Determining the distribution, population size and temporal trend of threatened species are key actions for assessing their conservation status (Schemske et al. 1994; DG Environment 2023) and for their inclusion in the different categories established for their protection (Mace et al. 2008; IUCN Standards and Petitions Committee 2024). Likewise, these parameters are fundamental for determining the need to implement conservation measures, and for defining their design (IUCN – SSC Species Conservation Planning Sub-Committee 2017). Both, species distribution and abundance, are among the 70 Essential Biodiversity Variables (EBVs) identified by EuropaBON, a standardised framework for biodiversity monitoring and reporting, as relevant for policy-making in the European Union (Junker et al. 2023). Moreover, they are also considered important for the implementation of species conservation initiatives on a global scale (IUCN 2023).

Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora (commonly known as “Habitats Directive”) established Natura 2000, a coherent European ecological network of Special Areas of Conservation (SAC), comprising sites hosting the types of natural habitats and habitats of species that must be protected, which are listed in Annexes I and II respectively (European Commission 1992). The aim of this network is to maintain the conservation status of these habitats as favourable or, where needed, to restore it in their natural range.

For the purposes of the Habitats Directive, the conservation status of a species is determined by the sum of the influences likely to affect the long-term distribution and abundance of its populations, which must be assessed every six years. For each biogeographical region, the species is characterised according to four parameters: range, population, habitat and future prospects. Each of these parameters, as well as the overall conservation status, must be included in one of four available categories: favourable, unfavourable-inadequate, unfavourable-bad, or unknown (DG Environment 2023).

One of the most valuable parameters to evaluate a species conservation status is the variation of its population size over time (trend), since the definition of an overall favourable conservation status can only result from a stable or increasing trend in its population and/or range. For this reason, DG Environment (2023) recommends paying special attention to the methodology of monitoring programmes in order to standardise data acquisition and to improve the quality of trend estimates.

Although the ideal situation would be to census all populations for each species, this is difficult for many taxa, either because they grow in places that are challenging to access, the specimens are difficult to distinguish or have cryptic life stages, or their distribution and population size are too large to allow a complete census (Elzinga et al. 1998). Very often, population sizes must be estimated from a representative sample. However, the uncertainty within this information can lead to flawed estimates and compromise effective conservation actions for threatened species if not properly accounted for, biasing conservation priorities or misguiding the allocation of conservation efforts between species (Regan et al. 2005; Conroy et al. 2008; De Ornellas et al. 2011; Hermoso and Kennard 2012; Robinson et al. 2022). Obtaining reliable information for guiding and assessing species management actions and policies is particularly important when the distribution and abundance of those species may be undergoing significant changes (Jetz et al. 2019).

Data uncertainty can arise from one or more of the following sources: natural variability, semantic uncertainty and measurement error (Akçakaya et al. 2000; Regan et al. 2002; Harstone et al. 2012). Semantic uncertainty arises from vagueness in the definition of terms or from a lack of consistency in their use. Measurement error [epistemic uncertainty sensu Regan et al. (2002) and van der Bles et al. (2019)] is often the largest source of uncertainty and arises from a lack of accurate information, either due to inaccuracies in estimating values or due to the lack of knowledge (Akçakaya et al. 2000).

Uncertainty in population size can be particularly relevant for many species that grow in rocky environments and can be difficult to monitor (García et al. 2019; Aronne et al. 2023). In most cases, difficult access to their locations precludes a full census, and their population size must be estimated indirectly from partial censuses (Alfaro-Saiz et al. 2019a, b) or, more generally, from local estimates of their density multiplied by the total area available to them (Goñi et al. 2006). In these cases, the uncertainty of the results can be greatly increased if the sampling scheme is not statistically robust and if the sampling areas are not representative of the whole area (Elzinga et al. 1998; Iriondo 2011; IUCN Standards and Petitions Committee 2024).

