Plant diversity and generation of ecosystem services at the landscape scale: expert knowledge assessment

Authors


Correspondence author. E-mail: squijas@oikos.unam.mx

Summary

1. In spite of the increasing amount of experimental evidence on the importance of plant species richness for ecosystem functioning at local scales, its role on the generation of ecosystem services at scales relevant for management is still largely unknown. To foster research on this topic, we assessed expert knowledge on the role of plant diversity in the generation of services at the landscape scale.

2. We developed a survey that included three levels of organization and seven components of plant diversity; four provisioning, six regulating and four cultural services; as well as three resources and three conditions among key abiotic factors that are likely to provide a contribution to service generation equalling that of plant diversity. Eighty experts in areas of biodiversity, ecosystem functioning and services answered the survey.

3. The experts identified species diversity within a community and diversity of communities within the landscape as the most important levels of organization for service generation, both with positive effects. Composition and number of species were considered to be the most relevant components of plant diversity, the latter with a positive effect on services. Water availability was identified as the most important abiotic resource.

4. Our results suggest different approaches to management for sustaining the generation of services at the landscape scale. Provisioning services were perceived as largely influenced by abiotic resources and less so (although positively) by plant diversity. Regulating services were expected to strongly depend on both plant diversity and abiotic factors. A particularly strong positive effect of plant diversity was expected for the generation of cultural services. Some variation in answers could be attributed to expert background.

5.Synthesis and applications. The expert survey generated detailed information and new hypotheses on the relationship between plant diversity and services at the landscape scale. Future research is needed to test these hypotheses, yet the areas of agreement identified in this study can be used immediately, with caution, as synthetic expert knowledge at spatial scales that are relevant for management, to guide technological and policy interventions ensuring the maintenance of biodiversity and ecosystem service delivery.

Introduction

The study of the influence of biodiversity on ecosystem functioning has developed over the past 20 years into an important area of ecological research (Naeem et al. 2009). Results from this research clearly point to the critical role that biodiversity plays for the generation of ecosystem services and thus human well-being. To date, qualitative (Diaz et al. 2006) and quantitative (Balvanera et al. 2006; Worm et al. 2006; Quijas, Schmid & Balvanera 2010) syntheses derived from experimental manipulation of biodiversity and ecosystem functioning have shown a consistently positive effect of diversity in the generation of ecosystem services for a range of organisms, habitats and services.

However, these syntheses are limited because they mostly refer to the outcomes of small-scale experiments, are confined to a limited range of ecosystem services, deal mostly with species richness and not with other components of diversity, do not address the role played by abiotic factors in such relationships and potentially generate equivocal messages as a result of confounding effect directions from a range of ecosystem service providers and habitats. Experiments manipulating diversity have largely been developed at local (<10 ha) scales (Balvanera et al. 2006), and yet ecosystems are managed and services delivered at landscape (10–1000 ha) to regional (>1000 ha) scales (Kremen 2005; Duffy 2009). The syntheses are also limited by their focus on only a few types of ecosystem services, mainly regulating and provisioning ones (Balvanera et al. 2006; Worm et al. 2006; Quijas, Schmid & Balvanera 2010), cultural services being poorly described (Worm et al. 2006). Studies in this area only focus on the species level of organization, and on richness as the main measure of species diversity (Balvanera et al. 2006), thus ignoring the relative contributions of other biodiversity attributes, such as level of organization (e.g. genotype, populations, communities), species evenness, species composition and functional diversity (Kremen 2005; Diaz et al. 2007; Luck et al. 2009). Furthermore, the relative importance of abiotic factors with respect to diversity on ecosystem functioning or services has seldom been analysed (Diaz et al. 2007). Finally, a wide variety of organisms that may function as ecosystem service providers (Luck et al. 2009) have simultaneously been considered in syntheses (Balvanera et al. 2006; Worm et al. 2006), yet their individual contribution to ecosystem services is likely to be different (Luck et al. 2009; Quijas, Schmid & Balvanera 2010).

Future research on the relationship between biodiversity, functioning and services should embrace a broader range of spatial scales, biodiversity components and ecosystem services; include the relative roles of abiotic factors with respect to biodiversity; and dissect the roles of specific ecosystem service providers to be applied to management decisions. In particular, focusing on terrestrial plants as ecosystem service providers can be quite useful given their fundamental role in ecosystem functioning, their direct contribution to the generation of many ecosystem services and the unsurprising majority of studies focussing on the relationships between their diversity and ecosystem functioning (Quijas, Schmid & Balvanera 2010; Cardinale et al. 2011).

