Data for this work were collected between June 2019 to December 2019 employing a cross-sectional research design informed by social network analysis approach and mixed quantitative and qualitative methods. The approach allowed mapping and triangulation of data collection methods, i.e. one method complementing the other in terms of data produced ( Bryman, 2006 ). Prior to actual data collection, pilot study was done to increase validity and reliability of the data. The pilot study was done between January 2019 to March 2019.

Network data relies on explanation about relations amongst actors, and not on individual actors. Thus, to enhance population representation of SOHFOs under the two NGOs, simple random sampling was used to select a sample of SOHFOs as a network boundary for studying networks amongst them. Again, with the interest of focusing on actor (node) level of analysis, particularly the difference in importance and influence of various SOHFOs, the chosen sample was allowed to identify the relationships they have with SOHFOs out of the sample, whether within or outside the selected NGOs. The sample for SOHFOs was determined by Yamane formula for finite population ( Yamane, 1967 ; Israel, 2003 ). Yamane’s formula is:

$n=N/1+N{\left(\text{e}\right)}^2$

where: n is the sample size, and N is estimated number of SOHFOs in two selected NGOs and e = Level of estimation (0.05)².

Therefore, $n=N/1+N{\left(e\right)}^2=286/1+286{\left(0.05\right)}^2=166.7638484=167$.

We selected a sample from a population of 79 SOHFOs and 207 SOHFOs under SAT and Floresta Tanzania respectively ( Lyassa, 2018 ; John, 2018 ), making a total of 286 SOHFOs. The sample involves 46 SOHFOs from SAT (operating from the same office) and 121 SOHFOs from Floresta Tanzania that operates from diverse offices including 48 in Moshi Rural, 48 again in Siha, 13 in Hai, and 12 in Rombo Districts with the headquarters’ office found in Moshi Urban District. The sample was obtained using a proportionate random sampling ( Hansen et al., 1953 ). The formula is as follows: $a=n/N\ast b$where: a is sample size for each point in the NGO, n is a total number of SOHFOs in a single point in NGOs, N is the total number of SOHFOs found in all points in NGOs and b is target (sampled) SOHFOs in all points in NGOs. In these SOHFOs, a total of 321 respondents were purposefully selected to represent their SOHFOs. The criteria for their selection were being a leader and knowledgeable on the matters of the study.

In the prior plan, the study expected to establish the network boundaries by selecting purposely 40 vertical supporters of SOHFOs under NGOs with strong inter-organizational networks that can result to their change as second-order sample (second sample selected to confirm information from first sample, particularly in networking actors). These organizations were responsible for confirming their networking with SOHFOs (as a general information on provision of services to such kind of FOs). However, the most mentioned organizations were not having any informed data on their relationships with SOHFOs. Thus, in collection of network data, advice given by Molina and Borgatti (2021) in the circumstance of missing some networking data due to some unavoidable reasons was used, in that case, they have advised on balancing between scientific research ethics and researcher decision for relevance and risks minimization. Thus, data from 167 SOHFOs on how they network amongst themselves and with other OHVCAs were collected with no confirmation of second-order sample. Instead, data from only managing organizations were used to serve the purpose by confirming the available networks. Thus, the most mentioned organizations responsible for managing (directing and controlling) SOHFOs were used instead of these other vertical organizations. The purposive selection of 10 managing organizations was done. These organizations were represented by 18 individual respondents. The organizations include 7 District councils with 14 individual respondents, SAT with 2, Floresta Tanzania with 1 and TOAM with 1 respondent.

The simplification of data collection and analysis to allow use of social network analysis was done. This was by preparation of the list of possible SOHFOS (network members) selected randomly from a sample frame given from the officers of the respective NGOs to be visited prior to the commencement of the data collection exercise ( Goetz et al., 2017 ). Based on the nature of the study, the respondents were told earlier via mobile phone communication to compile data on important matters of the study, i.e. networks they have built since their establishment. The study employed structured interviews and Focus Group Discussions (FGDs) for target group while semi-structured interviews were used for key informants as the main methods for data collection. Again, for the accuracy of information given, the respondents were allowed to use phone interview to contact with the knowledgeable members of their SOHFOs. These data collection methods were triangulated ( Bryman, 2006 , 2016 ) to generate comprehensive data that are credible. The structured interviews (same set of questions asked to all respondents in the same order) using interview guide attached as Appendix were conducted to 321 representative members of 167 SOHFOs. The criterion for their selection was experience and leadership in SOHFOs. These representatives gave the information on networks of particular SOHFOs with other SOHFOs or other SOHFOs outside their intermediary organization and with other OHVCAs. Semi-structured interviews (open-ended and close-ended questions were asked to get in-depth information) using interview guide were conducted to earlier mentioned 18 key informants from 10 organizations with vertical networks with SOHFOs in the OHVC and those managing the organic agriculture sector. In observing ethical issues, all respondents were asked for their consent before data collection exercise started and in FGD sessions on the use of an audio recorder.

