Partnerships between organizations that manage protected land in California are associated with groups with environmentally oriented missions

Partnerships between organizations that engage in land protection are promoted as a way to improve the efficiency of limited conservation budgets. However, limited empirical exploration of the types of organizations involved in partnering and their organizational objectives precludes a holistic understanding of how to integrate partnering into planning for improved conservation outcomes. Using data on protected areas from California, United States, we explored the frequency and extent of partnering between managing organizations. In addition, we analyzed mission statements of partnering and non-partnering organizations to explore whether organizational objectives were related to observed partnering behavior. We estimated that partnerships managed about 7 million acres, comprising 8% of total protected land area, in California. Furthermore, the organizations that we observed partnering tended to use more environmental themed language in their mission statements, while non-partnering organizations tended to use language with socioeconomic themes


| INTRODUCTION
Expanding terrestrial protected area is a global policy priority, with increasing interest in protected areas that serve social as well as ecological purposes (e.g., Watson et al., 2014). For instance, the recent "Conserving and Restoring America the Beautiful Report" outlines a plan to conserve 30% of land in the United States (US) by 2030 while also promoting the importance of a collaborative and inclusive approach to considering the desired outcomes of protecting land for conservation (DOI, 2021). Such priorities reflect a mounting understanding of how achieving the desired public benefits of land protection activities relies on planning and policy that accounts for the social and institutional context in which the land protection actions are occurring (Guerrero & Wilson, 2017).
An important social characteristic related to land protection is the identity and diversity of organizations involved in the activity. Specifically, many organizations are not explicitly conservation focused. Organizations that manage protected lands in the United States include a wide range of entities such as federal, state, and local governments, charitable organizations that work on conservation, wildlife, or community issues, and private organizations such as cemeteries and homeowner associations (e.g., Jackson et al., 2021;Wilkins et al., 2021).
These organizations often manage protected lands on their own, but they may also partner to co-manage lands as a way to improve outcomes related to their organizational objectives. Although the organizations may have varying objectives, and their particular incentives for managing land are likely to be equally heterogeneous (e.g., Baldwin & Leonard, 2015;Kroetz et al., 2014), potential complementarities between management goals could lead to mutually beneficial partnerships (Patney, 2000). For example, partnering may occur when one organization owns land with conservation potential (such as the US Department of Defense) and another organization with conservation expertise (such as a conservation NGO) agrees to manage it (Patney, 2000). Additional scenarios include when conservation easements on a property are held or managed by an organization other than the owner (Merenlender et al., 2004), or if different organizations share ownership or management tasks for a site (Radcliffe, 2020). Such co-management partnerships or other forms of collaboration are thought to be instrumental in both establishing (Kark et al., 2015) and managing (Moore et al., 2011) protected areas.
Despite the growing evidence of heterogeneity of stakeholders involved in protected lands and the prevalence of partnering in land protection efforts, partnering is still rarely considered in formal conservation planning theory (Bode et al., 2011;Iacona et al., 2016). There has been little systematic analysis of the frequency of partnerships in land management (e.g., Wilkins et al. (2021)). Furthermore, other than several case studies focused on conservation organizations (e.g., Gerber & Rissman, 2012;Wilkins et al., 2021), there has been little development of empirical evidence regarding how partnering corresponds to organizational type or objectives, although it has been hypothesized that organizational objectives may be important for partnering for conservation (Iacona et al., 2016).
Thus, in this study we conducted a systematic assessment of the prevalence of one type of land management partnership in the state of California, US and compare organizations that do and do not partner to provide insights into the types of organizations that partner and their objectives. First, to gauge the prevalence of protected area co-management, we quantified the extent to which documented partnering is occurring in land management using the Protected Area Database (PAD) for California. Second, we examined the types of organizations that engage in these partnerships. Third, to explore whether organizational objectives relate to co-management partnerships, we collected and analyzed the text of mission statements from the subset of land management organizations with available mission statements and compared mission statements of those who partnered and those who did not. Our study extends prior work by (1) exploring partnering in land management beyond explicitly conservationfocused entities and (2) developing insights into the choice to partner through comparing partnering and nonpartnering organization types and missions.
We draw from mission statements to develop insights into the partnering decision because they aim to embody the main values and goals of organizations (Garnett et al., 2016) and guide the organization's operations and programs through setting geographic, financial, and target audience boundaries (Campagna & Fernandez, 2007;Kirk & Beth Nolan, 2010). Prior research on environmentally oriented NGOs has used analysis of mission statements to highlight the diversity of organization priorities (Partelow et al., 2020). Prior work also indicates that mission statements may help to attract partner organizations with overlapping priorities (Dayer et al., 2016) because they provide meaningful data on the priorities and interactions of the organizations (Partelow et al., 2020).

