Governance at the river basin scale started formally with the application of the Integrated Water Resources Management (IWRM) approach, the first notable example of which is the Tennessee Valley Authority (TVA) in 1933 (GWP, 2002). This was followed by the establishment of the United States Bureau of Reclamation that was instrumental in facilitating the subsidized expansion of irrigated agriculture in the Western part of the US as well as construction of large public works (Snellen and Schrevel, 2004). The resulting pathology of water resource management was a reduction in the natural variation in a system creating a loss of resilience (Holling and Meffe, 1996). The Global Water Partnership (GWP) definition of IWRM included integration across development and management for economic and social welfare in an equitable manner without compromising the sustainability of vital ecosystems (GWP-TEC, 2000; Medema and Jeffrey, 2005).

Criticisms about IWRM rest on contentions that it remains a normative theory with very little empirical evidence of purported benefits (Jeffrey and Gearey, 2004). Furthermore, there is no agreement on fundamental issues like what aspects should be integrated, how, by whom, or even if such integration in a wider sense is possible (Biswas, 2004). Adaptive management has been put forward as an enhancement to the prevailing IWRM. Adaptive management has its roots in ecosystem management and refers to a systemic process of continually improving management policies and practices through learning, with the outcome of increasing the adaptive capacity of the system with proposed institutional design requirements outlined (Pahl-Wostl and Sendzimir, 2005). Furthermore, significant literature and theories have been developed toward adaptive governance (Huntjens et al., 2011, 2012), complexity governance (Pahl-Wostl et al., 2010), and social learning in water systems (Pahl-Wostl, 2009; Johannessen et al., 2019).

Long-term institutional analysis, especially of those for managing natural resource use or allocations have been subject to extensive study resulting in multiple diagnostic frameworks (Ostrom, 1990, 2009, 2011; Anderies et al., 2004). The dominant paradigm of environment as a consumer falls short when the environment itself (both biotic and abiotic contributions) is a fundamental precondition to the existence and vitality of the resource, as in rivers and streams. To address complex and dynamic issues and manage large transboundary social-ecological systems and their resources, collaborative institutional structures that unite governmental, private, and non-governmental organizations, as well as individuals into both formal and informal arrangements is needed (Powell, 2010).

Arguments have been made that institutional scholarship has become overly concerned with explaining institutions and institutional processes, notably at the level of the organizational field, rather than with using them to explain and understand organizations (Greenwood et al., 2014). There is considerable fluidity and confusion on the boundaries between institutions and organizations (North, 1990; Edquist and Johnson, 1997). Organizations are created with purposive intent in consequence of the opportunity set resulting from an existing set of constraints, and in the course of attempts to accomplish their objectives are a major agent of institutional change (North, 1990). Furthermore, organizations are embedded in an institutional context of socio-cultural ideas and beliefs that prescribe appropriate and socially legitimate ways of doing things and are therefore subject to institutional influences in either positive or negative ways (Greenwood et al., 2014). Lastly, organizations are formal structures with an explicit purpose and they are consciously created, whereas, institutions may develop spontaneously and are often not characterized by a specific purpose (Edquist and Johnson, 1997).

In the United States, new organizational systems specifically for river restoration and stewardship have begun proliferating over the past couple of decades (Lant, 2003). The primary drivers have been the necessity of compliance with the Endangered Species Act of 1979 and public mobilization of political support to preserve the ecosystem services the river provides. With the creation of such organizations also comes the question, can these organizations sustain themselves over the long-term? What would keep them going despite changes occurring internally and disturbances from the environment externally?

