Cultivating Diversity: Field Scattering as Agricultural Risk Management in Cuyo Cuyo, Department of Puno, Peru

By Carol Goland, 1993.

Chapter 8 - Agricultural Yield and Variability

(page 291)

potatoes were growing, they were growing well, and then all of a sudden the disease appeared. Two, three, days BLACKENED!

Although evidence is scanty, I suspect that one effect of the growing seriousness of the late blight has been to alter planting dates. Currently, planting begins in August and is completed by the Festival of Rosario, usually the first week of October. Although I discussed former planting schedules with older Cuyo Cuyeños, I was unable to discern a consensus opinion. While many claimed that planting dates have remained unchanged throughout their lifetimes, the following from a 65 year old man is suggestive:

Cuando estoy joven, chiquillo y sembrabamos el mes de Octubre. Octubre todavía! . . Cuando no habia nada, sanito las plantas. Y terminabamos sembrar para Todos Santos . . . [Empezabamos a sembrar] después del día Rosario. El día Rosario, sembrabamos, diciendo "Parawan. Ya vamos a sembrar" decimos. Pero, daba, daba pues mejores, eso - el sembrio de ese día.

When I was young, little, we planted in the month of October. October! . . When there was nothing [i.e., no diseases], the plants were healthy. And we finished planting for Todos Santos [November 1]. We began the planting after Rosario. The day of Rosario we planted, saying "It's raining. Now we're going to plant," we said. But, it yielded, it yielded the best - the planting from that day.

In other words, there is some indication that perhaps 40 or 50 years ago, planting may have occurred as much as two months later in the year.

Cuyo Cuyo farmers are aware that the greatest impact of rancha occurs when it befalls the crop prior to flowering. Because rancha attacks are exacerbated by wet conditions, the potato crop is more at risk once the rainy season begins in earnest. If planting is moved earlier (relative to the rainy season), then there is a longer period available for tuber bulking before rancha defoliates the crop, reducing photosynthesis and, ultimately, yield. Of course, in doing so the crop may also be displaced from its optimum growing season in terms of moisture, temperature, and radiation regimes.⁷

While rancha seems to be the greatest threat to potato production, it is not the only one. Unfortunately, none of the Quechua names provided to me can be unequivocally related to their scientific identification. Golden cyst nematodes (Globodera pallida and G. rostochiensis) occur throughout the Andes. The most effective control for them is through long crop rotations: seven or eight years are needed between potato cultivation. This has been posited as a major benefit of sectoral fallowing in the Andes (Orlove and Godoy 1986). I have no clear evidence that golden nematodes are among the pests in Cuyo Cuyo, but given their ubiquitous distribution in the Andes I presume they are present. The fact that Cuyo Cuyo farmers did not identify them possibly is related to their subtle symptomology; the effects on the above ground portion of the plant are generalized (reduced foliage, yellowing, etc.). The only diagnostic indication are the appearance of tiny (0.5 - 1.0 mm) cysts which sometimes occur on the roots and tubers (International Potato Center 1982).

Cuyo Cuyeños recognize a variety of other pests which reduce production. These include both worms (kuru) and caterpillars (silwi). Most cause their damage by consuming the

(page 292)

aerial foliage; several burrow into the tubers themselves. In the Awi Awi valley, the worm laqhakuru itself poses no threat to the crop, but skunks forage for them, digging into the root zone and dislodging plants.

It often is difficult to disentangle the reports of farmers, to determine whether a term refers to an actual disease, or whether it describes the condition of the tuber, regardless origin. While part of this confusion may be a product of my limited Quechua, another part of it undoubtedly reflects the lack of consensus among Cuyo Cuyo farmers. For example, several families reported papa ismu in some of their fields. While some see ismu as a disease (pathogen) which causes rot, others use ismu simply to denote the rotted condition of tubers, blaming it (sometimes) on humid soil conditions. It is clear, however, that excessive humidity is more often cited as negatively affecting yields than is insufficient moisture. But it is difficult to use this observation (covering only two years' time) to generalize about the relative importance of these factors over the long run.

