Introduction

“The explanation of the amazingly high standard of rice cultivation in Bali”, wrote a senior Dutch colonial officer, F.A. Liefrinck, in 1887, “is to be found in Montesquieu’s conclusion that ‘the yield of the soil depends less on its richness than on the degree of freedom enjoyed by those who till it’” (1969:3). Liefrinck echoed the conclusion of Sir Stamford Raffles, who visited Bali in 1811 and was surprised to discover that the Balinese kings were merely one group of landowners among many others: “The sovereign [that is, the Raja of Buleleng] is not here considered the universal landlord; on the contrary, the soil is almost invariably considered as the private property of the subject, in whatever manner it is cultivated or divided” (Raffles 1817:234). But their impression that Balinese kings bore little responsibility for the island’s spectacular rice terraces was later disputed, launching a debate on the respective roles of farmers and kings that has continued for more than a century. The Dutch completed their conquest of the island in the first decade of the twentieth century. In Priests and Programmers (1991), one of us (Lansing) argued that the “restoration” of control over irrigation by the Dutch successors to the Balinese kings provided an attractive justification for colonial rule, and the role of the kings in irrigation and agricultural taxation became a frequent topic in colonial journals. As time went on this issue became moot, and disappeared altogether after the retreat of the Dutch from the Indies in mid-century. But later on the advent of the “Green Revolution” in agriculture triggered a new round of debates over Balinese irrigation. In the 1970s a national campaign of “massive guidance” in agriculture required Balinese farmers to give up the right to set their own irrigation schedules, and practice continuous cultivation of Green Revolution rice.

In earlier publications we have suggested that the ensuing chaos in irrigation inadvertently exposed the functional role of networks of water temples, a point also made by the Balinese sociologist Nyoman Sutawan and his colleagues in the Faculty of Agriculture of Bali’s Udayana University (Sutawan et al. 1985). We argued that water temple networks enabled the farmers to implement staggered irrigation schedules that create an optimal balance between water sharing and pest control (via synchronized fallow periods) for entire watersheds. This idea has attracted several thoughtful critiques, to which we will respond here.

Lansing’s fieldwork in Bali began in 1971, when the Green Revolution was just beginning, and was mostly carried out in small irrigation systems in the upper reaches of the Oos and Petanu watersheds in central Bali. Critics have questioned whether our analysis is relevant to conditions that existed before the Green Revolution, or to the larger irrigation systems of southern and western Bali. If (as we have argued) Balinese irrigation systems developed from the bottom up through a process of trial and error, we should not be surprised to see variation from one watershed to the next, and indeed the breadth of local variation in governance systems is a prominent theme in the colonial literature on Bali. Thus, V.E. Korn suggested that “…the royal interference is not to be regarded as the beginning of the whole system, but as the completion of an existing system of irrigation” (1932: 271). Although Lansing dismissed this observation in 1991, we were prepared to accept that new evidence could reopen the question of the role of kings, farmers and water temples in Balinese irrigation. Returning to Bali in the summer of 2010, we began to follow up on the specific questions and issues raised by our critics.

The Role of Water Temples in Pest Control

We begin with A. P. Vayda and D. Falvo’s questions about the role of water temple networks in pest control. This point is central to our model: we contend that Balinese farmers knowingly attempt to control rice pests by synchronizing irrigation schedules so that swathes of flooded paddies become fallow at the same time, depriving rice pests of their habitat. In 2009, Vayda offered the following critique:

“Most important is evidence suggesting that both green leafhoppers and brown planthoppers, the two main insect pests of Balinese rice fields after the Green Revolution of the 1960s, were much less a threat to the fields in earlier times, making it unlikely that the system of coordinated irrigation schedules was, centuries ago, directed to controlling these pests…As for alternative causal possibilities to help account for whatever balance (not necessarily optimal) was achieved between water allocation and pest control either before or after the intensified rice cropping of the 1970s and 1980s, the possibilities mentioned by other critics of Lansing’s work include force or the use of force by larger and stronger downstream populations in order to get water from those upstream (Falvo 2000:645–6) and farmers eschewing continuous cropping of rice not simply or necessarily for the sake of pest control but, perhaps as much or more, for the sake of growing commercially profitable crops like chili peppers, garlic and shallots after rice harvests” (Helmreich 2000:323–4, citing Nakatani 1999:206–7).

Vayda’s first point concerns the control of pests by synchronized fallow episodes.Footnote 1 We argue that to reduce pest losses by fallowing, farmers must take into account the characteristics of the most damaging local rice pests: how fast can they reproduce and migrate, and how much damage can they cause? Synchronized fallowing has implications for water sharing because peak irrigation demand in rice paddies occurs at the beginning of the cultivation sequence. Water sharing and pest control by synchronous harvests are thus opposing constraints. Consequently, irrigation schedules in a Balinese watershed are like the pieces of a jigsaw puzzle. If they fit together well, the resulting pattern optimizes harvests for everyone by minimizing pest infestations, and reducing water shortages. But if too many fields go fallow at the same time, there will be water shortages downstream. Conversely, if the areas of synchronized fallow are too small, pests can migrate to adjacent fields that are still in cultivation.Footnote 2 Water temple networks enable the farmers to adjust their irrigation schedules in response to local conditions, solving the jigsaw puzzle for entire watersheds that may include dozens of weirs and local irrigation systems.

