Waterbirds increase more rapidly in Ramsar-designated wetlands than in unprotected wetlands

Authors


Summary

  1. There is a general lack of information on how international conservation treaties affect biodiversity. The Ramsar convention on the protection of internationally important wetlands is such an international conservation policy. It initiated the world-wide establishment of over 2000 protected areas currently covering more than 200 million ha. The convention came into force in 1975, but to date, it remains unknown whether it actually produces biodiversity benefits.
  2. We analysed 21 years of survey data from a wide range of waterbird species in over 200 Moroccan wetlands and examined whether Ramsar designation as well as a national protected areas programme Sites d'IntérêtBiologique et Ecologique (SIBE) positively affected bird abundance. Furthermore, habitat characteristics of wetlands in protected areas and control sites were compared and bird abundance was related to habitat characteristics.
  3. After designation, waterbird species richness and abundance increased more rapidly in Ramsar wetlands than in non-designated wetlands. In SIBE sites, increases in bird abundance were intermediate.
  4. Waterbird abundance was significantly related to habitat characteristics, most importantly cover of water or bare ground. Compared to control sites, Ramsar sites had significantly higher cover of habitat types favoured by most waterbird species. It remained unclear, however, whether these differences were caused by conservation management or were already present prior to conservation designation.
  5. Surprisingly, waterbird population trends in Moroccan wetlands were markedly positive. Trends were found to be related to precipitation levels in the Sahel which may have caused changes in migratory strategies.
  6. Synthesis and applications. This study demonstrates a powerful approach to systematically evaluating biodiversity responses to major international conservation policies. The results highlight that data on management and habitat quality are critical for reaching general conclusions about the effectiveness of conservation instruments. Site managers or conservation authorities should be encouraged to collect standardized data on conservation actions to help interpret the mechanisms behind population responses and thus extrapolate findings beyond the study system. These findings represent a first small step towards unravelling the contribution of international conservation tools towards global policy objectives of halting the biodiversity decline.

Introduction

International conservation treaties designating sites or species as protected are important instruments to halt the on-going decline in biodiversity. Yet there is little evidence available on the effectiveness of such instruments in protecting biodiversity (Kleijn et al. 2011; but see Donald et al. 2007; Hiley et al. 2013). Effective conservation requires a good understanding of the key drivers of ecosystem change because management needs to address the factors having the most negative impact on biodiversity (Williams et al. 2009). Furthermore, ecosystems are rarely stable as they are subject to changing environmental conditions driven by, for example, climate change or human population pressure. Changing environmental conditions may change the effectiveness of conservation management (Kleijn et al. 2010). Monitoring and evaluation is therefore pivotal to effective conservation (Kleijn & Sutherland 2003), and failure to evaluate can lead to the implementation of ineffective conservation practices (Sutherland et al. 2004).

The Ramsar Convention came into force in 1975 and targets wetlands of international importance (Ramsar Convention Bureau 2002). The implementation of the convention has been very successful, and currently, 165 countries have signed the convention, which now lists 2161 designated sites with a total surface area of 205 682 155 ha (September 2013; Ramsar 2013); an area equivalent to the size of Mexico. Yet, we know very little about how populations are changing over time in Ramsar sites after conservation designation, and how this relates to trends in unprotected areas. A systematic assessment of the biodiversity effects of the Ramsar convention has never been made (Castro, Chomitz & Thomas 2002), and almost four decades after the convention came into force, it is still unknown whether this conservation tool actually protects or promotes biodiversity.

Here, we use a unique 21-year data set of Moroccan wetland surveys to evaluate the response of waterbirds to Ramsar designation. In Morocco, the first four Ramsar sites were designated in 1980, prior to the start of the waterbird surveys, but in January 2005, a further 20 sites were designated allowing us to compare waterbird numbers in sites with and without Ramsar designation before and after designation took place. Such a replicated Before-After-Control-Implementation (BACI) approach is an effective way to evaluate the impact of conservation management (Underwood 1994) but, to our knowledge, has never been used to examine biodiversity responses to the establishment of protected areas.

