top of page

Modelling Green Hound's-tongue Cynoglossum germanicum 

translocation success to inform future site selection 

Green Hound's-tongue Cynoglossum germanicum is a Schedule 8, nationally scarce woodland herb belonging to the Boraginaceae. Following a translocation and seeding effort in woodlands in Hampshire, environmental data was gathered from translocation sites in attempts to discern key predictors of success. The hope is that the results of this analysis will guide future translocation efforts. 

Large numbers of the Schedule 8 plant Green Hound’s-tongue Cynoglossum germanicum were found within a development footprint in Hampshire, UK. As such, a licence was sought from Natural England to enable translocation of viable numbers of C. germanicum to neighbouring habitats. From September to December 2022, 2,988 plants were translocated into 20 plots. In addition, approximately 18,700 seeds were sown (some adjacent to translocation sites and some into new plots). Quarterly monitoring visits were subsequently undertaken by myself and other colleagues at Aspect Ecology, and staff from the Species Recovery Trust.

Translocations were variable in success, ranging from increased plot numbers (as of Autumn monitoring 2023) to plot failure. To inform locations of future translocation and seeding, I wanted to know which environmental variables were the most important for plot success. Plot data was collected in September 2023. The analysis comprised of Kruskal-Wallis tests for ordinal variables (moss cover, woodland community class) and linear mixed modelling for continuous variables. The LMM fitted ‘clusters’ (plots in close proximity) and ‘woodland’ (specific woodland each plot belonged to) as nested random effects.

Headline Results



Kruskal-Wallis testing revealed no significant difference in number of plants for:

a) different soil types;

b) different levels of moss cover; and

c) different woodland communities


Mean pH, mean moisture, bulk density, and rhododendron proximity varied between ‘clusters’ of plots (grouped by proximity)


Large amounts of variation remained unexplained by the LMM, with individual explanatory variables accounting for relatively small amounts of variation


Plot type (translocation vs. seeded) accounts for large amounts of variation – see point 2 in Next Steps


In a non-linear approach, pH and moisture had the highest (though non-significant) estimates


Variation within explanatory variables was minimal – environmental variables tended to be similar at each plot


AICc scores equivalent for all models of varying complexity


  • In light of no clear answer on key environmental drivers of translocation success, new seed plots have been sown in linear strips (50m) to encapsulate a wider range of conditions. These strips have primarily been sown along woodland paths in the hope that dogs and other mammals will help spread the sticky burs of next year’s plants.

  • Even if masked by large amounts of unexplained variation, higher pH appeared to have a small negative effect on plant numbers (along with moisture). As such, consideration is being given to additional translocation / seeding attempts in calcareous soils. Though found thriving here in slightly acidic soils, C. germanicum is traditionally reported as a calcareous woodland species and working in similar habitats should aid future translocation.

  • Similar, bulk density appears to be an important variable. Soil compaction was related to plant numbers cubically, with peaks at intermediate and high levels of compaction. Future translocations can use this to guide site selection going forward.

Next Steps

  • Fit an LMM with random slopes as well as random intercepts – analysis ongoing

  • Separate the data – future population boosting efforts will likely comprise seeding, so the effect of environment on only seed plots should be considered – analysis ongoing

Click here to download a full method report IN PREPARATION.


Click here to access my GitHub repository (with annotated code). 

bottom of page