Cannabis Grow

While these farmers represent a case study for building a successful, organic farm within one generations, the results of this study beg the question: What advancements in farm management and soil management could be possible with multiple generations of farmer knowledge transfer on the same land? Rather than re-learning the ins and outs of farming every generation or two, as new farmers arrive on new land, farmers could have the opportunity to build on existing knowledge from a direct line of farmers before them, and in this way, potentially contribute to breakthroughs in alternative farming. In this sense, moving forward agriculture in the US has a lot to learn from agroecological farming approaches with a deep multi-generational history . To this end, in most interviews—particularly among older farmers—there was a deep concern over the future of their farm operation beyond their lifetime. Many farmers lamented that no one is slated to take over their farm operation and that all the knowledge they had accumulated would not pass on. There exists a need to fill this gap in knowledge transfer between shifting generations of farmers in order to safeguard farmer knowledge and promote adaptations in alternative agriculture into the future.Most studies often speak to the scalability of approach or generalizability of the information presented.

While aspects of this study are generalizable particularly to similar farming systems in California such as the Central Coast region, cannabis dry rack the farmer knowledge presented in this study is not generalizable and not scalable to other regions in the US. To access farmer knowledge, relationship building with individual farmers leading up to interviews as well as the in-depth interviews themselves require considerable time and energy. While surveys often provide a way to overcome time and budget constraints to learn about farmer knowledge, this study shows that to achieve specificity and depth in analysis of farmer knowledge requires an interactive approach that includes—at a minimum—relationship building, multiple field visits, and in-depth, multi-hour interviews. Accessing farmer knowledge necessitates locally interactive research; this knowledge may not be immediately generalizable or scalable without further locally interactive assessment in other farming regions.Local knowledge among farmers in US alternative agriculture has often been dismissed or overlooked by the scientific community, policymakers, and agricultural industry experts alike; however, this study makes the case for inclusion of farmer knowledge in these arenas. In-depth interviews established that farmers provide an important role in translating theoretical aspects of agricultural knowledge into practice. It is for this reason that farmer knowledge must be understood in the context of working farms and the local landscapes they inhabit.

As one of the first systematic assessments of farmer knowledge of soil management in the US, this research contributes key insights to design future studies on farmer knowledge and farmer knowledge of soil. Specifically, this study suggests that research embedded in local farming communities provides one of the most direct ways to learn about the substance of farmer knowledge; working with the local UCCE advisor in combination with community referrals provided avenues to build rapport and relationships with individual farmers—relationships that were essential to effective research of farmer knowledge. Farmer knowledge of soil management for maintaining healthy soils and productive, resilient agriculture represents an integral knowledge base in need of further scientific research. This study provides a place-based case study as a starting point for documenting this extensive body of knowledge among farmers. It is our hope that this research will inspire future studies on 26 farmer knowledge in other contexts so that research in alternative agriculture can widen its frame to encompass a more complete understanding of farming systems and management motivations—from theory to practice.A fundamental challenge in agriculture is to limit the environmental impacts of nitrogen losses while still supplying adequate nitrogen to crops and achieving a farm’s expected yields . To balance among such environmental, ecological, and agronomic demands, it is essential to establish actual availability of nitrogen to crops . A holistic, functional understanding of plant N availability is particularly imperative in organic agriculture, as in this farming context, synthetic fertilizers are not applied and instead, production of inorganic N—the dominant form of N available to crops—depends on internal soil processes . In organic agricultural systems, farmers may seasonally apply cover crops or integrate livestock as alternative sources of nitrogen to crops—in addition to or in place of using organic fertilizers. In applying these alternative sources of nitrogen to soil, organic farmers rely on the activity of soil microbes to transform organic N into inorganic forms of N that are more readily available for crop uptake .

Currently, the predominant way crop available N is measured in organic agricultural systems tends to examine pools of inorganic N in soil . Inorganic N, or more specifically ammonium and nitrate , represents the predominant forms of N taken up by crop species in ecosystems where N is relatively available, such as in non-organic agricultural systems that apply inorganic fertilizers . However, in organic systems, crop available N is largely controlled by complex soil processes not adequately captured by simply measuring pools of ammonium and nitrate. First, because nitrogen made available to crops is controlled by soil microbes—wherein crops only have access to inorganic forms of N after microbial N transformations occur to first meet microbial N demand—pinpointing the flow of N moving through inorganic N pools as a result of these microbial N transformations is necessary to accurately measure actual N availability to crops . Second, extensive recycling of N among components of the plant-soil-microbe system complicates relying solely on measurements of inorganic N pools, which do not reflect these dynamics . As an example, one previous study in organic vegetable systems showed examples where inorganic N pool sizes in the soil were measured to be low, yet there existed high production and consumption rates of inorganic N . This outcome highlighted that if the turnover of inorganic N is high—for instance, high rates of soil ammonium production exist in the soil with simultaneously high rates of immobilization by soil microbes and high rates of uptake by plants—measured pools of inorganic N may still be low . This study also showed that conversely, there may also exist situations when inorganic N pools are low and rates of ammonium and nitrate production are also low, in which case N availability would limit crop production. In organic systems especially, higher carbon availability as a result of organic management can increase these microbially mediated gross N flows, thereby increasing N cycling and turnover of inorganic N . Thus, trimming tray we hypothesize that measuring total production of ammonium from organic N, or gross N mineralization, and subsequent total production of nitrate from ammonium, or gross N nitrification, may provide a more complete characterization of crop available N in the context of organic systems . Though the application of such diverse management practices on organic farms is known to affect rates of N cycling in soil , measuring N flow rates as a proxy for crop available N is currently uncommon on working organic farms. The current historical emphasis on measuring inorganic pools of N in organic agriculture was originally imported from non-organic farming, wherein the Sprengel-Liebig Law of the Minimum was a widely accepted agronomic principle . In practice, this Law of the Minimum placed particular importance on using artificial fertilizers to overcome so-called “limiting” nutrients—namely, inorganic forms of N. Because inorganic N is relatively straightforward to measure, focus on quantifying pools of inorganic N has since become common practice among agronomists and agricultural researchers . However, the continued acceptance of the Law of the Minimum in organic agriculture underscores the gap in a functional understanding of organic agricultural systems, in particular the role of soil microbes in mediating N cycling. To understand crop available N more holistically, there is a need to measure actual flow rates of soil N—in addition to—static pools of inorganic N . Soil indicators that adequately capture N availability to crops are therefore necessary to move beyond the legacy of the Law of the Minimum in organic agriculture.

Unpacking the soil processes that mediate flows of N may ultimately provide a more accurate characterization of soil N cycling and in turn, N availability to crops. Unfortunately, gross N mineralization and nitrification rates are very difficult to measure in practice, particularly on working organic farms . While net N flows are easier to measure in comparison to gross N flows and can provide a useful measure of N cycling dynamics as a complement to measurements of inorganic N pools, net N flows still pose serious limitations— namely that net rates cannot detect plant-soil-microbe interactions and therefore are not adequate as metrics for determining crop available N . In particular, relying on net N flows as a measure of N availability does not account for the ability of plants to compete for inorganic N, and assumes plants take up inorganic N only after microbial N demands are satisfied . It is also possible that measuring soil organic matter pools could help indicate N availability because SOM supports microbial abundance and activity, and because SOM is also the source of substrates for N mineralization . Several studies have proposed measuring soil organic matter levels to complement measuring inorganic N pools, understand soil N cycling, and infer N availability . Assessing the total quantity of organic carbon and nitrogen within soil organic matter represents one established method for measuring levels of soil organic matter, and is morereadily measurable than gross N rates. Additional indicators for quantifying “labile” pools of organic matter, such as POXC and soil protein, have also become more widely studied in recent years, and applied on organic farms as well . When used in combination with more established soil indicators that measure organic C and N pools , this suite of indicators may potentially provide added insight to understanding crop available N . Importantly, applied together these four indicators for soil organic matter levels may also more readily and accurately serve as a proxy for soil quality—generally defined as a soil’s ability to perform essential ecological functions key to sustaining a farm operation . Despite the availability of these soil indicators, very few studies have systematically examined the way in which SOM levels on working farms compare to N cycling processes, and specifically how SOM levels compare to microbially mediated gross N rates. Further, it is still unclear to what degree the interactions between soil edaphic characteristics and soil management influence N cycling and N availability to crops . For instance, soil texture may play a mediating role in N cycling, where soils high in clay content may limit substrate availability as well as access to oxygen, which in turn, may restrict the efficiency of N cycling . In this sense, it is important to understand the role that soil edaphic characteristics play in order to identify the underlying baseline limits imposed by the soil itself. Compared to controlled experiments, soil management regimes on working farms can be more complex and nonlinear in nature due to multiple interacting practices applied over the span of several years, and even multiple decades. To date, a handful of studies conducted on working farms have examined tradeoffs among different management systems , though few such studies examine the cumulative effects of multiple management practices across a gradient of working organic farms. However, understanding the cumulative effects of management practices is key to link soil management to N cycling on working farms . Likewise, it is important to examine the ways in which local soil edaphic characteristics may limit farmers’ ability to improve soil quality through management practices. Though underutilized in this context, the development of farm typologies presents a useful approach to quantitatively integrate the heterogeneity in management on working organic farms . Broadly, typologies allow for the categorization of different types of organic agriculture and provide a way to synthesize the complexity of agricultural systems . Previous studies that make use of farm typologies found that differences in total soil N across farms are largely defined by levels of soil organic matter.To address these questions, we conducted field research at 27 farm field sites in Yolo County, California, USA, and used four commonly available indicators of soil organic matter to classify farm field sites into farm types via k-means cluster analysis.

The evidence in table 6 that mother’s bargaining power influences the amount of house-work that daughters take over when mothers get employed suggests that mothers have influence over daughters’ time. But should daughters themselves and fathers also be seen as decision makers participating in decisions about daughters’ time use? In table 7, I interact proxies for the mother’s, father’s and the oldest daughter’s preferences – answers to survey questions about each of the three family-members’ attitude towards girls’ schooling – with the treatment. In households in which the mother considers girls’ schooling more important, the negative effect of mother’s employment on daughters’ schooling is significantly smaller. The father’s attitude towards girls schooling appears to have less influence on time use substitution between mothers and daughters, and a daughter’s own preferences have no significant effect on the amount of house-work she is expected to take over when her mother gets employed. It thus appears that daughters in rural Ethiopia have little control over their own time use in times of need.Analysis of how households select into mother’s versus father’s employment is important in its own right but also represents a powerful auxiliary test of the main message of the framework above. If, as this paper has argued, a key determinant of rural Ethiopians’ time use is gender-specific, greenhouse benches intra-household labor substitution, greenhouse benches then household characteristics that influence the impact of mother’s and father’s employment on other family-members – such as the gender composition of a couple’s children – should also influence selection into mother’s versus father’s employment.

The sample analyzed consists of households in which either the mother or the father applied to a flower farm. To pool the two sub-samples and explore selection into the two groups, we must thus assume that, in for example a household in which the mother applied, the father would have applied had the mother not done so. While this assumption is ultimately untestable, it is arguably reasonable. As noted, there were only seven households in which both spouses applied – for most households the relevant choice options appear to have been for one or none of the two spouses to apply. There are few households in the sample in which the spouse of the applicant was already formally employed. In table 8 I investigate the comparability of the two sub-samples. Excluding the right-hand-side variables that the framework predicts should influence selection into the two groups , the only significant difference is that husbands are one year older in households in which the mother applied. I thus control for husband’s age in the analysis below. As we saw in table 6, perhaps the most important variable governing heterogeneity in the impact of mother’s employment on daughters’ time-use is the gender composition of the couple’s children because the presence of more daughters means that house-work can be shared between more hands. As such, we would expect the number of daughters to have an important influence on selection into mother’s versus father’s employment. But testing for a causal relationship is possible only if the number of daughters is exogenous conditional on the total number of children. If parents follow differential stopping rules – that is, if the probability of having another child depends on the gender composition of existing children – then the number of daughters is not exogenous even conditional on family size, as pointed out by Clark and discussed in detail in Washington .

