“Value and benefit distribution of pollination services provided by bats in the production of cactus fruits in central Mexico”

7 Despite providing important ecosystem services in both natural and agricultural systems in the 8 tropics, bats are often disregarded or considered pests; and research quantifying their importance as 9 pollinators is scarce. We quantified the value and benefit distribution of bat pollination in the 10 production of a major fruit crop in Mexico (pitayas, Stenocereus queretaroensis ). We used exclusion 11 experiments to quantify the effect of bat pollinators on crop yield and quality. We then used yield 12 analysis to assess the market value of pollination services, combined with value chain analysis to 13 assess the distribution of these economic benefits among actors. Bat pollination services to pitaya 14 production are worth approximately US$2,500 per ha through increases in both fruit yield and size, 15 with bats contributing around 40% of gross income across producers. Participation in the pitaya 16 value chain provides a key seasonal source of cash income at a time of low agricultural activity, 17 supporting livelihoods and household activities of the rural poor. However, the commercialisation of 18 the pitaya has concentrated economic benefits with privileged groups who have access to land and 19 markets. Our novel approach to valuing pollination services is transferable to other crops and 20 pollinator species to demonstrate disaggregated socio-economic consequences of losing pollinators.


Introduction 25
Pollinators provide many benefits to humans, improving food production and security, and is not officially registered with the government. 126

Pitaya value chain 127
The key stages in pitaya production are cultivation, processing (harvesting, peeling fruits, making 128 products), marketing, and consumption. Pitaya production in Techaluta de Montenegro is dominated 129 almost entirely by small commercial plantations and home gardens (Pimienta- Barrios, 1999). The 130 value chain is short, due to the high perishability of the fruit (fruits must be eaten within one to two 131 days of harvest) and subsequent localised market (Pimienta- Barrios, 1999). Most fruits are sold 132 fresh, but a small but increasing proportion is used to make products. Producers largely sell fruits 133 directly to the consumer, either at the roadside or at a market. Actors commonly have multiple 134 functions in the value chain, and the use of intermediaries (defined here as an agent that buys fruit 135 from producers to sell to vendors) is rare (see Supporting Information S1 for a more detailed 136 overview of the stages in the pitaya chain). 137

Methods and data collection 138
We conducted our fieldwork in Techaluta de Montenegro (20.074°, -103.550°) during 2016 and 139 2017. Section 3.1 summarises the exclusion experiments we carried out to generate empirical data on 140 changes in yield and fruit size between openly pollinated and pollinator-excluded pitaya crops. Next, 141 we collected quantitative production and marketing data from 61 pitaya producers (Section 3.2). We 142 combined these data to estimate the economic value of bat pollination to the pitaya sector in 143 Techaluta de Montenegro (Section 3.3). Then, to assess the distribution of economic benefits 144 resulting from bat pollination services, we analysed economic data collected through structured 145 interviews with a sample of representatives from each actor group involved in pitaya production 146 (Sections 3.2. and 3.4). 147

Effect of bat pollinators on pitaya crop yield and quality 148
We carried out exclusion experiments in 2016 to estimate crop yield under several pollination 149 systems, whereby different flowers were exposed to certain pollinators using bags of different mesh 150 sizes placed during the day or at night. This method has been used to determine effective pollinator 151 taxa in many columnar cacti species in Latin America (e.g. Molina

Data collection: economic valuation and value chain analysis 166
We identified actor groups involved in the production of pitayas in Techaluta de Montenegro using 167 semi-structured interviews with key informants, people previously identified to have expert or broad 168 knowledge about the pitaya production sector (Newing, 2010). During the production season in 169 2017, we collected contact details of potential participants from each actor group by approaching 170 actors at random in both the production area (Techaluta de Montenegro) and subsequent market areas 171 (e.g. Guadalajara). We also used a snowball sampling technique whereby existing participants were 172 asked to recommend other potential participants. Additionally, we randomly approached registered 173 producers from a list of 189 provided by the municipality. 174 We then conducted structured interviews, using a standard set of pre-prepared interview questions 175 (Appendix B). We asked participants for: characteristics of pitaya plantations and harvest; marketing 176 and fruit prices; a detailed breakdown of financial costs and time spent on pitaya-related activities by 177 both family members and employees; and details of socio-economic background. These topics were 178 selected so we could fully determine aspects of income for each actor group (Kaplinsky and Morris, 179 2001;M4P, 2008;Sanogo, 2010). To validate responses, we asked each respondent several questions 180 relating to total and monthly income, prices and profits. Interviews allowed accurate data collection 181 while allowing participants privacy to discuss personal issues (Newing, 2010). We carried out pilot 182 interviews in a neighbouring production town (Amacueca) in June 2017 to check and refine 183 interview questions. 184 We carried out 124 interviews between July and August 2017. Interviews were conducted by trained 185 volunteers and lasted between 40 minutes and 3 hours. Prior to starting the interview, we provided 186 details of the project, data storage, and issues relating to anonymity and confidentiality, and obtained 187 written consent from each participant. We had ethics approval from the University of Southampton 188 ethics committee prior to carrying out data collection. 189