Given the lack of information described above, one approach is to synthesize present understanding, define key knowledge gaps and stimulate further hypothesis testing through expert assessments (Schläpfer, Schmid & Seidl 1999). Expert elicitation, a technique used to synthesize opinions of experts, has been in use for several decades in the disciplines of economics, sociology, political science, social psychology and public opinion research (De Vaus 2002), and more recently it is also being incorporated in studies of ecology and conservation biology (Halpern et al. 2007; Donlan et al. 2010). Expert judgment is not intended to be a substitute for scientific research, but to define the current knowledge that may not otherwise be easily accessible, illustrate the current sense of expert knowledge, reveal areas of greater or lesser agreement and help drive future applied research (Halpern et al. 2007; Donlan et al. 2010). An initial survey of expert knowledge aimed at understanding the relationship between biodiversity and its components on ecosystem processes, and the generation of ecosystem services, was conducted by Schläpfer, Schmid & Seidl (1999). In the last decade, biodiversity and ecosystem functioning research has added realism by including other components of diversity (e.g. Wilsey & Potvin 2000), increasing the spatial scale for analysis of processes related to ecosystem services (e.g. Wohl, Arora & Gladstone 2004), using more realistic extinction scenarios (e.g. Zavaleta & Hulvey 2004), simultaneously measuring multiple functions (e.g. Hector & Bagchi 2007) and incorporating natural (e.g. Fukami & Wardle 2005) and anthropogenic gradients of disturbance (e.g. Zavaleta et al. 2003). A new expert assessment would allow us to synthesize perceptions and advances relevant for managing both biodiversity and ecosystem services to ensure their maintenance.

In this article, we use an expert assessment approach to synthesize current understanding of how plant diversity relates to the generation of 14 ecosystem services at the landscape spatial scale. We focus on (i) direction of effect (DE) and relative importance of different levels of organization of plant diversity in the generation of ecosystem services, (ii) DE and relative importance of different components of plant diversity in the generation of ecosystem services and (iii) importance of plant diversity relative to the abiotic resources and conditions that can have direct effects on the generation of ecosystem services. We also assessed potential for biases in results given by the background of participating experts.

Materials and methods

Spatial Scale

Understanding the spatial scales at which ecosystems are managed and services are delivered to people will be essential to developing landscape-level conservation and management plans (Kremen 2005). For that reason, this assessment is focused at the landscape scale, defined here as areas ranging from 10 to 1000 ha (0·01 to 10 km2; 24 710 to 2 471 000 acres). We used analogies to help experts visualize theses scales (e.g. the Principality of Monaco or Niagara Falls has an area of 100 ha, whilst the city of Beverly Hills has an area of 1000 ha).

Components of the Survey

A summary of the approach that we used in this survey is shown in Fig. 1. We divided the survey into five sections. The first and second sections evaluated understanding about the DE and relative importance (RI) of the levels of organization (LO) of plant diversity on service generation. The third and fourth sections evaluated understanding about the DE and RI of components of plant diversity (CD) on service generation. The fifth section evaluated understanding about RI of resources and conditions (R&C). We considered four provisioning services, six regulating services and four cultural services. We did not include supporting services, as we considered them to be ecosystem processes that indirectly benefit societies by supporting one of the other three types of services. We identified three LO relevant for the generation of services and six CD. We considered three resources and three conditions that are likely to modify both plant diversity and service generation. A glossary with all definitions can be found in Appendix S1.

Figure 1.

 Framework for the design of the survey on plant diversity and the generation of ecosystem services.

Identification of the Relationship Between Diversity and Services

The DE of plant diversity on the generation of ecosystem services was assessed using five types of relationships: (i) the more diversity, the more service (+), (ii) the more diversity, the less service (−), (iii) there is a diversity effect on the service, but it is not possible to determine its direction or the direction is unknown (1), (iv) no influence of diversity on the generation of the service (0), (v) unknown whether there is an influence of plant diversity for the generation of the service (?). For example, ‘higher genetic diversity provides better pest regulation’, and thus, the more diversity there is, the more service (+); or ‘a higher number of species does not influence soil fertility’; thus, there is no influence of diversity on the generation of service (0).