It turns out that parents in the sample do not follow such stopping rules: neither a variable equal to the total number of children, nor dummies for having a given number of children, predict the proportion of daughters, as seen in table 9. The explanation may be that desired family sizes in rural Ethiopia are so large that almost all couples have one or more sons through “natural” fertility behavior. Parents with son preference typically want “at least X number of sons” , where X is a positive but relatively low number.We can thus test if the gender composition of a couple’s children has a causal effect on the probability that a mother seeks employment. I do so in table 10, including interactions with the proxies for mother’s weight on daughters’ well-being and mother’s bargaining power to mirror the heterogeneity regressions in table 6. The selection analysis results are supportive of the idea that female time use substitution is key to household employment and schooling decisions in Ethiopia. For example, one additional daughter increases the probability that the mother applies by 8 percentage points, or 12 percent, controlling for the total number of children, in households with low weight on daughter well-being and low mother’s bargaining power. The results also indicate that the higher the weight on daughters’ well-being, the lower the influence of the number of daughters on the couple’s employment decision. The reason appears to be that highly valued daughters are expected to take over less household work when mothers get employed. Finally, mother’s bargaining power at baseline has a marginally significant positive effect on the influence of the number of daughters on the probability that the mother applies. The presence of daughters has a direct influence on the mother’s well-being under mother’s employment relative to father’s employment because a mother can likely decrease her time spent on house-work when employed more when more daughters are present. A father’s well-being under mother’s versus father’s employment may, in contrast, be less dependant on the gender composition of the couple’s children because “male” house-work is less time consuming.

It appears that greater bargaining power for the mother therefore increases the weight given to the gender composition of the couple’s children when the employment decision is made. The findings in table 10 thus suggest that parents take into account substitutability between a mother’s and daughters’ time use when making adult employment decisions. If daughters taking over house-work duties when mothers get employed is difficult to avoid, it may be that the best way to take daughters’ well-being into account is at the employment decision stage.Domestic violence represents a serious violation of women’s rights and imposes substantial costs on society. In parts of Ethiopia, 71 percent of ever-partnered women have been physically assaulted by a male partner . In the U.S., domestic violence assault is more common than all other forms of violence combined . But despite its prevalence throughout much of the world, the nature of physical abuse of women remains poorly understood. Little is therefore known about how to address the issue. In this paper, growers equipment we analyze the effect of female employment on domestic violence through a field experiment in rural Ethiopia that randomized job offers, the first of its kind. Conventional economic models of domestic violence are “optimistic” in the sense of predicting a decrease in abuse when women get employed; we find the opposite. We then begin to distinguish between “pessimistic” models. We find limited support for models in which violence is used as a tool to gain control over household resources, growers equipment and more support for models that allow men to see violence as a way to restore their dominance in the household. The sample consists of 329 households in which an adult woman applied to a flower farm job and was deemed acceptable for hiring by the farm. The treatment and control groups were re-surveyed 5 – 7 months after employment commenced. Our research design has important advantages. Because we directly vary job offers, we can attribute changes in violence to the causal effect of employment. There is to our knowledge no existing experimental evidence from poor countries on the effects of permanent female employment, by many thought to be the most effective way to reduce physical abuse. Policy and arguments are therefore made on the basis of assumptions on which clear-cut causal evidence is largely missing: the World Health Organization argues, for example, that “women’s access to. . . employment should. . . be strongly supported as part of overall anti-violence efforts” . In the absence of sufficient evidence, there is little consensus on which model of domestic violence best describes reality. In the main result of the paper, we find a 13 percent increase in the probability that a woman is experiencing physical domestic violence, when she gets employed. We also find a 34 percent increase in emotional abuse, and a 32 percent increase in the number of violent incidents per month. As discussed below, the effects are unlikely to represent a change in reporting behavior. Our results are hard to reconcile with conventional models, most of which are optimistic in the sense that employment and other forms of economic empowerment of women is predicted to decrease abuse. We thus explore the ability of more recent, pessimistic violence models to explain our findings. Authors of instrumental violence models argue that a husband may turn more violent when his wife’s income goes up in order to counteract a rise in her bargaining power, or to increase the husband’s slice of a bigger income pie. But there is no indication that violent husbands in our sample have greater control over household resources, neither before nor after female employment.

Alternatively, physical abuse may be seen as a way to restore a traditional order in the household; either used by husbands to influence wives’ behavior, or generating direct, expressive utility for husbands. We argue that a natural adjustment to existing expressive violence models would allow the marginal utility that a husband derives from violence to increase when he is “disempowered” by his wife’s employment. Consistent with this, the increase in the incidence of violence is greater in households in which the newly employed wife was likely to end up further ahead of her husband in income because her baseline income was comparatively high relative to her husband’s. This paper’s findings have significant implications for theory and policy. We document that the form of female empowerment most forcefully advocated in the effort to reduce abuse of women – employment – increases rather than decreases domestic violence in the context of rural Ethiopia, and that the reason appears to be that men act upon the emotional costs implied by deviations from traditional household roles. We do not attempt to survey the literature on domestic violence here, but briefly summarize some of the most relevant papers. There are two cross-cutting dichotomies of domestic violence models: optimistic versus pessimistic models, and instrumental models in which violence is used to gain control over household resources versus models in which violence is not used to gain control over resources. Examples of conventional optimistic models include Chwe and Aizer . In Chwe , a male principal can use financial disincentives to discourage low effort from a high income female agent but must instead use costly violence disincentives to motivate a low income female. In Aizer , improvements in a woman’s expected utility outside of marriage, for example due to employment, is expected to reduce the level of violence she is willing to “offer” a husband who derives utility from violence. Aizer finds that decreases in the male-female wage gap in the U.S. reduce violence against women. There are several potential reasons why Aizer’s findings differ from ours. One possibility is that, in more male-dominated cultures such as that of many developing countries, the marginal utility men derive from violence may increase as women’s standing improves. Though not all the findings of previous studies can necessarily be interpreted causally, our results add to increasing evidence that nominal empowerment of women in poor countries can increase domestic violence. Eswaran and Malhotra find that women in India who work outside of the home are subjected to more violence. Gonzalez-Bernes concludes that female labor force participation in Zambia, Rwanda and Tanzania is not associated with lower levels of violence. The evidence for middle income countries is mixed at best. Instrumental models typically argue that men use violence as a tool to gain control over household resources, rather than as an end in itself. Examples of pessimistic instrumental violence models include Bloch and Rao and Bobonis et al. . Alternatively, men may derive “expressive” utility directly from violence, in which case physical abuse can be triggered by events that have purely symbolic meaning . This paper’s findings are most supportive of the expressive “male backlash” theories emphasized by sociologists .

The estimates in table 7 provide a clear picture. In a sub-sample of teams consisting of workers from two different tribes categorized as belonging to the same tribal bloc, little if any discrimination against non-coethnic processors occurs. The output of vertically mixed teams is for example not significantly different from that of homogeneous teams in the Luo – Luhya sub-sample. But within two different sub-samples of teams consisting of workers of two specific tribes categorized as belonging to different tribal blocs here, discrimination is pervasive and of an extent similar to that seen in the full-sample analysis in table 4. There are only minor differences across the Kikuyu – Luo and the Kikuyu – Luhya sub-samples, analyzed in columns 1 – 2 and 3 – 4 of table 7 respectively. So far we have seen strong evidence indicating that team-level ethnic diversity lowers productivity in the context of factory production in Kenya. If diversity effects are driven by discriminatory preferences, then we would expect the negative effect of ethnic diversity on private sector output to vary with factors that influence taste for discrimination, such as the political climate and relations between groups. A shift in taste for discrimination should differentially lower the output of mixed teams. In the next sub-section, I analyze differences in output between homogeneous and mixed teams during the period of ethnically-based, political conflict in Kenya in early 2008.The two coalitions in Kenya’s December 27 2007 presidential election were ethnically based. In advance of the election, grow lights for cannabis opinion polls predicted that the coalition led by Luo challenger Raila Odinga would oust the sitting Kikuyu- led coalition represented by incumbent president Mwai Kibaki. But results were delayed and the Kibaki victory announced on December 29 disputed by the opposition and the international community.

Widespread violence against Kikuyu and Kikuyu- allied tribes erupted, and counter-attacks soon followed. More than 1,200 people were killed and 500,000 displaced in the months that followed . On February 28, a peace agreement was reached, though violence continued in many areas, and it was not until after April 3 when the two sides reached an agreement on the composition of a power-sharing government that the political crisis ebbed. The conflict period significantly disrupted life in parts of Kenya. However, plant supervisors reported that logistics and worker absence at the farm was largely unaffected and that production continued as usual. Because the workers live on the farm in a gated community it was safest to remain on the farm. If the plant’s ability to operate was nevertheless affected, a decrease in productivity, as measured by the econometrician, should be observed in all teams. The model predicts an increase in the gap between the average output of homogeneous and mixed teams if attitudes towards workers of the other ethnic group worsened when conflict began. I interpret a possible increase in taste for discrimination as a decrease in the weight attached to the well-being of non-coethnics. In table 8, the difference in output between mixed and homogeneous teams before and after conflict began is compared. Data from 2007 and the first six weeks of 2008 is used. There was no significant change in the output of homogeneous teams when conflict began. If suppliers have social preferences, the impact of conflict on the productivity of homogeneous teams will reflect a combination of two factors. First, farm-wide disruption effects may have negatively affected output in all teams. Second, it is possible that conflict led to an increase in workers’ weight on the utility of coethnics: the findings of Eifert, Miguel, and Posner suggest that Africans increasingly identify with coethnics during times of heightened political competition between groups. I cannot rule out general disruption effects or an increase in the utility workers derive from coethnics’ output and income. But the combination of supervisors’ reports and a conflict coefficient for homogeneous teams that is essentially precisely zero points to little farm-wide disruption effects and little effect on workers’ weight on coethnics’ utility.

The output gap between homogeneous and vertically mixed teams nearly doubled in early 2008. Output in vertically mixed teams decreased by seven percent when conflict began. The results in table 8 thus indicate that upstream workers undersupply non-coethnic downstream workers to a significantly greater extent during times of ethnic conflict, as predicted by the model if taste for discrimination increased. Output in horizontally mixed teams decreased by four percent when conflict began, but there was a small but significant increase in the output of coethnic processors in horizontally mixed teams. An increase in upstream discrimination against workers of other ethnic groups thus appears to increase the supply of flowers to those downstream workers who belong to the same ethnic group as suppliers, as predicted by the model. The relative benefits of flowers supplied to coethnic processors in horizontally mixed teams go up if conflict lowers the utility upstream workers derive from non-coethnics’ output, even if suppliers’ weight on coethnics’ utility is unaffected. In light of the model presented above, the results for the conflict period thus suggest that discriminatory attitudes towards co-workers of other ethnic groups worsened in Kenya in early 2008. It appears that the economic costs of ethnic diversity vary with the political environment. A back-of-the-envelope calculation suggests that the increase in supplier discrimination during conflict may have cost the farm as much as US$560,000 in profit per year, had it not responded. Firms may be forced to take measures to limit distortions that arise from internal, ethnic discrimination, especially in times of conflict. In the next sub- section, I analyze how the gap in output between homogeneous and mixed teams was affected when the plant six weeks into the conflict period changed the pay system for processors and thereby altered the incentives facted by biased upstream workers.On February 11 2008, the farm began paying processors w per rose finalized by the team, rather than 2w per rose finalized by the processor herself as before.