Economic valuation 190
To estimate the economic value of bat pollination Vb in pitaya production, we used a production 191 value method (Winfree et al. 2011), which estimates the value of bat pollination assuming that there 192 are no substitutes. This economic value is estimated using the following general model: 193 where Vb is the economic value of bat pollination in pitaya fruit production, is the crop's 195 dependency on bat pollination (i.e. the fractional reduction in crop yield or quality in the absence of 196 bat pollinators), is crop price (expressed in Mex$ per fruit) and is crop yield (in fruits per 197 producer). 198 Our exclusion experiments showed that bat pollination affects both fruit yield ( ), and fruit quality, 199 in terms of size ( ). Thus, there are two separate elements to the crop's dependency on 200 pollination: and . We derived from the mixed effects model parameter estimates (see 201 Section 3.1), indicating the difference between pitaya fruit set when bats were excluded (diurnal 202 pollinators only) and fruit set with bats present, which varies across pitaya types k. We derived 203 from empirical data collected on changes in fruit weights in the absence of bat pollinators in 204 exclusion experiments (see Sections 3.1 and 3.3) and the subsequent impact on price, which varies 205 across producers w and pitaya type k. Hence, has two additive components: 206 where is the value of the fruit yield attributed to bat pollination (Eq. 3); and is the value of 208 the fruit quality attributed to bat pollination (Eq. 4). 209 To calculate the value of the fruit yield attributed to bat pollination for each producer, we multiplied 210 the proportion of fruits produced of each pitaya type ( ) by the crop yield dependency specific to 211 each pitaya type ( ). We then summed the change in fruit yield across pitaya types and multiplied 212 this proportion by the gross revenues from selling pitaya fruits ( ). To calculate , we then 213 summed the value of the change in yield attributable to bats across all pitaya producers (W) in the 214 study area, i.e.: 215 was inferred from total fruit production reported by the producer multiplied by the proportion of 217 the cultivar/wild cacti under production. 1 The value of therefore varies across producers, 218 depending on each producer's total fruit production for each pitaya type ( ), as well as their gross 219 revenues from selling the fruits ( ). We assumed an equal price for all fruits sold by each producer 220 (i.e. the proportion of fruits sold per variety was taken as a proxy for the proportion of revenues per 221 variety), as we did not have data on the number of fruits sold per producer in each price category or 222 per cultivar. In reality, prices received by producers varied according to both fruit size and time of 223 season; however, as producers sold the bulk of their fruits during the peak season for one price, and 224 had fruit production dominated largely by one pitaya type (and therefore of a similar size), we deem 225 this assumption defensible. 226 To calculate the value of the fruit quality attributable to bat pollination for each producer, we 227 multiplied proportion of cacti produced of each pitaya type ( ) by the crop quality dependency 228 specific to each pitaya type and producer ( ). We then summed the change in fruit quality across 229 pitaya types ( ), and multiplied this proportion by the value remaining after subtracting the value of 230 fruit yield attributable to bats from gross revenues from pitaya sales, − . To calculate , we 231 then summed the value of the change in quality attributable to bats across all pitaya producers ( ) in 232 the study area, i.e.: 233 We assigned a null value for unstudied cultivars for both increase in fruit yield and size, which 235 accounted for 13% of cacti under production overall. 236 To calculate , we first collected data on the size of ten fruits in each of the small, medium and 237 large size bands sold by the roadside in Techaluta de Montenegro in June 2018 to calibrate the 238 weight ranges of fruits in different price categories. We then compared the proportion of fruits in 239 small, medium and large size bands under the nocturnal and diurnal pollination treatments in our 240 exclusion experiments for each pitaya type, and calculated the proportion of fruits that would drop to 241 lower size bands for each pitaya type k in the absence of bat pollinators (  Blanco  0  9  24  2  9  56   Mamey  33  0  47  0  0  20   Tenamaxtle  6  25  62  0  7  0   Wild  0  0  16  0  21  63   251 The drop in size bands implies that the total value of pitaya fruits V would be lower in the absence of 252 bat pollination because the fruits would be smaller, and producers would obtain lower prices per 253 fruit. We weighted prices received by each producer at the beginning, middle and end of the season 254 by the approximate volume sold in each time-band. Dependency values were therefore specific to 255 each producer and depended on the weighted prices that each producer could negotiate at each size 256 band: for example, a producer that received the same price for large and medium fruits would have a 257 lower dependency value attributable to the decrease in fruit size in the absence of bat pollination than 258 a producer that sold large fruits for a higher price than medium fruits. We calculated by 259 multiplying the percentage of fruits that would change size in the absence of bat pollination for each 260 price-size category for each pitaya type by the difference in prices received by each producer. 261 We then summed the differences across the price-size categories (see Supporting Information S2 for 262 an example of this calculation): 263 where 0 is the fractional change in price received for each pitaya type for each producer, with 265 0 indicating the price received per fruit in the absence of bat pollination (for size band 0 ), and 266 indicating the price received per fruit with bat pollination (for size band ). is based on the 267 information in Table 1, and is the percentage difference in the number of fruits moving between each 268 size band q per variety k in the absence of bat pollination. 269 To assess the contribution of bat pollination to employment in the pitaya sector, we estimated total 270 extra jobs generated by bat pollination by multiplying the total number of employees of each 271 producer by the proportion of revenue attributable to bats . For example, we assumed that a 272 decreased revenue of 35% would result in a workforce decrease of 35%. Thus: 273 where is total extra jobs generated by bat pollination, and is the number of employees of each 275 producer. 276 To estimate the total gross value of bat pollination services to the pitaya sector in Techaluta de 277 Montenegro, we identified all likely Stenocereus queretaroensis plantations within the municipal 278 boundaries of Techaluta de Montenegro, using satellite imagery. We marked the plantations as 279 polygons and exported them to ArcGIS to calculate the area covered in hectares (Google Earth, 280 2019). 281

