A co-development approach to conservation leads to informed habitat design and rapid establishment of amphibian communities

U SIR is a digi t al collec tion of t h e r e s e a r c h ou t p u t of t h e U nive r si ty of S alford. Whe r e copyrigh t p e r mi t s, full t ex t m a t e ri al h eld in t h e r e posi to ry is m a d e fre ely availabl e online a n d c a n b e r e a d , dow nloa d e d a n d copied for no nco m m e rcial p riva t e s t u dy o r r e s e a r c h p u r pos e s . Ple a s e c h e ck t h e m a n u sc rip t for a ny fu r t h e r copyrig h t r e s t ric tions.

Pl e a s e c h e c k t h e m a n u s c ri p t fo r a n y fu r t h e r c o py ri g h t r e s t ri c tio n s.
Fo r m o r e info r m a tio n, in cl u di n g o u r p olicy a n d s u b mi s sio n p r o c e d u r e , pl e a s e c o n t a c t t h e R e p o si to ry Te a m a t: u si r@ s alfo r d. a c. u k .

INTRODUCTION
Increased recognition of the global biodiversity crisis and the need for biodiversity conservation have in the 21st century coincided with higher economic uncertainties (Díaz et al., 2019;IMF, 2020), leading to reduced investment in conservation (Hirsch et al., 2020). As conditions worsen, there is an increasing need for effective interventions that are well-executed and supported by empirical evidence (Addison et al., 2013). Co-development of projects by land managers, local citizens, conservation practitioners, policy makers and researchers has been proposed as a means of implementing successful and efficient conservation actions (Vercammen & Burgman, 2019;Wauters & Mathijs, 2013). The evidence-based conservation approach appears to be a good fit with co-development as it recommends the systematic evaluation of peer-reviewed and grey literature along with practitioners' and local people's experience (Sutherland, Pullin, Dolman, & Knight, 2004). Conservation research that is both practical and focuses on finding solutions is likely to appeal to a broad range of stakeholders, as is a systematic approach that includes social and economic, as well as ecological, considerations (Gilby et al., 2020).
The decline of amphibians is a major global biodiversity concern and has been linked to anthropogenic drivers including habitat loss, land use change, pollution, over-exploitation, disease and invasive species (Beebee & Griffiths, 2005;Scroggie et al., 2019), often acting in combination. Remediation of such threats can be particularly challenging outside of protected areas such as in places primarily managed for economic purposes (Denoël et al., 2019;Hartel, Scheele, Rozylowicz, Horcea-Milcu, & Cogȃlniceanu, 2020). Indeed, land managed for forestry and agriculture, rather than for conservation per se, comprises a greater proportion of rural land globally and provides vital habitat for many species (e.g. Irwin et al., 2014;Pywell et al., 2012). Amphibians in particular are characterized by limited powers of dispersal, and land managed for economic gain may thus be the only connection between nature reserves and other sites managed for conservation (Pickett & Thompson, 1978). In such situations, proactive habitat management outside protected areas benefits a range of amphibian species and aids in population recovery (Sterrett et al., 2019).
One widely promoted, and effective, form of intervention in agricultural areas, where traditional nature reserves may be too prescriptive, is the conservation or creation of 'pondscapes' (ecologically meaningful clusters of ponds) designed to support local wildlife, including aquatic-breeding amphibian communities (Hill et al., 2018;Peterman, Anderson, Drake, Ousterhout, & Semlitsch, 2013;Rannap, Lõhmus, & Briggs, 2009). Pondscapes may be funded through agri-environmental incentive schemes, or as part of mitigation measures following anthropogenic developments. However, such interventions can appear to be piecemeal, particularly for amphibians, which has led to calls for landscape scale conservation planning to consider the ecological requirements and connectivity needs of target species (Brown, Street, Nairn, & Forstner, 2012;Rannap et al., 2009).
Interventions for amphibian conservation must generally take account of both aquatic and terrestrial habitats to provide breeding, feeding and dispersal opportunities (Schmidt, Arlettaz, Schaub, Lüscher, & Kröpfli, 2019). However, pond construction and restoration projects have had mixed success, and relatively few have been evaluated for their overall effectiveness (Smith, Meredith, & Sutherland, 2019). Ecological features of constructed and restored ponds can significantly influence the species richness and viability of residing amphibians (Peterman et al., 2013;Shulse, Semlitsch, Trauth, & Gardner, 2012), which is important for intervention projects because these features can markedly differ between established and newly created sites (Drayer & Richter, 2016;Korfel, Mitsch, Hetherington, & Mack, 2010). Studies which document the colonization and temporal persistence of amphibians in created or restored ponds indeed show that such sites often fail to reach typical community composition or might not attain community diversity index values typical of more established ponds (Lesbarrères, Fowler, Pagano, & Lodé, 2010). Even when available, empirical information is rarely considered in the design phase of habitat management.
This study summarizes a two-step research and restoration effort on a five-species amphibian assemblage in a rural area in Northern Scotland. The two sequential objectives are to (i) quantify local amphibian habitat preferences to inform the design of constructed and restored ponds and (ii) document amphibian community diversity of these ponds in comparison with existing sites, to determine whether habitat assessment-based design promotes their rapid colonization.

