3.1 Methods of habitat quality assessment

Measuring demographic parameters

By definition, the quantification of habitat quality depends on estimating a site’s contribution to survival and reproduction. Therefore, any method that directly measures one or both of these parameters will be free from error propagation caused by imperfect correlation between a measured attribute and these variables. However, cautious interpretation is still required when only one of these attributes is measured, because sites with similar reproductive output can have divergent population trajectories if the population size is governed by adult survival, and vice versa (Cohen et al., 2009). Similarly, emigration or immigration at a site may also obscure the signal arising from measures of reproduction and survival (Cohen et al., 2009). Measuring demographic rates can be a lengthy, costly, and logistically challenging process. In waterbird research, directly measuring a site’s contribution to survival and reproduction may be unachievable owing to the mobility of waterbird populations. Although our structured search returned examples of studies that did quantify survival (e.g., Alves et al., 2013, Rice et al., 2007, Swift et al., 2020) and/or reproduction (e.g., Hunt et al., 2017, Powell and Powell, 1986, Swift et al., 2020), most studies used proxies for one or both of these measures.

Estimating food abundance and availability

Many of the proxies in the reviewed studies assumed that a high-quality habitat provided waterbird individuals with a high net energy intake rate. The corollary assumption was that high net energy intake results in increased survival and reproductive performance. Methods used to infer net energy intake rate included measures of prey abundance, prey accessibility, and waterbird physiology or morphology as an indicator of past foraging returns (Table 1). Habitat quality assessments that are based on habitat attributes are appealing because results are independent of variation in bird behaviour caused by factors unrelated to local habitat quality (e.g., current wind and rain conditions can determine which sites waterbirds use at very local scales (Kelly, 2001) and these short term changes are not typically useful for managers). For this reason, measuring the abundance or biomass of food was used widely in the reviewed studies to assess waterbird habitat quality. There is support for this method being an appropriate proxy for habitat quality because waterbirds preferentially forage at sites with the highest prey biomass and density (Guerra et al., 2016, Rose and Nol, 2010). Moreover, prey availability has a positive influence on reproductive performance and survival (Herring et al., 2010, Holopainen et al., 2014, Swift et al., 2020). However, there are also situations where prey biomass at a site can be a poor indicator of habitat quality. For example, sites with high prey biomass are not always favoured foraging sites (Hagy and Kaminski, 2015), and although these sites might have high occupancy, they do not necessarily support high waterbird abundance (Gillespie and Fontaine, 2017). This suggests that factors such as predation risk, forager condition, and prey accessibility modulate the effect of prey biomass on habitat quality (Hagy and Kaminski, 2015). Such a relationship is also dependent on waterbirds having a perfect knowledge of the distribution of prey resources (Reurink et al., 2015), which may not always be the case (Lewis et al., 2010), and relies on researchers correctly identifying dietary preferences and requirements of focal species.
The presence of suitable water levels and variation in water levels was also used as a proxy for habitat quality in the reviewed studies. These habitat attributes can influence accessibility of prey and foraging energetics (Ma et al., 2010). In some cases only a small proportion of a wetland provides suitable water levels for waterbirds to access prey (Collazo et al., 2002). This suggests that there is value in quantifying either prey biomass or the amount of suitable habitat through water level measurements. However, the two attributes will interact to influence the net rate of energy intake possible at a site meaning studies that measure both variables may have a greater likelihood of teasing apart meaningful habitat quality relationships and informing appropriate management (Herring and Gawlik, 2013). Similarly, remotely sensed measures of primary productivity (e.g., NDVI) are expected to be correlated with prey abundance. Yet, the relationship between net energy intake rate and primary productivity is dependent on changes in primary productivity causing changes in prey abundance (e.g., invertebrates, seeds, tubers) as well as those prey items being available to feeding waterbirds (Guan et al., 2016, Zhang et al., 2017). This suggests there is a hierarchy in the ability of proxies to provide precise habitat quality estimates based on how direct the link between the variable being measured and net energy intake rates is (Figure 1).