Based on background and experiences on matters of the study, six FGD sessions were conducted using FGD guide and audio recorder. Except for Morogoro Urban and Mvomero District, which were combined, one FGD was conducted in each District. Combination was possible because participants in these areas seemed to have similar background and experiences. Each FGD session was composed of six to eight members making a total of 44 leaders of SOHFOs. Documentary review entailed the use of various sources including SOHFOs quarterly progress reports, books, research reports and journals to obtain information on networks of SOHFOs and other OHVCAs, and the role played by the actors to enhance change in such SOHFOs.

In analyzing and mapping social network data, Gephi 0.9.2 software was used ( Bastian et al., 2009 ). Amongst its peculiar attributes over some other social network analytical software is ability to produce analysed data and data visualizations; and has the capacity to handle large-size data ( Jacomy et al., 2014 ). To ensure any weaknesses that were encountered during data collection and analysis are minimized, raw data were managed in terms of being coded and cleaned ( Bryman, 2016 ). These relational data particularly centrality measures included co-existence of flow outcome perspective (position of a node based on the network based on structural properties) and induced outcome perspective (position of a node based on other attributes beyond structural properties). The measures used were in terms of Normalized Closeness Centrality (NCC), Normalized Betweenness Centrality (NBC) and Degree Centrality (DC). This has been done with the aim of assessing the networks that have been made between SOHFOs under local umbrella NGOs and other OHVCAs in their various capacities and positions in the OHVC ( Freeman, 1977 , 1979 ; Borgatti & Bonasich, 1989 ; Wasserman & Faust 1994 ; Landherr et al., 2010 ; Eboli, 2019 ; Borgatti & Everreti, 2020 ). The results from the social network analysis, particularly the NBC scores and NCC scores were used to establish the networks between SOHFOs under local umbrella NGOs and networks between SOHFOs under local umbrella NGOs and other OHVCAs in the OHVC in Tanzania. On the other hand, the scores from DC were analysed using frequencies and percentages to establish role played by OHVCAs (organizations) towards SOHFOs under local umbrella NGOs in Tanzania. The NBC, NCC and DC scores were classified into strong and weak networks. Whereby strong networks included levels of networking of 0.5 or above and weak networks included those scores below 0.5.

Closeness centrality of a node V is calculated as:

$C_C\left(V\right)={\displaystyle \sum \frac{1}{\text{Dist}\left(v,t\right)}}$

whereby Dist v and t are nodes from the graph G. According to Freeman (1979) and Wasserman and Faust (1994) , the normalized closeness centrality of a node v from a graph G is calculated as $\left(n-1\right)/\left(\left|G\right|-1\right)$ where n is the number of nodes in the connected part of graph containing the node.

Betweenness centrality of a node V is calculated as:

$C_B\left(V\right)={\displaystyle \underset{s\ne v\ne t}{\sum }\frac{\sigma st\left(V\right)}{\sigma st}}$

Whereby s = source;

t = destination;

$\sigma st$ = total number of shortest paths from node s to node t;

$\sigma st\left(V\right)$ = number of shortest paths between (s, t) that pass through/intersects v.

Since the network is directed and therefore asymmetrical, to obtain the normalized values the formula is divided by $\left(n-1\right)$ $\left(n-2\right)$ where n is the number of nodes in the network ( Freeman, 1977 ; Borgatti & Bonacich, 1989 ).

Normalized centralities were established by asking individual SOHFO (that fall in production node), which is also known as actor (node) in duration of establishment to the year 2018, whether it has given the resources (source), and the value chain actors that fall on production, packaging and storage, processing, distribution and marketing nodes have received the resources (target); or it has received the resources (target), and the value chain actors who fall on production, packaging and storage, processing, distribution and marketing nodes have given the resources (sources). The nodes (actors) were observed based on networks that allow flow of knowledge and information resources and physical resources. (This tool for data collection didn’t show vividly the categorization of value chain actors based on their stages in the value chain. This was done further during data analysis).

Apart from establishing normalized closeness and normalized betweenness centralities of SOHFOs, degree centrality was also calculated.