In most countries, additional conservation policies operate (e.g. Natura 2000 in the EU, Conservation Reserve Program in the USA) that may augment or interact with Ramsar designation. In Morocco, Sites of Biological and Ecological Interest (‘Sites d'Intérêt Biologique et Ecologique’, SIBE) were established in 1996 as a national priority for biodiversity conservation and promotion of ecotourism (Fishpool & Evans 2001). All 24 Ramsar sites are also designated as SIBE (for simplicity, in this paper these sites will be referred to as Ramsar sites); however, only 54% and 41% of Ramsar and SIBE wetlands, respectively, are located in formal protected areas (e.g. Ramsar 2013). Sites may therefore be covered by multiple conservation policies but, since Ramsar designation is not based on a regulatory regime and has no punitive sanctions for violations of treaty commitments (Adaman, Hakyemez & Oëzkaynak 2009), at the same time do not benefit from any legal form of protection. For comparative purposes, we therefore include SIBE designation in the study and compare sites with Ramsar or SIBE designation with control sites; non-designated wetlands do not even have informal protection against wildlife threats such as hunting, habitat degradation or destruction. We hypothesize that, for conservation designation to be more than just a sign at the edge of a wetland, differences in biodiversity components between designated and non-designated wetlands should be quantifiable and examine whether species richness and abundance of waterbirds are affected by conservation designation in Moroccan wetlands.

The creation of protected areas generally does not target species directly but aims to indirectly protect species by preventing destruction of their habitat and removal of threats to either the habitat (e.g. overgrazing, reed cutting, pollution) or the species (e.g. hunting). Analysing the population response to the implemented conservation practices can therefore help us identify the most effective management strategies. However, data on conservation activities that have actually been implemented in sites are rarely collected in a systematic way in protected areas and was unavailable for the Moroccan study sites. Nevertheless, to obtain insight into the drivers of differences in bird species richness and abundance between sites, we surveyed wetland habitat characteristics in a single year and examined whether associations between waterbirds and habitat characteristics could explain potential effects of conservation designation on waterbirds.

Materials and methods

Birds are a species group specifically targeted by the Ramsar Convention. Wetlands are classified as being of international importance if they regularly support 20 000 or more waterbirds or if they regularly support 1% of the individuals of a population of waterbird species or subspecies (Ramsar Convention Bureau 2002). Morocco is an important stopover and wintering site for many western Palearctic migratory waterbird species. Birds crossing the Mediterranean at the Gibraltar channel often follow the Moroccan Atlantic coast line and use the many wetlands as staging sites thus largely circumventing the Saharan Desert (e.g. Delaney et al. 2009; Vardanis et al. 2011), while other species use Moroccan wetlands as wintering grounds. Additionally, these sites offer a year-round home to many waterbirds including a number of endangered species such as bald ibis Geronticus eremita, marbled teal Marmaronetta angustirostris and white-headed duck Oxyura leucocephala.

Data

We used published data from waterbird surveys in Moroccan wetlands carried out between 1985 and 2000 by the Scientific Institute of the Mohammed V University in Rabat (e.g. Dakki et al. 2002; for a complete list see Appendix S1 in Supporting Information). Additionally, for the period 2007–2011, survey data collected by one of the co-authors (IC) were used. All surveys were carried out by experienced ornithologists, during the wet season, in the Palearctic winter period (November–February) with the majority of the surveys in January in line with the International Waterbird Census (IWC; Wetlands International 2010). Count plots (‘sites’) generally consisted of clearly delineated wetlands (e.g. lakes, estuaries, bays) that could easily be counted in a standardized way in successive years. Many large wetlands consisted of complexes of distinct habitats such as marshes, salines or lakes, which were often surveyed separately. In each site, the number of individuals of all observed waterbird species was recorded using binoculars and telescopes. Zero counts were given to all waterbird species that were not observed. The number and identity of surveyed sites varied between years. The complete data set contained 1533 surveys carried out in 209 different sites (Fig. S1, Supporting information). The number of times each site was surveyed ranged from 1 to 40. Sites were located between 22·9831 and 35·7206 decimal degrees latitude and −2·2000 and −16·1726 longitude. More than 200 different bird species were observed but not all species were waterbirds, and only 105 waterbird species were observed regularly enough (at least five surveys and ten individuals) to be useful for the purpose of our study. These species were from eight orders and 21 families and include both migratory and sedentary species (Table S1, Supporting information). Seven regularly observed species have an unfavourable conservation status (IUCN 2013).

Study design

To examine whether Ramsar or SIBE designation affected bird abundance and population trends, we collected information on the type of designation of the wetland in which each count plot was located. This information was obtained for 109 count plots that had been surveyed in the winter of 2010–11. Forty-one of these count plots were located within Ramsar wetlands, 33 sites were only designated as SIBE and 35 sites had no conservation designation (Fig. S1, Supporting information). For reference, Birdlife International has identified Important Bird Areas (IBA) in Morocco (Fishpool & Evans 2001). In our selection of study sites, IBA designation always coincided with Ramsar and/or SIBE designation. Sites without designation were therefore not in any way identified as being of special value for wetland habitats or birds.