Value chain analysis 282
We used the data collected through interviews with different actors to understand the production, 283 processing, marketing, and consumption stages of the pitaya value chain (Appendix B). To better 284 understand the distribution of economic benefit provided by bat pollination services, we assessed the 285 proportion of income attributable to bats, profit, and hourly earnings across actors. 286 We first estimated the proportion of income attributable to bats for each actor. For all actors that 287 produced fruits themselves we extracted values for the percentage of income attributable to bats from 288 changes in both yield and quality , from our individual level data collected through 289 interview questions on production and marketing (section 3.3). A mixed model from the exclusion 290 experiment detailed in section 3.1 provided an average estimate of for individuals that did not 291 produce fruits themselves. For actors whose income depended on the quantity but not quality of 292 pitaya fruits, we assumed the proportion of their income attributable to bats was equivalent to . 293 This was assumed for waged workers (work availability depends on fruit volume, but we had no data 294 on the specific volumes of fruits of each cultivar handled by their employers) and plantation owners 295 that rented plantations to others (rent is calculated by number of fruits). For actors whose income 296 depended on both quality and quantity of fruits (e.g. intermediaries and all types of vendors), but that 297 did not produce fruits themselves, we calculated profit margins for small, medium and large fruits 298 during peak production (as the bulk of fruits are sold during this time) by subtracting costs of buying 299 fruits from prices received when selling fruits. We then inferred the overall volume of fruits of each 300 cultivar in the market from the overall proportion of each cultivar under production across our 301 sampled producers; and used data collected in section 3.3 on the proportion of fruits of each cultivar 302 in each of the small, medium and large size categories (Table 1) to estimate the overall proportions 303 of fruits in the market of each size category with and without bat pollination. We multiplied the 304 proportion of fruits in each size category by the profit margin calculated for each actor, in scenarios 305 of selling 100 fruits in both bat pollinator presence and absence, and took the difference between the 306 two as the per cent increase in profit attributable to increased fruit quality with bat pollination. The 307 proportion of income attributable to bats for product makers was assumed to be equivalent to , as 308 the prices of products did not vary according to the size of fruit used to make them. 309 We then calculated profit earned by each individual interviewed by subtracting direct costs incurred 310 by pitaya-related activities (costs of renting pitaya plantations, agricultural inputs, salaries and 311 compensations for employees or family members, marketing, transport, tools and equipment, loans, 312 buying pitayas) from gross pitaya income (the sum of any income generated by selling pitaya fruits 313 V, pitaya flowers, and/or pitaya products, as well as income generated by renting out pitaya 314 plantations). Fixed costs e.g. of establishing pitaya plantations were not included in our calculations 315 of costs and profits. For waged workers, costs (e.g. commuting, food, tools and equipment, 316 maintenance) were subtracted from the hours worked in the season multiplied by the hourly wage 317 received. 318 Finally, we calculated the profit attributable to bats by multiplying profit by the proportion of income 319 estimated to be attributable to bat pollination services. Estimates of profit attributable to bats 320 involved an assumption of constant variable costs per fruit (though we acknowledge that marketing 321 and transport costs will probably not decrease linearly with decreased production). 322 To incorporate the number of dependents reliant on pitaya-generated income across actor groups, we 323 calculated the per capita monthly income of actors by dividing monthly income by the number of 324 people living in each household. To elucidate the trade-off between profits, working hours and 325 reliance on unpaid labour by family members, we calculated the hourly wages of each actor group by 326 dividing total profit by total hours worked unsalaried on pitaya-related activities by the respondent or 327 family members; except for waged workers where fixed hourly wages received are reported. 328 To understand the importance of pitaya-generated income, we collected data on whether respondents 329 used it for direct household provisioning or were able to save or invest it for long-term benefit, for 330 example by spending it on school fees. We also asked about other income generating activities 331 throughout the year, and the proportion of yearly income generated by the pitaya. We evaluated 332 constraints to access profitable roles in the pitaya chain by combining qualitative interview data with 333 quantitative costs data. 334 We tested for differences between groups in profit, hourly wage and per capita monthly income with 335 a Kruskal-Wallis test followed by non-parametric (Dunn) pairwise tests (using R packages 'FSA' and 20 owned home gardens (some respondents produced fruit under more than one system). The 375 total area under production for each producer ranged in size from 0.03 to 12 ha (mean = 2.58 ha), and 376 fruit production ha -1 ranged from 4,200 fruits ha -1 per season to 633,300 ( Table 2). The most 377 commonly managed cultivars of Stenocereus queretaroensis were Mamey (63% of total cacti under 378 production across producers interviewed), Tenamaxtle (7%) and Blanco (7%); as well as wild cacti 379 (10%). 380 Bigger fruits command higher prices than smaller fruits (Fig. 2b). Vendors separate fruits into large, 381 medium and small categories, with some adding categories at the extreme (tiny, jumbo). There is no 382 minimum size for a pitaya fruit to enter the market. No other fruit characteristics (e.g. cultivar) 383 affected fruit price at markets we visited. Weights of small fruits measured at markets in 2018 ranged 384 between 21.7 and 42.1g (n = 10), medium fruits between 56.3 and 69.5g (n = 10), and large fruits 385 between 68.1 and 90.6g (n = 10). Fruit prices are highest at the beginning of the season (late May), 386 when there is less fruit available and consumer demand is greatest (Fig. 2b). Prices are lowest during 387 peak production (June). June; end = early July. 394 Increased fruit yield resulting from bat pollination across the 61 producers interviewed had a mean 395 total value (before costs) of Mex$39,900 per producer (range: Mex$600 to 320,300 / US$32 to 396 16,700; Table 2). The mean value of increased fruit size resulting from bat pollination was 397 Mex$39,500 (range: Mex$0 to 298,400 / US$0 to 12,500; Table 2) per producer interviewed. Thus, 398 by increasing fruit yield and size, bat pollination has a mean total market value of Mex$79,300 399 per producer, or Mex$48,400 (US$2,530) per ha (range: Mex$1700 to 246,400 / US$87 to 12,900; 400 Table 2). 401 The percentage of gross crop value attributable to bat pollination ranged from 5% to 58% across 402 interviewed producers, with bats contributing a mean 39% (± 12 SD) of gross revenues from fruit 403 sales per producer (Table 2), or 42% of total gross income summed across producers. Producers with 404 a higher proportion of Mamey and wild cacti were more dependent on bats for total income, because 405 fruit yield increased with bat pollination relative to diurnal pollination for Mamey and wild cacti, but 406 not Tenamaxtle and Blanco. Additionally, producers that received higher prices for large Mamey and 407 Tenamaxtle fruits than medium or small fruits benefited more from bat pollination, as fruits dropped 408 one or two size-price bands in the absence of bat pollination. We estimate that income attributable to bats for the 61 producers interviewed generated 430 approximately 129 extra jobs further down the production chain (e.g. peelers, harvesters), though we 431 acknowledge that job creation is not linearly associated with income. The number of paid workers 432 employed by producers ranged from 0 to 33. 433 Jobs generated by pitaya production are a chief source of employment in an area lacking many other 443 opportunities and provide an important source of income and a strategy to diversify livelihoods (see 444 Table S3 for a description of all actors and their roles). The pitaya was cited as the principal source 445 of income by 49% of respondents, though only one household was completely reliant on the pitaya; 446 all other households had multiple income streams. Participation in the pitaya chain is therefore a 447 'gap-filling activity' for most people: one that provides a seasonal income during the period of low 448 agricultural activity, thus increasing its relative importance and compatibility with other livelihood 449 activities (Marshall et al. 2006). The actor groups most heavily dependent on pitaya-generated 450 income over the year, and therefore bat pollination services, were intermediaries and market vendors 451 (an estimated 55% and 46% of yearly income respectively), with waged workers reporting between 452 15% (drivers) and 26% (harvesters) of yearly income coming from work with pitayas (Table 3). 453 However, the pitaya chain is characterised by informal, verbal contracts: just 33% of fruit sellers and 454 45% of waged workers had a contract arranged prior to the fruiting season, and all were verbal. 455 Participation in the pitaya value chain thus precludes permanent, formal work with benefits such as 456 health insurance and pensions that only accrue to workers in continuous employment, creating a lack 457 of social security for most actors. Despite this, the lack of technical entry requirements, instant 458 generation of cash at low times of the year, and higher wages relative to other low-skilled jobs, 459 makes the pitaya sector an attractive employment option for resource-poor people. Working with 460 pitayas offers a higher daily rate during the pitaya season than many other concurrent available job 461 opportunities, such as agricultural day labouring (Mex$200 per day) or jobs tending plants in large 462 greenhouses that grow berries for the export market (Mex$120 per day). 463 The discrepancy between the highest and lowest mean hourly wages of actors in the value chain 464 (Gini coefficient = 0.67) indicates inequality in the distribution of both economic benefits and labour 465 costs between actors. The low agricultural requirements of the cacti result in a low labour cost for 466 landowners, particularly those that rent plantations to others for the production season. Actors that 467 had multiple functions in the value chain, such as market vendors that produced and sold fruit 468 themselves, commonly worked very long hours of up to 22 hours a day. The mean hourly wage of 469 plantation owners who rented plantations to others was 22.6 times higher than that of peelers and 5.4 470 times higher than that of market vendors (Mex$543, Mex$24 and Mex$101 per hour respectively; 471  to bats for each actor (for waged workers, 'profit' is wage received multiplied by hours worked, minus costs), and b) the mean percentage of pitaya-491 generated income estimated to be attributable to bats for each actor group. 492 493 494  495 Wages and benefits are a major cost for all the different actors except intermediaries (Table S4). 496