Degree centrality of a node V is calculated as:

$C_D\left(V\right)=\frac{k_v}{n-1}={\displaystyle \underset{j\ne G}{\sum }\frac{a_{vj}}{n-1}}$

whereby k is the degree of a node and n is the total number of the nodes.

The DC results were analysed using descriptive analysis (frequencies and percentages) to establish the role played by other actors, i.e. OHVCAs (organizations) in flow of resources towards SOHFOs for the general positive results in the OHVC in Tanzania. Data from key informants and FGDs were analyzed by content analysis through development of themes ( Vaismoradi et al., 2016 ). In theme development, data were transcribed, coded (putting texts and phrases into relevant categories) and examined to find meaningful textual strings (relating themes) to established knowledge.

Actors positioning is paramount attribute in determining the characteristics of a network ( Freeman, 1979 ; Borgatti, 2009 ; Borgatti et al., 2018 ; Alarcão & Sacomano Neto, 2016 ). Therefore, we analyzed the patterns of interaction of OHVCAs for the purpose of unveiling the way SOHFOs under local umbrella NGOs participate in the OHVC and the realization of the benefits from horizontal and vertical networks for enhanced horticultural organic sector in Tanzania. In the subsequent parts we discussed the networks patterns of various actors of SOHFOs under NGOs with other OHVCAs. The networks are particularly on dissemination of knowledge and information, and spread of physical resources to other actors in the network. As a way of managing data (numerous actors) the discussion of networks (closeness and betweenness centralities) lies on highly ranked actors using the majority rule approach as used by Niimi (2018) in large set of data in credit card transactions. Furthermore, in using the normalized centrality measures of the study, all networks in OHVCAs with 0.5 or above scores are regarded as strong networking and those below 0.5 as weak networking. Thus, discussion lies on the five highly ranked SOHFOs and five other OHVCAs with the highest scores as representatives of other actors to observe their interaction patterns and their common attributes in the OHVC. In case, two similar OHVCAs with the same role to SOHFOs (e.g. Kiziga Agrovet and Amani Agrovet) were found in high rank, only the first one with higher rank was taken as representative to observe another type of actor based on its role to SOHFOs under particular NGO.

This part discusses the networking patterns of SOHFOs in disseminating knowledge and information and spreading physical resources. This is done with the aim of establishing how close SOHFOs are to other OHVCAs. The results in this part are from two local umbrella NGOs which are SAT and Floresta Tanzania. Since networks in this study implies exchange of resources (all initiatives made by SOHFOs to network (including exploring of markets from abroad by involving SOHFOs representatives by SAT and Floresta Tanzania NGOs) but have failed to result into exchange of resources are not part of this discussion.

To obtain the general picture of the ability of SOHFOs at SAT to transfer resources easily, i.e. to disseminate knowledge and information to other OHVCAs, we ranked them to obtain those with strong and weak networking. Results ( Table 1 ) show that in dissemination of knowledge and information at SAT, the leading SOHFOs are Twende pamoja-Kimambila, Upatacho-Langali, Tumaini Menge-Vianzi, Tushikamane-Kifulu and Lamka Mangala-Ludewa with higher closeness centralities of 0.548, 0.548, 0.541, 0.530 and 0.530 respectively. These SOHFOs have strong networks with other OHVCAs particularly SOHFOs in disseminating required knowledge and information. For Floresta Tanzania, in dissemination of knowledge and information, out of all 178 OHVCAs only 1 (0.6%) OHCVA had strong networks and the rest 177 (99.4%) OHCVAs including all SOHFOs had weak networks. In such circumstance, the leading SOHFO is Ufunuo-Makiwaru with the weak networks expressed in terms of closeness centrality of 0.475.

NB: Level of networking is denoted by 1 = closeness degree scores of 0.5 and above = strong networks and 2 = closeness degree scores of below 0.5 = weak networks.

The findings suggest that most of SOHFOs under local umbrella NGOs have limited capacity to disseminate technological/marketing knowledge and information to other OHVCAs particularly their fellow producers (SOHFOs). Some of the methods of networks used in exchange of knowledge and information include meetings, short seminars, observations, demonstrations and farmer study tours. However, only few actors are involving themselves in enabling SOHFOs or use methods that can allow networks between SOHFOs. This consequently affects their capacity to create higher products standards and participate in the marketing campaigns initiated by themselves.