Plot size and habitat characteristics

To correct for differences in plot size and to relate bird abundance to habitat characteristics, plot size and cover of different habitat types were estimated between 14 November and 16 December 2010 during a waterbird survey in 109 sites across Morocco. The delineation of each count plot was marked on maps after which count plot size was estimated using GIS (ArcMap 10·0, ESRI, Redlands, USA). Additionally, the percentage surface area covered by (i) water, (ii) tall herbaceous vegetation (e.g. reedbeds, tall sedges), (iii) short herbaceous vegetation (e.g. saltmarshes, wet meadows, Salicornia vegetation), (iv) bare ground (e.g. sandy beaches, mud flats, rocky outcrops), (v) forests, (vi) agricultural land and (vii) built-up areas was estimated in each plot.

Analyses

For analyses at the species level, we used data from 51 waterbird species that had been recorded in at least 25 sites and 8 years. The remaining 54 species were included in species richness estimates or when species with unfavourable conservation status were considered as a group (Table S1, Supporting information). In some winter periods, multiple surveys were available for the same site. The total number of observed waterbirds did not differ in any systematic way between months (F3,1290 = 0·68, = 0·564). We therefore determined the average number of waterbirds observed per site per winter period. For most species, the distribution of the observed counts was heavily skewed with maximum counts of up to 45 460 birds (Table S2, Supporting information). The usual Poisson assumption for analysing bird counts is then not appropriate. In all analyses described below, we therefore used log(bird abundance+1) data and used standard regression models assuming normal error distribution. Model outcomes were checked by plotting residuals versus fitted values, confirming that assumptions were met acceptably.

Ramsar and SIBE sites have been designated for their high biodiversity levels, so straightforward analyses comparing bird abundance in sites with conservation designation with control sites cannot be used to infer impacts of conservation designation. Analyses examining the impact of conservation status on bird abundance were therefore restricted to the 20 Ramsar sites that had been designated in 2005. For these sites, we analysed whether the mean difference in bird abundance between sites with conservation designation and control sites was higher after designation (years 2007–2011) than before (1985–2000; i.e. whether after designation, the change in bird abundance was more positive in sites with conservation designation than in control sites). Because the data set was unbalanced, we used linear mixed models to correct for site and year differences. These models included site as a random variable, count plot size as a fixed variable and year as a fixed factor. Additionally, conservation designation, period (before and after designation) and their interaction were included as fixed factors with the interaction term being indicative of conservation effects of designation. This part of the study focussed primarily on the effects of Ramsar designation, with the SIBE sites that were designated in 1996 serving mainly as reference. Differences between means were tested using t-tests.

When analysing the response of the 51 individual species, we chose not to correct for multiple testing. Correction reduces type I error, but tends to inflate type II error (Field et al. 2004). Instead, we critically interpret statistical results of analyses with individual species. To investigate the effect of conservation status, we examined whether the change in standardized population abundance, averaged over all species, was significantly affected by conservation designation. Analysis was carried out using anova with conservation designation, period and their interaction as explanatory factors and using the individual species as a blocking factor (Kleijn et al. 2008). Standardized population abundances were obtained by predictions of individual species models that correct for site and year differences.

Habitat quality is known to affect population trends of migratory waterbirds (Gill et al. 2001). We therefore checked whether responses in bird numbers were affected by intrinsic differences in habitat quality by testing whether, prior to Ramsar designation, per site bird population trend was related to per site mean bird abundance. This approach assumes that waterbird abundance reflects habitat quality. This is not always the case (van Horne 1983) but since the abundance of staging waterbirds generally reflects abundance in food supply (e.g. Norris, Bannister & Walker 1998; Gill, Sutherland & Norris 2001), it seems to be a reasonable assumption here. For each site and species combination for which we had at least five observations of which at least two were positive, we calculated mean population size and population trend (slope of the relationship between log(bird abundance+1) and year). We subsequently analysed for each species the relationship between population trend and mean population size using the different sites as experimental units.