The RI of plant diversity, as well as that of R&C, on the generation of services was assessed using four categories: (1) of little importance, (2) of intermediate importance, (3) very important and (?) of unknown importance.

Building the Survey

Two drafts of the survey were developed before reaching the final version used here (for details see Appendix S2). In the final version, we included questions on the background of the expert to evaluate potential biases on the final results owing to individual perspectives; the questions related to subject of expertise (e.g. population ecology, management), type of work (e.g. basic research, decision maker), type of organization (e.g. institute, environments NGOs), number of years working with plant diversity and ecosystem processes and/or services they know, and the focus ecosystems they study (e.g. tropical rainforest, grassland).

Expert Selection

We defined experts as individuals who had carried out observational or experimental research on the links between plant diversity and ecosystem functioning or ecosystem services at different spatial scales. We used the contributors to the Millennium Ecosystem Assessment (MA 2005) and researchers suggested by the co-authors of this work as a source of candidates. The survey was sent by email to the 344 experts in that list for which emails could be obtained between July 2007 and March 2008; up to two reminder emails were sent. Some (56 experts) did not participate in the survey, because of time constraints (17), or to self-stated lack of expertise on plant diversity and ecosystem services (38). A total of 80 experts (23%) from 27 countries and working in 14 types of ecosystems responded to the survey (more details see Appendix S3).

Analysis

We registered the frequencies of the different types of responses and then addressed the following questions:

  • 1 Do experts tend to recognize a particular DE or relative importance of plant diversity in the generation of ecosystem services when considering all levels of organization or all components of plant diversity?
  • 2 What is the DE and relative importance of the different levels of organization and components of plant diversity most commonly mentioned by experts when considering all services?
  • 3 What is the relative importance of resources and conditions with respect to plant diversity most frequently mentioned when considering all services?
  • 4 What is the DE and relative importance of the different levels of organization most frequently recognized by experts for the generation of each of services?
  • 5 What is the DE and relative importance of the different components of plant diversity most frequently recognized by experts for the generation of each of services?
  • 6 What is the relative importance of resources and conditions with respect to plant diversity most frequently recognized for the generation of each of services?
  • 7 How does the background of experts explain the differences in how frequently they chose different DE and relative importance they assigned to plant diversity and resources and conditions for the generation of services?

To test for differences in answer probabilities, we assumed as a null model that all answers to a question would be equally likely. Thus, the five possible answers (−, 1, +, 0, ?) about DE and the four possible answers (1,2,3,?) about the RI of plant diversity and R&C on the generation of services were all considered equally likely. Whilst this may not always have been the best null model, we had no further information to develop more specific null models. Significant deviation from the expected equal answer probabilities was detected with generalized linear and chi-square tests (Sokal & Rohlf 1995; for details see Appendix S4). We used Bonferroni corrections to account for the large numbers of test performed (360 hypotheses tested; critical < 0·00014 for individual tests, corresponding to an overall significance level of < 0·05). We used adjusted residuals (residuals divided by their variance) as an a posteriori test for identifying particular frequencies responsible for the significant chi-square values, and to explore whether particular answer frequencies were larger (or smaller) than expected from the null model (Everitt 1992).

We assessed biases caused by different backgrounds of experts on frequencies of responses regarding the links between plant diversity and ecosystem services. We used chi-square tests to test for independence of answers with respect to expert characteristics (e.g. five types of relationships vs. type of ecosystem). We used Bonferroni correction on the critical P values as a priori test for the numbers of test performed (25 hypotheses tested; critical < 0·002). Adjusted residuals were also used to identify particular frequencies that differed from the null model (calculated using the same Bonferroni correction).

Results

We found that most experts recognized a positive effect of plant diversity on the generation of ecosystem services when all LO (Fig. 2a) and all CD (Fig. 2b) were pooled together. However, there was no consensus on the RI of plant diversity in the generation of services. We found large discrepancies between experts, with opinions for importance varying between little importance and very important often considered for levels of organization (χ2 = 8·1, = 0·01, d.f. = 2). Yet, most CD were frequently considered to be of little importance (χ2 = 22·8, < 0·0001, d.f. = 2).

Figure 2.

