Fish Losses for Whom? A Gendered Assessment of Post-Harvest Losses in the Barotse Floodplain Fishery, Zambia

Round 1

Reviewer 1 Report

This paper is a study of post-harvest fish losses in the Barotse Floodplain fishery in Zambia, focusing on the higher rate of losses experienced by women compared with men. For example, the authors report that “Female processors lost three times the mass of their fish consignments compared to male processors” (Abstract). I found this paper fascinating, original, easy to read, coherently argued, methodologically sound, and faultlessly presented, and I strongly recommend its publication virtually as it stands. However, I have one question which I would like the authors to answer before their paper is accepted for publication: why did they not ask their respondents more insistently to explain why women lose more fish than men?

 

In some places, the authors allude to potential explanations for women’s poorer performance. For example, in the Abstract, they assert that “Technical and social constraints create gender gaps in post-harvest losses”, and in lines 520-526, they state:

 

“it is argued here that fish losses are partly caused by deeper-rooted gender issues that especially complicate or constrain women’s abilities to adequately process fish with minimal losses in the processing node. These studies highlight unequal differences in women’s and men’s access to assets and their decision-making powers, reinforced by unequal gender norms, attitudes and beliefs surrounding women’s roles and capabilities. These gender constraints could partly explain why our study found differences in post-harvest fish losses between women and men”.

But the authors do not explain what these “deeper-rooted gender issues” are. One of their quotations shows that at least one respondent gave one reason – that women cannot spend as much time as men on selling fish:

“We [women] cannot wait there [in the market] and sell our fish as we have to use the money that day to buy food for our families and still get back [home]. I have to get back [home] and give my children nshima [food], I cannot wait at the market too long, otherwise my fish will rot and so I have to sell my fish for a lower price (Tangatanga 5/5/2015)” [lines 400-403]

 

But the authors do not follow up this explanation, by for example, noting that there are two reasons offered here - one is lack of time; the other is the risk of rotting – and only the former reason is gender-based. Nor do the authors seem to ask their respondents for other explanations for gender differences in fish losses.

 

The authors refer to other studies which have found gender-based differences in fish activities in the Barotse Floodplain and elsewhere, implying that their own study has shown similar findings:

“Women’s gendered role as caretakers of their homes, where they disproportionately perform the majority of unpaid tasks (e.g. cooking, washing clothes, caring for children, the sick and the elderly), for example, influences how, when, and where women process, store, transport, and trade fish. Women’s limited decision-making powers in certain rural African contexts dictate how much time, effort, and money they can invest in fish value chain-related activities. This is evident in gender studies carried out in the Barotse Floodplain. Power relations can further constrain groups, especially women, from accessing key resources. Power relations also influence women’s abilities to make decisions to forgo the completion of domestic (unpaid) tasks that they are expected to perform, and instead, engage in work outside the home that generates income, which is also evident in the Barotse Floodplain”. [lines 540-550]

But what evidence do they provide, for example, that “power relations” played a part in the gender differences in the fish losses they found in their study? Since the central focus of the research is this gender difference, it is surprising that the authors did not ask respondents for their perceptions of the reasons for this difference. The authors seem to leave this issue to future researchers to focus on:

“Since such social and technical constraints are evident in other studies situated in the Barotse Floodplain, we can surmise that they are a likely contributing factor to why women may have experienced higher losses than men from our sample. Further research should aim to uncover causal links, that are social and technical, in analyses of food production, loss and waste”. [lines 560-564]

But why did they not focus on these “causal links”?

My recommendation is that the authors should be asked why they did not ask respondents to give their perceptions of reasons for the gender differences in fish losses before this otherwise impressive paper is accepted for publication.

Author Response

We are delighted with the reviewer’s comment and thank the reviewer for the constructive feedback. Naturally, we regret that we were not able to provide a larger generalizable sample and that we were unable to provide more nuanced, or statistically verifiable, causal links as to “why” this gender disparity in fish losses is occurring. However, we must point out that this study was just one component of a much larger project, which we failed to mention in the previous draft for various reasons, and which we now describe in our new version. We have provided some background to this project with references in Section 3: Materials and methods (lines 205 – 216). Some of the “causal links” the reviewer would like to see in this study were captured under a different instrument called “the Women’s Empowerment in Fisheries Index (WEFI)”, which can be accessed here: https://doi.org/10.1080/09718524.2020.1729480 - and which we have included now as a citation in this study.  

We had previously included part of this WEFI assessment in this paper to help explain some of the gendered differences, though we could only verify that the fishery, indeed, had unequal gender dynamics. We were never able to statistically verify a causal link (i.e. we were able to prove time allocation differences between women and men, but we were unable to prove that these time allocation differences physically impacted on fish handling). We decided therefore, to remove the WEFI and place it in a separate paper (see above). The problem is that the paper mentioned above is light on the fact that there is a huge gender disparity evident in the amount of fish handled and lost by women and men in this fishery, hence the need for this paper. We decided to showcase this by focusing on the “exploratory and descriptive” elements of the QLAM and EFLAM approach and went back to our qualitative analysis and extrapolated these results to existing studies done in the fish value chain and broader gender literature (much of it conducted in the Barotse Floodplain – see lines 1045 - 1054 for some minor additions).