To get insight into the factors underlying differences in waterbird abundances between sites, we determined which habitat variables were most strongly related to waterbird abundance and whether habitat characteristics varied systematically between sites with different conservation status. A more direct approach of including habitat characteristics into the analyses relating bird abundance to conservation status of sites was not possible because bird data covered the period 1985–2011, while habitat data were only available for the year 2011. We determined which habitat variables were most strongly related to waterbird abundance in 2011 using an Akaike Information Criterion framework (adjusted for small sample sizes, AICc; Burnham & Anderson 2002). Because the seven habitat variables sum to 100%, the habitat variable ‘built-up areas’ was excluded from analyses to avoid aliasing with the other variables. For each of the 51 bird species, we calculated Akaike weights for the 63 models containing all possible subsets of the remaining six habitat variables (Burnham & Anderson 2002; Crampton et al. 2011). To estimate the relative importance of variables in explaining bird abundance, the Akaike weights for each model containing the variable of interest were summed. To assess whether variables were, on average, positively or negatively related to bird abundance, we calculated weighted model-averaged estimates for each variable. Differences in cover of habitat types between sites with different conservation designation were analysed using generalized linear models assuming a binomial error distribution and a logit-link function.

All models were fitted using standard facilities in GenStat (Payne et al. 2002).

Results

On average, the individual waterbird species selected for analysis were observed in 374 surveys or 24% of the total number of surveys (Table S2, Supporting information). The mean number of observed birds per species per survey ranged from 0·1 (common kingfisher Alcedo atthis) to 504 (European coot Fulica atra) with an average of 55. The maximum number of observed birds per species varied between 7 and 45 460. During the study period, the mean number of observed waterbirds per site increased considerably, with low numbers during the 1980s, a rapid increase during the 1990s and a more moderate increase in the 2000s (Fig. 1).

Figure 1.

Trends in the total number of waterbirds observed per site (bars) and the amount of rainfall in the Sahel in the preceding rainy season (line). The trend in annual waterbird abundance is based on 1533 surveys in 209 sites. Effects of different sites being surveyed in different years have been corrected for using generalized linear models relating log(abundance+1) to year as explanatory factor and site as correcting factor. Waterbird abundance therefore represents the predicted back-transformed mean abundance per site per year for the average site. Rainfall anomalies indicate the difference with the 1900–2011 mean for the months June–October, the rainy season in the Sahel zone (

source: JISAO, http://jisao.washington.edu/data/sahel/#values

).

After designation, the number of species observed per site increased more rapidly in Ramsar and SIBE sites than in non-designated sites (interaction designation.period F2,858·9 = 19·99, < 0·001; Fig. 2a). Species richness increased by 33% in non-designated sites against 80% and 84% in SIBE and Ramsar sites, respectively. Before designation, species richness was already higher in Ramsar sites although not significantly at α = 0·05 (t95·85 = 1·968, = 0·052). In the period after designation, the difference in species richness between non-designated and Ramsar sites was highly significant (t95·85 = 6·852, < 0·001). The contrast with non-designated sites before and after designation was even larger for SIBE sites (before: t95·85 = 0·093, = 0·923; after: t95·85 = 2·917, = 0·004).

Figure 2.

Differences in (a) species richness and (b) abundance of waterbirds between sites with no conservation designation, Sites of Biological and Ecological Interest as identified by Moroccan authorities (SIBE) or Moroccan wetlands of international importance under the Ramsar convention (Ramsar) before and after conservation designation. Species richness indicates the mean number of species per site of a total of 105 regularly observed waterbirds. Abundance is expressed as the mean (± SE) log10-transformed abundance per site of 51 individual species that were analysed in detail (Table S3, Supporting information). Before designation: 1985–2000; After designation: 2007–2011.

A similar pattern was observed for the mean abundance of individual waterbird species (Fig. 2b, Table S3, Supporting information). Averaged over the 51 examined species, after designation of Ramsar sites, waterbird abundance increased more rapidly in Ramsar (98%) and SIBE wetlands (108%) than in non-designated wetlands (27%; interaction designation.period F2,250 = 11·72, < 0·001). Already before designation, waterbird abundance was significantly higher in Ramsar sites than in non-designated sites (t250 = 2·206, = 0·028), but after designation, this contrast became much more pronounced (t250 = 8·997, < 0·001). Before 2001, waterbird abundance in SIBE sites was not significantly different from that in non-designated sites (t250 = 1·078, = 0·282), but after the year 2006, SIBE sites supported significantly larger numbers of waterbirds (t250 = 3·075, = 0·002).