 Total frequency of expert assessment on direction of effect of levels of organization and components of plant diversity on the generation of ecosystem services. In (a), the total frequencies of answers for each type of effect were obtained by adding up both all levels of organization and all services; in (b), the total frequencies of answers were obtained by adding up both all components and all services. Arrows indicate that frequencies were significantly higher (↑) or lower (↓) than expected from a null model of equal frequencies of all answers; **** = < 0·0001.

Levels of Organization

Genetic diversity within species, species diversity within a community and the diversity of the communities in the landscape were consistently identified as having a positive effect on ecosystem service generation (genetic χ2 = 601·4, < 0·00014, d.f. = 4; species χ2 = 1980·1, < 0·00014, d.f. = 4; community χ2 = 1609·7, < 0·00014, d.f. = 4; Table 1). However, the RI changed between LO, because plant diversity at the level of species (χ2 = 594·6, < 0·00014, d.f = 3) and communities in the landscape (χ2 = 474·4, < 0·00014, d.f. = 3) levels were most often considered as very important, whilst diversity at the genetic (χ2 = 1110·57, < 0·00014, d.f. = 3) level of organization was most frequently recognized as of little importance.

Table 1.   Expert assessment of direction of effect and relative importance of levels of organization and components of plant diversity on the generation of ecosystem services. Cells with a square indicate levels of organization and components that were significantly more frequently mentioned than would be expected from a null model of equal frequencies of all answers, both for effects and for relative importance; < 0·00014 (▪); cells with no square indicate no significantly different frequencies from those expected from the null model. (--) is used to show that the direction of species composition effect on the generation of services could not be assessed (see text for details)
Attributes of plant diversity Direction of effectRelative importance
 1+0?123?
  1. Effects: (−) the more diversity the less service; (1) there is a diversity effect on the service, but it is not possible to determine its direction or the direction is unknown; (+) the more diversity the more service; (0) no influence of diversity on the generation of the services; (?) unknown whether there is an influence of plant diversity for the generation of services. Relative importance: (1) little importance; (2) intermediate importance; (3) very important; (?) unknown importance.

Levels of organizationGenetic diversity within species      
Species diversity within a community      
Diversity of communities within in the landscape       
ComponentsNumber of species       
Species evenness       
Species composition----------  
Functional diversity       
Spatial turnover       
Structural diversity      

Components of Diversity

Positive effects of different CD on service generation were reported, but their RI differed between components (Table 1). Clear positive effects were attributed to number of species (χ2 = 1575·7, < 0·00014, d.f. = 4), species evenness (χ2 = 395, < 0·00014, d.f. = 4), functional diversity (χ2 = 661,< 0·00014, d.f. = 4), spatial turnover (χ2 = 411·5, < 0·00014, d.f. = 4) and structural diversity (χ2 = 1044·8, < 0·00014, d.f. = 4). Whilst number of species was consistently considered as very important (χ2 = 272, < 0·00014, d.f. = 3), plant species composition and structural diversity were thought to be of intermediate importance to very important (composition χ2 = 531, < 0·00014, d.f. = 3; structural χ2 = 227·7, < 0·00014, d.f. = 3). Other CD including species evenness, functional diversity and spatial turnover were rated as less important for the generation of services.

Abiotic Resources and Conditions

Plant diversity together with water availability was most frequently considered by experts as the most important factors for the generation of ecosystem services (plant diversity χ2 = 79·7, < 0·00014, d.f. = 3; water χ2 = 37·1, < 0·00014, d.f. = 3; Table 2). Disturbance intensity was identified to be of only intermediate importance; soil type and position on the landscape were rated of intermediate to little importance. Energy and nutrient availability were recorded as the least important for the generation of services.

Table 2.   Expert assessment on the relative importance of plant diversity with respect to that of resources and conditions for the generation of ecosystem services. Presentation of cells follow those of Table 1; < 0·00014 (▪)
  Relative importance
123?
  1. Relative importance: (1) little importance; (2) intermediate importance; (3) very important; (?) unknown importance.

 Plant diversity  
ResourcesWater availability   
Energy availability   
Nutrient availability   
ConditionsSoil type  
Position within the landscape  
Disturbance intensity   

Levels of Organization and the Types of Ecosystem Services

Experts consistently regarded species diversity within a community as the most important LO, with positive effects on provisioning services (Table 3). Genetic diversity was recognized as the least important for these services, with either positive effects or non-effects most commonly reported.