Study area and study design
The study was based on 88 reference ponds and 25 intervention ponds  The construction or restoration of intervention ponds was based on two amphibian conservation projects co-developed by practitioners (including two of the authors, DOB and KOB), local farmers, foresters, a golf club, governmental organizations and a regional conservation NGO (Highland Amphibian and Reptile Project -HARP). Lengthy discussions were held with each potential stakeholder to explain the rationale for amphibian conservation and to understand the constraints and opportunities that conservation measures might pose for their businesses. Interventions took place at four nodes within the study area, which were selected using criteria including connectivity, soil/geology and compatibility with economic use (Figure 1; Table A1 in the F I G U R E 1 Distribution of intervention and reference ponds across the study area. The four pond nodes are named in the map, and magnified below (the complete list of intervention ponds is shown in table A1 in the online Appendix A).
TA B L E 1 Selected designable habitat characteristics of the reference ponds used for the statistical analyses. Abbreviations used in Figure 2 are given in brackets. Detailed information about all variables is given in Appendix A in the Supporting Information

Assessment of local amphibian habitat preferences
The 88 reference ponds served as a basis to quantify local habitat preferences, before applying these findings to develop criteria to the design and siting of the intervention ponds (see below). Pond surface areas ranged from 2 to 168,500 m 2 (median 975 m 2 ), and altitudes To investigate which habitat features were most important for amphibian community composition, we collected habitat data from 129 variables for these ponds, 88 derived through fieldwork in 2014 and 41 through desk study (for further details on habitat descriptors, see Appendix A in the Supporting Information and Miró et al., 2017).
Topographical features were obtained from GIS using 1:25,000 maps from the British mapping agency Ordnance Survey. Given the conservation management context of the study and the proximity of occupied and non-occupied control ponds, we did not include spatial autocorrelation variables to avoid unnecessarily complexity and collinearity in the models.

Intervention pond construction and surveys
The 25 intervention ponds were constructed or restored (i.e. former ponds identified from historic maps) for amphibian conservation in Autumn-Winter 2014-2015 (except one pond constructed in Winter 2017-2018), using a 13-t excavator on low-pressure tracks. Based on the findings from habitat preferences for the reference ponds (see below), particular attention was paid to land use, soil type, bank slope, and terrestrial habitat when constructing or restoring these ponds (Table A1 in