The positive response of waterbirds after Ramsar designation was mainly caused by wader, gull and tern species (Fig. 3, Table S3, Supporting information). Compared to non-designated sites, wader and gull abundance showed a more pronounced increase in Ramsar and SIBE sites after designation of Ramsar sites in 2005 (interaction designation.period waders: F2,30 = 5·87, = 0·007; gulls and terns: F2,100 = 16·21, < 0·001). Before Ramsar designation, wader abundance in future Ramsar sites or in SIBE sites was not significantly different from that in non-designated sites (respectively, t100 = 1·419, = 0·159 and t100 = 0·044, P = 0·965). After designation, wader abundance was significantly higher in sites with both types of conservation designation than in non-designated sites (Ramsar: t100 = 9·379, < 0·001; SIBE = t100 = 2·975, = 0·004). Before designation of Ramsar sites, gulls and terns were present in (marginally) significantly higher numbers in future Ramsar sites than in non-designated sites (t30 = 2·126, = 0·042), and after designation, this difference was much more pronounced (t30 = 6·216, < 0·001). Gull and tern abundance in SIBE sites was not affected by Ramsar designation. Relative abundance of ducks in sites with different conservation was significantly different in the two examined periods (interaction designation.period ducks: F2,40 = 7·43, = 0·002); however, this was primarily due to SIBE sites. Before the year 2005, duck abundance was significantly lower in SIBE sites compared to non-designated sites (t40 = 2·353, = 0·024), after 2005, the reverse was the case (t40 = 3·096, = 0·004). Birds from other orders showed similar increases in the three types of sites (interaction designation period other species F2,65 = 0·21, = 0·808).

Figure 3.

Differences in abundance of four functional groups of waterbirds between sites with no conservation designation, Sites of Biological and Ecological Interest as identified by Moroccan authorities (SIBE) or Moroccan wetlands of international importance under the Ramsar convention (Ramsar) before and after conservation designation. Abundance is expressed as the mean (± SE) log10-transformed abundance per site of the number of individual species that were analysed in detail (indicated between brackets in the panels). The species included in each functional group are indicated in Table S1 (Supporting information).

The abundance of species with unfavourable conservation status was not significantly affected by Ramsar designation (interaction designation.period F2,876·3 = 2·66, P = 0·071). Before and after 2005, their abundance was significantly higher in Ramsar sites than in non-designated sites (respectively, t93·24 = 3·252, = 0·002 and t93·24 = 5·151, < 0·001; Fig. S2, Supporting information). Abundance of waterbirds with unfavourable conservation status was not affected by SIBE designation in either period (before: t93·24 = 0·342, = 0·733; after t93·24 = 1·524, = 0·131).

Prior to Ramsar designation, population trends of nine species were significantly positively related to mean population abundance (Table S4, Supporting information). However, averaged over all species, the slope of this relationship was close to zero (r = 0·022). Furthermore, no clear difference in response between species groups was observed (Fig. S3, Supporting information), while positive responses to Ramsar designation were restricted to waders and gulls (Fig. 3).

Count plots in Ramsar wetlands had lower proportions of water and built-up areas and higher proportions of short herbaceous vegetation and bare ground than count plots in non-designated wetlands (Fig. 4, Table S5, Supporting information). Proportional cover of water and bare soil were generally the most important variables explaining waterbird abundance in Moroccan wetlands (Table 1). Water cover was mostly negatively related to bird abundance. For example, best models indicated that only ruddy shelduck Tadorna ferruginea and tufted duck Aythya fuligula were significantly positively related to percentage water, all other significant relationships were negative (Table S6, Supporting information). Bare soil was both positively and negatively related to a considerable number of species. Gulls, terns and plovers were generally positively related, and ducks and grebes were mostly negatively related to bare soil. Short and tall herbaceous vegetation, although less important as explanatory variables, were, when significant, exclusively positively related to waterbird abundance (Table 1, Table S6, Supporting information).