Moreover, SOHFOs which are active in disseminating technological knowledge and information to other OHVCAs are the ones with experience in virtual objects such as web platforms that are strategically employed to enhance networking between FOs. The web platforms including WhatsApp are mainly used by SOHFOs in exchanging technological knowledge and scanty on marketing information. Others are those SOHFOs established in agro-ecological settings where there is water availability and with high geographical proximity (low functional distance between SOHFOs). This is because the availability of water increases the duration for a SOHFO to work in organic horticultural production and therefore results in more networks.

Again, to obtain the general snapshot of the ability of SOHFOs at SAT to share physical resources easily, we ranked them to obtain those with strong and weak networks. Results ( Table 2 ) indicate that SOHFOs Umoja Mafumba-Mkuyuni, Nguvu Kazi-Lukonde, Tutogole-Kiloka, Vijana Bomba-Kungwe and Twende Pamoja-Kimambila had closeness centrality measures of 1, 1, 1, 0.480 and 0.472 respectively. Though not all highly ranked SOHFOs had strong relationships, the aforesaid SOHFOs had higher ability to spread farm inputs (including seeds and seedlings) to other SOHFOs and mainly organic horticultural products to other OHVCAs. All these SOHFOs except Tutugole-Kiloka are in the WhatsApp group project. This suggests that flow of information and knowledge facilitates flow of physical resources. In spreading of physical resources by SOHFOs at Floresta Tanzania the findings indicate that, Siangicha-Mrimbo Uuwo, Kimangaro-Kimangaro, Nuru-Kokrie, Ebeneza-Lawate, Shalomu-Mlangoni are among SOHFOs with higher closeness centralities of 1 each.

NB: Level of networking is denoted by 1 = closeness degree scores of 0.5 and above = strong networks and 2 = closeness degree scores of below 0.5 = weak networks.

This shows that, some SOHFOs have strong networks at Floresta Tanzania to spread physical resources including farm inputs to their fellow SOHFOs and to spread mainly organic horticultural products to other OHVCAs. The findings suggest that generally, only some of SOHFOs under local umbrella NGOs have strong ability to spread inputs and supply implements (organic seeds, seedlings and organic matters for production of manure) to other SOHFOs and mainly market their organic horticultural products (including Chinese cabbage, sweet peppers and collard greens) to other OHVCAs. The results are in line with the FGD findings where FGD participants argued that:

“There is problem in marketing of organic horticultural products for SOHFOs under local umbrella NGOs due to minimal networking amongst us. Use of deliberate networks of SOHFOs to capture the marketing opportunity obtained by individual farmer in a single SOHFO can be amongst possible solutions” (FGD, Mvomero District, December 2019).

Moreover, most active SOHFOs in spreading physical resources (mainly organic horticultural products) to other OHVCAs are found in agro-ecological settings with water availability or experience with virtual objects such as web platforms that are strategically employed to enhance networking between them. As earlier stated, the presence of water enables SOHFOs to spend more time in production of organic horticultural products and consequently more networks.

In observing the consequences of networking amongst SOHFOs in realizing value chain benefits, participants in FGD concurred on how crucial the arranged efforts in SOHFOs under local umbrella NGOs could increase their sales and decrease their costs of production:

“In a situation whereby SOHFOs unites and interacts for exchange of information, horticultural products, farm inputs during production or marketing of their products; negotiations could be stronger due to power of collective bargaining and costs of production and marketing can decrease” (FGD, Hai District, October 2019).

The findings contend that the minimal networking between SOHFOs under local umbrella NGOs reduces their capacity to cut their transaction costs, increase their bargaining power and their economies of scale in purchase/sale of their goods and hence decrease the competitiveness of their horticultural products. The findings contrast the evidence from Iringa, in Tanzania where strong networks of SOHFOs and NGOs have enabled them to certify, purchase, process and export organic pineapples via Dabaga fruits and vegetables canning company ( Temu & Marwa, 2007 ).

The results ( Table 2 ) also indicate some influential OHVCAs such as Floresta Tanzania to have lower closeness centrality compared to others in terms of physical resources networking. As argued by Newman (2005) , the closeness centrality is emanated from the length of paths of actors that are interacting. In a circumstance that, the networks to various actors can be huge or sparingly distributed, the shortest path to those nodes must be different ( Landherr et al., 2010 ). For instance, one of the reasons that empirically have contributed to the position of the Floresta Tanzania is its less involvement in sales and marketing of horticultural products of its SOHFOs. However, it should be noted that results at closeness centrality measure are not directly and simply interpreted compared to degree centrality measures. The same argument is recurrently made by Freeman (1979) and Wasserman and Faust (1994) who asserted that if the node is somehow disconnected (not fully connected), its value of closeness centrality will be computed with regard to each connected part separately. This consequently will lead to numerous gauges based on the size of those specified parts.