Table 3.   Expert assessment of direction of effect and relative importance of levels of organization of plant diversity on the generation of ecosystem services. Presentation of cells follow those of Table 1; < 0·00014 (▪)
TypeServicesDirection of effectRelative importance
GeneticSpeciesCommunityGeneticSpeciesCommunity
1+0?1+0?1+0?123?123?123?
  1. Effects: (−) the more diversity the less service; (1) there is a diversity effect on the service, but it is not possible to determine its direction or the direction is unknown; (+) the more diversity the more service; (0) no influence of diversity on the generation of the services; (?) unknown whether there is an influence of plant diversity for the generation of services. Relative importance: (1) little importance; (2) intermediate importance; (3) very important; (?) unknown importance.

ProvisioningFood, fodder, fibre and biofuel intensive production                     
Timber production                       
Firewood production                       
Diverse products                      
RegulatingSoil fertility                       
Plant pests                       
Resistance to plant invasion                      
Response of the ecosystem to extreme events                     
Water availability                     
Climate regulation and air quality                     
CulturalScenic beauty                     
Source of inspiration                     
Recreation and tourism                     
Traditional use                      

Species diversity within a community and the diversity of the communities in the landscape were most often considered to be from intermediate importance to very important for the generation of regulating services; in contrast, genetic diversity was considered the least important. Positive effects of species diversity and diversity of the communities were indicated for all regulating services. Positive or non-existent effects of genetic diversity on regulating services were most often chosen (Table 3).

Experts considered plant diversity at the species and community levels of intermediate importance to very important for the generation of cultural services, and positive effects were the most frequent in both cases. Genetic diversity was recognized as the least important for cultural services, with no-effect most commonly reported (Table 3).

Components of Diversity and the Types of Ecosystem Services

Species composition, that is, specific combination of the species or presence/absence of particular species within a plant community, was the only DC that was most frequently identified by experts as very important for provisioning services (Table 4); most of the other components of plant diversity were expected to have positive effects of little importance on provisioning services.

Table 4.   Expert assessment of direction of effect and relative importance of components of plant diversity on the generation of ecosystem services. Presentation of cells as Table 1; < 0·00014 (▪)
TypeServicesDirection of effectRelative importance
Number of speciesSpecies evennessFunctional diversitySpatial turnoverStructural diversityNumber of speciesSpecies evennessSpecies compositionFunctional diversitySpatial turnoverStructural diversity
1+0?1+0?1+0?1+0?1+0?123?123?123?123?123?123?
  1. Effects: (−) the more diversity the less service; (1) there is a diversity effect on the service, but it is not possible to determine its direction or the direction is unknown; (+) the more diversity the more service; (0) no influence of diversity on the generation of the services; (?) unknown whether there is an influence of plant diversity for the generation of services. Relative importance: (1) little importance; (2) intermediate importance; (3) very important; (?) unknown importance.

ProvisioningFood, fodder, fibre and biofuel intensive production                                              
Timber production                                          
Firewood production                                          
Diverse products                                          
RegulatingSoil fertility                                            
Plant pests                                          
Resistance to plant invasion                                          
Response of the ecosystem to extreme events                                          
Water availability                                             
Climate regulation and air quality                                             
CulturalScenic beauty                                        
Source of inspiration                                           
Recreation and tourism                                         
Traditional use                                          

For regulating services, species composition and number of species were considered to be the most important CD; positive effects were reported for the generation of soil fertility, regulation of plant pests and invasion resistance (Table 4). The functional diversity was considered important with for the regulation of the response of the ecosystem to extreme events. Positive effects of functional diversity and structural diversity were identified for water-related services and for climate regulation and air quality.

Species composition was most often considered as the most important CD for the generation of various cultural services. Structural diversity was identified as important for scenic beauty, and number of species was important for traditional use. Most of the CD were identified as having positive effects on the generation of all cultural services, except for functional diversity, which was considered to have no effects on these services (Table 4).

Abiotic Resources and Conditions and the Types of Ecosystem Services

The relative importance of R&C differed markedly among types of services (Table 5). Water, energy and nutrient availability were considered most important for the majority of provisioning services. Plant diversity was most important for three regulating services, water availability for one service, nutrient availability, soil type position within the landscape for one service and disturbance intensity for two services. In the case of cultural services, only water availability was identified as important for the generation of recreation and tourism services; in all other cases, only plant diversity was considered important.