Statistical analyses
The relative importance of habitat characteristics in shaping amphibian communities was assessed by redundancy analysis (RDA; Wollenberg, 1977 (Legendre & Legendre, 2012). We sequentially computed a variance inflation factor (VIF) for all variables and rejected the ones with the highest value until none of them was above 3.
Seventeen variables obtained VIF values below 3, the threshold indicative of worrisome collinearity in regression analyses (Zuur, Ieno, Walker, Saveliev, & Smith, 2009). The detailed RDA variable selection procedure can be found in Appendix A in the Supporting Information.
Since the design of intervention ponds was based on reference ponds, we tested for the similarity of habitat characteristics between both groups of ponds by applying Kruskal-Wallis non-parametric tests for group homogeneity, followed by post hoc pairwise comparisons with Bonferroni adjustment. Since B. bufo is favoured by different variables than the other amphibian species and prefers larger, deeper ponds commonly containing fish (Hartel et al., 2007;Knapp, 2005;Miró, Sabás, & Ventura, 2018;Winandy, Darnet, & Denoë, 2015), we split the reference ponds into two groups: with amphibians present, and without amphibians or with B. bufo only. We illustrated the data using violin plots (Hintze & Nelson, 1998).
We investigated temporal changes of diversity patterns in the intervention ponds by computing 3 yearly diversity measures. First, amphibian richness was computed as the number of amphibian species per pond, based on all records for a given year. Second, amphibian pooled richness was defined as the total number of species found in the pool of intervention ponds. Finally, amphibian community dissimilarity (or beta diversity) was defined as the total variance of the site-byspecies (rows and columns, respectively) amphibian community table and computed as the total sum of the matrix squared deviations from the column means (Legendre & De Cáceres, 2013), again using the Euclidean distance of Hellinger transformed data (Legendre & Gallagher, 2001 (Cohen, 1988). We

Habitat assessments based on reference ponds
Analysing the pond habitat characteristics most associated with the taxonomic structure of the amphibian community in the reference ponds, we obtained a significant RDA (F = 4.681, p = 0.001), which explained 35.1% of the total variance ( Figure 2a). All nine variables identified by the forward selection process were significant, and most  (Table A2). The occurrence of R. temporaria was positively linked to terrestrial habitat richness and proportion of adjacent thickets and scrub, and negatively linked to proportion of organic mud substrate (Table A2). Among salamanders, L. helveticus was positively associated with macrophyte surface coverage and proportion of adjacent mixed woodland, and negatively associated with fish presence, proportion of tree coverage of the shore, and adjacent thickets and scrub (Table A2). The occurrence of T. cristatus was positively linked to proportion of slightly sloping bank, organic mud substrate and adjacent mixed woodland, and negatively linked to fish presence (Table A2). L.
vulgaris was found in insufficient sites to allow a clear evaluation of habitat preferences.

Habitat characteristics of intervention ponds
Once favourable amphibian habitat in the study area was assessed, the

Diversity patterns and community composition in intervention ponds
We recorded 51 colonization events in the 25 intervention ponds over 4 years (Table 2)