Lansing’s observations began soon after the onset of the Green Revolution in the 1970s. In the 1980s he was joined by ecologist James Kremer, and they collected survey data on losses to pests from farmers in the Oos and Petanu watersheds. In the 1990s they began to collaborate with staff from the Ministry of Agriculture to conduct field studies of pest population dynamics in both synchronous and asynchronous fields in several regions of south Bali (summarized in Latham 1999). Local populations of all species of pests were estimated by biweekly sweep census, and field experiments were also undertaken to determine the three rates mentioned above (pest migration, reproduction and damage). Populations of all pests were always significantly higher in asynchronous fields. This is consistent with the observations of Balinese agricultural researchers, summarized in Lansing 1991, that asynchronous planting mandated by the Green Revolution programs triggered the observed explosions of rice pests.

But Vayda’s question is, were pests a problem before the Green Revolution? We revisited this issue, and confirmed that references to pest damage are abundant in reports from colonial officials in Bali, beginning soon after the conquest of the island. For example, a 1916 study of agricultural practices in Bali describes the sequence of agricultural rites that are still practiced today, including “njoengsoeng, so that the fields will not be destroyed by rats, crickets or worms” (Van Naerssen 1916:34). There are also frequent references to rice pests in Balinese lontar manuscripts having to do with agriculture, such as this passage from the Dharma Pamaculan: “the God of the Masceti temple, who controls the rats, must be given offerings and the God of Sakenan Temple, who controls grasshoppers, should be given offerings. If there is a problem at the weir, perform the balik sumpah ritual at the Ulun Swi temple” (Lansing 1991:64). Harvest losses from rats are often noted in the reports of colonial officers until World War II. Rats can multiply very quickly, and the traditional remedies include synchronous fallowing, rituals such as rat cremations (which sometimes accompany widespread synchronous fallows), and incentives to kill rats in the fields. In 1879, R. van Eck reported that “Now and then in Gianyar mice that destroy the sawahs or rice fields are caught in great numbers and burned in the same way as dead (human) bodies. Except two of them are left alive, who are then dressed in a little suit of white linen, and after people have bowed to them as if before a godhead, are then set free” (van Eck 1879: 125). More accounts of rat cremations in the early twentieth century are collected in an unpublished manuscript by V.E. Korn entitled “Abenan bikul” (rodent cremations) in his unpublished papers (Korn n.d.) Today, rats continue to be a problem for many subaks. In 2010, subak Jaka Dayang in Tabanan paid a bounty of 2000 rupiah per dead rat, which cost the subak six million rupiah.Footnote 3 Overall, both Balinese and Dutch records dating from before the Green Revolution contain references to harvest losses from rice pests, the role of temples in pest control, and the practice of synchronous fallowing.Footnote 4 Undoubtedly both brown planthoppers and green leafhoppers became more abundant after the onset of the Green Revolution. But rodents can be quite as devastating as insects (causing very heavy losses in some regions of Bali in 2011). Synchronous fallowing is an effective strategy to control both insects and rodents.

Vayda’s second point is to suggest two alternative explanations for synchronous irrigation schedules: either the use of force by downstream populations to obtain water from upstream groups, or the desire of farmers to plant other crops after rice harvests. He does not offer evidence that either actually occurred. Instead, the farmers spoke of returning to the traditional method of pest control by synchronized fallowing. An easily accessible example of this explanation occurs in a scene filmed in 1979 in our film “The Three Worlds of Bali”, in which the head of the village of Sukawati describes the decision of the local farmers to return to synchronized planting at the water temple Masceti Er Jeruk:

If I’m not mistaken this occurred after the pest attacks began, around '73 or '74. Just after we were released from the IPEDA [agricultural] taxes. At that time not everyone was getting a harvest. The rats were always coming. So they had a meeting and asked themselves, why are we having these loses? At first the yield was good. They were planting so quickly, rice crop after rice crop. But then the pests increased, and there were water shortages. The rats kept coming. So they had a meeting and asked themselves, why have we abandoned a legacy that was already good? And they returned to this (gestures at the temple), planting all together.

The same point is made in a 1985 report by the Department of Public Works of the Regency of Tabanan (Lansing 1991:114):

I. Background

Concerning the explosion of pests and diseases which recently attacked the rice crops, such as brown planthoppers, rodents, tungro virus, and other insects, in the Tabanan regency; and also with regard to the frequent problems which began to arise at about the same time concerning water sharing during the dry season, various groups are now urgently working to get on top of the problem. The result has been acknowledgment of the following factors which caused the explosion of pests and diseases:

  1. 1.