As in standard incentive models, the framework above predicts that processors will freeride on each others’ effort when paid in part based on the output of the other processor. Free riding should negatively affect output in all teams, but in horizontally mixed teams an offsetting positive effect is expected. Under team pay, suppliers are unable to influence the relative pay of the two processors through relative supply. If the higher output for processors of the supplier’s ethnic group observed under individual pay is driven by suppliers’ taste for discrimination, indoor cannabis grow system a decrease in the output gap between coethnic and non-coethnic processors in horizontally mixed teams is thus expected when team pay is introduced. To test these predictions, I consider the period after processors’ pay system was changed and through the remainder of 2008 as a single team pay period. Figure 8 displays team and individual output during the three sample periods: pre-conflict , conflict , and the team pay period . The decrease in output in mixed teams during conflict is apparent. Comparing the second and third periods, the figure also clearly indicates that the introduction of team pay had a positive effect on output in horizontally mixed teams.Corresponding regression results are in table 9. The results indicate that team pay leads to some degree of free riding among processors: output in homogeneous and vertically mixed teams is 1 percent lower under team pay. The modest magnitude of this effect is noteworthy and interesting in itself. Output in horizontally mixed teams is four percent higher under team pay, as seen in columns 3-4 and 7-8 in table 9. The difference in output between horizontally mixed and homogeneous teams thus decreased significantly when team pay was introduced. The introduction of team pay essentially canceled out the effect of conflict on output in horizontally mixed teams, returning the difference in output between homogeneous and horizontally mixed teams to pre-conflict levels. The increase in horizontally mixed teams’ output appears to be due to horizontal favoritism being eliminated when biased suppliers’ ability to increase the relative income of favored processors through relative supply was removed, as predicted by the model. There is no statistically significant difference in the output of the coethnic processor and the non-coethnic processor in horizontally mixed teams during the last ten and a half months of 2008. An output gap of 32 percent between processors of the supplier’s ethnicity and processors who are not of the supplier’s ethnicity in horizontally mixed teams was eliminated by the introduction of team pay. The positive impact on output in horizontally mixed teams, which make up half of all teams, led to an overall increase in output when team pay was introduced. However, output in horizontally mixed teams remains lower than in homogeneous teams under team pay, and output in vertically mixed teams still lower. Under team pay a biased supplier continues to derive greater benefit from flowers supplied the more downstream workers belong to her tribe. The ranking of output of teams of different ethnicity configurations observed under team pay is thus due to incentives for vertical discrimination remaining in place, it appears. The model presented above, in which the productivity effect of ethnic diversity in teams arises from a taste for discrimination on the part of upstream workers, thus predicts the output response to the introduction of team pay well. Approximately one fourth of the yearly expected profit loss due to the impact of conflict on misallocation of flowers was avoided through the change in suppliers’ contractual incentives. It is difficult to imagine a standard economic model of joint production that would predict an increase in output when team pay is introduced.In the previous sub-section we saw that the economic costs of ethnic diversity vary with the political environment. The results in this sub-section suggest that, in high-cost environments, firms adopt “second best” policies to limit the distortions caused by ethnic favoritism. Group-based pay leads to free riding and reduces output in homogeneous teams, but the new pay system introduced by the plant during the conflict period in Kenya in early 2008 was likely designed to remove the ability of biased upstream workers to increase one processor’s pay relative to the other’s through differential allocation of flowers. Distortionary discrimination fell and the net effect was positive. Interestingly, La Ferrara also finds that ethnically diverse cooperatives in Nairobi are more likely to adopt group-pay. It thus appears that ethnic diversity has an important influence on how firms organize production in the private sector.Note first that informational and technological diversity effects are unlikely to explain this paper’s results. Suppose that the higher output observed in homogeneous teams during the pre-conflict, individual pay period was due to inferior technology or information in diverse teams. In that case it is difficult to see why output in mixed teams would fall differentially during conflict, and why the output of the two processors in horizontally mixed teams would be equalized under team pay. Cooperational effects have proven difficult to distinguish from social preferences , in part because such theories typically have few testable implications. Some forms of cooperational diversity effects could explain the observed decrease in mixed teams’ output during conflict. If trust for example facilitates cooperation, an erosion of trust between workers of different ethnic groups during times of ethnic antagonism could lead to a decrease in mixed teams’ output. Other forms of cooperational diversity effects could explain the observed increase in the output of non-coethnic processors in horizontally mixed teams under team pay. Coethnic processors that can exert effective social pressure on the upstream worker may for example induce the supplier to supply more to non-coethnic processors in horizontally mixed teams under team pay because processors derive benefits from the output of the other processor under team pay. It is, however, difficult to think of cooperational or other forms of non-taste-based ethnic diversity effects that can simultaneously explain a decrease in mixed teams’ output during conflict, equalization of processors’ output in horizontally mixed teams when team pay is introduced, and the other results seen in this paper. Though I cannot rule out that other forms of ethnic diversity effects also play a role, I thus conclude that the leading explanation for the lower output observed in ethnically diverse teams at the plant is taste-based discrimination on the part of suppliers. 

The two most commonly used summer cover crops in our region are annual buckwheat and sudan grass. Buckwheat is the fastest-growing summer cover crop, and when planted at a high enough density and irrigated up, annual buckwheat will outgrow and “smother” most of the fast-growing summer annual weeds such as pig weed and lambs quarter. Once incorporated, buckwheat residue breaks down easily, allowing for seedbed preparation soon after incorporation. Growers always comment on the noticeable improvement in soil quality following buckwheat. Note that buckwheat is not very drought tolerant and therefore must be sprinkler irrigated at least every 7 to 10 days on the Central Coast. If planting buckwheat with a drill, a good seeding rate is around 60 lbs per acre. If broadcasting, increase this amount to 80 lbs per acre. The advantages of Sudan are that it can be mowed and incorporated 40 days after planting when the plants are in full bloom and it is a good weed competitor. Sudan grass grows best during very warm weather, so during cool foggy periods its growth may be less than optimal for good biomass production and weed suppression. At the UCSC Farm we have had good luck intercropping vetch and sudan grass as a summer cover crop. The advantage of the intercrop is that if the weather is unseasonably warm during the initial growth stage the sudan will dominate and, conversely, vertical grow system if the weather is unseasonably cool the vetch will dominate.

In either scenario the cover crop will provide good biomass and weed suppression. Typically, when intercropping two different species it is advisable to plant each at half the recommended seeding rate. If planting a sudan grass/vetch mix with a drill, a good seeding rate is around 20 lbs per acre of each seed type. If broadcasting, increase this amount to 30 lbs of each per acre. AGS104 rye or Merced rye will both germinate well and provide excellent weed suppression when grown as summer cover crops in our region, and can be mowed numerous times to keep biomass manageable and to knock off developing seed heads of escaped weeds. Cereals like rye and oats are typically drilled at around 80 lbs per acre and broadcast at between 100 and 120 lbs per acre. Residue breakdown and subsequent seedbed preparation will depend on the length of time the rye, sudan or vetch covers are allowed to grow prior to termination. Sudan grass can be drilled at 40 to 50 lbs per acre and broadcast at 60 lbs per acre. The higher the seeding rate the finer the stem will be and the easier the breakdown will be at time of incorporation.The best tool for planting cover crops is either a no-till or conventional grain drill. Depending on the scale of operation, either three point or wider pull-behind drills can be used. All grain drills have single or double disc seedline openers, which facilitate planting into high residue situations often encountered when cover crops are planted following high residue cash crops such as corn or broccoli. Some drills, such as the no-till drills, have press wheels that run behind the disc openers, which help to re-establish capillarity to aid in bringing deeper soil moisture up to the seed; this feature greatly facilitates planting to moisture. The press wheels, which can be adjusted with spring tensioners, also facilitate accurately setting the planting depth, which is a critical factor as well when planting to moisture.

Accurate seed depth is also easily set with a drill and information on how to set depth can be found in the operator’s manual. Drills typically put down seed lines spaced from 6 to 7.5 inches apart, providing a close enough spacing for good early cover crop canopy closure, which will greatly reduce weed competition during the critical early cover crop establishment phase. Drills typically have adjustable seed drop openers that allow for some level of accuracy in setting seeding rates. It is advisable to “calibrate” a drill to improve the accuracy of seeding, and thus avoid either over planting and running out of seed or under planting and having seed left over . Drills are fast and efficient at field planting cover crops once the proper seed delivery rate has been determined. The double disc openers seldom clog, but it is not uncommon for clogging to take place in the drop tubes especially when the drill is being used to plant large seeds at a high rate. The drill operator must continually monitor the output of the drops to ensure that no clogging is taking place. It is advisable to check drop output visually from the tractor seat at the end of each pass. Another common problem is for the disc openers to pick up field trash that can jam the openers. The drill operator must also be cautious about not allowing the drill to move backwards while soil engaged, as this will often clog the openers with soil. A clogged drop is easily detected in the field since seed from the clogged tube will start to overflow at the top of the drop tube near the hopper, and a mindful operator will notice this overflow while running the drill. When using the drill after it has been parked for extended periods it is critical to blow out each of the drop tubes with compressed air or high-pressure water to clear out spider webs and other debris that can cause clogging.

When going into a field with the drill it often takes several feet of ground wheel operation for the seed to make its way through the delivery mechanism, down the drop tubes, and into the soil in the slot opened up by the disc opener. For this reason it is always advisable to make a final perpendicular pass along both edges of the field, filling in the areas that were potentially skipped as the drill entered and exited the field.If a drill is not available the next best option is to broadcast the cover crop seed with a relatively inexpensive, three-point tractor mounted broadcaster. Small-scale growers planting out small areas can effectively broadcast cover crop seed using commonly available and inexpensive hand cranked broadcast seeders. Seeding rates are challenging to set with broadcast seeders. Depending on the area to be broadcast it is often advisable to set the seed opening smaller than anticipated and make numerous passes over a field to improve overall uniformity of seed application. Note that it is important to measure out your field sizes and estimate the acreage prior to broadcasting so that you know exactly how much seed needs to be broadcast on each block. Recommended seeding rates are typically increased by 30% when cover crops are broadcast and harrowed, compared to drilling, to compensate for the lack of seeding depth uniformity. Once the seed is broadcast at the desired rate the grower must go back over the field with some type of secondary tillage implement to improve soil/seed contact to ensure adequate germination and minimize seed exposed on the surface. Secondary tillage implements commonly used to cover broadcast seed include spring tooth and spike tooth harrows and three point rototillers. Implement choice will often depend on the amount of residue in the field, since spring tooth and spike tooth harrows tend to bring residue to the surface, vertical grow system which can negatively impact cover crop stand establishment. The drawback to rototillers is that they are best operated at much slower ground speeds than other types of harrows, so covering large areas can be time consuming. Discs can also be used to cover broadcast cover crop seed, but setting the correct depth is critical to avoid placing the seed too deep and/or over mixing the soil. Tandem discs are better than offset discs for covering broadcast cover crop seed since they tend to move less soil and are less aggressive. If available, it is advantageous to pull either a ring roller or drag bar behind the disc or spring tooth harrow when covering cover crop seed to leave a uniform soil surface. When broadcasting and harrowing cover crop seed, it is inevitable that some seed ends up getting buried too deep and some seed may be left on the surface where it is less likely to germinate due to either bird feeding pressure or inadequate moisture.

Growers have several effective tools for dealing with weeds in cover crops. Perhaps the best tool is selecting the optimum seeding density and having the ability to plant uniformly, in terms of both density and seeding depth, in mid to late October for fall planted cover crops, when soil temperatures are conducive to quick cover crop germination; this allows the cover crop to effectively out compete weed seeds. Timing in relation to soil temperature is critical for success since cooler soil temperatures later in the fall will favor the success of winter weeds over the cover crops. Soil temperatures are not an issue with summer planted cover crops. Other weed management tools include the use of cover crops known for their ability to outcompete weeds through allelopathy. Good examples of these include mustards and many of the cereals—most notably cereal rye—when planted as monocrops. Though effective at outcompeting weeds, growers must be mindful of the challenges of spring incorporation of cereal cover crops when planted as pure stands. Although often difficult to achieve, one of the most effective winter cover crop planting strategies for good weed management is to drill cover crop seed into ground that has had a flush of weeds from either a light overhead irrigation or early rainfall event in the fall. Light tillage with a spring tooth cultivator or under-cutter bar at time of weed seed emergence will knock out the newly emerged weeds. If done correctly the cover crop seed can then be planted into residual moisture and will germinate without additional rainfall or irrigation. This scenario will provide a strong and weed free legume/cereal mix cover crop stand. This technique is dependent on the use of a drill for planting . Soil moisture is critical as well since too much moisture will have a potentially negative impact on soil compaction. An effective way to deal with emerging weeds in a newly planted cover crop is to go over the field very quickly with either a rotary hoe or a tine weeder just as the cover crop is emerging. This technique is referred to as “blind” cultivation and can effectively clean up a weedy cover crop field. If the timing is right, the cultivation from the rotary hoe or tine weeder will not negatively impact the emerging cover crop seed but will effectively disrupt, kill, and/or desiccate the newly emerged weed seeds that are much smaller and closer to the surface than the drilled cover crop seed. This technique depends entirely on timing in terms of the stage of development of the cover crop and the stage of development of the weed seeds as well as soil moisture. Tine weeders work best when they can be run perpendicular to the drill lines—particularly on soils prone to crusting.On the Central Coast of California, incorporation of high residue cover crops can be extremely challenging. Because of our mild maritime-influenced winters and relatively high rainfall rates , a legume/cereal mix cover crop may produce 2 to 3 tons per acre of residue calculated as “dry weight.” The average weight per acre of a standing legume/cereal mix cover crop just prior to incorporation can be over 20 tons per acre . At time of incorporation this residue typically has a very high moisture level and, depending on the level of maturity, can be carbonaceous and lignified. Because of these conditions it is advisable to flail mow the residue prior to incorporation to break up the stems into manageable sizes to facilitate incorporation into the soil. Timing of incorporation is directly linked to soil moisture and the level of maturation of the cover crop. Every spring is different and cover crop incorporation timing often involves a lot of guessing about potential rainfall patterns and soil moisture and cover crop maturation dynamics. A standing cover crop can transpire a tremendous amount of water and soil moisture can vary at different depths, making incorporation decisions challenging.