DISCUSSION
Our study found a total of 43 colonizations across the 25 intervention ponds during the first and the second breeding season after intervention works, increasing to 51 by 2018. In addition, intervention ponds reached amphibian diversity patterns and community compositions comparable with reference ponds in the study area by the second breeding season after intervention, a generally more rapid increase compared to other restoration projects (e.g. Lesbarrères et al., 2010;Petranka, Harp, Holbrook, & Hamel, 2007;Rannap et al., 2009). We propose that this rapid recovery is mostly due to our application , and the projection, in the same PCA space, of the yearly centroid (mean and SE) for the community composition of the intervention ponds across the study period (e). In graphic b1, we highlight the position occupied by the lakes without amphibians (square), and without amphibians or with B. bufo present only (inverted triangle). In graphic e, we add the two reference categories of amphibians absent, and amphibians absent or B. bufo present only, and the p-value of the permutational multivariate analysis of variance (PERMANOVA) main analysis. Groups of categories with different letters showed significantly different amphibian communities in the post-hoc pairwise PERMANOVA test.
of knowledge of local amphibian habitat preferences as acquired through a quantitative analysis prior to intervention. This has made the project effective in biodiversity terms, and efficient in financial (the total project cost was under £20,000, although we estimate the cost in volunteer time for monitoring the intervention project as being around 400-500 worker-hours, excluding survey of the reference ponds, which had been carried out as a previous project) and land-use terms.
We suggest that the project's success is also due to bringing together co-development and conservation evidence approaches.
Three out of the five landowners had not previously undertaken conservation work on their land, and only one (the state forest managers at FLS) had done so on a formal basis. Participants had different motivations. Land managers wished to be seen as good stewards, protecting habitats and species whose survival represents part of a legacy of traditional land-management systems handed down from their ancestors; one of the families has owned the land since at least the 17th century.
For both farm managers, there was desire not only to integrate conservation with profitable businesses, but to be seen to be doing so by their peers. Indeed, one of the participants joined the project after being convinced by another land manager. Whilst economic considerations were important, neither had participated in formal agri-environmental schemes due to the bureaucracy involved. A major attraction of the current work was the lack of form-filling, or any perception of loss of control over their own land.
The intervention at the golf club was more challenging, and ultimately depended on the relationship built with the club professional and ground staff. Some members were opposed to pond creation because of negative publicity about the presence of great crested newts limiting people's control over their land (Jehle, Thiesmeier, & Foster 2011). However, the club professional, a local man who remembered catching newts as a boy, was able to persuade members that a new water feature would benefit the game, without adverse impacts on site management. As with the farming community, the respect of local stakeholders was vital to implementation.
Government agency staff wanted to protect biodiversity whilst ensuring prudent use of public money. Volunteers were motivated by the desire to protect local species and the challenges of wildlife recording. With any co-development, failure includes the risk not only of wasting money and other resources (such as land removed from production and volunteer surveyors' time), but also of dampening enthusiasm for future biodiversity interventions. These motivations are complimentary but could easily have led to a project based on methods that were inefficient or ineffective. Evidence-based interventions are more likely to be effective as they build on a rigorously compiled knowledge base (Sutherland et al., 2004). However, any expert-based system runs the risk of alienating people living in the area of the intervention, as they are seen as something external being done to them and their land, and failing to consider their values and objectives. This criticism has in the past been levelled at evidence-based conservation (Adams & Sandbrook, 2013). By combining the two methodological approaches, we believe we have avoided their principal drawbacks.