    In areas with sufficient irrigation water, farmers are now planting continuously throughout the year.

  2. 2.

    In areas with insufficient water, farmers are planting without a coordinated schedule.

    In other words, the farmers/subaks have ceased to follow the centuries-old cyclical cropping patterns …

We turn now to Vayda’s idea that synchronized irrigation schedules could be the result of the use of force by larger and stronger downstream populations in order to get water from those upstream. There are three problems with this explanation: the demographic premise is incorrect (many downstream subaks are small); there is no evidence that downstream subaks threaten upstream subaks; and the probability that this behavior could produce the actual pattern of synchronized irrigation schedules at the scale of whole watersheds is infinitesimal. The difficulties facing downstream farmers attempting to forcefully appropriate water from upstream farmers may be imagined by a glance at Fig. 1, which shows the location of the subaks belonging to a single dam, one of dozens along this river. If scores or hundreds of downstream subaks regularly threaten their upstream neighbors with force in order to obtain water, this has left no trace in either scholarly publications or the memory of farmers. In reality the opposite occurs: downstream subaks routinely “borrow water” from upstream subaks, a practice noted by the Dutch irrigation engineer Van Naerssen (1916:39) which continues today. For example, the head of subak Gadon 1 in Tabanan describes borrowing water from distant upstream neighbors in 1988–9. This entailed an agreement extending upstream from the Gadon damFootnote 5 to the dams of Tapesan, Sungsan and Cangi. The following year, the Gadon subaks moved up their planting schedule to allow their upstream neighbors to obtain more water during the dry season, and so repay the loan. Several farmers in Gadon and Cangi confirmed this account. They noted that between the Gadon dam and the Cangi dam there are two other dams, at Tapesan and Sungsan. So for Gadon to receive extra water from Cangi, it was necessary for Tapesan and Sungsan to allow the extra flow to continue downstream (Fig 2).

Fig. 1
figure 1

Subaks of the Mambal (Gumasih) irrigation system, corresponding to Fig. 2: Some water temples of the Mambal irrigation system

Full size image
Fig. 2
figure 2

Some water temples of the Mambal irrigation system. Not all subak temples are included

Full size image

Finally, we consider Vayda’s premise that the complex spatial patterning of irrigation schedules that exists today could have resulted from a process other than the trade-off between pest control and water sharing that drives our model of the Oos and Petanu rivers. Evidence that the farmers have this trade-off in mind when they decide on irrigation schedules is provided by our survey of 149 farmers in 10 subaks. The farmers were asked “which is worse, water shortages or pest damage?” Upstream farmers chose pests and downstream farmers water shortages; the results were significant at P < 0.00005 (Lansing and Miller 2005). That decisions based on this logic will produce an expanding network of synchronized irrigation schedules is shown in our budding model for Balinese water temple networks (Lansing et al. 2009), and by our study of the irrigation schedules of 172 subaks along the Oos and Petanu rivers (Lansing and Kremer 1993; Lansing 2006). These simulations begin with every subak following its own independent irrigation schedule. Through a process of trial and error, they respond to pest damage and water shortages by imitating the irrigation schedules of their most successful neighbors. After 6–8 years, a global pattern of interlocking irrigation schedules appears which optimizes harvest yields for the whole watershed, and closely resembles the actual irrigation schedules followed by these subaks. By “global pattern” we mean one that includes all the subaks in an entire watershed, which are interconnected by the flows of both water and pests in each local neighborhood. Our models show that globally optimal patterns exist and can easily be discovered by a process of trial and error pursued by every subak, provided they respond to pest damage as well as water shortages. The Green Revolution unintentionally created an experimental test of this model, which can be replicated by removing the pest constraint and changing the goal to maximizing the continuous cropping of rice. The immediate result, in both the model and the real world, is an explosion of pests and chaotic water shortages.

Dynastic and Colonial Involvement in Irrigation

Historian H. Schulte Nordholt raises two objections to our portrayal of the functional role of water temples:

[Lansing] emphasizes the autonomous and bottom-up nature of local irrigation management, but adds an important qualification by arguing that regional irrigation management centered upon extended temple networks which culminated in the Batur temple complex in central Bali. In doing so, he seems to downplay, or even ignore, the role played by regional dynasties and noble lords in the pre-colonial period, while he also seems to underestimate the structural changes which took place at the beginning of the colonial period during the first decades of the twentieth century.