Once the tube is filled, a small hand-held suction pump is used to remove air bubbles from the tube. The lid of the reservoir is then retightened, sealing the lower tube. It is important to follow all of the manufacturer’s recommendations for installation and maintenance, including the use of an additive to minimize algal contamination of the water in the tensiometer. When used properly, tensiometers will provide accurate “soil/water tension” readings on a range of crops. These readings provide the irrigation manager with critical information that can be used to establish irrigation schedules adequate to maintain soil moisture at levels conducive to good crop growth and productivity.In many ways electrical resistance sensing devices are similar to tensiometers—the main difference is the method used to measure soil moisture. ERSDs utilize two “electrodes” cast into a porous material . The two electrodes in the “block” are attached to wires that run from the ERSD to the surface. These wires are often protected within a ½-inch PVC tube that is attached to the ERSD. The ESRDs are buried in the soil at various depths and locations, similar to tensiometers, and like the tensiometer, a soil/water slurry is used when the ERSD is installed to establish good soil contact with the instrument. To get a reading from the ERSD the irrigator uses a small, inexpensive, growing cannabis indoors hand-held electrical resistance meter that is temporarily connected to the wire leads from the buried ERSD. The meter allows a very low electrical current to flow between the two electrodes in the ERSD and displays an electrical resistance reading.

This reading reflects the amount of moisture within the porous material, since the buried ERSD takes on the moisture properties of the surrounding soil. Due to the electrical conductivity potential of water, the higher the concentration of moisture within the porous block the lower the resistance and, conversely, the lower the concentration of moisture within the block the higher the resistance. At field capacity the block is wet; as the growing plants start to extract moisture from the soil, the moisture is also pulled from the ERSD and the conductivity reading will reflect this change in soil moisture. Note that high salt concentrations in the soil solution will affect the accuracy of the reading, since salts increase electrical conductivity. This potential salt impact needs to be taken into account when deciding which monitoring tool is best suited to your farm. Electrical resistance sensing devices are relatively inexpensive and easy to install and monitor. Like tensiometers, they are left in the field for the duration of the cropping cycle and provide critical irrigation scheduling information that enables the irrigation manager to make informed decisions about irrigation frequency and quantity based on site-specific data.Central California’s Mediterranean climate creates the conditions that make dry farming possible. In normal years Central Coast rainfall is generated by storms that develop in the Gulf of Alaska and sweep south and then east, moving from the Pacific Ocean across the region from November through February and into March. High pressure then dominates the region from April through September and often into October, pushing rainfall to the north during the Central Coast’s long “summer drought.” Thus the region rarely receives significant rainfall from May through September.

Rainfall amounts vary considerably across the Central Coast, influenced in large part by the location, height, and orientation of the area’s numerous mountain ranges. Steeper ranges parallel to the coast can cause significant orographic lifting of moisture-laden air, resulting in high rainfall amounts on the west side of these slopes. These ranges also create rain shadows on the east sides, reducing rainfall in these areas. From San Luis Obispo County in the south to San Mateo County in the north, rainfall amounts vary from approximately 8 inches up to approximately 35 inches per year depending on the effects of the mountain ranges and specific storm dynamics.Higher afternoon temperatures and ET rates in the range of .33 inches per day, typically encountered in the more inland valleys with less marine influence, are much less suited to dry farming, especially of tomatoes, since it can be difficult for the plants to access deeper moisture quickly enough to maintain turgidity during periods of high evapotranspiration. However, some crops can be successfully dry farmed in inland valleys: although not within the scope of this article, wine grapes, olives, and apricots are successfully dry farmed in California on small acreages in areas with little or no maritime influence.The best soils for dry farming have relatively high clay content. Sandy loam soils or loam soils that overlay deeper clay soils also work well for dry farming. Soils higher in sand content do not hold soil moisture as well as clay and clay loam soils and therefore are typically not used for dry farming. And because organic matter increases the soil’s porosity, it does not improve conditions for dry farming. A grower considering dry farming should bore numerous holes up to 4 feet deep throughout the production area using a 2-inch slide hammer and soil probe to obtain soil “plugs”: soils suitable for dry farming will exhibit continuity within the different horizons and a loam or sandy loam upper horizon going directly to clay.

Horizons with a larger particle size, e.g., containing sand or gravel, will impede water’s ability to be drawn upward to the plant’s root zone, thus making dry farming less feasible. Preparing and planting a small area of the field is the best way to determine whether the site and conditions are suited to dry farming.Soil preparation that conserves or “traps” winter rainfall is critical for successful dry farming. In the spring, prior to planting, residual rain moisture is typically lost from the root zone as water percolates down through the soil horizon with the help of gravity. High clay content in the soil, and to a lesser extent soil organic matter , greatly facilitates the soil’s ability to hold water in the root zone against the pull of gravity. As the weather warms, soil moisture is also lost through surface evaporation. Evaporation occurs as water is drawn upward via small channels between soil particles; these channels can be thought of as capillaries within the soil horizon. Polar bonds between water molecules and the forces of cohesion facilitate water’s upward movement through the soil: as water near the soil surface evaporates, growing indoor cannabis water lower in the soil is pulled nearer the surface, much like liquid being drawn through a straw. Thus in fields destined for dry farming it is critical to break up the capillaries near the surface to minimize the evaporative loss of residual rain moisture during late spring and summer. This breaking of capillaries is typically accomplished with relatively shallow mechanical soil tillage. Commonly used tillage tools include rototillers and disc harrows, often followed by secondary tillage implements such as spring tooth harrows. The resultant tilled zone is called a “dust mulch.” This dust mulch provides an effective barrier to the potential evaporative loss of residual rain moisture held within the root zone of the soon-to-be-planted dry-farmed crop. When creating the initial dust mulch, timing is critical: the grower must trap as much rain moisture in the soil as possible, yet avoid working the soil when it is too wet. Wet soils, especially “heavier” soils high in clay content, are subject to clod formation and compaction caused by tractor operations. It is also important to minimize tillage depth when preparing soil for planting annual dry-farmed crops, since deeper tillage could disrupt the lower soil capillaries that are critical for soil water movement below the tilled zone. The dust mulch needs to be maintained with fairly frequent and light tillage operations from the time of initial tilling until the crops are too large to cultivate effectively. Although dry farming relies on winter rainfall, several scenarios can necessitate irrigation prior to planting. During dry springs it is sometimes necessary to pre-irrigate the beds before planting using either overhead irrigation or drip lines in order to establish an optimal stand. When a mechanical spader is used to incorporate a high residue cover crop prior to dry farming it is often necessary, in the absence of post-tillage rain events, to pre-irrigate with overhead sprinklers to facilitate the cover crop’s breakdown. On a garden scale, you may need to hand water the newly planted plants to assist in rooting and uniform establishment.In any dry farming system, variety selection is absolutely critical. Varieties that do well as dry-farmed crops typically have an aggressive root system capable of reaching deep into the soil horizon to tap the stored rain moisture. It is interesting to note that growers in the Central Coast region have trialed literally hundreds of varieties of heirloom, open pollinated and hybrid tomatoes and, to date, none have compared to ‘Early Girl’ in their ability to set roots deep and consistently produce a high yield of high quality, flavorful, and marketable fruits with no irrigation. ‘New Girl’, a recently introduced variety, is closely related to ‘Early Girl’ and appears to have many of the same favorable characteristics.Dry-farmed crops with extensive root systems can effectively extract deep residual rain moisture from a fairly large area within their roots’ grasp.

Competition from other nearby crop plants or weeds can result in water-stressed plants that produce very little fruit and remain stunted. For this reason it is critical to plant out dry-farmed crops in a much wider spacing than is typically used for irrigated crops of the same type. Good weed management in a dry farm system is also critical, since most weeds have aggressive root systems capable of outcompeting most crop plants for both water and nutrients. As an example of plant spacing, irrigated tomatoes are commonly spaced 2 feet apart within the row with rows spaced 4 feet apart, a density of roughly 5,400 plants per acre. A typical spacing for dry-farmed tomatoes would be 6 feet between rows and 6 feet between plants, for a total plant population of 1210 plants per acre. As you can see from this example a significant yield reduction can be expected from most dry-farmed crops simply based on per acre plant populations. A higher price premium for dry-farmed tomatoes will often make up for the yield loss related to wider spacing.As a rotation within a diverse irrigated cropping system, dry farming has many advantages. The lack of irrigation in a dry-farmed production block can lead to improved soil tilth, since dry surface soil is not prone to compaction or clod formation from both foot traffic associated with harvest and tractor compaction from cultivation operations. Problem weeds are much easier to deal with when irrigation is eliminated for a season and weed seed development is easily minimized in a dry-farmed block. If water is a limited resource on a farm then dry farming makes perfect sense as a means of maintaining production while eliminating the need for irrigation. Forcing deep rooting of dry-farmed crops can also facilitate the extraction of nutrients that have leached below the root zone of most irrigated crops through excessive rainfall or irrigation. This is particularly true of tomatoes, which are highly sought after by savvy consumers and the Central Coast region’s chefs. As a result, the production and sale of dry-farmed tomatoes has become an important and economically viable niche market for small-scale organic specialty crop growers on the Central Coast. Finally, although dry farming may not be appropriate for every cropping system and region, understanding the basic principles of dry farming can lead to a greater knowledge of the complexities of water and soil dynamics, tillage, weed management, and fertility management. This knowledge can in turn lead to a greater understanding of your particular production system. In regions where conserving water is critical, applying dry farming principles to irrigated systems can result in improved water use efficiencies, better weed management, and improved soil tilth and productivity.While these statistics clearly illustrate the enormous quantity of water used in agriculture, they also suggest that irrigation has far-reaching consequences on water quality. In an effort to maximize crop yields, many farmers apply nitrogen-based synthetic fertilizers. More than half of the nitrogen applied may go unused by crops, ending up in surface water runoff or leaching into groundwater and causing severe water quality and other public health concerns for rural communities, many populated by poor, immigrant farm workers.3 4 As this supplement illustrates, how farmers use irrigation and apply fertilizers affects not only their crops, but also their neighbors.