Habitat assessment and design of intervention ponds
Although widespread generalist species like R. temporaria and B. bufo have been well-studied, knowledge of species' habitat needs often relates to their core range and may not be relevant throughout their distribution (Arntzen & Themudo, 2008;Gomez-Mestre & Tejedo, 2003;Zanini, Pallet, & Schmidt, 2009). For example, a previous study of T. cristatus habitat preferences in the Scottish Highlands revealed differences from its core range (Miró et al., 2017). An assessment and analysis of amphibians' local habitat preferences was therefore needed to select potential intervention sites likely to be colonized and then retain breeding populations. Our selection of reference ponds was only partly randomized, and one subset was biased towards T. cristatus presence, including ponds with especially rich amphibian communities rather than a random selection. This should not have markedly affected our inferences, as T. cristatus, while otherwise rare in Scotland, most commonly occurs in ponds in agricultural areas (e.g. Miró et al., 2017;Orchard, Tessa, & Jehle, 2019) and regularly co-occurs with all other investigated amphibians whenever their ranges overlap (e.g. Arntzen, Abrahams, Meilink, Iosif, & Zuiderwijk, 2017;Denoël, Perez, Cornet, & Ficetola, 2013). However, the corollary is that we have compared our intervention ponds with a set of reference ponds with higher than average amphibian diversity, making the rapid establishment of rich communities more remarkable.
Colonization by T. cristatus was seen as a particular success. The species has special protection under European and UK law, and prior to the intervention was found in only 44 ponds in the region. B. bufo was not a specific target of the project, although it colonized two ponds: this species is locally typical of large lochs of glacial origin and artificial water bodies created for fishing. It is common and often abundant in these habitats, and there is no evidence of any local declines (McInerny & Minting, 2016).
Pond design needed to both meet amphibians' ecological needs and complement local hydrology. Land-managers provided hydrological knowledge, based on experience of which areas were most likely to be waterlogged. The habitat needs revealed through our study, for example a large proportion of gently sloping banks, were easily incorporated by the construction operative. Sloping banks create shallow margins where water warms quickly, thus speeding larval development and, later, offer easy egress for metamorphs (e.g. Parris, 2006; see also Shulse et al., 2012 for constructed ponds). Such banks also make ponds less dangerous for livestock and humans, an important safety consideration on land managed for agriculture or where recreational access is likely.
Fish presence had been identified as an important negative factor for all local amphibians except B. bufo. Outside of flood plains, small water bodies in the region tend to be fishless, with human introduction seemingly the most common means of spread (Maitland, 1977). Whilst all stakeholders are aware of the need to keep ponds fish-free, introduction of fish by others, particularly in public access lands remains a risk. For some of these ponds, we have tried to keep visitors at a distance through allowing development of dense vegetation such as the spikey shrub Ulex europaeus between the pond margin and adjacent paths. In the long term, succession to wet woodland may become an issue. Subject to licensing, this may potentially be mitigated by reexcavating ponds on a rotational basis, thus keeping a variety of successional stages within the pondscape at each site which offer a variety of habitats.
Of the 20 newly created ponds, only one has not been colonized by amphibians and is unlikely to be suitable for them due to miscommunication during construction. Forestry staff assumed that lying deadwood is often beneficial for wildlife and thus piled large quantities of brash along the margins. However, the species in question (the non-native Sitka spruce Picea sitchensis) produces highly acidic runoff, which resulted in low pH, sphagnum-dominated waters. Removing the brash would be logistically challenging, with an adverse impact on the acidophilous plants and invertebrates which have colonized the pond, potentially including the nationally endangered dragonfly Leucorrhinia dubia (unpublished data).
Not all characteristics of intervention ponds were more favourable for specific amphibians than the characteristics of reference ponds (for a similar finding, see, e.g. Korfel et al., 2010), although some of these will improve with time as the ponds mature. As the ponds were excavated with a digger, substrates are largely composed of mineral soils. Proportions of organic mud should increase through natural processes over time. Terrestrial habitat diversity can also be expected to increase, particularly in the forestry sites. Prior to construction, nine of the 13 forest sites (site codes beginning 'B' in Table A1 in the Supporting Information) were areas of clear-fell following harvesting of non-native Sitka spruce (Picea sitchensis). Regeneration is already taking place, with a broader variety of habitats developing, for example temperate thickets and scrub (EUNIS code F3.1), mixed Pinus sylvestris -Betula woodland (G4.4), broadleaved deciduous woodland (G1), Pinus sylvestris woodland (G3.4), along with surface running waters (C2), mires, bogs and fens (D). The farmland sites have seen less of a change as they were already adjacent to habitats that we had determined to favour amphibians, and the pond restorations have led to a more natural ecotone transition from grassland through mire to standing water.