Schulte Nordholt offers three sources of evidence for this critique: descriptions of Balinese irrigation by colonial administrators and engineers (P.L.E Happé, F.A. Liefrinc E. van Naerssen Naerssen and O.W. Sorensen); the Kidung Nderet, an eighteenth century Balinese manuscript, and interviews recorded during his own fieldwork in Bali. Schulte Nordholt’s major publications concern the history of the royal dynasty of Mengwi, the largest Balinese kingdom of the nineteenth century. Initially we were prepared to be persuaded that the rajahs of western Bali played a larger role in irrigation than in the smaller systems of Gianyar and Bangli that we studied. But the evidence from these sources appears to be inconsistent with S.N.’s conclusions. We begin by reviewing his sources.

An article on “Gegevens over Irrigatie en Waterverdeeling” was originally published by “E. v. N.” in 1916, and subsequently republished in 1918 in an Adatrechtbundel, attributed to E. J. van Naerssen.Footnote 6 This article mentions the arrival in Bali of a Dutch irrigation engineer, presumably O.W. Sorensen, who also wrote about Balinese irrigation. But to our surprise, neither van Naerssen nor Sorenson describe any role for Balinese lords or kings in the construction of dams or irrigation systems. Apart from taxation, van Naerssen’s sole reference to dynastic involvement in irrigation is this remark (van Naerssen 1918:34):

Heading the waterschap of a kingdom is the Sedahan Agoeng, presently in the service of the (colonial) government which puts him in that position. This person is in charge of getting tax revenues (landrente) and is the highest functionary to settle differences between subaks (the irrigation area that is watered by one and the same dam). The Sedahan Agoeng is assisted by pangloerahs.

Both van Naerssen and Sorensen provide extensive descriptions of the creation of new subaks and irrigation works, including maps and (in Sorensen’s case) engineering diagrams. Van Naerssen observes that small dams are adequate for small streams, but large rivers require large dams, up to 25 m in the regencies of Tabanan, Badung and Gianyar. He notes that construction of these dams requires the work of hundreds of people and may take decades: “to avoid the need to build high dams, they construct them upstream (where the flows are smaller).” The process is indeed “bottom-up” (van Naerssen 1918:29):

Should several owners of a complex of unirrigated farmlands [tegal] wish to turn these into irrigated sawah, a farmer’s association [landbouwersvereeningen] is created. A president, secretary, treasurer and several commissaries are chosen. A certain amount of money and shares in the eventual income are promised to him, who can deliver irrigation water to the aforementioned ground [sewa mara]. An association of technical people, consisting of several tunnel builders and water engineers, start to explore, bring their plan to the farmer’s association, and the conditions for an eventual construction are negotiated.

However, Schulte Nordholt’s argument is that while this method may have been adequate for the construction of small upstream dams, “large scale irrigation in the precolonial period was unthinkable without dynastic involvement.” To our knowledge, the only detailed account of the construction of a large dam is Sorenson’s description of the building of the Pejeng dam (1921). But like van Naerssen, Sorenson says nothing about rajahs or princes; instead his article is a tribute to the organizational and engineering skills of the subaks. We recently translated and summarized his account in another publication and so will not repeat this description here (Lansing et al. 2009:115–119). Van Naerssen concludes his description of Balinese irrigation in 1916 with laudatory remarks about the subaks. There is no mention of active involvement by rulers. Instead, Van Naerssen observes that “if due to lack of water not all areas can get water, then they create a turn-taking which is decided upon during the monthly meetings.” The article ends with an ironic comparison of centralized governance in Java with the ineffectual rajahs of Bali, “where popular institutions under a “bad” indigenous government were able to develop freely”(p. 39).

Schulte Nordholt’s other historical evidence for dynastic involvement in the construction of the larger dams is taken from the traditional Balinese lontar manuscript Kidung Nderet. He attaches much significance to this source, noting that “without the Mengwi dynasty the irrigation system of the Gumasih dam was simply unthinkable”; further “In a private conversation in 1991 the late professor Clifford Geertz admitted reluctantly that he would have given a different picture of the nineteenth century Balinese Negara if he had known the Kidung Nderet ”(Schulte Nordholt 2011: footnote 8: “The maintenance and repairs of the dams and canals rests with the subak members, who at the charge of the pangloerah, the head of one or more waterschaps [water districts], and under the supervision of the heads of the subaks, the pekasehs, must execute the necessary labor. The same with the construction of new dams: the execution and the costs are borne by those who later will profit from the dam.”).