Table 1. The relative importance (summed parameter weights) of different habitat types in explaining the abundance of waterbirds in Morocco. Analyses are based on data from 109 sites surveyed in winter 2011 for both birds and habitat cover. Weights represent the combined support for all models containing a given habitat type as an explanatory variable and range between 0 and 1. Parameter weights close to 1 indicate that this variable was included in all well-supported models. Low parameter weights suggest lower support for models containing these variables as predictors. Signs indicate whether on average the relationship was positive or negative as indicated by the model average parameter estimates. Bold figures indicate for each species the habitat variable with the highest summed parameter weight
Scientific nameEnglish nameWaterTall herbaceousShort herbaceousBare groundForestAgricultural
Tachybaptus ruficollis Little grebe−0·820·550·50 −0·85 −0·420·53
Podiceps cristatus Great crested grebe0·420·320·40 −0·81 −0·31−0·34
Phalacrocorax carbo Great cormorant−0·80−0·340·42 0·94 0·53−0·35
Ardeola ralloides Squacco heron−0·52 0·91 0·85−0·50−0·400·87
Egretta garzetta Little egret −0·99 0·330·34−0·790·300·32
Ardea cinerea Grey heron −0·82 0·500·52−0·79−0·430·65
Ciconia ciconia White stork−0·700·570·61−0·70−0·43 0·88
Platalea leucorodia Eurasian spoonbill −0·93 0·870·47−0·470·390·35
Phoenicopterus ruber Greater flamingo −0·77 0·510·450·480·29−0·47
Tadorna ferruginea Ruddy shelduck 0·61 −0·290·45−0·46−0·29−0·51
Anas penelope Eurasian wigeon0·320·270·36−0·34 −0·52 −0·28
Anas platyrhynchos Mallard−0·720·450·44 −0·85 −0·330·43
Marmaronetta angustirostris Marbled teal−0·74 0·98 0·520·48−0·29−0·32
Anas acuta Northern pintail −0·67 0·590·400·400·28−0·34
Anas clypeata Northern shoveler−0·460·73 0·79 −0·48−0·390·39
Anas crecca Common teal−0·63 0·76 0·63−0·61−0·340·53
Aythya ferina Common pochard0·400·450·37 −0·80 −0·720·31
Aythya fuligula Tufted duck0·570·35−0·32 −0·64 −0·31−0·30
Circus aeruginosus Marsh harrier −0·79 0·530·58−0·790·430·65
Pandion haliaetus Osprey −0·64 0·320·340·58−0·35−0·29
Gallinula chloropus Common moorhen −0·83 0·66−0·65−0·820·410·54
Fulica atra Common coot−0·420·43−0·33 −0·94 0·300·40
Haematopus ostralegus Eurasian oystercatcher−0·68−0·490·72 0·95 0·750·53
Burhinus oedicnemus Stone-curlew0·760·700·670·71 0·89 0·67
Himantopus himantopus Black-winged stilt −0·94 −0·360·45−0·840·300·36
Recurvirostra avosetta Avocet −0·91 0·330·62−0·41−0·28−0·32
Charadrius hiaticula Greater ringed plover −0·95 −0·390·340·53−0·330·31
Charadrius dubius Little ringed plover−0·30−0·540·310·30 0·56 0·39
Charardrius alexandrinus Kentish plover−0·60−0·350·69 0·94 0·320·35
Pluvialis squatarola Grey plover−0·67−0·370·76 0·98 −0·32−0·38
Arenaria interpres Ruddy turnstone−0·77−0·340·45 0·99 0·54−0·34
Calidris alba Sanderling−0·58−0·370·71 0·97 0·87−0·39
Calidris minuta Little stint −0·91 −0·460·480·780·32−0·39
Philomachus pugnax Ruff−0·680·580·74 −0·68 −0·540·64
Numenius arquata Eurasian curlew −0·77 0·410·700·51−0·570·38
Numenius phaeopus Whimbrel−0·55−0·470·81 0·92 0·420·71
Limosa limosa Black-tailed godwit −0·78 0·400·77−0·58−0·350·48
Tringa totanus Common redshank −0·91 −0·300·520·37−0·28−0·31
Tringa nebularia Common greenshank −0·80 −0·380·53−0·600·750·69
Tringa glareola Wood sandpiper−0·420·34 0·70 −0·440·280·34
Actitis hypoleucos Common sandpiper−0·470·41−0·31 0·71 −0·30−0·29
Tringa ochropus Green sandpiper−0·750·520·51 −0·89 −0·32−0·37
Gallinago gallinago Common snipe −0·87 0·380·51−0·75−0·330·56
Larus ridibundus Black-headed gull −0·92 −0·390·340·36−0·790·40
Larus audouinii Audouin's gull−0·640·370·60 0·95 0·93−0·37
Larus michahellis Yellow-legged gulll−0·76−0·370·42 0·95 0·410·38
Larus fuscus Lesser black-backed gull−0·89−0·48−0·42 0·91 −0·32−0·37
Gelochelidon nilotica Gull-billed tern −0·40 −0·290·400·310·280·30
Sterna sandvicensis Sandwich tern−0·73−0·400·54 0·98 0·350·36
Sterna caspia Caspian tern0·350·310·81 0·99 0·280·31
Alcedo atthis Common kingfisher−0·81 0·82 −0·55−0·690·380·56
Figure 4.