As expected, the dendrograms pruned with the ensemble of simulations that accounted for both measurement error and longitudinal consistency of the underlying behavioral pattern produced encodings that were far more granular. A total of 13 clusters were returned for the unweighted Euclidean metric, 17 for KL distance, and 14 for both the noise-penalized and the plasticity-penalized dissimilarity metrics. In Figure 5B, the heatmap visualization of these pruning results for the plasticity-penalized dissimilarity metric reveal an encoding that is coarser but ultimately quite well balanced, with the pruning heights modulated to produce cluster sizes that were reasonably uniform across the domain of support. Closer inspection revealed that this final encoding largely matched the order of bifurcations in the original tree, except that this pruning strategy left no animals isolated in anomalous clusters. It should be noted, however, that the granularity of this encoding is not entirely intrinsic to this system, but was dependent on the size of the subsample used to calculate the overall time budget in each simulation. While we can expect cows that were more inconsistent in their daily time budgets to be subjected to a stronger penalty with this estimator due to relatively higher rates of sampling error imposed by the subsampling routine, we can also anticipate that the overall scale of the sampling error imposed on all cows should grow as the size of the subsample is reduced. This would in turn modulate how quickly the underlying behavioral signals would be drowned out by simulated noise within the tree. This suggest that, cannabis trim tray for larger samples where a greater range of subsample sizes can be utilized, this simulation value can also be treated as a meta-parameter to tune the granularity of the final encoding.

Given that the plasticity-penalized mimicry was created for this data set by subsampling only 14 out of 65 observation days, the resolution achieved in the pruned encodings for all four dissimilarity metrics reinforces that this herd was overall fairly consistent in their daily time budgets, and that this data set will support fairly detailed inferences against a strong underlying behavioral pattern.Encodings of the overall time budgets produced using both the noise and plasticity penalized dissimilarity estimators, wherein both were pruned using the more conservative plasticity-penalized ensemble, produced similar behavioral insights when compared against longitudinal patterns in parlor entry position across the herd. For the bivariate analyses run with encodings for all 177 cows with complete records, highly significant associations with entry order were recovered for both the noise-penalized and plasticity-penalized time budget encodings. The bivariate relationship was optimized for both time budget encodings with a five-cluster encodings of entry order patterns. The noise-penalized encoding produced the strongest associations with entry order with seven time budget clusters, whereas the plasticity-penalized encoding performed better with a finer encoding of nine clusters, the key difference being the degree of stratification among animals with the most moderate time budgets. Visualization of the contingency tables for the optimized encodings colored by their PMI estimates revealed that the significant overall association between the two data streams was driven predominantly by animals in the latter half of the milking queue. Figure 5 displays the results for the noise-penalized encoding. We see first that cows that entered consistently at the very rear of the queue were significantly over-represented in the time budget cluster characterized by moderate time spent eating, low time nonactive, and high rates of rumination . Cows that entered nearer the back of the queue , just ahead of the cows that consistently brought up the rear, were also over-represented in the same time budget cluster ± a trend that was statistically significant for the plasticity-penalized encoding but only marginally significant for the noise-penalized encodings.

In fact, very few animals that entered in the front half of the queue were found to have this time budget pattern, with cows entering just behind the leaders being significantly under-represented in this time budget cluster. One potential interpretation of this pattern might be that, if these cows are prioritizing time investments in rumination, then this strategy may include hanging back towards the later part of the queue, where they may be able to chew their cud while avoiding the more serious contention for parlor entry position. Further analysis that could facilitate visualization of the cyclical patterns in this time budget data would be needed, however, to confirm this suspicion, and will be left for future work.While this more moderate tradeoff between rumination and non-activity demonstrated a fairly straightforward and progressive trend across the milking queue, which might readily have been captured by a linear model, more complex dynamics were found for the time budget cluster characterized by extremely low time spent eating and high time spent ruminating and nonactive . Cows that consistently entered at the very end of the queue were significantly under-represented in this extreme time budget, while the cows that entered just ahead of them were significantly over-represented. While an extreme tradeoff in eating and ruminating might be explained by issues with sensor placement, that such cows are not evenly dispersed across the herd may instead indicate a biological driver. Health status naturally comes to mind with such an extreme time budget, and indeed several previous studies have reported higher rates of health complications amongst animals in the latter part of the milking queue , but health status alone would not necessarily explain the inversion in association pattern between these two adjacent queue groups.

Time budgets provide a convenient and intuitive means of quantitatively summarizing the behavioral tradeoffs of animals, but multinomial-distributed data present a number of analytical challenges. The results of this analytical case study have highlighted how a novel simulation based approach may be employed to simultaneously accommodate both the codependency structures fundamental to multivariate-distributed data formats and the complex multi-faceted sources of measurement uncertainty that may be encountered across a broader ranger of PLF data streams. While such simulations may be more computationally expensive than closed form estimators, we have demonstrated that an ensemble of data mimicries can be efficiently repurposed throughout the analytical pipeline to improve not only the visualization of these behavioral tradeoffs, but also the compression of such information into robust empirically defined discrete encodings. It should be noted, however, cannabis trimming tray that the utility of these novel clustering techniques are not restricted to time budget data. The ensemble-penalized dissimilarity estimator and ensemble cut algorithm that we have introduced in this case study are both fundamentally non-parametric. This means that their implementation is in no way intrinsically restricted to any particular class of data. Subsequently, the choices that a user makes in constructing an appropriate error simulation model are restricted only by their own creativity, allowing this analytical framework to be easily generalized to a much wider array of PLF data streams and the wider array of complex error structures that they have to offer. Additionally, while discrete data is typically seen as an impediment to statistical analysis in most model-based approaches, we hope that this analytical case study has served to demonstrate the comparable ease with which insights may be extracted from encoded data when an information theoretic approach is employed. For large, structurally complex, and often informationally redundant PLF data streams, an efficient encoding may be far easier to achieve than a comprehensive model that can fully accommodate the temporal dynamics of behavioral responses in dynamic farm environments. This may be especially true for data sets where all the factors driving such behavioral responses are not measurable. While more formal model-based inferences may be warranted for further analysis of the underlying causes of this relationship, the exploratory analysis tools provided by the LIT pipeline have undoubtably served to create a more comprehensive picture of the complex behavioral dynamics hiding within these two underutilized data streams.Precision Livestock Farming technologies create new opportunities to record the behaviors of large groups of socially housed animals in complex working farm environments. The datasets produced by such technologies, however, often contain measurements collected more frequently and over longer observation windows than is possible with direct human observation, resulting in complex structural features that in turn present new logistical challenges in extracting usable knowledge from these rich data streams . In previous work, we introduced the concept of ensemble mimicries, wherein minimally parametric simulation techniques can be integrated into standard hierarchical clustering pipelines in order to account for complex nested sources of measurement error often encountered in PLF applications. In analyses of overall time budget data, this not only allowed us to account for heterogeneous variance in sensor noise, but also to penalize animals whose behavioral responses were more variable over the observation window. This subsequently produced encodings that gave relatively more weight to cows with the most clearly defined behavioral responses that were pervasive across environmental contexts .

In previous work with milking order records extracted from metadata produced by RFID-equipped rotary milking parlors, however, we also demonstrated that interesting ethological insights can be gleaned by recovering systematic temporal variations in behavioral responses using a data mechanics clustering approach, even when the environmental factors eliciting changes in queueing patters were not recorded and only the behavioral artifacts remained in the response data . In light of these collective results, an algorithmic framework is needed to disaggregate overall time budgets to determine if variability that was previously penalized using ensemble mimicry approaches might in fact be hiding systematic heterogeneity in daily time budgets that might provide further insights into the behavioral responses of this herd of cows across management contexts. In previous work with entry order records, we showed that data mechanics can be conceptualized as a form of nonlinear PCA, and thus used as a discrete but more readily explainable alternative to manifold embedding algorithms . Its efficacy has been demonstrated for information compression for multivariate response data and for repeated measures of univariate response , but daily time budget records are a multivariate repeated measure. While it would be possible to flatten daily time budget records in order to analyzed this dataset as a 2D matrix using the standard data mechanics analysis pipeline, this would afford no control or differentiation in weighting the behavioral and temporal axes in the final encodings. Forcing the algorithm to learn, from the data itself, fundamental structural features that could easily be specified a priori, is inherently inefficient. This chapter will therefore explore how data mechanics algorithms can be extended, not only to incorporate information about measurement error from new ensemble mimicry techniques, but also to utilize tensor constructs to more naturally accommodate the complex temporal structures often encountered in PLF data streams. The result will be an algorithmic framework that will allow ethologist to leverage simultaneously the multivariate and temporal richness of similar PLF data streams, in order to provide an approach tomodel-free knowledge discovery that is not only more statistically powerful, but will also provide more wholistic ethological inferences.To explore the efficacy this proposed analytical framework, data was repurposed from a feed trial conducted on a working dairy farm assessing the efficacy of an organic fat supplement on cow health during early lactation . All animal handling and experimental protocols were approved by the Colorado State University Institution of Animal Care and Use Committee . The study was conducted in 2017, from January through July, on a USDA Certified Organic dairy in Northern Colorado. Cows were housed in a closed herd in an opensided free-stall barn that was stocked at roughly half capacity with respect to both feed bunk spaces and stalls, with free access to an adjacent outdoor dry lot. The grazing season began in April, at which point cows were moved onto pasture each night to comply with organic grazing standards. Cows had free access to TMR between milkings while in their home pen, and were head locked each morning to facilitate daily health checks and heat detection. For more details on feed trial protocols, the housing environment, and the day to day management of animals, see Manriquez et al. and Manriquez et al. . In order to facilitate direct comparisons with previously reported work, daily time budget records were calculated for this methods paper using the same CowManager ear tag accelerometer records used previously to analyze overall time budgets . Data scrubbing decisions used to exclude incomplete records were identical to those reported previously to calculate overall time budgets, with the only appreciable difference in data wrangling being that here hourly time budgets were aggregated conditional onday of observation.

In more recent work, cows may have been motivated to access limited feed bunk space on commercial dairies or to obtain prime pasture . In this study, as all animals were locked following milking to facilitate feeding treatments and health checks, cows would have had ample access silage regardless of queue position. Alternatively, Rathore suggested greater intermammary pressure might motivate high yielding animals to be milked earlier. As this herd was milked three times daily, however, this biological driver may also have been attenuated. Indeed, among modern studies with herds milked thrice daily, Polikarpus et al. found no significant correlation and Grasso et al. also found high yielding cows frequented the rear of the queue. Ultimately, as yield is influenced by a wide range of health and management factors, any number of confounding variables might be implicated in explaining this somewhat unexpected result. In this study a significant linear association between ageand entry position was not found. Recent work by Berry and McCarthy , which identified a nonlinear trend across parity, and by Grasso et al. , which highlighting significant interactions of parity with other biological drivers of queue position, suggests that a linear effect may not adequately capture the underlying biological relationship. A larger and more structured sample may be necessary to bring more complex age dynamics into clearer resolution.Visual inspection of means plots produced from mixed model analysis of sensor records recovered only a handful of statistically significant differences between queue quartiles when hour and day effects were assessed individually, trimming trays but several global trends were still readily visible.

With respect to minutes recorded as active, the 1st-3rd queue quartiles were visually indistinguishable in their cyclical behavioral patterns, but cows in the fourth queue quartile were consistently more active, particularly during the night and morning lounging period. With respect to longitudinal trends across days, fourth queue quartile animals were again more active across the observation window, whereas cows in the first queue quartile were consistently the least active. These patterns were somewhat mirrored in the longitudinal and cyclical analysis of high activity minutes, but the pattern was both less distinct and less consistent. No clear qualitative insights could be drawn for cyclical or longitudinal patterns in nonactivity. Cyclical patterns in minutes spent eating were not seen overnight or in the afternoon, but first queue quantile cows may have spent slightly more time eating after the morning milking. Longitudinal analysis of eating patterns suggested cows in the fourth queue quartile spent relatively less time eating, whereas the cows in the first and second queue quartile consistently spent more time at the bunk. This contrasted with longitudinal results for minutes spent ruminating, where the cows in the second queue quartile were consistently low. No clear distinctions between groups were recovered in cyclical rumination patterns. Temperature patterns were, surprisingly, the most visually distinct of all the sensorparameters. Cows in the first queue quartile were consistently lower in body temperature in both the longitudinal and cyclical time dimensions as compared with the remainder of the herd.