Kidungs are a metrical genre of Balinese literature, rich with metaphors and allusions; interpreting them is regarded as a specialized skill. For this reason we checked our reading of the key passages pertaining to kings and dams with three Balinese kidung experts. Two episodes that occur early in the story appear to be most relevant to S.N.’s thesis; the first concerns taxes and the second the construction of the Gumasih dam. Before turning to the translation, two points of clarification are needed. First, as noted above by van Naerssen, the Sedahan Agoeng is the collector of agricultural taxes and “the highest functionary to settle differences between subaks.” Second, two kinds of taxes are mentioned in the story: ordinary taxes on agricultural land and soewinih, which one of S.N.’s sources (Happé) succinctly defines as a “counter-prestation for the use of the water that belongs to the gods”.Footnote 7

When the story begins, the greedy and unscrupulous lord Nderet is appointed Sedahan Agoeng by the young rajah of Mengwi. Earlier, for his services to the old rajah, another individual named I Made Tibung was given exemption from agricultural taxes. The new rajah is persuaded by the villainous Nderet to forget about this exemption, and hand over the right to collect these taxes to Nderet, since he is now the Sedahan Agoeng. Responding to this challenge, Made Tibung goes to see the young king and reminds him that his father’s gift of exemption from agricultural taxes was sworn and sanctified in a temple. After Made Tibung departs, Nderet returns to argue his case before the rajah:

19. It’s right for Lord King to take taxes from Made Tibung to the palace, the soewinih can be given to Made Tibung, and that will be enough for them. By the time it is gone there will be another harvest. [Lord king] can do so because the market will continue (e.g. there will be more harvests).

20. [The rajah] took back the taxes from Made Tibung. He returned the soewinih to the royal court [puri]. Made Tibung objected, realizing that the rajah had forgotten his gift.

(Later, Tibung argues his case:)

30. Like the way fleas get food, they live on a tiger, who will dare to challenge a tiger, but the tiger does not know that his blood is being drunk by little fleas, making the tiger thin, while the fleas grow in numbers.

31. The tiger remains unaware that it is sick because it has lots of fur, while the fleas are still drinking its blood and getting fat. But the tiger does not feel them, or realize that he is being eaten by fleas, used as their source of food.

Tibung’s complaints about the greed and unfairness of the blood-sucking Nderet are echoed by others in the court of Mengwi, who describe the unhappiness of the people. The story then pauses as the narrator explains the origins of the crisis with Nderet, concerning events that took place in the time of the old king:

41. I tell (now) what happened earlier, that the people of Padang Luah made a dam at Gumasih. Because everyone was working at the dam, the village was destroyed because when the enemy came everyone was gone and no one stopped them. I Made Tibung ran away.

42. Some ran north until worn out. The people working at the dam stopped because the day was old (late). All ran south, hungry and tired. At dusk they wanted to go home. After they got to Buduk they encountered many cannons of Badoeng.

The story continues with a description of an attack by the mercenaries of the king of Badoeng, and goes on for another hundred verses. There are no further references to the building of dams or royal involvement in irrigation. Hence we offer two observations: First, lines 41 and 42 explicitly describe the construction of the Gumasih dam as the work of the village of Padang Luah (see Fig. 1: map). Second, the role of the king with respect to irrigation consists in the receipt of taxes collected by his Sedahan.Footnote 8

Royal Irrigation Temples?

The second part of S.N.’s critique has to do with the respective roles of dynasties and temples in irrigation: “Dynasties, and not temples, were the key institutions that guaranteed the continuity of the system”, according to S.N. As evidence he offers the example of the irrigation system that originates at the Gumasih dam: “without the Mengwi dynasty the irrigation system of the Gumasih dam was simply unthinkable.” Accordingly we focused our investigations on this irrigation system. This dam is now known as the Mambal dam.Footnote 9 It is located on the Ayung river, and presently includes 43 subaks (Table 1). Locations of most of these subaks and their water temples are shown in Fig. 1. We begin by exploring the specific functions of these temples with respect to irrigation and pest management, and then discuss the role of the sedahans, who would have been the instruments of dynastic control over irrigation.

Table 1 Subaks and water temples in the Mambal (formerly Gumasih) irrigation system, listed according to their administrative district. Source: Sedahan Agung, Denpasar, Bali
Full size table

As is true everywhere in Bali, the farmers of this region treat irrigation water as a shared resource bestowed by the Goddess of the Lake(s). In return for her gift, subaks make annual pilgrimages to the lake temples to present offerings at her shrines. In return, they receive vessels of holy water representing her continued blessing, to be used to grow the next crops. There are four crater lakes, and which lake temple one goes to depends on where one’s irrigation water is believed to originate (Fig. 3). The Ayung river is thought to come from Lake Batur, while the Sungi river descends from Mount Beratan. The Mambal subaks happen to be located between these two rivers. Hence Mambal subaks that obtain their water from the Ayung river require holy water from the temple of the Goddess at Lake Batur. Other subaks go elsewhere: 20 Badung subaks that obtain water from the Sungi river bring their offerings to Lake Beratan; while most Tabanan subaks worship the Goddess in her temple at Lake Tamblingan.