We examine the effect of both biophysical and institutional disturbances on organizations through a study of the extent of attitudinal diversity in response to organizational agendas. Attitudinal diversity is defined as the varying attitudes around the yearly program agendas, ranging from supportive to critical. The cases include the Upper Colorado River Endangered Fish Recovery Program (UCREFRP), and the Lower Colorado River Multi-Species Conservation Program (LCR-MSCP). For the purposes of this study, the two programs represent a polycentric organizational construct with multiple organizations working toward a common goal. The study answers the following research questions:

Because a disturbance-response framework is utilized in the conceptualization of organizational resilience, a characterization of the types of disturbances hitting the systems being studied is detailed. Disturbances, in an ecological sense, are defined as any relatively discrete event in time that disrupts ecosystem, community or population structure and changes resources, substrate availability, or the physical environment (Pickett and White, 1985). There has been a growing emphasis in social-ecological systems science that complex dynamics often lead to unexpected outcomes with long-term effects. The disturbances that occur in social-ecological systems can be sudden events that are large in magnitude and/or infrequent, termed as “pulse” dynamics or extensive, pervasive and subtle change termed as “press” dynamics (Collins et al., 2011). Furthermore, institutional disturbances as utilized in this study can be those rooted in broad socio-cultural norms and values, legislative and regulatory measures or market factors, while the biophysical disturbances refer to cases of episodic “hydrologic” drought, climate change and continuing chronic “megadrought” as well as the presence of invasive species (Young, 2010).

Organizational resilience, in essence, is defined as the maintenance of positive adjustment under challenging conditions (i.e., shocks, disruptions, stresses, and strains) such that the organization emerges from those conditions strengthened and more resourceful (Vogus and Sutcliffe, 2007). In other words, the process of “resiling” in the face of ongoing strain and discrete jolts is due to the presence of latent resources that can be activated, combined and recombined in new situations as challenges arise. The implication is that resilience relies upon past learning and fosters future learning, but also exists independently of learning and is embedded within a broader store of constitutive capabilities or endowments (Vogus and Sutcliffe, 2007). These characteristics correspond with the three dimensions of organizational response resilience not only to current issues (concurrent action) or the past (reactive action), but also to the future (anticipatory action) (Duchek, 2014). Furthermore, past learning speaks to a reliance on organizational memory that is posited to consist of mental and structural artifacts that have consequential effects on performance (Walsh and Ungson, 1991).

Anticipatory capabilities refer to the ability of organizations to detect critical developments and to adapt proactively to future changes before they happen. This involves activities such as looking for signals or environmental scanning, which is a process of acquiring information in preparation for inevitable surprises (Duchek, 2014). Both anticipatory learning and organizational memory are rooted in actors' various capabilities, knowledge skills, processes, and routines that facilitate access to resources (Duchek, 2014).

Capabilities for durability or coping capabilities, which is another feature of organizational resilience, refers to the endowments actors possess prior to adversity that shape their capacity for positive adjustment. Endowments facilitate resilience by enabling adaptability and may include financial capability, cognitive capability, behavioral capability, emotion-regulation capability, and relational capability endowments (Williams et al., 2017). This speaks to resilience at the organizational level referring to an organization's ability (embodied in the existence of resources, ideologies, routines and structures) to absorb a discrete environmental jolt and restore prior order (Williams et al., 2017). The authors also speak of an extension of this concept in the systems tradition to include dynamic processes rooted in relational patterns between actors that may affect a reorganization following a disturbance, while still retaining essential structures, functions and identity.

Adaptive capability is strongly linked to strategic actions that aim toward a reconfiguration of organizational resources, competencies and routines to meet demands and opportunities in a changing environment (Kaehler et al., 2014). Adaptive capability can be associated with multiple organizational identities, which in public organizations can take the form of networks, projects, hierarchies and so on. As time goes by, each unit learns to be better at what they are doing and grow into stronger identities (Aagaard, 2012). Adaptive capability can be studied through the processes of integration (involving cultural consensus and clarity), differentiation (involving subcultures and islands of clarity), and fragmentation (involving rituals loaded with ambiguity; Aagaard, 2012).

Laughlin (1991) argues that organizations, for possible psycho-social reasons are naturally change resistant, with a strong tendency to “inertia” and will only change when forced, “kicked” or disturbed into doing something. However, once disturbed, he argues that the track which the disturbance takes through the organization and the degree of transformation it will generate in the pathway it follows will differ over time and across different organizations based on certain organizational characteristics. These organizational characteristics comprise three distinct elements: sub-systems (tangible elements such as buildings, people, machine, etc.), design archetypes (intangible structures, accounting processes, and systems) and interpretive schemes (core values, norms, culture, beliefs, rules, missions statements, etc.), with the less tangible elements being more central to organizational functioning (Laughlin, 1991).