Estimating food intake rate

The behaviour and habitat use patterns of waterbirds themselves were often used in the reviewed studies to infer underlying patterns of habitat quality (Table 1). Indicators of prey intake rate (be it current, past or expected future foraging returns) were frequently used metrics of habitat quality. Variables including peck rate, capture success rate, and the proportion of time a bird spent foraging were commonly measured to assess the current rate of energy intake supported by a habitat. Defecation rate is significantly correlated with peck rate in a visually foraging shorebird, supporting the assumption that peck rate represents a valid indicator of intake rate (Rose and Nol, 2010). Likewise, sites with a higher peck rate or probe rate had a higher rate of successful prey captures in a study where capture success could be visually verified (Kuwae et al., 2010). However, different prey items have different energy content and different processing costs within the digestive system (Dugger et al., 2007, Jorde et al., 1995). This means that the net rate of energy intake will depend on the prey type consumed. This may not be an issue in studies of diet specialists, but it may confound the interpretation of peck rate and capture success data for diet generalists. In situations where the diet of the population being studied is not well understood, investigating the prey community composition to determine prey encounter rates, or dietary studies (e.g., metabarcoding of prey DNA sequences in faecal samples) will inform whether differences in peck rate between sites or across time genuinely reflect changes in energy returns.
Intake rates over the recent and more distant past were inferred from a variety of variables including body condition, blood metabolites, and indicators of feather growth rate. These have the advantage that they reflect assimilated energy rather than gross intake including energy lost via excretion or through processing costs. However, the longer timeframe of integration meant that studies using these methods were rarely site-specific, rather they tended to assess habitat quality at regional scales (e.g., Aharon-Rotman et al., 2016b). In cases where individuals use only a small geographic area (e.g., when nesting constrains movements, or individuals have strong residency patterns) these measures may provide insights into site-specific habitat quality. For example, Swift et al. (2020) found that visually-scored body condition of non-breeding Hudsonian Godwits Limosa haemastica was correlated with pecking rate at individual non-breeding sites. This suggests that these birds were resident at sites long enough to integrate site-specific habitat quality information in the form of body condition. Importantly, birds with higher body condition had higher survival and reproductive output the following breeding season, indicating that body condition reliably influenced demographic rates (Swift et al., 2020).

Predation pressure

Given the direct link between predation pressure and survival, it was surprising that predation pressure was estimated relatively infrequently in the reviewed studies. This is perhaps reflective of the difficulties of censusing predator populations due to predators of waterbirds typically occurring at low density and predation events on adult waterbirds being rare. Where predation pressure was quantified, these studies often focused on nest predation (e.g., Kenow et al., 2009, Pehlak and Lõhmus, 2008, Trinder et al., 2009). Most studies that inferred an influence of predation pressure on habitat quality assumed that the abundance of predators was correlated with predation rate without explicitly testing this assumption, which may be problematic when generalist predators are involved. Some studies also assessed predation pressure by using vigilance or escape behaviours of waterbirds (Fernández and Lank, 2010, Gunness et al., 2001). This has the advantage of integrating information on the degree of lost foraging time as a result of predation pressure because lost foraging opportunities will affect reproductive performance as well as survival (Castillo-Guerrero et al., 2009).

Physical habitat attributes

Many of the reviewed studies measured various physical and/or chemical attributes of waterbird habitats to infer habitat quality. The attributes measured were purported to influence habitat quality via their contribution to supporting viable prey populations (e.g., water pH, water conductivity, sediment grain size), enabling access to sufficient quantities of food (e.g., water area, pond density in the local area and vegetation composition, as well as water level which we discussed previously), or providing shelter from predators (e.g., vegetation structure). In most cases, these environment attributes are linked indirectly to demographic rates (Figure 1) and the mechanisms governing their effects may be difficult to disentangle (Raquel et al., 2016). Nonetheless, physical attributes of the habitat may provide waterbirds with visual cues as to the quality of a site and play a role in determining patterns of site use, which can have flow-on effects on demographic rates (Buderman et al., 2020).

Other methods

A variety of other methods were used infrequently in the reviewed studies (Table 1). These included estimates of levels of human disturbance, individual movement data (e.g., home range size), and the spatial distribution of individuals in different age classes. Despite their infrequent use, these methods may provide meaningful habitat quality information. Factors such as the cost of obtaining the data or the difficulty of obtaining the data (e.g., challenges distinguishing between age classes in the field) probably contributed to their infrequent use.

