/ absence from household (% of year). Average AEU per household in this sample is # Dependence ratio ratio of dependants to non-dependants. Dependants, children 0 14 years and adults older than 65. Non-dependants, years old. Figures chosen here are those used by the Lao National Statistical Centre 28. $ Land index [total land (ha)]. [quality of land]. Quality score: 1, good; 0.8, average; 0.5, poor. Rice index index for rice production over previous 5 years. [(a + b + c)/5] + 0.5, where a = [not enough rice, -0.5].[no. years]; b = [enough rice, 0]. [no. years]; c = [surplus rice, 0.5]. [no. years]. they migrated from the rice fields back to rivers, streams and small waterbodies. This pattern has been described for other parts of Lao PDR and Southeast Asia 29,19,30,31 and was confirmed by statistical analysis, with the Tamhane T2 test showing that catches in October/November were significantly greater than all other periods except December/January (P<0.05), and catches in April/May were significantly less than any other period (P <0.05). Total annual catches The average weekly household catch across the year [and 95% C.I.] was 1.27 kg [ ]. This was equivalent to 0.3 kg [ ] per household member (in terms of AEU s). Multiplying these figures across the year gives an average annual catch per household member of just less than 16 kg and a household average annual total catch of approximately 66 kg. Variation in catches between villages and socioeconomic groups Inter-village variation Comparative analysis of village catches revealed that while the villages showed similar seasonal trends, with the exception of differences between Bung Xiang and Xieng Hom, large and statistically significant differences (Tamhane T2, P< 0.05) existed between the size of catches. (This test was based on average weekly catches throughout the whole year and not per period). The average catch per household in Kong Knak was more than double that of catches in Bung
9 138 Caroline Garaway Table 4. Average household characteristics for each socioeconomic group Variables N = 103 Mean rich Mean middle Mean poor AEU ** Bicycle Bicycle/member Buffalo ** Cattle ** Dependence ratio Land (ha) ** Land index ** Rice index ** ANOVA ** (P<0.05) Xiang and Xieng Hom and five times that of catches in Nong Hong. Average weekly and annual catches per household and per AEU member are shown in Table 5. Reasons for variation in catches could have been due to differences in household effort. In fact, analysis showed that households went fishing less often in Nong Hong village than in other villages (P< 0.05 Tamhane T2 test), contributing to the explanation of lower catches here, but other villages were almost identical. No other significant differences between villages were found in their fishing effort (either total fishing trips or time spent fishing), suggesting that resource availability/quality was another contributing factor to variance in total catches. The three villages in Sonbuli were very close to each other and hence it might be supposed that they had access to similar resources. However, villagers are constrained by what is available in their immediate vicinity for two reasons. First, fishing is often combined with other activities that require household members to be in or near their village or rice fields and therefore households do not have the time to travel to waterbodies further away where they are allowed to fish. Secondly, within their area, villagers tend to fish in waterbodies over which their village have de facto rights of ownership, rights that are generally well respected. These customary rights are only attached to the discrete perennial waterbodies (i.e. not the rivers or rice fields) and permitted only village members the rights to Average catch/household/week (g) April/ May June/ July Aug/ Sept (est.) Oct/ Nov Monitoring period 96/97 fish without prior special consent from the village administration. Inter-socioeconomic group variation Dec/ Jan Feb/ March Fig. 2. Seasonal variation in household average catches: April 1996 March 1997 Table 6 shows the average weekly catches for households from different socioeconomic groups and associated statistics. Three things are immediately clear. First, any catch variation between socioeconomic groups was substantially less than differences in catches between villages (a difference between highest and lowest of 0.56 kg per household per week as opposed to 1.54 kg between villages). Secondly, there was no linear relationship between socioeconomic status and household catches, as it was the middle households who caught, on average, the least fish. Thirdly, the general perception put forward in the literature concerning small-scale common pool resources, that it was the poorest households that utilised the fishery the most was not supported in this study, as it was the richer households that caught Table 5. Variation in weekly and annual catches between villages Catch household -1 week -1 (kg) Catch household -1 year -1 (kg) Catch household member (village specific AEU) -1 week -1 (kg) Catch household member -1 year -1 (kg) Bung Xiang (N = 25) [95% C.I.] Kong Knak (N = 43) [95% C.I.] Nong Hong (N = 16) [95% C.I.] Xieng Hom (N = 19) [95% C.I.] 0.82 [ ] 1.94 [ ] 0.40 [ ] 0.95 [ ]