Table 5.   Expert assessment on the relative importance of plant diversity with respect to that of abiotic resources and conditions for the generation of ecosystem services. Presentation of cells as Table 1; < 0·00014 (▪)
TypeServicesPlant diversityResourcesConditions
Water availabilityEnergy availabilityNutrient availabilitySoil typePosition within the landscapeDisturbance intensity
123?123?123?123?123?123?123?
  1. Relative importance: (1) little importance; (2) intermediate importance; (3) very important; (?) unknown importance.

ProvisioningFood, fodder, fibre and biofuel intensive production                        
Timber production                       
Firewood production                        
Diverse products                           
RegulatingSoil fertility                         
Plant pests                       
Resistance to plant invasion                         
Response of the ecosystem to extreme events                          
Water availability                          
Climate regulation and air quality                           
CulturalScenic beauty                        
Source of inspiration                        
Recreation and tourism                       
Traditional use                          

Biases Given by Expert Background

Biases associated with the background of experts were found in the frequencies of responses for the DEs and relative importance (RI) of plant diversity on service generation. When pooling together all LO, we found an effect of the type of organization (χ2 = 54·6, < 0·002, d.f. = 12) and of the focus ecosystem that scientists studied (χ2 = 205·5, < 0·002, d.f. = 52): experts working for NGOs indicated that plant diversity had no effects on ecosystem services more frequently than expected from a null model, whilst those working in agroecosystems more frequently chose an unknown influence. The RI of plant biodiversity on ecosystem service provision was influenced by the focus ecosystem (χ2 = 104·5, < 0·002, d.f. = 39); experts working in agroecosystems and successional forest chose an unknown relative importance of plant diversity more frequently. When pooling together all CD, we found that subject of expertise (χ2 = 347, < 0·002, d.f. = 24), type of organization (χ2 = 203·3, < 0·002, d.f. = 12), years of experience (χ2 = 87·8, P < 0·002, d.f. = 12) and focus ecosystem (χ2 = 516·6, < 0·002, d.f. = 52) influenced expert assessments: plant and community ecologists identified a positive relationship between plant diversity and service generation more frequently, whilst managers, individuals working for NGOs and governmental agencies, experts with less experience (<10 years) and those focusing on agroecosystems and successional forests tended to report no effects and unknown relationships more often. Responses to the RI of plant diversity were biased by subject of expertise (χ2 = 120·3, < 0·002, d.f. = 18) and focus ecosystem (χ2 = 103·8, < 0·002, d.f. = 39); experts studying population ecology and agroecosystems tended to choose the unknown importance option more frequently than expected. Also, respondents working in agroecosystems reported unknown importance of plant diversity relative to R&C on service generation more frequently (χ2 = 72·2, < 0·002, d.f. = 39). For more details see Appendix S5.

Discussion

Overarching Patterns

Our work synthesized perceptions of experts on the relationships between plant diversity and ecosystem services at the landscape scale. The assessment showed that when all the services were pooled together, a positive effect of plant diversity on service provision was consistently found. The generality of this positive effect is consistent with previous publications on the contribution of diversity to a stable supply of ecosystem services as spatial and temporal variability increases, which typically occurs over larger areas, such as the ones addressed here (Loreau, Mouquet & Gonzalez 2003; Diaz et al. 2006).

However, there was no consensus on the relative importance of plant diversity on service provision. The relative importance given to diversity changed among services. This lack of consensus may derive from different underlying mechanisms, the effect of context or just the scarcity of information in this topic.

Consistent with previous qualitative reviews (Kremen 2005; Diaz et al. 2006), experts consistently identified diversity of species within a community and diversity of communities within a landscape as the most important levels of organization for service generation. In both cases, a positive relationship was suggested. Composition and number of species were considered to be the most relevant components of plant diversity. However, experts interviewed saw little importance for the role played by genetic diversity in the provision of services, despite the growing evidence of its role in generating useful plant products such as food, fodder and fibre (Jackson, Pascual & Hodgkin 2007), regulation of plant pests (Schweitzer et al. 2005) and resistance to plant invasion (Barberi et al. 2010) at the landscape scale. Our results reflected consistency only for what appear to be the more widely known levels of organization and components of plant diversity among ecologists (Hooper et al. 2005).