Animal intrusions (both wild and domestic) cause frequent damage to the crop. Mice, foxes, and unattended cows, llamas, horses, burros, or sheep may enter fields and feed on the foliage, devastating production. Seven of nine families in Ura Ayllu, and eight of nine in Puna Ayllu cited crop damage by animals in at least one of their fields. Farmers like to have fields located in the center of a manda sector because fields near the edge are more at risk of damage from domestic animals on adjacent fallow open for grazing. Fields located near paths are also vulnerable. Sometimes humans and animals combine to create damage. One family ruefully described how during the festival of Carnival enthusiastic celebrants danced off the path into their field during a procession, trampling the habas stalks. Because of this, mice invaded the chakra to feast on the easy pickings. Theft is also widespread. The majority of families in each community claimed to have had thefts in at least one field.

When trying to explain reduced yields, some families cited their own faults, having done a certain task at the wrong time (barbecho or planting too late, or planting or aporque on a bad day). Given the heavy demands of labor in Cuyo Cuyo, it seems evident that families are sometimes forced to compromise work schedules relative to what they believe to be optimal.

Cuyo Cuyeños are consistent in declaring q'echa hallp'a the finest soil. Frequently this soil is described as yana hallp'a (black soil) and urina hallp'a (soil which produces). Q'echa hallp'a is distinguished as "pure soil" (pura tierra), emphasizing a deep topsoil layer and the absence of stone.⁸ These two characteristics stand out in descriptions of other soils: q'ara hallp'a (impoverished or eroded soils) is distinguished by its high content of pebbles and stones and relative shallowness, as is saya hallp'a. The latter term appears to be reserved for soils on steep slopes, qhata. Cuyo Cuyeños express their concern over the moisture retention capabilities of the soils of different fields. One man described his terrace as langra, ch'ata, q'ara. Langra describes the loosely textured, somewhat rocky, light colored soil. The quality of being ch'ata is related to this texture: water runs directly over the field without percolating;

(page 293)

it retains moisture poorly. The last quality, q'ara, is a consequence of both of the preceding factors. It describes eroded soil. As the water runs over the field, it takes the top soil with it.

Another factor affecting yields which families cite is the origin of the seed itself. In the second year of the study many families received a loan of a small amount of habas seed from PISA. This seed performed poorly, and many families believed that the reason for this was that they were habas de afuera (habas from outside). This reasoning is not unique to habas. Reflecting on a field where some potato seed from PISA had been planted, a woman said "perhaps this potato isn't accustomed to this soil" ("Quizás esta papa no ve acostumbrado a la tierra").

Cuyo Cuyeños readily make the distinction between those potatoes which are nativas de la zona (native to the zone), and those which are de afuera (from the outside). They recognize the mutual compatibility between native potatoes and soils of the valley, expressed as an "affection" which binds potato to land. When potatoes are removed from the right habitat they cease to thrive. This is more than a statement about adaptation of potato varieties to microhabitats. It is underlain by belief in the interlinked sentience of the potatoes and the earth (pachamama). A potato that is removed from its proper habitat may become resentful. Relocated, it will come to miss its home. The pachamama will be angered at the removal of her children. People are careful about those to whom they give potato seed. Although some varieties can be shared with anyone, special classes of potato can only be obtained from, or should only be given to, trustworthy friends and relatives. Unfamiliar people may not care for the potato properly, with the consequence that it and its land will grow resentful, and fail to produce.⁹ These beliefs are not absent with respect to other crops, but they find their fullest and most frequent expression in association with potatoes.

The crops of Cuyo Cuyo, especially potatoes, are conceptualized very much like children. The earth is their mother, pachamama. Labor inputs are higher for potato than other crops. The potato crop is lavished with care and affection; it matures to reward its growers. The pachamama is a mother who gives birth to the crops of Cuyo Cuyo. The image of foods issuing forth from the womb of the pachamama seems especially apt in this region where the majority of agricultural products are tubers, and their harvest, literally, involves opening up the earth and removing produce from it. The meaning of pachamama, and the beliefs and behaviors of agricultural practice, unite in this interpretational context.