To present networks in diverse forms, we mapped the network of OHVCAs in SAT and Floresta Tanzania respectively. The mapping of the network of OHVCAs was done particularly by observing the bridging/coordination function in terms of betweenness centrality of the actor. Though the network maps in Figures 3-6 show the importance of the actors in terms of degree centrality (since they are very related to bridging function), they are not part of this discussion. In directed networks an actor with high betweenness centrality (bridging function) has the following attributes. The actor may receive resources from many actors who don’t receive their resources from the same actor (good receiver). Otherwise, the actor may give to few actors but connect them to many actors that could be distant (good giver) ( Golbeck, 2013 ). In understanding the actual situation, the direction of flow of resources was the key.

The results, as shown in Figure 3 and Figure 4 indicate that in bridging flow of knowledge and information; and physical resources at SAT, there is no SOHFO with high betweenness centrality of more than 0.5. Despite the fact that all SOHFOs have low betweenness centrality, SOHFOs or other OHVCAs at SAT with leading betweenness centrality in flow of knowledge and information are Twende Pamoja-Kimambila, Upatacho-Langali, Tushikamane-Kifulu, Nguvu Kazi-Tulo and Masimbu Station-Kimambila. For physical resources flow, the SOHFOs with leading betweenness centrality are Tughetse-Langali, Twende Pamoja-Kimambila, Jitahidi-Kiloka, Vijana Bomba-Kungwe and Tujikomboe-Mfumbwe. Again, the SOHFOs using virtual objects are the ones with the leading betweenness centrality in both knowledge and information resources and physical resources. For the case of other OHVCAs, SAT is the organization with the highest betweenness centrality in flow of knowledge and information resources and physical resources.

In SOHFOs at Floresta Tanzania, results ( Figure 5 and Figure 6 ) show that in bridging flow of knowledge and information; and physical resources there are no SOHFOs with high betweenness centrality of more than 0.5. Furthermore, the leading SOHFOs with higher betweenness centrality in flow of knowledge and information include Ufunuo-Makiwaru, Alfa-Koboko, Mshikamano-Lyasongoro, Faraja-Samanga and Mashora Da-Kware. In bridging flow of physical resources, SOHFOs with leading betweenness centrality includes Nuru-Kokrie, Ebeneza-Lawate, Shalomu-Mlangoni and Mahida-Nguduni. Amongst OHVCAs, Floresta-Tanzania has high betweenncss centrality in flow of knowledge and information. For the case of flow of physical resources, again there are no OHVCAs with high betweenness centrality. However, Mwika Market, Africado-Sanya Juu and Oloshaik-Magadini dispensary are OHVCAs with higher betweenness centrality.

The findings (from Figures 3-6 ) indicate that SOHFOs under local umbrella NGOs allow passage of knowledge and information, and physical resources. However, there is no SOHFO with strong networks in terms of bridging function in flow of knowledge and information; and physical resources to other OHVCAs. This may be caused by the fact that though both NGOs have done great job to promote networks of SOHFOs, there are no institutionalized methods established by NGOs or SOHFOs themselves that allows strategic networks amongst SOHFOs; and between SOHFOs and other OHVCAs for intended flows of network resources. In comparison between capacity to spread physical resources, disseminate knowledge and information and bridging (controlling) physical and knowledge and information resources, SOHFOs under NGOs have weaker networking capacity in bridging (controlling) resources rather than in spread physical resources and disseminate knowledge and information. This indicates somehow evenly distribution of resources within the SOHFOs under NGOs which leads on structural equivalence in flow of resources among various SOHFOs. At the same time, it shows failure of NGOs to build capacity of SOHFOs to recognize, invest and share peculiar opportunities they can have. For instance, some of the SOHFOs were networking with companies that could buy their organic horticultural products that is Africado-Sanyajuu, Seeds of Expertise for the Vegetable Sector of Africa (SEVIA)-Moshi, Home Veg Exporting Company-Arusha. Furthermore, other SOHFOs are nearer to the market places, or networking with remarkable actors in input supply (opportunities that are not known or available to other SOHFOs). The presence of mechanisms that could allow shift of these opportunities is paramount. In actual sense, in the first place the mechanisms will create structural holes (presence of SOHFOs with strong networks in controlling resources in the network, however in a long run the resources will be evenly distributed again. This is the case if social capital, which refers to the capacity to receive resources is built among SOHFOs).