Combination of methods

Many of the reviewed studies recorded data on multiple proxies for habitat quality. Multiple lines of evidence allowed researchers to tease apart complex relationships among various parameters in their respective study systems and provide powerful insight to conservation managers (Cohen et al., 2009, Hunt et al., 2017, Swift et al., 2020). In these studies, it was often possible to pinpoint factors that were limiting habitat quality, providing managers with priorities to address in order to improve habitat quality. For example, Cohen et al. (2009) recommended that restoring Piping Plover, Charadrius melodus, habitat adjacent to bayside intertidal flats would improve habitat quality by increasing the number of breeding pairs that could occupy a site. However, this action must be carried out in conjunction with predator management in order to achieve the desired increase in reproductive output.

Factors influencing the choice of variables to measure

Staying within the project’s scope

Our synthesis of the habitat quality literature indicates that there is a hierarchy of data quality from directly monitoring demographic rates to measuring parameters that are increasingly indirectly linked to demography. Yet, practitioners typically face a trade-off between the need for accuracy of the habitat quality estimate and their particular study’s aims and constraints. If it is feasible, measuring demographic rates directly generally involves extended field time, individually marked birds, limited spatial scale, and substantial costs (Buderman et al., 2020). Other factors may also influence the suitability of a proxy for the habitat quality assessment at hand including ethical considerations (Hunt et al., 2013), and the availability of appropriately trained personnel. Physiological and morphological measurements used in the reviewed studies typically required birds to be handled (but see the abdominal profile index method; Swift et al., 2020), which imposes stress on the study subjects (Karlíková et al., 2018), and capturing a large sample size of birds can be time-consuming. This may mean that methods requiring birds to be handled, including individually marking birds for quantifying demographic rates, are not feasible within the scope of a project.

Spatial and temporal scales of assessments

Another consideration that must be made prior to implementing a study on habitat quality is whether the habitat quality measure being used returns data at a relevant spatial and/or temporal scale. For example, prey abundance measures typically provide very local scale (both spatial and temporal) information on habitat conditions, but may not be representative of habitat quality across the entire wetland or extended timeframes (e.g., the entire non-breeding period). For example, Fonseca and Navedo (2020) reported a 43% reduction in invertebrate prey biomass as a result of shorebird foraging in study plots over the course of three days. Consequently, habitat quality assessments either side of this three-day period could yield vastly different inferences about local habitat quality and neither may be representative of habitat quality over an extended timeframe. The accuracy of these methods in terms of returning habitat quality data at time-scales meaningful for management will therefore be increased by repeated sampling (Murray et al., 2010). This was reflected in a number of the reviewed studies, especially those aimed at specifying management regimes, repeating sampling both spatially, and intra- and inter-annually (e.g., Gillespie and Fontaine, 2017). Whereas methods that relied on measuring attributes of the habitat typically provided snapshot estimates of habitat quality, methods reliant on waterbird body condition or physiology (e.g., abdominal profile index or red blood cell heat shock protein concentrations) often provide information integrated over longer timeframes (Herring and Gawlik, 2013). They may therefore be unsuitable for site-specific and/or instantaneous habitat quality questions, but may be applied to questions informing management of a regional wetland complex over broader timeframes. Similarly, remotely sensed measures of primary productivity offer the potential to rapidly and cost-effectively monitor habitat conditions at large spatial and temporal scales. For example, Wen et al. (2016) used remotely sensed primary productivity data to inform an assessment of waterbird habitat quality across a 810,000 km2 study area in multiple years.
There is no rule that governs whether the spatial or temporal scale of a particular proxy is appropriate for a particular application because even labour-intensive or costly methods that return site-specific information may be suitable for large-scale projects if the budget enables sufficiently widespread sampling (e.g., sites and time points). We provide some recommendations as to the spatial and temporal scales that methods for habitat quality assessments are typically carried out at (Table 1). Readers may also find papers such as Behney and colleagues’ (2014) guide to determining the optimum number of benthic core samples to collect useful for planning how much field effort is likely to be involved when planning a sampling regime.

What makes for a good habitat quality assessment?