The overall objective of this work is to assess whether the different methods used to determine the population size of threatened flora growing in rocky habitats provide sufficiently reliable information to allow proper management for its conservation. For this purpose, we used Petrocoptis grandiflora Rothm. as a model species. Petrocoptis (Caryophyllaceae) is a small genus of rupicolous species endemic to the Iberian Peninsula (Cires and Fernández Prieto 2015; Calvo-Yuste et al. 2024). Petrocoptis grandiflora is included in Annex II of the Habitats Directive, and is therefore a species of Community interest, whose conservation requires the designation of SACs (so far, three SACs have been designated for its protection), as well as periodic assessments of its conservation status. According to Eionet (2025), P. grandiflora populations could have decreased as much as 43.9% between 2009 and 2017. Therefore, this species was chosen as a relevant model species to reach the overall objective of evaluating the implications of data uncertainties in the assessment of conservation status, since it is a typical narrow endemism growing in rocky walls difficult to survey, it is the subject of special conservation and monitoring measures at European level and, according to some authors, its population trend over the last years is considered highly negative (Tapia et al. 2019).

In addition, this study had the following specific objectives: (1) to analyse the degree of variability and uncertainty in the estimates of the population size of P. grandiflora; (2) to evaluate the effectiveness of a complete census compared to estimates; and (3) to assess the need to adapt the methodologies designed to monitor the conservation status of threatened chasmophytes.

Study area and model species

The fieldwork was carried out in the Serra da Enciña da Lastra Natural Park (province of Ourense, Galicia, Spain) and in neighbouring areas of El Bierzo Region (province of León), all of them in NW Iberian Peninsula (Fig. 1). Serra da Enciña da Lastra has a botanical singularity that was one of the main reasons for its designation as a protected area in 2002 (Xunta de Galicia 2002). El Bierzo region also hosts a very high botanical diversity and many endemic calcicolous chasmophytes are present in both areas (e.g., Armeria rothmaleri, Campanula arvatica subsp. adsurgens, Leontodon farinosus, Petrocoptis grandiflora, Rhamnus pumila subsp. legionensis) (Giménez de Azcárate and Amigo 1996; Llamas et al. 2003; Acedo et al. 2008).

Petrocoptis grandiflora is a narrow endemic chasmophyte that only occurs in a small number of locations between the Autonomous Communities of Galicia and Castilla y León, with an altitudinal range of 380–950 m.a.s.l. Climate is subhumid Mediterranean (Rodriguez and Ramil 2018) with annual precipitation below 1,000 mm and a summer drought lasting 3 months. It is a chamaephyte whose specimens can reach up to 30 cm in length and generally grow isolated in cracks in vertical or overhanging walls associated with the limestone of La Aquiana, preferably north-facing, shady and moist. These walls generally exceed 50–100 m in height (Fig. 2), although specimens are also observed on lower outcrops surrounded by shrub and tree vegetation.

Individual plants are not difficult to detect and identify in bloom, occurring from March to June (mainly in May) (Guitián and Sánchez 1992), when they produce bright purple flowers (see Fig. 3a). The conspicuous flowers and the fact that the individuals grow in low densities in walls with very little vegetation cover make it possible to count every individual when observing walls at close range.

Estimates of population size

In June and July of 2007, Acedo and Miranda (co-authors) surveyed the study area to locate populations of P. grandiflora, visiting several sites and counting all the individuals in selected walls. Most of the observations were done at close distance, and only in particular cases at distances up to 100 m using 8 × 30 binoculars. At each observation point, geographical coordinates were recorded using a hand-held GPS (accuracy < 5.0 m). The area of the available habitat for the species in each wall was measured using Google Earth Pro 4.1.7087 (Google 2007), and was then used to calculate the plant density for each wall. The number of individuals in the remaining zones was estimated by multiplying the area of suitable habitat (calcareous walls and outcrops, also calculated using Google Earth Pro), by the population density of the censused wall with the most similar environmental conditions in each case, following Iriondo (2011).

The results of the 2007 survey were compared with previous estimates of the population size of the species that were carried out in 1990 (Guitián et al. 1993), 1996 (Navarro 1996) and those obtained as part of the Atlas of the Threatened Vascular Flora of Spain project in 2009 (Carbajal et al. 2010) and 2017 (Tapia et al. 2019).

Since each author considered a different number of sites and also named them differently, an equivalence between each author’s sites was defined to account for semantic uncertainty (see Table S1 in Supplementary Material). For this, the spatial information in each of the studies was used when available, or otherwise was extracted from other relevant sources from the same authors mentioning their original investigations. See Fig. S1 in Supplementary Material for further details.