Fig. 3
figure 3

Relationship of water temples to lakes and rivers. Subaks that obtain water from the Ayung river owe annual offerings (sarin tahun)to the Goddess at her temple at Lake Batur. A delegation of subak leaders brings the offerings from the Ulun Swi temples, and returns with holy water that is distributed to the farmers. Twenty subaks that obtain water from the Sungi river make offerings at a shrine in the royal temple Taman Ayun, and receive holy water there

Full size image

Before the Dutch built roads on Bali and put an end to the incessant armed conflicts between rajahs, the annual pilgrimages to the lake temples would have been a difficult journey.Footnote 10 This would have been particularly true for the subaks located furthest downslope, such as the Mambal subaks. For the 20 subaks most closely attached to the old Mengwi royal dynasty (not the Mambal subaks), fulfilling their responsibilities to the lake temple of Beratan was made easier by the creation of a shrine to the Deity of Mount Beratan, maintained by the dynasty of Mengwi within their royal temple, Taman Ayun. According to the chronicle of the Mengwi dynasty, this grand moated temple was constructed in 1634 A.D. Its major shrines are dedicated to the ancestors of the dynasty and to the gods most directly concerned with the welfare of the kingdom. Among them is the shrine to the Deity of the lake temple of Mount Beratan. Each year, a prince of the Mengwi dynasty accompanies a small delegation of subaks and priests on a pilgrimage to the lake temple of Beratan. They obtain a small quantity of holy water from the lake temple, and carry it back to the shrine in the royal temple of Taman Ayun. There, over a period of 3 days 20 Mengwi subaks bring their offerings. In return they receive a few drops of holy water from the lake temple, augmented by the blessings of the dynasty’s deified ancestors.

However, the subaks belonging to the Mambal irrigation system obtain their irrigation water from the Ayung river, which is thought to originate from Lake Batur rather than Lake Beratan. Consequently, the Mambal subaks cannot avail themselves of the holy water from the royal temple of Mengwi. Instead, delegations of Mambal subak leaders make an annual pilgrimage to the temple of the Goddess at Lake Batur, and return with holy water to be shared with all subak members at their Ulun Swi and Ulun Carik temples (Fig. 2 and Table 1). The rites performed in these water temples- offerings to the gods and distribution of holy water from the lake- are identical to those that take place at the royal temple of Taman Ayun, except that they lack ritual blessings from the ancestors of the Mengwi dynasty.

Thus because of their water source, the Mambal subaks have no direct involvement in the subak rituals that take place in the royal temple of the Mengwi dynasty. However, they share responsibility with many other subaks for the support of the temple of the Goddess at Lake Batur. This temple is generally regarded as Bali’s supreme water temple. It is located atop Mount Batur, near the juncture of four nineteenth century Balinese kingdoms: Mengwi, Buleleng, Bangli, and Tabanan. Another of our critics, B. Hauser-Schäublin, has argued that “the temple at Batur was unquestionably a royal temple” (2003:168).Footnote 11 And indeed it is easy to imagine that control of the supreme “irrigation temple” could have been a key political goal for Balinese rajahs, like the Mengwi dynasty. The reality, however, was more interesting (Lansing 2005). To our knowledge, the earliest description of this temple by a European was written by Heinrich Zollinger, after spending several weeks in the vicinity in September 1857 (Zollinger 1866:532):

On the north side of the village there is a temple, one of the biggest and most spacious that I have seen on Bali and also one of the most famous, because delegates come from far and wide to pray here; several kings (vorsten) even have their own houses at Batoer since every year they send delegates there.

Evidently no single king claimed supremacy at the great water temple (or perhaps they took turns?).Footnote 12 Instead many princes and rajahs sent offerings to the temple and sometimes made donations of land, which are recorded in traditional palm-leaf lontar manuscripts kept in the temple. In 1919, after a volcanic eruption damaged the temple, a Dutch official requested that the Governor General provide funds for repairs because “Batoer is of importance to the population of the whole of Bali, and from almost all parts of the island smaller or larger shrines have been built there, or the people have paid a share in their construction” (Moojen 1930:21). Again, there is no mention of control by a single dynasty or king. The importance of the lake temples is stressed in many colonial-era documents, for example in the final report of the senior Dutch official in Bali in April 1937, which describes the annual pilgrimages of the subaks to the lake temples to receive the blessings of the Goddess of the Lake as “keeper of irrigation water” (de Haze Winkelman 1937:44).

While the lake temples and some regional Ulun Swi and Ulun Carik temples are focal points for the distribution of holy water, other temples are associated with the control of agricultural pests (nangluk merana). For this purpose the spatial location of the temples is also determinative: each group of subaks has its own specific temples for pest control. For all the subaks of the Regency (former kingdom) of Badung, the traditional venue for rituals having to do with rats and other pests is the Ulun Swi temple of Cemagi. For the 20 subaks of the old kingdom of Mengwi, the venue is the royal temple of Taman Ayun. At Taman Ayun, one shrine is dedicated to the sea god of the temple of Peti Tenget, who is associated with seasonal pestilence, and another welcomes the deity of the Ulun Swi (head of the ricefields) temple at Jembaran. These shrines are known as penyawangan or “way-stations,” where a distant deity may receive offerings from a local congregation. They provide a way for the farmers to make offerings to these gods without making annual long-distance pilgrimages to their home temples.