Laughlin (1991) described four pathways that characterize organizational responses to environmental disturbances:

Intrinsic to these models is the distinction between the first and second-order change precipitated by environmental disturbances that has parallels to single and double-loop learning (Argyris and Schon, 1978). First-order rebuttal and reorientation pathways are responses to morphostatic change (making things look different while remaining the same) and are therefore termed “transitions,” whereas second-order changes including colonization and evolution are morphogenetic (penetrating deeply into the genetic code) and represent “transformations.”

What kind of organizational response is invoked and what pathway is chosen is dependent on the extent of the environmental disturbance, which can range from low to high. Figure 1 shows the organizational response patterns in relation to environmental disturbances (Tucker, 2013).

The seven basin states came together in 1922 in order to resolve the equitable division and apportionment of the use of the waters of the Colorado River system. The result of this was the apportionment of 7.5 million acre-feet per year equally to the Upper and Lower Basin states, respectively, for exclusive beneficial consumptive use, with Mexico getting the surplus over and above the aggregate of the quantities specified. The requirement of the Colorado River Compact was that Upper basin states not cause the river flow at Lee's Ferry to be depleted below 75 million acre-feet for any period of 10 consecutive years nor withhold water to Lower basin states.

Later amendments to the agreement included the 1928 Boulder Canyon Project Act authorizing the construction of the Hoover Dam, the 1944 Mexican Water Treaty which allocated 1.5 million acre-feet to Mexico, and the 1948 Upper Basin Compact which solidified the water allocation amounts between the Upper Basin states. Additionally, the Colorado River Storage Project of 1956 authorized the construction of Glen Canyon, Flaming Gorge, Navajo and Curecanti dams for river regulation and power production as well as for irrigation and other uses. The Colorado River Basin Project Act of 1968 authorized the construction of a number of water development projects in both the upper and lower basins, including the Central Arizona Project (CAP). It also made CAP water supply subordinate to California in times of water shortage. Finally, Minute 242 of the US-Mexico International Boundary and Water Commission of 1973 required the United States to take actions to reduce the salinity of water being delivered to Mexico at Morelos Dam. This resulted in the Colorado River Basin Salinity Control Act of 1974 which authorized desalting and salinity control projects including the Yuma Desalting Plant, to improve Colorado River quality (USDOI, 2008).

The main problem arising out of these was that these amounts were overallocations based on measurements taken at Lee's Ferry during greater than average wet years. Among the Upper Basin states, Colorado has senior rights with 3.9 million acre-feet allocated, Utah –1.7 million acre-feet, Wyoming −1.0 million acre-feet, and New Mexico −0.85 million acre-feet. The Lower Basin states includes California as a senior rights state with 4.4 million acre-feet, followed by Arizona at 2.85 million acre-feet and finally, Nevada with 0.30 million acre-feet (CRS, 2020). Another problem is that since implementation of the 1974 Salinity Control Act, measures have been put in place to reduce the annual salt load of the river by more than 1.3 million tons. The salinity concentrations at Hoover, Parker and Imperial dams has been reduced by more than 100 mg/l. However, even with these efforts, the quantified damages to U.S. users are still ~$454 million per year, with projected damages to increase to $574 million per year by 2035 if the Program does not continue to be aggressively implemented (Keeler, 2017).

Diversity analysis in organizational and management literature has been heavily reliant on horizontal surveys and questionnaires that elicit people's attitudes and reactions to certain issues (Miller et al., 1998; Simons et al., 1999; Kilduff et al., 2000; Ricardo, 2000). Some studies rely on focus group discussions and interviews with organizational personnel (Mor Barak et al., 1998; Gilbert and Ivancevich, 2000; McIntosh and Morse, 2015). Fiol (1994) used codes to measure consensus building around the content and framing of meanings to determine how diverse interpretations fostered innovation and collective learning. We use a similar method to analyze attitudinal diversity based on content analysis.

The choice of the two cases was the result of a combination of various factors including the geographic proximity of the systems being studied, the availability of appropriate program documentation for coding, the ease of initiating stakeholder interviews to supplement archival analysis and a specific focus on institutions for river restoration. A few other case studies were considered but did not meet one or more of these criteria and, therefore, were excluded from the analysis.