Costs
Transport costs (predominantly petrol) and rent are important costs for marketing actors. The costs 497 incurred by intermediaries and market vendors are the highest, while plantation owners have among 498 the lowest costs, thanks to the low agricultural inputs required (Table S4) buying cactus branches to plant. Furthermore, there is then a lag time before fruit production of up to 504 10 years. Access to formal credit is low: six percent of waged workers had access to credit and 505 thirteen percent of non-waged workers. There was no significant difference between actor groups in 506 per capita monthly income (Table 3), though those that earned the highest (plantation owners that 507 rent their plantations out to other people, Mex$3,770 ± 1444 SE) had a per capita monthly income of 508 nearly four times those who earned the lowest (peelers, Mex$1,000 ± 198 SE), indicating that access 509 to land may be captured disproportionately by an already economically privileged group. 510 The majority of the income (84%) associated with pitayas accrues to the local community and is 511 retained as cash income, supporting household activities (Table S5). Cash income generated from the 512 pitaya was allocated to: household food (71% of respondents), rent and bills (54%), investment back 513 into pitaya or other businesses (40%), savings (37%), household goods (36%), childrens' education 514 (30%) and other uses including medical bills and paying debts (19%). Little pitaya-generated income 515 is passed onto the government (7%) as few taxes are paid; most government revenue results from 516 actors buying petrol from the state-owned distributor (Table S5). External agents, for example 517 suppliers of packaging or agricultural inputs, accounted for the remaining 9% of pitaya-generated 518 income (Table S5). 519