With this background, the functional role of the temples with regard to irrigation can be clarified. The heads of the Mambal subaks are responsible for coordinating the ritual cycles held in their water temples, focused on the collection and distribution of offerings and holy water. They also meet to fine-tune their irrigation and cropping schedules, claiming to be guided by the principle that “come better or worse we’re together.”Footnote 13 As conditions require they may also negotiate with their peers who represent upstream subaks along the Ayung river, to arrange the “turn-taking” in irrigation schedules between dams described by van Naerssen. Holy water from the lake temples is collected by subak leaders and distributed at the larger Ulun Swi and Ulun Carik temples, from which it is taken to the Bedugul temples possessed by every subak, and from thence to each farmer’s fields. The sequence of rituals performed in the temples and the fields synchronizes the planting and irrigation schedules at appropriate scales. Along with the hierarchy of temples associated with each river, regional multi-subak temples such as Ulun Swi Jembaran and Ulun Swi Cemagi provide a venue for rituals intended to manage agricultural pests. As conditions warrant, the scale of activities such as rat killings or synchronized fallows can be expanded to include more subaks. The farmers refer to this scheduling as kerta masa (which can be roughly translated as “good time”), in contrast to continuous or opportunistic planting (tulak sumur).Footnote 14

Turning now to the role of the Mengwi dynasty in irrigation, we can clarify our points of disagreement with S.N. Like Vayda, he does not consider the functional importance of the interdependency of the subaks and irrigation systems. With regard to Mambal system, he notes that “the entire system was spread over the domains of several lords.” More generally, he states that “Satellites of the dynasty and other lords were in control of their own irrigation systems” (2011:22). Our question is, what kind of control would be possible at this scale?

Once again we point to the hydrological and ecological submodels in our simulations and to the experience of the Green Revolution, both of which show that irrigation control at the scale of single dams is inadequate. During the Green Revolution, the Mambal subaks obeyed the instruction to abandon coordinated irrigation schedules and plant the new fast-growing dwarf rice as often as possible. The older farmers we interviewed reported that losses to pests soon approached 100%. Harvests began to recover only after the subaks returned to synchronized irrigation schedules.

With regard to the role of sedahans in the management of irrigation, some insights can be acquired from examining the present management of the Mambal system. Table 1 was provided to us by the staff of the Sedahan Agung of Badung, who asked us to note that their jurisdiction is limited to the subaks located in Badung, as shown in the table. Today the Mambal system stretches across the territories of three administrative units: the Regencies of Badung, Tabanan and the administrative district of Denpasar. Each has its own sedahan, consequently three sedahans are involved in the management of the Mambal system. Their respective staff collect taxes from the subaks in each jurisdiction, some of which are earmarked as contributions to water temple rituals. They also help to facilitate the pilgrimages of the subak leaders to the lake temples, and other major ceremonies. But the three Sedahans do not attempt to manage or control irrigation at Mambal.Footnote 15 Were they to do so, they would be faced with the problem that the seasonal water flow at Mambal depends on the release flow from upstream dams on the Ayung river, which lie beyond their jurisdiction.Footnote 16 The same problem would have confronted the sedahans of the precolonial era, with the added difficulty that these kingdoms were often at war.

More broadly, the activities of the Mengwi dynasty and the temple of Taman Ayun fit into the overall pattern of irrigation management by subaks and water temples. For the reasons explained above, the Mambal subaks do not participate in the distribution of holy water or other rites at the Taman Ayun temple, but other subaks do. From a comparative perspective, the involvement of the Taman Ayun in the distribution of holy water to the subaks appears to be unique; elsewhere its irrigation-related functions are carried out by Ulun Swi, Masceti or Ulun Carik water temples. The inclusion of subak rituals within the dynastic temple of Taman Ayun helps to clarify precisely how the lake temples differ from a royal or dynastic temple that has taken on a role in the system of rituals and beliefs pertaining to irrigation, fertility and the management of pests. We note that Taman Ayun does not claim supreme authority over any of these functions; instead it provides a venue for the distribution of holy water from Lake Beratan, and for mobilizing the supernatural powers of the ruler over pests.

The Antiquity of the Subaks

The third part of S.N.’s critique concerns the antiquity of the subak system. Briefly summarized, he argues that “F.A. Liefrinck’s description of the northern Balinese irrigation association (subak) was taken as the normative standard for the reorganization of irrigation management in south Bali.” Hence “it was this colonial intervention which was instrumental in achieving the colonial image of the ‘original’ Balinese subak as a strictly local and ‘autonomous’ institution.”