Patterns for Different Types of Ecosystem Services

Positive relationships between plant diversity and the generation of provisioning services at the landscape scale were consistently reported, thus suggesting that experts expected clear local synergy between the maintenance of biodiversity (in conserved habitats or agroecosystems) and obtaining provisioning services to satisfy human needs. The role of species evenness and functional diversity on the generation of provisioning services (e.g. food, fodder, fibre and intensive biofuel production) was not recognized. This is inconsistent with the growing evidence critical role that agrobiodiversity plays in much of the world’s food supply (Perfecto & Vandermeer 2010).

Positive effects of plant diversity were consistently recognized for some regulating services, and not for others. We recognized two main groups: those with answer consistency and those with no consistency. The first group corresponds to services that have been widely studied at the local scale and studied less at the landscape scale (e.g. soil fertility, plant pest, resistance to plant invasion; Landis et al. 2005; Vacher et al. 2008). The second group, showing a lack of consistency for the questions on DE and relative importance, corresponds to services for which these links have not been documented (e.g. water amount, quality and temporal variability). Lack of consistency for the links between plant diversity and regulating services has been explored from the processes that underpin them (e.g. Quétier et al. 2007; more detailed references in Appendix S6), and the importance of functional diversity on regulating services at landscape scales (e.g. maintenance of soil fertility, landslide and avalanche risk; Diaz et al. 2007; Lavorel et al. 2011). The lack of consistency in these factors identified in the current study suggests either a lack of understanding of the ecological processes associated with service generation, a lack of clear patterns emerging from such literature or a lack of awareness of a larger set of biophysical literature outside the field of ecology.

Finally, experts consistently recognized a positive effect of plant diversity on the generation of cultural services. This is surprising given ecologists’ limited understanding of these services (Chan et al. 2011), although the role of functional diversity at the landscape scale has been shown for these services (Diaz et al. 2007; Lavorel et al. 2011). The consensus may reflect connections between cultural diversity and biodiversity, which are widely known in the fields of economic botany, indigenous use of native plants, and anthropology and cultural services (e.g. cultural heritage, land stewardship; Quétier et al. 2007).

Relative Importance of Abiotic Resources and Conditions

The relative importance of resources and conditions with respect to plant diversity on service provision at the landscape scale varied greatly among types of services. Provisioning services were perceived to be largely influenced by abiotic resources and little influenced (although positively) by plant diversity; regulating services were thought to depend on both plant diversity and abiotic resources and conditions; abiotic factors were not considered relevant for cultural services. These results differ from previous syntheses (Loreau et al. 2001; Hooper et al. 2005; Balvanera et al. 2006) and may indicate that experts assumed a greater influence of abiotic factors among than within sites, thus increasing the relative importance of resources and conditions at the larger landscape scale (Loreau, Mouquet & Gonzalez 2003). Further research is needed to fully understand how resources and conditions are related to biodiversity and service provision as spatial and temporal scale increase.

Limitations of Assessment: Representativeness and Biases of Experts and Data Analyses

The assessment was aimed at identifying people who have carried out research on plant diversity and ecosystem services and is thus not based on a random sample of people. The percentage of experts that responded to the survey (23%) is similar to that reported by other expert opinion studies on biodiversity (Schläpfer, Schmid & Seidl 1999; Halpern et al. 2007). The majority of experts worked in community and ecosystem ecology at research institutes and had more than 10 years of experience with plant diversity or ecosystem processes/services at the time of the survey. Analysing the characteristics that described the experts in this study (see Appendix S3) showed that they have thought deeply about the subject and were thus likely to provide authoritative estimates on plant diversity and ecosystem services. Furthermore, their understanding of the topic was very likely to have included experience beyond their particular research publications.

Some biases emerged from expert background. As explained above, experts working in management, in NGOs or governmental agencies, those focusing on transformed ecosystem (e.g. cropland, pasture, exotic pine and eucalyptus plantations, irrigated rice fields, secondary tropical wet forest, savanna transition) and those with less experience (<10 years) tended to report no influence of plant diversity on service generation (nature of effects and relative importance) more frequently. This suggests that experts dealing with real-world conditions are either more sceptical of emerging research findings derived from the experimental literature or less aware of them or that information relevant to their needs is lacking; for instance, a rigorous meta-analysis of the DE of diversity in agricultural environments has only been recently published (Letourneau et al. 2011).