Differences in habitat composition between sites with different conservation status in Morocco. Bars indicate means ± standard errors. Ramsar: sites designated as wetlands of international importance under the Ramsar convention; SIBE: Sites of Biological and Ecological Interest as identified by Moroccan authorities. Asterisks indicate significant difference with non-designated sites: *< 0·05, ***< 0·001, see Table S5 (Supporting information) for exact test statistics.

Discussion

Conservation generally targets biodiversity hotspots. It is cost-efficient to concentrate limited conservation budgets to locations that hold high densities of species and individuals (Myers et al. 2000). This does, however, pose serious challenges for studies that aim to evaluate the ecological impact of conservation because it makes often-used space for time designs (e.g. Kleijn et al. 2006) inappropriate. Baseline data of biodiversity levels prior to designation are rarely available for protected areas, and suitable control sites are difficult to find. This may explain why studies that systematically evaluate the ecological effects of conservation strategies such as protected areas are scarce. In this study, we took advantage of an existing long-term data set of waterbird communities in well-defined habitats with contrasting conservation strategies. Using a replicated BACI approach, we showed that, after designation, waterbird species richness and abundance increased more rapidly in Ramsar wetlands than in non-designated control wetlands.

These results could indicate that conservation initiatives taken after designation successfully improved habitat conditions for waterbirds in Ramsar sites relative to unprotected sites and thus that Ramsar designation produced conservation benefits. However, lack of information on the management practices implemented and the protective measures taken in the studied wetlands make it difficult to draw firm conclusions about the effects of Ramsar on waterbirds. Ramsar designation specifically targets wetlands with high biodiversity levels. Prior to designation, Ramsar wetlands indeed hosted larger waterbird populations which could indicate higher initial habitat quality. A possible alternative explanation could therefore be that Ramsar sites are intrinsically higher quality and that, when populations rise for any reason, they rise more rapidly at higher-quality sites. On the other hand, we found no systematic evidence that, prior to designation, population trends of waterbirds were more positive in sites supporting higher population densities than in sites supporting lower population densities. The species group showing the most pronounced response to Ramsar designation, waders, actually showed the weakest relationship between population trend and initial population size (Fig. S3, Supporting information). It is therefore unlikely that the positive effects of Ramsar designation are explained solely by a more pronounced increase in waterbird populations in sites with higher initial densities. Differences in initial site quality furthermore could not explain the more rapid increase in waterbird population numbers and species richness in SIBE wetlands since these sites had densities similar to control sites prior to designation, suggesting that they were of similar quality. It is thus likely that the positive response of waterbirds is at least partially related to activities associated with conservation designation.

Conservation designation of Moroccan wetlands did not eliminate all threats to biodiversity. In Ramsar sites, the most important threats observed during the surveys were overgrazing, hunting, tourism and reed cutting (IC, personal observations). Other studies similarly find that the establishment of protected areas does not remove all threats from habitats. For example, Buchanan et al. (2009) found that African Important Bird Areas lying inside protected areas were more threatened by overgrazing than those lying outside protected areas. In this study, Ramsar sites were covered by a higher proportion of habitat types that were preferred by waterbirds (Fig. 4, Table 1). Habitat type and vegetation structure are known to be important determinants of abundance of migratory birds (Deppe & Rotenberry 2008) but, as far as we know, no other studies exist that link population responses after conservation designation to differences in habitat quality. Ramsar sites had a significantly higher cover of short herbaceous vegetation and bare soil, the habitat types preferred by wader, gull and tern species. The fact that these species groups showed the most pronounced positive response to Ramsar designation suggests that this might be a key factor explaining the higher bird numbers in Ramsar wetlands after designation. Unfortunately, the general lack of data on conservation management actions makes it impossible to determine whether the higher cover of habitats preferred by waterbirds was the result of Ramsar designation or whether habitat differences were already in place before conservation designation and (partly) drove the observed responses of waterbirds.

Wetlands that were designated as SIBE in 1996 showed a response that was intermediate to those observed in Ramsar and non-designated sites. In the period 1985–2000, bird abundance in SIBE sites was similar to that in non-designated sites, but in the period 2007–2011, numbers were significantly higher in SIBE wetlands than in control wetlands. In line with this, habitat characteristics at the end of the study period were intermediate between Ramsar and non-designated sites with, compared to non-designated sites, significantly higher cover of the preferred habitat types short herbaceous vegetation and bare soil. The SIBE programme appears to be successful in conserving waterbirds, and the two key wetland conservation policies implemented in Morocco seem to supplement each other, with Ramsar designation effectively targeting the key sites with the highest numbers of waterbirds and SIBE designation additionally boosting waterbird trends in a range of wetlands with lower initial densities of waterbirds.