Our evidence for the antiquity of the subak system comes from three sources: the dated inscriptions issued by Balinese kings from the ninth to the fourteenth centuries (overviews in Lansing 2006, 2007); an archaeological investigations of water temples and irrigation in the Sebatu area in south Bali (Scarborough et al. 1999, 2000) and of early rice (Lansing et al. 2004); and studies of the historical demography of 20 subaks using neutral genetic markers (Lansing et al. 2008; Karafet et al. 2005; Lansing et al. 2011). Because none of our critics refer to these studies, we briefly summarize them here. The role of subaks in water management is attested in royal inscriptions beginning in the eleventh century (Lansing et al. 2008:376–7)Footnote 17:

Because of Bali’s steep volcanic topography, “the spatial distribution of Balinese irrigation canals, which by their nature cross community boundaries, made it impossible for irrigation to be handled at a purely community level” (Christie 1992, 2007). Thus even the earliest Balinese irrigation systems required a system of management that could extend beyond the boundaries of villages. The problem was solved by a new institution called subak, which begins to appear in 11th-century royal inscriptions…Typical ancient irrigation systems included multiple subaks, strung like melons on a vine to take advantage of water flowing through irrigation tunnels, aqueducts, and canals. For example, an inscription dated AD1072 refers to a single subak comprising fields located in 27 named hamlets (Atmodjo 1986). Lansing et al. 2008:376–7.

Archaeologists Vernon Scarborough and John Schoenfelder surveyed the temple of Gunung Kawi Sebatu and the irrigation system which originates in a spring enclosed by the temple, and analyzed the composition of 18 soil cores. They concluded that “over a period of centuries, small teams of farmers were continuously engaged in skilled micro-engineering to maintain control of the flows of water and sediment. There was no evidence for the mobilization of large teams of laborers, or the use of more sophisticated technology than that available to the villagers” (Scarborough et al. 1999, 2000).

Further insights into the historical spread of irrigation in Bali were obtained from studies of neutral genetic markers from 507 farmers in 20 subaks (Lansing et al. 2009:120):

According to Schulte Nordholt, the mobilization of labor to expand irrigation into previously forested regions of western Bali was accomplished by the lords of these ‘micro kingdoms’. This scenario contrasts with the budding model we have proposed for the Sebatu region, in which irrigation works were created by farmers, with new settlements budding off downstream as a result of demographic pressure. The budding process makes specific predictions about the population genetic structure of the villages, and can therefore be tested through molecular genetic analysis.

Analysis of the demographic composition of the subaks indicated that “Although village formation can form in ways other than population budding, there is little evidence that the formation of new subaks is anything but a simple downstream budding process” (Lansing et al. 2009:124; Lansing and Fox 2011).

Conclusion

“The amazingly high standard of rice cultivation in Bali” is an enduring theme in the annals of the Dutch empire in the Indies. The critiques we have considered here help to clarify the functional role of Bali’s water temple networks in this achievement.Footnote 18 As George Box famously observed, all models are wrong, but some are useful, particularly those that are cogent enough to be easily falsifiable (Box and Draper 1987). Vayda’s questions about the relationship between synchronized fallowing and pest control provided an alternative hypothesis to test against both historical evidence and our analytical model of water temple networks. But on re-examination it is clear that pest control by synchronized fallowing antedates the Green Revolution. Indeed, the Green Revolution provided a real-world test of our model, with remarkably clear results.

Schulte Nordholt proposed a plausible counter-argument to our account of the respective roles of rulers and subaks in the construction and management of large-scale irrigation systems. Our response to his critique is twofold: with regard to the historical evidence, we observe that both colonial-era publications and the Kidung Nderet give credit to the subaks, not the kings. We suggest that this fact lends additional support to our analytical model of the ecological role of temple networks, in particular the significance of the interdependency of irrigation systems at scales that range from fields to subaks, dams, contiguous subaks and entire watersheds.Footnote 19 The question of whether the kings played a role in coordinating this system from the top has been investigated by Hauser-Schäublin in her restudy of the lake temple at Mount Batur. But so far she has published no evidence that any dynasty ever claimed supremacy at this temple.

What kind of evidence would count against our model? Hypothetically, if evidence turned up showing that a particular dam or two was built at the instigation of a prince, this would have little significance unless it were part of a larger pattern in which dynasties took over control of either the construction or management of irrigation. Since Balinese irrigation is embedded in a religious context, any assertion of dynastic control over irrigation would likely involve temples and shrines. To our knowledge, the only dynastic temple in which irrigation plays a role is the temple of Taman Ayun in the former kingdom of Mengwi. But the rituals held in Taman Ayun do not assert the control of even local irrigation by the king. Instead, they are one cog in a much larger regional system of water temple rites. If Balinese kings fought for control over the right to distribute holy water from the lake temples, or mobilized large-scale irrigation projects like the Khmer kings; if either pests or water could be effectively managed at the scale of single dams; if water temple rites were unconnected to irrigation schedules and to the imperatives of cooperation; or if, as Vayda proposes, downstream subaks routinely reacted to water shortages with threats, the assumptions of our model would need to be reassessed.