10 Fish, fishing and the rural poor 139 Table 6. member Differences between socioeconomic groups in average weekly catches per household and per household Richer [95% C.I.] Middle [95% C.I.] Poorer [95% C.I.] Catch household -1 week -1 (kg) 1.58 [ ] 1.00 [ ] 1.43 [ ] Catch household -1 year -1 (kg) Catch household member -1 (group 0.32 [ ] 0.24 [ ] 0.42 [ ] specific AEU) week -1 (kg) Catch household member -1 year -1 (kg) more. In fact, statistical analysis showed none of the household differences between groups to be significant (P< 0.05). Analysis of catches at the individual, rather than household, level produced somewhat different results. In order to test the a priori hypothesis, that the poorest in the community caught the most fish, individual t tests (not assuming equal variance) were used. Households from the rich and middle wealth groups were grouped together for this analysis. When differences in household size were taken into account, the poorest households caught significantly more (per household member per week (AEU) (P< 0.05)). The mean difference was small, however, at approximately 0.15 kg. Analysis of fishing trips showed that individual household members (AEU) from poorer households did go fishing more often than other socioeconomic groups, which may partly have contributed to this result. Non-homogeneity of group variances meant that a one-way ANOVA could not be carried out but the t test showed a significant difference of 0.17 trips per household member per week between the richest and poorest households (P < 0.05). Results of an analysis into the time spent fishing per trip were inconclusive. In summary then, there was variation in total catches between seasons, villages and socioeconomic groups. Catch variation between villages was substantial. Socioeconomic variation was far less than that between villages and not significant at the household level. In fact, trends suggested that richer households were catching more. Owing to the smaller household size of poorer groups (in terms of AEU), a significant difference was discernible at the individual level, with data showing poorer individuals catching slightly, but only slightly, more. This finding shows the importance of analysis at the sub-household level. However, in general, these results question the assumption that poorer groups are catching substantially more than others. Utilisation of fish caught As a second aspect of the role of fisheries, households were asked in the monitoring survey how they utilised the fish they had caught from their most recent fishing trips. There were 895 separate records, all used here. On the assumption that these events were representative of all fishing events in the monitoring period, estimates for the quantity of fish annually eaten, given away or sold were also calculated. In terms of axes of importance, this information investigated whether poorer households relied more on the fishery for income generation. Overall utilisation Results of utilisation of fish caught over the whole sample are presented in Table 7. It shows that, as expected, most fish was eaten by households themselves (75%) as opposed to being sold or given away. The mean quantity of fish eaten per household per year was estimated at 49.4 kg. Approximately one-fifth of total catch was sold though it was not a common activity occurring in only 11% of recorded events. Finally, giving away fish was a more frequent occurrence (> one in five cases), but the average amount each time was small, leading to an annual amount of approximately 4 kg per household. Variation in utilisation between villages and socioeconomic groups Figures 3 and 4 show how the utilisation of fish varied between socioeconomic groups and villages respectively. Data on socioeconomic differences showed that in all cases most of the fish were consumed by households themselves, true of differences between villages also. With regards to other uses, whilst Fig. 3 shows that poorer groups sold a higher percentage of their catch than the other two groups (and at an average of just under 30% of total catch, this was not an insignificant activity), the most significant difference in how fish was utilised was between villages. Kong Knak village sold significantly more than any other, with one village, Nong Hong, selling none, and therefore significantly less. The mean percentage of fish sold in Kong Knak was also just under 30%. In summary, results show that those that have already been shown to be catching the most fish