Diversity and community composition in the intervention ponds
Amphibians now breed in 24 out of 25 (96%) intervention ponds compared with 83% of reference ponds. This compares favourably to a recent review, which found that only five out of nine studies showed similar or higher numbers of species and reproductive activities in created ponds relative to natural ponds (occupancy of created ponds was 64-100%, with reproductive activity in 64-68% of them; Smith et al., 2019). All colonization events were within 600 m of known populations, with the exception of one T. cristatus colonization at 1840 m from the nearest known pond. This is further than the known colonization range for the species (Haubrock & Altrichter, 2016), and may have originated from an undocumented source, as there are unsurveyed ponds nearby.
The dominant species in our intervention ponds were R. temporaria (23 ponds, 92%) and L. helveticus (20 ponds, 80%; see Table A3 in the Supporting Information), both representing a higher occupancy than that documented for amphibian ponds across Scotland (70% and 42%, respectively (Table A4 in Appendix A in the Supporting Information; Wilkinson & Arnell, 2013). The proportions of ponds with breeding L. vulgaris and T. cristatus were both an order of magnitude higher than in the Scotland-wide survey (Wilkinson & Arnell, 2013; Table A3 in the Supporting Information). However, both species have very restricted ranges in Scotland and so conclusions from such a comparison are limited. B. bufo's low occupancy rate (two ponds, 8% relative to 35% of Scottish ponds, Wilkinson & Arnell, 2013) was unsurprising, as the species has different habitat preferences to the other native amphibians; typically breeding in larger, deeper ponds where fish are present (Minting, 2016). Indeed, Harper et al. (2020) found a negative association between T. cristatus and B. bufo in a study of over 500 ponds in England. Overall amphibian species richness was also greater in intervention ponds than in the reference group (Table A4 in the Supporting Information), despite the high proportion of the latter known to hold the European protected species T. cristatus (Miró et al., 2017). Thus, our comparison was with high biodiversity value ponds, rather than with already degraded ecosystems. A random sample would likely have included species-depleted ponds and possibly would have led to a failure to identify a shifting baseline resulting from biodiversity losses that occurred in past generations (Papworth, Rist, Coad, & Milner-Gulland, 2009).
The lower beta-diversity of the intervention ponds was expected, as they were constructed to be within a perceived ideal size and geological range for amphibians, rather than representing the wider range of natural ponds. That said, medium-sized ponds appear to be the sizeclass most likely to have been lost in the last hundred years or so (Wood, Greenwood, & Agnew, 2003) and confirmed by our review of historic maps. Indeed, small ponds (<400 m 2 ) may even have increased in number through the creation of garden ponds (Banks & Laverick, 1986;Williams et al., 2007) and, at the other end of the scale, large water bodies tend not to be drained, despite often being heavily modified for power generation or water abstraction. Although created for amphibians, a pondscape featuring medium-sized ponds may host a range of species with similar habitats preferences and would benefit from detailed study of their fauna and flora (see e.g. Sayer et al., 2012).

4.3
The future A key factor in any conservation intervention is its long-term stability. While agri-environmental schemes can deliver benefits, they can be inflexible and include constraints which make them unsustainable after the initial funding period (Burton & Paragahawewa, 2011). Our development approach allowed land managers to allocate land that was unproductive, or where a pond would be an asset. Ponds can add to the aesthetic appeal of locations (Milburn, Brown, & Mulley, 2010), encouraging tourism (two of the sites offer cabin rentals), or be used for hunting water birds. Two farmers neighbouring the project sites have subsequently constructed ponds for amphibians without any incentivization (DOB, personal observation). Globally, not all amphibians breed in ponds, however, we believe that the approach we have used is equally applicable to other habitat types.
The project, although highly successful, will need continued monitoring and adaptive management to manage habitat changes that might threaten future amphibian persistence in the intervention ponds (e.g. Petranka et al., 2007). Examples of potential changes include macrophyte surface coverage (Fardell et al., 2018), shore tree coverage/shading (Oldham, Keeble, Swan, & Jeffcote, 2000), emergence of shrubs leading to long-term pond loss (Erõs, Maloș, Horváth, & Hartel, 2020), frequency of desiccation/ hydroperiod (Swartz, Lowe, Muths, & Hossack, 2020) and fish introductions (Hazell, Hero, Lindenmayer, & Cunningham, 2004). While land managers are aware of the need to avoid such developments, ongoing monitoring is carried out by volunteers from the local citizen science group HARP. At the date of submission (2020), management at all 25 ponds and their surrounding terrestrial habitats remain appropriate for wildlife conservation and all landowners remain committed to the project.

ACKNOWLEDGMENTS
We The project was funded by FLS and SNH/NatureScot. Survey work was carried out under licence 31980.

AUTHORS' CONTRIBUTIONS
All authors conceived the ideas and designed methodology; all authors took part in field work including data collection; DOB and KOB led on pre-intervention surveys and project management; AM analysed the data; DOB and JH led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

DATA AVAILABILITY STATEMENT
All data are available through the Dryad Digital Repository https://doi. org/10.5061/dryad.s1rn8pk6d (O'Brien et al., 2020). Original species records, including records of non-target species are also publicly available via the National Biodiversity Network (https://nbnatlas.org/).