| Identification of protected areas and managing organizations
We used California as an empirical context to explore our research questions because it has high levels of biodiversity and over 23% of the state is protected lands (Wilson et al., 2014). We identified 21,369 protected areas in California and the 1144 organizations that manage them using the Protected Areas Database of the United States (PAD-US Version 2.0). In PAD-US 2.0, protected areas are defined as areas "dedicated to the preservation of biological diversity and to other natural (including extraction), recreation and cultural uses, managed for these purposes through legal or other effective means" (U.S. Geographical Survey, 2018).
For each protected area, PAD-US 2.0 lists the name and type of the managing organization(s). Managing organizations are not necessarily conservation-focused. Instead, a large variety of organizations are listed as managers of protected areas, including federal, state, and local governments, charitable organizations such as The Nature Conservancy, and private organizations such as cemeteries, homeowner's associations, and businesses. Broad organization types included in PAD-US for California are federal (FED), state (STAT), local (LOC), district (DIST), non-governmental organizations (NGO), private (PVT), and unknown (UNK).
While the PAD-US 2.0 database is extensive, there are some gaps in coverage. It does not include tribal land or land managed by the Bureau of Indian Affairs (BIA). Neither does PAD-US 2.0 include temporary forms of protection such as that provided by rental contracts. For example, the United States government pays food producers to retire land from production and to maintain perennial cover under 10-15 year rental contracts through its Conservation Reserve Program (Claassen et al., 2008). In addition, the coverage of protected areas-and by extension the organizations involved in protecting these lands is incomplete. This is particularly true with regard to the data on easements and local and NGO managed lands.

| Identification of partnering
We identified "partnering" in management of protected area in California documented in the PAD-US database by identifying all parcels in our dataset that had multiple manager names recorded and that were larger than 100 acres. We expect these records to represent actual instances of co-management because PAD-US 2.0 had been extensively updated to remove spurious overlapping boundaries between adjacent protected areas. In addition, we performed further processing to remove duplicate records of the same protected area and resolve inconsistencies in owner/manager information. This approach yields rich data on land management partnerships, but does not include other types of partnering, such as donor-NGO partnerships, a point we return to in Section 4.
With this approach we aim to capture a variety of potential situations that could result in multiple managers being associated with the same parcel. For instance, protected areas that have been owned or managed by different agencies either over time (could be because one partner sells or transfers the parcel to another partner) or at the same time (could indicate co-ownership or management or one owner and a different manager). We focused on managers rather than owners because the manager field is better represented in the dataset (Johnson, L., & Croft, M., personal communication, May 19, 2019). In addition, within PAD 2.0, the "manager type" field for easements represents the holder of the easement (Johnson, L., & Croft, M., personal communication, May 19, 2019). Easements are one of the primary instances where we expect owner and manager of a protected area to differ, so we needed to differentiate these properties in our analysis.
This process of parcel manager identification allowed us to estimate the extent of partnering within protected area management in California and also allowed us to construct two groups of organizations for our analysis: organizations that we observe engaging in at least one comanagement partnership ("partnering organizations") and those that we never observe partnering ("nonpartnering organizations"). To visually summarize the spatial extent of land managed by the group of partnering organizations, we constructed a map to show the geographic location of parcels meeting our definition of being managed in partnership and those that did not meet our definition. We also used network analysis to identify the frequency and extent of relationships between organizations of different types that were observed partnering within our dataset. Specifically, we plotted network diagrams where manager organization types were the network nodes, and the partnerships between organizations of the two types were the edgeswith edge weights representing the number of times the linked organizations partnered on different protected areas.