11 140 Caroline Garaway Table 7. Utilisation of fish over all villages. Figures in square brackets are the 95% C.I. limits for mean catch for each use (kg) as a percentage of mean total catch Mean caught fish utilised, expressed as % of total weight caught % of cases in which activity occurred (from 895 records) Estimate of mean quantity of fish (kg year -1 ) Eaten Given away Sold [ ] [ ] [ ] (both in the cases of villages and socioeconomic groups) were using the local fishery for income generation as well as household subsistence. At 30% of total catch in these cases, this selling was not an insignificant activity. However, in all cases consumption was the principal use and fishing for income generation was not an activity for the poorer groups alone. Reliance on fishing as a means of acquiring fish The final measure of importance looks at the contribution of other sources of fish to the household and leads to an estimate of household reliance on fishing. Greater dependence of poorer groups could have been linked to their inability to get fish from other sources. Overall reliance on fishing and total fish acquired by households Table 8 shows the average quantities per monitoring period for each activity bringing fish into the household along with 95% C.I. Results show that, unsurprisingly, Mean % of catch utilised and 95% CI Rich (eaten) Middle (eaten) Poor (eaten) Rich (sold) Middle (sold) Poor (sold) Rich (given) lower 95 CI mean upper 95 CI Middle (given) Poor (given) Utilsation between different socio-economic groups Fig. 3. Variation in the use of fish between socioeconomic groups (the 95% C.I.s represent the 95 confidence limits for mean catch for each use (kg) as a percentage of mean total catch) the most important source of fish was that caught by the households themselves (73% of total) indicating that reliance on fishing was high amongst the whole sample population. After household fishing, purchasing fish was the next most important supply route (less than 20%), with fish being received (from friends/ relatives) and earned (payment in kind) contributing only a small fraction of total supply. From these results, estimates, on a weekly and yearly basis, for households and household members (AEU) were calculated (Table 8). Results showed that the average fish supplied for each household member per week and per year was approximately 0.4 and 21.5 kg respectively. Differences between villages and socioeconomic groups Figures 5 and 6 show how percentage composition of fish supply and total quantities per member per week varied between villages and socioeconomic groups respectively. As can be seen from Fig. 5(a), the reliance on the different supply routes was very similar for all Mean % of catch utilised and 95% CI Bx (eaten) Kn (eaten) NH (eaten) XH (eaten) Bx (sold) Kn (sold) Nh (sold) Xh (sold) Utilisation between villages lower 95 CI mean upper 95 CI Bx (given) Kn (given) Nh (given) Xh (given) Fig. 4. Village variation in utilisation of fish (BX = Bung Xiang, NH = Nong Hong, KN = Kong Knak, XH = Xieng Hom). (The 95% C.I.s represent the 95% confidence limits for mean catch for each use (kg) as a percentage of mean total catch)
12 Fish, fishing and the rural poor 141 a) % Composition b) Average quantity/household member/week (g) Bung Xiang Kong Knak Nong Hong Xieng Hom Caught Bought Received Earnt Activity Xieng Hom Nong Hong Kong Knak Bung Xiang 0 Caught Bought Received Earnt Activity Fig. 5. Variation in fish acquisition activities across villages villages, with no differences being significant (Kruskal Wallace test (a non-parametric equivalent to ANOVA). This information, together with average quantities shown in Fig. 5(b), showed that villages differed in the quantities of total fish coming into the household, and not where it came from. In other words, villagers did not compensate for catching less by buying more. This could be due to an inability to buy more fish (due to lack of market supply, or household economic constraints) or due to villagers being able to supplement their fish diet with other sources of animal protein. With regards to socioeconomic groups (Fig. 6), looking at percentage composition, socioeconomic groups showed similar trends. Reliance on fishing was very similar for all groups (as was earnings). Reliance on bought fish was highest for the richest socioeconomic group, and lowest for the poorest, as would be expected given the lower purchasing power of poorer socioeconomic groups. Results show the opposite with respect to receiving fish, with the poorest receiving more. This was also unsurprising, with strong social norms encouraging wealthier households to help the poorer, particularly within kinship groups. However, analysis of differences in composition (Kruskal Wallace test) showed no evidence of significant differences between any groups. In terms of quantities (Fig. 6(b)) comparative statistical analysis (using a log transformation and subsequent t test) showed that the rich do buy significantly more per member than the poor (P <0.05) and the poor received significantly more per member than the rich (P<0.05). In conclusion, no group was more reliant than any other on fishing as a supply of fish to the household, which was