For the purposes of analysis, we utilized archival material from the program websites, specifically the meeting minutes for every year of the existence and functioning of both the UCREFR and LCR-MSC programs. The number of meetings over the years has varied somewhate with the UCREFRP having an average of two meetings per year and the LCR-MSCP had an average of three meetings per year. The LCR-MSCP generally conduct deliberations around a water-year from October to September. For the purposes of this study, we have grouped deliberations into a calendar year, instead to ensure consistency in cross-case comparisons. The meetins are conducted each year from program inception The meeting is generally called by the Implementation or Steering Committee Chair who is elected by common vote every few years, from one of the participating agencies. The length of the meeting minutes are variable, with some being short and some long duration, depending on issues on the agenda. Because of the dispersal of the meetings, a significant number of them are carried out remotely, with some being carried out at specific chosen locations depending on upper or lower basin region.

We coded the meeting minutes of the programs using defined codes for assessing the diversity of attitudes present in each meeting, ranging from positive to negative, including neutral expressions. The UCREFRP has a much richer and more extensive dataset than the LCR-MSCP, which is quite sparse. The datasets used are available at https://osf.io/j36ry/. Assessment of attitudinal diversity is seen as a variance of views/opinions/articulations around each meeting agenda and the goal and task interdependence for that period. The meeting agenda is taken as a baseline around which attitudinal variation is measured. Attitudes are divided into four categories for a more fine-grained analysis—Supportive, Clarifying, Conditional, and Critical.

A “Supportive” code was assigned to text entries where all statements in a discussion on the agenda that are positive or supportive of the agenda based on linguistic qualifiers that include but are not limited to “should be supported/accepted,” “should not be rejected,” “suggested/asked that,” “proposed that,” “….if deemed necessary,” “agreed that” and so on.

A “Clarifying” code was assigned to text entries where all statements in a discussion on the agenda are seeking further information or clarification on the topic and/or questioning issues further but have not yet expressed a marked positive reaction. Linguistic qualifiers include but are not limited to “asked if/why…,” “expected that…,” “responded that….,” “noted/stated that….,” “clarified that” and so on. This code is viewed as neutral because it illustrates the general question-answer and clarification that is sought in any discussion of the agenda topics.

A “Conditional” code was assigned to text entries where all statements in a discussion on the agenda that emphasize the fulfillment of certain conditions before the theme of the agenda is accepted. Since the tone of the code is more of a conditional acceptance with negative connotations, it is taken as a negative-leaning attribute. Linguistic qualifiers include but are not limited to “before a decision is made…,” “what is gained/intended by…?,” “if…can be changed/modified/developed/excluded…,” “support, but….,” and so on.

A “Critical” code was assigned to text entries where all statements in a discussion on the agenda that are critical to/of the agenda. This code is the most negative-leaning in tone. Linguistic qualifiers include but are not limited to “not yet convinced….,” “…could make things more difficult/challenging,” “concerned about/expressed concern about…,” “reconsider, pending…,” and so on.

Intercoder reliability rating was conducted on 30% of the documents that were analyzed using the guidelines for computing inter-reliability rating (Hallgren, 2012). A total of 35 segments, taken from random document sections, were analyzed by the second coder independently. The total code numbers for each attribute for each segment were entered into a spreadsheet and intraclass correlation (ICC) calculated for the four main attitudinal diversity attributes before the codes were reconciled through discussion. ICC was chosen because the purpose was to assess the consistency and reproducibility of the measurements made by independent coders as some codes were clear-cut and some more ambiguous and contextual. The ICC was run on SPSS using a two-way mixed effects model with an absolute agreement definition. Table 1 below lists the descriptive statistics for ICC for each of the four main attitudinal variable categories.