| Collection and analysis of mission statements
For each organization listed in PAD-US 2.0 that managed more than 100 acres of land within California, we searched for a mission statement on their organization's website. In total, we collected mission statements for 596 land management organizations including 114 partnering, and 482 non-partnering organizations. This comprised 78.62% of the partnering organizations and 48.23% of non-partnering organizations that we observed in California. Many of the organizations without mission statements were cities, cemeteries, and homeowner's association (HOAs), which are not likely to have a formal mission statement like an NGO or government agency. We provide further details about the organization types for which we were not able to obtain a mission statement in Appendix S1. The remainder of Section 2 and Section 3.3 pertains only to this subset of organizations that have mission statements.
To process the mission statement text, we first cleaned the mission statements to prepare them for text mining and to avoid retrieving irrelevant characters when performing our analyses. Using the "tm" package in R, we transformed everything to lowercase and removed non-alphanumeric words, English stopwords (e.g., "a" and "the"), punctuation, numbers, and whitespace. In addition, we stemmed the words to standardize tense and transform words to their root (Feinerer & Hornik, 2020). The stemming process groups similar words into a single base form (Welbers et al., 2017). For example, the word "community" conveys the same concept as "communities," so stemming groups them together as "communiti." This allowed us to focus on the content rather than the specific wording within the text. All data processing was performed in the R programming language using the packages tidyverse and dplyR (R Core Team, 2020; Wickham et al., 2019).
We then used several qualitative and quantitative methods to analyze the text of mission statements, enabling us to examine whether partnering and nonpartnering organizations differ in organizational objectives.

| Frequency of phrases
We used a key-word-in-context (KWIC) approach to identify key phrases in mission statements. KWIC identifies the string of words surrounding a certain word (Bernard & Ryan, 2010), which reveals a richer semantic meaning than single words (Weber, 1990). We pulled key phrases (unique 2-word phrases) from the cleaned mission statements using the tokenizer function in R . Once we had the list of phrases from all mission statements, we counted the number of times a phrase was used by partnering and nonpartnering organizations. These frequency lists were visualized as word clouds using the wordcloud package in R (Lang & Chein, 2018). To create the word clouds, we only displayed phrases that were used more than once and limited the output to 50 phrases displayed in the word cloud. Word clouds with a limit of 200 phrases are included in Figure S1.

| Themes
We grouped the phrases into thematic categories, following the approach of Bernard and Ryan (2010). In reviewing the frequency lists and word clouds, we inductively observed that different phrases corresponded to different aspects of sustainability (e.g., Mensah, 2019). We subsequently constructed two themes-environmental and socioeconomic-to deductively test whether these aspects were correlated with partnering behavior. The environmental theme included phrases relating to environment or conservation, and the socioeconomic theme included phrases related to social, economic, or governance issues (see Tables S7-S9 for a list of phrases and their classification). Although we were interested in all three pillars of sustainability, we were unable to differentiate between strictly economic and social phrases and thus combined them into the one socioeconomic theme. Next, we identified all phrases that occurred in mission statements six or more times (n = 168) and categorized them into themes based on the context and semantics of the phrase. We excluded from our analysis phrases that did not fit either theme or could apply to either theme.