Profits 520
The distribution of profits between actors was unequal (Gini coefficient = 0.60). The highest profits 521 (income minus direct costs) were gained by market vendors who both produced fruits and sold them 522 directly to the consumer, achieving the highest final fruit prices (Table 3; Fig. S1). However, 523 intermediaries, producers and plantation owners all earned a higher hourly wage (Table 3)  Producers that sold peeled fruits to other vendors could earn very high profits but there was 532 substantial variation across respondents (Table 3). Profits earned by this group in our study are 533 biased by one producer that had a very high production and took the fruits to Guadalajara to sell 534 direct to market vendors; producers that sold to vendors or intermediaries in Techaluta earned much 535 lower profits. The localised nature of the pitaya market results in a good level of market information 536 throughout the chain and enables direct market access by most actors. This increases the power of 537 producers to earn a fair price and results in intermediaries being uncommon, who frequently earn 538 excessive profits in value chain assessments (Marshall et al. 2006). Nonetheless, the few 539 intermediaries active in the pitaya chain earn a high profit due to the large number of fruits traded, 540 despite earning the lowest profit margin on fruits (Table 4) and having the highest costs (Table S4). 541 A substantial part of pitaya-generated profit for all actor groups could be attributable to the impacts 542 of bat pollination on crop yield and quality ( Fig. 3a and b). Actors whose profits depended on the 543 quality of fruits as well as quantity were more dependent on bat pollination services than actors who 544 depended on quantity only, as profit margins per fruit decreased with fruit size (Table 4), and fruits 545 were smaller in the absence of bat pollination. Intermediaries, and ambulant, roadside and market 546 vendors had the largest mean percentage of profits attributable to bat pollination (62, 56, 47 and 46% 547 of profits respectively; Fig. 3b). Actors with the highest value of profit attributable to bat pollination 548 services however, were those that earned the most from working with pitayas: market vendors, 549 producers and plantation owners ( Fig. 3a; Table 3). 550 551 Table 4. Profit margin (Mex$) per fruit of each size category during peak production (± SD) for 552 actors buying fruit to sell rather than producing their own (cost of buying fruit subtracted from sale 553 price received for fruit). 554 555