We regret that we could not probe further on “why” these differences occurred since the EFLAM was an exploratory tool used at first to frame the QLAM and the WEFI, as well as to identify the participants for the larger pilot study who were to test new processing technologies. We agree with the reviewer that the “why” is only insinuated in this manuscript rather than fully verified. We have attempted to make linkages to other literature more explicit to help readers as per the Reviewer’s suggestion. Regardless, the power of this paper is to showcase the oft-missing gender dimension in post-harvest loss assessments, and we hope that, should it be published, future studies will see the need to include gender as a crucial variable and get to the bottom of the story. Due to the reviewer’s excellent comments we now feel more confident to showcase our larger approach and highlight the shortcomings but also the relevance of this paper. We feel more self-assured now that this paper, despite its gaps, helps to tell a powerful story and we sincerely thank the reviewer for these comments.

Reviewer 2 Report

1. The abstract should indicate why female processors experience higher losses than men. The abstract should also separate and unpack technical and social constraints. The recommendation should be specific in terms of how the constraints should be addressed.
2. Lines 67-70. "Sex-disaggregated......" is not clear.
3. Line 71. The research question should be formulated in such a way that it can be answered with a "Yes" or a "No"
4. Line 139. Indicate how the rest of the paper is organized.
5. Section 3.1. Data collection and analysis. This is the major weakness of the paper. The statistical method used for sampling fishers/participants should be properly described. The sample size should be based on statistical methods. The paper should just stick to the qualitative approach.
6. The authors need to read and understand value chain analysis and why it is carried out. The presentation and application is too casual for a journal article
7. Results. The results are poorly presented and are less informative. The paper looks like an undergraduate research project. The paper should have analyzed and summarized the losses occurring at each key node of the value chain. In addition, the analysis should have been grouped into two scenarios: (1) when water levels are low; (2) when water levels are high. Each scenario should be separately combined with dried fish or fresh fish.
8. The quantitative results are not useful because they are based on a flawed sampling technique.

Author Response

We sincerely thank the Reviewer for these constructive comments. We have tended to each comment in the same numbered order that the Reviewer provided:

 

1. Thank you for this comment. We have updated the abstract to explain why female processors experience higher losses. We have tried to unpack the technical and social constraints by showing examples, however, since the abstract is limited to 200 words, we feel that this is better unpacked in the discussion.
2. Sex-disaggregated means that this study was disaggregated by the experiences of females and males. We have tried to make this clearer. We provide a reference for why this is vital in any study of fisheries (Lines 116 – 117)
3. Thank you for this comment. We have changed this to: Do women experience more losses than men in the Barotse Floodplain? (Line 118)
4. Thank you for the suggestion. We have provided an update as to how the paper is organized (Lines 197-203).
5. We appreciate the reviewer’s concern with the sampling and lack of representativeness and we explicitly acknowledge this concern in Section 3.2 ‘Limitations of the study’. We must strongly reaffirm two main points to support the sampling strategy and thus the overall inclusion of the quantitative data. First, we hope that the reviewer can appreciate the complexity in small-scale floodplain fisheries, such as the Barotse Floodplain, as well as the complete absence of population data for fisherfolk living within these temporary camps. People live temporarily on small islands in what is effectively an enormous swamp (around 10,750 km², roughly the size of Lebanon) and they can change location from year to year. We stress in our paper that we were unable to define this population and thus unable to produce generalizable results. Instead, we wanted to capture the descriptive reality of “fish loss and waste” as experienced by people living in the temporary fishing camp rather than the whole value chain. Secondly, the EFLAM and QLAM methods (generally used in tandem) were created by the FAO as an exploratory tool to assess and quantify post-harvest losses, often found in practitioner manuals rather than academic journals. The goal of the research was a mixed methods design, and thus the intention was simply descriptive (such as those often seen in value chain analyses). Our intention was not to do a post-harvest loss assessment of the entire value chain, but only on the temporary fishing camps, which we now make explicit in our methods section (see Lines 219-232). We use this mixed methods approach to produce some validity in the absence of statistical reliability. We were able to capture all people who were on the camps on the days we visited, and we found both women and men performing value chain activities. We assert that given the novelty in being the first sex-disaggregated study of fish losses, that the results still have immense value. We hope this highlights the importance of collecting sex-disaggregated and gender data in such assessments, which is often overlooked, and that new studies will employ more rigorous designs in the future.
6. We agree that perhaps we were not clear in our use of the term “vale chain” and we have now fixed this (line 225-228). We would like to highlight that we never actually used the phrase “value chain analysis” and never described our study as such. This was, in fact, a post-harvest assessment (or food loss and waste assessment), utilising tools developed by the FAO, where “value chain” is a concept used to categorize losses. We now explain this difference in lines 228 - -232. The term “value chain” is used often in food loss and waste assessments but this does not mean that a “value chain analysis” design is used per se, including the broader economic, social and environmental peculiarities that this pertains. We deliberately avoided framing our study as a “value chain analysis” precisely because we only aimed to provide a descriptive analysis and measurement of fish loss and waste in a small part (i.e. fishing camps) of a much larger chain that stretches across southern Africa. To validate our approach, we have included a recent article on post-harvest losses (Kruijssen et al. (2020)), which states…