SOHFOs with water availability in their agro-ecological settings or involved in WhatsApp group project portray more ability in coordination/bridging function. Despite the fact that the results promote the use of virtual aids, particularly WhatsApp groups that are strategically deployed as means of communication; it should be noted that the use of the virtual aids without working on emerging challenges in the system may bring unintended effects. This was raised by FGD findings, where FGD participants argued that:

“Use of WhatsApp groups has enhanced the knowledge of SOHFOs’ members on fighting against crop diseases, raise awareness on organic practices and sometimes realizing possible markets for their crops. However, the method can face a number of challenges like failure to use smart phones properly by some of members of other SOHFOs, poor networks that sometimes is caused by type of phone or line used and false pictures especially in environment where the officer is calling the SOHFO member and ask him/her to send a picture of their products while he/she is far from the farm. It is better to blend the WhatsApp groups with effective physical officers’ visitations which is also currently used by NGOs can help as a way of solving the issue” (FGD, Mvomero District, December 2019).

The study findings are in line with those of Lemma et al. (2008) who did their study on Ethiopian smallholder dairy value chain for innovation system. They argued on importance of intensifying Information and Computer Technology (ICT) infrastructures for enhancing inter-organizational networks in the value chain. Despite the fact that value chains for various products are different, it is plausible that their findings can apply to SOHFOs under NGOs and their OHVC in Tanzania. Other OHVCAs which portray more ability in coordination/bridging function include local agrovets, business companies, i.e. Dynamo-Arusha (that deals with vegetable and fruits drying machine), Africado-Sanya Juu (for purchasing avocado) and local markets.

When focusing on various discrepancies found between the two NGOs so as to get more insights on context variability, a number of issues can be raised. Results ( Table 1 and Table 2 ) show that in overall comparison amongst OHVCAs, SAT and Floresta Tanzania, have higher closeness centrality measures of 0.721 and 0.858 respectively in flow of information and knowledge. Again, in flow of physical resources, similarly SAT and Floresta Tanzania have higher closeness centrality measures of 0.808 and 0.792 respectively amongst all OHVCAs. The findings suggest that, amongst OHVCAs, local umbrella NGOs which act as intermediary organizations are close to SOHFOs and therefore are key actors in disseminating various kinds of knowledge and information and spread numerous physical resources to SOHFOs. This indicates a need for various OHVCAs who can strengthen inputs supply, production, packaging storage, processing distribution and marketing. The findings concur with those of Phuong (2018) who asserts that, NGOs were key actors in Thailand to support local farmers in adopting organic farming practices and employing standards and certification systems.

When observing the little discrepancy between local umbrella NGOs in flow of knowledge and information resources; and physical resources, in flow of knowledge and information resources the closeness centrality of SAT was lower compared to that of Floresta Tanzania. This is due to SAT having more organizations including FOs, local and international NGOs which while they give their knowledge and information resources to SOHFOs at SAT, they did not receive any of such resources. On the other hand, Floresta Tanzania has lower closeness centrality in flow of physical resources compared to that of SAT. Again, this was consequently brought by Floresta Tanzania giving physical resources to SOHFOs but not receiving any from those SOHFOs.

In spreading physical resources at SOHFOs of local umbrella NGOs, results ( Table 1 and Table 2 ) show that at SAT in the networks of SOHFOs with other OHVCAs, most of OHCVAs including SOHFOs had weak networks. In Floresta Tanzania, in spreading of physical resources between SOHFOs and other OHVCAs the findings ( Table 1 and Table 2 ) show that for OHVCAs including SOHFOs, comparatively to SAT most of them had strong networks. Based on the findings, at Floresta Tanzania more SOHFOs have strong ability to spread inputs and implements supply (whether hoes, watering facilities, organic seeds and organic matters for production of manure) and marketing of organic horticultural products (including Chinese cabbage, sweet peppers, carrots and collard greens) to other OHVCAs than at SAT. One reason for such discrepancy is different ways of handling physical resources in the said NGOs. While both NGOs were involved in providing farm inputs to their SOHFOs, Floresta Tanzania had not been purchasing the organic horticultural products from SOHFOs. This consequently has influenced the nature of SOHFOs at Floresta Tanzania in aggressiveness if seeking markets for their products. For the case of SAT, SAT purchases directly some of the organic horticultural products from SOHFOs. Consequently, this resulted into SOHFOs at Floresta Tanzania increasing their networks with other actors who could purchase their products. The findings suggest that when local umbrella NGOs strengthen SOHFOs to sell their products to other OHVCAs, they can produce many networks and thus resulting into better positioning of SOHFOs in the OHVC. Another factor that has helped SOHFOs at Floresta Tanzania to market their organic horticultural products is availability of private business companies including Seeds of Expertise for the Vegetable Sector of Africa (SEVIA)-Moshi, Home Veg Exporting Company-Arusha, Africado-Sanya Juu and Serengeti Fresh. This shows the importance of various stakeholders to create conducive environment for SOHFOs that allows them to access various OHVCAs for remarkable results.