| Statistical tests
We conducted three statistical tests to examine differences in mission statements between partnering and non-partnering organizations. First, we used a chisquared test to test for differences in the relative frequency of specific phrases in mission statements across partnering and non-partnering organizations. We conducted this test using all 8777 phrases identified.
Second, we examined thematic differences between partnering and non-partnering groups. We started by identifying the 20 most-used phrases within each theme. The choice to focus on a subset of the most frequently used phrases avoids results being influenced by infrequently used phrases (Weber, 1990). The socioeconomic theme had three phrases that tied for inclusion in the top 20; the results are qualitatively similar for all three phrases, but we present results for the first phrase based on alphabetical order in the main text and results for the other phrases in Table S6 and Figures S2 and S3. Then we calculated the mean rate of usage of environmental and socioeconomic phrases within the mission statements of organizations observed partnering versus those not observed partnering in California. Lastly, we used two-sample t-tests to test whether partnering and nonpartnering organizations were statistically similar in their usage of environmental and socioeconomic phrases.
Third, since the groups of partnering and nonpartnering organizations had different mixes of organizational types, we wanted to ensure that our results did not merely reflect this underlying heterogeneity. Therefore, we repeated our two-sample t-tests of thematic differences between observed partnering and non-partnering organizations for each of the four primary types of organizations in our dataset (state, NGO, local, and district). The federal, private, and unknown groups did not contain sufficient observations for statistical testing.

| Frequency of partnering by land management organizations in California
We found evidence of partnering occurring between land management organizations in California (Table S1). Of the 1144 organizations that managed protected areas in the state, we observed 145 engaged in management partnerships. These partnerships managed over 7 million acres, or 7.54% of the total protected area in the state (Figure 1).

| Identity of land management organizations that partner in California
Protected lands in California were managed by multiple types of organizations. Although they made up a small percentage of the total number of organizations involved in protected land management in California, federal organizations were involved in managing the vast majority of land area, about 92% of protected land (Table S1). Local, district, NGO, and private organizations comprised the bulk of the organizations involved in land management, but were responsible for only about 4% of land area (Table S1). Finally, state organizations made up about 2% of the organizations and managed about 4% of the area (Table S1).
There was heterogeneity in observed partnering patterns based on organization type. About 78% of the relatively few federal organizations partnered, and about 58% of state organizations partnered (Table S1). NGOs were also active in partnering, with about 35% engaged in at least one partnership. Furthermore, these partnerships occurred between many different types of organizations ( Figure 2). The most frequently occurring type of partnership in terms of number of parcels managed was between federal and state organizations, followed by federal and NGO. In terms of area covered, federal and state organizational partnerships covered the most area.

| Comparison of mission statements
For the subset of organizations for which we were able to obtain missions statements, we observed differences in the language between those that partnered and those that did not. First, the word clouds revealed differences in mission statement language between the mission statements of the observed partnering and non-partnering organizations. Additionally, the frequency lists revealed substantial differences in key phrases, especially for the more commonly used phrases (Table 1; Figure 3). For partnering organizations, "open space" and "futur generat" were the most frequently occurring phrases ( Figure 3; Table S2). In contrast, "qualiti life" was the dominant phrase for non-partnering organizations ( Figure 3; Table S2). Our chi-squared test confirmed that F I G U R E 1 Map of protected areas within the state of California with area managed by partnerships highlighted. the two types of organizations used phrases in their mission statements with different relative frequencies (p-value = .00; Table S5).
Thematic analysis provided further evidence of a difference in organizational objectives. Partnering and nonpartnering organizations used environmental and socioeconomic themes with different average frequencies (both p-values .00). Partnering organizations used environmental phrases 160% more frequently, on average, than non-partnering organizations, with partnering organizations using environmental phrases an average of 0.86 times and non-partnering organizations an average of 0.33 times in their mission statements (Figure 4). In contrast, non-partnering organizations used socioeconomic phrases 75% more, on average, than partnering organizations. Non-partnering organizations used socioeconomic phrases an average of 1.07 times and partnering organizations used socioeconomic phrases an average of 0.61 times in their mission statements (Figure 4). These results were qualitatively robust to including either of the two socioeconomic phrases that tied for inclusion in the top 20 (Table S6).
Lastly, when examining thematic differences between partnering and non-partnering organizations for each type of organization, we found no evidence that heterogeneity in organizational type was driving our results. On average, partnering organizations of each type used more environmental phrases (differences in mean use of 0.70, 0.21, 0.10, and 0.54 for State, NGO, Local, and District, respectively), while non-partnering organizations of the NGO, Local, and District types used more socioeconomic phrases (differences in mean use of 0.02, 0.07, and 0.31 for NGO, Local, and District, respectively). These results were qualitatively robust to including either of the two socioeconomic phrases that tied for inclusion in the top 20 ( Figures S2 and S3). We note, however, that while these differences have the sign that we would expect from the aggregate results, they are not statistically significant. Moreover, partnering state organizations used slightly more socioeconomic phrases on average than nonpartnering state organizations (0.04), though again the difference was not statistically significant. It is possible that more observations, potentially at a country-wide scale, would provide more statistical power when disaggregating this analysis by organizational-type.