While the preceding analyses revealed few statistically significant differences at individual time points, collective analysis of days and subsets of the 24-hour management cycle would uundoubtedlyreturn statistically significant differences for the broader qualitative trends visually identified via mean plots. Within a linear modeling framework, however, this constitutes no small task. Group-by-date interaction effects were also significant for activity, high activity, and temperature models . This suggests that these models should not be simplified to a single cyclical or longitudinal trend, which would allow overall differences between groups to be tested via a single group intercept term. Targeted hypotheses comparing comprehensive trends between groups would instead require formulation of linear contrasts ± a daunting task with so many fixed effects terms used to accommodate the high sampling frequency and extended observation period of this dataset. Further, as with the linear models with cow attributes, behavioral synchronization due to social cohesion or compensatory use of physical resources in the pen could again create non-independence between animals in such sensor records. Any such issues in estimation of model degrees of freedom, compounded with the inability to fit behaviorally and empirically compelling correlation and variance models, would only serve to further confound the estimation of appropriate p-values from these models. Fortunately, the qualitative trends identified via the preceding means plots largely aligned with the significant bivariate associations identified by mutual conditional entropy tests summarized in Table 2. Activity again proved to be the most distinctive behavioral axis. Significant associations were identified for all three lounging periods when analyzed both independently and in aggregate, with the afternoon lounging periods being the most distinct. High activity also showed a significant relation to queue records, but this association may have been driven predominantly by overnight lounging period. Whereas no clear qualitative patterns were identified for non-activity data via the means plots, a significant association with queue records was identified during the afternoon lounging period.

A highly significant relationship was identified for time spent eating for the full sensor record, but given that time budgets recorded by this platform were segmented somewhat arbitrarily at the start of each hour, this result may simply reflect a lag in the arrival of cows to the feed bunk after exiting the parlor. Significant associations were not found during the lounging periods at the standard 0.05 cutoff, though records from the afternoon lounging period approached significance. These results were mirrored in rumination patterns, where again no significant association was recovered, but the afternoon lounging period approached significance. Finally, as with the linear modeling results, temperature proved highly distinct between queue subgroups for all subperiods.Visual inspection of tube plots produced with median queue subgroup values again yielded insights comparable to the linear modeling results . Based on the results of the MCE tests, all behavioral axes cows were clustered into two subgroups based on queueing records, with Group 1 cows consisting of 80 animals at the front of the queue, and Group 2 cows constituting the 34 animals in the rear. Tube plots of minutes spent active revealed Group 2 cows to be more active across all three lounging periods. This pattern was the most consistent in the morning and overnight lounging periods, though this difference was ultimately quite subtle and seldom constituted more than a few minutes. In the afternoon subperiod there was evidence of several periods with anomalously high activity levels, most of which occurred post-pasture access. The significant association recovered for minutes spent highly active in the overnight subperiod appeared to be largely driven by increased activity immediately following the evening milking, trim tray which could reflect divergent home pen behaviors, but might also have been driven by delays in milking. To complement these results for active and high active minutes, the significant association for afternoon non-activity records appears to have been driven by increased non-activity among the Group 1 cows during the three hours immediately preceding the night milking. As anticipated, differences in time spent eating were largely restricted to the 2-3 hours immediately following milking. Cows only lingered at the feed bunk during the morning lounging period, where median eating times for Group 1 cows were perhaps slightly higher. Similarly, differences in rumination also appeared restricted to time periods immediately following milking, with no clear differences seen during the lounging period with this coarse stratification of animals.

Finally, as with the mean plots, body temperature values again produced surprisingly distinctive results. More finely segmented into five queue groups by the mutual conditional entropy test, the tube plots proved a slightly cumbersome means of comparing temperature records, but a clear visual distinction could still be made between the Group 2 animals and the remainder of the herd. For all three lounging periods, this relatively small cluster of 17 cows that constituted the very front of the milking queue demonstrated lower median body temperature values, a distinction seen most clearly at night.The strong agreement between the results of these two analytical pipelines suggests that UML and conventional linear modeling approaches could be used interchangeably or in concert to glean preliminary insights from exploratory analyses of large sensor-based datasets that may inform future hypothesis-driven studies. Perhaps the most surprising result of these analyses, that cows frequenting the back of the queue are consistently more active between milkings, may indeed warrant further exploration. In much of the prior literature, health challenges that impede movement have been identified as the main driver of delayed entry into the parlor . In fact, this mechanism is so well-established that it has even been proposed that milk order records might be incorporated into genetic evaluations to improve estimates of health traits . As these analyses were run on the subset of animals with no recorded health events, however, it is possible that this dataset has brought other behavioral mechanisms into focus. Previous studies have found that animals of low social status frequent the rear of the herd in voluntary movements , and social dominance is known to impact resource access in spatially constricted conditions such as those found at the entrance to the milking parlor. If low dominance animals are in turn also forced to wait longer or walk farther to access resources in the home pen, this could potentially explain the increased activity levels of animals found in the rear of the queue. While the early literature has found the relationship between dominance value and milking order to be tenuous at best , it is possible that such social mechanisms may have been confounded by health status, with linear analyses of limited sample size failing to disentangle these mechanisms in non-disaggregated data.On this farm, where resources are not severely restricted and animals are frequently remixed, energetic investments in a dominance hierarchy may offer few returns . Such a behavioral strategy might also explain why it is high-yielding multiparous cows and not heifers that occupy the end of the queue. Both these hypotheses are ultimately purely speculative interpretations of these exploratory results; however, if proposals to incorporate milk order records into genetic indices are progressed, any correlations between queueing position and consistent individual differences in home pen behaviors likely warrant closer inspection to mitigate the risk of unintended and potentially deleterious selection pressures.As with previous studies of milk order records, these analyses perhaps raise more questions than answers. As dairy record management systems grow to accommodate an ever wider range data streams, perhaps future work considering more herds from a wider range of management strategies will succeed in further untangling the complex web of explanatory variables at the individual, herd, and farm levels that drive variation in queueing patterns. This dataset has, none the less, demonstrated the utility of unsupervised machine learning tools in ethological studies using sensor platforms to study larger groups of animals over extended periods of time. While these analyses recovered no evidence of social cohesion amongst large or temporally consistent subgroups, information theoretic approaches succeeded in clarifying the underlying pattern of heterogeneity in error variance between animals and also demonstrated an advantage in recovering evidence of non-uniform patterns in temporal nonstationary over basic EDA tools. After incorporating these insights into the structure of subsequent linear models, these model-free tools then showed some capacity to confirm inferential results where probabilistic assumptions were not strictly met, as well as an aptitude for recovering significant associations not captured by a simple linear effect. This flexible clustering-based approach to identifying significant bivariate associations was then easily extended to accommodate two high dimensional behavioral axes, providing equivalent insights to more computationally taxing mixed effect models. While UML approaches are by no means infallible, as seen here with artifacts produced by the spectral embeddings, these analyses have demonstrated that such tools can add value at every stage of the standard hypothesis-driven linear analysis pipeline, and may even offer an advantages over model-based approaches in early-stage exploratory projects. While many new methodological developments are doubtless on the ethological horizon, we hope this algorithmic toolset will provide a meaningful step forwards to meet the challenges of a future defined by ever larger and more complex data.Precision Livestock Farming technologies produce prodigious amounts of data . While the behaviors encoded by such sensors are often much simpler than those that can be quantified by a human observer, the measurement granularity and perseverance provided by these technologies create new opportunities to study complex behavioral patterns across time and in a wider range of contexts.

This model-free framework can subsequently be extended to a multivariate estimator with two or more discrete variables. In the bivariate case, the distribution of one variable is compared across each encoded level of the other in order to decompose the total entropy in the joint encoding into three terms: the conditional entropy unique to the first variable, the conditional entropy unique to the second variable, and the mutual information that is redundant between the two encodings . This mutual information estimate in turn reflects how much information we learn about one encoded variable if we know the value of the other, and subsequently can be used to reflect the strength of a bivariate association between two sets of encoded data regardless of the underlying dynamic ±linear, quadratic, exponential, etc. Suppose our farmer with the overstocked cows, now fully aware of the welfare issues this management choice has created, reduces their stocking rate to a 1:1 ratio and continues to monitor the lying time of the animals to provide proof to their milk buyer that the issue has been resolved. After several months at this lower stocking rate, they review the data and are dismayed to find that there are still days where animals have inadequate lying times. To solve this new problem, they hire yet another consultant to help them track down the source of the problem. Suppose that this new dataset were collected in the summer, high quality shelving systems and so naturally, this consultant includes Temperature-Humidity Index as one of many candidate variables to consider as a potential source of the continued welfare concerns for this herd .

Using stochastic sampling techniques, the full details of which are provided in Supplemental Materials, we have simulated a fairly straight forward but nonlinear dynamic between these two variables. On days where the observed THI values are low, animals are not heat stressed and so spend the majority of their day out on pasture grazing. As the THI rises, cows become heat stressed for progressively larger proportions of the day, resulting in a gradual increase in the proportion of each day that cows spend lying down in the shade of their free stall barn. Above a certain high THI threshold, however, cows struggle to thermoregulate while lying down, causing them to stand for extended periods of time. When lying time is plotted against THI, as in Figure 7A, we can see that there is a clear nonrandom pattern in this data that is perhaps best characterized by a threshold model ±a dynamic that is commonly found when a single behavioral response is subject to the influence of competing underlying behavioral response mechanisms. If a simple linear effect were utilized to probe for a significant bivariate association between these two variables, however, a near-zero slope would be returned, as shown by the red line overlaid in Figure 7A. In this case, not only would a Pearson correlation test fail to identify this nonrandom but also nonlinear pattern, but because this pattern is also not monotonically increasing, even anonparametric Spearman Rank correlation test would fail to identify THI as a significant influence on lying times within this herd. Suppose both the THI and lying time measurements are discretized using simple equal-sized binning rules. If we compare the mutual information estimate from the two observed encodings against estimates generated from a simple permutation, the resulting p-value for this test of bivariate association would be highly significant .

To further characterize this dynamic, a simple contingency table, wherein each cell represents the total number of observations for each joint encoding, can be easily visualized as in Figure 7B. The mutual information estimate for the overall bivariate association can subsequently be decomposed into point wise mutual information estimates to reflect how much each cell in the observed table differs from the counts that would be expected by multiplying the marginal probabilities, which would be the distribution of joint observations anticipated if no association existed between the two encodings . Here blue cells indicate that there are fewer observations with the corresponding joint encoding than would be expected if no association between these variables were present, whereas orange cells are over-represented relative to the null. From this visualization we can clearly see that the probability of observing a given lying pattern is being shifted in different directions based on the level of the THI encoding. Thus, absent any prior intuition or assumptions about the relationship between these two variables, an information theoretic approach has successfully identified a significant bivariate association and provided insights into the underlying dynamic to inform further interpretation of the underlying behavioral mechanisms at play and subsequently the correct management interventions needed to remediate this welfare concern.For much of its history, ethological research in livestock has relied on human observers to encode behaviors of interest . While developing a detailed ethogram and observer training protocols constitute no simple task, there are several inherent advantages to this approach for subsequent statistical analyses. Continuous involvement of a human in the incoming data stream allows many erroneous data points to be identified and excluded from downstream analyses that they might otherwise destabilize. Extensive involvement of research personnel in the data collection phase also nurtures a deeper familiarity with the system under study. This not only aids in the specification of an appropriate statistical model and interpretation of results, but is often critical in identifying unexpected behavioral patterns that can inspire novel hypotheses. Unfortunately, the inherent quality of such data imposes practical limitations on the quantity that can be produced. This can restrict both the number of animals utilized in a study and the period of time over which they are observed. Restrictions on the number of animals that can be studied, on the other hand, can fundamentally alter the behavioral mechanisms at play in a herd. For example, the linearity of dominance hierarchies are known to change with group size .As commercial herds and flocks become ever larger, this only serves to broaden the gap between experimental findings and the welfare challenges they are meant to inform. Subsampling of animals or observations windows may be employed to reduce the number of observations collected without restricting the size of the study system. If the pre-existing base of scientific literature does not provide clear guidance on the selection of target animals or focal periods, however, such strategies may risk overlooking finer-grain behavioral patterns and skewing inferences about the collective behavior of the group . In recent years, livestock sensor technologies have become a popular alternative to visual observation . While the behaviors recorded are neither as complex or as detailed as those quantified via an observational ethogram, such devices have the capacity to continuously monitor hundreds or even thousands of animals for extended periods of time. Such a substantial expansion in the bandwidth capacity of ethological studies creates many new opportunities to better understand the behavior of livestock, particularly in large-scale commercial settings, commercial drying rack but also raises new methodological challenges.