Complete census

During 2023 and 2024, a complete census of P. grandiflora was carried out. First, the sites where the species had previously been described were identified and their boundaries defined. The search included limestone outcrops where the species had not been previously cited, and two additional sites with the presence of the species that were not mentioned in previous surveys (“Peña del Rego” and “Cubelas”) were identified. Thus, in this study, P. grandiflora was censused at 13 sites (Fig. 1). The distance between the study sites varies considerably (Table S2 in Supplementary Material), but they are all geographically distinct areas (“locations” sensu IUCN 2012), regardless of whether they belong to the same or different subpopulations in terms of demographic or genetic exchange.

In 2023, all P. grandiflora individuals present in 12 of the 13 known sites were counted, and in 2024 the remaining site (“Arroyo de la Balouta”) was visited. Censuses were mainly carried out in May, during the flowering peak of the species.

In each of the study sites, to facilitate counting, the whole cliff area was divided into clearly definable sections by using easily identifiable shapes, structures or stone colours, or through other landscape elements. The size of the sections differed notably depending on the observation distance and also on the density of P. grandiflora individuals, and ranged from 10 m² (very close-range, high-density sections) up to 1,000 m² (long-range, low-density sections). Possible “shadow zones” were minimised by choosing observation points that were as close as possible to the specimens, but that also provided the best view of the area (Goñi et al. 2006).

The routes followed to reach the observation points were registered using a hand-held GPS (accuracy 1.8 m), recording also the geographical coordinates of each observation point. The distance from the observation point to the identified individuals was measured with a laser rangefinder (accuracy ± 0.3 m, < 1,000 m; ± 1.0 m, ≥ 1,000 m) when the distance to the specimens was greater than 3 m. At each observation point, two operators counted all the visible individuals, excluding seedlings, either by direct observation at short distances (< 3 m), or by using 8 × 30 binoculars or a 20-60 × 77 spotting scope at longer range. Due to the presence of sensitive bird species (Aquila chrysaetos, Falco peregrinus, Neophron percnopterus) that nest in some of the cliffs, their accessibility could not be further improved, for example by using UAVs (Unmanned Aerial Vehicles) (Strumia et al. 2020; Rominger and Meyer 2021; Tavilla et al. 2024). Counts were cross-checked and repeated when needed until the number of individuals was agreed upon for each section before moving on to the next observation point. All the counts were made on days with good visibility. All the limestone outcrops in the area were surveyed, and absences (sections checked but where no individuals were observed) were also recorded.

To facilitate the comparison of the results with those of previous works, P. grandiflora individuals in flower (or flowered) (defined as “reproductive” individuals) were distinguished from those where reproductive structures were not observed (named as “vegetative”). During the fieldwork in 2023 many dry individuals were observed, most probably because the year prior to the census was extraordinarily dry and hot in Galicia (AEMET 2023; Meteogalicia 2023), and there is some evidence of “important” mortalities for other species of the genus Petrocoptis following periods of severe drought (Guardiola and Sáez 2019). Given that this information may be helpful to understand the adaptations and future responses of cliff communities to climate change (March-Salas et al. 2025), the number of dry individuals was counted. A “dry” individual was defined as a plant that had recently dried-up (within the last one to two years) with its full vegetative structure (including leaves) still intact on the wall (see Fig. 3b). However, we did not count old basal branch remains (Fig. 3c) as dry individuals.

The number of counted individuals was considered as the minimum population size at each site. Correction factors (CF) were also defined for those observations made with binoculars or spotting scope (Goñi et al. 2006). The maximum number of individuals at each monitoring section was estimated by multiplying the number of individuals spotted (“visual units”) by the corresponding CF.

The appropriate authorisation for carrying out the fieldwork in Serra da Enciña da Lastra Natural Park was obtained from Xunta de Galicia (EB-040/2023). Whenever possible, unrestricted roads and paths were used, which was always the case for fieldwork in Castilla y León. In addition, access to the walls with presence of sensitive fauna was avoided during the breeding period.