The actors mentioned to work with SOHFOs are identified and the role they play while participating in flow of knowledge and information; and physical resources is well articulated. Networks amongst SOHFOs are not covered here, since they were largely discussed in previous parts. Furthermore, in flow of physical resources, some of the activities of value chain nodes (for example distribution) are not well covered in networks. This is due to the fact that, though SOHFOs members are transporting their organic horticultural products, they are not aware of any transport agent who is commonly used to distribute their products except for local umbrella NGOs.

The results ( Table 3 ) indicate that there was a flow of knowledge and information between SOHFOs under NGOs and other OHVCAs, including higher learning and research institutions; local governments; local, international and foreign NGOs, associations, business companies; and local agrovets. However, most of the other OHVCAs including certification and standardization organizations (on participatory guarantee system as locally developed certification system), particularly TOAM, have managed to interact with below 50% of the SOHFOs under NGOs. Only local umbrella NGOs were able to interact with all SOHFOs. Furthermore, most organizations have disseminated the knowledge and information on organic practices to SOHFOs that mainly fall under production as shown in Table 3 . Few were focusing on inputs supply, processing and marketing information and none was disseminating knowledge and information on distribution, packaging and storage of organic horticultural products produced in SOHFOs. This makes the activities found in node 5 related to marketing and distribution of the OHVC to be fulfilled partially, whereby information is disseminated on marketing and not on distribution. The findings imply that other OHVCAs, particularly local umbrella NGOs are the key actors with strong networks with all SOHFOs. Furthermore, most of the knowledge and information on organic practices disseminated to SOHFOs is mainly focused on production. This is revealed by the knowledge found in SOHFOs on how to practice organic production and introduction of new types of vegetables and fruits that were not present prior to NGOs interventions. The fact that there is scant information that is disseminated on inputs supply, processing and marketing and none on distribution, packaging and storage, connote limited potentiality for growth of SOHFOs in such areas.

The subsequent numbers denote the type of knowledge and information provided by the OHVCAs particularly the fruits and vegetables and the role they play in supporting SOHFOs in in terms of 1 = inputs supply; 2 = production; 3 = packaging and storage; 4 = processing and 5 = distribution and marketing. Key for networking = above 50% networking = Strong networks, below 50% networking = weak networks.

The findings concur with the results of interview made to DCDO in Rombo District who said that:

“Floresta Tanzania is familiar to me, its leaders came and introduced themselves to me; they have been giving written reports. For effective results, I recommend relationships via meetings and special platform to harmonize availability of various knowledge and information; and physical resources for SOHFOs. Mobilization of efforts between District Council, Floresta Tanzania and other actors in the sector is important. The relationships provide a chance of diversity and avoiding duplication. One advantage of the platform is to seek markets for SOHFOs” (KII, Rombo, October 2019).

This continues to indicate that, there was poor value chain coordination due to lack of a platform that can harmonize the flow of knowledge and information; and physical resources flow between SOHFOs and other OHVCAs in OHVC. Again, the study findings concur with those of Lemma et al. (2008) who did their study on Ethiopian smallholder dairy value chain for innovation system. They authors argued on weak networks in terms of poor organization between local dairy producers and other vertically coordinated value chain actors which hampers more progress in the sector.

Nevertheless, some few concerns were evident in production information disseminated, including presence of reject fruits and vegetables. For instance, in an interview with one SOHFO member, it was revealed that, out of 700kgs of avocados that were to be sold by that SOHFO, 200kgs were rejected. The findings are in line with the study by VSO-ICS Volunteers (2015) which shows that smallholder farmers on horticultural value chain in Zanzibar face challenges of unreliable information on marketing, prices uncertainty and access to low-priced way to transport their products to the markets. With these findings, it is plausible for SOHFOs in the similar value chain to come across the same challenges. The challenges of timely and more relevant information on marketing consequently affect their capacity in various ways including provision of quality and marketable products. This is also argued by Liverpool-Tasie et al. (2020) and Saarelainen and Sievers (2012) who assert that SFOs’ operations in developing countries including Sub-Saharan Africa countries have continued to be affected by various matters including capacity constraints in terms of technical knowledge, poor participation in the value chain, high transaction costs, financial constraints, and buyers’ market power. As earlier observed, most of the said challenges in SFOs are perpetuated by information asymmetry and poor access to the reliable market. Again, this is expected to be so for the SOHFOs under NGOs in Tanzania.