In the month of August at the time that the earth is said to be open, Cuyo Cuyeños make payment to the earth (santatierraman pagasum, or pago a la tierra). The payment is undertaken by individual families with the help of one of the local curanderos. The payment consists of a store-bought offering (recado), a package containing crackers, candies, noodles, rice, beans, flour, sugar, and ornaments. To this are added llama fat, flowers, coca leaves, herbs, and alcohol. Gathered together in a swirl of coca, cigarettes, and alcohol, the participants assist in the construction of the offering (ofrenda) under the direction of the curandero. At midnight, after ascending near to the mountain peaks, the offering is burned to the pachamama and the powerful apus (god-spirits resident in the mountain peaks). If the offering is accepted, it will be fully carbonized by the fire; anything less suggests displeasure, and possible menace from the spirits. The participants remain together until daybreak, in rikch'ay (to stay awake until the dawn).

(page 294)

malevolent powers of the apus and pachamama: Apu Chuchu, Alqamarini, with all our hearts we offer you this. Receive us with good will. Make this year a good year, that there is not too much rain, that the sun should be moderated, the rain that which is necessary, and with your power blow the diseases along with the wind into other valleys.

And to the pachamama in particular:

You have given us good years, these past years, that's the reason for these foods [i.e., the offering], so that you will give us good years. Look also little mother, Santa Tierra, Maria Ankari we offer you so that you will lavishly give us to eat from your womb, and give us always our food, so that there is an abundance of food this year. And don't anger and don't send your anger upon us. The rain should rain controlled, the wind blow as it should blow, with calm, no more than necessary, the sun, shine as it should shine, no more than this.¹⁰

As noted by Urton (1981) there is a strong--and unsurprising--relation between the moon and female biological cycles in the Andes. In the Cuzco town of Misminay where Urton worked, the association between potato planting and moon phases was explicit: potatoes should be planted during the waning moon. Urton suggests that this can be extended to all crops which produce underground. The planting of crops that produce above ground, in contrast, is associated with the waxing moon. Buechler and Buechler (1971) also note an association between femaleness and certain crops (potatoes and oca), and maleness and other crops (habas, wheat, and barley).

In Cuyo Cuyo there are strong prohibitions concerning not only planting, but also weeding (aporque), in relation to the moon phases. In Spanish, the phrase most often used in discussing this is "no tocar la tierra" (to not touch the earth). This sense of fragility or vulnerability of the earth during certain phases of the moon is most certainly related to biological cycles. This is underscored by the constant equation of female and moon in Andean ritual and symbolism (Arnold n.d.; Harrison 1989; Isbell 1978).

One man described to me the alarm of his neighbors when he had fallen behind in his weeding and found himself in his maizal on Christmas Day:

Entonces, yo estoy cultivando pues, ese día, fiesta, como dicen mamayawaniqyoq (la sangre de la mama, por haber dado luz) antes decíamos esto, Navidad, entonces "Cómo vas a cultivar hoy día? Nuestra madre está con sangre. Vas a malograr a su cuerpo." . . . me han dicho "que se malogra su maíz, va a malograrse todo, no va a dar, no va a producir y vas a perder."

Well, I was weeding, that day, fiesta, as they say mamayawaniqyoq (the blood of the mother, for having given birth), before that's how we said "Christmas." Well "How are you going to weed today? Our mother is bloody. You're going to ruin her body." . . . they said to me "your corn will be ruined, all of it's going to be ruined, it's not going to yield, it's not going to produce and you're going lose [it]."

(page 295)

Though powerful, the pachamama is also vulnerable. Along with routine care and respect for the pachamama, Cuyo Cuyeños must also be sensitive to cyclical or irregular changes in her condition. For many Cuyo Cuyo farmers this threat to agricultural production, rooted in Andean cosmology, is equal to the risks created by the ecological and climatological factors discussed below.