Evaluation of management implications of uncertainty

Given that P. grandiflora is a species of Community interest, the uncertainty about its population size may have relevant implications for the periodic assessment of its conservation status, which in turn could have an impact on the definition of the necessary management measures. To determine the relevance of the uncertainties in population size estimates of for the assessment of P. grandiflora conservation status, the assessment reports prepared under Article 17 of the Habitats Directive for the periods 2007–2012 and 2013–2018 (Eionet 2025), which are based on estimates from 2009 (Carbajal et al. 2010) and 2017 (Tapia et al. 2019), were compared to an assessment based on the results obtained in the present work. For this purpose, the long-term trend in population size was calculated using the direct counts made in 2007 and in 2023–2024. Since only certain walls were fully counted in 2007, only those areas that were censused on both occasions were included in the analysis. Those sections include 16% of the total population of P. grandiflora in 2023–2024. Line graphs showing trends in estimated or censused population size were created with R 4.5.3 (R Core Team 2026).

Additionally, to evaluate the impact of spatial uncertainty on conservation management, during the 2023–2024 census we identified which sections were located inside Natura 2000 sites, and compared our results with the reported Natura 2000 coverage for P. grandiflora in the last two assessment periods (Eionet 2025). The boundaries of the Natura 2000 sites were obtained from the Spatial Data Infrastructure of MITERD (2025), the information on the SAC designated for the protection of P. grandiflora from the Natura 2000 Viewer (European Environment Agency 2025), and the limits of the Natural Park from Xunta de Galicia (2025). Spatial information was processed in QGIS 3.34.12 (QGIS Development Team 2025).

Estimates of population size

Early studies by Guitián et al. (1993) and Navarro (1996) identified sites of different population sizes, and reported the number of Petrocoptis grandiflora individuals at each site on a semi-quantitative scale. Therefore, the total population size could not be calculated from these data (see Table S3 in Supplementary Material).

Although it was clear in Navarro (1996) that obtaining precise estimates of the population size of P. grandiflora was not one of the main goals of this work, the density of individuals was measured in several plots and the area available to the species was estimated at each study site. This allowed us to estimate the total population size at 15,409 − 18,696 reproductive individuals (see Table S4 in Supplementary Material). This is close to the figures given by Izco (2005) who, also based on Navarro (1996), estimated a total population of “18,000–20,000 reproductive individuals”.

During our fieldwork in 2007, a total of 3,996 living P. grandiflora individuals were counted in the censused walls, and the presence of a further 7,085 individuals was estimated in the remaining walls in the surveyed locations, bringing the total number of individuals in the eight visited sites to 11,081 (Table 1). Density values per site (0.005–0.77 ind./m², median = 0.12 ind./m²) are notably different from those in Navarro (1996) (0.21–0.66 ind./m², median = 0.43 ind./m²), as are the suitable area and the estimated population size (see Tables S4 and S5 in Supplementary Material).

In contrast, the population size estimated in 2009 by Carbajal et al. (2010) was much higher (48,747 individuals) than those reported in previous studies. Even a partial estimate for two of the sites exceeded 21,000 individuals (Table S6 in Supplementary Material).

Finally, Tapia et al. (2019) estimated a total population size of between 27,422 and 33,982 individuals in 2017, resulting in an estimated mean change of -37.2% between 2009 and 2017. The change rate was also identical across all sites (Table S6 in Supplementary Material).

Even in equivalent sites, published estimates of the population size of P. grandiflora varied greatly between different studies, and were also different from our own estimates (Fig. 4a).

Complete census

A total of 37 days of fieldwork were required in 2023–2024 to complete the census, 11 of which were dedicated exclusively to identifying and characterising the sites, and 26 to counting individuals. The 26 counting days represented 136 h of work, 33 of which were spent doing the counting in 157 sections, while the remainder were needed for transport and accessing the observation points.

In total, 14,231 living individuals of P. grandiflora were counted at the 13 monitoring sites, 11,183 of which were reproductive and 3,048 vegetative (Table 1). Additionally, 2,412 dry individuals were counted, with higher percentages in the SW-oriented sites. Consequently, a total of 16,643 P. grandiflora specimens were counted in this study (see Table S7 in Supplementary Material). After adjusting the raw observations where necessary using CF, the maximum total number of P. grandiflora was found to be 16,588 living (13,183 reproductive and 3,405 vegetative) and 2,587 dry individuals (see Table S7 in Supplementary Material). These results differ notably from previous estimates, both for individual sites and for total values (see Table S8 in Supplementary Material).