Finally, a sample of codes are displayed for each category of codes to show the best examples of coding that fall under that category, in the table in Appendix A. In other words, a small sample of those codes that clearly fall into one category based on the operational definitions of that category are listed to provide a clearer depiction of how the codes were used. Definite examples of statements are those that clearly depict linguistic qualifiers for a particular category. Ambiguous examples are coded based on the context of the discussion about the agenda. The latter category is why, for example, a linguistic qualifier such as “countered” would be taken as supportive because the countering is in support of the agenda rather than critical of it or a linguistic qualifier like “said that” would fall into a condition category rather than a clarifying one because it is conditional on a particular action.

Figures 2, 3 shows the temporal patterns of distribution of attitudinal diversity for the UCREFRP and LCR-MSCP, respectively. Both programs have their inception in a desire to avoid jeopardy or financial penalties associated with the impacts of continued water development activities on critically endangered fish species as well as on incidental take from recreational fishing activities. Attitudinal diversity variation tends to lessen over time in the UCREFRP case. We observe that critical and conditional attitudes were more prevalent over the first 10-year period starting from 1988 and lessened after that. The 1994–1997 period was not showing success in terms of native fish protection and there was concern about further investment until the situation showed improvement. Overall, there is growing convergence toward agreement based on the higher supportive and clarifying attitudes around the agenda issues and a diminution of critical attitudes over time.

It is evident that the period from the inception of the UCREFRP to 2000 showed a greater attitudinal diversity than the period post-2000 to present. The years 2008 and 2018 show absence of good data for analysis and are discounted. The period 1988–1998 was the period when the most important issues such as institutional membership, native fish recovery, non-native fish management and capital investment projects were closely discussed with variable success over the period. The 1994–1997 period depicts a period of setbacks in program goal achievements and resulted in increased efforts to incorporate science-based approaches to flow management to increase propagation and survival of listed native fish species. The period from 1988 to 2000 therefore signifies a phase when consensus was less easily achieved due to implementation challenges. However, post-2000 there appears to be greater positive responses with the agreement being higher.

The LCR-MSCP does not have sufficient and consistent data over time. But what can be observed from Figure 3 is that critical attitudes have persisted or occurred at discrete intervals over time. This is supported by evidence that details that a building dissension around key program goals resulting in legal settlement of a water right dispute centered around acquiring water leases for fish conservation (Arizona Department of Water Resources vs. Mohave County, 2015). During the period 2009–2010, the acquisition of land for habitat construction was being considered, toward housing listed endangered species including the Southwestern Willow Flycatcher. One of the potential land acquisitions considered was Planet Ranch, which is located ~20 miles east of Parker, Arizona upstream of the Bill Williams River National Wildlife Refuge. It was a site owned by Freeport McMoran Mineral Corporation. The land is 3,418 acres with 5,549 acre-feet of water rights attached to it. Initial discussions merely kept the proposal under consideration.

When the acquisition became more viable, Mohave County Water Authority and the City of Bullhead registered opposition to the acquisition on grounds that there was lack of involvement of the local government in the final agreement, particularly in relation to settlement provisions and water rights transfers. The LCR-MSCP has arbitration measures where the Steering Committee calls for a deciding majority vote on an issue if agreement cannot be reached through discussion. This vote found overwhelmingly in favor of acquisition of Planet Ranch for the program. However, the matter was further taken to the courts with the final decision being made in favor of the Program. In December 2015, LCR-MSCP acquired a lease for Planet Ranch to be used as a conservation area. There was a gradual build-up of concern and then opposition until the 2013–2014 period when the acquisition was brought to the table for implementation.

Table 2 below illustrates the various environmental disturbances that have affected the two organizations over time, how the organizations responded; how attitudinal diversity mediates the leveraging of organizational resilience capabilities to these disturbances is discussed in the following section. Environmental disturbances are characterized as either press or pulse events and are either institutional or biophysical in nature. Press disturbances are long-term disturbances with long-term impacts, while pulse disturbances are temporary in nature and while they may cause substantial impacts, there is potential for the system to rebound or recover. Institutional disturbances can take the form of broad socio-cultural norms and values, legislation and regulatory policies and market factors, while biophysical disturbances include drought (both episodic and chronic), climate change and the presence of invasive species.

This section is organized into three parts. The first part concisely discusses the links between attitudinal diversity and organizational resilience capabilities and how this process has been operationalized in the study. The second part describes the role of attitudinal diversity in organizational responses to institutional disturbances and how resilience is displayed, while the third part discusses the same in the context of organizational responses to biophysical disturbances.