Medium
Mex$  production currently is largely small-scale and organic; however, production is expanding yearly, 626 with attempts to export the fruits internationally. Pesticide exposure can have various lethal and sub-627 lethal effects on bats, including disruption of hormones and the immune system, reproductive failure, 628 and changes to behaviour (Bayat et al. 2014). We found consumers of pitayas to have a higher 629 monthly income and level of education than any of the actors involved in the production chain (Table  630 S3), suggesting that they can afford to contribute to initiatives such as a 'bat-friendly' pitaya label 631 Additionally, the production value method assumes that crop prices will be unaffected by decreased 655 supply in the case of pollinator loss, and that farmers cannot compensate for reduced pollination 656 supply by reducing input costs or employing substitutive pollination (Winfree et al. 2011). Techaluta 657 de Montenegro contributes 40% of registered pitaya production in Jalisco (SIAP, 2018) and therefore 658 price increases may be seen with decreased fruit supply. However, the pitaya is already a highly 659 priced luxury fruit, and 67% of consumers interviewed in our study said that they would buy fewer 660 pitayas if the price increased. Input costs are already low for pitaya producers and it is unlikely they 661 could be reduced further without loss of employment. Furthermore, bats are wild pollinators that 662 cannot be replaced by a managed service, e.g. from rented bee hives; and the cost of hand-pollination 663 is likely to be prohibitive (Partap and Ya, 2012), though cost estimates are not available for this crop. The main actors at the cultivation stage are plantation owners and renters, and home garden owners 4 (Table S3 describes all actors and their roles). Additionally, day labourers are employed in the 5 management of the plantations. Agricultural inputs are low: only 24% of plantation owners 6 interviewed used pesticides (mostly a powder applied to deter ants) or fertilisers, and 16% irrigated the 7 cacti. Over half (54%) of plantation owners also grew crops other than pitayas. The timing of pitaya 8 harvest in the dry season (April -June) allows producers to grow other crops during the rainy season 9 (July -September), such as maize (22% of plantation owners) and beans (14%). 35% of plantation 10 owners also grew fruit trees such as guamuchil, mango, avocado, orange and guava.