Management implications of uncertainty

The Petrocoptis grandiflora assessment report for 2007–2012 (Eionet 2025) was based entirely on the results of Carbajal et al. (2010), where the population size was estimated at 48,747 individuals. While the population trend from 2001 to 2012 was assessed as “stable” and the population size as “favourable”, the overall conservation status was considered “unknown” due to a purported lack of information regarding the species’ habitat.

Similarly, the assessment report for 2013–2018 was based exclusively on the results of the National Monitoring Project (Carbajal et al. 2010; Tapia et al. 2019). Consequently, the reported population size remained consistent with that reported by Tapia et al. (2019), and the estimated population trend ranged from a minimum of -30,5% and a maximum of -43,9%. Given the negative trend in population size, the conservation status of P. grandiflora was categorised as “unfavourable – inadequate”.

In the present work, the counts made in 2007 and 2023–2024 in four selected sections were very similar (see Table S9 in Supplementary Material), and accordingly the long-term trend in population size can be considered stable, at least in these sections (Fig. 4b). Therefore, based on this stable trend in population size, the conservation status of P. grandiflora should be considered “favourable”.

Finally, the assessment report for the period 2007–2012 concluded that all 48,747 estimated individuals of P. grandiflora were located within Natura 2000 sites. The assessment report for the period 2013–2018, expressed Natura 2000 coverage for P. grandiflora as 3–5 1 × 1 UTM squares out of 26 within the SAC network. However, according to the fine-scale information collected during the 2023–2024 census, only 6,454 of the 14,231 living P. grandiflora specimens counted (45.4%) were found within Natura 2000 sites.

The simplest approach to determine whether the population size of a rare or threatened species is increasing, decreasing or stable is to conduct a census of the number of individuals through time (Schemske et al. 1994). However, depending on the species, its habitat, or its growth pattern, a complete count is not always feasible, and different methods must be employed to estimate population size (Elzinga et al. 1998; DG Environment 2023). In such cases, great care must be taken to ensure the robustness of the sampling scheme, and to account for uncertainty in data collection and the presentation of results (IUCN Standards and Petitions Committee 2024).

In this study we used an endangered, narrow-endemic chasmophyte as a model species to compare the robustness of indirect methods of estimating population size and trend (based on density and available area) with the censused total abundance and its trend. The results in previous studies suggest significant variation in population size over time, which is not a typical characteristic of stenoecious long-lived rock-dwelling species, such as Petrocoptis grandiflora (García et al. 2021; Múgica et al. 2024). Our results show that estimates vary considerably between studies and are subject to high levels of the most common sources of uncertainty (Akçakaya et al. 2000).

One of the main sources of uncertainty detected is semantics. In this case, spatial information was used to reduce ambiguity (Elith et al. 2002), although the equivalence will not be exact due to the lack of detailed spatial information in previous studies. It is therefore important to emphasise the need to include this type of information (at least GPS coordinates of observation points and polygon features representing the areas surveyed or occupied by the species) in all studies on the conservation status of threatened species, so that they can be compared over time.

Other major source of uncertainty was the methodology used to estimate the population size of P. grandiflora. Indirect methods are based on sampling, so in order to correctly estimate the species abundance, samples have to be representative of its spatial reality. In the case of P. grandiflora, the density of individuals varies greatly depending on the structure of the substrate rock, even reaching a density of up to 4 individuals/m² (Guitián and Sánchez 1992). This high spatial variability must have been the source of the large differences in density observed between our study and previous works, which suggests that a greater sampling effort could be necessary to improve the representativeness of the samples taken. Also, the area considered available to the species can vary considerably between studies. Understanding the environmental or biological needs of the species to be surveyed must be a priority when beginning this monitoring.

The effectiveness of conservation management can be greatly enhanced by focusing monitoring efforts on key information needs (Nichols and Williams 2006). Following recent proposals to control uncertainty in the study of population size trends of threatened plants over time (Miranda-Cebrián et al. 2025), we propose that monitoring approaches should be tailored on a case-by-case basis (e.g., according to the species or the objective), while still being based on widely standardised methodologies, in order to minimise the uncertainty of the information collected.