The pachamama is critical to the fertility of the valley. A key element of the pago a la tierra was a llama fetus, considered the mark of a high quality offering. Like so many other relationships in the Andes, this one is built on reciprocity; here the fetus offers a life in exchange for a return of produce harvested from the womb, pachamama. The term for a year of poor agricultural production, sullunwata, comes directly from the image of harvest as the fruit of pachamama's womb. Wata means year, while sullun comes from the verb sulluy, "to abort." Agricultural failure is starkly equated with aborted pregnancy.

STATISTICAL ANALYSES OF VARIANCE IN CROP YIELD

Crop yields are not systematically different when the two communities, the two years of the study, or the different croppings (monocrops and polycrops) are compared. But this does not mean that yields are uniform from field to field. On the contrary, field-to-field variance in crop yield is large. For all of the crops the standard deviations are nearly as large as the means. Part of the explanation lies in field location: fields of the estancia on average produce more. But this observation in itself reveals little about the causes of variance in crop yields.

Crop production may be reduced or enhanced by a series of environmental and agronomic factors. These include high- and low-temperature stresses, insufficient or excessive moisture regimes, hail, wind, insufficient quantity of solar radiation, poor seed quality, inappropriate plant population density, deficiencies in soil texture and nutrients, pests and diseases, weed competition, and inappropriate planting date. Many of these factors are directly or indirectly related to the ability of the crop to maximize its rate of photosynthesis, the fundamental factor of yield (Tivy 1990). The rate of photosynthesis is determined in large part by receipt of incident radiation, which in turn is a product of the structure of the crop canopy. Any factor which reduces the Leaf Area Index (area of leaf surface per unit of ground surface) to sub-optimal levels will also reduce yield. If temperature and water stresses, or pests, defoliate the plants (especially in the earlier stages of their growth), yields will be reduced. If seeding occurs at such low density that bare ground exists between individual plants in the stand, yield (per area) will be below optimum. Conversely, if seeding density is so high that there is competition between plant neighbors, yield (per individual plant) will be reduced. Planting date must be calibrated to optimize the timing and duration of incident radiation (as well as thermal and moisture conditions).

In the heterogeneous agricultural landscape of the Andes, these factors may differ areally on micro-geographic spatial scales. The temperature which a given field experiences is affected by altitude and micro-climatic setting; moisture regime will be influenced by slope and soil texture; incident radiation will differ according to slope and aspect (and further affected by planting date, in turn mediated by altitude), and will vary depending on cloud cover and shadow; pest and disease factors are patchy; plant population densities vary and may be determined by availability and condition of seed (in turn, affected by the quality of the previous harvest and storage conditions). Given all of this, the extreme field-to-field variance in crop yields in Cuyo Cuyo is not surprising.

(page 296)

In order to understand the underlying causes of yield variance, I constructed multiple regression models for yields of three crops: potato, oca, and habas. A series of factors which could potentially induce systematic change in crop yields were examined, alone and in various combinations. Among them are: community, calendar year, rotation year, cropping (for potatoes, i.e., milli, panq'o, or maincrop), field location (estancia or distant), altitude, fertilizer input, labor inputs, planting date, field area, and seed density. The first five variables were included in the models as control ("dummy") variables, while the last six are continuous variables. I also included more precise community/location control variables (i.e., Puna Ayllu-Far, Puna Ayllu-estancia, Ura Ayllu-Far, Ura Ayllu-estancia), and several interaction terms composed of combinations of control and continuos variables (i.e., Community x Fertilizer, or Near x Labor).

Many of the continuous variables were significantly correlated with yields when examined as a univariate linear model (altitude was the notable exception - see below). Although significant, the univariate models accounted for only small proportions of the variance in yield. This fact suggested the need to analyze the variance with multiple regression models.