A discussion of the results reveals two main trends in organizational responses to disturbances and the resilience capabilities that are leveraged. The resilience capabilities in response to institutional disturbances reveals the exercise of predominantly anticipatory capabilities, while the main resilience capability leveraged in response to biophysical disturbances is coping capability. The proactive nature of the organizational responses to institutional drivers in the form of regulations is consistent with prevailing literature detailing the presence of sanctions as a driver in self-organization around common pool resource governance (Ostrom, 1990).

The disturbance categories covered in this study are not intended to be an exhaustive list of all possible institutional and/or biophysical disturbances. They represent the most significant ones that we judged to be relevant to the study. As mentioned earlier, other institutional disturbances could take the form of socio-cultural norms or market factors. One of the prevailing socio-cultural norms in the Upper Basin, is the illegal stocking of invasive fish species in certain sections for the purposes of recreational fishing, a continuing legacy of past authorized fish stocking (Johnson et al., 2009). The UCREFRP undertakes invasive fish species containment through educating the public, putting out a blanket kill policy on invasive fish species and also using chemical containment measures in key sections to reduce invasive species presence and facilitate the propagation of endangered fish species. This creates significant ambiguities for our assessments and it was not included. Other disturbances internal to the organization, such as turnover, were similarly not included. The list of disturbances in this study is a first attempt to categorize the impacts of external environmental disturbances on organizations for river restoration. We hope that in future research, the list could be expanded to include more complex and cascading disturbances and their impacts.

The predominant prevalence of coping capabilities as resilience capabilities in response to biophysical disturbances also illustrates the fact that while the principle behind the formation and operation of the programs remains adaptive management, it has not been sufficiently sensitive to changing biophysical conditions. Organizational responses to these still hinge on short-term, stop-gap solutions that mitigate immediate adverse effects to endangered species recovery.

Also significant is the fact that the UCR-EFRP displays adaptive capabilities in response to both institutional and biophysical pulse disturbance events, but not to either of the press disturbance events. This supports the previous contention that the adaptive capabilities are leveraged more in response to short-term disturbances, with very little planned adaptive capacity being built to deal with more complex and long-lasting press disturbances. The LCR-MSCP covers an intensively managed section of the river and apart from anticipatory capabilities leveraged to comply with the ESA, the unique and aggressive water politics of the Lower Basin states (Morrison et al., 1996; Huckleberry and Potts, 2019) as well as the path dependence engendered by extensive dams and diversions and rigid water rights and allocation regimes, there are significant constraints for the program to incorporate adaptive resilience capabilities to changing biophysical press conditions.

A point to highlight here is the fact that the institutional press disturbance in the form of ESA follows a pre-set process for listing that lags far behind the more urgent biophysical press disturbances of megadrought, climate change and continued invasive species management. This time lag can be costly and detrimental to future endangered species recovery efforts. While anticipatory capabilities can help to deal with institutional disturbances, these capabilities need to be leveraged to take into account biophysical disturbances that are already making themselves felt. This is in addition to building adaptive capacity to deal with longer-term press disturbances.

As river restoration in arid regions takes on new and more urgent meaning, and in light of the recent federal shortage declaration on the Colorado River, it is imperative that organizations formed to facilitate the goal of river restoration and healthy, living rivers are constituted in a long-term sustainable manner and are resilient to various disturbances. This study used an in-depth content analysis method to tie the presence of certain intra-organizational diversity attributes to organizational resilience. A key implication the study identified was the need for increased cognitive diversity in decision-making through the incorporation of local and indigenous ecological knowledge that better engages with the conditions of the river systems on the ground. This study contributes to a deeper understanding of organizational resilience, points to vulnerabilities that can be addressed, and highlights the necessity for stronger sensitivity to biophysical conditions and incorporating feedback from them to ensure long-term organizational sustainability for river restoration.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

This manuscript was conceptualized and written by JS under the supervision and guidance of MS and DW. TL's contribution was in implementing the methods and undertaking intecoder reliability assessment, as well as in providing manuscript feedback. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.