an unexpected result. The role of received fish was higher than anticipated and in fact in part compensated for the lower ability to buy fish amongst the poorer households. Summary Results show that fishing in small waterbodies was an essential economic activity for all households, not just those from the poorer groups, with an average of 70% of total household fish supply being caught by Table 8. Average quantities of fish coming into households from different sources per 8-week monitoring period and, from this, estimates of average amount of fish supplied to households and household members by each acquisition activity Unit Caught Bought Received Earned Total acquired Average weight per 8 week monitoring period (kg) [95% C.I.] Household week -1 (kg) Household year -1 (kg) Household member (AEU) week -1 (kg) Household member (AEU) year -1 (kg) [ ] 2.82 [ ] 0.85 [ ] 0.06 [ ] [ ]
13 142 Caroline Garaway a) % Composition b) Average quantity/household member/week (g) Richest Middle Poorest Caught Bought Received Earnt Activity Poorest Middle Richest 0 Caught Bought Received Earnt Activity Fig. 6. Variation in fish acquisition activities across socioeconomic groups household members. The fishery was utilised all year and provided the majority of household fish all year. Reliance on caught fish, as a source of household fish supply did not significantly vary between socioeconomic groups. Mean annual catches per household were estimated at approximately 66 kg per year (16 kg per AEU). Total fish coming into the household was higher at 90 kg per year (19.5 kg per AEU) of which 74 kg (75%) was eaten by the household (17.5 kg per AEU). The rest is either given away to relatives and friends or sold to generate household income. Fish consumption figures varied, but at an average of 17.5 kg per household member per year, they are far greater than the frequently cited national figure of 7 kg per person peryear 32 ), though larger site specific estimates do exist (e.g. 20,33 ). With fish being widely believed to be the major source of animal protein available in rural Lao, the importance of such fisheries becomes even more apparent. Variation in average catches did exist between socioeconomic groups but that between villages was far greater, and it was only when household size was taken into account that a small but significant difference could be found between the poorest and other groups. With regards to the utilisation of fish, results show that those catching the most fish (found in this research to be one particular village, Kong Knak, and the poorer groups in general) were selling substantially more and therefore using the local fishery for income generation as well as household subsistence. At approximately 30% of total catch in both cases, this selling was not an insignificant activity. Again though, such activity was not obviously only the domain of poorer groups. Given prices at the time (and the percentage of small fish and shrimp versus larger fish being caught) the average value of these catches was estimated at US$ perhousehold per year. Whilst appearing low, taking into consideration the economic context of the time, it was not insignificant. Forty-six dollars was approximately the same as 26 days wage labour (if such employment could be found) and full time Provincial government staff salaries were 30US$ per month. Finally, per capita income for the whole country was low, estimated at 295US$ in , with most of this income generated in Vientiane and the larger provincial capitals. A lower estimate of US$ for rural households in Lao PDR and neighbouring countries has been suggested 35. Discussion In this example, Lao PDR does not fit the common perception of fishing being of particular importance to the poorest groups. On all aspects of possible importance/dependence, little evidence of this could be found. Whilst this common perception is sometimes, perhaps even usually, the case, it is not, therefore, a foregone conclusion. This was also one of the results of a far less detailed study in NE Thailand, where it was found that it was middle as opposed to poorer groups that utilised the local fishery the most 9. One aspect this study has not considered is the relative importance of fish versus other forms of animal protein (e.g. livestock). Whilst being an important aspect that should be investigated, there is little doubt that at the time of this study, fish was both the preferred and the most commonly eaten source of animal protein. Evidence here shows small-scale fisheries can have an important role for all members of the rural community and therefore any fishery or agricultural development initiatives that alter the nature of, or command over, the fisheries resources could have widespread effect. Such findings have implications for development policy in these areas. Whilst recognizing the importance of the socioinstitutional changes that often accompany development, we should not assume, on the basis of greater dependence, that it is