Replacing nuanced human intuition with basic computer logic may increase the risk of erroneous data points, an issue that is only further compounded by the scale of data produced by such technologies, which renders many conventional visualizations techniques ineffective in identifying outliers. Observations recorded over extended time periods with high sampling frequency from large heterogeneous social groups may also contain a range of complex stochastic features ±autocorrelation, temporal nonstationary, heterogeneous variance structures, non-independence between experimental units, etc. ±that can lead to spurious inferences when not appropriately specified in a conventional liner model. The hands-off and somewhat black-boxed nature of many sensor platforms, however, do not nurture the intuition needed to identify many of these model structures a priori. Such insights must instead be drawn directly from the data itself, but here again, standard visualization tools may not scale to such large datasets. Unsupervised machine learning tools offer a distinct empirical approach to knowledge discovery that are purpose built for large and complex datasets . Whereas conventional linear models excel at providing answers to targeted experimental hypotheses, UML algorithms strive to identify and characterize the nonrandom patterns hiding beneath the stochastic surface of a dataset using model-free iterative techniques that impose few structural assumptions. This open-ended and highly flexible approach to data exploration may offer an empirical means by which to recover much of the familiarity with a study system that is lost with the shift from direct observation to sensor platforms. The purpose of this research was contrast the behavioral insights gleaned from UML algorithms with those recovered using conventional exploratory data analysis techniques, and to then explore how such information could be best integrated into standard linear analysis pipelines. Milking order, or the sequence in which cows enter the parlor to be milked, is recorded in all RFID equipped milking systems, making such records one of the most universal automated data streams to be found on modern dairies. Despite their ubiquity, such records are seldom used to inform individual or herd-level management strategies. This lack of utility, however, has not been for lack of study. The modest base of scientific literature that has since been compiled on this topic, however, has struggled to recover repeatable evidence of such associations . While such inconsistency may simply reflect non-uniformity in the behavioral strategies driving queueing patterns across different herds and farm environments, misspecification of the linear models used to describe this system could also contribute to volatility in these statistical inferences. The objective of this methodological case study will be visualize the various stochastic aspects of such records using UML tools in an effort to identify erroneous data points and heterogeneous variance structures that may not be recovered using conventional EDA techniques.Data for this case study was repurposed from a feed trial assessing the effect of an organic fat supplement on cow health and productivity through the first 150 days of lactation . All animal handling and experimental protocols were approved by the Colorado State University Institution of Animal Care and Use Committee . The study ran from January to July of 2017 on a certified organic dairy in Northern Colorado. A total of 200 mixed-parity Holstein cows were enrolled over a 1.5 month period as study-eligible animals calved. Cows were maintained in a closed herd for the duration of the study, with sick animals temporarily removed to a hospital pen when necessary. The study pen was an open-sided free stall barn, stocked at just above half capacity with respect to bunk space and beds, with free access to an adjacent outdoor dry lot. At roughly the midpoint of the trial, cows were moved overnight to a grass pasture that conformed with organic grazing requirements . Cows had access to total mixed ration ration following each milking. Animals were temporarily split into two subsections of the pen following the morning milking to facilitate administration of control and treatment diets. Cows remained locked for roughly 45 minutes following this division so that farm and research staff could collect health and fertility data. Additionally, all animals were fitted with CowManager® ear tag accelerometers . This commercial sensor platform, while designed and optimized for disease and heat detection, also provided hourly time budget estimates for total time engaged in a range of behaviors – eating, rumination, non-activity, activity, and high activity – as well as average skin temperature.Raw milk logs were exported from the rotary parlor following each morning milking , and were processed using data wrangling tools available in R version 3.5.1 . To account for missing records due to illnesses and RFID reader errors, ordinal entry positions were normalized by the total number of cows recorded in a given milking . For example, if a cow were always the last animal to enter the parlor, her ordinal entry position might vary widely with herd size, but her entry quantile would always be 1. The first 55 days of records were excluded from analyses to allow all animals to enter the herd over the rolling enrollment period and become established in their parlor entry position . To avoid irregularities in cow movements, several observation days surrounding management changes were also dropped, including: the two days preceding transition to pasture, the four days following pasture access, and the final seven days on trial.

Sordariomycetes enrichment may indicate other community shifts that are ultimately the cause for enhanced fruit quality. Endophytes in the Hypocreales class, which was enriched in dry farm fields, are known to increase drought resistance and decrease pest pressure in their hosts, though none of the specific species known to exhibit this behavior were enriched in dry farm soils. On the other hand, Nectriaceae, the family that contains the Fusarium genus, was found to be enriched, though similarly no known pathogenic species were enriched in dry farm soils.Our study explored dry farm management practices and their influence on soil nutrient and fungal community dynamics in 7 fields throughout the Central Coast region of California, allowing us to explore patterns across a wide range of management styles, soil types, and climatic conditions. Though we were able to sample from a large swath of contexts in which tomatoes are dry farmed, we are also aware that conditions will vary year to year, especially as climates change and farmers can no longer rely on “typical” weather conditions in the region. While we are confident in the patterns we observed and the recommendations below, racks industries we also encourage further study across multiple years to better understand the full scope of the decision space in which dry farm growers are acting.

Given the scope of our current findings, we outline several management and policy implications for dry farmers and dry farming. Though we aim these implications towards the context of dry farm tomatoes in coastal California, we expect that they are likely to generalize to other dry farm crops grown in other regions with Mediterranean climates. First, given the expense and possibility that it is detrimental to fruit quality, we do not advise AMF inoculation for dry farm tomato growers. Second, we note the importance of nutrients below 60cm and the complexities of subsurface fertility management, and we recommend experimentation with organic amendments and deeply rooted cover crops that may be able to deliver nutrient sources that persist at depth, as well as planning several seasons in advance to build nutrients deeper in the soil profile. Finally, given our finding that dry farm soils develop a fungal signature that increases over time and its association with improved fruit quality, we encourage farmers to experiment with rotations that include only dry farm crops and even consider setting aside a field to be dry farmed in perpetuity. However, fully dry farmed rotations currently do not exist, likely due to a lack of commercially viable options for crops to include in a dry farm rotation. In order to experiment with potential dry farm rotations, as well as cover crops that can best scavenge excess nitrates and soil management regimes that can increase soil fertility at depth, farmers must be given both research support and a safety net for their own on-farm experimentation. Funding to mitigate the inherent risk in farmers’ management explorations will be key in further developing high-functioning dry farm management systems. Expanding land access to farmers who are committed to exploring dry farm management can additionally benefit these explorations.

Dry farm tomato systems on the Central Coast point to key management principles that can both help current growers flourish and provide guidance for how irrigation can be dramatically decreased in a variety of contexts without harming farmer livelihoods. In these systems, managing nutrients at depth–at least below 30cm and ideally below 60cm–is necessary to influence outcomes in fields where surface soils dry down quickly after transplant. Fostering locally-adapted soil microbial communities that are primed for water scarcity can improve fruit quality. Farmers can otherwise manage nutrients to maximize either yields or quality, giving latitude to match local field conditions to desired markets. As water scarcity intensifies in California agriculture and around the globe, dry farm management systems are positioned to play an important role in water conservation. Understanding and implementing dry farm best management practices will not only benefit fields under strict dry farm management, but will provide an increasingly robust and adaptable example for how farms can continue to function and thrive while drastically reducing water inputs.Unlike other forms of dryland farming , in this region dry farm tomatoes are grown over a summer season where there is a near guarantee of no rainfall. Farmers plant tomatoes into moisture from winter rains, counting on soils to hold on to enough water to support the crops over the course of the entire dry summer and fall. While some farmers irrigate 1-3 times in the first month after transplant, severe water restriction is what gives the fruits their intense flavor, and farmers trade water cuts that lower yields for price premiums that consumers are more than willing to pay for higher quality fruits. Beyond Bay Area consumer’s enthusiasm for high-quality local produce, dry farm tomatoes also trace their origins to a richer food culture of justice-oriented and farmer-centric food distribution in the region.

From the Black Panther Party’s Free Breakfast Program to strong community support for worker-owned and consumer food cooperatives , the Bay Area has become a hub of alternative values-based supply chains in a country largely dominated by an industrialized food system . Following this tradition, dry farm tomatoes originally found their footing in the United States in the Central Coast region 30 miles south of the Bay. In the 1970’s and 1980’s, innovative growers in small-scale cooperatives and teaching farms adapted an Italian and Spanish legacy of vegetable dry farming to the region’s Mediterranean climate, maritime influence, and high-clay soils . While these environmental features were necessary to grow tomatoes under dry farm management, the movement that sparked the reemergence of local farmer’s markets in the 1980’s also provided the access to direct to-consumer marketing that small farms needed to win consumer attention and loyalty, allowing farmers to both grow and sell this niche product. With their origins in local food distribution networks and local adaptations to a unique climate, dry farm tomatoes are now a signature of small, diversified, organic farms on the Central Coast and are a feature of many such operations’ business models. To this point, dry farming has largely followed its initial course and is only practiced at a small scale in the region, both in terms of geographic scope, and farm size. Dry farming may therefore be to playing a role in an agroecological transition in the region, buoying small-scale, thought-intensive management styles with access to a steady income source and consumer base. However, with recent droughts and water shortages in California, dry farming has recently begun to take a more prominent role in social and policy visions for the future of the state’s agricultural system. From the Sustainable Groundwater Management Act to emergency orders in drought years, farmers, researchers, policymakers, and the general public have become acutely aware of California’s currently unsustainable agricultural water use and the economic ramifications of water shortages . As an option that holds promise for maintaining farmer livelihoods while dramatically cutting water use, vertical growing systems journalists and policy groups have touted dry farming as an important system to target for significant expansion . Farmers have been considering how to use dry farming to adapt to drier futures for decades, lighting the way for researchers and policymakers’ more recent interest. However, up to this point, farmers’ thoughts and knowledge about dry farming have not been clearly elicited or formally incorporated into conversations about the future of the practice. Grounding conversations about future expansion of the practice in the knowledge of those who are most intimately familiar with its implementation is essential. At this moment of enthusiasm for dry farming, we look to practitioners to better understand the current state of dry farming on the Central Coast and its potential for expansion across California, along with the benefits and harms that expansion may carry.