“In addition, loss assessments are costly and time-consuming, so few examples of full chain assessments exist. We therefore categorize losses by stage of the value chain, rather than pull out single chains, since studies most often collect and present information in this fashion. Losses are reported for the ‘entire chain’ for studies that do not separate loss estimates by value chain stage, recognizing that these different estimates may not reflect value chains with a similar number of nodes depending on if fish are sold directly to consumers or undergo numerous processing or marketing steps”. 

7. With respect, we do summarize the losses at each node of the value chain as far as it extends on the camp (all tables and figures are presented by node and sex). We regret the reviewer’s unfortunate view that this data is only of an undergraduate standard. We must stress the enormity and diversity of the team that collected these data over such a large geographic area, and we have provided more background as to how this study forms a much larger project in Section 3 Materials and methods. We have presented the data in a new structure in our results section (Section 4), which follows some constructive feedback provided by another reviewer. Despite its shortcoming, we feel this paper still tells a powerful story that has relevance for the literature and various disciplines (gender studies, fisheries science, food science, food security). We explicitly highlight that we were unable to collect panel data of losses within the different periods of the fishing season and we stress the importance of seasonality for future studies. We do not feel that this justifies rejecting the entire quantitative dataset.
8. We appreciate the reviewer’s view; however, we must strongly argue that mixed-methods approaches are valid and by dropping the quantitative dataset we miss an opportunity to validate some of the gendered differences that were surfaced from the qualitative inquiry. We do not believe the whole quantitative dataset should be dropped because it provides verification of what was surfaced in the qualitative results, and this would nullify the whole “mixed-methods” approach.

Reviewer 3 Report

The topic of the paper is interesting, but some revisions are necessary. I highly suggest systematizing the responses from the people. Although these anecdotal responses are valid and they add very interesting points to the discussion, I think the paper would have a better structure if these responses would be dissected and key elements grouped together under distinct categories. For example: food security issues, gender roles, time scarcity, lack of training in processing, etc. Otherwise, it seems these comments are randomly added in the paper without a purpose.

Author Response

We sincerely thank the reviewer for this insightful comment. We agree that we could have done a better job at presenting the themes and narrative of the qualitative results. As such, we have followed the reviewer’s advice and made six headings separating Section 4. Results. Specifically, for the themes used in the organization of the qualitative data, we have now settled on 1) Gender roles and division of labour; 2) Decision-making and risk; 3) Gender norms and beliefs. We have reorganised and re-written parts of the results to make these themes more explicit. Particularly, we re-wrote Section 4.1 (line 376 ) to reflect a general description of the value chain and included pictures in a new Figure 2 to be able to highlight the descriptive and exploratory approach we used. Then we introduced and re-organised the text for Section 4.2, 4.3 and 4.4 (lines 434 – 818) according to the themes provided above, and which was suggested by the reviewer. Finally, we summarise the exploratory qualitative inquiry with Figure 3 in Section 4.5 (lines 822 – 911). We hope that this new structure is met with approval and we thank the reviewer again for this great comment.

Round 2

Reviewer 2 Report

Clarify whether the enumerators picked the males and females at random?

Author Response

We thank the reviewer again for the attention to detail. We see the need to highlight how people were chosen for both the EFLAM and the QLAM. Thank you for highlighting this.

We have now added a few sentences to make clear the process in two areas. For the EFLAM we added the following (Lines 274 – 277):

Meetings to introduce the research study to people living in the camps were scheduled by a DoF officer a day earlier via telephone. The research team arrived the next day and conducted the FGD with most of the people who joined the introductory meetings.

This is preceded by the sentence before which states that around 15 people made up each FGD and that 30% were females.

For the QLAM we added the following (Lines 324):

The QLAM was purposively administered in July and August 2015 to fishers, processors and traders on the six fishing camps.

And then a few sentences further we added the sentences (Lines 241-344):

Therefore, as was executed for the EFLAM, introductory meetings on the fishing camps were scheduled via telephone the day before. At the meeting the following day, enumerators introduced the study and selected all the people who were present and willing to participate in the research.

This is followed then by showing that on those days we found 176 people (33% of whom were women).

We trust that this shows how we purposively sampled those people who were present on the camps on the day and that in both the EFLAM and the QLAM this was around 30% women. Whilst this cannot be generalized, this finding makes sense since these are primarily “fishing” camps where men reside close to lagoons and sections of the river where they can catch fish, and women travel between the two sporadically carrying supplies and fish from the camps to the markets.