A simple and robust method to determine the population trend of P. grandiflora (or other chasmophytes) would be an annual count of the number of individuals in a series of fixed plots representative of the main sites where it occurs, supported by photo-sampling (García et al. 2021; Harrison et al. 2024). By focusing on a smaller proportion of the population, more attention can be given to improving the accuracy of the data (Elzinga et al. 1998), and the amount of uncertainty in monitoring can be reduced by improving quality control and fieldwork protocols (García et al. 2021; Miranda-Cebrián et al. 2025).

The true population size could be obtained with a complete census, varying the degree of precision as little as possible over the years in the long-term (Morris and Doak 2002) and maintaining the quality and consistency of data (Lovett et al. 2007). Due to the inconsistency of data in previous studies, in this study we have included several improvements on the monitoring methodology applied to make the global censuses: assessing survey effort, subdividing the study area to improve count accuracy, using double-counting to reduce observation error, registering and reporting all the information with sufficient detail. These implementations could be extensible to the study of any narrow endemic chasmophyte.

Two of the main mechanisms of protection and conservation of the species in Annex II of the Habitats Directive are the designation of SACs (and the establishment of the necessary conservation measures within them) and surveillance and reporting of their conservation status. We have shown that the effectiveness of these measures could be jeopardised if uncertainty in estimates of the population size of P. grandiflora and its trend is not adequately accounted for. The spatial components of the semantic (“underspecificity” sensu Elith et al. 2002) and measurement (“positional” sensu Moudrý et al. 2024) uncertainties prevent correct assignment to a SAC in certain cases, resulting in the assignment to Natura 2000 of all plots (report 2007–2012), only 20% of the UTM squares (report 2013–2018), or 50% of the population (this study, using fine-scale spatial information). The results of the periodic assessment of the conservation status of P. grandiflora were also highly influenced by measurement error. The categorisation of the overall conservation status of this endangered species has proven to be highly dependent on the reliability of the population trend based on estimates, and the use of complete counts eliminated the effect of uncertainty. Other studies have found similar results of the effect of survey methodology on conservation assessments (Teale et al. 2025). In this way, uncertain estimates of the total population size, and misleading trends derived from them, could misguide the allocation of conservation efforts, transferring the consequences of uncertainty to other endangered species and becoming “opportunity costs” for them (Conroy et al. 2008).

Rocky outcrops are highly diverse formations that require particular monitoring approaches to improve the mapping of species distribution and the quantification of the populations that inhabit them (Fitzsimons and Michael 2017; March-Salas et al. 2023; Harrison et al. 2024). Calcareous or siliceous rock outcrops with chasmophytic vegetation are among the habitats of Community interest whose conservation requires the designation of SACs, and many of the plant species that inhabit them are rare or threatened endemic species. In order to promote adequate knowledge of them and to allow their conservation and the adoption of appropriate management measures, it is essential to have more accurate long-term monitoring data (Bayraktarov et al. 2019), and to better account for and communicate data uncertainty (Huntley et al. 2016; White et al. 2023).

Assessing the conservation status of threatened species is a crucial tool in conservation decision-making, as it allows prioritising the allocation of limited resources and determining the need for and design of future conservation actions.

Nevertheless, through the case of the narrow-endemic rock-dwelling Petrocoptis grandiflora, we have observed that high levels of uncertainty caused by different cumulative factors in obtaining population size estimates can prevent reliable conclusions about the conservation status of a species, and can have other major implications for conservation management.

If the sources of uncertainty are not correctly identified during data collection and if the necessary measures are not taken to minimise them, the biases arising from this uncertainty are transferred to the decision-making process, which ultimately compromises the success of existing mechanisms for species management and, therefore, for their conservation.

Authors

1. Elisa Gago
2. Marcos Lado
3. Carmen Acedo
4. Bernardo Miranda
5. Elvira Sahuquillo

Cite this article

Gago, E., Lado, M., Acedo, C. et al. Highly variable and uncertain estimates of the population size of endangered narrow endemic chasmophytes have profound implications for conservation management. Biodivers Conserv 35, 203 (2026). https://doi.org/10.1007/s10531-026-03399-5

Download citation

• Received: 30 June 2025

• Revised: 09 June 2026

• Accepted: 16 June 2026

• Published: 30 June 2026

• Version of record: 30 June 2026

• DOI: https://doi.org/10.1007/s10531-026-03399-5

Keywords