12
The main actors at the processing stage are waged workers that harvest and peel the fruits, and product 13 makers. Pitayas are harvested manually within a day of ripening, usually in the early hours of the 14 morning to sell the fruits the same day. Pitayas are packed with foliage (e.g. alfalfa) to keep them fresh 15 for transportation to markets. The spines are removed from fruits by peelers before they are sold to the 16 consumer. Most fruits are sold to the consumer fresh, but a small proportion are increasingly used to 17 make products such as cakes, jams and punch.

19
The main marketing actors are roadside and market vendors, ambulant sellers that sell on foot or from 20 a vehicle, drivers that transport fruits, and sales assistants. The market area for pitayas is very 21 localised, with 98% of fruits sold by interviewed vendors within the state of Jalisco, both in the 22 production area and nearby towns and cities. 69% of all fruits sold by interviewed vendors were sold at 23 markets in Guadalajara, the largest city in Jalisco (80km north-east of Techaluta de Montenegro). The 24 most important of these is a traditional seasonal market dedicated to pitayas, 'Las 9 esquinas', which 25 accounted for 23% of total fruits sold. 10% of total fruits are sold to consumers in Techaluta de 26 Most fruits are sold direct to the consumer for immediate payment. Markets are informal, and 28 contractual agreements with commercial enterprises are rare, due to difficulties with transport and 29 packaging, and lag-times in payment for goods by large companies to vendors with high overheads 30 to pay. Some fruits are exchanged for goods in shops or with travelling salesmen. 34 Below is an example of how we calculated the dependency of fruit quality on bat pollination 35 for one producer, 'Producer A': 36 Producer A receives 0.5 pesos for small fruits, 3 pesos for medium fruits and 3 pesos for large fruits.

37
He produces 10,000 fruits each year. He has 600 Blanco cacti (6%), 8400 Mamey cacti (84%), 1000 38 Tenamaxtle cacti (10%) and 0 wild cacti (0%). We use these percentages of cacti as a proxy for the 39 percentage of fruits of each pitaya type. His total income from selling fruits was Mx$20,000, of 40 which Mx$2,268 can be attributed to increased fruit yield resulting from bat pollination.  the absence of bat pollination, so we would multiply this price difference by zero (Table A2). 9% of 52 the fruits moved from the large category to the medium category, but there is no difference in price 53 received by the producer between large and medium fruits. We therefore multiply the proportion of 54 price change (0.00) by the proportion of fruits to make this category change (0.09; Table A2). There  (Table A2).  Thus, the total for producer A for Blanco fruits is 0.27i.e. the value of his Blanco fruits drops 65 by 27% in the absence of bats. We get this total by adding up the changes in potential earnings due to 66 size category change (last row of Table A1).

67
To calculate the value of bat pollination resulting from increased fruit quality , we multiply the is attributable to increases in quality (size) of Blanco fruits as a result of bat pollination. 76 We repeat this process for each of the pitaya types, and sum them to get the total value associated 77 with increases in fruit size for each producer. You can see that if the producer charged more for 78 the large fruits than for the medium fruits, then there would be a higher overall difference in the 79 amount of money that he/she would lose from having smaller fruits in the absence of bat pollination.

Consumption:
Consumers 20 ▪ Consume fruits; mostly passers-by for home consumption ▪ Some restaurants buy fruits to make into e.g. drinks or desserts ▪ Mean monthly income around $12,500; a higher income than any of the actors involved in the rest of the value chain ▪ 78% had reached a level of schooling of undergraduate degree or above ▪ Indicates that pitayas are a luxury fruit ▪ Travelled a mean time of 38 minutes to buy pitayas, almost always by car n.b. Home garden owners did not sell fruits at market or roadside, nor rent plantations. Market vendors did not sell fruits by the roadside and vice versa. Plantation owners did not also sell fruits at market or by the road, or rent plantations. Intermediaries did not also rent or own plantations. Main cost categories for each actor type are emphasised in bold.   8 Ahora haremos preguntas para describir de manera detallada los costos asociados con el trabajo de la pitaya. Esta información es muy importante ya que nos permitirá entender los ingresos promedios atribuibles a las pitayas en Techaluta.