| DISCUSSION
Our results provide new insight into the prevalence and characteristics of partnerships between organizations that engage in protected area land management through comparing organizations that are observed to partner to those that were not observed partnering. We found that over 7 million acres of California protected land are managed by partnerships and that these partnerships occur between different organization types. Organization mission was also varied across those that were observed partnering versus those not observed partnering, with partnering organizations being more focused on environmental themes.
We observed that partnerships are involved in the protection of about 8% of the conservation estate in CA and occur between many different organizations of every type. Examples of this include The Nature Conservancy purchasing and then donating a subset of the land currently comprising the Fairfield Osborn Preserve to Sonoma State University but maintaining an easement on the land after the transfer (Sonoma State University, n.d.) and the Catalina Island Conservancy F I G U R E 2 Partnerships between organizations that manage 100 or more acres of land for conservation in California represented as network diagrams. (Panel a) Node size represents the number of entities of a type of organization that partner and the width of the edge indicates the number of shared protected areas managed. (Panel b) Nodes size represents the area of protected area managed by a type of organization and the width of the edge indicates the shared area managed.
managing Catalina Island land along with Los Angeles County holding a 50-year easement limiting development (Los Angeles County Department of Regional Planning, 2009). In fact, several of the partnering organizations express their desire to work with partners or to partner as part of their mission statement. For example, the Wildlife Conservation Board's mission includes "partnership with conservation groups, government agencies and the people of California" (WCB, 2021). Further work could explore the extent to which mission statements mention partnerships or other pathways to achieving conservation-oriented objectives.
Our definition of partnerships used in the analysis is limited, however. It excludes organizations that are involved in the process of protecting land but are not listed as a manager on a fee, easement, or designation. Data on land managed through conservation rental contracts, which are by definition partnerships between land owner (usually private) and renter (usually government), are not available through PAD-US. The largest such program in the United States, the Conservation Reserve Program, protects more land in the lower 48 than either the National Park Service or the Fish and Wildlife Service (Jackson et al., 2021). In addition, our dataset provides no information on partnerships that are not directly involved in land management, such as those developed to advocate for, facilitate, and monitor land protection. For example, it is common for protected areas to arise from partnerships between an organization that initially protected an area and then transferred the area to another organization. The Mojave Desert Land Trust, for example, acquires inholdings (private lands nestled within public lands) and holds onto them until a federal T A B L E 1 Frequency of top 20 most frequent stemmed phrases in mission statements of organizations that fall in the environmental and socioeconomic theme. Note: There are three phrases that each have 15 occurrences. For the main text we present results using "life communiti". The results are qualitatively similar when using either of the other two phrases-"life resid" or "recreat opportun" instead (see Table S6 and Figures S2 and S3).