Measuring habitat quality enables conservation managers to assess the need for or effectiveness of management actions (e.g., Schultz et al., 2020). The ultimate objective of conservation management is to influence demographic parameters of conservation targets to improve conservation status. Therefore, assessments of habitat quality inherently must determine a site’s contribution to survival probability and/or reproductive output. This requires there to be a link between the variable, or combination of variables, used to measure habitat quality and demographic rates (Figure 1). Before commencing an assessment of habitat quality, the researcher must carefully consider whether the selected measure does actually influence demographic rates. For example, quantifying the time budgets of waterbirds is a commonly used method for inferring differences in habitat quality (Dugger and Feddersen, 2009, van der Kolk et al., 2019). However, the inferences derived from time budget comparisons may not actually reflect changes in underlying habitat quality. Time budgets can be flexible to buffer intrinsic changes in requirements (Mallory et al., 1999). For example, this may be due to individuals dedicating more time to foraging to meet the metabolic demands of producing a clutch of eggs (Mallory et al., 1999), or dedicating more time to feeding to fatten up for migration (Castillo-Guerrero et al., 2009). That is not to say that time budgets are unsuitable for quantifying habitat quality, but care must be taken to ensure that appropriate comparison groups are being used (e.g., sampling at the same time of year).
Researchers must also be aware that inferences made about populations that are not at equilibrium may depart from theoretical relationships underpinning many habitat quality proxies. For example, populations that have been reduced below carrying capacity by historical or offsite factors may not show any temporal differences in various local habitat quality proxies (e.g., foraging success, stress markers, body condition, and time budgets) because individuals are easily able to meet their resource requirements even if local habitat quality is declining. Similarly, there may be differences in the relevance of some habitat quality proxies depending on whether the conservation target is a resident population, or a dispersive or migratory population (Loewenthal et al., 2015). Abundance and density are clearly linked to local habitat quality for resident populations, but may not be truly reflective of local habitat quality for populations that undertake large-scale movements exposing individuals to factors that limit population size elsewhere in the range. For example, Jia et al. (2018) reported declines in abundance of migratory shorebirds at a migratory staging site, but none of the measured proxy variables for habitat quality could explain these declines. They suggest that factors in other parts of the migratory range may be responsible for driving the observed declines in abundance rather than changes in habitat quality at their study site.
Many of the habitat quality proxies identified in this review assume individuals have perfect knowledge of the resource distribution at a site and behave such that the net rate of energy gain is being maximised at any given time (Reurink et al., 2015). Several factors can result in waterbirds using their habitat in ways that do not conform to these assumptions. The choice of foraging site for many waterbirds is strongly influenced by conspecific attraction (Gawlik and Crozier, 2007, Herring et al., 2015, Smith, 1995). This is also true for the selection of nest sites (Sebastián-González et al., 2010c). Furthermore, fidelity to areas that have provided favourable habitat conditions in the past may decouple patterns of waterbird habitat use from current habitat conditions (O’Neil et al., 2014). Waterbird habitat requirements may also change with breeding stage (Holopainen et al., 2014), and during less energetically demanding parts of the annual cycle, such as the non-breeding period, individuals may be less selective in their habitat use decisions (Sebastián-González et al., 2010b).
Most of the reviewed studies provided a relative assessment of habitat quality (i.e., they compared waterbird habitat quality at a site to previous points in time, or made comparisons between sites). These studies allow researchers to determine habitat quality trends or identify the best and worst sites in a landscape, but do not enable managers to determine whether the habitat quality is sufficient to maintain viable waterbird populations. There were some studies that sought to determine whether the habitat quality at a site was sufficient to support population growth or whether the site represented a sink habitat (e.g., Roy et al., 2019, Sabatier et al., 2010). These studies do enable managers to determine whether management intervention is necessary rather than arbitrarily setting a reference site as the standard against which to decide whether management is warranted. In particular, studies seeking to identify whether a site had sufficient habitat quality to support population growth tended to focus directly on reproductive output or survival data (Roy et al., 2019, Weiser et al., 2018), or in some cases focused on energetic demands relative to prey resources (West et al., 2005).
Together, the potentially confounding factors mean that there is no universally applicable habitat quality proxy. Yet, with careful consideration and a detailed understanding of the ecology of the study system, waterbird researchers and management practitioners can derive meaningful measures of habitat quality.