The future will usher in new questions on biodiversity and ecosystem services and allow access to new empirical data or additional expert opinion. Given the broad reach of the Internet, Web-based expert opinion surveys are a strategic way to aggregate information that can help set priorities for conservation and management action plans and related research (Donlan et al. 2010). Priority setting for maintenance of biodiversity and management of ecosystem services at the landscape scale cannot wait for exhaustive empirical research. Instead, survey instruments can be easily replicable and quickly updated to include new sources of information, control for expert bias and refine the results from DE and relative importance of plant diversity on services generation.

The statistical methods used here were very useful to address the questions posed. The Bonferroni adjustments may have increased the probability of not rejecting some null hypothesis when it would have been appropriate to do so (Moran 2003). However, changes in the observed patterns with this correction were only found for the most complex assessments of the effects on individual ecosystem services (Tables 4 and 5), and so the adjustments helped us to more clearly identify the major inconsistencies among experts (Appendix S5). The use of generalized mixed effect models with individual experts as a random factor and expert background categories as fixed contrasts within this random factor would have been an alternative analysis possibility (De Vaus 2002). Owing to the complexity of the design, we used the described chi-square tests without random factors instead.

Future Research: Main Questions and Management Implications

The results of our expert knowledge assessment can provide different hypotheses about the relationships between plant diversity and services generation that could be tested further. Future research should focus more on the relative importance of plant diversity on services, rather than the direction of its effects, which are better known.

Plant genetic diversity effects seem to differ between ecosystem services types. Also, the relative role of plant diversity on service provision may change across types of ecosystems. The role of components of plant diversity such as functional diversity may change across different types of regulating and cultural services. Finally, exploring the relative role of plant diversity with respect to that of abiotic resources and conditions for different types of ecosystem services is particularly relevant for management. This set of hypotheses should help to identify unexamined research questions that would lead to a novel approach to observational and experimental studies for the foundation of rigorous and science-based evidence for the management, conservation and sustainable use of biodiversity and ecosystem services at the landscape scale.

Our approach and results can be used in a number of ways to inform and aid management decisions. This expert assessment can identify themes of agreement that may be used with caution as a synthesis of expert knowledge to guide technological and policy interventions. It also highlights themes for which closer communication between scientists and managers is needed.

Conclusions

Our survey revealed the attributes of plant diversity that, in the eyes of the interviewed experts, are most likely to have effects on the generation of services at landscape scales. Expert assessment identified diversity of species within a community and diversity of communities within a landscape as the most important levels of organization for service generation with positive effects; the same can be said for composition and number of species among the components of plant diversity. Water availability was perceived to be the most important abiotic resource for service generation at the landscape scale; but this was not the case for all services. Provisioning services were thought to be largely influenced by abiotic resources and little influenced (although positively) by plant diversity. Sustaining the generation of regulating services was expected to depend on both plant diversity and abiotic resources and conditions. A very important positive effect was attributed to plant diversity for the generation of cultural services. Most experts know and seem to trust the results of observational and experimental research showing that plant diversity increases ecosystem functions; this pattern was true even for experts carrying out more real-world work, who were among the most sceptical about the existence of such links. Key areas of future research to guide ecosystem management include the role of plant genetic diversity and that of abiotic factors on landscape (i.e. management-relevant) scales. Overall, the experts interviewed agreed that, at the landscape scale, the importance of maintaining plant diversity is crucial if the management goal is to ensure and sustain provision of ecosystem services for human well-being.

Acknowledgements

The authors thank all the experts that answered the survey. M. Martínez-Ramos, V. Jaramillo, F. Espinosa, R. Linding, L. Barraza, M. Quesada, J. Benítez, E. del Val, H. Paz, D. Pérez, N.M. Montaño and L. Romero provided suggestions to earlier drafts of the survey. We thank Alberto Valencia, Heberto Ferreira and Atzimba López for technical support. We would also like to thank three anonymous referees who greatly helped to improve this manuscript. This article constitutes a partial fulfilment of the Graduate Programme in Biological Sciences of the National Autonomous University of México (UNAM). S.Q. acknowledges the scholarship and financial support provided by the National Council of Science and Technology (CONACyT), and UNAM. Research was funded by grant SEP-CONACYT-2005-50955 to P.B; P.B acknowledges sabbatical support from PASPA-UNAM and the Department of Biology, Stanford University.

Ancillary