(a) Partnering Organizations (b) Non-partnering Organizations
F I G U R E 3 Phrase frequencies in mission statements of organizations that manage protected land in California that partner on land management (a-left) and that do not partner (b-right).
agency is ready to purchase (Contreras, 2020). Another example of a type of partnership we are missing in our analysis is large-scale multi-organizational partnerships with specific conservation-oriented objectives. For example, our dataset shows no partnerships within the Channel Islands, but the Channel Island Biosecurity Program is a well-known and established partnership between Channel Islands National Park managers, The Nature Conservancy, and the United States Navy focused on protecting native species (NPS, n.d.). We also were unable to differentiate between partnerships related to acquisition strategy and those related to management need. These and other potential gaps could be filled by future work looking more broadly at the types of partnerships and role of partnerships in land protection. Our result that, for the organizations whose mission statements we were able to obtain, partnering organizations tended to use more environmental themed language in their mission statements, while non-partnering organizations tended to use language with socioeconomic themes, suggests opportunities for further exploration. Our evidence that organizations with environmental goals are engaged in partnering raises questions about whether they are pursuing mutual goals or attempting to generate complementary benefits. Additionally, although we find that partnering organizations overall use more environmentally themed language, there are also partnering organizations that do not, suggesting that partnerships between organizations with different objectives merit further exploration as a pathway toward improving conservation outcomes. The occurrence of these partnerships also points to the importance of exploring how partnering in protected land management fits within the broader socio-environmental system that includes socioeconomic drivers for land protection.
Our results also suggest that additional work to understand partnerships could enhance conservation outcomes. Our documentation of these patterns of comanagement provides empirical support to supplement theoretical work on partnering within the systematic, large-scale conservation planning literature (e.g., Bode et al., 2011;Iacona et al., 2016;Gordon et al., 2013). We also see this work as supplementing existing empirical work exploring the many dimensions of partnerships in conservation activities such as efforts to understand the diversity of stakeholders involved in caring for local environments (e.g., the STEW-MAP project from the US Forest Service) and work mapping and analyzing the social networks of conservation actors (e.g., Guerrero et al., 2020). Further work is also needed to explore how partnering can be planned as a potential mechanism to improve biodiversity outcomes while strategically managing land (Gerber & Rissman, 2012) and adhering to best practices related to collaborative conservation (Sanderson et al., 2022).
We acknowledge several potential biases in this work could arise due to presence of and/or accuracy of mission statements. First, bias could exist from the many organizations with missing mission statements, especially among those that were non-partnering. However, as it was primarily cities, cemeteries, and HOAs that were missing mission statements (see Tables S3 and S4), and these generally tended to not partner, we expect that our conclusions are robust. Relatedly, 48.23% of nonpartnering organizations as opposed to 78.62% of partnering organizations had a mission statement at all. This may be consistent with hypotheses from previous research that organizations may be missing opportunities for partnerships with other organizations by not having a specific mission statement (Dayer et al., 2016). Additionally, it is possible that the length, detail, and/or other aspects of the mission statements could be influenced by other factors such as organization size and budget. For example, smaller organizations may have less budget and/or time available to develop a mission statement. Finally, our analysis is constrained by the extent to which the mission statements reflect true organization objectives, which is likely limiting given past work that has found that land trusts generating wildlife and habitat conservation benefits are not necessarily including these outcomes as goals in their mission statements (Dayer et al., 2016).
Our work suggests that partnerships are worth exploring as a determinant of protected area siting, and that understanding how these partnerships arise could provide important insights into how to improve conservation outcomes. Our analysis confirms that partnering is occurring in California conservation land management and suggests that patterns of partnering are consistent with pursuit of environmental objectives by the participating organizations. Therefore, an improved understanding of conservation partnerships may assist with planning, policy and practice moving forward. An important step to F I G U R E 4 Average use of themed phrases by partnering status. meeting this goal is improving protected area databases to provide more detailed information on how areas ended up protected, in addition to more traditional information on the spatial extent of protection (e.g., Bartuska et al., 2022).