In flow of physical resources ( Table 4 ) (including giving inputs, i.e. hoes, watering facilities, organic seeds and purchasing of organic products, i.e. Chinese cabbage, sweet peppers, carrots and collard greens), the results show that while local umbrella NGOs interact with all SOHFOs, most OHVCAs including local markets, local agrovets, learning institutions, international NGOs, associations, business companies, sole proprietors, secondary and primary schools, dispensaries, local food vendors and religious institutions have received or given physical resources to less than 50% of the SOHFOs. With regard to OHVC activities in flow of physical resources as shown in Table 4 , most of OHVCAs have focused on inputs supply, production and marketing. Few have dealt with processing and distribution and no organization has dealt with packaging and storage.

The study findings show that in flow of physical resources, local umbrella NGOs are key actors in interacting with SOHFOs followed by local markets and local agrovets. This implies that, SOHFOs under NGOs receive or give physical resources that were mainly focusing on production, inputs supply (including hoes, watering facilities, seeds) and marketing (organic horticultural products). Furthermore, the results indicate scant services by OHVCAs which are available for SOHFOs on processing and distribution of their horticultural products. In the same vein, there is lack of services and facilities by OHVCAs dealing with packaging and storage of horticultural products. The findings concur with those of Temu and Marwa (2007) who compiled evidence on changes in governance of global value chains of horticultural products in Sub-Saharan Africa countries and argued that challenges on availability of facilities, equipments and other infrastructures for horticultural production affect the participation of smallholder farmers in the global value chain. With these findings it is plausible for the challenges to happen to SOHFOs in the OHVC whether for countries’ internal or external markets of organic horticultural products. Again, similar findings in other way are insisted by Liverpool-Tasie et al. (2020) , who argued about persistence of problems such as high transaction costs, buyer’s market power, poor involvement of SFOs in the value chain is accelerated by lack of reliable facilities, equipments and infrastructures.

In flow of physical resources ( Table 4 ) (including giving inputs, i.e. hoes, watering facilities, organic seeds and purchasing of organic products, i.e. Chinese cabbage, sweet peppers, carrots and collard greens), the results show that while local umbrella NGOs interact with all SOHFOs, most OHVCAs including local markets, local agrovets, learning institutions, international NGOs, associations, business companies, sole proprietors, secondary and primary schools, dispensaries, local food vendors and religious institutions have received or given physical resources to less than 50% of the SOHFOs. With regard to OHVC activities in flow of physical resources as shown in Table 4 , most of OHVCAs have focused on inputs supply, production and marketing. Few have dealt with processing and distribution and no organization has dealt with packaging and storage.

The subsequent numbers denote the type of knowledge and information provided by the OHVCAs, particularly the fruits and vegetables and the role they play in supporting SOHFOs in in terms of 1 = inputs supply; 2 = production; 3 = packaging and storage; 4 = processing and 5 = distribution and marketing. Key for networking = above 50% networking = strong networks Below 50% networking = weak networks.

The study findings show that in flow of physical resources, local umbrella NGOs are key actors in interacting with SOHFOs followed by local markets and local agrovets. This implies that, SOHFOs under NGOs receive or give physical resources that were mainly focusing on production, inputs supply (including hoes, watering facilities, seeds) and marketing (organic horticultural products). Furthermore, the results indicate scant services by OHVCAs which are available for SOHFOs on processing and distribution of their horticultural products. In the same vein, there is lack of services and facilities by OHVCAs dealing with packaging and storage of horticultural products. The findings concur with those of Temu and Marwa (2007) who compiled evidence on changes in governance of global value chains of horticultural products in Sub-Saharan Africa countries and argued that challenges on availability of facilities, equipments and other infrastructures for horticultural production affect the participation of smallholder farmers in the global value chain. With these findings it is plausible for the challenges to happen to SOHFOs in the OHVC whether for countries’ internal or external markets of organic horticultural products. Again, similar findings in other way are insisted by Liverpool-Tasie et al. (2020) , who argued about persistence of problems such as high transaction costs, buyer’s market power, poor involvement of SFOs in the value chain is accelerated by lack of reliable facilities, equipments and infrastructures.