We interviewed ten dry farmers, representing over half of the commercial dry farm tomato operations on the Central Coast, in order to collaboratively answer two central research questions. First, what business and land stewardship practices characterize successful tomato dry farming on California’s Central Coast? And second, what is the potential for dry farming to expand beyond its current adoption while maintaining its identity as a diversified practice that benefits small-scale operations? The majority of these farmers were part of an ongoing participatory research project in which field data were collected to better understand soil fungal communities and nutrient management in dry farm systems . These interviews were extensions of conversations and relationships fostered with farmers throughout the research process. We synthesized farmer insights into nine key themes that broadly describe how dry farming is currently practiced on the Central Coast, its potential to expand in scope , and the opportunities that farmers see as particularly provident for the practice. We also used the constraints identified by farmers to map areas most likely to be suitable for future dry farming. At this juncture of a high-functioning, low-water management system and urgent political interest in decreasing agricultural water use–in California and across the globe–we conclude by asking how dry farming can be a model for developing systems that decrease water use, and also how dry farming itself may be scaled out to other small-scale, thought-intensive operations without jeopardizing these same farms’ ability to continue profitably growing dry farm produce.Interviews were done with farmers who have commercial operations in California’s northern Central Coast region , as well as one farm with operations in Marin and Sonoma counties. Ranges of coastal mountains govern both climate and land use, trapping cool, moist air, and concentrating farming operations in valleys with fertile, alluvial soils. The Central Coast is known for its agricultural production–particularly berries, lettuce, and artichokes–that thrive in its fertile soils and mild climates that allow for year-round cultivation. Agricultural revenue in the region totals over $8 billion annually , making it a larger agricultural producer than most countries. This intensive production has led to both high land values and environmental degradation–largely in the form of water contamination–that shape both farmer decision-making and policy interventions . Within this landscape, farms often operate at industrial scales, though many small farms persist. Though cropland is consolidated into fewer, large operations , many smaller farms have found niches selling to local markets.After building relationships over the course of a year-long participatory field research process with eight tomato dry farmers , we conducted semistructured interviews with all farmers involved in that study. We interviewed two additional dry farmers who were not involved in the field project–one whose farm is in Sonoma County , and one whose farm could not participate in the field study due to extensive fire damage–for a total of ten farmers representing eight operations. Interviews were done in person , over the phone , and on Zoom in winter and fall 2022. Because there is no official record of tomato dry farmers in the Central Coast region, we used a snowball approach to identify farms that might be candidates for inclusion, asking each interviewee what other dry farm operations they knew of in the area. We can identify two dry farm tomato growers in the region who were not interviewed in this study, and we estimate that our interview subjects represented 50-75% of commercial dry farm tomato operations on California’s Central Coast. Interviews lasted 1-2 hours and focused on dry farm management practices, environmental constraints, support, water/land access, and economics . Interviews were recorded and transcribed, then analyzed through an interactive process of open, axial, and selective coding . Data were grouped into three overarching categories , with key themes in each category. In order to identify areas that might be suitable for future tomato dry farm management, we used farmer-described constraints to make a suitability map using publicly available datasets. We first compiled the environmental constraints on tomato dry farming described in each interview , which fell into three main categories: precipitation, temperature, and soil texture. We limited our analysis to California as the region these farmers are most familiar with to avoid extrapolating constraints beyond the context in which they were given. We used PRISM 30-year climate normals to characterize California’s temperature and precipitation . We used the average constraint named by the farmers; however, because these normals are a 30 year average and will stray significantly from these averages in individual years, particularly in the case of precipitation, we expect that we overestimate the extent of suitable areas.

They show that Kenyan exporters selectively allocated limited deliveries across contracts to maintain credibility in seller’s reliability. In a study of relational contracts in the Rwandan coffee industry, the same authors instrument the placement of mills to show that more potential competition from other mills reduces the use relational contracts with farmers making farmers worse off and reducing the quantity of coffee supplied to mills. Second-best competition is thus not necessarily welfare reducing.Biological simplification has accompanied agricultural intensification across the world, resulting in vast agricultural landscapes dominated by just one or two crop species. The Midwestern US is a prime example, where corn currently dominates at unprecedented spatial and temporal scales. An area the size of Norway is planted in corn in the Midwest in any given year with little variation in crop sequence; over half of Midwestern cropland is dedicated to corn-soy rotations and corn monoculture. Directly and indirectly, this agricultural homogeneity causes environmental degradation that harms ecosystem health while also contributing to climate change and increasing vulnerability to climate shocks. Agricultural diversification in space and time reverses this trend towards homogeneity with practices like crop rotations that vary which harvested crops are grown in a field from year to year. Crop rotations are a traditional agricultural practice with ample evidence that complex rotations— ones that include more species that turn over frequently—benefit farmers, crops, and ecosystems.

As one of the principles underlying agricultural soil management, diverse croprotations promote soil properties that provide multiple ecosystem services including boosting soil microbial diversity, grow trays enhancing soil fertility, improving soil structure and reducing pest pressur. These soil benefits combine to increase crop yields and stabilize them in times of environmental stress. Crop rotations’ environmental and economic benefits typically increase with the complexity of the rotation , while conversely, biophysical aspects like soil structure and microbial populations are degraded as rotations are simplified. Despite its benefits, crop rotational complexity continues its century-long decline in the Midwestern US. Corn-soy rotations increasingly dominate over historical crop sequences that included small grains and perennials, with corn monocultures also on the rise1. This increasing simplification is in part the result of a set of interlocking, long-standing federal policies aimed at maximizing production of a handful of commodity crops that distort farmers’ economic incentives. Regional rotation simplification is clear from analyses of crop frequency, county-level data, and farmer interviews. However, fine-grained patterns that more completely reflect farmers’ rotational choices across the region, and how those choices relate to influences from policy and biophysical factors that play out across agricultural landscapes, remain largely unstudied. This knowledge is essential for understanding how national agricultural policy manifests locally and interacts with biophysical phenomena to erode—or bolster—soil and environmental health, agricultural resilience, and farmers’ livelihoods. Bio-fuel mandates and concerted efforts to craft industrial livestock systems as end-users of these corn production systems make corn lucrative above other commodities, while federal crop insurance programs push farmers to limit the number of crops grown on their farms.

These policies, along with the current corporate food regime, drive pervasive economic incentives to grow corn, and farmers must increasingly choose between growing corn as often as possible to provide a source of government guaranteed income, and maximizing soil benefits and annual yields through diversified rotations. These policies both alter agricultural economics at a national level by boosting corn prices and manifest locally in grain elevators and bio-fuel plants that create pockets of high corn prices with rising demand closer to each facility. Biophysical factors like precipitation and land capability that are highly localized and spatially heterogeneous can catalyze or impede this simplification trend. For example, increasing rotational complexity is one strategy that farmers may employ to manage marginal soils or greater probability of drought, while ideal soil and climate conditions allow for rotation simplification to be profitable, at least in the short run5. As these top-down and bottom-up forces combine, we ask: how do farmers optimize crop rotational diversity in complex social-ecological landscapes, with top-down policy pressures to simplify intertwined with bottom-up biophysical incentives to diversify? . Because biophysical factors and even policy influences vary greatly at the field scale at which management decisions occur, an approach is needed to assess patterns of crop rotation that can capture simplification and diversification at this scale. Though remotely sensed data on crop types can now show fine-scale crop sequences, previous approaches to quantifying rotational complexity have relied on classifying rotations based on how often a certain crop appears in a region over a given time period, aggregating over large areas, or examining short sequences. To date, methods to capture rotational complexity have therefore been unable to address management decisions at the field scale , and/or lose valuable information about the number of crops present in a sequence and the complexity of their order .

At the other end of the spectrum, farmer surveys have impressively detailed the economic and biophysical considerations that go into farmers’ rotation decisions35, yet are limited by the number of farmers they can reach and who chooses to respond. Here, we explore how aspects of farm landscapes influence field-scale patterns of crop rotational complexity across the Midwestern US. We developed the first field-scale dataset of rotational complexity in corn-based rotations, covering 1.5 million fields in eight states across the Midwest and ranking crop sequences based on their capacity to benefit soils. We examined rotations from 2012-2017 to coincide with the introduction of the Renewable Fuel Standard, or “bio-fuel mandate,” which took full effect in 2012. We then correlated fields’ rotational complexity with biophysical and policy outcomes factors, using bootstrapped linear mixed models to account for spatial auto correlation in the data. By identifying spatially explicit predictors of rotational complexity, we illuminate how top-down policy pressures combine with biophysical conditions to create fine-scale simplification patterns that threaten the quality and long-term productivity of the United States’ most fertile soils.We focused our analysis on the eight Midwestern states with the highest corn acreage 2. We considered the six-year period from 2012 to 2017, which coincides with the introduction of the Renewable Fuel Standard in 2012. After deriving a novel field-scale rotational complexity index , we used spatially blocked bootstrapped regression to assess how key landscape factors associated with this indicator. These statistical methods account for overly confident parameter estimates that arise in naive models due to spatial autocorrelation in the data. All analyses were conducted in R47.To test for a relationship between RCI and predictive factors, all variables were centered and RCI was regressed against a set of covariate data in a linear mixed model including US state as a random effect to account for regional differences . We included interactions for which we had a priori hypotheses . The model was estimated using the R package `lme4`64. Two model assumptions are violated in the above model, requiring updated estimates of the parameters’ standard errors. First, because RCI is a derived statistic with an unusual domain, the index is not distributed according to a known distribution family and violates the assumption of normality in the residuals. Second, residuals showed high spatial autocorrelation at multiple scales and with an unknown structure, necessitating a nonparametric approach. Both violations are likely to shrink standard errors of the estimated parameters, leading to overconfident estimates; to illustrate, in the case of spatial autocorrelation, if the explanatory variables are randomly located in relation to crop rotation, spatial autocorrelation in crop rotation would falsely inflate significance. We used nonparametric spatial block bootstrapping to correct for this overconfidence. An algorithm for sparsely distributed spatial data, derived by Lahiri 2018, was implemented in R . Spatial block bootstrapping involves iteratively resampling data in spatial blocks to mimic the generation of autocorrelated data. Choice of block size is nontrivial, cannabis drying rack and choosing the optimal block is an open question, but blocks should be larger than the scale at which autocorrelation operates. Using the R package `gstat` to compute a variogram of the residuals generated by the naive LMM, we determined that range was 400815m. We used this as the dimension of each spatial block . We repeated this bootstrap with a range of possible spatial block sizes and found that this inference on parameters was robust to the choice of block size .Complexity of Corn-based Rotations in the Midwestern US: RCI values calculated for corn-based rotations create the first map, to our knowledge, that quantifies field-scale rotational complexity across the Midwestern US . RCI values from 2012-2017 range from 0-5.2 , and are positively skewed . Corn monoculture accounts for 4.5% of the study area and 3.3% of fields, suggesting that larger fields are more likely to be managed as monocultures .

The mode RCI score corresponds to a corn-soy rotation and dominates the region, covering over half of the study area. Two thirds of the area with this score was a CSCSCS or SCSCSC sequence, while the remaining third corresponds to other rotations that yield the same RCI .RCI scores have statistically clear correlations with land capability, mean rainfall, distance to the nearest bio-fuel plant, and field size, as well as with several interactions between these variables . Standard errors from the spatially blocked bootstrap were much larger than uncorrected naive confidence intervals, reflecting that accounting for spatial non-independence is necessary to estimate uncertainty of parameter estimates. Rotational complexity decreased with NCCPI, a proxy for land capability. We find that land of higher inherent capability is more likely to be used for lower complexity rotations. Fields with ample precipitation during the growing season are more likely to have simplified rotations. Though the relationship between the proximity of the nearest grain elevator and a field’s rotational complexity is not statistically clear , RCI showed a clear increase with distance to the nearest bio-fuel plant. Fields that are closer to bio-fuel plants are therefore more likely to have simplified rotations. Rotational complexity decreased with field size, with larger fields being more likely to have simplified rotations. Two of the interactions included in the model show statistically clear relationships. There is a positive interaction between land capability and field size, with higher quality land associated with decreasing RCI on small fields and slightly increasing RCI on large fields . The interaction between land capability and rainfall variance show a negative effect on RCI, with highly variable rainfall accentuating land capability’s impact on RCI . Interpretations of the relationship that each variable has with rotational complexity are shown in Table 4. Though each change is associated with a small shift in average RCI across the region, these can represent massive shifts in regional land management.As crop rotations continue to simplify in the Midwestern US despite robust evidence demonstrating yield and soil benefits from diversified rotations, our ability to explain and understand these trends will come in part from observing the biophysical and policy influences on farmers’ crop choices at one key scale of management: the field. By developing a novel metric, RCI, that can classify rotational complexity over large areas at the field scale, we open the door to regional analyses that can address the unique landscape conditions that impact farmers’ field-level management choices and their subsequent influence on rotational simplification. We find that as farmers are pushed towards simplification by broad federal policies , physical manifestations of these policies like bio-fuel plants are correlated with intensified simplification pressures. Similarly,we see that the pressure to build soils and boost crop yields through diversified rotations intensifies in fields with lower land capability, while conversely the negative effects of cropping system simplifications are accentuated on the region’s best soils.RCI uses the sequence of cash crops on a given field as a proxy for crop rotation, and sorts these sequences into scores based on the sequence’s complexity and potential for agro-ecosystem health. Because this metric has not been used in previous analyses, we verified RCI’s validity through comparisons to previous estimates of rotational prevalence in the region. For example, two separate surveys of farmers in the Midwestern US showed that between 24% and 46% report growing “diversified rotations” which we consider to be an RCI of greater than 2.24 . In the present study, 34% of fields had an RCI greater than 2.24. This and further comparisons of RCI to previous work show that RCI is capable of capturing previously-noted trends in the region.