Analysis of Variance in Potato Yields

Table 8.12 presents descriptive statistics for the variables used in this analysis. Table 8.13 presents the correlation coefficients of each of the independent variables with potato yields, for both years of the sample combined as well as each year alone. Seed density and distance (i.e., in the estancia or outside of it) are significantly correlated with potato yields in each individual year and both combined. Fertilizer input (NPK) is significantly correlated only in the composite sample and 1986-87. Labor is significantly correlated with yield in 1986-87, the only year for which information is available. As I will describe below, labor, fertilizer, and distance are the variables which together explain the highest proportion of the variance in potato yields.

Table 8.12 also presents information on control variables: distance (Near), use of fumigation (Fumig), cropping (Non-panq'o), and year of study (Calyear). Neither fumigation or calendar year alone accounts for a significant proportion of the variance in yields. The non-significance of calendar year is particularly important for the following reason: as discussed in Chapter 6, only labor data for the 1986-87 agricultural cycle were analyzed. Since labor appears to be an important independent variable in accounting for yield, any model which includes labor will by default exclude the preceding year. Exclusion of one year is not as serious a problem as it may first appear. Common sense would dictate that the two years of crop production should be analyzed separately, since even if yields did not significantly differ, the factors affecting those yields might.

The control variable which accounts for distance ("near") is significantly correlated with yield, though this was something already known from data presented earlier in this chapter. Again, the near fields (i.e., estancia fields) show significantly greater yields, while the distant fields in both communities show significantly lower yields, contrasted to the non-excluded categories.

(page 299)

Several of the independent variables are correlated among themselves. The problems posed by such collinearity in regression models can be serious. Potentially, collinearity can create large estimation errors and highly unstable coefficients (Rawlings 1988). However, among the independent variables of primary interest (seed/ha, NPK, labor, and distance), the highest correlation (between NPK and seed/ha) was 0.4473 (for both years of data combined). All other correlations are below 0.40, so the models will not be affected by the problem of multicollinearity.

Notwithstanding the potential problems created for the regression analysis, the fact that seed, labor, fertilizer inputs, and distance are highly correlated is itself of interest. Table 8.14 displays the correlation coefficients among these four variables. It is not surprising that they should be correlated: a higher density crop will remove greater quantities of nutrients from the soil and hence require greater fertilizer inputs. The labor inputs are those for field preparation and weeding. (Planting was excluded since it would be expected to highly correlate with seed density; harvest time was excluded since it should properly be viewed as the product rather than cause of yield variance.) While preparation time should vary independently from seed density, weeding time is clearly a function of density. Since plants receive individual attention in the weeding and hilling process, an increase in seed density may increase weeding labor (at least up to the point where the crop becomes so thick that it shades out weeds). The correlation of distance (here scored for "near") with inputs echoes material presented in previous chapters on labor and fertilizer inputs: fields in the estancia are apparently tended more carefully.

There are several other interesting correlations among independent variables. Date of planting, altitude, and NPK are correlated among themselves. A correlation of 0.5806 between altitude and planting date substantiates material already discussed: lower elevation fields are used for planting the early potato crop, papa milli, while maincrop fields at higher elevation are planted later in the season, presumably due to fear of frost. The correlation of altitude and NPK (0.3848) underscores the heavier applications of fertilizers in the Awi Awi potato fields. Thus the correlation between fertilizer and date of planting (0.2307) is presumably a product of the influence of altitude on both.

Of the five continuous variables, three (seed/ha, prep/weed labor, and NPK) demonstrate significant correlations with yield (Table 8.13). When each is regressed on yield, the independent variable which alone explains the greatest amount of variance in potato yields is preparation and weeding labor (measurement units used are days). Seed/ha and NPK, although significant, alone can account for little of the variance in yield. The correlation of altitude and yield is not significant. This is not to say that altitude is unimportant. There are two possible explanations for this. First, it may be that alone, altitude is unable to account for variance in the absence of information about other important variables (such as labor, fertilizer inputs, and seed density). Second, and equally if not more likely, altitude and yield may not have a linear relationship. If there is some optimum elevation for potato production in Cuyo Cuyo, then dispersion of fields both above and below this optimum could produce a curvilinear relationship between elevation and yield. But if specific potato varieties are planted in elevational ranges where they produce best, or if other production decisions can be geared to altitude, then this source of variance may be circumvented. When the relationship between yield and altitude was examined over a narrow segment of the gradient (the estancia only), no linear relationship was found. The correlation of yield and planting date is not significant.