Surprisingly, virtually all population trends were positive, despite alarming reports of pronounced population declines for a number of the investigated species (Sanderson et al. 2006; Saino et al. 2011). Rainfall is known to influence bird migration (Gordo et al. 2005; Saino et al. 2007; Zwarts et al. 2009), though exploratory analyses showed that precipitation in Morocco in the rainy period preceding each waterbird census (September–December) was not related to predicted bird abundance (see Fig. 1 for approach; t17 = −0·07; = 0·941). However, precipitation in the Sahel zone was significantly positively related to the mean number of birds observed in Moroccan wetlands (t19 = 2·83; = 0·011; Fig. 1). Rainfall in the Sahel was exceptionally low in the 1980s but increased rapidly in the 1990s and fluctuated around the long-term average during the rest of the study period (Hastenrath & Polzin 2011). It is unclear exactly why precipitation in the Sahel zone should be positively related to the number of waterbirds visiting Morocco during spring migration. However, habitat conditions along the flyway may influence the migration strategy of birds. In years with average rainfall, birds wintering south of the Sahara might use wetlands in the Sahel zone for the energetically favourable hopping strategy (Piersma 1987) and use a westerly detour to cross the Sahara at its narrowest near the Atlantic coast. This would let birds pass over the entire length of Morocco. In dry years, Sahel wetlands would have dried up and the lack of suitable stop-over sites could make the hopping strategy impossible and birds might instead cross the Sahara in a single jump. Under this scenario, birds would cross just the northern tip of Morocco, if at all. Such a rainfall-induced shift in migration patterns could explain why after 2007, even a declining species such as Eurasian curlew Numenius arquata was observed almost six times more frequently than before 2000.

The overall positive population trends in this study are therefore most likely the result of spatiotemporal changes in the migratory routes used by birds and not indicative of actual changes in population numbers. Consequently, our results should not be seen as an indication that all is well with Palearctic-African migratory waterbirds. The lack of a positive response of species with an unfavourable conservation status is a clear signal that the mechanisms underlying the more positive response to conservation designation of waterbirds in general were apparently not strong enough for the most critical species. The numerous threats observed during this study in Ramsar sites furthermore indicate that conservation designation at best slows down habitat degradation to the extent that it attracts a higher proportion of the birds staging or wintering in Moroccan wetlands.

Synthesis and applications

This study shows that it is possible to systematically evaluate the conservation benefits of major international conservation policies using both existing data sets and targeted field studies. It also highlights an important limitation of the approach. While target species groups are often monitored, data on management activities and habitat characteristics are rarely systematically collected. This makes it difficult to interpret the causes of population responses and frustrates the drawing of robust conclusions about the effectiveness of conservation instruments. Moreover, if we do not understand the mechanisms driving species responses to conservation, it will be hazardous to generalize the results of evaluation studies and to learn from successes or failures. Site managers or authorities responsible for the implementation of conservation instruments should therefore consider ways to collect standardized data on the actions taken per site to safeguard biodiversity in protected areas. The type of data collection can range from simple to detailed, depending on constraints of time and money, as long as they provide insights in the general type and frequency of actions. A partial solution to this problem could be to use remote sensing techniques to retrospectively obtain detailed data on changes in habitat quality before and after conservation designation. Although this still does not provide a causal link between conservation activities and habitat quality, it will help us better understand any observed response in the target species group to conservation designation.

So far, we have a very poor understanding to what extent key conservation instruments help achieve the global policy objectives of halting the biodiversity decline (Kleijn et al. 2011; but see Donald et al. 2007). There is a general lack of studies linking the effects of national conservation instruments to global population trends of target species groups. This study could be seen as a first small step towards unravelling the contribution of international conservation tools to halting the biodiversity decline. For a long time and for obvious reasons, the focus of conservation initiatives has been on establishing protected areas. In many countries, the time has now come to shift focus towards evaluating the impact of establishment of these areas and to assess how the conservation benefits of such instruments can be optimized.

Acknowledgements

We thank J. Bovenschen, L. Fishpool, J.A. Gill, S. Nagy, N. Van den Brink, G. Van Dijk and J. Thissen for their help, comments and discussions. This project was part of the Econet project BO-10-011-003. The data reported in this paper are available in the reports cited in the SOM (1985–2000) and available on request from the authors (IC, 2007–2011).

Ancillary