As noted above, there is a significant linear relation between seed density and potato yields. However, the relationship becomes even stronger when the log transform of seed density is used. This suggests that the relationship between seed density and yield is curvilinear: yield rises with increasing seed density, eventually reaching an asymptote at which point the increase in yield tapers off or even begins to decrease. This may be due to increased

(page 301)

competition between plants, especially with regard to intercepting solar radiation for photosynthesis.

The curvilinear relation between seed density and yield prompts several additional observations. The average seed density in Cuyo Cuyo potato fields is somewhat higher than the recommended values. In commercial European and American potato production, seed densities range from 1500-2500 kg/ha for a normal crop (ware potatoes), higher if a seed crop is planted (3000-5000 kg/ha). But this refers to high yield commercial production, which vastly exceeds the yield of Cuyo Cuyo fields. For lower yielding areas, lower seed densities are recommended (Beukema and van der Zaag 1979). The average seed input per hectare in Cuyo Cuyo is 2315 kg/ha (Table 8.12), high for the conditions of production prevailing there. Some fields are seeded at much higher densities (the maximum in the sample is 5200 kg/ha). The decreased yield of fields planted at higher densities suggests that these in fact have exceeded the optimum.

I tentatively suggest that this apparent over-seeding in Cuyo Cuyo may be interpreted as a risk minimization strategy. It is important to recall that the years represented by this sample were good years for potato production. Under more adverse conditions, seed germination rates may be reduced. If a minimum (worst-case) goal of farmers is to recuperate their seed investment in a field, then high seeding rates might be seen as an attempt to at least maintain seed stock in the face of nearly catastrophic yield. If this were the case, then in years of extremely good production seed density may actually exceed the optimum. Yields somewhat reduced from the theoretical optimum during especially good years seem to be a reasonable compromise when the alternative worst-case scenario is one of failing to recuperate seed in years when germination rate is low and growing conditions are adverse. In the absence of comparable data from Cuyo Cuyo for years with poor potato production, this argument cannot be tested. Analyses of oca yields however, provide some additional support (see below).

It is not entirely clear, though, at what point increases in seed density will cause yield to diminish (i.e., the exact curvilinear form of the relationship). While yield is fairly stable over a wide range of planting densities, it is certain that tuber size is considerably more plastic (Beukema and van der Zaag 1979; Hay and Walker 1989). With increasing plant population, yield increases (at least to the point of some plateau), but the size of individual tubers decreases. Commercial seed potato growers plant at high population densities in order to obtain tubers of the appropriate size (35-80 g). I suggest that, even though seed potato production is not the primary concern of Cuyo Cuyo farmers, harvest of relatively smaller tubers is also to their benefit, for several reasons. First, a portion of the crop is processed into ch'uño. Because of loss of fresh tubers in storage, it is prudent to preserve a portion of the crop into the more durable form of ch'uño. Since the freeze-dry method works best with diminutive potatoes (frost penetration is more complete), planting densities may also be geared towards production of tubers which in size are appropriate for processing as well as for seed. An equally compelling case for the production of small potatoes has to do with cooking time. Fuels are scarce: kerosene is costly, and wood currently must be hauled from long distances. Altitude exacerbates the problem, as foods take longer to cook due to the lower boiling temperature of water. In this setting, small potatoes are advantageous, since a staple dish, wayk'u (boiled tubers), utilizes whole tubers. It makes sense to cook many small potatoes, thereby conserving fuel. There are thus several advantages to the production of small tubers. This may explain in part the high seed densities observed.

MULTIPLE REGRESSIONS OF POTATO YIELDS